Hugging Face's logo

Datasets:
FineEnvs
/
HF_ML_Tasksmith

RL Environment
Languages:
English
Size:
n<1K
Tags:
reinforcement-learning
coding
harbor
repo2rlenv
tasksmith
Libraries:
Harbor
License:
Dataset card Files
xet
 Community
AdithyaSK HF Staff commited on
Commit
62e8dff
·
verified ·
1 Parent(s): 44c4a73

Stage Harbor release files 104001-104500

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/miss/llama-3.2-3B-bat/adapter_config.json +18 -0
  2. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/miss/llama-3.2-3B-default/adapter_config.json +18 -0
  3. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/miss/llama-3.2-3B-mini/adapter_config.json +18 -0
  4. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/oft/llama-3.2-3B-rank32/adapter_config.json +27 -0
  5. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/osf/llama-3.2-3B-rank128/adapter_config.json +28 -0
  6. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/osf/llama-3.2-3B-rank128/training_params.json +7 -0
  7. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prefixtuning/llama-3.2-3B-lr_0.001/adapter_config.json +15 -0
  8. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prefixtuning/llama-3.2-3B-lr_0.001/training_params.json +6 -0
  9. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-default/adapter_config.json +17 -0
  10. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-lr_0.001/adapter_config.json +17 -0
  11. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-lr_0.001/training_params.json +6 -0
  12. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-sample_vocab-lr_0.001/adapter_config.json +17 -0
  13. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-sample_vocab-lr_0.001/training_params.json +6 -0
  14. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/psoft/llama-3.2-3B-default/adapter_config.json +26 -0
  15. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/psoft/llama-3.2-3B-fast/adapter_config.json +26 -0
  16. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/ptuning/llama-3.2-3B-default/adapter_config.json +17 -0
  17. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/pvera/llama-3.2-3B-default/adapter_config.json +20 -0
  18. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/pvera/llama-3.2-3B-default/training_params.json +6 -0
  19. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/randlora/llama-3.2-3B-default/adapter_config.json +22 -0
  20. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/road/llama-3.2-3B-lr_0.001/adapter_config.json +12 -0
  21. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/road/llama-3.2-3B-lr_0.001/training_params.json +5 -0
  22. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/shira/llama-3.2-3B-lr_0.0003-random_seed_42/adapter_config.json +15 -0
  23. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/shira/llama-3.2-3B-lr_0.0003-random_seed_42/training_params.json +6 -0
  24. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/trainable_tokens/llama-3.2-3B-sos+eos/adapter_config.json +7 -0
  25. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/trainable_tokens/llama-3.2-3B-sos+eos/training_params.json +5 -0
  26. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/vblora/llama-3.2-3B-default/adapter_config.json +26 -0
  27. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/vera/llama-3.2-3B-default/adapter_config.json +20 -0
  28. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/vera/llama-3.2-3B-default/training_params.json +6 -0
  29. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/waveft/llama-3.2-3B-n_frequency-5000/adapter_config.json +26 -0
  30. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/requirements.txt +4 -0
  31. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/.gitkeep +0 -0
  32. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/adalora--llama-3.2-3B-rank32.json +4080 -0
  33. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/adaptionprompt--llama-3.2-3B-lr_0.0005.json +343 -0
  34. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/boft--llama-3.2-3B-default.json +356 -0
  35. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/bone--llama-3.2-3B-bat.json +352 -0
  36. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/bone--llama-3.2-3B-default.json +352 -0
  37. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/c3a--llama-3.2-3B-default.json +352 -0
  38. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/delora--llama-3.2-3B-rank32.json +355 -0
  39. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/fourierft--llama-3.2-3B-default.json +356 -0
  40. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/fourierft--llama-3.2-3B-n_frequency-5000.json +356 -0
  41. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/full-finetuning--llama-3.2-3B-lr_0.00001.json +332 -0
  42. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/gralora--llama-3.2-3B-rank32.json +355 -0
  43. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/ia3--llama-3.2-3B-default.json +353 -0
  44. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/ia3--llama-3.2-3B-lr_0.001.json +352 -0
  45. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/ln_tuning--llama-3.2-3B-default.json +348 -0
  46. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/loha--llama-3.2-3B-rank32.json +357 -0
  47. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/lokr--llama-3.2-3B-rank32.json +360 -0
  48. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/lora--llama-3.2-3B-rank10-target-mlp.json +375 -0
  49. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/lora--llama-3.2-3B-rank14-target-mlp-bdlora.json +388 -0
  50. tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/lora--llama-3.2-3B-rank32-dora.json +374 -0
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/miss/llama-3.2-3B-bat/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "bias": "none",
5
+ "exclude_modules": null,
6
+ "inference_mode": false,
7
+ "init_weights": "bat",
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "mini_r": 1,
11
+ "miss_dropout": 0.0,
12
+ "modules_to_save": null,
13
+ "peft_type": "MISS",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": null,
17
+ "task_type": null
18
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/miss/llama-3.2-3B-default/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "bias": "none",
5
+ "exclude_modules": null,
6
+ "inference_mode": false,
7
+ "init_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "mini_r": 1,
11
+ "miss_dropout": 0.0,
12
+ "modules_to_save": null,
13
+ "peft_type": "MISS",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": null,
17
+ "task_type": null
18
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/miss/llama-3.2-3B-mini/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "bias": "none",
5
+ "exclude_modules": null,
6
+ "inference_mode": false,
7
+ "init_weights": "mini",
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "mini_r": 64,
11
+ "miss_dropout": 0.0,
12
+ "modules_to_save": null,
13
+ "peft_type": "MISS",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": null,
17
+ "task_type": null
18
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/oft/llama-3.2-3B-rank32/adapter_config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": null,
5
+ "bias": "none",
6
+ "block_share": false,
7
+ "coft": false,
8
+ "eps": 6e-05,
9
+ "exclude_modules": null,
10
+ "fan_in_fan_out": false,
11
+ "inference_mode": false,
12
+ "init_weights": true,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "module_dropout": 0.0,
16
+ "modules_to_save": null,
17
+ "oft_block_size": 0,
18
+ "peft_type": "OFT",
19
+ "r": 32,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "q_proj",
24
+ "v_proj"
25
+ ],
26
+ "task_type": null
27
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/osf/llama-3.2-3B-rank128/adapter_config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_type": null,
3
+ "peft_type": "OSF",
4
+ "auto_mapping": null,
5
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
6
+ "revision": null,
7
+ "inference_mode": false,
8
+ "effective_rank": null,
9
+ "target_modules": [
10
+ "q_proj",
11
+ "k_proj",
12
+ "v_proj",
13
+ "o_proj",
14
+ "gate_proj",
15
+ "down_proj",
16
+ "up_proj"
17
+ ],
18
+ "rank_pattern": {
19
+ "q_proj": 2944,
20
+ "o_proj": 2944,
21
+ "k_proj": 896,
22
+ "v_proj": 896,
23
+ "gate_proj": 2944,
24
+ "down_proj": 2944,
25
+ "up_proj": 2944
26
+ }
27
+ }
28
+
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/osf/llama-3.2-3B-rank128/training_params.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "autocast_adapter_dtype": false,
3
+ "optimizer_kwargs": {
4
+ "lr": 5e-5
5
+ }
6
+ }
7
+
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prefixtuning/llama-3.2-3B-lr_0.001/adapter_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "encoder_hidden_size": 3072,
5
+ "inference_mode": false,
6
+ "num_attention_heads": 24,
7
+ "num_layers": 28,
8
+ "num_transformer_submodules": 1,
9
+ "num_virtual_tokens": 200,
10
+ "peft_type": "PREFIX_TUNING",
11
+ "prefix_projection": false,
12
+ "revision": null,
13
+ "task_type": "CAUSAL_LM",
14
+ "token_dim": 3072
15
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prefixtuning/llama-3.2-3B-lr_0.001/training_params.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "optimizer_kwargs": {
3
+ "lr": 1e-3
4
+ }
5
+ }
6
+
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-default/adapter_config.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "inference_mode": false,
5
+ "num_attention_heads": 24,
6
+ "num_layers": 28,
7
+ "num_transformer_submodules": 1,
8
+ "num_virtual_tokens": 200,
9
+ "peft_type": "PROMPT_TUNING",
10
+ "prompt_tuning_init": "RANDOM",
11
+ "prompt_tuning_init_text": null,
12
+ "revision": null,
13
+ "task_type": "CAUSAL_LM",
14
+ "token_dim": 3072,
15
+ "tokenizer_kwargs": null,
16
+ "tokenizer_name_or_path": null
17
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-lr_0.001/adapter_config.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "inference_mode": false,
5
+ "num_attention_heads": 24,
6
+ "num_layers": 28,
7
+ "num_transformer_submodules": 1,
8
+ "num_virtual_tokens": 200,
9
+ "peft_type": "PROMPT_TUNING",
10
+ "prompt_tuning_init": "RANDOM",
11
+ "prompt_tuning_init_text": null,
12
+ "revision": null,
13
+ "task_type": "CAUSAL_LM",
14
+ "token_dim": 3072,
15
+ "tokenizer_kwargs": null,
16
+ "tokenizer_name_or_path": null
17
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-lr_0.001/training_params.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "optimizer_kwargs": {
3
+ "lr": 1e-3
4
+ }
5
+ }
6
+
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-sample_vocab-lr_0.001/adapter_config.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "inference_mode": false,
5
+ "num_attention_heads": 24,
6
+ "num_layers": 28,
7
+ "num_transformer_submodules": 1,
8
+ "num_virtual_tokens": 200,
9
+ "peft_type": "PROMPT_TUNING",
10
+ "prompt_tuning_init": "SAMPLE_VOCAB",
11
+ "prompt_tuning_init_text": null,
12
+ "revision": null,
13
+ "task_type": "CAUSAL_LM",
14
+ "token_dim": 3072,
15
+ "tokenizer_kwargs": null,
16
+ "tokenizer_name_or_path": null
17
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/prompt_tuning/llama-3.2-3B-sample_vocab-lr_0.001/training_params.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "optimizer_kwargs": {
3
+ "lr": 1e-3
4
+ }
5
+ }
6
+
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/psoft/llama-3.2-3B-default/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "r": 256,
5
+ "psoft_alpha": 256,
6
+ "psoft_dropout": 0.0,
7
+ "exclude_modules": null,
8
+ "fan_in_fan_out": false,
9
+ "ab_svd_init": "psoft_init",
10
+ "psoft_svd": "full",
11
+ "psoft_svd_lowrank_niter": 10,
12
+ "psoft_orth": true,
13
+ "psoft_mag_a": true,
14
+ "psoft_mag_b": true,
15
+ "use_cayley_neumann": false,
16
+ "num_cayley_neumann_terms": 5,
17
+ "cayley_neumann_eps": null,
18
+ "init_weights": true,
19
+ "layers_pattern": null,
20
+ "layers_to_transform": null,
21
+ "modules_to_save": null,
22
+ "peft_type": "PSOFT",
23
+ "revision": null,
24
+ "target_modules": null,
25
+ "task_type": null
26
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/psoft/llama-3.2-3B-fast/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "r": 256,
5
+ "psoft_alpha": 256,
6
+ "psoft_dropout": 0.0,
7
+ "exclude_modules": null,
8
+ "fan_in_fan_out": false,
9
+ "ab_svd_init": "psoft_init",
10
+ "psoft_svd": "lowrank",
11
+ "psoft_svd_lowrank_niter": 10,
12
+ "psoft_orth": true,
13
+ "psoft_mag_a": true,
14
+ "psoft_mag_b": true,
15
+ "use_cayley_neumann": true,
16
+ "num_cayley_neumann_terms": 5,
17
+ "cayley_neumann_eps": null,
18
+ "init_weights": true,
19
+ "layers_pattern": null,
20
+ "layers_to_transform": null,
21
+ "modules_to_save": null,
22
+ "peft_type": "PSOFT",
23
+ "revision": null,
24
+ "target_modules": null,
25
+ "task_type": null
26
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/ptuning/llama-3.2-3B-default/adapter_config.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "encoder_dropout": 0.0,
5
+ "encoder_hidden_size": 3072,
6
+ "encoder_num_layers": 2,
7
+ "encoder_reparameterization_type": "MLP",
8
+ "inference_mode": false,
9
+ "num_attention_heads": 24,
10
+ "num_layers": 28,
11
+ "num_transformer_submodules": 1,
12
+ "num_virtual_tokens": 20,
13
+ "peft_type": "P_TUNING",
14
+ "revision": null,
15
+ "task_type": "CAUSAL_LM",
16
+ "token_dim": 3072
17
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/pvera/llama-3.2-3B-default/adapter_config.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "bias": "none",
5
+ "d_initial": 0.1,
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": false,
8
+ "init_weights": true,
9
+ "layers_pattern": null,
10
+ "layers_to_transform": null,
11
+ "modules_to_save": null,
12
+ "peft_type": "PVERA",
13
+ "projection_prng_key": 0,
14
+ "r": 256,
15
+ "revision": null,
16
+ "save_projection": true,
17
+ "target_modules": null,
18
+ "task_type": null,
19
+ "pvera_dropout": 0.0
20
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/pvera/llama-3.2-3B-default/training_params.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "optimizer_kwargs": {
3
+ "lr": 3e-3
4
+ }
5
+ }
6
+
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/randlora/llama-3.2-3B-default/adapter_config.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": false,
7
+ "init_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "modules_to_save": null,
11
+ "peft_type": "RANDLORA",
12
+ "projection_prng_key": 0,
13
+ "r": 32,
14
+ "randlora_alpha": 640,
15
+ "randlora_dropout": 0.0,
16
+ "revision": null,
17
+ "save_projection": true,
18
+ "sparse": false,
19
+ "target_modules": null,
20
+ "task_type": null,
21
+ "very_sparse": false
22
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/road/llama-3.2-3B-lr_0.001/adapter_config.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "group_size": 64,
5
+ "inference_mode": false,
6
+ "init_weights": true,
7
+ "peft_type": "ROAD",
8
+ "revision": null,
9
+ "target_modules": null,
10
+ "task_type": null,
11
+ "variant": "road_2"
12
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/road/llama-3.2-3B-lr_0.001/training_params.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "optimizer_kwargs": {
3
+ "lr": 1e-3
4
+ }
5
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/shira/llama-3.2-3B-lr_0.0003-random_seed_42/adapter_config.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "fan_in_fan_out": false,
5
+ "inference_mode": false,
6
+ "init_weights": true,
7
+ "mask_type": "random",
8
+ "modules_to_save": null,
9
+ "peft_type": "SHIRA",
10
+ "r": 32,
11
+ "random_seed": 42,
12
+ "revision": null,
13
+ "target_modules": null,
14
+ "task_type": null
15
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/shira/llama-3.2-3B-lr_0.0003-random_seed_42/training_params.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "optimizer_kwargs": {
3
+ "lr": 3e-4
4
+ }
5
+ }
6
+
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/trainable_tokens/llama-3.2-3B-sos+eos/adapter_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "peft_type": "TRAINABLE_TOKENS",
5
+ "token_indices": [128000, 128001],
6
+ "task_type": "CAUSAL_LM"
7
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/trainable_tokens/llama-3.2-3B-sos+eos/training_params.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "optimizer_kwargs": {
3
+ "lr": 0.2
4
+ }
5
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/vblora/llama-3.2-3B-default/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "bias": "none",
5
+ "exclude_modules": null,
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": false,
8
+ "init_logits_std": 0.1,
9
+ "init_vector_bank_bound": 0.02,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "modules_to_save": null,
13
+ "num_vectors": 256,
14
+ "peft_type": "VBLORA",
15
+ "r": 4,
16
+ "revision": null,
17
+ "save_only_topk_weights": false,
18
+ "target_modules": [
19
+ "v_proj",
20
+ "q_proj"
21
+ ],
22
+ "task_type": null,
23
+ "topk": 2,
24
+ "vblora_dropout": 0.0,
25
+ "vector_length": 256
26
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/vera/llama-3.2-3B-default/adapter_config.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "bias": "none",
5
+ "d_initial": 0.1,
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": false,
8
+ "init_weights": true,
9
+ "layers_pattern": null,
10
+ "layers_to_transform": null,
11
+ "modules_to_save": null,
12
+ "peft_type": "VERA",
13
+ "projection_prng_key": 0,
14
+ "r": 256,
15
+ "revision": null,
16
+ "save_projection": true,
17
+ "target_modules": null,
18
+ "task_type": null,
19
+ "vera_dropout": 0.0
20
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/vera/llama-3.2-3B-default/training_params.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "optimizer_kwargs": {
3
+ "lr": 1e-3
4
+ }
5
+ }
6
+
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/experiments/waveft/llama-3.2-3B-n_frequency-5000/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": null,
4
+ "bias": "none",
5
+ "exclude_modules": null,
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": false,
8
+ "init_weights": true,
9
+ "layers_pattern": null,
10
+ "layers_to_transform": null,
11
+ "modules_to_save": null,
12
+ "n_frequency": 5000,
13
+ "n_frequency_pattern": {},
14
+ "peft_type": "WAVEFT",
15
+ "proportional_parameters": false,
16
+ "random_loc_seed": 777,
17
+ "revision": null,
18
+ "scaling": 25.0,
19
+ "target_modules": [
20
+ "q_proj",
21
+ "v_proj"
22
+ ],
23
+ "task_type": "CAUSAL_LM",
24
+ "use_idwt": true,
25
+ "wavelet_family": "db1"
26
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/requirements.txt ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ bitsandbytes
2
+ datasets
3
+ numpy
4
+ tqdm
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/.gitkeep ADDED
File without changes
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/adalora--llama-3.2-3B-rank32.json ADDED
@@ -0,0 +1,4080 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T17:02:52+00:00",
4
+ "total_time": 1339.9807203459786,
5
+ "experiment_name": "adalora/llama-3.2-3B-rank32",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "ADALORA",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "r": 8,
42
+ "target_modules": [
43
+ "q_proj",
44
+ "v_proj"
45
+ ],
46
+ "exclude_modules": null,
47
+ "lora_alpha": 8,
48
+ "lora_dropout": 0.0,
49
+ "fan_in_fan_out": false,
50
+ "bias": "none",
51
+ "use_rslora": false,
52
+ "modules_to_save": null,
53
+ "init_lora_weights": true,
54
+ "layers_to_transform": null,
55
+ "layers_pattern": null,
56
+ "rank_pattern": {
57
+ "model.layers.0.self_attn.q_proj.lora_E": [
58
+ false,
59
+ false,
60
+ false,
61
+ false,
62
+ false,
63
+ false,
64
+ false,
65
+ false,
66
+ false,
67
+ false,
68
+ false,
69
+ false,
70
+ false,
71
+ false,
72
+ false,
73
+ false,
74
+ false,
75
+ false,
76
+ false,
77
+ false,
78
+ false,
79
+ false,
80
+ false,
81
+ false,
82
+ false,
83
+ false,
84
+ false,
85
+ false,
86
+ false,
87
+ false,
88
+ false,
89
+ false,
90
+ false,
91
+ false,
92
+ false,
93
+ false,
94
+ false,
95
+ false,
96
+ false,
97
+ false,
98
+ false,
99
+ false,
100
+ false,
101
+ false,
102
+ false,
103
+ false,
104
+ false,
105
+ false,
106
+ false,
107
+ false,
108
+ false,
109
+ false,
110
+ false,
111
+ false,
112
+ false,
113
+ false,
114
+ false,
115
+ false,
116
+ false,
117
+ false,
118
+ false,
119
+ false,
120
+ false,
121
+ false
122
+ ],
123
+ "model.layers.0.self_attn.v_proj.lora_E": [
124
+ true,
125
+ true,
126
+ true,
127
+ true,
128
+ true,
129
+ true,
130
+ true,
131
+ true,
132
+ true,
133
+ true,
134
+ true,
135
+ true,
136
+ true,
137
+ true,
138
+ true,
139
+ false,
140
+ true,
141
+ true,
142
+ true,
143
+ true,
144
+ true,
145
+ true,
146
+ true,
147
+ true,
148
+ true,
149
+ true,
150
+ true,
151
+ true,
152
+ true,
153
+ true,
154
+ true,
155
+ true,
156
+ true,
157
+ true,
158
+ true,
159
+ true,
160
+ true,
161
+ true,
162
+ true,
163
+ true,
164
+ true,
165
+ true,
166
+ true,
167
+ true,
168
+ true,
169
+ true,
170
+ true,
171
+ true,
172
+ true,
173
+ true,
174
+ false,
175
+ true,
176
+ true,
177
+ true,
178
+ true,
179
+ true,
180
+ true,
181
+ true,
182
+ true,
183
+ true,
184
+ true,
185
+ true,
186
+ true,
187
+ true
188
+ ],
189
+ "model.layers.1.self_attn.q_proj.lora_E": [
190
+ false,
191
+ false,
192
+ false,
193
+ true,
194
+ false,
195
+ true,
196
+ true,
197
+ false,
198
+ false,
199
+ false,
200
+ false,
201
+ true,
202
+ true,
203
+ false,
204
+ true,
205
+ true,
206
+ true,
207
+ true,
208
+ false,
209
+ false,
210
+ false,
211
+ false,
212
+ false,
213
+ false,
214
+ true,
215
+ false,
216
+ true,
217
+ true,
218
+ false,
219
+ false,
220
+ true,
221
+ true,
222
+ true,
223
+ false,
224
+ true,
225
+ false,
226
+ false,
227
+ false,
228
+ true,
229
+ true,
230
+ true,
231
+ false,
232
+ false,
233
+ false,
234
+ true,
235
+ false,
236
+ true,
237
+ true,
238
+ true,
239
+ true,
240
+ false,
241
+ true,
242
+ true,
243
+ true,
244
+ false,
245
+ false,
246
+ true,
247
+ true,
248
+ false,
249
+ false,
250
+ true,
251
+ true,
252
+ false,
253
+ false
254
+ ],
255
+ "model.layers.1.self_attn.v_proj.lora_E": [
256
+ true,
257
+ true,
258
+ true,
259
+ true,
260
+ true,
261
+ true,
262
+ true,
263
+ true,
264
+ true,
265
+ true,
266
+ true,
267
+ true,
268
+ true,
269
+ true,
270
+ true,
271
+ true,
272
+ true,
273
+ true,
274
+ false,
275
+ true,
276
+ true,
277
+ true,
278
+ true,
279
+ true,
280
+ true,
281
+ true,
282
+ true,
283
+ true,
284
+ true,
285
+ true,
286
+ true,
287
+ true,
288
+ true,
289
+ true,
290
+ true,
291
+ true,
292
+ true,
293
+ true,
294
+ true,
295
+ true,
296
+ true,
297
+ true,
298
+ false,
299
+ true,
300
+ true,
301
+ true,
302
+ true,
303
+ true,
304
+ true,
305
+ true,
306
+ true,
307
+ true,
308
+ true,
309
+ true,
310
+ true,
311
+ true,
312
+ true,
313
+ true,
314
+ true,
315
+ true,
316
+ true,
317
+ true,
318
+ true,
319
+ true
320
+ ],
321
+ "model.layers.2.self_attn.q_proj.lora_E": [
322
+ true,
323
+ false,
324
+ true,
325
+ false,
326
+ false,
327
+ false,
328
+ true,
329
+ true,
330
+ true,
331
+ true,
332
+ false,
333
+ true,
334
+ true,
335
+ true,
336
+ false,
337
+ false,
338
+ true,
339
+ false,
340
+ false,
341
+ true,
342
+ false,
343
+ false,
344
+ true,
345
+ false,
346
+ true,
347
+ true,
348
+ false,
349
+ false,
350
+ false,
351
+ false,
352
+ true,
353
+ false,
354
+ false,
355
+ false,
356
+ false,
357
+ false,
358
+ false,
359
+ true,
360
+ true,
361
+ true,
362
+ false,
363
+ false,
364
+ false,
365
+ true,
366
+ true,
367
+ false,
368
+ false,
369
+ false,
370
+ true,
371
+ false,
372
+ true,
373
+ true,
374
+ false,
375
+ true,
376
+ false,
377
+ false,
378
+ false,
379
+ false,
380
+ true,
381
+ false,
382
+ true,
383
+ true,
384
+ false,
385
+ false
386
+ ],
387
+ "model.layers.2.self_attn.v_proj.lora_E": [
388
+ false,
389
+ false,
390
+ false,
391
+ false,
392
+ true,
393
+ true,
394
+ true,
395
+ true,
396
+ false,
397
+ true,
398
+ true,
399
+ true,
400
+ true,
401
+ true,
402
+ false,
403
+ true,
404
+ false,
405
+ true,
406
+ false,
407
+ true,
408
+ false,
409
+ true,
410
+ true,
411
+ true,
412
+ true,
413
+ true,
414
+ false,
415
+ false,
416
+ false,
417
+ false,
418
+ false,
419
+ false,
420
+ true,
421
+ false,
422
+ false,
423
+ false,
424
+ false,
425
+ false,
426
+ false,
427
+ true,
428
+ false,
429
+ false,
430
+ true,
431
+ true,
432
+ false,
433
+ false,
434
+ true,
435
+ true,
436
+ true,
437
+ false,
438
+ false,
439
+ true,
440
+ true,
441
+ true,
442
+ false,
443
+ false,
444
+ false,
445
+ true,
446
+ true,
447
+ false,
448
+ true,
449
+ false,
450
+ true,
451
+ true
452
+ ],
453
+ "model.layers.3.self_attn.q_proj.lora_E": [
454
+ false,
455
+ false,
456
+ false,
457
+ false,
458
+ false,
459
+ false,
460
+ false,
461
+ false,
462
+ false,
463
+ false,
464
+ false,
465
+ false,
466
+ false,
467
+ false,
468
+ false,
469
+ false,
470
+ false,
471
+ false,
472
+ false,
473
+ false,
474
+ false,
475
+ false,
476
+ false,
477
+ false,
478
+ false,
479
+ false,
480
+ false,
481
+ false,
482
+ false,
483
+ false,
484
+ false,
485
+ false,
486
+ false,
487
+ false,
488
+ false,
489
+ false,
490
+ false,
491
+ false,
492
+ false,
493
+ false,
494
+ false,
495
+ false,
496
+ false,
497
+ false,
498
+ false,
499
+ false,
500
+ false,
501
+ false,
502
+ false,
503
+ false,
504
+ false,
505
+ false,
506
+ false,
507
+ false,
508
+ false,
509
+ false,
510
+ false,
511
+ false,
512
+ false,
513
+ false,
514
+ false,
515
+ false,
516
+ false,
517
+ false
518
+ ],
519
+ "model.layers.3.self_attn.v_proj.lora_E": [
520
+ false,
521
+ false,
522
+ false,
523
+ false,
524
+ false,
525
+ true,
526
+ false,
527
+ true,
528
+ false,
529
+ false,
530
+ false,
531
+ false,
532
+ true,
533
+ false,
534
+ false,
535
+ false,
536
+ false,
537
+ true,
538
+ false,
539
+ true,
540
+ true,
541
+ false,
542
+ false,
543
+ false,
544
+ false,
545
+ false,
546
+ false,
547
+ false,
548
+ false,
549
+ false,
550
+ false,
551
+ false,
552
+ true,
553
+ false,
554
+ false,
555
+ true,
556
+ false,
557
+ false,
558
+ false,
559
+ false,
560
+ true,
561
+ false,
562
+ true,
563
+ true,
564
+ false,
565
+ true,
566
+ false,
567
+ false,
568
+ true,
569
+ false,
570
+ false,
571
+ false,
572
+ false,
573
+ false,
574
+ false,
575
+ false,
576
+ true,
577
+ false,
578
+ false,
579
+ false,
580
+ false,
581
+ true,
582
+ true,
583
+ true
584
+ ],
585
+ "model.layers.4.self_attn.q_proj.lora_E": [
586
+ false,
587
+ false,
588
+ false,
589
+ false,
590
+ false,
591
+ true,
592
+ false,
593
+ false,
594
+ true,
595
+ true,
596
+ false,
597
+ true,
598
+ false,
599
+ false,
600
+ false,
601
+ false,
602
+ false,
603
+ false,
604
+ false,
605
+ true,
606
+ true,
607
+ false,
608
+ true,
609
+ false,
610
+ true,
611
+ false,
612
+ false,
613
+ false,
614
+ false,
615
+ false,
616
+ true,
617
+ false,
618
+ false,
619
+ false,
620
+ false,
621
+ false,
622
+ false,
623
+ false,
624
+ true,
625
+ true,
626
+ false,
627
+ false,
628
+ true,
629
+ false,
630
+ false,
631
+ false,
632
+ true,
633
+ true,
634
+ false,
635
+ false,
636
+ false,
637
+ false,
638
+ false,
639
+ false,
640
+ false,
641
+ false,
642
+ false,
643
+ false,
644
+ false,
645
+ false,
646
+ true,
647
+ false,
648
+ true,
649
+ false
650
+ ],
651
+ "model.layers.4.self_attn.v_proj.lora_E": [
652
+ true,
653
+ false,
654
+ true,
655
+ true,
656
+ false,
657
+ false,
658
+ true,
659
+ false,
660
+ false,
661
+ false,
662
+ true,
663
+ true,
664
+ true,
665
+ true,
666
+ false,
667
+ true,
668
+ false,
669
+ true,
670
+ true,
671
+ false,
672
+ true,
673
+ true,
674
+ true,
675
+ false,
676
+ true,
677
+ true,
678
+ true,
679
+ false,
680
+ false,
681
+ false,
682
+ false,
683
+ false,
684
+ false,
685
+ false,
686
+ true,
687
+ false,
688
+ true,
689
+ false,
690
+ false,
691
+ true,
692
+ true,
693
+ true,
694
+ true,
695
+ true,
696
+ false,
697
+ false,
698
+ false,
699
+ false,
700
+ false,
701
+ true,
702
+ false,
703
+ true,
704
+ true,
705
+ true,
706
+ true,
707
+ true,
708
+ false,
709
+ true,
710
+ true,
711
+ false,
712
+ true,
713
+ true,
714
+ true,
715
+ true
716
+ ],
717
+ "model.layers.5.self_attn.q_proj.lora_E": [
718
+ false,
719
+ true,
720
+ false,
721
+ false,
722
+ false,
723
+ false,
724
+ false,
725
+ false,
726
+ false,
727
+ false,
728
+ false,
729
+ false,
730
+ false,
731
+ false,
732
+ false,
733
+ false,
734
+ false,
735
+ false,
736
+ false,
737
+ false,
738
+ false,
739
+ false,
740
+ false,
741
+ false,
742
+ false,
743
+ false,
744
+ false,
745
+ false,
746
+ false,
747
+ false,
748
+ false,
749
+ false,
750
+ false,
751
+ false,
752
+ false,
753
+ false,
754
+ false,
755
+ false,
756
+ false,
757
+ false,
758
+ false,
759
+ false,
760
+ false,
761
+ false,
762
+ false,
763
+ false,
764
+ false,
765
+ false,
766
+ false,
767
+ false,
768
+ false,
769
+ false,
770
+ false,
771
+ false,
772
+ false,
773
+ false,
774
+ false,
775
+ false,
776
+ false,
777
+ false,
778
+ false,
779
+ false,
780
+ false,
781
+ false
782
+ ],
783
+ "model.layers.5.self_attn.v_proj.lora_E": [
784
+ true,
785
+ true,
786
+ true,
787
+ true,
788
+ true,
789
+ true,
790
+ true,
791
+ true,
792
+ false,
793
+ true,
794
+ false,
795
+ false,
796
+ true,
797
+ false,
798
+ false,
799
+ true,
800
+ false,
801
+ true,
802
+ false,
803
+ false,
804
+ false,
805
+ false,
806
+ true,
807
+ true,
808
+ false,
809
+ false,
810
+ false,
811
+ false,
812
+ true,
813
+ false,
814
+ true,
815
+ false,
816
+ true,
817
+ true,
818
+ false,
819
+ false,
820
+ true,
821
+ true,
822
+ true,
823
+ true,
824
+ false,
825
+ false,
826
+ true,
827
+ false,
828
+ true,
829
+ false,
830
+ false,
831
+ true,
832
+ true,
833
+ true,
834
+ false,
835
+ true,
836
+ true,
837
+ false,
838
+ false,
839
+ true,
840
+ true,
841
+ true,
842
+ true,
843
+ false,
844
+ false,
845
+ false,
846
+ true,
847
+ true
848
+ ],
849
+ "model.layers.6.self_attn.q_proj.lora_E": [
850
+ false,
851
+ false,
852
+ true,
853
+ true,
854
+ false,
855
+ false,
856
+ true,
857
+ true,
858
+ false,
859
+ false,
860
+ false,
861
+ true,
862
+ false,
863
+ true,
864
+ false,
865
+ false,
866
+ false,
867
+ false,
868
+ false,
869
+ false,
870
+ false,
871
+ false,
872
+ true,
873
+ true,
874
+ false,
875
+ true,
876
+ false,
877
+ true,
878
+ false,
879
+ true,
880
+ false,
881
+ false,
882
+ false,
883
+ true,
884
+ true,
885
+ false,
886
+ false,
887
+ false,
888
+ false,
889
+ true,
890
+ true,
891
+ false,
892
+ false,
893
+ false,
894
+ true,
895
+ false,
896
+ true,
897
+ false,
898
+ false,
899
+ true,
900
+ true,
901
+ false,
902
+ false,
903
+ false,
904
+ true,
905
+ false,
906
+ false,
907
+ true,
908
+ false,
909
+ false,
910
+ false,
911
+ true,
912
+ false,
913
+ false
914
+ ],
915
+ "model.layers.6.self_attn.v_proj.lora_E": [
916
+ true,
917
+ true,
918
+ true,
919
+ true,
920
+ true,
921
+ true,
922
+ true,
923
+ true,
924
+ true,
925
+ true,
926
+ false,
927
+ true,
928
+ true,
929
+ true,
930
+ true,
931
+ true,
932
+ false,
933
+ true,
934
+ true,
935
+ true,
936
+ false,
937
+ false,
938
+ false,
939
+ false,
940
+ false,
941
+ true,
942
+ false,
943
+ false,
944
+ false,
945
+ false,
946
+ true,
947
+ true,
948
+ false,
949
+ true,
950
+ true,
951
+ true,
952
+ false,
953
+ true,
954
+ true,
955
+ true,
956
+ false,
957
+ true,
958
+ true,
959
+ true,
960
+ true,
961
+ false,
962
+ false,
963
+ false,
964
+ true,
965
+ true,
966
+ false,
967
+ false,
968
+ true,
969
+ true,
970
+ true,
971
+ true,
972
+ true,
973
+ true,
974
+ false,
975
+ true,
976
+ false,
977
+ true,
978
+ false,
979
+ true
980
+ ],
981
+ "model.layers.7.self_attn.q_proj.lora_E": [
982
+ false,
983
+ false,
984
+ false,
985
+ false,
986
+ false,
987
+ true,
988
+ false,
989
+ false,
990
+ false,
991
+ false,
992
+ false,
993
+ false,
994
+ false,
995
+ false,
996
+ true,
997
+ false,
998
+ false,
999
+ false,
1000
+ false,
1001
+ false,
1002
+ false,
1003
+ false,
1004
+ false,
1005
+ false,
1006
+ false,
1007
+ false,
1008
+ false,
1009
+ false,
1010
+ false,
1011
+ false,
1012
+ false,
1013
+ false,
1014
+ false,
1015
+ false,
1016
+ false,
1017
+ false,
1018
+ false,
1019
+ false,
1020
+ false,
1021
+ false,
1022
+ false,
1023
+ false,
1024
+ false,
1025
+ false,
1026
+ false,
1027
+ false,
1028
+ false,
1029
+ false,
1030
+ false,
1031
+ false,
1032
+ true,
1033
+ false,
1034
+ false,
1035
+ false,
1036
+ false,
1037
+ false,
1038
+ false,
1039
+ false,
1040
+ false,
1041
+ false,
1042
+ false,
1043
+ false,
1044
+ false,
1045
+ false
1046
+ ],
1047
+ "model.layers.7.self_attn.v_proj.lora_E": [
1048
+ false,
1049
+ false,
1050
+ true,
1051
+ true,
1052
+ true,
1053
+ true,
1054
+ true,
1055
+ true,
1056
+ true,
1057
+ true,
1058
+ true,
1059
+ true,
1060
+ true,
1061
+ true,
1062
+ true,
1063
+ true,
1064
+ true,
1065
+ true,
1066
+ true,
1067
+ true,
1068
+ false,
1069
+ false,
1070
+ true,
1071
+ true,
1072
+ true,
1073
+ true,
1074
+ true,
1075
+ false,
1076
+ true,
1077
+ false,
1078
+ false,
1079
+ true,
1080
+ true,
1081
+ true,
1082
+ true,
1083
+ false,
1084
+ false,
1085
+ false,
1086
+ true,
1087
+ false,
1088
+ false,
1089
+ true,
1090
+ true,
1091
+ true,
1092
+ false,
1093
+ true,
1094
+ true,
1095
+ true,
1096
+ true,
1097
+ true,
1098
+ true,
1099
+ false,
1100
+ true,
1101
+ true,
1102
+ true,
1103
+ true,
1104
+ true,
1105
+ false,
1106
+ false,
1107
+ false,
1108
+ true,
1109
+ true,
1110
+ true,
1111
+ true
1112
+ ],
1113
+ "model.layers.8.self_attn.q_proj.lora_E": [
1114
+ false,
1115
+ true,
1116
+ false,
1117
+ false,
1118
+ false,
1119
+ false,
1120
+ false,
1121
+ true,
1122
+ false,
1123
+ false,
1124
+ false,
1125
+ false,
1126
+ true,
1127
+ true,
1128
+ false,
1129
+ false,
1130
+ false,
1131
+ false,
1132
+ false,
1133
+ false,
1134
+ false,
1135
+ false,
1136
+ false,
1137
+ false,
1138
+ false,
1139
+ false,
1140
+ false,
1141
+ false,
1142
+ false,
1143
+ false,
1144
+ false,
1145
+ false,
1146
+ false,
1147
+ true,
1148
+ false,
1149
+ false,
1150
+ false,
1151
+ false,
1152
+ true,
1153
+ false,
1154
+ false,
1155
+ true,
1156
+ false,
1157
+ false,
1158
+ false,
1159
+ false,
1160
+ true,
1161
+ false,
1162
+ false,
1163
+ false,
1164
+ false,
1165
+ false,
1166
+ true,
1167
+ false,
1168
+ false,
1169
+ false,
1170
+ false,
1171
+ false,
1172
+ false,
1173
+ false,
1174
+ false,
1175
+ false,
1176
+ false,
1177
+ true
1178
+ ],
1179
+ "model.layers.8.self_attn.v_proj.lora_E": [
1180
+ false,
1181
+ true,
1182
+ false,
1183
+ false,
1184
+ false,
1185
+ true,
1186
+ false,
1187
+ false,
1188
+ false,
1189
+ false,
1190
+ true,
1191
+ true,
1192
+ true,
1193
+ true,
1194
+ true,
1195
+ false,
1196
+ false,
1197
+ true,
1198
+ true,
1199
+ true,
1200
+ true,
1201
+ true,
1202
+ true,
1203
+ true,
1204
+ true,
1205
+ true,
1206
+ false,
1207
+ true,
1208
+ true,
1209
+ false,
1210
+ false,
1211
+ true,
1212
+ false,
1213
+ false,
1214
+ false,
1215
+ true,
1216
+ false,
1217
+ true,
1218
+ false,
1219
+ true,
1220
+ true,
1221
+ false,
1222
+ true,
1223
+ false,
1224
+ true,
1225
+ true,
1226
+ false,
1227
+ false,
1228
+ true,
1229
+ true,
1230
+ false,
1231
+ true,
1232
+ true,
1233
+ true,
1234
+ false,
1235
+ true,
1236
+ true,
1237
+ true,
1238
+ true,
1239
+ false,
1240
+ false,
1241
+ false,
1242
+ false,
1243
+ true
1244
+ ],
1245
+ "model.layers.9.self_attn.q_proj.lora_E": [
1246
+ true,
1247
+ false,
1248
+ true,
1249
+ true,
1250
+ false,
1251
+ false,
1252
+ true,
1253
+ true,
1254
+ false,
1255
+ true,
1256
+ true,
1257
+ false,
1258
+ false,
1259
+ false,
1260
+ true,
1261
+ true,
1262
+ false,
1263
+ true,
1264
+ false,
1265
+ true,
1266
+ false,
1267
+ false,
1268
+ false,
1269
+ true,
1270
+ false,
1271
+ true,
1272
+ false,
1273
+ true,
1274
+ false,
1275
+ true,
1276
+ false,
1277
+ true,
1278
+ true,
1279
+ false,
1280
+ false,
1281
+ true,
1282
+ true,
1283
+ false,
1284
+ false,
1285
+ false,
1286
+ false,
1287
+ true,
1288
+ true,
1289
+ true,
1290
+ false,
1291
+ false,
1292
+ false,
1293
+ false,
1294
+ true,
1295
+ true,
1296
+ false,
1297
+ false,
1298
+ false,
1299
+ false,
1300
+ false,
1301
+ false,
1302
+ true,
1303
+ false,
1304
+ true,
1305
+ false,
1306
+ false,
1307
+ true,
1308
+ false,
1309
+ false
1310
+ ],
1311
+ "model.layers.9.self_attn.v_proj.lora_E": [
1312
+ true,
1313
+ true,
1314
+ false,
1315
+ true,
1316
+ true,
1317
+ true,
1318
+ true,
1319
+ true,
1320
+ true,
1321
+ false,
1322
+ true,
1323
+ true,
1324
+ true,
1325
+ true,
1326
+ true,
1327
+ false,
1328
+ true,
1329
+ true,
1330
+ true,
1331
+ true,
1332
+ true,
1333
+ true,
1334
+ true,
1335
+ true,
1336
+ false,
1337
+ true,
1338
+ true,
1339
+ true,
1340
+ true,
1341
+ true,
1342
+ true,
1343
+ true,
1344
+ false,
1345
+ true,
1346
+ true,
1347
+ false,
1348
+ true,
1349
+ false,
1350
+ true,
1351
+ true,
1352
+ true,
1353
+ true,
1354
+ false,
1355
+ false,
1356
+ true,
1357
+ true,
1358
+ false,
1359
+ false,
1360
+ false,
1361
+ true,
1362
+ true,
1363
+ true,
1364
+ true,
1365
+ true,
1366
+ true,
1367
+ true,
1368
+ true,
1369
+ true,
1370
+ true,
1371
+ true,
1372
+ true,
1373
+ true,
1374
+ false,
1375
+ false
1376
+ ],
1377
+ "model.layers.10.self_attn.q_proj.lora_E": [
1378
+ false,
1379
+ false,
1380
+ false,
1381
+ false,
1382
+ false,
1383
+ false,
1384
+ false,
1385
+ false,
1386
+ false,
1387
+ false,
1388
+ true,
1389
+ false,
1390
+ false,
1391
+ false,
1392
+ false,
1393
+ false,
1394
+ false,
1395
+ false,
1396
+ false,
1397
+ false,
1398
+ false,
1399
+ false,
1400
+ true,
1401
+ false,
1402
+ false,
1403
+ false,
1404
+ false,
1405
+ false,
1406
+ false,
1407
+ false,
1408
+ false,
1409
+ false,
1410
+ false,
1411
+ false,
1412
+ false,
1413
+ false,
1414
+ false,
1415
+ false,
1416
+ false,
1417
+ true,
1418
+ false,
1419
+ false,
1420
+ false,
1421
+ false,
1422
+ false,
1423
+ false,
1424
+ false,
1425
+ false,
1426
+ false,
1427
+ false,
1428
+ true,
1429
+ false,
1430
+ false,
1431
+ false,
1432
+ false,
1433
+ false,
1434
+ false,
1435
+ false,
1436
+ false,
1437
+ false,
1438
+ false,
1439
+ false,
1440
+ false,
1441
+ false
1442
+ ],
1443
+ "model.layers.10.self_attn.v_proj.lora_E": [
1444
+ true,
1445
+ true,
1446
+ false,
1447
+ false,
1448
+ false,
1449
+ true,
1450
+ true,
1451
+ false,
1452
+ false,
1453
+ true,
1454
+ true,
1455
+ true,
1456
+ true,
1457
+ true,
1458
+ true,
1459
+ true,
1460
+ true,
1461
+ true,
1462
+ true,
1463
+ true,
1464
+ true,
1465
+ true,
1466
+ true,
1467
+ true,
1468
+ true,
1469
+ false,
1470
+ false,
1471
+ false,
1472
+ false,
1473
+ true,
1474
+ true,
1475
+ true,
1476
+ true,
1477
+ false,
1478
+ false,
1479
+ false,
1480
+ false,
1481
+ true,
1482
+ false,
1483
+ false,
1484
+ false,
1485
+ true,
1486
+ true,
1487
+ true,
1488
+ false,
1489
+ true,
1490
+ true,
1491
+ false,
1492
+ true,
1493
+ false,
1494
+ true,
1495
+ true,
1496
+ true,
1497
+ false,
1498
+ false,
1499
+ true,
1500
+ true,
1501
+ true,
1502
+ true,
1503
+ true,
1504
+ true,
1505
+ true,
1506
+ false,
1507
+ true
1508
+ ],
1509
+ "model.layers.11.self_attn.q_proj.lora_E": [
1510
+ true,
1511
+ false,
1512
+ false,
1513
+ false,
1514
+ false,
1515
+ true,
1516
+ false,
1517
+ false,
1518
+ false,
1519
+ true,
1520
+ true,
1521
+ false,
1522
+ false,
1523
+ false,
1524
+ false,
1525
+ false,
1526
+ false,
1527
+ false,
1528
+ false,
1529
+ false,
1530
+ false,
1531
+ true,
1532
+ false,
1533
+ false,
1534
+ true,
1535
+ true,
1536
+ false,
1537
+ false,
1538
+ false,
1539
+ false,
1540
+ true,
1541
+ false,
1542
+ false,
1543
+ false,
1544
+ false,
1545
+ true,
1546
+ true,
1547
+ false,
1548
+ false,
1549
+ false,
1550
+ true,
1551
+ false,
1552
+ true,
1553
+ false,
1554
+ true,
1555
+ false,
1556
+ false,
1557
+ false,
1558
+ true,
1559
+ true,
1560
+ false,
1561
+ false,
1562
+ false,
1563
+ false,
1564
+ true,
1565
+ true,
1566
+ false,
1567
+ true,
1568
+ false,
1569
+ true,
1570
+ true,
1571
+ false,
1572
+ false,
1573
+ false
1574
+ ],
1575
+ "model.layers.11.self_attn.v_proj.lora_E": [
1576
+ false,
1577
+ true,
1578
+ true,
1579
+ true,
1580
+ false,
1581
+ true,
1582
+ true,
1583
+ false,
1584
+ true,
1585
+ false,
1586
+ true,
1587
+ true,
1588
+ true,
1589
+ true,
1590
+ true,
1591
+ true,
1592
+ true,
1593
+ false,
1594
+ false,
1595
+ true,
1596
+ false,
1597
+ true,
1598
+ true,
1599
+ true,
1600
+ true,
1601
+ false,
1602
+ false,
1603
+ false,
1604
+ true,
1605
+ true,
1606
+ true,
1607
+ false,
1608
+ true,
1609
+ true,
1610
+ true,
1611
+ true,
1612
+ true,
1613
+ true,
1614
+ true,
1615
+ true,
1616
+ true,
1617
+ false,
1618
+ false,
1619
+ false,
1620
+ false,
1621
+ false,
1622
+ true,
1623
+ false,
1624
+ false,
1625
+ true,
1626
+ true,
1627
+ false,
1628
+ false,
1629
+ true,
1630
+ true,
1631
+ true,
1632
+ true,
1633
+ true,
1634
+ false,
1635
+ true,
1636
+ true,
1637
+ true,
1638
+ false,
1639
+ false
1640
+ ],
1641
+ "model.layers.12.self_attn.q_proj.lora_E": [
1642
+ false,
1643
+ false,
1644
+ false,
1645
+ false,
1646
+ false,
1647
+ false,
1648
+ false,
1649
+ false,
1650
+ false,
1651
+ false,
1652
+ false,
1653
+ false,
1654
+ false,
1655
+ true,
1656
+ false,
1657
+ true,
1658
+ false,
1659
+ false,
1660
+ true,
1661
+ false,
1662
+ false,
1663
+ false,
1664
+ false,
1665
+ false,
1666
+ false,
1667
+ false,
1668
+ false,
1669
+ false,
1670
+ true,
1671
+ false,
1672
+ false,
1673
+ false,
1674
+ false,
1675
+ false,
1676
+ false,
1677
+ false,
1678
+ false,
1679
+ true,
1680
+ true,
1681
+ false,
1682
+ true,
1683
+ false,
1684
+ false,
1685
+ false,
1686
+ false,
1687
+ true,
1688
+ false,
1689
+ true,
1690
+ false,
1691
+ false,
1692
+ true,
1693
+ false,
1694
+ true,
1695
+ false,
1696
+ false,
1697
+ true,
1698
+ false,
1699
+ false,
1700
+ false,
1701
+ false,
1702
+ false,
1703
+ false,
1704
+ false,
1705
+ false
1706
+ ],
1707
+ "model.layers.12.self_attn.v_proj.lora_E": [
1708
+ true,
1709
+ true,
1710
+ true,
1711
+ true,
1712
+ false,
1713
+ true,
1714
+ true,
1715
+ false,
1716
+ true,
1717
+ true,
1718
+ true,
1719
+ true,
1720
+ false,
1721
+ true,
1722
+ true,
1723
+ true,
1724
+ true,
1725
+ false,
1726
+ false,
1727
+ false,
1728
+ true,
1729
+ true,
1730
+ true,
1731
+ false,
1732
+ false,
1733
+ true,
1734
+ true,
1735
+ true,
1736
+ true,
1737
+ true,
1738
+ true,
1739
+ true,
1740
+ true,
1741
+ true,
1742
+ true,
1743
+ false,
1744
+ false,
1745
+ true,
1746
+ false,
1747
+ true,
1748
+ true,
1749
+ true,
1750
+ false,
1751
+ true,
1752
+ false,
1753
+ false,
1754
+ true,
1755
+ false,
1756
+ true,
1757
+ true,
1758
+ true,
1759
+ true,
1760
+ true,
1761
+ true,
1762
+ true,
1763
+ true,
1764
+ false,
1765
+ true,
1766
+ false,
1767
+ true,
1768
+ false,
1769
+ true,
1770
+ true,
1771
+ false
1772
+ ],
1773
+ "model.layers.13.self_attn.q_proj.lora_E": [
1774
+ true,
1775
+ true,
1776
+ true,
1777
+ true,
1778
+ true,
1779
+ true,
1780
+ false,
1781
+ false,
1782
+ true,
1783
+ true,
1784
+ false,
1785
+ true,
1786
+ false,
1787
+ true,
1788
+ false,
1789
+ true,
1790
+ false,
1791
+ false,
1792
+ true,
1793
+ false,
1794
+ true,
1795
+ true,
1796
+ true,
1797
+ true,
1798
+ true,
1799
+ true,
1800
+ true,
1801
+ false,
1802
+ false,
1803
+ false,
1804
+ true,
1805
+ false,
1806
+ false,
1807
+ true,
1808
+ false,
1809
+ true,
1810
+ false,
1811
+ true,
1812
+ true,
1813
+ true,
1814
+ false,
1815
+ false,
1816
+ false,
1817
+ false,
1818
+ true,
1819
+ true,
1820
+ true,
1821
+ true,
1822
+ false,
1823
+ true,
1824
+ false,
1825
+ true,
1826
+ true,
1827
+ true,
1828
+ false,
1829
+ true,
1830
+ false,
1831
+ true,
1832
+ true,
1833
+ false,
1834
+ false,
1835
+ false,
1836
+ true,
1837
+ false
1838
+ ],
1839
+ "model.layers.13.self_attn.v_proj.lora_E": [
1840
+ true,
1841
+ false,
1842
+ true,
1843
+ true,
1844
+ true,
1845
+ false,
1846
+ true,
1847
+ true,
1848
+ true,
1849
+ false,
1850
+ true,
1851
+ true,
1852
+ true,
1853
+ true,
1854
+ true,
1855
+ false,
1856
+ true,
1857
+ false,
1858
+ true,
1859
+ true,
1860
+ true,
1861
+ true,
1862
+ true,
1863
+ true,
1864
+ true,
1865
+ true,
1866
+ true,
1867
+ false,
1868
+ true,
1869
+ true,
1870
+ false,
1871
+ false,
1872
+ true,
1873
+ true,
1874
+ false,
1875
+ false,
1876
+ true,
1877
+ false,
1878
+ false,
1879
+ true,
1880
+ false,
1881
+ false,
1882
+ true,
1883
+ true,
1884
+ true,
1885
+ true,
1886
+ true,
1887
+ true,
1888
+ true,
1889
+ false,
1890
+ true,
1891
+ false,
1892
+ false,
1893
+ true,
1894
+ true,
1895
+ true,
1896
+ false,
1897
+ true,
1898
+ true,
1899
+ false,
1900
+ true,
1901
+ false,
1902
+ true,
1903
+ true
1904
+ ],
1905
+ "model.layers.14.self_attn.q_proj.lora_E": [
1906
+ false,
1907
+ true,
1908
+ false,
1909
+ true,
1910
+ true,
1911
+ false,
1912
+ false,
1913
+ false,
1914
+ true,
1915
+ false,
1916
+ false,
1917
+ true,
1918
+ false,
1919
+ false,
1920
+ true,
1921
+ true,
1922
+ true,
1923
+ true,
1924
+ false,
1925
+ true,
1926
+ false,
1927
+ false,
1928
+ false,
1929
+ true,
1930
+ false,
1931
+ false,
1932
+ false,
1933
+ true,
1934
+ false,
1935
+ false,
1936
+ false,
1937
+ true,
1938
+ true,
1939
+ true,
1940
+ true,
1941
+ false,
1942
+ false,
1943
+ true,
1944
+ true,
1945
+ false,
1946
+ true,
1947
+ true,
1948
+ false,
1949
+ false,
1950
+ false,
1951
+ false,
1952
+ false,
1953
+ false,
1954
+ true,
1955
+ false,
1956
+ true,
1957
+ true,
1958
+ true,
1959
+ false,
1960
+ true,
1961
+ true,
1962
+ true,
1963
+ false,
1964
+ false,
1965
+ true,
1966
+ false,
1967
+ true,
1968
+ true,
1969
+ true
1970
+ ],
1971
+ "model.layers.14.self_attn.v_proj.lora_E": [
1972
+ true,
1973
+ true,
1974
+ true,
1975
+ false,
1976
+ false,
1977
+ false,
1978
+ true,
1979
+ false,
1980
+ false,
1981
+ false,
1982
+ false,
1983
+ true,
1984
+ true,
1985
+ false,
1986
+ false,
1987
+ true,
1988
+ false,
1989
+ false,
1990
+ true,
1991
+ true,
1992
+ false,
1993
+ true,
1994
+ false,
1995
+ false,
1996
+ true,
1997
+ false,
1998
+ true,
1999
+ false,
2000
+ true,
2001
+ true,
2002
+ false,
2003
+ true,
2004
+ false,
2005
+ true,
2006
+ false,
2007
+ false,
2008
+ true,
2009
+ false,
2010
+ false,
2011
+ true,
2012
+ false,
2013
+ true,
2014
+ true,
2015
+ false,
2016
+ true,
2017
+ true,
2018
+ true,
2019
+ false,
2020
+ true,
2021
+ true,
2022
+ true,
2023
+ true,
2024
+ true,
2025
+ true,
2026
+ false,
2027
+ false,
2028
+ true,
2029
+ true,
2030
+ true,
2031
+ true,
2032
+ true,
2033
+ true,
2034
+ true,
2035
+ false
2036
+ ],
2037
+ "model.layers.15.self_attn.q_proj.lora_E": [
2038
+ true,
2039
+ true,
2040
+ true,
2041
+ true,
2042
+ true,
2043
+ true,
2044
+ false,
2045
+ true,
2046
+ false,
2047
+ true,
2048
+ false,
2049
+ true,
2050
+ false,
2051
+ true,
2052
+ true,
2053
+ true,
2054
+ true,
2055
+ true,
2056
+ true,
2057
+ false,
2058
+ true,
2059
+ true,
2060
+ false,
2061
+ true,
2062
+ false,
2063
+ true,
2064
+ false,
2065
+ true,
2066
+ true,
2067
+ true,
2068
+ false,
2069
+ true,
2070
+ false,
2071
+ false,
2072
+ true,
2073
+ true,
2074
+ true,
2075
+ true,
2076
+ true,
2077
+ false,
2078
+ true,
2079
+ true,
2080
+ false,
2081
+ true,
2082
+ false,
2083
+ true,
2084
+ false,
2085
+ false,
2086
+ true,
2087
+ true,
2088
+ false,
2089
+ true,
2090
+ false,
2091
+ true,
2092
+ false,
2093
+ true,
2094
+ true,
2095
+ true,
2096
+ true,
2097
+ true,
2098
+ false,
2099
+ true,
2100
+ true,
2101
+ true
2102
+ ],
2103
+ "model.layers.15.self_attn.v_proj.lora_E": [
2104
+ true,
2105
+ true,
2106
+ true,
2107
+ true,
2108
+ false,
2109
+ true,
2110
+ true,
2111
+ true,
2112
+ true,
2113
+ false,
2114
+ true,
2115
+ true,
2116
+ false,
2117
+ true,
2118
+ true,
2119
+ false,
2120
+ false,
2121
+ true,
2122
+ true,
2123
+ true,
2124
+ true,
2125
+ true,
2126
+ true,
2127
+ false,
2128
+ false,
2129
+ true,
2130
+ true,
2131
+ true,
2132
+ true,
2133
+ false,
2134
+ false,
2135
+ false,
2136
+ false,
2137
+ true,
2138
+ true,
2139
+ true,
2140
+ true,
2141
+ true,
2142
+ true,
2143
+ true,
2144
+ true,
2145
+ true,
2146
+ true,
2147
+ false,
2148
+ true,
2149
+ true,
2150
+ true,
2151
+ true,
2152
+ true,
2153
+ false,
2154
+ true,
2155
+ true,
2156
+ true,
2157
+ true,
2158
+ false,
2159
+ false,
2160
+ false,
2161
+ true,
2162
+ true,
2163
+ true,
2164
+ true,
2165
+ true,
2166
+ true,
2167
+ true
2168
+ ],
2169
+ "model.layers.16.self_attn.q_proj.lora_E": [
2170
+ false,
2171
+ false,
2172
+ false,
2173
+ false,
2174
+ false,
2175
+ false,
2176
+ false,
2177
+ true,
2178
+ true,
2179
+ false,
2180
+ true,
2181
+ false,
2182
+ true,
2183
+ false,
2184
+ true,
2185
+ true,
2186
+ true,
2187
+ false,
2188
+ true,
2189
+ false,
2190
+ false,
2191
+ true,
2192
+ true,
2193
+ false,
2194
+ false,
2195
+ false,
2196
+ true,
2197
+ true,
2198
+ false,
2199
+ false,
2200
+ true,
2201
+ false,
2202
+ true,
2203
+ true,
2204
+ true,
2205
+ true,
2206
+ false,
2207
+ true,
2208
+ false,
2209
+ false,
2210
+ true,
2211
+ true,
2212
+ false,
2213
+ true,
2214
+ true,
2215
+ true,
2216
+ true,
2217
+ false,
2218
+ true,
2219
+ true,
2220
+ false,
2221
+ true,
2222
+ true,
2223
+ false,
2224
+ true,
2225
+ false,
2226
+ false,
2227
+ true,
2228
+ true,
2229
+ true,
2230
+ false,
2231
+ false,
2232
+ true,
2233
+ true
2234
+ ],
2235
+ "model.layers.16.self_attn.v_proj.lora_E": [
2236
+ true,
2237
+ false,
2238
+ true,
2239
+ false,
2240
+ true,
2241
+ false,
2242
+ true,
2243
+ false,
2244
+ false,
2245
+ true,
2246
+ false,
2247
+ true,
2248
+ false,
2249
+ true,
2250
+ true,
2251
+ true,
2252
+ true,
2253
+ false,
2254
+ true,
2255
+ false,
2256
+ true,
2257
+ true,
2258
+ true,
2259
+ false,
2260
+ true,
2261
+ true,
2262
+ false,
2263
+ false,
2264
+ false,
2265
+ true,
2266
+ true,
2267
+ true,
2268
+ true,
2269
+ true,
2270
+ false,
2271
+ false,
2272
+ true,
2273
+ false,
2274
+ false,
2275
+ true,
2276
+ true,
2277
+ false,
2278
+ true,
2279
+ false,
2280
+ true,
2281
+ true,
2282
+ true,
2283
+ false,
2284
+ true,
2285
+ true,
2286
+ false,
2287
+ false,
2288
+ true,
2289
+ true,
2290
+ true,
2291
+ false,
2292
+ true,
2293
+ true,
2294
+ false,
2295
+ true,
2296
+ false,
2297
+ true,
2298
+ false,
2299
+ false
2300
+ ],
2301
+ "model.layers.17.self_attn.q_proj.lora_E": [
2302
+ true,
2303
+ true,
2304
+ true,
2305
+ true,
2306
+ false,
2307
+ true,
2308
+ false,
2309
+ true,
2310
+ false,
2311
+ true,
2312
+ true,
2313
+ true,
2314
+ true,
2315
+ false,
2316
+ true,
2317
+ false,
2318
+ true,
2319
+ true,
2320
+ true,
2321
+ true,
2322
+ true,
2323
+ true,
2324
+ true,
2325
+ true,
2326
+ true,
2327
+ true,
2328
+ true,
2329
+ true,
2330
+ true,
2331
+ true,
2332
+ true,
2333
+ true,
2334
+ false,
2335
+ false,
2336
+ true,
2337
+ true,
2338
+ true,
2339
+ true,
2340
+ true,
2341
+ true,
2342
+ true,
2343
+ false,
2344
+ true,
2345
+ true,
2346
+ false,
2347
+ true,
2348
+ true,
2349
+ true,
2350
+ false,
2351
+ true,
2352
+ false,
2353
+ true,
2354
+ true,
2355
+ true,
2356
+ true,
2357
+ false,
2358
+ true,
2359
+ true,
2360
+ true,
2361
+ true,
2362
+ true,
2363
+ true,
2364
+ true,
2365
+ true
2366
+ ],
2367
+ "model.layers.17.self_attn.v_proj.lora_E": [
2368
+ false,
2369
+ true,
2370
+ true,
2371
+ true,
2372
+ true,
2373
+ true,
2374
+ true,
2375
+ true,
2376
+ true,
2377
+ true,
2378
+ false,
2379
+ true,
2380
+ true,
2381
+ false,
2382
+ false,
2383
+ false,
2384
+ false,
2385
+ true,
2386
+ false,
2387
+ true,
2388
+ true,
2389
+ false,
2390
+ false,
2391
+ true,
2392
+ true,
2393
+ false,
2394
+ false,
2395
+ false,
2396
+ false,
2397
+ true,
2398
+ true,
2399
+ true,
2400
+ true,
2401
+ true,
2402
+ true,
2403
+ true,
2404
+ false,
2405
+ true,
2406
+ true,
2407
+ true,
2408
+ true,
2409
+ false,
2410
+ false,
2411
+ true,
2412
+ true,
2413
+ true,
2414
+ true,
2415
+ true,
2416
+ false,
2417
+ true,
2418
+ true,
2419
+ true,
2420
+ true,
2421
+ true,
2422
+ true,
2423
+ true,
2424
+ true,
2425
+ true,
2426
+ false,
2427
+ false,
2428
+ true,
2429
+ true,
2430
+ true,
2431
+ true
2432
+ ],
2433
+ "model.layers.18.self_attn.q_proj.lora_E": [
2434
+ false,
2435
+ true,
2436
+ false,
2437
+ true,
2438
+ false,
2439
+ true,
2440
+ false,
2441
+ true,
2442
+ true,
2443
+ true,
2444
+ true,
2445
+ true,
2446
+ true,
2447
+ true,
2448
+ false,
2449
+ true,
2450
+ true,
2451
+ true,
2452
+ true,
2453
+ true,
2454
+ false,
2455
+ false,
2456
+ true,
2457
+ true,
2458
+ true,
2459
+ true,
2460
+ true,
2461
+ true,
2462
+ false,
2463
+ true,
2464
+ true,
2465
+ true,
2466
+ true,
2467
+ true,
2468
+ true,
2469
+ false,
2470
+ true,
2471
+ true,
2472
+ false,
2473
+ true,
2474
+ false,
2475
+ true,
2476
+ false,
2477
+ false,
2478
+ true,
2479
+ false,
2480
+ true,
2481
+ true,
2482
+ false,
2483
+ false,
2484
+ true,
2485
+ false,
2486
+ false,
2487
+ true,
2488
+ false,
2489
+ true,
2490
+ false,
2491
+ false,
2492
+ true,
2493
+ false,
2494
+ true,
2495
+ true,
2496
+ true,
2497
+ true
2498
+ ],
2499
+ "model.layers.18.self_attn.v_proj.lora_E": [
2500
+ true,
2501
+ true,
2502
+ true,
2503
+ false,
2504
+ true,
2505
+ true,
2506
+ true,
2507
+ true,
2508
+ false,
2509
+ true,
2510
+ true,
2511
+ false,
2512
+ false,
2513
+ false,
2514
+ true,
2515
+ false,
2516
+ true,
2517
+ true,
2518
+ true,
2519
+ true,
2520
+ true,
2521
+ true,
2522
+ true,
2523
+ true,
2524
+ true,
2525
+ true,
2526
+ true,
2527
+ true,
2528
+ true,
2529
+ false,
2530
+ true,
2531
+ true,
2532
+ true,
2533
+ false,
2534
+ true,
2535
+ true,
2536
+ true,
2537
+ true,
2538
+ true,
2539
+ true,
2540
+ true,
2541
+ true,
2542
+ false,
2543
+ true,
2544
+ true,
2545
+ false,
2546
+ true,
2547
+ true,
2548
+ true,
2549
+ true,
2550
+ true,
2551
+ true,
2552
+ false,
2553
+ true,
2554
+ false,
2555
+ true,
2556
+ true,
2557
+ true,
2558
+ true,
2559
+ true,
2560
+ true,
2561
+ false,
2562
+ false,
2563
+ true
2564
+ ],
2565
+ "model.layers.19.self_attn.q_proj.lora_E": [
2566
+ false,
2567
+ true,
2568
+ false,
2569
+ true,
2570
+ false,
2571
+ true,
2572
+ false,
2573
+ true,
2574
+ true,
2575
+ true,
2576
+ true,
2577
+ true,
2578
+ true,
2579
+ true,
2580
+ true,
2581
+ true,
2582
+ true,
2583
+ true,
2584
+ true,
2585
+ true,
2586
+ false,
2587
+ true,
2588
+ true,
2589
+ true,
2590
+ true,
2591
+ true,
2592
+ true,
2593
+ false,
2594
+ false,
2595
+ true,
2596
+ false,
2597
+ true,
2598
+ true,
2599
+ false,
2600
+ true,
2601
+ false,
2602
+ true,
2603
+ false,
2604
+ true,
2605
+ true,
2606
+ false,
2607
+ false,
2608
+ false,
2609
+ true,
2610
+ true,
2611
+ true,
2612
+ true,
2613
+ false,
2614
+ false,
2615
+ false,
2616
+ false,
2617
+ true,
2618
+ true,
2619
+ true,
2620
+ true,
2621
+ true,
2622
+ true,
2623
+ false,
2624
+ false,
2625
+ true,
2626
+ true,
2627
+ false,
2628
+ true,
2629
+ true
2630
+ ],
2631
+ "model.layers.19.self_attn.v_proj.lora_E": [
2632
+ false,
2633
+ true,
2634
+ false,
2635
+ false,
2636
+ true,
2637
+ true,
2638
+ false,
2639
+ true,
2640
+ false,
2641
+ false,
2642
+ false,
2643
+ false,
2644
+ false,
2645
+ true,
2646
+ true,
2647
+ false,
2648
+ true,
2649
+ false,
2650
+ false,
2651
+ false,
2652
+ false,
2653
+ false,
2654
+ false,
2655
+ true,
2656
+ true,
2657
+ true,
2658
+ true,
2659
+ true,
2660
+ false,
2661
+ true,
2662
+ true,
2663
+ true,
2664
+ true,
2665
+ true,
2666
+ false,
2667
+ false,
2668
+ false,
2669
+ false,
2670
+ false,
2671
+ true,
2672
+ false,
2673
+ false,
2674
+ false,
2675
+ false,
2676
+ false,
2677
+ false,
2678
+ true,
2679
+ true,
2680
+ false,
2681
+ false,
2682
+ false,
2683
+ true,
2684
+ true,
2685
+ true,
2686
+ true,
2687
+ false,
2688
+ true,
2689
+ true,
2690
+ false,
2691
+ true,
2692
+ false,
2693
+ true,
2694
+ true,
2695
+ true
2696
+ ],
2697
+ "model.layers.20.self_attn.q_proj.lora_E": [
2698
+ false,
2699
+ true,
2700
+ false,
2701
+ false,
2702
+ false,
2703
+ false,
2704
+ true,
2705
+ false,
2706
+ false,
2707
+ false,
2708
+ false,
2709
+ false,
2710
+ false,
2711
+ false,
2712
+ false,
2713
+ true,
2714
+ false,
2715
+ true,
2716
+ false,
2717
+ false,
2718
+ false,
2719
+ false,
2720
+ false,
2721
+ false,
2722
+ false,
2723
+ false,
2724
+ false,
2725
+ true,
2726
+ false,
2727
+ false,
2728
+ false,
2729
+ false,
2730
+ false,
2731
+ true,
2732
+ false,
2733
+ true,
2734
+ true,
2735
+ false,
2736
+ false,
2737
+ false,
2738
+ false,
2739
+ true,
2740
+ false,
2741
+ true,
2742
+ false,
2743
+ true,
2744
+ false,
2745
+ false,
2746
+ false,
2747
+ false,
2748
+ false,
2749
+ true,
2750
+ false,
2751
+ false,
2752
+ true,
2753
+ true,
2754
+ false,
2755
+ false,
2756
+ false,
2757
+ false,
2758
+ false,
2759
+ true,
2760
+ false,
2761
+ false
2762
+ ],
2763
+ "model.layers.20.self_attn.v_proj.lora_E": [
2764
+ true,
2765
+ false,
2766
+ true,
2767
+ true,
2768
+ false,
2769
+ false,
2770
+ false,
2771
+ true,
2772
+ true,
2773
+ false,
2774
+ false,
2775
+ true,
2776
+ true,
2777
+ true,
2778
+ false,
2779
+ true,
2780
+ false,
2781
+ true,
2782
+ false,
2783
+ false,
2784
+ false,
2785
+ false,
2786
+ true,
2787
+ false,
2788
+ false,
2789
+ false,
2790
+ true,
2791
+ false,
2792
+ true,
2793
+ false,
2794
+ true,
2795
+ true,
2796
+ true,
2797
+ false,
2798
+ false,
2799
+ false,
2800
+ true,
2801
+ false,
2802
+ true,
2803
+ false,
2804
+ true,
2805
+ true,
2806
+ true,
2807
+ true,
2808
+ false,
2809
+ false,
2810
+ false,
2811
+ false,
2812
+ false,
2813
+ false,
2814
+ false,
2815
+ false,
2816
+ true,
2817
+ false,
2818
+ true,
2819
+ false,
2820
+ true,
2821
+ false,
2822
+ false,
2823
+ true,
2824
+ true,
2825
+ false,
2826
+ true,
2827
+ true
2828
+ ],
2829
+ "model.layers.21.self_attn.q_proj.lora_E": [
2830
+ false,
2831
+ true,
2832
+ true,
2833
+ true,
2834
+ true,
2835
+ true,
2836
+ false,
2837
+ true,
2838
+ true,
2839
+ false,
2840
+ false,
2841
+ true,
2842
+ true,
2843
+ true,
2844
+ true,
2845
+ true,
2846
+ false,
2847
+ false,
2848
+ false,
2849
+ false,
2850
+ true,
2851
+ true,
2852
+ false,
2853
+ true,
2854
+ true,
2855
+ true,
2856
+ true,
2857
+ false,
2858
+ false,
2859
+ false,
2860
+ false,
2861
+ false,
2862
+ false,
2863
+ false,
2864
+ true,
2865
+ true,
2866
+ true,
2867
+ false,
2868
+ true,
2869
+ false,
2870
+ true,
2871
+ false,
2872
+ true,
2873
+ false,
2874
+ false,
2875
+ false,
2876
+ true,
2877
+ false,
2878
+ true,
2879
+ true,
2880
+ true,
2881
+ true,
2882
+ true,
2883
+ false,
2884
+ false,
2885
+ true,
2886
+ true,
2887
+ false,
2888
+ true,
2889
+ true,
2890
+ true,
2891
+ false,
2892
+ true,
2893
+ true
2894
+ ],
2895
+ "model.layers.21.self_attn.v_proj.lora_E": [
2896
+ true,
2897
+ true,
2898
+ true,
2899
+ true,
2900
+ true,
2901
+ false,
2902
+ true,
2903
+ true,
2904
+ false,
2905
+ true,
2906
+ true,
2907
+ false,
2908
+ false,
2909
+ true,
2910
+ true,
2911
+ false,
2912
+ true,
2913
+ true,
2914
+ false,
2915
+ true,
2916
+ true,
2917
+ true,
2918
+ true,
2919
+ false,
2920
+ false,
2921
+ false,
2922
+ true,
2923
+ false,
2924
+ true,
2925
+ true,
2926
+ true,
2927
+ false,
2928
+ true,
2929
+ true,
2930
+ false,
2931
+ true,
2932
+ true,
2933
+ false,
2934
+ true,
2935
+ true,
2936
+ false,
2937
+ false,
2938
+ false,
2939
+ true,
2940
+ true,
2941
+ false,
2942
+ true,
2943
+ false,
2944
+ true,
2945
+ true,
2946
+ false,
2947
+ true,
2948
+ true,
2949
+ false,
2950
+ false,
2951
+ false,
2952
+ false,
2953
+ true,
2954
+ true,
2955
+ false,
2956
+ false,
2957
+ false,
2958
+ false,
2959
+ false
2960
+ ],
2961
+ "model.layers.22.self_attn.q_proj.lora_E": [
2962
+ false,
2963
+ true,
2964
+ false,
2965
+ true,
2966
+ true,
2967
+ false,
2968
+ true,
2969
+ false,
2970
+ false,
2971
+ true,
2972
+ false,
2973
+ false,
2974
+ false,
2975
+ false,
2976
+ false,
2977
+ false,
2978
+ false,
2979
+ false,
2980
+ true,
2981
+ true,
2982
+ false,
2983
+ false,
2984
+ false,
2985
+ false,
2986
+ true,
2987
+ true,
2988
+ true,
2989
+ false,
2990
+ false,
2991
+ true,
2992
+ false,
2993
+ false,
2994
+ false,
2995
+ false,
2996
+ true,
2997
+ true,
2998
+ false,
2999
+ false,
3000
+ false,
3001
+ false,
3002
+ true,
3003
+ false,
3004
+ false,
3005
+ false,
3006
+ true,
3007
+ false,
3008
+ true,
3009
+ false,
3010
+ false,
3011
+ false,
3012
+ true,
3013
+ false,
3014
+ true,
3015
+ true,
3016
+ true,
3017
+ false,
3018
+ false,
3019
+ true,
3020
+ false,
3021
+ false,
3022
+ true,
3023
+ true,
3024
+ false,
3025
+ true
3026
+ ],
3027
+ "model.layers.22.self_attn.v_proj.lora_E": [
3028
+ false,
3029
+ true,
3030
+ true,
3031
+ true,
3032
+ true,
3033
+ false,
3034
+ false,
3035
+ true,
3036
+ true,
3037
+ false,
3038
+ true,
3039
+ false,
3040
+ true,
3041
+ true,
3042
+ false,
3043
+ true,
3044
+ true,
3045
+ true,
3046
+ true,
3047
+ true,
3048
+ true,
3049
+ true,
3050
+ true,
3051
+ true,
3052
+ false,
3053
+ false,
3054
+ true,
3055
+ true,
3056
+ false,
3057
+ true,
3058
+ true,
3059
+ false,
3060
+ true,
3061
+ false,
3062
+ false,
3063
+ false,
3064
+ false,
3065
+ true,
3066
+ false,
3067
+ false,
3068
+ false,
3069
+ true,
3070
+ true,
3071
+ false,
3072
+ false,
3073
+ false,
3074
+ true,
3075
+ true,
3076
+ false,
3077
+ false,
3078
+ true,
3079
+ true,
3080
+ true,
3081
+ true,
3082
+ true,
3083
+ true,
3084
+ true,
3085
+ true,
3086
+ false,
3087
+ false,
3088
+ true,
3089
+ false,
3090
+ true,
3091
+ true
3092
+ ],
3093
+ "model.layers.23.self_attn.q_proj.lora_E": [
3094
+ true,
3095
+ false,
3096
+ true,
3097
+ false,
3098
+ true,
3099
+ true,
3100
+ true,
3101
+ true,
3102
+ true,
3103
+ true,
3104
+ true,
3105
+ true,
3106
+ true,
3107
+ true,
3108
+ true,
3109
+ true,
3110
+ true,
3111
+ true,
3112
+ false,
3113
+ false,
3114
+ true,
3115
+ true,
3116
+ true,
3117
+ true,
3118
+ true,
3119
+ false,
3120
+ true,
3121
+ true,
3122
+ true,
3123
+ true,
3124
+ true,
3125
+ true,
3126
+ true,
3127
+ true,
3128
+ true,
3129
+ true,
3130
+ false,
3131
+ true,
3132
+ true,
3133
+ true,
3134
+ false,
3135
+ true,
3136
+ true,
3137
+ false,
3138
+ true,
3139
+ false,
3140
+ true,
3141
+ false,
3142
+ true,
3143
+ true,
3144
+ true,
3145
+ true,
3146
+ true,
3147
+ true,
3148
+ true,
3149
+ false,
3150
+ true,
3151
+ true,
3152
+ true,
3153
+ true,
3154
+ false,
3155
+ false,
3156
+ true,
3157
+ true
3158
+ ],
3159
+ "model.layers.23.self_attn.v_proj.lora_E": [
3160
+ false,
3161
+ true,
3162
+ false,
3163
+ true,
3164
+ false,
3165
+ false,
3166
+ true,
3167
+ false,
3168
+ true,
3169
+ true,
3170
+ true,
3171
+ true,
3172
+ true,
3173
+ true,
3174
+ true,
3175
+ true,
3176
+ false,
3177
+ true,
3178
+ true,
3179
+ true,
3180
+ true,
3181
+ true,
3182
+ true,
3183
+ true,
3184
+ true,
3185
+ true,
3186
+ false,
3187
+ true,
3188
+ false,
3189
+ false,
3190
+ false,
3191
+ true,
3192
+ false,
3193
+ true,
3194
+ true,
3195
+ false,
3196
+ true,
3197
+ false,
3198
+ true,
3199
+ true,
3200
+ true,
3201
+ true,
3202
+ true,
3203
+ true,
3204
+ false,
3205
+ true,
3206
+ true,
3207
+ true,
3208
+ true,
3209
+ true,
3210
+ false,
3211
+ true,
3212
+ true,
3213
+ false,
3214
+ false,
3215
+ true,
3216
+ false,
3217
+ true,
3218
+ true,
3219
+ false,
3220
+ true,
3221
+ false,
3222
+ false,
3223
+ true
3224
+ ],
3225
+ "model.layers.24.self_attn.q_proj.lora_E": [
3226
+ true,
3227
+ true,
3228
+ true,
3229
+ true,
3230
+ true,
3231
+ true,
3232
+ false,
3233
+ true,
3234
+ true,
3235
+ true,
3236
+ true,
3237
+ false,
3238
+ true,
3239
+ true,
3240
+ true,
3241
+ false,
3242
+ true,
3243
+ true,
3244
+ true,
3245
+ true,
3246
+ true,
3247
+ true,
3248
+ false,
3249
+ false,
3250
+ true,
3251
+ true,
3252
+ true,
3253
+ true,
3254
+ false,
3255
+ true,
3256
+ true,
3257
+ true,
3258
+ true,
3259
+ true,
3260
+ true,
3261
+ true,
3262
+ true,
3263
+ true,
3264
+ false,
3265
+ true,
3266
+ true,
3267
+ true,
3268
+ false,
3269
+ false,
3270
+ true,
3271
+ true,
3272
+ true,
3273
+ false,
3274
+ true,
3275
+ false,
3276
+ true,
3277
+ true,
3278
+ true,
3279
+ true,
3280
+ false,
3281
+ false,
3282
+ false,
3283
+ true,
3284
+ true,
3285
+ true,
3286
+ true,
3287
+ false,
3288
+ false,
3289
+ true
3290
+ ],
3291
+ "model.layers.24.self_attn.v_proj.lora_E": [
3292
+ true,
3293
+ true,
3294
+ true,
3295
+ false,
3296
+ true,
3297
+ false,
3298
+ false,
3299
+ true,
3300
+ true,
3301
+ true,
3302
+ false,
3303
+ true,
3304
+ true,
3305
+ false,
3306
+ false,
3307
+ true,
3308
+ false,
3309
+ false,
3310
+ false,
3311
+ true,
3312
+ true,
3313
+ true,
3314
+ true,
3315
+ false,
3316
+ false,
3317
+ true,
3318
+ false,
3319
+ true,
3320
+ false,
3321
+ true,
3322
+ false,
3323
+ true,
3324
+ true,
3325
+ false,
3326
+ true,
3327
+ true,
3328
+ false,
3329
+ false,
3330
+ false,
3331
+ true,
3332
+ false,
3333
+ false,
3334
+ true,
3335
+ true,
3336
+ true,
3337
+ true,
3338
+ true,
3339
+ false,
3340
+ false,
3341
+ true,
3342
+ true,
3343
+ true,
3344
+ true,
3345
+ false,
3346
+ false,
3347
+ true,
3348
+ true,
3349
+ true,
3350
+ false,
3351
+ true,
3352
+ false,
3353
+ true,
3354
+ true,
3355
+ true
3356
+ ],
3357
+ "model.layers.25.self_attn.q_proj.lora_E": [
3358
+ false,
3359
+ false,
3360
+ false,
3361
+ false,
3362
+ true,
3363
+ true,
3364
+ false,
3365
+ true,
3366
+ true,
3367
+ false,
3368
+ false,
3369
+ false,
3370
+ false,
3371
+ false,
3372
+ false,
3373
+ false,
3374
+ true,
3375
+ false,
3376
+ false,
3377
+ true,
3378
+ false,
3379
+ false,
3380
+ true,
3381
+ false,
3382
+ false,
3383
+ false,
3384
+ false,
3385
+ false,
3386
+ false,
3387
+ false,
3388
+ false,
3389
+ false,
3390
+ false,
3391
+ false,
3392
+ false,
3393
+ false,
3394
+ false,
3395
+ false,
3396
+ false,
3397
+ false,
3398
+ false,
3399
+ false,
3400
+ false,
3401
+ false,
3402
+ true,
3403
+ false,
3404
+ false,
3405
+ false,
3406
+ false,
3407
+ false,
3408
+ true,
3409
+ false,
3410
+ false,
3411
+ false,
3412
+ true,
3413
+ true,
3414
+ false,
3415
+ false,
3416
+ false,
3417
+ false,
3418
+ false,
3419
+ false,
3420
+ false,
3421
+ false
3422
+ ],
3423
+ "model.layers.25.self_attn.v_proj.lora_E": [
3424
+ false,
3425
+ false,
3426
+ false,
3427
+ true,
3428
+ false,
3429
+ false,
3430
+ false,
3431
+ true,
3432
+ true,
3433
+ false,
3434
+ false,
3435
+ true,
3436
+ false,
3437
+ true,
3438
+ true,
3439
+ true,
3440
+ false,
3441
+ false,
3442
+ false,
3443
+ false,
3444
+ true,
3445
+ false,
3446
+ false,
3447
+ false,
3448
+ true,
3449
+ true,
3450
+ true,
3451
+ true,
3452
+ false,
3453
+ false,
3454
+ false,
3455
+ false,
3456
+ true,
3457
+ false,
3458
+ false,
3459
+ false,
3460
+ false,
3461
+ true,
3462
+ false,
3463
+ true,
3464
+ false,
3465
+ false,
3466
+ false,
3467
+ false,
3468
+ false,
3469
+ false,
3470
+ false,
3471
+ false,
3472
+ false,
3473
+ false,
3474
+ false,
3475
+ false,
3476
+ true,
3477
+ false,
3478
+ true,
3479
+ false,
3480
+ false,
3481
+ false,
3482
+ false,
3483
+ false,
3484
+ false,
3485
+ false,
3486
+ false,
3487
+ false
3488
+ ],
3489
+ "model.layers.26.self_attn.q_proj.lora_E": [
3490
+ true,
3491
+ false,
3492
+ false,
3493
+ true,
3494
+ false,
3495
+ false,
3496
+ false,
3497
+ false,
3498
+ false,
3499
+ false,
3500
+ true,
3501
+ false,
3502
+ true,
3503
+ false,
3504
+ true,
3505
+ true,
3506
+ true,
3507
+ false,
3508
+ false,
3509
+ true,
3510
+ true,
3511
+ true,
3512
+ false,
3513
+ false,
3514
+ true,
3515
+ true,
3516
+ false,
3517
+ false,
3518
+ true,
3519
+ false,
3520
+ true,
3521
+ true,
3522
+ false,
3523
+ false,
3524
+ false,
3525
+ true,
3526
+ true,
3527
+ false,
3528
+ true,
3529
+ false,
3530
+ false,
3531
+ true,
3532
+ false,
3533
+ false,
3534
+ true,
3535
+ false,
3536
+ false,
3537
+ false,
3538
+ true,
3539
+ false,
3540
+ true,
3541
+ true,
3542
+ false,
3543
+ true,
3544
+ false,
3545
+ false,
3546
+ true,
3547
+ true,
3548
+ true,
3549
+ false,
3550
+ true,
3551
+ true,
3552
+ true,
3553
+ true
3554
+ ],
3555
+ "model.layers.26.self_attn.v_proj.lora_E": [
3556
+ false,
3557
+ false,
3558
+ true,
3559
+ false,
3560
+ true,
3561
+ false,
3562
+ false,
3563
+ false,
3564
+ true,
3565
+ false,
3566
+ false,
3567
+ false,
3568
+ false,
3569
+ false,
3570
+ true,
3571
+ false,
3572
+ false,
3573
+ false,
3574
+ false,
3575
+ false,
3576
+ false,
3577
+ false,
3578
+ true,
3579
+ false,
3580
+ false,
3581
+ false,
3582
+ false,
3583
+ true,
3584
+ false,
3585
+ true,
3586
+ true,
3587
+ false,
3588
+ false,
3589
+ false,
3590
+ false,
3591
+ false,
3592
+ false,
3593
+ false,
3594
+ false,
3595
+ false,
3596
+ true,
3597
+ false,
3598
+ false,
3599
+ false,
3600
+ false,
3601
+ false,
3602
+ false,
3603
+ true,
3604
+ true,
3605
+ true,
3606
+ false,
3607
+ false,
3608
+ false,
3609
+ true,
3610
+ false,
3611
+ true,
3612
+ false,
3613
+ true,
3614
+ false,
3615
+ false,
3616
+ false,
3617
+ true,
3618
+ false,
3619
+ false
3620
+ ],
3621
+ "model.layers.27.self_attn.q_proj.lora_E": [
3622
+ true,
3623
+ false,
3624
+ false,
3625
+ true,
3626
+ true,
3627
+ false,
3628
+ false,
3629
+ true,
3630
+ true,
3631
+ false,
3632
+ false,
3633
+ false,
3634
+ true,
3635
+ true,
3636
+ false,
3637
+ true,
3638
+ false,
3639
+ false,
3640
+ true,
3641
+ false,
3642
+ false,
3643
+ true,
3644
+ true,
3645
+ true,
3646
+ true,
3647
+ false,
3648
+ false,
3649
+ true,
3650
+ true,
3651
+ false,
3652
+ false,
3653
+ false,
3654
+ false,
3655
+ true,
3656
+ true,
3657
+ true,
3658
+ true,
3659
+ false,
3660
+ false,
3661
+ false,
3662
+ false,
3663
+ false,
3664
+ false,
3665
+ true,
3666
+ true,
3667
+ true,
3668
+ false,
3669
+ false,
3670
+ false,
3671
+ true,
3672
+ true,
3673
+ true,
3674
+ true,
3675
+ true,
3676
+ false,
3677
+ false,
3678
+ false,
3679
+ false,
3680
+ false,
3681
+ false,
3682
+ false,
3683
+ false,
3684
+ true,
3685
+ false
3686
+ ],
3687
+ "model.layers.27.self_attn.v_proj.lora_E": [
3688
+ false,
3689
+ false,
3690
+ true,
3691
+ true,
3692
+ true,
3693
+ true,
3694
+ true,
3695
+ true,
3696
+ true,
3697
+ true,
3698
+ false,
3699
+ false,
3700
+ true,
3701
+ false,
3702
+ false,
3703
+ false,
3704
+ true,
3705
+ true,
3706
+ false,
3707
+ false,
3708
+ false,
3709
+ true,
3710
+ false,
3711
+ true,
3712
+ true,
3713
+ true,
3714
+ true,
3715
+ true,
3716
+ false,
3717
+ true,
3718
+ false,
3719
+ false,
3720
+ true,
3721
+ false,
3722
+ true,
3723
+ true,
3724
+ false,
3725
+ true,
3726
+ true,
3727
+ false,
3728
+ false,
3729
+ true,
3730
+ false,
3731
+ true,
3732
+ true,
3733
+ false,
3734
+ false,
3735
+ true,
3736
+ false,
3737
+ true,
3738
+ true,
3739
+ true,
3740
+ false,
3741
+ false,
3742
+ false,
3743
+ false,
3744
+ false,
3745
+ true,
3746
+ true,
3747
+ true,
3748
+ true,
3749
+ true,
3750
+ false,
3751
+ true
3752
+ ]
3753
+ },
3754
+ "alpha_pattern": {},
3755
+ "megatron_config": null,
3756
+ "megatron_core": "megatron.core",
3757
+ "trainable_token_indices": null,
3758
+ "loftq_config": {},
3759
+ "eva_config": null,
3760
+ "corda_config": null,
3761
+ "use_dora": false,
3762
+ "alora_invocation_tokens": null,
3763
+ "use_qalora": false,
3764
+ "qalora_group_size": 16,
3765
+ "layer_replication": null,
3766
+ "lora_bias": false,
3767
+ "target_parameters": null,
3768
+ "use_bdlora": null,
3769
+ "arrow_config": null,
3770
+ "ensure_weight_tying": false,
3771
+ "target_r": 32,
3772
+ "init_r": 64,
3773
+ "tinit": 200,
3774
+ "tfinal": 500,
3775
+ "deltaT": 1,
3776
+ "beta1": 0.85,
3777
+ "beta2": 0.85,
3778
+ "orth_reg_weight": 0.5,
3779
+ "total_step": 5000
3780
+ },
3781
+ "error_msg": ""
3782
+ },
3783
+ "train_info": {
3784
+ "accelerator_memory_reserved_avg": 14915761654,
3785
+ "accelerator_memory_max": 22796042240,
3786
+ "accelerator_memory_reserved_99th": 20617101312,
3787
+ "train_time": 1097.5522341161268,
3788
+ "file_size": 35131056,
3789
+ "num_trainable_params": 18353664,
3790
+ "num_total_params": 3231103544,
3791
+ "status": "success",
3792
+ "metrics": [
3793
+ {
3794
+ "step": 250,
3795
+ "valid accuracy": 0.0,
3796
+ "train loss": 1.3241328690052032,
3797
+ "train samples": 1000,
3798
+ "train time": 33.437529707327485,
3799
+ "eval time": 12.192411294032354,
3800
+ "tokens / sec": 6331.777552143872,
3801
+ "mem allocated avg": 7294713141.248,
3802
+ "mem reserved avg": 15068431384.576,
3803
+ "elapsed time": 71.46285747201182
3804
+ },
3805
+ {
3806
+ "step": 500,
3807
+ "valid accuracy": 0.42,
3808
+ "train loss": 1.0195916182994842,
3809
+ "train samples": 2000,
3810
+ "train time": 35.78730485920096,
3811
+ "eval time": 12.20008662098553,
3812
+ "tokens / sec": 5811.977203042275,
3813
+ "mem allocated avg": 7287517972.48,
3814
+ "mem reserved avg": 14679560683.52,
3815
+ "elapsed time": 123.03286996803945
3816
+ },
3817
+ {
3818
+ "step": 750,
3819
+ "valid accuracy": 0.3,
3820
+ "train loss": 0.788285866856575,
3821
+ "train samples": 3000,
3822
+ "train time": 36.553958278673235,
3823
+ "eval time": 12.192845579993445,
3824
+ "tokens / sec": 5865.329230982038,
3825
+ "mem allocated avg": 7297823625.216,
3826
+ "mem reserved avg": 14887958872.064,
3827
+ "elapsed time": 175.27128310705302
3828
+ },
3829
+ {
3830
+ "step": 1000,
3831
+ "valid accuracy": 0.34,
3832
+ "train loss": 0.7407853524684906,
3833
+ "train samples": 4000,
3834
+ "train time": 36.186631593853235,
3835
+ "eval time": 12.179883008997422,
3836
+ "tokens / sec": 5757.264238857439,
3837
+ "mem allocated avg": 7288626950.144,
3838
+ "mem reserved avg": 14961325637.632,
3839
+ "elapsed time": 227.3045465090545
3840
+ },
3841
+ {
3842
+ "step": 1250,
3843
+ "valid accuracy": 0.32,
3844
+ "train loss": 0.7283406076431275,
3845
+ "train samples": 5000,
3846
+ "train time": 35.77542787248967,
3847
+ "eval time": 12.210806805000175,
3848
+ "tokens / sec": 5829.084721034463,
3849
+ "mem allocated avg": 7289293629.44,
3850
+ "mem reserved avg": 14984788574.208,
3851
+ "elapsed time": 278.867415188055
3852
+ },
3853
+ {
3854
+ "step": 1500,
3855
+ "valid accuracy": 0.38,
3856
+ "train loss": 0.7160038756132125,
3857
+ "train samples": 6000,
3858
+ "train time": 35.90073716780171,
3859
+ "eval time": 12.185589247965254,
3860
+ "tokens / sec": 5830.827345454697,
3861
+ "mem allocated avg": 7291158284.288,
3862
+ "mem reserved avg": 14897169563.648,
3863
+ "elapsed time": 330.540843091032
3864
+ },
3865
+ {
3866
+ "step": 1750,
3867
+ "valid accuracy": 0.34,
3868
+ "train loss": 0.7057664946317673,
3869
+ "train samples": 7000,
3870
+ "train time": 35.91021893679863,
3871
+ "eval time": 12.154321324953344,
3872
+ "tokens / sec": 5829.95610158939,
3873
+ "mem allocated avg": 7292672542.72,
3874
+ "mem reserved avg": 15298061139.968,
3875
+ "elapsed time": 382.3315215060138
3876
+ },
3877
+ {
3878
+ "step": 2000,
3879
+ "valid accuracy": 0.34,
3880
+ "train loss": 0.7058077181577682,
3881
+ "train samples": 8000,
3882
+ "train time": 35.90797309036134,
3883
+ "eval time": 12.198829908971675,
3884
+ "tokens / sec": 5784.119295103047,
3885
+ "mem allocated avg": 7288192765.952,
3886
+ "mem reserved avg": 14933509013.504,
3887
+ "elapsed time": 434.1038082460291
3888
+ },
3889
+ {
3890
+ "step": 2250,
3891
+ "valid accuracy": 0.36,
3892
+ "train loss": 0.6997898343801499,
3893
+ "train samples": 9000,
3894
+ "train time": 36.599770584725775,
3895
+ "eval time": 12.184865101007745,
3896
+ "tokens / sec": 5872.932987446225,
3897
+ "mem allocated avg": 7298940741.632,
3898
+ "mem reserved avg": 15212841271.296,
3899
+ "elapsed time": 486.43084904103307
3900
+ },
3901
+ {
3902
+ "step": 2500,
3903
+ "valid accuracy": 0.36,
3904
+ "train loss": 0.6983125779628754,
3905
+ "train samples": 10000,
3906
+ "train time": 35.72843996837037,
3907
+ "eval time": 7.310388852027245,
3908
+ "tokens / sec": 5764.791303016259,
3909
+ "mem allocated avg": 7285925011.456,
3910
+ "mem reserved avg": 14699911446.528,
3911
+ "elapsed time": 533.0108368530055
3912
+ },
3913
+ {
3914
+ "step": 2750,
3915
+ "valid accuracy": 0.32,
3916
+ "train loss": 0.6912899543046951,
3917
+ "train samples": 11000,
3918
+ "train time": 36.1996374403825,
3919
+ "eval time": 12.285541804973036,
3920
+ "tokens / sec": 5853.124920075475,
3921
+ "mem allocated avg": 7295115497.472,
3922
+ "mem reserved avg": 15075444260.864,
3923
+ "elapsed time": 585.0133225410245
3924
+ },
3925
+ {
3926
+ "step": 3000,
3927
+ "valid accuracy": 0.36,
3928
+ "train loss": 0.6850721148252488,
3929
+ "train samples": 12000,
3930
+ "train time": 35.86890918528661,
3931
+ "eval time": 12.206184039998334,
3932
+ "tokens / sec": 5819.2737036347135,
3933
+ "mem allocated avg": 7291246731.264,
3934
+ "mem reserved avg": 15016489123.84,
3935
+ "elapsed time": 636.763230385026
3936
+ },
3937
+ {
3938
+ "step": 3250,
3939
+ "valid accuracy": 0.34,
3940
+ "train loss": 0.6939100304841995,
3941
+ "train samples": 13000,
3942
+ "train time": 36.13155907928012,
3943
+ "eval time": 12.18468267400749,
3944
+ "tokens / sec": 5837.030157963555,
3945
+ "mem allocated avg": 7293795069.952,
3946
+ "mem reserved avg": 14881256374.272,
3947
+ "elapsed time": 688.5670087640174
3948
+ },
3949
+ {
3950
+ "step": 3500,
3951
+ "valid accuracy": 0.36,
3952
+ "train loss": 0.6824952847957612,
3953
+ "train samples": 14000,
3954
+ "train time": 35.909013247233815,
3955
+ "eval time": 12.309029697033111,
3956
+ "tokens / sec": 5841.151873371449,
3957
+ "mem allocated avg": 7290872406.016,
3958
+ "mem reserved avg": 14875896053.76,
3959
+ "elapsed time": 740.3315773730283
3960
+ },
3961
+ {
3962
+ "step": 3750,
3963
+ "valid accuracy": 0.32,
3964
+ "train loss": 0.6794799867868423,
3965
+ "train samples": 15000,
3966
+ "train time": 36.30529697693419,
3967
+ "eval time": 12.151031351007987,
3968
+ "tokens / sec": 5968.908617871319,
3969
+ "mem allocated avg": 7301860868.096,
3970
+ "mem reserved avg": 15216507092.992,
3971
+ "elapsed time": 792.3454526530113
3972
+ },
3973
+ {
3974
+ "step": 4000,
3975
+ "valid accuracy": 0.3,
3976
+ "train loss": 0.6966380437612534,
3977
+ "train samples": 16000,
3978
+ "train time": 35.68916123179952,
3979
+ "eval time": 12.212951252993662,
3980
+ "tokens / sec": 5726.472490418209,
3981
+ "mem allocated avg": 7282899265.536,
3982
+ "mem reserved avg": 14899249938.432,
3983
+ "elapsed time": 843.872047640034
3984
+ },
3985
+ {
3986
+ "step": 4250,
3987
+ "valid accuracy": 0.32,
3988
+ "train loss": 0.6774853245019913,
3989
+ "train samples": 17000,
3990
+ "train time": 36.08229462581221,
3991
+ "eval time": 12.197503683972172,
3992
+ "tokens / sec": 5858.524303739224,
3993
+ "mem allocated avg": 7294883461.12,
3994
+ "mem reserved avg": 14887791099.904,
3995
+ "elapsed time": 895.6108071450144
3996
+ },
3997
+ {
3998
+ "step": 4500,
3999
+ "valid accuracy": 0.32,
4000
+ "train loss": 0.6867766143083572,
4001
+ "train samples": 18000,
4002
+ "train time": 35.802349225094076,
4003
+ "eval time": 12.20918653300032,
4004
+ "tokens / sec": 5804.591165049559,
4005
+ "mem allocated avg": 7287909275.648,
4006
+ "mem reserved avg": 14847987154.944,
4007
+ "elapsed time": 947.2645250619971
4008
+ },
4009
+ {
4010
+ "step": 4750,
4011
+ "valid accuracy": 0.32,
4012
+ "train loss": 0.6804986503124237,
4013
+ "train samples": 19000,
4014
+ "train time": 32.32421511563007,
4015
+ "eval time": 12.151784093992319,
4016
+ "tokens / sec": 6494.790337491783,
4017
+ "mem allocated avg": 7071402708.992,
4018
+ "mem reserved avg": 14653581164.544,
4019
+ "elapsed time": 995.2198103830451
4020
+ },
4021
+ {
4022
+ "step": 5000,
4023
+ "valid accuracy": 0.34,
4024
+ "train loss": 0.6861251518726349,
4025
+ "train samples": 20000,
4026
+ "train time": 31.86748417984927,
4027
+ "eval time": 12.234077127010096,
4028
+ "tokens / sec": 6535.815592612781,
4029
+ "mem allocated avg": 7067250501.632,
4030
+ "mem reserved avg": 14337473249.28,
4031
+ "elapsed time": 1042.8103677670006
4032
+ },
4033
+ {
4034
+ "step": 5000,
4035
+ "test accuracy": 0.38362395754359363,
4036
+ "train loss": 0.6861251518726349,
4037
+ "train samples": 20000,
4038
+ "train total tokens": 4198051,
4039
+ "forgetting": 0.0754537582397461
4040
+ }
4041
+ ]
4042
+ },
4043
+ "meta_info": {
4044
+ "model_info": {
4045
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
4046
+ "created_at": "2024-09-18T15:23:48+00:00"
4047
+ },
4048
+ "dataset_info": {
4049
+ "metamath": {
4050
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
4051
+ "created_at": "2023-09-21T17:22:46+00:00"
4052
+ },
4053
+ "gsm8k": {
4054
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
4055
+ "created_at": "2022-04-12T10:22:10+00:00"
4056
+ }
4057
+ },
4058
+ "package_info": {
4059
+ "transformers-version": "4.57.1",
4060
+ "transformers-commit-hash": null,
4061
+ "peft-version": "0.18.1.dev0",
4062
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
4063
+ "datasets-version": "4.2.0",
4064
+ "datasets-commit-hash": null,
4065
+ "bitsandbytes-version": "0.46.0",
4066
+ "bitsandbytes-commit-hash": null,
4067
+ "torch-version": "2.9.0+cu128",
4068
+ "torch-commit-hash": null
4069
+ },
4070
+ "system_info": {
4071
+ "system": "Linux",
4072
+ "release": "6.14.0-1016-aws",
4073
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
4074
+ "machine": "x86_64",
4075
+ "processor": "x86_64",
4076
+ "accelerator": "NVIDIA L40S"
4077
+ },
4078
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
4079
+ }
4080
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/adaptionprompt--llama-3.2-3B-lr_0.0005.json ADDED
@@ -0,0 +1,343 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T17:25:16+00:00",
4
+ "total_time": 1466.8854283409892,
5
+ "experiment_name": "adaptionprompt/llama-3.2-3B-lr_0.0005",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0005
22
+ },
23
+ "lr_scheduler": "cosine",
24
+ "use_amp": false,
25
+ "autocast_adapter_dtype": true,
26
+ "generation_kwargs": {
27
+ "max_length": 800,
28
+ "max_new_tokens": 300
29
+ },
30
+ "attn_implementation": null
31
+ },
32
+ "peft_config": {
33
+ "task_type": "CAUSAL_LM",
34
+ "peft_type": "ADAPTION_PROMPT",
35
+ "auto_mapping": null,
36
+ "peft_version": "0.18.1.dev0@UNKNOWN",
37
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
38
+ "revision": null,
39
+ "inference_mode": false,
40
+ "target_modules": "self_attn",
41
+ "adapter_len": 100,
42
+ "adapter_layers": 28
43
+ },
44
+ "error_msg": ""
45
+ },
46
+ "train_info": {
47
+ "accelerator_memory_reserved_avg": 14458868570,
48
+ "accelerator_memory_max": 22445817856,
49
+ "accelerator_memory_reserved_99th": 20222836736,
50
+ "train_time": 1180.5937879550038,
51
+ "file_size": 17210384,
52
+ "num_trainable_params": 8601628,
53
+ "num_total_params": 3221351452,
54
+ "status": "success",
55
+ "metrics": [
56
+ {
57
+ "step": 250,
58
+ "valid accuracy": 0.0,
59
+ "train loss": 1.3200566630363464,
60
+ "train samples": 1000,
61
+ "train time": 34.53835600853199,
62
+ "eval time": 14.21398706396576,
63
+ "tokens / sec": 6129.967504756138,
64
+ "mem allocated avg": 6848584845.312,
65
+ "mem reserved avg": 14547960201.216,
66
+ "elapsed time": 71.65906680002809
67
+ },
68
+ {
69
+ "step": 500,
70
+ "valid accuracy": 0.08,
71
+ "train loss": 1.1533236474990844,
72
+ "train samples": 2000,
73
+ "train time": 34.522620833537076,
74
+ "eval time": 14.150563488015905,
75
+ "tokens / sec": 6024.890201787426,
76
+ "mem allocated avg": 6841784819.712,
77
+ "mem reserved avg": 14164156219.392,
78
+ "elapsed time": 123.70079839800019
79
+ },
80
+ {
81
+ "step": 750,
82
+ "valid accuracy": 0.2,
83
+ "train loss": 0.9019568083286286,
84
+ "train samples": 3000,
85
+ "train time": 35.55624677182641,
86
+ "eval time": 14.130223647051025,
87
+ "tokens / sec": 6029.910900772695,
88
+ "mem allocated avg": 6852599943.168,
89
+ "mem reserved avg": 14414614888.448,
90
+ "elapsed time": 176.709548267012
91
+ },
92
+ {
93
+ "step": 1000,
94
+ "valid accuracy": 0.18,
95
+ "train loss": 0.8574914336204529,
96
+ "train samples": 4000,
97
+ "train time": 35.06597913411679,
98
+ "eval time": 14.383076103986241,
99
+ "tokens / sec": 5941.2571713220295,
100
+ "mem allocated avg": 6843962202.112,
101
+ "mem reserved avg": 14454611771.392,
102
+ "elapsed time": 229.6261993580265
103
+ },
104
+ {
105
+ "step": 1250,
106
+ "valid accuracy": 0.22,
107
+ "train loss": 0.8493955388069153,
108
+ "train samples": 5000,
109
+ "train time": 35.032775730942376,
110
+ "eval time": 14.193959789990913,
111
+ "tokens / sec": 5952.6542116333285,
112
+ "mem allocated avg": 6843453933.568,
113
+ "mem reserved avg": 14489709707.264,
114
+ "elapsed time": 282.2174852700555
115
+ },
116
+ {
117
+ "step": 1500,
118
+ "valid accuracy": 0.2,
119
+ "train loss": 0.8379271342754364,
120
+ "train samples": 6000,
121
+ "train time": 35.1143401027075,
122
+ "eval time": 14.287394939980004,
123
+ "tokens / sec": 5961.410619926743,
124
+ "mem allocated avg": 6845303027.712,
125
+ "mem reserved avg": 14413767639.04,
126
+ "elapsed time": 335.0856573909987
127
+ },
128
+ {
129
+ "step": 1750,
130
+ "valid accuracy": 0.24,
131
+ "train loss": 0.8321145892143249,
132
+ "train samples": 7000,
133
+ "train time": 34.62294518906856,
134
+ "eval time": 14.048951104981825,
135
+ "tokens / sec": 6046.712631081982,
136
+ "mem allocated avg": 6846216781.824,
137
+ "mem reserved avg": 14785382973.44,
138
+ "elapsed time": 387.15798614604864
139
+ },
140
+ {
141
+ "step": 2000,
142
+ "valid accuracy": 0.24,
143
+ "train loss": 0.836554151058197,
144
+ "train samples": 8000,
145
+ "train time": 34.390015198267065,
146
+ "eval time": 14.16389181703562,
147
+ "tokens / sec": 6039.427397824062,
148
+ "mem allocated avg": 6842828976.128,
149
+ "mem reserved avg": 14449343725.568,
150
+ "elapsed time": 439.18930962705053
151
+ },
152
+ {
153
+ "step": 2250,
154
+ "valid accuracy": 0.2,
155
+ "train loss": 0.8322268238067627,
156
+ "train samples": 9000,
157
+ "train time": 35.45279946445953,
158
+ "eval time": 14.172650399035774,
159
+ "tokens / sec": 6062.934471944297,
160
+ "mem allocated avg": 6854121762.816,
161
+ "mem reserved avg": 14755116875.776,
162
+ "elapsed time": 492.21291138301603
163
+ },
164
+ {
165
+ "step": 2500,
166
+ "valid accuracy": 0.24,
167
+ "train loss": 0.8304620375633239,
168
+ "train samples": 10000,
169
+ "train time": 35.034967973362654,
170
+ "eval time": 14.244871944014449,
171
+ "tokens / sec": 5878.897910127911,
172
+ "mem allocated avg": 6840198320.128,
173
+ "mem reserved avg": 14230501720.064,
174
+ "elapsed time": 544.8923244239995
175
+ },
176
+ {
177
+ "step": 2750,
178
+ "valid accuracy": 0.22,
179
+ "train loss": 0.8325318789482117,
180
+ "train samples": 11000,
181
+ "train time": 35.15703216131078,
182
+ "eval time": 14.128618655027822,
183
+ "tokens / sec": 6026.703250371868,
184
+ "mem allocated avg": 6849819580.416,
185
+ "mem reserved avg": 14618449674.24,
186
+ "elapsed time": 597.5583866710076
187
+ },
188
+ {
189
+ "step": 3000,
190
+ "valid accuracy": 0.16,
191
+ "train loss": 0.8273461208343506,
192
+ "train samples": 12000,
193
+ "train time": 35.226769833650906,
194
+ "eval time": 14.172337282972876,
195
+ "tokens / sec": 5925.3516852574585,
196
+ "mem allocated avg": 6844943210.496,
197
+ "mem reserved avg": 14524555984.896,
198
+ "elapsed time": 650.4424432980013
199
+ },
200
+ {
201
+ "step": 3250,
202
+ "valid accuracy": 0.16,
203
+ "train loss": 0.832145271062851,
204
+ "train samples": 13000,
205
+ "train time": 34.82297695695888,
206
+ "eval time": 14.088282125012483,
207
+ "tokens / sec": 6056.374797039126,
208
+ "mem allocated avg": 6846756392.96,
209
+ "mem reserved avg": 14383988080.64,
210
+ "elapsed time": 702.644515115011
211
+ },
212
+ {
213
+ "step": 3500,
214
+ "valid accuracy": 0.2,
215
+ "train loss": 0.8267167332172394,
216
+ "train samples": 14000,
217
+ "train time": 35.251085491792765,
218
+ "eval time": 14.197126877959818,
219
+ "tokens / sec": 5950.171379795793,
220
+ "mem allocated avg": 6845619677.184,
221
+ "mem reserved avg": 14408566702.08,
222
+ "elapsed time": 755.514794363
223
+ },
224
+ {
225
+ "step": 3750,
226
+ "valid accuracy": 0.24,
227
+ "train loss": 0.8225111892223358,
228
+ "train samples": 15000,
229
+ "train time": 35.429980689892545,
230
+ "eval time": 14.131252312974539,
231
+ "tokens / sec": 6116.373641203281,
232
+ "mem allocated avg": 6856933664.768,
233
+ "mem reserved avg": 14748112388.096,
234
+ "elapsed time": 808.5010599610396
235
+ },
236
+ {
237
+ "step": 4000,
238
+ "valid accuracy": 0.18,
239
+ "train loss": 0.8426167027950286,
240
+ "train samples": 16000,
241
+ "train time": 34.42210436682217,
242
+ "eval time": 14.11721703200601,
243
+ "tokens / sec": 5937.260483033845,
244
+ "mem allocated avg": 6838090446.848,
245
+ "mem reserved avg": 14423599087.616,
246
+ "elapsed time": 860.4732752880082
247
+ },
248
+ {
249
+ "step": 4250,
250
+ "valid accuracy": 0.22,
251
+ "train loss": 0.8195950338840484,
252
+ "train samples": 17000,
253
+ "train time": 35.29558803292457,
254
+ "eval time": 14.072754080989398,
255
+ "tokens / sec": 5989.105488278345,
256
+ "mem allocated avg": 6848445681.664,
257
+ "mem reserved avg": 14423037050.88,
258
+ "elapsed time": 913.1631882850197
259
+ },
260
+ {
261
+ "step": 4500,
262
+ "valid accuracy": 0.22,
263
+ "train loss": 0.8332238008975983,
264
+ "train samples": 18000,
265
+ "train time": 34.60890661936719,
266
+ "eval time": 14.16370828700019,
267
+ "tokens / sec": 6004.754853587451,
268
+ "mem allocated avg": 6843256502.272,
269
+ "mem reserved avg": 14372839620.608,
270
+ "elapsed time": 965.424148318998
271
+ },
272
+ {
273
+ "step": 4750,
274
+ "valid accuracy": 0.2,
275
+ "train loss": 0.8247757973670959,
276
+ "train samples": 19000,
277
+ "train time": 34.65050247934414,
278
+ "eval time": 14.110966334003024,
279
+ "tokens / sec": 6058.7577373560125,
280
+ "mem allocated avg": 6846388692.992,
281
+ "mem reserved avg": 14453437366.272,
282
+ "elapsed time": 1017.51672834903
283
+ },
284
+ {
285
+ "step": 5000,
286
+ "valid accuracy": 0.24,
287
+ "train loss": 0.8316655712127685,
288
+ "train samples": 20000,
289
+ "train time": 34.58433834684547,
290
+ "eval time": 14.152012452017516,
291
+ "tokens / sec": 6022.379202723645,
292
+ "mem allocated avg": 6842473314.304,
293
+ "mem reserved avg": 14115619733.504,
294
+ "elapsed time": 1069.6098215640523
295
+ },
296
+ {
297
+ "step": 5000,
298
+ "test accuracy": 0.21228203184230476,
299
+ "train loss": 0.8316655712127685,
300
+ "train samples": 20000,
301
+ "train total tokens": 4198051,
302
+ "forgetting": -0.9543380737304688
303
+ }
304
+ ]
305
+ },
306
+ "meta_info": {
307
+ "model_info": {
308
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
309
+ "created_at": "2024-09-18T15:23:48+00:00"
310
+ },
311
+ "dataset_info": {
312
+ "metamath": {
313
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
314
+ "created_at": "2023-09-21T17:22:46+00:00"
315
+ },
316
+ "gsm8k": {
317
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
318
+ "created_at": "2022-04-12T10:22:10+00:00"
319
+ }
320
+ },
321
+ "package_info": {
322
+ "transformers-version": "4.57.1",
323
+ "transformers-commit-hash": null,
324
+ "peft-version": "0.18.1.dev0",
325
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
326
+ "datasets-version": "4.2.0",
327
+ "datasets-commit-hash": null,
328
+ "bitsandbytes-version": "0.46.0",
329
+ "bitsandbytes-commit-hash": null,
330
+ "torch-version": "2.9.0+cu128",
331
+ "torch-commit-hash": null
332
+ },
333
+ "system_info": {
334
+ "system": "Linux",
335
+ "release": "6.14.0-1016-aws",
336
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
337
+ "machine": "x86_64",
338
+ "processor": "x86_64",
339
+ "accelerator": "NVIDIA L40S"
340
+ },
341
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
342
+ }
343
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/boft--llama-3.2-3B-default.json ADDED
@@ -0,0 +1,356 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T17:49:47+00:00",
4
+ "total_time": 10059.440659368003,
5
+ "experiment_name": "boft/llama-3.2-3B-default",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "BOFT",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "boft_block_size": 4,
42
+ "boft_block_num": 0,
43
+ "boft_n_butterfly_factor": 1,
44
+ "target_modules": [
45
+ "v_proj",
46
+ "q_proj"
47
+ ],
48
+ "exclude_modules": null,
49
+ "boft_dropout": 0.0,
50
+ "fan_in_fan_out": false,
51
+ "bias": "none",
52
+ "modules_to_save": null,
53
+ "init_weights": true,
54
+ "layers_to_transform": null,
55
+ "layers_pattern": null
56
+ },
57
+ "error_msg": ""
58
+ },
59
+ "train_info": {
60
+ "accelerator_memory_reserved_avg": 17188165936,
61
+ "accelerator_memory_max": 24410849280,
62
+ "accelerator_memory_reserved_99th": 22345154560,
63
+ "train_time": 7190.453845723998,
64
+ "file_size": 3225360,
65
+ "num_trainable_params": 802816,
66
+ "num_total_params": 3213552640,
67
+ "status": "success",
68
+ "metrics": [
69
+ {
70
+ "step": 250,
71
+ "valid accuracy": 0.0,
72
+ "train loss": 1.2914534993171691,
73
+ "train samples": 1000,
74
+ "train time": 168.16101133066695,
75
+ "eval time": 142.30593162897276,
76
+ "tokens / sec": 1259.025491846513,
77
+ "mem allocated avg": 6795307169.792,
78
+ "mem reserved avg": 17261339344.896,
79
+ "elapsed time": 344.37489816197194
80
+ },
81
+ {
82
+ "step": 500,
83
+ "valid accuracy": 0.12,
84
+ "train loss": 1.065816521883011,
85
+ "train samples": 2000,
86
+ "train time": 169.6081039829878,
87
+ "eval time": 143.82841559901135,
88
+ "tokens / sec": 1226.3270157236266,
89
+ "mem allocated avg": 6787530438.656,
90
+ "mem reserved avg": 16960280592.384,
91
+ "elapsed time": 662.5959406199981
92
+ },
93
+ {
94
+ "step": 750,
95
+ "valid accuracy": 0.42,
96
+ "train loss": 0.8761289038658142,
97
+ "train samples": 3000,
98
+ "train time": 170.50346154812723,
99
+ "eval time": 143.55063587299082,
100
+ "tokens / sec": 1257.4583416271698,
101
+ "mem allocated avg": 6797964732.416,
102
+ "mem reserved avg": 17119546703.872,
103
+ "elapsed time": 981.3382467220072
104
+ },
105
+ {
106
+ "step": 1000,
107
+ "valid accuracy": 0.44,
108
+ "train loss": 0.8187025270462036,
109
+ "train samples": 4000,
110
+ "train time": 170.603982186527,
111
+ "eval time": 143.48508089897223,
112
+ "tokens / sec": 1221.1672748190563,
113
+ "mem allocated avg": 6789360863.232,
114
+ "mem reserved avg": 17173468676.096,
115
+ "elapsed time": 1300.2320007149829
116
+ },
117
+ {
118
+ "step": 1250,
119
+ "valid accuracy": 0.38,
120
+ "train loss": 0.7968714184761048,
121
+ "train samples": 5000,
122
+ "train time": 170.08946738566738,
123
+ "eval time": 143.52085828600684,
124
+ "tokens / sec": 1226.0488741913275,
125
+ "mem allocated avg": 6788928741.376,
126
+ "mem reserved avg": 17237641527.296,
127
+ "elapsed time": 1618.561195612012
128
+ },
129
+ {
130
+ "step": 1500,
131
+ "valid accuracy": 0.42,
132
+ "train loss": 0.7768479048013687,
133
+ "train samples": 6000,
134
+ "train time": 171.03791188867763,
135
+ "eval time": 142.4387985090143,
136
+ "tokens / sec": 1223.8865505809376,
137
+ "mem allocated avg": 6790983270.4,
138
+ "mem reserved avg": 17204464582.656,
139
+ "elapsed time": 1936.8438452039845
140
+ },
141
+ {
142
+ "step": 1750,
143
+ "valid accuracy": 0.34,
144
+ "train loss": 0.7638826340436935,
145
+ "train samples": 7000,
146
+ "train time": 169.5759706167737,
147
+ "eval time": 142.81106396700488,
148
+ "tokens / sec": 1234.5793996551745,
149
+ "mem allocated avg": 6792170039.296,
150
+ "mem reserved avg": 17521872732.16,
151
+ "elapsed time": 2253.9779722079984
152
+ },
153
+ {
154
+ "step": 2000,
155
+ "valid accuracy": 0.3,
156
+ "train loss": 0.7574314765930176,
157
+ "train samples": 8000,
158
+ "train time": 169.80084089771844,
159
+ "eval time": 142.94605548703112,
160
+ "tokens / sec": 1223.1741545090943,
161
+ "mem allocated avg": 6788512305.152,
162
+ "mem reserved avg": 17193743941.632,
163
+ "elapsed time": 2571.564919157012
164
+ },
165
+ {
166
+ "step": 2250,
167
+ "valid accuracy": 0.4,
168
+ "train loss": 0.747995015501976,
169
+ "train samples": 9000,
170
+ "train time": 169.89184381318046,
171
+ "eval time": 142.61838847299805,
172
+ "tokens / sec": 1265.2049396577568,
173
+ "mem allocated avg": 6799572060.16,
174
+ "mem reserved avg": 17437013573.632,
175
+ "elapsed time": 2888.81089515402
176
+ },
177
+ {
178
+ "step": 2500,
179
+ "valid accuracy": 0.3,
180
+ "train loss": 0.7451818664073944,
181
+ "train samples": 10000,
182
+ "train time": 169.05202275590273,
183
+ "eval time": 142.17101406800793,
184
+ "tokens / sec": 1218.3645995020095,
185
+ "mem allocated avg": 6785213417.472,
186
+ "mem reserved avg": 17006149500.928,
187
+ "elapsed time": 3204.817744602973
188
+ },
189
+ {
190
+ "step": 2750,
191
+ "valid accuracy": 0.26,
192
+ "train loss": 0.7368416948318481,
193
+ "train samples": 11000,
194
+ "train time": 169.79861720540794,
195
+ "eval time": 142.90656082698843,
196
+ "tokens / sec": 1247.837017092338,
197
+ "mem allocated avg": 6795389593.6,
198
+ "mem reserved avg": 17333305212.928,
199
+ "elapsed time": 3522.2434823570075
200
+ },
201
+ {
202
+ "step": 3000,
203
+ "valid accuracy": 0.42,
204
+ "train loss": 0.7283624488115311,
205
+ "train samples": 12000,
206
+ "train time": 169.40914075018372,
207
+ "eval time": 142.4628674259875,
208
+ "tokens / sec": 1232.1117920537804,
209
+ "mem allocated avg": 6790662739.968,
210
+ "mem reserved avg": 17229924007.936,
211
+ "elapsed time": 3838.9266921749804
212
+ },
213
+ {
214
+ "step": 3250,
215
+ "valid accuracy": 0.4,
216
+ "train loss": 0.7360379147529602,
217
+ "train samples": 13000,
218
+ "train time": 169.2894103731378,
219
+ "eval time": 142.70766545902006,
220
+ "tokens / sec": 1245.8014918661738,
221
+ "mem allocated avg": 6792828252.16,
222
+ "mem reserved avg": 17116644245.504,
223
+ "elapsed time": 4155.603474072996
224
+ },
225
+ {
226
+ "step": 3500,
227
+ "valid accuracy": 0.38,
228
+ "train loss": 0.7245079389810563,
229
+ "train samples": 14000,
230
+ "train time": 169.62816236459184,
231
+ "eval time": 142.80769011599477,
232
+ "tokens / sec": 1236.528162989657,
233
+ "mem allocated avg": 6790204116.992,
234
+ "mem reserved avg": 17182729699.328,
235
+ "elapsed time": 4472.8496758749825
236
+ },
237
+ {
238
+ "step": 3750,
239
+ "valid accuracy": 0.32,
240
+ "train loss": 0.7198032699823379,
241
+ "train samples": 15000,
242
+ "train time": 169.99743940570625,
243
+ "eval time": 143.21545482298825,
244
+ "tokens / sec": 1274.742729993885,
245
+ "mem allocated avg": 6802659770.368,
246
+ "mem reserved avg": 17454201831.424,
247
+ "elapsed time": 4790.820539581007
248
+ },
249
+ {
250
+ "step": 4000,
251
+ "valid accuracy": 0.36,
252
+ "train loss": 0.7386080088615418,
253
+ "train samples": 16000,
254
+ "train time": 169.4405590199167,
255
+ "eval time": 142.85203308903147,
256
+ "tokens / sec": 1206.1633954829977,
257
+ "mem allocated avg": 6782729986.048,
258
+ "mem reserved avg": 17107366445.056,
259
+ "elapsed time": 5107.90697863797
260
+ },
261
+ {
262
+ "step": 4250,
263
+ "valid accuracy": 0.4,
264
+ "train loss": 0.7167673094272613,
265
+ "train samples": 17000,
266
+ "train time": 169.4217613734072,
267
+ "eval time": 142.6639407540206,
268
+ "tokens / sec": 1247.7086667402577,
269
+ "mem allocated avg": 6793673689.088,
270
+ "mem reserved avg": 17128346353.664,
271
+ "elapsed time": 5424.600868158974
272
+ },
273
+ {
274
+ "step": 4500,
275
+ "valid accuracy": 0.36,
276
+ "train loss": 0.7279775400161743,
277
+ "train samples": 18000,
278
+ "train time": 169.55897833802737,
279
+ "eval time": 142.51284795097308,
280
+ "tokens / sec": 1225.6384299845254,
281
+ "mem allocated avg": 6789383653.376,
282
+ "mem reserved avg": 17117466329.088,
283
+ "elapsed time": 5741.48882275197
284
+ },
285
+ {
286
+ "step": 4750,
287
+ "valid accuracy": 0.4,
288
+ "train loss": 0.7207833558321,
289
+ "train samples": 19000,
290
+ "train time": 169.6458221563953,
291
+ "eval time": 142.71257099701324,
292
+ "tokens / sec": 1237.513528664789,
293
+ "mem allocated avg": 6791604719.616,
294
+ "mem reserved avg": 17139536756.736,
295
+ "elapsed time": 6058.54882254597
296
+ },
297
+ {
298
+ "step": 5000,
299
+ "valid accuracy": 0.36,
300
+ "train loss": 0.7267992464303971,
301
+ "train samples": 20000,
302
+ "train time": 169.51394822495058,
303
+ "eval time": 142.66391144797672,
304
+ "tokens / sec": 1228.6894511099795,
305
+ "mem allocated avg": 6788474263.552,
306
+ "mem reserved avg": 16838276677.632,
307
+ "elapsed time": 6375.395246869011
308
+ },
309
+ {
310
+ "step": 5000,
311
+ "test accuracy": 0.36239575435936316,
312
+ "train loss": 0.7267992464303971,
313
+ "train samples": 20000,
314
+ "train total tokens": 4198051,
315
+ "forgetting": 0.3084230422973633
316
+ }
317
+ ]
318
+ },
319
+ "meta_info": {
320
+ "model_info": {
321
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
322
+ "created_at": "2024-09-18T15:23:48+00:00"
323
+ },
324
+ "dataset_info": {
325
+ "metamath": {
326
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
327
+ "created_at": "2023-09-21T17:22:46+00:00"
328
+ },
329
+ "gsm8k": {
330
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
331
+ "created_at": "2022-04-12T10:22:10+00:00"
332
+ }
333
+ },
334
+ "package_info": {
335
+ "transformers-version": "4.57.1",
336
+ "transformers-commit-hash": null,
337
+ "peft-version": "0.18.1.dev0",
338
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
339
+ "datasets-version": "4.2.0",
340
+ "datasets-commit-hash": null,
341
+ "bitsandbytes-version": "0.46.0",
342
+ "bitsandbytes-commit-hash": null,
343
+ "torch-version": "2.9.0+cu128",
344
+ "torch-commit-hash": null
345
+ },
346
+ "system_info": {
347
+ "system": "Linux",
348
+ "release": "6.14.0-1016-aws",
349
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
350
+ "machine": "x86_64",
351
+ "processor": "x86_64",
352
+ "accelerator": "NVIDIA L40S"
353
+ },
354
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
355
+ }
356
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/bone--llama-3.2-3B-bat.json ADDED
@@ -0,0 +1,352 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T20:37:31+00:00",
4
+ "total_time": 1743.6652406350477,
5
+ "experiment_name": "bone/llama-3.2-3B-bat",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "BONE",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "r": 64,
42
+ "target_modules": [
43
+ "q_proj",
44
+ "v_proj"
45
+ ],
46
+ "exclude_modules": null,
47
+ "init_weights": "bat",
48
+ "layers_to_transform": null,
49
+ "layers_pattern": null,
50
+ "bias": "none",
51
+ "modules_to_save": null
52
+ },
53
+ "error_msg": ""
54
+ },
55
+ "train_info": {
56
+ "accelerator_memory_reserved_avg": 17289858514,
57
+ "accelerator_memory_max": 25209864192,
58
+ "accelerator_memory_reserved_99th": 23320330240,
59
+ "train_time": 1432.8082630151184,
60
+ "file_size": 29367552,
61
+ "num_trainable_params": 7340032,
62
+ "num_total_params": 3220089856,
63
+ "status": "success",
64
+ "metrics": [
65
+ {
66
+ "step": 250,
67
+ "valid accuracy": 0.38,
68
+ "train loss": 0.8741573710441589,
69
+ "train samples": 1000,
70
+ "train time": 47.08210420422256,
71
+ "eval time": 17.23298390500713,
72
+ "tokens / sec": 4496.804116520603,
73
+ "mem allocated avg": 6899156936.704,
74
+ "mem reserved avg": 17432425005.056,
75
+ "elapsed time": 92.3484000210301
76
+ },
77
+ {
78
+ "step": 500,
79
+ "valid accuracy": 0.44,
80
+ "train loss": 0.694805077791214,
81
+ "train samples": 2000,
82
+ "train time": 45.921670439012814,
83
+ "eval time": 10.156194142997265,
84
+ "tokens / sec": 4529.343075100717,
85
+ "mem allocated avg": 6892228743.168,
86
+ "mem reserved avg": 17057630388.224,
87
+ "elapsed time": 153.09530606598128
88
+ },
89
+ {
90
+ "step": 750,
91
+ "valid accuracy": 0.46,
92
+ "train loss": 0.6675743660926818,
93
+ "train samples": 3000,
94
+ "train time": 46.44820640789112,
95
+ "eval time": 9.883839346992318,
96
+ "tokens / sec": 4615.915588154449,
97
+ "mem allocated avg": 6902447476.736,
98
+ "mem reserved avg": 17196562513.92,
99
+ "elapsed time": 213.98306120501366
100
+ },
101
+ {
102
+ "step": 1000,
103
+ "valid accuracy": 0.38,
104
+ "train loss": 0.6475769439935685,
105
+ "train samples": 4000,
106
+ "train time": 46.272512987605296,
107
+ "eval time": 11.084571123996284,
108
+ "tokens / sec": 4502.370555405226,
109
+ "mem allocated avg": 6893826957.312,
110
+ "mem reserved avg": 17298584764.416,
111
+ "elapsed time": 276.0952120529837
112
+ },
113
+ {
114
+ "step": 1250,
115
+ "valid accuracy": 0.42,
116
+ "train loss": 0.6444406238794327,
117
+ "train samples": 5000,
118
+ "train time": 45.26340117654763,
119
+ "eval time": 8.693854385986924,
120
+ "tokens / sec": 4607.210120746516,
121
+ "mem allocated avg": 6894148816.896,
122
+ "mem reserved avg": 17275818082.304,
123
+ "elapsed time": 334.61537264700746
124
+ },
125
+ {
126
+ "step": 1500,
127
+ "valid accuracy": 0.52,
128
+ "train loss": 0.6367908464670181,
129
+ "train samples": 6000,
130
+ "train time": 45.64225660829106,
131
+ "eval time": 12.539341575000435,
132
+ "tokens / sec": 4586.342033798003,
133
+ "mem allocated avg": 6895285663.744,
134
+ "mem reserved avg": 17295631974.4,
135
+ "elapsed time": 397.52763449802296
136
+ },
137
+ {
138
+ "step": 1750,
139
+ "valid accuracy": 0.38,
140
+ "train loss": 0.6280897797346116,
141
+ "train samples": 7000,
142
+ "train time": 46.028662396303844,
143
+ "eval time": 17.17660208797315,
144
+ "tokens / sec": 4548.361588209252,
145
+ "mem allocated avg": 6897027553.28,
146
+ "mem reserved avg": 17636528226.304,
147
+ "elapsed time": 465.4344937889837
148
+ },
149
+ {
150
+ "step": 2000,
151
+ "valid accuracy": 0.4,
152
+ "train loss": 0.6281794626712799,
153
+ "train samples": 8000,
154
+ "train time": 46.304100580571685,
155
+ "eval time": 17.161395469971467,
156
+ "tokens / sec": 4485.477471667926,
157
+ "mem allocated avg": 6892820834.304,
158
+ "mem reserved avg": 17282487025.664,
159
+ "elapsed time": 533.6705477859941
160
+ },
161
+ {
162
+ "step": 2250,
163
+ "valid accuracy": 0.44,
164
+ "train loss": 0.6160554866790772,
165
+ "train samples": 9000,
166
+ "train time": 47.66154300532071,
167
+ "eval time": 14.93811830598861,
168
+ "tokens / sec": 4509.883366050575,
169
+ "mem allocated avg": 6904737810.432,
170
+ "mem reserved avg": 17579468914.688,
171
+ "elapsed time": 600.9623893949902
172
+ },
173
+ {
174
+ "step": 2500,
175
+ "valid accuracy": 0.48,
176
+ "train loss": 0.6119242373704911,
177
+ "train samples": 10000,
178
+ "train time": 45.28331206360599,
179
+ "eval time": 17.183108089026064,
180
+ "tokens / sec": 4548.408466913683,
181
+ "mem allocated avg": 6890091653.12,
182
+ "mem reserved avg": 17013791522.816,
183
+ "elapsed time": 668.064512034005
184
+ },
185
+ {
186
+ "step": 2750,
187
+ "valid accuracy": 0.48,
188
+ "train loss": 0.6009454892873765,
189
+ "train samples": 11000,
190
+ "train time": 46.55525783565827,
191
+ "eval time": 17.103973756020423,
192
+ "tokens / sec": 4551.172302555977,
193
+ "mem allocated avg": 6899537018.88,
194
+ "mem reserved avg": 17462959538.176,
195
+ "elapsed time": 736.392622578016
196
+ },
197
+ {
198
+ "step": 3000,
199
+ "valid accuracy": 0.42,
200
+ "train loss": 0.5898327060937881,
201
+ "train samples": 12000,
202
+ "train time": 45.69055778736947,
203
+ "eval time": 17.16533613699721,
204
+ "tokens / sec": 4568.361825902261,
205
+ "mem allocated avg": 6895349350.4,
206
+ "mem reserved avg": 17354150903.808,
207
+ "elapsed time": 804.0024728610297
208
+ },
209
+ {
210
+ "step": 3250,
211
+ "valid accuracy": 0.5,
212
+ "train loss": 0.5984167819023132,
213
+ "train samples": 13000,
214
+ "train time": 46.50175124424277,
215
+ "eval time": 17.302273807988968,
216
+ "tokens / sec": 4535.334570353648,
217
+ "mem allocated avg": 6898310023.168,
218
+ "mem reserved avg": 17226862166.016,
219
+ "elapsed time": 872.4092298070318
220
+ },
221
+ {
222
+ "step": 3500,
223
+ "valid accuracy": 0.48,
224
+ "train loss": 0.5800097153186798,
225
+ "train samples": 14000,
226
+ "train time": 45.426418964751065,
227
+ "eval time": 17.125486647011712,
228
+ "tokens / sec": 4617.357141067116,
229
+ "mem allocated avg": 6895650811.904,
230
+ "mem reserved avg": 17232507699.2,
231
+ "elapsed time": 939.6246801590314
232
+ },
233
+ {
234
+ "step": 3750,
235
+ "valid accuracy": 0.42,
236
+ "train loss": 0.577217645406723,
237
+ "train samples": 15000,
238
+ "train time": 47.63793423282914,
239
+ "eval time": 17.120611401973292,
240
+ "tokens / sec": 4548.958797013947,
241
+ "mem allocated avg": 6907398975.488,
242
+ "mem reserved avg": 17587723304.96,
243
+ "elapsed time": 1009.0847758559976
244
+ },
245
+ {
246
+ "step": 4000,
247
+ "valid accuracy": 0.5,
248
+ "train loss": 0.5854997953176498,
249
+ "train samples": 16000,
250
+ "train time": 44.550334901490714,
251
+ "eval time": 17.11661865498172,
252
+ "tokens / sec": 4587.462708235699,
253
+ "mem allocated avg": 6887509987.328,
254
+ "mem reserved avg": 17280842858.496,
255
+ "elapsed time": 1075.4839722960023
256
+ },
257
+ {
258
+ "step": 4250,
259
+ "valid accuracy": 0.48,
260
+ "train loss": 0.5725951849222183,
261
+ "train samples": 17000,
262
+ "train time": 45.552318896050565,
263
+ "eval time": 17.120852815976832,
264
+ "tokens / sec": 4640.576048002853,
265
+ "mem allocated avg": 6898683543.552,
266
+ "mem reserved avg": 17244771844.096,
267
+ "elapsed time": 1142.6874302860233
268
+ },
269
+ {
270
+ "step": 4500,
271
+ "valid accuracy": 0.48,
272
+ "train loss": 0.5793783943653107,
273
+ "train samples": 18000,
274
+ "train time": 46.363164163019974,
275
+ "eval time": 17.14258968201466,
276
+ "tokens / sec": 4482.394671538814,
277
+ "mem allocated avg": 6894085709.824,
278
+ "mem reserved avg": 17169358258.176,
279
+ "elapsed time": 1210.9217723699985
280
+ },
281
+ {
282
+ "step": 4750,
283
+ "valid accuracy": 0.46,
284
+ "train loss": 0.5683046000003815,
285
+ "train samples": 19000,
286
+ "train time": 45.238315581926145,
287
+ "eval time": 17.115985165000893,
288
+ "tokens / sec": 4640.7342382101415,
289
+ "mem allocated avg": 6896364773.376,
290
+ "mem reserved avg": 17282084372.48,
291
+ "elapsed time": 1277.8335477730143
292
+ },
293
+ {
294
+ "step": 5000,
295
+ "valid accuracy": 0.46,
296
+ "train loss": 0.5765521181821823,
297
+ "train samples": 20000,
298
+ "train time": 44.84296027891105,
299
+ "eval time": 17.095471853972413,
300
+ "tokens / sec": 4644.653223260795,
301
+ "mem allocated avg": 6893230526.464,
302
+ "mem reserved avg": 16886980935.68,
303
+ "elapsed time": 1344.3281179019832
304
+ },
305
+ {
306
+ "step": 5000,
307
+ "test accuracy": 0.5064442759666414,
308
+ "train loss": 0.5765521181821823,
309
+ "train samples": 20000,
310
+ "train total tokens": 4198051,
311
+ "forgetting": 0.3088569641113281
312
+ }
313
+ ]
314
+ },
315
+ "meta_info": {
316
+ "model_info": {
317
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
318
+ "created_at": "2024-09-18T15:23:48+00:00"
319
+ },
320
+ "dataset_info": {
321
+ "metamath": {
322
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
323
+ "created_at": "2023-09-21T17:22:46+00:00"
324
+ },
325
+ "gsm8k": {
326
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
327
+ "created_at": "2022-04-12T10:22:10+00:00"
328
+ }
329
+ },
330
+ "package_info": {
331
+ "transformers-version": "4.57.1",
332
+ "transformers-commit-hash": null,
333
+ "peft-version": "0.18.1.dev0",
334
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
335
+ "datasets-version": "4.2.0",
336
+ "datasets-commit-hash": null,
337
+ "bitsandbytes-version": "0.46.0",
338
+ "bitsandbytes-commit-hash": null,
339
+ "torch-version": "2.9.0+cu128",
340
+ "torch-commit-hash": null
341
+ },
342
+ "system_info": {
343
+ "system": "Linux",
344
+ "release": "6.14.0-1016-aws",
345
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
346
+ "machine": "x86_64",
347
+ "processor": "x86_64",
348
+ "accelerator": "NVIDIA L40S"
349
+ },
350
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
351
+ }
352
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/bone--llama-3.2-3B-default.json ADDED
@@ -0,0 +1,352 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T21:06:39+00:00",
4
+ "total_time": 1113.975204212009,
5
+ "experiment_name": "bone/llama-3.2-3B-default",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "BONE",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "r": 64,
42
+ "target_modules": [
43
+ "q_proj",
44
+ "v_proj"
45
+ ],
46
+ "exclude_modules": null,
47
+ "init_weights": true,
48
+ "layers_to_transform": null,
49
+ "layers_pattern": null,
50
+ "bias": "none",
51
+ "modules_to_save": null
52
+ },
53
+ "error_msg": ""
54
+ },
55
+ "train_info": {
56
+ "accelerator_memory_reserved_avg": 13443749694,
57
+ "accelerator_memory_max": 20275265536,
58
+ "accelerator_memory_reserved_99th": 18498977792,
59
+ "train_time": 931.1479183980846,
60
+ "file_size": 29367496,
61
+ "num_trainable_params": 7340032,
62
+ "num_total_params": 3220089856,
63
+ "status": "success",
64
+ "metrics": [
65
+ {
66
+ "step": 250,
67
+ "valid accuracy": 0.36,
68
+ "train loss": 0.8770946979522705,
69
+ "train samples": 1000,
70
+ "train time": 28.682230734440964,
71
+ "eval time": 8.853944770002272,
72
+ "tokens / sec": 7381.538833580775,
73
+ "mem allocated avg": 6897714161.664,
74
+ "mem reserved avg": 13519852404.736,
75
+ "elapsed time": 59.13298718299484
76
+ },
77
+ {
78
+ "step": 500,
79
+ "valid accuracy": 0.4,
80
+ "train loss": 0.6949568554162979,
81
+ "train samples": 2000,
82
+ "train time": 28.45853918744251,
83
+ "eval time": 6.328033077006694,
84
+ "tokens / sec": 7308.702622788837,
85
+ "mem allocated avg": 6890609534.976,
86
+ "mem reserved avg": 13203794821.12,
87
+ "elapsed time": 96.60603074397659
88
+ },
89
+ {
90
+ "step": 750,
91
+ "valid accuracy": 0.44,
92
+ "train loss": 0.6667997059822083,
93
+ "train samples": 3000,
94
+ "train time": 29.355493688082788,
95
+ "eval time": 6.721548973000608,
96
+ "tokens / sec": 7303.607368287546,
97
+ "mem allocated avg": 6900360769.536,
98
+ "mem reserved avg": 13417939206.144,
99
+ "elapsed time": 135.35112743999343
100
+ },
101
+ {
102
+ "step": 1000,
103
+ "valid accuracy": 0.42,
104
+ "train loss": 0.6474654762744904,
105
+ "train samples": 4000,
106
+ "train time": 29.110337335790973,
107
+ "eval time": 7.787894255015999,
108
+ "tokens / sec": 7156.770380116902,
109
+ "mem allocated avg": 6891688468.48,
110
+ "mem reserved avg": 13453288800.256,
111
+ "elapsed time": 175.05626158800442
112
+ },
113
+ {
114
+ "step": 1250,
115
+ "valid accuracy": 0.4,
116
+ "train loss": 0.6438281389474869,
117
+ "train samples": 5000,
118
+ "train time": 28.465142201224808,
119
+ "eval time": 6.192574144050013,
120
+ "tokens / sec": 7326.0831976812315,
121
+ "mem allocated avg": 6892015517.696,
122
+ "mem reserved avg": 13493218574.336,
123
+ "elapsed time": 212.38757716899272
124
+ },
125
+ {
126
+ "step": 1500,
127
+ "valid accuracy": 0.5,
128
+ "train loss": 0.6369395072460174,
129
+ "train samples": 6000,
130
+ "train time": 28.41225368680898,
131
+ "eval time": 7.941400229989085,
132
+ "tokens / sec": 7367.630963297592,
133
+ "mem allocated avg": 6894076573.696,
134
+ "mem reserved avg": 13404207054.848,
135
+ "elapsed time": 251.46505238697864
136
+ },
137
+ {
138
+ "step": 1750,
139
+ "valid accuracy": 0.46,
140
+ "train loss": 0.6280212769508362,
141
+ "train samples": 7000,
142
+ "train time": 28.603689649200533,
143
+ "eval time": 11.7747625039774,
144
+ "tokens / sec": 7319.1606596057245,
145
+ "mem allocated avg": 6895286140.928,
146
+ "mem reserved avg": 13746437095.424,
147
+ "elapsed time": 294.6126683760085
148
+ },
149
+ {
150
+ "step": 2000,
151
+ "valid accuracy": 0.48,
152
+ "train loss": 0.6282699835300446,
153
+ "train samples": 8000,
154
+ "train time": 28.49381335376529,
155
+ "eval time": 6.387318613997195,
156
+ "tokens / sec": 7289.161244279513,
157
+ "mem allocated avg": 6891065575.424,
158
+ "mem reserved avg": 13428383023.104,
159
+ "elapsed time": 332.2536138329888
160
+ },
161
+ {
162
+ "step": 2250,
163
+ "valid accuracy": 0.44,
164
+ "train loss": 0.6164305354356766,
165
+ "train samples": 9000,
166
+ "train time": 29.641954794351477,
167
+ "eval time": 7.702494629949797,
168
+ "tokens / sec": 7251.478571209485,
169
+ "mem allocated avg": 6901582643.2,
170
+ "mem reserved avg": 13693672751.104,
171
+ "elapsed time": 372.30250247899676
172
+ },
173
+ {
174
+ "step": 2500,
175
+ "valid accuracy": 0.52,
176
+ "train loss": 0.6126096348762512,
177
+ "train samples": 10000,
178
+ "train time": 28.804422622197308,
179
+ "eval time": 9.238736133964267,
180
+ "tokens / sec": 7150.533885073516,
181
+ "mem allocated avg": 6888456951.808,
182
+ "mem reserved avg": 13231275900.928,
183
+ "elapsed time": 413.07279076898703
184
+ },
185
+ {
186
+ "step": 2750,
187
+ "valid accuracy": 0.46,
188
+ "train loss": 0.6011255930662155,
189
+ "train samples": 11000,
190
+ "train time": 28.807400498830248,
191
+ "eval time": 7.274331616004929,
192
+ "tokens / sec": 7355.089189967128,
193
+ "mem allocated avg": 6897535533.056,
194
+ "mem reserved avg": 13570335047.68,
195
+ "elapsed time": 451.849644197966
196
+ },
197
+ {
198
+ "step": 3000,
199
+ "valid accuracy": 0.5,
200
+ "train loss": 0.5905841100215912,
201
+ "train samples": 12000,
202
+ "train time": 28.54119487659773,
203
+ "eval time": 7.517624335014261,
204
+ "tokens / sec": 7313.323808007365,
205
+ "mem allocated avg": 6893574338.56,
206
+ "mem reserved avg": 13508544561.152,
207
+ "elapsed time": 490.6727877999656
208
+ },
209
+ {
210
+ "step": 3250,
211
+ "valid accuracy": 0.5,
212
+ "train loss": 0.5996895109415055,
213
+ "train samples": 13000,
214
+ "train time": 28.56189932534471,
215
+ "eval time": 11.817645783012267,
216
+ "tokens / sec": 7383.99773760335,
217
+ "mem allocated avg": 6896400535.552,
218
+ "mem reserved avg": 13368513527.808,
219
+ "elapsed time": 533.6943597999634
220
+ },
221
+ {
222
+ "step": 3500,
223
+ "valid accuracy": 0.48,
224
+ "train loss": 0.580611946105957,
225
+ "train samples": 14000,
226
+ "train time": 28.992627491941676,
227
+ "eval time": 6.298771312984172,
228
+ "tokens / sec": 7234.597832097099,
229
+ "mem allocated avg": 6893331273.728,
230
+ "mem reserved avg": 13393645797.376,
231
+ "elapsed time": 571.7360254359664
232
+ },
233
+ {
234
+ "step": 3750,
235
+ "valid accuracy": 0.46,
236
+ "train loss": 0.577835813999176,
237
+ "train samples": 15000,
238
+ "train time": 29.917891803139355,
239
+ "eval time": 8.922266290988773,
240
+ "tokens / sec": 7243.257694289169,
241
+ "mem allocated avg": 6904537260.032,
242
+ "mem reserved avg": 13701608374.272,
243
+ "elapsed time": 613.3215442419751
244
+ },
245
+ {
246
+ "step": 4000,
247
+ "valid accuracy": 0.46,
248
+ "train loss": 0.5866445735692978,
249
+ "train samples": 16000,
250
+ "train time": 28.4423630858073,
251
+ "eval time": 11.81282829795964,
252
+ "tokens / sec": 7185.514065179128,
253
+ "mem allocated avg": 6885427052.544,
254
+ "mem reserved avg": 13412377559.04,
255
+ "elapsed time": 656.3323608639766
256
+ },
257
+ {
258
+ "step": 4250,
259
+ "valid accuracy": 0.46,
260
+ "train loss": 0.5731088873147965,
261
+ "train samples": 17000,
262
+ "train time": 28.676882747968193,
263
+ "eval time": 11.7569903130061,
264
+ "tokens / sec": 7371.407898753475,
265
+ "mem allocated avg": 6897404037.12,
266
+ "mem reserved avg": 13391523479.552,
267
+ "elapsed time": 699.4078589889687
268
+ },
269
+ {
270
+ "step": 4500,
271
+ "valid accuracy": 0.42,
272
+ "train loss": 0.5799848650693893,
273
+ "train samples": 18000,
274
+ "train time": 28.678615707845893,
275
+ "eval time": 11.924760186986532,
276
+ "tokens / sec": 7246.444602385225,
277
+ "mem allocated avg": 6891705860.096,
278
+ "mem reserved avg": 13364461830.144,
279
+ "elapsed time": 742.7912658689893
280
+ },
281
+ {
282
+ "step": 4750,
283
+ "valid accuracy": 0.44,
284
+ "train loss": 0.5690632259845734,
285
+ "train samples": 19000,
286
+ "train time": 29.26655402261531,
287
+ "eval time": 11.802604599972256,
288
+ "tokens / sec": 7173.34196016971,
289
+ "mem allocated avg": 6894110355.456,
290
+ "mem reserved avg": 13432342446.08,
291
+ "elapsed time": 786.5075720959576
292
+ },
293
+ {
294
+ "step": 5000,
295
+ "valid accuracy": 0.42,
296
+ "train loss": 0.5775251245498657,
297
+ "train samples": 20000,
298
+ "train time": 28.85452529176837,
299
+ "eval time": 11.805582064029295,
300
+ "tokens / sec": 7218.278515897755,
301
+ "mem allocated avg": 6891173048.32,
302
+ "mem reserved avg": 13139571638.272,
303
+ "elapsed time": 829.8449935640092
304
+ },
305
+ {
306
+ "step": 5000,
307
+ "test accuracy": 0.49128127369219105,
308
+ "train loss": 0.5775251245498657,
309
+ "train samples": 20000,
310
+ "train total tokens": 4198051,
311
+ "forgetting": 0.2672872543334961
312
+ }
313
+ ]
314
+ },
315
+ "meta_info": {
316
+ "model_info": {
317
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
318
+ "created_at": "2024-09-18T15:23:48+00:00"
319
+ },
320
+ "dataset_info": {
321
+ "metamath": {
322
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
323
+ "created_at": "2023-09-21T17:22:46+00:00"
324
+ },
325
+ "gsm8k": {
326
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
327
+ "created_at": "2022-04-12T10:22:10+00:00"
328
+ }
329
+ },
330
+ "package_info": {
331
+ "transformers-version": "4.57.1",
332
+ "transformers-commit-hash": null,
333
+ "peft-version": "0.18.1.dev0",
334
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
335
+ "datasets-version": "4.2.0",
336
+ "datasets-commit-hash": null,
337
+ "bitsandbytes-version": "0.46.0",
338
+ "bitsandbytes-commit-hash": null,
339
+ "torch-version": "2.9.0+cu128",
340
+ "torch-commit-hash": null
341
+ },
342
+ "system_info": {
343
+ "system": "Linux",
344
+ "release": "6.14.0-1016-aws",
345
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
346
+ "machine": "x86_64",
347
+ "processor": "x86_64",
348
+ "accelerator": "NVIDIA L40S"
349
+ },
350
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
351
+ }
352
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/c3a--llama-3.2-3B-default.json ADDED
@@ -0,0 +1,352 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T21:25:18+00:00",
4
+ "total_time": 1366.5376071679639,
5
+ "experiment_name": "c3a/llama-3.2-3B-default",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.3,
22
+ "weight_decay": 1e-05
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "C3A",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "block_size": 64,
42
+ "target_modules": [
43
+ "v_proj",
44
+ "q_proj"
45
+ ],
46
+ "bias": "none",
47
+ "modules_to_save": null,
48
+ "layers_to_transform": null,
49
+ "layers_pattern": null,
50
+ "block_size_pattern": {},
51
+ "init_weights": false
52
+ },
53
+ "error_msg": ""
54
+ },
55
+ "train_info": {
56
+ "accelerator_memory_reserved_avg": 14377058251,
57
+ "accelerator_memory_max": 22166896640,
58
+ "accelerator_memory_reserved_99th": 20061356032,
59
+ "train_time": 1153.291345822101,
60
+ "file_size": 22027512,
61
+ "num_trainable_params": 5505024,
62
+ "num_total_params": 3218254848,
63
+ "status": "success",
64
+ "metrics": [
65
+ {
66
+ "step": 250,
67
+ "valid accuracy": 0.38,
68
+ "train loss": 0.8909596583843231,
69
+ "train samples": 1000,
70
+ "train time": 37.80949870531913,
71
+ "eval time": 11.432839135988615,
72
+ "tokens / sec": 5599.624624756394,
73
+ "mem allocated avg": 6869184182.272,
74
+ "mem reserved avg": 14482872991.744,
75
+ "elapsed time": 72.47078875597799
76
+ },
77
+ {
78
+ "step": 500,
79
+ "valid accuracy": 0.42,
80
+ "train loss": 0.7013351047039031,
81
+ "train samples": 2000,
82
+ "train time": 37.42838709271746,
83
+ "eval time": 8.331702384981327,
84
+ "tokens / sec": 5557.1456895739475,
85
+ "mem allocated avg": 6862181789.696,
86
+ "mem reserved avg": 14091510874.112,
87
+ "elapsed time": 121.3354702090146
88
+ },
89
+ {
90
+ "step": 750,
91
+ "valid accuracy": 0.46,
92
+ "train loss": 0.674084732890129,
93
+ "train samples": 3000,
94
+ "train time": 38.329594842216466,
95
+ "eval time": 9.41572032001568,
96
+ "tokens / sec": 5593.615087312568,
97
+ "mem allocated avg": 6872815554.56,
98
+ "mem reserved avg": 14353973641.216,
99
+ "elapsed time": 172.14090229501016
100
+ },
101
+ {
102
+ "step": 1000,
103
+ "valid accuracy": 0.44,
104
+ "train loss": 0.6545811468362808,
105
+ "train samples": 4000,
106
+ "train time": 37.05278595397249,
107
+ "eval time": 13.860594167024828,
108
+ "tokens / sec": 5622.681119276645,
109
+ "mem allocated avg": 6864588425.216,
110
+ "mem reserved avg": 14391965646.848,
111
+ "elapsed time": 226.19000285398215
112
+ },
113
+ {
114
+ "step": 1250,
115
+ "valid accuracy": 0.38,
116
+ "train loss": 0.6509714391231537,
117
+ "train samples": 5000,
118
+ "train time": 36.79460979660507,
119
+ "eval time": 8.696108918986283,
120
+ "tokens / sec": 5667.62363163425,
121
+ "mem allocated avg": 6863943450.624,
122
+ "mem reserved avg": 14405840404.48,
123
+ "elapsed time": 274.7266224689665
124
+ },
125
+ {
126
+ "step": 1500,
127
+ "valid accuracy": 0.52,
128
+ "train loss": 0.6443188463449478,
129
+ "train samples": 6000,
130
+ "train time": 37.157257698476315,
131
+ "eval time": 9.571486864006147,
132
+ "tokens / sec": 5633.650408183485,
133
+ "mem allocated avg": 6865824692.224,
134
+ "mem reserved avg": 14339352297.472,
135
+ "elapsed time": 324.5449085899745
136
+ },
137
+ {
138
+ "step": 1750,
139
+ "valid accuracy": 0.46,
140
+ "train loss": 0.6351410290002822,
141
+ "train samples": 7000,
142
+ "train time": 37.588308651640546,
143
+ "eval time": 13.673957422957756,
144
+ "tokens / sec": 5569.6839658376775,
145
+ "mem allocated avg": 6866604124.16,
146
+ "mem reserved avg": 14722149646.336,
147
+ "elapsed time": 378.88215018698247
148
+ },
149
+ {
150
+ "step": 2000,
151
+ "valid accuracy": 0.42,
152
+ "train loss": 0.6367183820009231,
153
+ "train samples": 8000,
154
+ "train time": 37.49864329502452,
155
+ "eval time": 9.148618383973371,
156
+ "tokens / sec": 5538.760385700621,
157
+ "mem allocated avg": 6862256517.12,
158
+ "mem reserved avg": 14371757490.176,
159
+ "elapsed time": 428.6968064419925
160
+ },
161
+ {
162
+ "step": 2250,
163
+ "valid accuracy": 0.4,
164
+ "train loss": 0.6258963789939881,
165
+ "train samples": 9000,
166
+ "train time": 38.2491970359697,
167
+ "eval time": 13.646375201002229,
168
+ "tokens / sec": 5619.673526685071,
169
+ "mem allocated avg": 6874371139.584,
170
+ "mem reserved avg": 14650997473.28,
171
+ "elapsed time": 483.6756669579772
172
+ },
173
+ {
174
+ "step": 2500,
175
+ "valid accuracy": 0.48,
176
+ "train loss": 0.6206584950685501,
177
+ "train samples": 10000,
178
+ "train time": 37.62900273769628,
179
+ "eval time": 13.673765863000881,
180
+ "tokens / sec": 5473.623668311165,
181
+ "mem allocated avg": 6859448377.344,
182
+ "mem reserved avg": 14140047360.0,
183
+ "elapsed time": 538.0801370120025
184
+ },
185
+ {
186
+ "step": 2750,
187
+ "valid accuracy": 0.5,
188
+ "train loss": 0.6105827242136002,
189
+ "train samples": 11000,
190
+ "train time": 37.365518155740574,
191
+ "eval time": 9.94396861596033,
192
+ "tokens / sec": 5670.495431560023,
193
+ "mem allocated avg": 6870005147.648,
194
+ "mem reserved avg": 14540318179.328,
195
+ "elapsed time": 588.4858504809672
196
+ },
197
+ {
198
+ "step": 3000,
199
+ "valid accuracy": 0.48,
200
+ "train loss": 0.5981686524152756,
201
+ "train samples": 12000,
202
+ "train time": 36.890319129568525,
203
+ "eval time": 9.116078838997055,
204
+ "tokens / sec": 5658.151106442905,
205
+ "mem allocated avg": 6865474818.048,
206
+ "mem reserved avg": 14431375327.232,
207
+ "elapsed time": 637.5967296099989
208
+ },
209
+ {
210
+ "step": 3250,
211
+ "valid accuracy": 0.58,
212
+ "train loss": 0.6065964109897614,
213
+ "train samples": 13000,
214
+ "train time": 36.91873880266212,
215
+ "eval time": 13.868325658957474,
216
+ "tokens / sec": 5712.573257914012,
217
+ "mem allocated avg": 6867266957.312,
218
+ "mem reserved avg": 14327809572.864,
219
+ "elapsed time": 691.4201774260146
220
+ },
221
+ {
222
+ "step": 3500,
223
+ "valid accuracy": 0.54,
224
+ "train loss": 0.5876260681152343,
225
+ "train samples": 14000,
226
+ "train time": 37.31881154741859,
227
+ "eval time": 13.723029766988475,
228
+ "tokens / sec": 5620.4898093682405,
229
+ "mem allocated avg": 6864456710.144,
230
+ "mem reserved avg": 14320821862.4,
231
+ "elapsed time": 745.5659840449807
232
+ },
233
+ {
234
+ "step": 3750,
235
+ "valid accuracy": 0.52,
236
+ "train loss": 0.5839207136631012,
237
+ "train samples": 15000,
238
+ "train time": 38.572791439597495,
239
+ "eval time": 8.601323624025099,
240
+ "tokens / sec": 5618.027420684421,
241
+ "mem allocated avg": 6877443866.624,
242
+ "mem reserved avg": 14659335749.632,
243
+ "elapsed time": 795.8486898440169
244
+ },
245
+ {
246
+ "step": 4000,
247
+ "valid accuracy": 0.56,
248
+ "train loss": 0.5916516542434692,
249
+ "train samples": 16000,
250
+ "train time": 36.93544478120748,
251
+ "eval time": 8.182845938019454,
252
+ "tokens / sec": 5533.248650737345,
253
+ "mem allocated avg": 6857094844.416,
254
+ "mem reserved avg": 14323103563.776,
255
+ "elapsed time": 844.0624020369723
256
+ },
257
+ {
258
+ "step": 4250,
259
+ "valid accuracy": 0.56,
260
+ "train loss": 0.5781249686479568,
261
+ "train samples": 17000,
262
+ "train time": 37.122223554353695,
263
+ "eval time": 9.616572241007816,
264
+ "tokens / sec": 5694.405662163205,
265
+ "mem allocated avg": 6869011593.216,
266
+ "mem reserved avg": 14312148041.728,
267
+ "elapsed time": 893.7695867909933
268
+ },
269
+ {
270
+ "step": 4500,
271
+ "valid accuracy": 0.56,
272
+ "train loss": 0.5846204442977905,
273
+ "train samples": 18000,
274
+ "train time": 37.07799867092399,
275
+ "eval time": 8.460158887959551,
276
+ "tokens / sec": 5604.887195893012,
277
+ "mem allocated avg": 6863795603.456,
278
+ "mem reserved avg": 14295052058.624,
279
+ "elapsed time": 942.4601713569718
280
+ },
281
+ {
282
+ "step": 4750,
283
+ "valid accuracy": 0.52,
284
+ "train loss": 0.5748073060512543,
285
+ "train samples": 19000,
286
+ "train time": 37.64343991095666,
287
+ "eval time": 8.636093506007455,
288
+ "tokens / sec": 5577.040793737192,
289
+ "mem allocated avg": 6865605898.24,
290
+ "mem reserved avg": 14353235443.712,
291
+ "elapsed time": 991.785048523976
292
+ },
293
+ {
294
+ "step": 5000,
295
+ "valid accuracy": 0.5,
296
+ "train loss": 0.5811544748544692,
297
+ "train samples": 20000,
298
+ "train time": 37.4456382485223,
299
+ "eval time": 8.431857075018343,
300
+ "tokens / sec": 5562.1965532452705,
301
+ "mem allocated avg": 6862668220.416,
302
+ "mem reserved avg": 14027497406.464,
303
+ "elapsed time": 1040.7068556279992
304
+ },
305
+ {
306
+ "step": 5000,
307
+ "test accuracy": 0.5026535253980288,
308
+ "train loss": 0.5811544748544692,
309
+ "train samples": 20000,
310
+ "train total tokens": 4198051,
311
+ "forgetting": 0.4295167922973633
312
+ }
313
+ ]
314
+ },
315
+ "meta_info": {
316
+ "model_info": {
317
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
318
+ "created_at": "2024-09-18T15:23:48+00:00"
319
+ },
320
+ "dataset_info": {
321
+ "metamath": {
322
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
323
+ "created_at": "2023-09-21T17:22:46+00:00"
324
+ },
325
+ "gsm8k": {
326
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
327
+ "created_at": "2022-04-12T10:22:10+00:00"
328
+ }
329
+ },
330
+ "package_info": {
331
+ "transformers-version": "4.57.1",
332
+ "transformers-commit-hash": null,
333
+ "peft-version": "0.18.1.dev0",
334
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
335
+ "datasets-version": "4.2.0",
336
+ "datasets-commit-hash": null,
337
+ "bitsandbytes-version": "0.46.0",
338
+ "bitsandbytes-commit-hash": null,
339
+ "torch-version": "2.9.0+cu128",
340
+ "torch-commit-hash": null
341
+ },
342
+ "system_info": {
343
+ "system": "Linux",
344
+ "release": "6.14.0-1016-aws",
345
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
346
+ "machine": "x86_64",
347
+ "processor": "x86_64",
348
+ "accelerator": "NVIDIA L40S"
349
+ },
350
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
351
+ }
352
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/delora--llama-3.2-3B-rank32.json ADDED
@@ -0,0 +1,355 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T21:48:09+00:00",
4
+ "total_time": 1277.7959856730304,
5
+ "experiment_name": "delora/llama-3.2-3B-rank32",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.001
22
+ },
23
+ "lr_scheduler": "cosine",
24
+ "use_amp": false,
25
+ "autocast_adapter_dtype": true,
26
+ "generation_kwargs": {
27
+ "max_length": 800,
28
+ "max_new_tokens": 300
29
+ },
30
+ "attn_implementation": null
31
+ },
32
+ "peft_config": {
33
+ "task_type": "CAUSAL_LM",
34
+ "peft_type": "DELORA",
35
+ "auto_mapping": null,
36
+ "peft_version": "0.18.1.dev0@UNKNOWN",
37
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
38
+ "revision": null,
39
+ "inference_mode": false,
40
+ "r": 32,
41
+ "delora_lambda": 15,
42
+ "module_dropout": 0.0,
43
+ "target_modules": [
44
+ "v_proj",
45
+ "q_proj"
46
+ ],
47
+ "exclude_modules": null,
48
+ "bias": "none",
49
+ "init_weights": true,
50
+ "layers_to_transform": null,
51
+ "layers_pattern": null,
52
+ "rank_pattern": {},
53
+ "lambda_pattern": {},
54
+ "modules_to_save": null
55
+ },
56
+ "error_msg": ""
57
+ },
58
+ "train_info": {
59
+ "accelerator_memory_reserved_avg": 14449004406,
60
+ "accelerator_memory_max": 22353543168,
61
+ "accelerator_memory_reserved_99th": 20126367744,
62
+ "train_time": 1058.4298028770136,
63
+ "file_size": 37417520,
64
+ "num_trainable_params": 9175096,
65
+ "num_total_params": 3221924920,
66
+ "status": "success",
67
+ "metrics": [
68
+ {
69
+ "step": 250,
70
+ "valid accuracy": 0.36,
71
+ "train loss": 0.7511726866960525,
72
+ "train samples": 1000,
73
+ "train time": 33.47864836757071,
74
+ "eval time": 13.131931535026524,
75
+ "tokens / sec": 6324.000828094447,
76
+ "mem allocated avg": 6927743098.88,
77
+ "mem reserved avg": 14590045847.552,
78
+ "elapsed time": 69.13025180797558
79
+ },
80
+ {
81
+ "step": 500,
82
+ "valid accuracy": 0.46,
83
+ "train loss": 0.7057771706581115,
84
+ "train samples": 2000,
85
+ "train time": 33.688454272574745,
86
+ "eval time": 13.079983315023128,
87
+ "tokens / sec": 6174.073714308867,
88
+ "mem allocated avg": 6921407948.8,
89
+ "mem reserved avg": 14192392273.92,
90
+ "elapsed time": 119.17276985396165
91
+ },
92
+ {
93
+ "step": 750,
94
+ "valid accuracy": 0.38,
95
+ "train loss": 0.6696414754390717,
96
+ "train samples": 3000,
97
+ "train time": 33.589053718489595,
98
+ "eval time": 11.759635296999477,
99
+ "tokens / sec": 6383.061630640097,
100
+ "mem allocated avg": 6930950647.808,
101
+ "mem reserved avg": 14421644541.952,
102
+ "elapsed time": 167.67709576297784
103
+ },
104
+ {
105
+ "step": 1000,
106
+ "valid accuracy": 0.36,
107
+ "train loss": 0.6491379799842835,
108
+ "train samples": 4000,
109
+ "train time": 33.53858558466891,
110
+ "eval time": 9.152838472975418,
111
+ "tokens / sec": 6211.830235775778,
112
+ "mem allocated avg": 6922103914.496,
113
+ "mem reserved avg": 14461423321.088,
114
+ "elapsed time": 213.7067174440017
115
+ },
116
+ {
117
+ "step": 1250,
118
+ "valid accuracy": 0.36,
119
+ "train loss": 0.6447645363807678,
120
+ "train samples": 5000,
121
+ "train time": 33.239838396781124,
122
+ "eval time": 13.123246687988285,
123
+ "tokens / sec": 6273.736878943261,
124
+ "mem allocated avg": 6922382727.168,
125
+ "mem reserved avg": 14487830659.072,
126
+ "elapsed time": 263.2440371370176
127
+ },
128
+ {
129
+ "step": 1500,
130
+ "valid accuracy": 0.46,
131
+ "train loss": 0.6379029039144516,
132
+ "train samples": 6000,
133
+ "train time": 33.83518080448266,
134
+ "eval time": 9.2000446600141,
135
+ "tokens / sec": 6186.78532293425,
136
+ "mem allocated avg": 6924438607.872,
137
+ "mem reserved avg": 14402577235.968,
138
+ "elapsed time": 309.58903181599453
139
+ },
140
+ {
141
+ "step": 1750,
142
+ "valid accuracy": 0.5,
143
+ "train loss": 0.6294291808605194,
144
+ "train samples": 7000,
145
+ "train time": 33.638763133611064,
146
+ "eval time": 13.089110154949594,
147
+ "tokens / sec": 6223.623596636268,
148
+ "mem allocated avg": 6925751805.952,
149
+ "mem reserved avg": 14803384926.208,
150
+ "elapsed time": 359.6268992089899
151
+ },
152
+ {
153
+ "step": 2000,
154
+ "valid accuracy": 0.42,
155
+ "train loss": 0.6335482016801834,
156
+ "train samples": 8000,
157
+ "train time": 33.628849623783026,
158
+ "eval time": 9.815332075988408,
159
+ "tokens / sec": 6176.125627952288,
160
+ "mem allocated avg": 6921476399.104,
161
+ "mem reserved avg": 14434068070.4,
162
+ "elapsed time": 406.38412045896985
163
+ },
164
+ {
165
+ "step": 2250,
166
+ "valid accuracy": 0.42,
167
+ "train loss": 0.6242709509134292,
168
+ "train samples": 9000,
169
+ "train time": 33.97552239638753,
170
+ "eval time": 13.092339784023352,
171
+ "tokens / sec": 6326.55467345675,
172
+ "mem allocated avg": 6932143722.496,
173
+ "mem reserved avg": 14720891355.136,
174
+ "elapsed time": 456.67317382100737
175
+ },
176
+ {
177
+ "step": 2500,
178
+ "valid accuracy": 0.46,
179
+ "train loss": 0.6215527185201645,
180
+ "train samples": 10000,
181
+ "train time": 33.24377211777028,
182
+ "eval time": 13.087968986015767,
183
+ "tokens / sec": 6195.656716401972,
184
+ "mem allocated avg": 6919181907.968,
185
+ "mem reserved avg": 14204631252.992,
186
+ "elapsed time": 506.2042880220106
187
+ },
188
+ {
189
+ "step": 2750,
190
+ "valid accuracy": 0.44,
191
+ "train loss": 0.6135006090402603,
192
+ "train samples": 11000,
193
+ "train time": 33.8344249399961,
194
+ "eval time": 11.559464724035934,
195
+ "tokens / sec": 6262.290562814704,
196
+ "mem allocated avg": 6928519376.896,
197
+ "mem reserved avg": 14587529265.152,
198
+ "elapsed time": 554.7988454389852
199
+ },
200
+ {
201
+ "step": 3000,
202
+ "valid accuracy": 0.5,
203
+ "train loss": 0.6046421718597412,
204
+ "train samples": 12000,
205
+ "train time": 33.518839641998056,
206
+ "eval time": 13.07255553000141,
207
+ "tokens / sec": 6227.274041386164,
208
+ "mem allocated avg": 6924324909.056,
209
+ "mem reserved avg": 14517383725.056,
210
+ "elapsed time": 604.7252621700172
211
+ },
212
+ {
213
+ "step": 3250,
214
+ "valid accuracy": 0.5,
215
+ "train loss": 0.6123787004947663,
216
+ "train samples": 13000,
217
+ "train time": 33.85599415731849,
218
+ "eval time": 12.407609467976727,
219
+ "tokens / sec": 6229.354808487008,
220
+ "mem allocated avg": 6927248498.688,
221
+ "mem reserved avg": 14381471498.24,
222
+ "elapsed time": 654.1598190389923
223
+ },
224
+ {
225
+ "step": 3500,
226
+ "valid accuracy": 0.46,
227
+ "train loss": 0.5954382091760635,
228
+ "train samples": 14000,
229
+ "train time": 33.584031070175115,
230
+ "eval time": 8.71236887795385,
231
+ "tokens / sec": 6245.52780938415,
232
+ "mem allocated avg": 6924072755.2,
233
+ "mem reserved avg": 14376438333.44,
234
+ "elapsed time": 699.7364873759798
235
+ },
236
+ {
237
+ "step": 3750,
238
+ "valid accuracy": 0.52,
239
+ "train loss": 0.5923196750879288,
240
+ "train samples": 15000,
241
+ "train time": 34.014798398071434,
242
+ "eval time": 8.587440980016254,
243
+ "tokens / sec": 6370.844755977933,
244
+ "mem allocated avg": 6935308263.424,
245
+ "mem reserved avg": 14727753236.48,
246
+ "elapsed time": 745.567757465993
247
+ },
248
+ {
249
+ "step": 4000,
250
+ "valid accuracy": 0.5,
251
+ "train loss": 0.6026594350337983,
252
+ "train samples": 16000,
253
+ "train time": 33.61519722332014,
254
+ "eval time": 10.390929970017169,
255
+ "tokens / sec": 6079.779887717532,
256
+ "mem allocated avg": 6916062009.344,
257
+ "mem reserved avg": 14388383711.232,
258
+ "elapsed time": 792.880163897993
259
+ },
260
+ {
261
+ "step": 4250,
262
+ "valid accuracy": 0.46,
263
+ "train loss": 0.5877031481266022,
264
+ "train samples": 17000,
265
+ "train time": 33.592281938006636,
266
+ "eval time": 8.453822323994245,
267
+ "tokens / sec": 6292.784764968063,
268
+ "mem allocated avg": 6928201375.744,
269
+ "mem reserved avg": 14382436188.16,
270
+ "elapsed time": 838.0397840199876
271
+ },
272
+ {
273
+ "step": 4500,
274
+ "valid accuracy": 0.5,
275
+ "train loss": 0.5929731521606445,
276
+ "train samples": 18000,
277
+ "train time": 33.91169841529336,
278
+ "eval time": 8.1252419599914,
279
+ "tokens / sec": 6128.20972441413,
280
+ "mem allocated avg": 6921858213.888,
281
+ "mem reserved avg": 14347757682.688,
282
+ "elapsed time": 883.4076476009795
283
+ },
284
+ {
285
+ "step": 4750,
286
+ "valid accuracy": 0.52,
287
+ "train loss": 0.583827891588211,
288
+ "train samples": 19000,
289
+ "train time": 33.430834216298535,
290
+ "eval time": 8.106015269993804,
291
+ "tokens / sec": 6279.801414517154,
292
+ "mem allocated avg": 6925002395.648,
293
+ "mem reserved avg": 14434042904.576,
294
+ "elapsed time": 928.1137036280124
295
+ },
296
+ {
297
+ "step": 5000,
298
+ "valid accuracy": 0.5,
299
+ "train loss": 0.591714503645897,
300
+ "train samples": 20000,
301
+ "train time": 33.46077763434732,
302
+ "eval time": 8.106805060000625,
303
+ "tokens / sec": 6224.601301142554,
304
+ "mem allocated avg": 6921049147.392,
305
+ "mem reserved avg": 14118002098.176,
306
+ "elapsed time": 972.8581771000172
307
+ },
308
+ {
309
+ "step": 5000,
310
+ "test accuracy": 0.5026535253980288,
311
+ "train loss": 0.591714503645897,
312
+ "train samples": 20000,
313
+ "train total tokens": 4198051,
314
+ "forgetting": 0.515233039855957
315
+ }
316
+ ]
317
+ },
318
+ "meta_info": {
319
+ "model_info": {
320
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
321
+ "created_at": "2024-09-18T15:23:48+00:00"
322
+ },
323
+ "dataset_info": {
324
+ "metamath": {
325
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
326
+ "created_at": "2023-09-21T17:22:46+00:00"
327
+ },
328
+ "gsm8k": {
329
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
330
+ "created_at": "2022-04-12T10:22:10+00:00"
331
+ }
332
+ },
333
+ "package_info": {
334
+ "transformers-version": "4.57.1",
335
+ "transformers-commit-hash": null,
336
+ "peft-version": "0.18.1.dev0",
337
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
338
+ "datasets-version": "4.2.0",
339
+ "datasets-commit-hash": null,
340
+ "bitsandbytes-version": "0.46.0",
341
+ "bitsandbytes-commit-hash": null,
342
+ "torch-version": "2.9.0+cu128",
343
+ "torch-commit-hash": null
344
+ },
345
+ "system_info": {
346
+ "system": "Linux",
347
+ "release": "6.14.0-1016-aws",
348
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
349
+ "machine": "x86_64",
350
+ "processor": "x86_64",
351
+ "accelerator": "NVIDIA L40S"
352
+ },
353
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
354
+ }
355
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/fourierft--llama-3.2-3B-default.json ADDED
@@ -0,0 +1,356 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T22:09:31+00:00",
4
+ "total_time": 2139.03748531203,
5
+ "experiment_name": "fourierft/llama-3.2-3B-default",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "FOURIERFT",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "n_frequency": 1000,
42
+ "scaling": 300,
43
+ "random_loc_seed": 777,
44
+ "fan_in_fan_out": false,
45
+ "target_modules": [
46
+ "q_proj",
47
+ "v_proj"
48
+ ],
49
+ "exclude_modules": null,
50
+ "bias": "none",
51
+ "modules_to_save": null,
52
+ "layers_to_transform": null,
53
+ "layers_pattern": null,
54
+ "n_frequency_pattern": {},
55
+ "init_weights": false
56
+ },
57
+ "error_msg": ""
58
+ },
59
+ "train_info": {
60
+ "accelerator_memory_reserved_avg": 15762228825,
61
+ "accelerator_memory_max": 23651680256,
62
+ "accelerator_memory_reserved_99th": 21596471296,
63
+ "train_time": 1726.894935239805,
64
+ "file_size": 231416,
65
+ "num_trainable_params": 56000,
66
+ "num_total_params": 3212805824,
67
+ "status": "success",
68
+ "metrics": [
69
+ {
70
+ "step": 250,
71
+ "valid accuracy": 0.0,
72
+ "train loss": 1.3261459288597106,
73
+ "train samples": 1000,
74
+ "train time": 54.26436190091772,
75
+ "eval time": 20.152891070989426,
76
+ "tokens / sec": 3901.62148016375,
77
+ "mem allocated avg": 6780939091.968,
78
+ "mem reserved avg": 15867958001.664,
79
+ "elapsed time": 99.62701734097209
80
+ },
81
+ {
82
+ "step": 500,
83
+ "valid accuracy": 0.0,
84
+ "train loss": 1.3400139141082763,
85
+ "train samples": 2000,
86
+ "train time": 53.535696760285646,
87
+ "eval time": 20.12735749600688,
88
+ "tokens / sec": 3885.16471414073,
89
+ "mem allocated avg": 6773629601.792,
90
+ "mem reserved avg": 15486259560.448,
91
+ "elapsed time": 176.75905558397062
92
+ },
93
+ {
94
+ "step": 750,
95
+ "valid accuracy": 0.0,
96
+ "train loss": 1.3044390759468079,
97
+ "train samples": 3000,
98
+ "train time": 53.646198650298174,
99
+ "eval time": 20.156346999981906,
100
+ "tokens / sec": 3996.573949211373,
101
+ "mem allocated avg": 6783680942.08,
102
+ "mem reserved avg": 15731836059.648,
103
+ "elapsed time": 253.90234332997352
104
+ },
105
+ {
106
+ "step": 1000,
107
+ "valid accuracy": 0.0,
108
+ "train loss": 1.311077395439148,
109
+ "train samples": 4000,
110
+ "train time": 52.92571030347608,
111
+ "eval time": 20.228597007051576,
112
+ "tokens / sec": 3936.3855261535678,
113
+ "mem allocated avg": 6776640890.88,
114
+ "mem reserved avg": 15724596690.944,
115
+ "elapsed time": 330.55366935697384
116
+ },
117
+ {
118
+ "step": 1250,
119
+ "valid accuracy": 0.0,
120
+ "train loss": 1.2998026401996612,
121
+ "train samples": 5000,
122
+ "train time": 53.52640636928845,
123
+ "eval time": 20.14151435403619,
124
+ "tokens / sec": 3895.9835741868837,
125
+ "mem allocated avg": 6775686537.216,
126
+ "mem reserved avg": 15805932634.112,
127
+ "elapsed time": 407.61837184696924
128
+ },
129
+ {
130
+ "step": 1500,
131
+ "valid accuracy": 0.0,
132
+ "train loss": 1.2868368306159974,
133
+ "train samples": 6000,
134
+ "train time": 53.88665995717747,
135
+ "eval time": 20.14792396198027,
136
+ "tokens / sec": 3884.6534590629794,
137
+ "mem allocated avg": 6777282525.184,
138
+ "mem reserved avg": 15735241834.496,
139
+ "elapsed time": 485.17054163996363
140
+ },
141
+ {
142
+ "step": 1750,
143
+ "valid accuracy": 0.0,
144
+ "train loss": 1.2805588779449464,
145
+ "train samples": 7000,
146
+ "train time": 53.43724203767488,
147
+ "eval time": 20.139051661011763,
148
+ "tokens / sec": 3917.773298487193,
149
+ "mem allocated avg": 6777839691.776,
150
+ "mem reserved avg": 16100456660.992,
151
+ "elapsed time": 562.2017202629941
152
+ },
153
+ {
154
+ "step": 2000,
155
+ "valid accuracy": 0.0,
156
+ "train loss": 1.2766080429553985,
157
+ "train samples": 8000,
158
+ "train time": 53.74037380289519,
159
+ "eval time": 20.10542684403481,
160
+ "tokens / sec": 3864.8037834975135,
161
+ "mem allocated avg": 6774405505.024,
162
+ "mem reserved avg": 15730795872.256,
163
+ "elapsed time": 639.6445371389855
164
+ },
165
+ {
166
+ "step": 2250,
167
+ "valid accuracy": 0.0,
168
+ "train loss": 1.2596942813396454,
169
+ "train samples": 9000,
170
+ "train time": 55.60716101707658,
171
+ "eval time": 20.199656157987192,
172
+ "tokens / sec": 3865.473368330941,
173
+ "mem allocated avg": 6785559302.144,
174
+ "mem reserved avg": 16046719238.144,
175
+ "elapsed time": 718.9049943919526
176
+ },
177
+ {
178
+ "step": 2500,
179
+ "valid accuracy": 0.0,
180
+ "train loss": 1.2683517651557923,
181
+ "train samples": 10000,
182
+ "train time": 53.73305493139196,
183
+ "eval time": 20.08102407801198,
184
+ "tokens / sec": 3833.152614586777,
185
+ "mem allocated avg": 6771125086.208,
186
+ "mem reserved avg": 15553687191.552,
187
+ "elapsed time": 796.1963960899739
188
+ },
189
+ {
190
+ "step": 2750,
191
+ "valid accuracy": 0.0,
192
+ "train loss": 1.254921770811081,
193
+ "train samples": 11000,
194
+ "train time": 55.10060942819109,
195
+ "eval time": 20.137201737961732,
196
+ "tokens / sec": 3845.347668543126,
197
+ "mem allocated avg": 6781653977.088,
198
+ "mem reserved avg": 15920974004.224,
199
+ "elapsed time": 874.8866917559644
200
+ },
201
+ {
202
+ "step": 3000,
203
+ "valid accuracy": 0.0,
204
+ "train loss": 1.2537723093032838,
205
+ "train samples": 12000,
206
+ "train time": 53.60143480350962,
207
+ "eval time": 20.126665371004492,
208
+ "tokens / sec": 3894.130833720389,
209
+ "mem allocated avg": 6776695320.576,
210
+ "mem reserved avg": 15812299587.584,
211
+ "elapsed time": 952.1623955519754
212
+ },
213
+ {
214
+ "step": 3250,
215
+ "valid accuracy": 0.0,
216
+ "train loss": 1.2485535402297974,
217
+ "train samples": 13000,
218
+ "train time": 54.74126313335728,
219
+ "eval time": 20.236808118002955,
220
+ "tokens / sec": 3852.6878615536516,
221
+ "mem allocated avg": 6778833948.672,
222
+ "mem reserved avg": 15685539332.096,
223
+ "elapsed time": 1030.4662453850033
224
+ },
225
+ {
226
+ "step": 3500,
227
+ "valid accuracy": 0.0,
228
+ "train loss": 1.247687777519226,
229
+ "train samples": 14000,
230
+ "train time": 53.67239992291434,
231
+ "eval time": 20.154386248032097,
232
+ "tokens / sec": 3907.9676016210988,
233
+ "mem allocated avg": 6776816490.496,
234
+ "mem reserved avg": 15717592203.264,
235
+ "elapsed time": 1107.7825586889521
236
+ },
237
+ {
238
+ "step": 3750,
239
+ "valid accuracy": 0.0,
240
+ "train loss": 1.2286142303943635,
241
+ "train samples": 15000,
242
+ "train time": 55.43350543506676,
243
+ "eval time": 20.094228011963423,
244
+ "tokens / sec": 3909.2422227174457,
245
+ "mem allocated avg": 6788307730.432,
246
+ "mem reserved avg": 16017065508.864,
247
+ "elapsed time": 1186.77993246197
248
+ },
249
+ {
250
+ "step": 4000,
251
+ "valid accuracy": 0.0,
252
+ "train loss": 1.2608506062030793,
253
+ "train samples": 16000,
254
+ "train time": 53.14804647688288,
255
+ "eval time": 20.161071109992918,
256
+ "tokens / sec": 3845.353000676958,
257
+ "mem allocated avg": 6769206956.032,
258
+ "mem reserved avg": 15759719792.64,
259
+ "elapsed time": 1263.6211953659658
260
+ },
261
+ {
262
+ "step": 4250,
263
+ "valid accuracy": 0.0,
264
+ "train loss": 1.2272750456333161,
265
+ "train samples": 17000,
266
+ "train time": 54.87057439738419,
267
+ "eval time": 20.140905109001324,
268
+ "tokens / sec": 3852.502043610417,
269
+ "mem allocated avg": 6780253456.384,
270
+ "mem reserved avg": 15731970277.376,
271
+ "elapsed time": 1341.9455141729559
272
+ },
273
+ {
274
+ "step": 4500,
275
+ "valid accuracy": 0.0,
276
+ "train loss": 1.2504476809501648,
277
+ "train samples": 18000,
278
+ "train time": 53.83008128858637,
279
+ "eval time": 20.128594986046664,
280
+ "tokens / sec": 3860.629503527497,
281
+ "mem allocated avg": 6775227097.088,
282
+ "mem reserved avg": 15675707883.52,
283
+ "elapsed time": 1419.423019112961
284
+ },
285
+ {
286
+ "step": 4750,
287
+ "valid accuracy": 0.0,
288
+ "train loss": 1.234989553451538,
289
+ "train samples": 19000,
290
+ "train time": 53.29879065020941,
291
+ "eval time": 20.26173467701301,
292
+ "tokens / sec": 3938.9073830547627,
293
+ "mem allocated avg": 6777659713.536,
294
+ "mem reserved avg": 15754426580.992,
295
+ "elapsed time": 1496.3555859409971
296
+ },
297
+ {
298
+ "step": 5000,
299
+ "valid accuracy": 0.0,
300
+ "train loss": 1.2481540043354034,
301
+ "train samples": 20000,
302
+ "train time": 53.355687880772166,
303
+ "eval time": 20.13233814504929,
304
+ "tokens / sec": 3903.6138089985725,
305
+ "mem allocated avg": 6774096742.4,
306
+ "mem reserved avg": 15385797591.04,
307
+ "elapsed time": 1573.2050500559853
308
+ },
309
+ {
310
+ "step": 5000,
311
+ "test accuracy": 0.000758150113722517,
312
+ "train loss": 1.2481540043354034,
313
+ "train samples": 20000,
314
+ "train total tokens": 4198051,
315
+ "forgetting": -0.012529373168945312
316
+ }
317
+ ]
318
+ },
319
+ "meta_info": {
320
+ "model_info": {
321
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
322
+ "created_at": "2024-09-18T15:23:48+00:00"
323
+ },
324
+ "dataset_info": {
325
+ "metamath": {
326
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
327
+ "created_at": "2023-09-21T17:22:46+00:00"
328
+ },
329
+ "gsm8k": {
330
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
331
+ "created_at": "2022-04-12T10:22:10+00:00"
332
+ }
333
+ },
334
+ "package_info": {
335
+ "transformers-version": "4.57.1",
336
+ "transformers-commit-hash": null,
337
+ "peft-version": "0.18.1.dev0",
338
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
339
+ "datasets-version": "4.2.0",
340
+ "datasets-commit-hash": null,
341
+ "bitsandbytes-version": "0.46.0",
342
+ "bitsandbytes-commit-hash": null,
343
+ "torch-version": "2.9.0+cu128",
344
+ "torch-commit-hash": null
345
+ },
346
+ "system_info": {
347
+ "system": "Linux",
348
+ "release": "6.14.0-1016-aws",
349
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
350
+ "machine": "x86_64",
351
+ "processor": "x86_64",
352
+ "accelerator": "NVIDIA L40S"
353
+ },
354
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
355
+ }
356
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/fourierft--llama-3.2-3B-n_frequency-5000.json ADDED
@@ -0,0 +1,356 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T22:45:14+00:00",
4
+ "total_time": 2150.229806937976,
5
+ "experiment_name": "fourierft/llama-3.2-3B-n_frequency-5000",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "FOURIERFT",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "n_frequency": 5000,
42
+ "scaling": 300,
43
+ "random_loc_seed": 777,
44
+ "fan_in_fan_out": false,
45
+ "target_modules": [
46
+ "v_proj",
47
+ "q_proj"
48
+ ],
49
+ "exclude_modules": null,
50
+ "bias": "none",
51
+ "modules_to_save": null,
52
+ "layers_to_transform": null,
53
+ "layers_pattern": null,
54
+ "n_frequency_pattern": {},
55
+ "init_weights": false
56
+ },
57
+ "error_msg": ""
58
+ },
59
+ "train_info": {
60
+ "accelerator_memory_reserved_avg": 15765951270,
61
+ "accelerator_memory_max": 23674748928,
62
+ "accelerator_memory_reserved_99th": 21602762752,
63
+ "train_time": 1735.5117494451115,
64
+ "file_size": 1127472,
65
+ "num_trainable_params": 280000,
66
+ "num_total_params": 3213029824,
67
+ "status": "success",
68
+ "metrics": [
69
+ {
70
+ "step": 250,
71
+ "valid accuracy": 0.0,
72
+ "train loss": 1.3799961261749267,
73
+ "train samples": 1000,
74
+ "train time": 54.63538205990335,
75
+ "eval time": 20.25719825399574,
76
+ "tokens / sec": 3875.126191446176,
77
+ "mem allocated avg": 6784255385.6,
78
+ "mem reserved avg": 15879072907.264,
79
+ "elapsed time": 99.3786606270005
80
+ },
81
+ {
82
+ "step": 500,
83
+ "valid accuracy": 0.0,
84
+ "train loss": 1.370254718542099,
85
+ "train samples": 2000,
86
+ "train time": 54.05832158756675,
87
+ "eval time": 20.27056284097489,
88
+ "tokens / sec": 3847.6037341092406,
89
+ "mem allocated avg": 6777261156.352,
90
+ "mem reserved avg": 15501593935.872,
91
+ "elapsed time": 177.16174678102834
92
+ },
93
+ {
94
+ "step": 750,
95
+ "valid accuracy": 0.0,
96
+ "train loss": 1.3024464647769929,
97
+ "train samples": 3000,
98
+ "train time": 54.38698652503081,
99
+ "eval time": 20.305425599042792,
100
+ "tokens / sec": 3942.1378844243395,
101
+ "mem allocated avg": 6787271151.616,
102
+ "mem reserved avg": 15726962278.4,
103
+ "elapsed time": 255.17251890699845
104
+ },
105
+ {
106
+ "step": 1000,
107
+ "valid accuracy": 0.0,
108
+ "train loss": 1.2703465101718903,
109
+ "train samples": 4000,
110
+ "train time": 54.011514401237946,
111
+ "eval time": 20.39393692498561,
112
+ "tokens / sec": 3857.251593656943,
113
+ "mem allocated avg": 6780041570.304,
114
+ "mem reserved avg": 15735644487.68,
115
+ "elapsed time": 333.0673065180308
116
+ },
117
+ {
118
+ "step": 1250,
119
+ "valid accuracy": 0.0,
120
+ "train loss": 1.2236051378250121,
121
+ "train samples": 5000,
122
+ "train time": 54.131226874596905,
123
+ "eval time": 20.295763457019348,
124
+ "tokens / sec": 3852.452863909948,
125
+ "mem allocated avg": 6779091253.248,
126
+ "mem reserved avg": 15815420149.76,
127
+ "elapsed time": 410.88266454002587
128
+ },
129
+ {
130
+ "step": 1500,
131
+ "valid accuracy": 0.0,
132
+ "train loss": 1.1791759724617004,
133
+ "train samples": 6000,
134
+ "train time": 54.805071563867386,
135
+ "eval time": 20.344787555979565,
136
+ "tokens / sec": 3819.5552715601325,
137
+ "mem allocated avg": 6781132302.336,
138
+ "mem reserved avg": 15728438673.408,
139
+ "elapsed time": 489.5477928170003
140
+ },
141
+ {
142
+ "step": 1750,
143
+ "valid accuracy": 0.0,
144
+ "train loss": 1.1484073193073272,
145
+ "train samples": 7000,
146
+ "train time": 54.526441418449394,
147
+ "eval time": 20.204474157013465,
148
+ "tokens / sec": 3839.5133545092,
149
+ "mem allocated avg": 6781927921.664,
150
+ "mem reserved avg": 16101270355.968,
151
+ "elapsed time": 567.73078815802
152
+ },
153
+ {
154
+ "step": 2000,
155
+ "valid accuracy": 0.06,
156
+ "train loss": 1.1230360004901887,
157
+ "train samples": 8000,
158
+ "train time": 53.57974702503998,
159
+ "eval time": 20.231831898970995,
160
+ "tokens / sec": 3876.390082673128,
161
+ "mem allocated avg": 6778186371.072,
162
+ "mem reserved avg": 15751582842.88,
163
+ "elapsed time": 645.0985403700033
164
+ },
165
+ {
166
+ "step": 2250,
167
+ "valid accuracy": 0.06,
168
+ "train loss": 1.0939874126911164,
169
+ "train samples": 9000,
170
+ "train time": 55.8729160525254,
171
+ "eval time": 20.28473745501833,
172
+ "tokens / sec": 3847.087554870596,
173
+ "mem allocated avg": 6789467377.664,
174
+ "mem reserved avg": 16052515766.272,
175
+ "elapsed time": 724.6561014120234
176
+ },
177
+ {
178
+ "step": 2500,
179
+ "valid accuracy": 0.08,
180
+ "train loss": 1.077495297908783,
181
+ "train samples": 10000,
182
+ "train time": 52.74528944323538,
183
+ "eval time": 20.17585710203275,
184
+ "tokens / sec": 3904.93638719458,
185
+ "mem allocated avg": 6774599432.192,
186
+ "mem reserved avg": 15540391247.872,
187
+ "elapsed time": 801.0244456710061
188
+ },
189
+ {
190
+ "step": 2750,
191
+ "valid accuracy": 0.08,
192
+ "train loss": 1.0567093884944916,
193
+ "train samples": 11000,
194
+ "train time": 53.79716244636802,
195
+ "eval time": 20.35781637695618,
196
+ "tokens / sec": 3938.516277902769,
197
+ "mem allocated avg": 6785299869.696,
198
+ "mem reserved avg": 15918188986.368,
199
+ "elapsed time": 878.562464024988
200
+ },
201
+ {
202
+ "step": 3000,
203
+ "valid accuracy": 0.12,
204
+ "train loss": 1.0359035170078277,
205
+ "train samples": 12000,
206
+ "train time": 53.633092043863144,
207
+ "eval time": 20.22181941999588,
208
+ "tokens / sec": 3891.8323006492337,
209
+ "mem allocated avg": 6780504604.672,
210
+ "mem reserved avg": 15821015351.296,
211
+ "elapsed time": 955.9252954589901
212
+ },
213
+ {
214
+ "step": 3250,
215
+ "valid accuracy": 0.1,
216
+ "train loss": 1.0238903017044068,
217
+ "train samples": 13000,
218
+ "train time": 54.95394577109255,
219
+ "eval time": 20.251423412002623,
220
+ "tokens / sec": 3837.7771976282793,
221
+ "mem allocated avg": 6782707699.712,
222
+ "mem reserved avg": 15711476908.032,
223
+ "elapsed time": 1034.4340154950041
224
+ },
225
+ {
226
+ "step": 3500,
227
+ "valid accuracy": 0.22,
228
+ "train loss": 1.0096496410369873,
229
+ "train samples": 14000,
230
+ "train time": 54.10011919972021,
231
+ "eval time": 20.23786943597952,
232
+ "tokens / sec": 3877.0709400042274,
233
+ "mem allocated avg": 6780696848.384,
234
+ "mem reserved avg": 15717843861.504,
235
+ "elapsed time": 1112.2632181590307
236
+ },
237
+ {
238
+ "step": 3750,
239
+ "valid accuracy": 0.26,
240
+ "train loss": 0.9940191543102265,
241
+ "train samples": 15000,
242
+ "train time": 55.7427657317603,
243
+ "eval time": 20.346782680018805,
244
+ "tokens / sec": 3887.5537866706554,
245
+ "mem allocated avg": 6792103485.44,
246
+ "mem reserved avg": 16009381543.936,
247
+ "elapsed time": 1191.7781133620301
248
+ },
249
+ {
250
+ "step": 4000,
251
+ "valid accuracy": 0.16,
252
+ "train loss": 1.0120926787853242,
253
+ "train samples": 16000,
254
+ "train time": 53.314123046409804,
255
+ "eval time": 20.31927741097752,
256
+ "tokens / sec": 3833.3745042020823,
257
+ "mem allocated avg": 6772785229.824,
258
+ "mem reserved avg": 15762924240.896,
259
+ "elapsed time": 1268.9340723860078
260
+ },
261
+ {
262
+ "step": 4250,
263
+ "valid accuracy": 0.16,
264
+ "train loss": 0.9847111780643463,
265
+ "train samples": 17000,
266
+ "train time": 54.80445172073087,
267
+ "eval time": 20.269200723967515,
268
+ "tokens / sec": 3857.1501650483247,
269
+ "mem allocated avg": 6784316258.304,
270
+ "mem reserved avg": 15734939844.608,
271
+ "elapsed time": 1347.3028064740356
272
+ },
273
+ {
274
+ "step": 4500,
275
+ "valid accuracy": 0.16,
276
+ "train loss": 0.9993657054901123,
277
+ "train samples": 18000,
278
+ "train time": 54.06312520604115,
279
+ "eval time": 20.218144676997326,
280
+ "tokens / sec": 3843.9879161254607,
281
+ "mem allocated avg": 6778886414.336,
282
+ "mem reserved avg": 15662697152.512,
283
+ "elapsed time": 1425.0894955290132
284
+ },
285
+ {
286
+ "step": 4750,
287
+ "valid accuracy": 0.2,
288
+ "train loss": 0.9890022552013398,
289
+ "train samples": 19000,
290
+ "train time": 54.60328221245436,
291
+ "eval time": 20.21068259398453,
292
+ "tokens / sec": 3844.8055042397323,
293
+ "mem allocated avg": 6781246834.688,
294
+ "mem reserved avg": 15759325528.064,
295
+ "elapsed time": 1503.2603250180255
296
+ },
297
+ {
298
+ "step": 5000,
299
+ "valid accuracy": 0.24,
300
+ "train loss": 0.9977171268463135,
301
+ "train samples": 20000,
302
+ "train time": 53.51452894380782,
303
+ "eval time": 20.269856880011503,
304
+ "tokens / sec": 3892.0271580583562,
305
+ "mem allocated avg": 6777745672.192,
306
+ "mem reserved avg": 15388339339.264,
307
+ "elapsed time": 1580.3933992580278
308
+ },
309
+ {
310
+ "step": 5000,
311
+ "test accuracy": 0.11675511751326763,
312
+ "train loss": 0.9977171268463135,
313
+ "train samples": 20000,
314
+ "train total tokens": 4198051,
315
+ "forgetting": 0.40204334259033203
316
+ }
317
+ ]
318
+ },
319
+ "meta_info": {
320
+ "model_info": {
321
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
322
+ "created_at": "2024-09-18T15:23:48+00:00"
323
+ },
324
+ "dataset_info": {
325
+ "metamath": {
326
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
327
+ "created_at": "2023-09-21T17:22:46+00:00"
328
+ },
329
+ "gsm8k": {
330
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
331
+ "created_at": "2022-04-12T10:22:10+00:00"
332
+ }
333
+ },
334
+ "package_info": {
335
+ "transformers-version": "4.57.1",
336
+ "transformers-commit-hash": null,
337
+ "peft-version": "0.18.1.dev0",
338
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
339
+ "datasets-version": "4.2.0",
340
+ "datasets-commit-hash": null,
341
+ "bitsandbytes-version": "0.46.0",
342
+ "bitsandbytes-commit-hash": null,
343
+ "torch-version": "2.9.0+cu128",
344
+ "torch-commit-hash": null
345
+ },
346
+ "system_info": {
347
+ "system": "Linux",
348
+ "release": "6.14.0-1016-aws",
349
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
350
+ "machine": "x86_64",
351
+ "processor": "x86_64",
352
+ "accelerator": "NVIDIA L40S"
353
+ },
354
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
355
+ }
356
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/full-finetuning--llama-3.2-3B-lr_0.00001.json ADDED
@@ -0,0 +1,332 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-09T23:21:09+00:00",
4
+ "total_time": 2347.218174619018,
5
+ "experiment_name": "full-finetuning/llama-3.2-3B-lr_0.00001",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 1e-05
22
+ },
23
+ "lr_scheduler": "cosine",
24
+ "use_amp": false,
25
+ "autocast_adapter_dtype": true,
26
+ "generation_kwargs": {
27
+ "max_length": 800,
28
+ "max_new_tokens": 300
29
+ },
30
+ "attn_implementation": null
31
+ },
32
+ "peft_config": null,
33
+ "error_msg": ""
34
+ },
35
+ "train_info": {
36
+ "accelerator_memory_reserved_avg": 34309618532,
37
+ "accelerator_memory_max": 38549848064,
38
+ "accelerator_memory_reserved_99th": 37237492285,
39
+ "train_time": 2154.846133010811,
40
+ "file_size": 6425499648,
41
+ "num_trainable_params": 3212749824,
42
+ "num_total_params": 3212749824,
43
+ "status": "success",
44
+ "metrics": [
45
+ {
46
+ "step": 250,
47
+ "valid accuracy": 0.3,
48
+ "train loss": 1.0749022269248962,
49
+ "train samples": 1000,
50
+ "train time": 91.49652248725761,
51
+ "eval time": 11.212762297014706,
52
+ "tokens / sec": 2313.956795784073,
53
+ "mem allocated avg": 26068575909.888,
54
+ "mem reserved avg": 34185951576.064,
55
+ "elapsed time": 124.61640950996662
56
+ },
57
+ {
58
+ "step": 500,
59
+ "valid accuracy": 0.4,
60
+ "train loss": 0.7238605138063431,
61
+ "train samples": 2000,
62
+ "train time": 91.91453206090955,
63
+ "eval time": 11.141172014991753,
64
+ "tokens / sec": 2262.917466219234,
65
+ "mem allocated avg": 26061334226.944,
66
+ "mem reserved avg": 34147850518.528,
67
+ "elapsed time": 230.16530759999296
68
+ },
69
+ {
70
+ "step": 750,
71
+ "valid accuracy": 0.42,
72
+ "train loss": 0.6648618532419205,
73
+ "train samples": 3000,
74
+ "train time": 92.5216026121634,
75
+ "eval time": 6.087067044980358,
76
+ "tokens / sec": 2317.3074606017867,
77
+ "mem allocated avg": 26070816337.92,
78
+ "mem reserved avg": 34269409837.056,
79
+ "elapsed time": 331.3244165930082
80
+ },
81
+ {
82
+ "step": 1000,
83
+ "valid accuracy": 0.4,
84
+ "train loss": 0.6407386429309845,
85
+ "train samples": 4000,
86
+ "train time": 91.41802564635873,
87
+ "eval time": 11.26410798204597,
88
+ "tokens / sec": 2278.9378629322678,
89
+ "mem allocated avg": 26063015718.912,
90
+ "mem reserved avg": 34418341183.488,
91
+ "elapsed time": 436.4462315859855
92
+ },
93
+ {
94
+ "step": 1250,
95
+ "valid accuracy": 0.46,
96
+ "train loss": 0.6343499521017074,
97
+ "train samples": 5000,
98
+ "train time": 91.51783608621918,
99
+ "eval time": 9.06717514997581,
100
+ "tokens / sec": 2278.6596462304447,
101
+ "mem allocated avg": 26061503963.136,
102
+ "mem reserved avg": 34260576632.832,
103
+ "elapsed time": 539.5919883140014
104
+ },
105
+ {
106
+ "step": 1500,
107
+ "valid accuracy": 0.58,
108
+ "train loss": 0.624906349182129,
109
+ "train samples": 6000,
110
+ "train time": 91.87651025888044,
111
+ "eval time": 11.171614417049568,
112
+ "tokens / sec": 2278.395200363706,
113
+ "mem allocated avg": 26064701278.208,
114
+ "mem reserved avg": 34224690167.808,
115
+ "elapsed time": 645.2011065029656
116
+ },
117
+ {
118
+ "step": 1750,
119
+ "valid accuracy": 0.54,
120
+ "train loss": 0.6175477323532105,
121
+ "train samples": 7000,
122
+ "train time": 90.95103025593562,
123
+ "eval time": 11.160093326005153,
124
+ "tokens / sec": 2301.843084249583,
125
+ "mem allocated avg": 26063144126.464,
126
+ "mem reserved avg": 34362062012.416,
127
+ "elapsed time": 749.8216157860006
128
+ },
129
+ {
130
+ "step": 2000,
131
+ "valid accuracy": 0.42,
132
+ "train loss": 0.6184427078962326,
133
+ "train samples": 8000,
134
+ "train time": 90.84291155746905,
135
+ "eval time": 11.153122653020546,
136
+ "tokens / sec": 2286.3203792032505,
137
+ "mem allocated avg": 26060605181.952,
138
+ "mem reserved avg": 34304549715.968,
139
+ "elapsed time": 854.322792426974
140
+ },
141
+ {
142
+ "step": 2250,
143
+ "valid accuracy": 0.46,
144
+ "train loss": 0.6107840378284455,
145
+ "train samples": 9000,
146
+ "train time": 92.96169786009705,
147
+ "eval time": 11.16127427702304,
148
+ "tokens / sec": 2312.221107702729,
149
+ "mem allocated avg": 26073147115.52,
150
+ "mem reserved avg": 34312711831.552,
151
+ "elapsed time": 960.9334694420104
152
+ },
153
+ {
154
+ "step": 2500,
155
+ "valid accuracy": 0.54,
156
+ "train loss": 0.6089447345733643,
157
+ "train samples": 10000,
158
+ "train time": 90.26249368820572,
159
+ "eval time": 11.146873010031413,
160
+ "tokens / sec": 2281.8669370189687,
161
+ "mem allocated avg": 26057948864.512,
162
+ "mem reserved avg": 34399810748.416,
163
+ "elapsed time": 1064.9780696959933
164
+ },
165
+ {
166
+ "step": 2750,
167
+ "valid accuracy": 0.56,
168
+ "train loss": 0.6021645392179489,
169
+ "train samples": 11000,
170
+ "train time": 91.39197911316296,
171
+ "eval time": 11.157545586000197,
172
+ "tokens / sec": 2318.376317659624,
173
+ "mem allocated avg": 26068202784.768,
174
+ "mem reserved avg": 34373529239.552,
175
+ "elapsed time": 1170.0117728059995
176
+ },
177
+ {
178
+ "step": 3000,
179
+ "valid accuracy": 0.48,
180
+ "train loss": 0.5950663160085679,
181
+ "train samples": 12000,
182
+ "train time": 91.94340823084349,
183
+ "eval time": 6.359558377007488,
184
+ "tokens / sec": 2270.2116879976475,
185
+ "mem allocated avg": 26062483675.136,
186
+ "mem reserved avg": 34305480851.456,
187
+ "elapsed time": 1270.7611719949637
188
+ },
189
+ {
190
+ "step": 3250,
191
+ "valid accuracy": 0.5,
192
+ "train loss": 0.6067559220790864,
193
+ "train samples": 13000,
194
+ "train time": 91.35254304017872,
195
+ "eval time": 6.207320767978672,
196
+ "tokens / sec": 2308.649469202422,
197
+ "mem allocated avg": 26064313700.352,
198
+ "mem reserved avg": 34178494103.552,
199
+ "elapsed time": 1370.8038186989725
200
+ },
201
+ {
202
+ "step": 3500,
203
+ "valid accuracy": 0.46,
204
+ "train loss": 0.5922972733974456,
205
+ "train samples": 14000,
206
+ "train time": 91.81345162418438,
207
+ "eval time": 7.237666736007668,
208
+ "tokens / sec": 2284.523632316533,
209
+ "mem allocated avg": 26062809985.024,
210
+ "mem reserved avg": 34401211645.952,
211
+ "elapsed time": 1472.3757120459923
212
+ },
213
+ {
214
+ "step": 3750,
215
+ "valid accuracy": 0.5,
216
+ "train loss": 0.5926397043466568,
217
+ "train samples": 15000,
218
+ "train time": 92.5030986299389,
219
+ "eval time": 11.144441604032181,
220
+ "tokens / sec": 2342.6566591777223,
221
+ "mem allocated avg": 26074840416.256,
222
+ "mem reserved avg": 34314205003.776,
223
+ "elapsed time": 1578.5238473899662
224
+ },
225
+ {
226
+ "step": 4000,
227
+ "valid accuracy": 0.52,
228
+ "train loss": 0.6050900548696518,
229
+ "train samples": 16000,
230
+ "train time": 90.88473943283316,
231
+ "eval time": 11.175483218976296,
232
+ "tokens / sec": 2248.7053522449546,
233
+ "mem allocated avg": 26056039493.632,
234
+ "mem reserved avg": 34580459421.696,
235
+ "elapsed time": 1683.2547346790088
236
+ },
237
+ {
238
+ "step": 4250,
239
+ "valid accuracy": 0.52,
240
+ "train loss": 0.5931364653110505,
241
+ "train samples": 17000,
242
+ "train time": 91.31777060282184,
243
+ "eval time": 7.878185064007994,
244
+ "tokens / sec": 2314.8725445720397,
245
+ "mem allocated avg": 26067197499.392,
246
+ "mem reserved avg": 34410673995.776,
247
+ "elapsed time": 1785.080884763971
248
+ },
249
+ {
250
+ "step": 4500,
251
+ "valid accuracy": 0.48,
252
+ "train loss": 0.6010294594764709,
253
+ "train samples": 18000,
254
+ "train time": 91.71872854570393,
255
+ "eval time": 6.507075314002577,
256
+ "tokens / sec": 2265.8185879282355,
257
+ "mem allocated avg": 26062461708.288,
258
+ "mem reserved avg": 34202108035.072,
259
+ "elapsed time": 1885.7456577859703
260
+ },
261
+ {
262
+ "step": 4750,
263
+ "valid accuracy": 0.5,
264
+ "train loss": 0.5937523410320282,
265
+ "train samples": 19000,
266
+ "train time": 91.9850392206572,
267
+ "eval time": 6.368945363035891,
268
+ "tokens / sec": 2282.31679606496,
269
+ "mem allocated avg": 26063243161.6,
270
+ "mem reserved avg": 34250082484.224,
271
+ "elapsed time": 1986.7437554029748
272
+ },
273
+ {
274
+ "step": 5000,
275
+ "valid accuracy": 0.46,
276
+ "train loss": 0.5987640290260315,
277
+ "train samples": 20000,
278
+ "train time": 90.72794272896135,
279
+ "eval time": 11.154609176970553,
280
+ "tokens / sec": 2295.654389764034,
281
+ "mem allocated avg": 26061294045.184,
282
+ "mem reserved avg": 34290171641.856,
283
+ "elapsed time": 2091.1681218339945
284
+ },
285
+ {
286
+ "step": 5000,
287
+ "test accuracy": 0.49507202426080366,
288
+ "train loss": 0.5987640290260315,
289
+ "train samples": 20000,
290
+ "train total tokens": 4198051,
291
+ "forgetting": 0.5911064147949219
292
+ }
293
+ ]
294
+ },
295
+ "meta_info": {
296
+ "model_info": {
297
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
298
+ "created_at": "2024-09-18T15:23:48+00:00"
299
+ },
300
+ "dataset_info": {
301
+ "metamath": {
302
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
303
+ "created_at": "2023-09-21T17:22:46+00:00"
304
+ },
305
+ "gsm8k": {
306
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
307
+ "created_at": "2022-04-12T10:22:10+00:00"
308
+ }
309
+ },
310
+ "package_info": {
311
+ "transformers-version": "4.57.1",
312
+ "transformers-commit-hash": null,
313
+ "peft-version": "0.18.1.dev0",
314
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
315
+ "datasets-version": "4.2.0",
316
+ "datasets-commit-hash": null,
317
+ "bitsandbytes-version": "0.46.0",
318
+ "bitsandbytes-commit-hash": null,
319
+ "torch-version": "2.9.0+cu128",
320
+ "torch-commit-hash": null
321
+ },
322
+ "system_info": {
323
+ "system": "Linux",
324
+ "release": "6.14.0-1016-aws",
325
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
326
+ "machine": "x86_64",
327
+ "processor": "x86_64",
328
+ "accelerator": "NVIDIA L40S"
329
+ },
330
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
331
+ }
332
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/gralora--llama-3.2-3B-rank32.json ADDED
@@ -0,0 +1,355 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-10T00:00:21+00:00",
4
+ "total_time": 1289.217340779025,
5
+ "experiment_name": "gralora/llama-3.2-3B-rank32",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0002
22
+ },
23
+ "lr_scheduler": "cosine",
24
+ "use_amp": false,
25
+ "autocast_adapter_dtype": true,
26
+ "generation_kwargs": {
27
+ "max_length": 800,
28
+ "max_new_tokens": 300
29
+ },
30
+ "attn_implementation": null
31
+ },
32
+ "peft_config": {
33
+ "task_type": null,
34
+ "peft_type": "GRALORA",
35
+ "auto_mapping": null,
36
+ "peft_version": "0.17.2.dev0@UNKNOWN",
37
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
38
+ "revision": null,
39
+ "inference_mode": false,
40
+ "r": 32,
41
+ "hybrid_r": 0,
42
+ "target_modules": [
43
+ "q_proj",
44
+ "v_proj"
45
+ ],
46
+ "alpha": 64,
47
+ "gralora_dropout": 0.0,
48
+ "gralora_k": 2,
49
+ "fan_in_fan_out": false,
50
+ "bias": "none",
51
+ "modules_to_save": null,
52
+ "init_weights": true,
53
+ "layers_to_transform": null,
54
+ "layers_pattern": null
55
+ },
56
+ "error_msg": ""
57
+ },
58
+ "train_info": {
59
+ "accelerator_memory_reserved_avg": 14431023425,
60
+ "accelerator_memory_max": 22307405824,
61
+ "accelerator_memory_reserved_99th": 20126367744,
62
+ "train_time": 1078.0007609191234,
63
+ "file_size": 36714960,
64
+ "num_trainable_params": 9175040,
65
+ "num_total_params": 3221924864,
66
+ "status": "success",
67
+ "metrics": [
68
+ {
69
+ "step": 250,
70
+ "valid accuracy": 0.36,
71
+ "train loss": 0.942215921163559,
72
+ "train samples": 1000,
73
+ "train time": 34.042564845993184,
74
+ "eval time": 13.884479815023951,
75
+ "tokens / sec": 6219.243495835461,
76
+ "mem allocated avg": 6927297665.024,
77
+ "mem reserved avg": 14540712443.904,
78
+ "elapsed time": 70.35818397701951
79
+ },
80
+ {
81
+ "step": 500,
82
+ "valid accuracy": 0.48,
83
+ "train loss": 0.707061094880104,
84
+ "train samples": 2000,
85
+ "train time": 33.86772509320872,
86
+ "eval time": 13.81671311397804,
87
+ "tokens / sec": 6141.392710244596,
88
+ "mem allocated avg": 6920123852.8,
89
+ "mem reserved avg": 14167075454.976,
90
+ "elapsed time": 121.25836294901092
91
+ },
92
+ {
93
+ "step": 750,
94
+ "valid accuracy": 0.42,
95
+ "train loss": 0.6730556710958481,
96
+ "train samples": 3000,
97
+ "train time": 34.94966330617899,
98
+ "eval time": 13.804657173983287,
99
+ "tokens / sec": 6134.5655356312,
100
+ "mem allocated avg": 6930530816.0,
101
+ "mem reserved avg": 14409984376.832,
102
+ "elapsed time": 173.17440196697135
103
+ },
104
+ {
105
+ "step": 1000,
106
+ "valid accuracy": 0.38,
107
+ "train loss": 0.6534032428264618,
108
+ "train samples": 4000,
109
+ "train time": 34.35864242265234,
110
+ "eval time": 13.949590890028048,
111
+ "tokens / sec": 6063.56902689048,
112
+ "mem allocated avg": 6921228693.504,
113
+ "mem reserved avg": 14450920783.872,
114
+ "elapsed time": 224.78821431397228
115
+ },
116
+ {
117
+ "step": 1250,
118
+ "valid accuracy": 0.4,
119
+ "train loss": 0.6493164491653443,
120
+ "train samples": 5000,
121
+ "train time": 33.983851948869415,
122
+ "eval time": 11.309890707023442,
123
+ "tokens / sec": 6136.3850193838225,
124
+ "mem allocated avg": 6921922877.44,
125
+ "mem reserved avg": 14466959802.368,
126
+ "elapsed time": 273.25096270401264
127
+ },
128
+ {
129
+ "step": 1500,
130
+ "valid accuracy": 0.34,
131
+ "train loss": 0.6408532521724701,
132
+ "train samples": 6000,
133
+ "train time": 34.35372261068551,
134
+ "eval time": 9.27557178097777,
135
+ "tokens / sec": 6093.400775579672,
136
+ "mem allocated avg": 6923754981.376,
137
+ "mem reserved avg": 14382721400.832,
138
+ "elapsed time": 320.11889754998265
139
+ },
140
+ {
141
+ "step": 1750,
142
+ "valid accuracy": 0.4,
143
+ "train loss": 0.6322305067777634,
144
+ "train samples": 7000,
145
+ "train time": 34.57727349648485,
146
+ "eval time": 9.888179413974285,
147
+ "tokens / sec": 6054.699484078268,
148
+ "mem allocated avg": 6925341489.152,
149
+ "mem reserved avg": 14776876924.928,
150
+ "elapsed time": 367.8946221729857
151
+ },
152
+ {
153
+ "step": 2000,
154
+ "valid accuracy": 0.52,
155
+ "train loss": 0.6344789278507232,
156
+ "train samples": 8000,
157
+ "train time": 33.59021200949792,
158
+ "eval time": 8.374654100975022,
159
+ "tokens / sec": 6183.229803410356,
160
+ "mem allocated avg": 6920861077.504,
161
+ "mem reserved avg": 14419908100.096,
162
+ "elapsed time": 413.12702070496744
163
+ },
164
+ {
165
+ "step": 2250,
166
+ "valid accuracy": 0.42,
167
+ "train loss": 0.6251693757772445,
168
+ "train samples": 9000,
169
+ "train time": 34.35726486198837,
170
+ "eval time": 13.802904761978425,
171
+ "tokens / sec": 6256.260527822477,
172
+ "mem allocated avg": 6931583428.608,
173
+ "mem reserved avg": 14699693342.72,
174
+ "elapsed time": 464.4872953959857
175
+ },
176
+ {
177
+ "step": 2500,
178
+ "valid accuracy": 0.46,
179
+ "train loss": 0.6212710882425309,
180
+ "train samples": 10000,
181
+ "train time": 33.39420525066089,
182
+ "eval time": 13.795758826017845,
183
+ "tokens / sec": 6167.746722941514,
184
+ "mem allocated avg": 6918615058.432,
185
+ "mem reserved avg": 14201795903.488,
186
+ "elapsed time": 514.8732995300088
187
+ },
188
+ {
189
+ "step": 2750,
190
+ "valid accuracy": 0.48,
191
+ "train loss": 0.6123416483402252,
192
+ "train samples": 11000,
193
+ "train time": 34.10195262962952,
194
+ "eval time": 7.583350939967204,
195
+ "tokens / sec": 6213.163284260355,
196
+ "mem allocated avg": 6927684456.448,
197
+ "mem reserved avg": 14572228444.16,
198
+ "elapsed time": 559.7389343840186
199
+ },
200
+ {
201
+ "step": 3000,
202
+ "valid accuracy": 0.46,
203
+ "train loss": 0.6028379487991333,
204
+ "train samples": 12000,
205
+ "train time": 34.35880611848552,
206
+ "eval time": 7.6943675089860335,
207
+ "tokens / sec": 6075.036463147065,
208
+ "mem allocated avg": 6923840229.376,
209
+ "mem reserved avg": 14495858556.928,
210
+ "elapsed time": 605.0798396450118
211
+ },
212
+ {
213
+ "step": 3250,
214
+ "valid accuracy": 0.54,
215
+ "train loss": 0.6112224916219712,
216
+ "train samples": 13000,
217
+ "train time": 34.423862263094634,
218
+ "eval time": 10.099462785990909,
219
+ "tokens / sec": 6126.593186671682,
220
+ "mem allocated avg": 6926499131.392,
221
+ "mem reserved avg": 14352186867.712,
222
+ "elapsed time": 652.7459372479934
223
+ },
224
+ {
225
+ "step": 3500,
226
+ "valid accuracy": 0.5,
227
+ "train loss": 0.5940833052396775,
228
+ "train samples": 14000,
229
+ "train time": 34.00928645499516,
230
+ "eval time": 7.006814954045694,
231
+ "tokens / sec": 6167.433129700158,
232
+ "mem allocated avg": 6923424174.08,
233
+ "mem reserved avg": 14361095569.408,
234
+ "elapsed time": 696.9908059639856
235
+ },
236
+ {
237
+ "step": 3750,
238
+ "valid accuracy": 0.52,
239
+ "train loss": 0.5911667649745941,
240
+ "train samples": 15000,
241
+ "train time": 34.607924775453284,
242
+ "eval time": 8.789539379009511,
243
+ "tokens / sec": 6261.6583168749585,
244
+ "mem allocated avg": 6934536548.352,
245
+ "mem reserved avg": 14719675006.976,
246
+ "elapsed time": 743.6320169619867
247
+ },
248
+ {
249
+ "step": 4000,
250
+ "valid accuracy": 0.52,
251
+ "train loss": 0.601644613146782,
252
+ "train samples": 16000,
253
+ "train time": 33.85560346476268,
254
+ "eval time": 7.726654959958978,
255
+ "tokens / sec": 6036.607801503638,
256
+ "mem allocated avg": 6915689891.84,
257
+ "mem reserved avg": 14388962525.184,
258
+ "elapsed time": 788.4833239429863
259
+ },
260
+ {
261
+ "step": 4250,
262
+ "valid accuracy": 0.6,
263
+ "train loss": 0.5881835660934448,
264
+ "train samples": 17000,
265
+ "train time": 34.50086922443006,
266
+ "eval time": 10.961278792994563,
267
+ "tokens / sec": 6127.063020496756,
268
+ "mem allocated avg": 6927453581.312,
269
+ "mem reserved avg": 14375817576.448,
270
+ "elapsed time": 837.071181916981
271
+ },
272
+ {
273
+ "step": 4500,
274
+ "valid accuracy": 0.58,
275
+ "train loss": 0.5937378851175308,
276
+ "train samples": 18000,
277
+ "train time": 33.85475521400804,
278
+ "eval time": 9.431811225018464,
279
+ "tokens / sec": 6138.517283209049,
280
+ "mem allocated avg": 6921458198.528,
281
+ "mem reserved avg": 14334059085.824,
282
+ "elapsed time": 883.6634952509776
283
+ },
284
+ {
285
+ "step": 4750,
286
+ "valid accuracy": 0.58,
287
+ "train loss": 0.586182700753212,
288
+ "train samples": 19000,
289
+ "train time": 34.06897705892334,
290
+ "eval time": 9.29696652397979,
291
+ "tokens / sec": 6162.1750379209825,
292
+ "mem allocated avg": 6924237260.8,
293
+ "mem reserved avg": 14424924487.68,
294
+ "elapsed time": 930.1821659649722
295
+ },
296
+ {
297
+ "step": 5000,
298
+ "valid accuracy": 0.6,
299
+ "train loss": 0.592198490023613,
300
+ "train samples": 20000,
301
+ "train time": 34.439023942220956,
302
+ "eval time": 7.703576382948086,
303
+ "tokens / sec": 6047.7904469486575,
304
+ "mem allocated avg": 6920309841.92,
305
+ "mem reserved avg": 14079011848.192,
306
+ "elapsed time": 975.4882056199713
307
+ },
308
+ {
309
+ "step": 5000,
310
+ "test accuracy": 0.5011372251705838,
311
+ "train loss": 0.592198490023613,
312
+ "train samples": 20000,
313
+ "train total tokens": 4198051,
314
+ "forgetting": 0.49779701232910156
315
+ }
316
+ ]
317
+ },
318
+ "meta_info": {
319
+ "model_info": {
320
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
321
+ "created_at": "2024-09-18T15:23:48+00:00"
322
+ },
323
+ "dataset_info": {
324
+ "metamath": {
325
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
326
+ "created_at": "2023-09-21T17:22:46+00:00"
327
+ },
328
+ "gsm8k": {
329
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
330
+ "created_at": "2022-04-12T10:22:10+00:00"
331
+ }
332
+ },
333
+ "package_info": {
334
+ "transformers-version": "4.57.1",
335
+ "transformers-commit-hash": null,
336
+ "peft-version": "0.18.1.dev0",
337
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
338
+ "datasets-version": "4.2.0",
339
+ "datasets-commit-hash": null,
340
+ "bitsandbytes-version": "0.46.0",
341
+ "bitsandbytes-commit-hash": null,
342
+ "torch-version": "2.9.0+cu128",
343
+ "torch-commit-hash": null
344
+ },
345
+ "system_info": {
346
+ "system": "Linux",
347
+ "release": "6.14.0-1016-aws",
348
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
349
+ "machine": "x86_64",
350
+ "processor": "x86_64",
351
+ "accelerator": "NVIDIA L40S"
352
+ },
353
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
354
+ }
355
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/ia3--llama-3.2-3B-default.json ADDED
@@ -0,0 +1,353 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-10T00:21:54+00:00",
4
+ "total_time": 1252.7159106170293,
5
+ "experiment_name": "ia3/llama-3.2-3B-default",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "IA3",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "target_modules": [
42
+ "k_proj",
43
+ "down_proj",
44
+ "v_proj"
45
+ ],
46
+ "exclude_modules": null,
47
+ "feedforward_modules": [
48
+ "down_proj"
49
+ ],
50
+ "fan_in_fan_out": false,
51
+ "modules_to_save": null,
52
+ "init_ia3_weights": true
53
+ },
54
+ "error_msg": ""
55
+ },
56
+ "train_info": {
57
+ "accelerator_memory_reserved_avg": 14786832524,
58
+ "accelerator_memory_max": 23240638464,
59
+ "accelerator_memory_reserved_99th": 20933771264,
60
+ "train_time": 1015.700665248034,
61
+ "file_size": 1157064,
62
+ "num_trainable_params": 286720,
63
+ "num_total_params": 3213036544,
64
+ "status": "success",
65
+ "metrics": [
66
+ {
67
+ "step": 250,
68
+ "valid accuracy": 0.0,
69
+ "train loss": 1.3155594708919525,
70
+ "train samples": 1000,
71
+ "train time": 29.70507554640062,
72
+ "eval time": 11.724285421019886,
73
+ "tokens / sec": 7127.367835482717,
74
+ "mem allocated avg": 6782600361.984,
75
+ "mem reserved avg": 14885492621.312,
76
+ "elapsed time": 63.95026855904143
77
+ },
78
+ {
79
+ "step": 500,
80
+ "valid accuracy": 0.0,
81
+ "train loss": 1.2050744531154634,
82
+ "train samples": 2000,
83
+ "train time": 29.590323601732962,
84
+ "eval time": 11.715296942042187,
85
+ "tokens / sec": 7029.155976780827,
86
+ "mem allocated avg": 6775425632.256,
87
+ "mem reserved avg": 14480272523.264,
88
+ "elapsed time": 108.9384753460181
89
+ },
90
+ {
91
+ "step": 750,
92
+ "valid accuracy": 0.12,
93
+ "train loss": 1.0194114191532135,
94
+ "train samples": 3000,
95
+ "train time": 30.585519072716124,
96
+ "eval time": 11.650421504979022,
97
+ "tokens / sec": 7009.885936225842,
98
+ "mem allocated avg": 6785668968.448,
99
+ "mem reserved avg": 14756492607.488,
100
+ "elapsed time": 154.82391193299554
101
+ },
102
+ {
103
+ "step": 1000,
104
+ "valid accuracy": 0.3,
105
+ "train loss": 0.9196182670593261,
106
+ "train samples": 4000,
107
+ "train time": 29.49092253309209,
108
+ "eval time": 11.840653097024187,
109
+ "tokens / sec": 7064.411083316362,
110
+ "mem allocated avg": 6777244168.192,
111
+ "mem reserved avg": 14776985976.832,
112
+ "elapsed time": 199.8590077710105
113
+ },
114
+ {
115
+ "step": 1250,
116
+ "valid accuracy": 0.3,
117
+ "train loss": 0.8686861591339111,
118
+ "train samples": 5000,
119
+ "train time": 29.44121315999655,
120
+ "eval time": 11.64985877397703,
121
+ "tokens / sec": 7083.1999641696975,
122
+ "mem allocated avg": 6777135923.2,
123
+ "mem reserved avg": 14839564992.512,
124
+ "elapsed time": 244.56394972099224
125
+ },
126
+ {
127
+ "step": 1500,
128
+ "valid accuracy": 0.3,
129
+ "train loss": 0.8333434355258942,
130
+ "train samples": 6000,
131
+ "train time": 29.930749938997906,
132
+ "eval time": 11.69840682001086,
133
+ "tokens / sec": 6993.844137772663,
134
+ "mem allocated avg": 6779035060.224,
135
+ "mem reserved avg": 14722384527.36,
136
+ "elapsed time": 289.89695524802664
137
+ },
138
+ {
139
+ "step": 1750,
140
+ "valid accuracy": 0.3,
141
+ "train loss": 0.8170836737155914,
142
+ "train samples": 7000,
143
+ "train time": 30.3804316627793,
144
+ "eval time": 11.654387208051048,
145
+ "tokens / sec": 6891.113408914859,
146
+ "mem allocated avg": 6779937654.784,
147
+ "mem reserved avg": 15161679151.104,
148
+ "elapsed time": 335.59190371102886
149
+ },
150
+ {
151
+ "step": 2000,
152
+ "valid accuracy": 0.3,
153
+ "train loss": 0.8073079003095627,
154
+ "train samples": 8000,
155
+ "train time": 29.52141359890811,
156
+ "eval time": 11.679582766955718,
157
+ "tokens / sec": 7035.435457863099,
158
+ "mem allocated avg": 6776053977.088,
159
+ "mem reserved avg": 14763992023.04,
160
+ "elapsed time": 380.5167595850071
161
+ },
162
+ {
163
+ "step": 2250,
164
+ "valid accuracy": 0.28,
165
+ "train loss": 0.795267580151558,
166
+ "train samples": 9000,
167
+ "train time": 30.315992950985674,
168
+ "eval time": 11.681108721997589,
169
+ "tokens / sec": 7090.25102187891,
170
+ "mem allocated avg": 6787896084.48,
171
+ "mem reserved avg": 15088715038.72,
172
+ "elapsed time": 426.1353191550006
173
+ },
174
+ {
175
+ "step": 2500,
176
+ "valid accuracy": 0.3,
177
+ "train loss": 0.7891060090065003,
178
+ "train samples": 10000,
179
+ "train time": 29.285075834020972,
180
+ "eval time": 11.683325350983068,
181
+ "tokens / sec": 7033.17284091594,
182
+ "mem allocated avg": 6772744847.36,
183
+ "mem reserved avg": 14541761019.904,
184
+ "elapsed time": 470.7990039800061
185
+ },
186
+ {
187
+ "step": 2750,
188
+ "valid accuracy": 0.3,
189
+ "train loss": 0.7823175694942475,
190
+ "train samples": 11000,
191
+ "train time": 30.584478964097798,
192
+ "eval time": 11.64602530299453,
193
+ "tokens / sec": 6927.729592801655,
194
+ "mem allocated avg": 6783104911.36,
195
+ "mem reserved avg": 14943508234.24,
196
+ "elapsed time": 516.6453109350405
197
+ },
198
+ {
199
+ "step": 3000,
200
+ "valid accuracy": 0.26,
201
+ "train loss": 0.7710221862792969,
202
+ "train samples": 12000,
203
+ "train time": 29.543955663975794,
204
+ "eval time": 11.677449156995863,
205
+ "tokens / sec": 7065.099960683823,
206
+ "mem allocated avg": 6778547337.216,
207
+ "mem reserved avg": 14865569677.312,
208
+ "elapsed time": 561.5630381150404
209
+ },
210
+ {
211
+ "step": 3250,
212
+ "valid accuracy": 0.3,
213
+ "train loss": 0.7756292035579682,
214
+ "train samples": 13000,
215
+ "train time": 29.600664581230376,
216
+ "eval time": 11.676535067963414,
217
+ "tokens / sec": 7124.873815628153,
218
+ "mem allocated avg": 6780569153.536,
219
+ "mem reserved avg": 14715774304.256,
220
+ "elapsed time": 606.3470957170357
221
+ },
222
+ {
223
+ "step": 3500,
224
+ "valid accuracy": 0.32,
225
+ "train loss": 0.7658540964126587,
226
+ "train samples": 14000,
227
+ "train time": 29.628978170803748,
228
+ "eval time": 11.7645476309699,
229
+ "tokens / sec": 7079.218148896091,
230
+ "mem allocated avg": 6778699649.024,
231
+ "mem reserved avg": 14727702904.832,
232
+ "elapsed time": 651.4321867520339
233
+ },
234
+ {
235
+ "step": 3750,
236
+ "valid accuracy": 0.36,
237
+ "train loss": 0.7585025093555451,
238
+ "train samples": 15000,
239
+ "train time": 30.480809888918884,
240
+ "eval time": 11.652502680022735,
241
+ "tokens / sec": 7109.489570314241,
242
+ "mem allocated avg": 6790036742.144,
243
+ "mem reserved avg": 15097195921.408,
244
+ "elapsed time": 697.2096698190435
245
+ },
246
+ {
247
+ "step": 4000,
248
+ "valid accuracy": 0.34,
249
+ "train loss": 0.7766863429546356,
250
+ "train samples": 16000,
251
+ "train time": 29.984013571229298,
252
+ "eval time": 11.786360332975164,
253
+ "tokens / sec": 6816.065484845664,
254
+ "mem allocated avg": 6770827464.704,
255
+ "mem reserved avg": 14748657647.616,
256
+ "elapsed time": 742.6605575150461
257
+ },
258
+ {
259
+ "step": 4250,
260
+ "valid accuracy": 0.32,
261
+ "train loss": 0.7541669148206711,
262
+ "train samples": 17000,
263
+ "train time": 30.13022418995388,
264
+ "eval time": 11.657879816018976,
265
+ "tokens / sec": 7015.845573113327,
266
+ "mem allocated avg": 6782400247.808,
267
+ "mem reserved avg": 14721461780.48,
268
+ "elapsed time": 787.9822572360281
269
+ },
270
+ {
271
+ "step": 4500,
272
+ "valid accuracy": 0.3,
273
+ "train loss": 0.7659742934703827,
274
+ "train samples": 18000,
275
+ "train time": 30.020536605035886,
276
+ "eval time": 11.697008132992778,
277
+ "tokens / sec": 6922.527826006246,
278
+ "mem allocated avg": 6776485750.784,
279
+ "mem reserved avg": 14703325609.984,
280
+ "elapsed time": 833.4761402930017
281
+ },
282
+ {
283
+ "step": 4750,
284
+ "valid accuracy": 0.32,
285
+ "train loss": 0.7579846435785293,
286
+ "train samples": 19000,
287
+ "train time": 30.37128037889488,
288
+ "eval time": 11.668571799993515,
289
+ "tokens / sec": 6912.4184881546,
290
+ "mem allocated avg": 6779605635.072,
291
+ "mem reserved avg": 14800801234.944,
292
+ "elapsed time": 879.1311652210425
293
+ },
294
+ {
295
+ "step": 5000,
296
+ "valid accuracy": 0.34,
297
+ "train loss": 0.7657218432426453,
298
+ "train samples": 20000,
299
+ "train time": 29.584271008439828,
300
+ "eval time": 11.69853776000673,
301
+ "tokens / sec": 7040.227556750738,
302
+ "mem allocated avg": 6776251918.336,
303
+ "mem reserved avg": 14395312701.44,
304
+ "elapsed time": 924.0385756980395
305
+ },
306
+ {
307
+ "step": 5000,
308
+ "test accuracy": 0.35178165276724793,
309
+ "train loss": 0.7657218432426453,
310
+ "train samples": 20000,
311
+ "train total tokens": 4198051,
312
+ "forgetting": 0.1822957992553711
313
+ }
314
+ ]
315
+ },
316
+ "meta_info": {
317
+ "model_info": {
318
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
319
+ "created_at": "2024-09-18T15:23:48+00:00"
320
+ },
321
+ "dataset_info": {
322
+ "metamath": {
323
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
324
+ "created_at": "2023-09-21T17:22:46+00:00"
325
+ },
326
+ "gsm8k": {
327
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
328
+ "created_at": "2022-04-12T10:22:10+00:00"
329
+ }
330
+ },
331
+ "package_info": {
332
+ "transformers-version": "4.57.1",
333
+ "transformers-commit-hash": null,
334
+ "peft-version": "0.18.1.dev0",
335
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
336
+ "datasets-version": "4.2.0",
337
+ "datasets-commit-hash": null,
338
+ "bitsandbytes-version": "0.46.0",
339
+ "bitsandbytes-commit-hash": null,
340
+ "torch-version": "2.9.0+cu128",
341
+ "torch-commit-hash": null
342
+ },
343
+ "system_info": {
344
+ "system": "Linux",
345
+ "release": "6.14.0-1016-aws",
346
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
347
+ "machine": "x86_64",
348
+ "processor": "x86_64",
349
+ "accelerator": "NVIDIA L40S"
350
+ },
351
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
352
+ }
353
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/ia3--llama-3.2-3B-lr_0.001.json ADDED
@@ -0,0 +1,352 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-10T00:42:51+00:00",
4
+ "total_time": 1158.928240164998,
5
+ "experiment_name": "ia3/llama-3.2-3B-lr_0.001",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.001
22
+ },
23
+ "lr_scheduler": "cosine",
24
+ "use_amp": false,
25
+ "autocast_adapter_dtype": true,
26
+ "generation_kwargs": {
27
+ "max_length": 800,
28
+ "max_new_tokens": 300
29
+ },
30
+ "attn_implementation": null
31
+ },
32
+ "peft_config": {
33
+ "task_type": null,
34
+ "peft_type": "IA3",
35
+ "auto_mapping": null,
36
+ "peft_version": "0.18.1.dev0@UNKNOWN",
37
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
38
+ "revision": null,
39
+ "inference_mode": false,
40
+ "target_modules": [
41
+ "down_proj",
42
+ "k_proj",
43
+ "v_proj"
44
+ ],
45
+ "exclude_modules": null,
46
+ "feedforward_modules": [
47
+ "down_proj"
48
+ ],
49
+ "fan_in_fan_out": false,
50
+ "modules_to_save": null,
51
+ "init_ia3_weights": true
52
+ },
53
+ "error_msg": ""
54
+ },
55
+ "train_info": {
56
+ "accelerator_memory_reserved_avg": 14786832524,
57
+ "accelerator_memory_max": 23240638464,
58
+ "accelerator_memory_reserved_99th": 20933771264,
59
+ "train_time": 969.7345388169051,
60
+ "file_size": 1157064,
61
+ "num_trainable_params": 286720,
62
+ "num_total_params": 3213036544,
63
+ "status": "success",
64
+ "metrics": [
65
+ {
66
+ "step": 250,
67
+ "valid accuracy": 0.2,
68
+ "train loss": 1.16681476187706,
69
+ "train samples": 1000,
70
+ "train time": 29.673775453935377,
71
+ "eval time": 11.714410117012449,
72
+ "tokens / sec": 7134.885829700566,
73
+ "mem allocated avg": 6782600361.984,
74
+ "mem reserved avg": 14885492621.312,
75
+ "elapsed time": 64.35547732701525
76
+ },
77
+ {
78
+ "step": 500,
79
+ "valid accuracy": 0.34,
80
+ "train loss": 0.8284487807750702,
81
+ "train samples": 2000,
82
+ "train time": 29.442405258771032,
83
+ "eval time": 11.714618961967062,
84
+ "tokens / sec": 7064.470384532775,
85
+ "mem allocated avg": 6775425632.256,
86
+ "mem reserved avg": 14480272523.264,
87
+ "elapsed time": 109.20030425902223
88
+ },
89
+ {
90
+ "step": 750,
91
+ "valid accuracy": 0.34,
92
+ "train loss": 0.7387913703918457,
93
+ "train samples": 3000,
94
+ "train time": 30.728236329741776,
95
+ "eval time": 7.293018446012866,
96
+ "tokens / sec": 6977.328529346211,
97
+ "mem allocated avg": 6785668968.448,
98
+ "mem reserved avg": 14756492607.488,
99
+ "elapsed time": 150.8779634690145
100
+ },
101
+ {
102
+ "step": 1000,
103
+ "valid accuracy": 0.36,
104
+ "train loss": 0.7031073143482208,
105
+ "train samples": 4000,
106
+ "train time": 29.568247662391514,
107
+ "eval time": 6.950251807982568,
108
+ "tokens / sec": 7045.936654034017,
109
+ "mem allocated avg": 6777244168.192,
110
+ "mem reserved avg": 14776985976.832,
111
+ "elapsed time": 191.10305509302998
112
+ },
113
+ {
114
+ "step": 1250,
115
+ "valid accuracy": 0.3,
116
+ "train loss": 0.6952721246480942,
117
+ "train samples": 5000,
118
+ "train time": 29.136579907499254,
119
+ "eval time": 7.7922670019906946,
120
+ "tokens / sec": 7157.257326084655,
121
+ "mem allocated avg": 6777135923.2,
122
+ "mem reserved avg": 14839564992.512,
123
+ "elapsed time": 231.62949198303977
124
+ },
125
+ {
126
+ "step": 1500,
127
+ "valid accuracy": 0.28,
128
+ "train loss": 0.6860846598148346,
129
+ "train samples": 6000,
130
+ "train time": 29.309486759477295,
131
+ "eval time": 8.780186134041287,
132
+ "tokens / sec": 7142.0902630549235,
133
+ "mem allocated avg": 6779035060.224,
134
+ "mem reserved avg": 14722384527.36,
135
+ "elapsed time": 273.4239038110245
136
+ },
137
+ {
138
+ "step": 1750,
139
+ "valid accuracy": 0.34,
140
+ "train loss": 0.677450892329216,
141
+ "train samples": 7000,
142
+ "train time": 29.44055196386762,
143
+ "eval time": 9.34235847304808,
144
+ "tokens / sec": 7111.109881939079,
145
+ "mem allocated avg": 6779937654.784,
146
+ "mem reserved avg": 15161679151.104,
147
+ "elapsed time": 315.85002974100644
148
+ },
149
+ {
150
+ "step": 2000,
151
+ "valid accuracy": 0.32,
152
+ "train loss": 0.6783282169103623,
153
+ "train samples": 8000,
154
+ "train time": 29.261146096221637,
155
+ "eval time": 8.788515535998158,
156
+ "tokens / sec": 7098.013157687588,
157
+ "mem allocated avg": 6776053977.088,
158
+ "mem reserved avg": 14763992023.04,
159
+ "elapsed time": 357.609911208041
160
+ },
161
+ {
162
+ "step": 2250,
163
+ "valid accuracy": 0.48,
164
+ "train loss": 0.6720160303115845,
165
+ "train samples": 9000,
166
+ "train time": 30.398827165656257,
167
+ "eval time": 10.552425866015255,
168
+ "tokens / sec": 7070.930691788078,
169
+ "mem allocated avg": 6787896084.48,
170
+ "mem reserved avg": 15088715038.72,
171
+ "elapsed time": 402.1721519330167
172
+ },
173
+ {
174
+ "step": 2500,
175
+ "valid accuracy": 0.44,
176
+ "train loss": 0.6704732036590576,
177
+ "train samples": 10000,
178
+ "train time": 29.788877881946974,
179
+ "eval time": 8.030864721979015,
180
+ "tokens / sec": 6914.224859903927,
181
+ "mem allocated avg": 6772744847.36,
182
+ "mem reserved avg": 14541761019.904,
183
+ "elapsed time": 443.68974046903895
184
+ },
185
+ {
186
+ "step": 2750,
187
+ "valid accuracy": 0.44,
188
+ "train loss": 0.6629449647665023,
189
+ "train samples": 11000,
190
+ "train time": 30.050681999477092,
191
+ "eval time": 8.138099809992127,
192
+ "tokens / sec": 7050.7883982029725,
193
+ "mem allocated avg": 6783104911.36,
194
+ "mem reserved avg": 14943508234.24,
195
+ "elapsed time": 485.477781796013
196
+ },
197
+ {
198
+ "step": 3000,
199
+ "valid accuracy": 0.38,
200
+ "train loss": 0.6557633522748947,
201
+ "train samples": 12000,
202
+ "train time": 29.61273902398534,
203
+ "eval time": 10.074375847005285,
204
+ "tokens / sec": 7048.689411369032,
205
+ "mem allocated avg": 6778547337.216,
206
+ "mem reserved avg": 14865569677.312,
207
+ "elapsed time": 528.8621876270045
208
+ },
209
+ {
210
+ "step": 3250,
211
+ "valid accuracy": 0.48,
212
+ "train loss": 0.6654598761796952,
213
+ "train samples": 13000,
214
+ "train time": 29.63859097607201,
215
+ "eval time": 7.940114627999719,
216
+ "tokens / sec": 7115.756621840281,
217
+ "mem allocated avg": 6780569153.536,
218
+ "mem reserved avg": 14715774304.256,
219
+ "elapsed time": 569.9359965579933
220
+ },
221
+ {
222
+ "step": 3500,
223
+ "valid accuracy": 0.46,
224
+ "train loss": 0.6528869969844818,
225
+ "train samples": 14000,
226
+ "train time": 30.20164098625537,
227
+ "eval time": 11.715904708020389,
228
+ "tokens / sec": 6944.986866622787,
229
+ "mem allocated avg": 6778699649.024,
230
+ "mem reserved avg": 14727702904.832,
231
+ "elapsed time": 615.5466552060097
232
+ },
233
+ {
234
+ "step": 3750,
235
+ "valid accuracy": 0.48,
236
+ "train loss": 0.6514250497817993,
237
+ "train samples": 15000,
238
+ "train time": 31.03051795932697,
239
+ "eval time": 9.986055641958956,
240
+ "tokens / sec": 6983.544402450578,
241
+ "mem allocated avg": 6790036742.144,
242
+ "mem reserved avg": 15097195921.408,
243
+ "elapsed time": 660.206704717013
244
+ },
245
+ {
246
+ "step": 4000,
247
+ "valid accuracy": 0.42,
248
+ "train loss": 0.6659475769996643,
249
+ "train samples": 16000,
250
+ "train time": 29.327483134518843,
251
+ "eval time": 11.69795022701146,
252
+ "tokens / sec": 6968.651181643686,
253
+ "mem allocated avg": 6770827464.704,
254
+ "mem reserved avg": 14748657647.616,
255
+ "elapsed time": 704.8852831150289
256
+ },
257
+ {
258
+ "step": 4250,
259
+ "valid accuracy": 0.42,
260
+ "train loss": 0.6488557003736496,
261
+ "train samples": 17000,
262
+ "train time": 29.780065669328906,
263
+ "eval time": 11.67279983102344,
264
+ "tokens / sec": 7098.338947509905,
265
+ "mem allocated avg": 6782400247.808,
266
+ "mem reserved avg": 14721461780.48,
267
+ "elapsed time": 749.8611966730095
268
+ },
269
+ {
270
+ "step": 4500,
271
+ "valid accuracy": 0.4,
272
+ "train loss": 0.6574098072052001,
273
+ "train samples": 18000,
274
+ "train time": 29.94875920598861,
275
+ "eval time": 7.102646674029529,
276
+ "tokens / sec": 6939.11886534666,
277
+ "mem allocated avg": 6776485750.784,
278
+ "mem reserved avg": 14703325609.984,
279
+ "elapsed time": 790.6740298850345
280
+ },
281
+ {
282
+ "step": 4750,
283
+ "valid accuracy": 0.44,
284
+ "train loss": 0.6505107415914535,
285
+ "train samples": 19000,
286
+ "train time": 30.134280904079787,
287
+ "eval time": 7.6594287430052646,
288
+ "tokens / sec": 6966.783135401682,
289
+ "mem allocated avg": 6779605635.072,
290
+ "mem reserved avg": 14800801234.944,
291
+ "elapsed time": 832.055415440991
292
+ },
293
+ {
294
+ "step": 5000,
295
+ "valid accuracy": 0.38,
296
+ "train loss": 0.6568637717962265,
297
+ "train samples": 20000,
298
+ "train time": 29.94096938223811,
299
+ "eval time": 9.314888863998931,
300
+ "tokens / sec": 6956.354596974339,
301
+ "mem allocated avg": 6776251918.336,
302
+ "mem reserved avg": 14395312701.44,
303
+ "elapsed time": 874.9385239290423
304
+ },
305
+ {
306
+ "step": 5000,
307
+ "test accuracy": 0.41243366186504926,
308
+ "train loss": 0.6568637717962265,
309
+ "train samples": 20000,
310
+ "train total tokens": 4198051,
311
+ "forgetting": 0.2050151824951172
312
+ }
313
+ ]
314
+ },
315
+ "meta_info": {
316
+ "model_info": {
317
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
318
+ "created_at": "2024-09-18T15:23:48+00:00"
319
+ },
320
+ "dataset_info": {
321
+ "metamath": {
322
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
323
+ "created_at": "2023-09-21T17:22:46+00:00"
324
+ },
325
+ "gsm8k": {
326
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
327
+ "created_at": "2022-04-12T10:22:10+00:00"
328
+ }
329
+ },
330
+ "package_info": {
331
+ "transformers-version": "4.57.1",
332
+ "transformers-commit-hash": null,
333
+ "peft-version": "0.18.1.dev0",
334
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
335
+ "datasets-version": "4.2.0",
336
+ "datasets-commit-hash": null,
337
+ "bitsandbytes-version": "0.46.0",
338
+ "bitsandbytes-commit-hash": null,
339
+ "torch-version": "2.9.0+cu128",
340
+ "torch-commit-hash": null
341
+ },
342
+ "system_info": {
343
+ "system": "Linux",
344
+ "release": "6.14.0-1016-aws",
345
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
346
+ "machine": "x86_64",
347
+ "processor": "x86_64",
348
+ "accelerator": "NVIDIA L40S"
349
+ },
350
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
351
+ }
352
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/ln_tuning--llama-3.2-3B-default.json ADDED
@@ -0,0 +1,348 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-10T01:02:15+00:00",
4
+ "total_time": 1158.007769990014,
5
+ "experiment_name": "ln_tuning/llama-3.2-3B-default",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "LN_TUNING",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "target_modules": [
42
+ "norm",
43
+ "input_layernorm",
44
+ "post_attention_layernorm"
45
+ ],
46
+ "exclude_modules": null,
47
+ "modules_to_save": null
48
+ },
49
+ "error_msg": ""
50
+ },
51
+ "train_info": {
52
+ "accelerator_memory_reserved_avg": 13811999938,
53
+ "accelerator_memory_max": 21179138048,
54
+ "accelerator_memory_reserved_99th": 19146997760,
55
+ "train_time": 933.4248713181005,
56
+ "file_size": 358288,
57
+ "num_trainable_params": 175104,
58
+ "num_total_params": 3212924928,
59
+ "status": "success",
60
+ "metrics": [
61
+ {
62
+ "step": 250,
63
+ "valid accuracy": 0.0,
64
+ "train loss": 1.3265766797065734,
65
+ "train samples": 1000,
66
+ "train time": 27.05523074005032,
67
+ "eval time": 11.134952866006643,
68
+ "tokens / sec": 7825.436864102909,
69
+ "mem allocated avg": 6781727356.928,
70
+ "mem reserved avg": 13917774413.824,
71
+ "elapsed time": 59.58366395003395
72
+ },
73
+ {
74
+ "step": 500,
75
+ "valid accuracy": 0.0,
76
+ "train loss": 1.3412033476829528,
77
+ "train samples": 2000,
78
+ "train time": 27.19167830166407,
79
+ "eval time": 11.076495647954289,
80
+ "tokens / sec": 7649.215237563,
81
+ "mem allocated avg": 6774720333.824,
82
+ "mem reserved avg": 13569009647.616,
83
+ "elapsed time": 100.96434171200963
84
+ },
85
+ {
86
+ "step": 750,
87
+ "valid accuracy": 0.0,
88
+ "train loss": 1.2789963104724884,
89
+ "train samples": 3000,
90
+ "train time": 27.539981623878703,
91
+ "eval time": 11.047948513005394,
92
+ "tokens / sec": 7785.081447334822,
93
+ "mem allocated avg": 6784869052.416,
94
+ "mem reserved avg": 13769388326.912,
95
+ "elapsed time": 142.60399866703665
96
+ },
97
+ {
98
+ "step": 1000,
99
+ "valid accuracy": 0.0,
100
+ "train loss": 1.2634205253124238,
101
+ "train samples": 4000,
102
+ "train time": 26.802510203735437,
103
+ "eval time": 11.213655479019508,
104
+ "tokens / sec": 7773.00329022781,
105
+ "mem allocated avg": 6775999932.416,
106
+ "mem reserved avg": 13779588874.24,
107
+ "elapsed time": 183.79537835903466
108
+ },
109
+ {
110
+ "step": 1250,
111
+ "valid accuracy": 0.0,
112
+ "train loss": 1.2484288651943207,
113
+ "train samples": 5000,
114
+ "train time": 26.859816205396783,
115
+ "eval time": 11.04810167697724,
116
+ "tokens / sec": 7763.939946770734,
117
+ "mem allocated avg": 6776329017.344,
118
+ "mem reserved avg": 13803437686.784,
119
+ "elapsed time": 224.77404374204343
120
+ },
121
+ {
122
+ "step": 1500,
123
+ "valid accuracy": 0.0,
124
+ "train loss": 1.231340999364853,
125
+ "train samples": 6000,
126
+ "train time": 27.544580540037714,
127
+ "eval time": 11.118913470010739,
128
+ "tokens / sec": 7599.716383254584,
129
+ "mem allocated avg": 6778265929.728,
130
+ "mem reserved avg": 13820894380.032,
131
+ "elapsed time": 266.58437842602143
132
+ },
133
+ {
134
+ "step": 1750,
135
+ "valid accuracy": 0.0,
136
+ "train loss": 1.2306929278373717,
137
+ "train samples": 7000,
138
+ "train time": 27.029298119421583,
139
+ "eval time": 11.036927138979081,
140
+ "tokens / sec": 7745.484143725154,
141
+ "mem allocated avg": 6779571120.128,
142
+ "mem reserved avg": 14093188595.712,
143
+ "elapsed time": 307.76138371700654
144
+ },
145
+ {
146
+ "step": 2000,
147
+ "valid accuracy": 0.0,
148
+ "train loss": 1.2307198526859284,
149
+ "train samples": 8000,
150
+ "train time": 26.702352902852,
151
+ "eval time": 11.090556368988473,
152
+ "tokens / sec": 7778.1909615094855,
153
+ "mem allocated avg": 6775307636.736,
154
+ "mem reserved avg": 13810257625.088,
155
+ "elapsed time": 348.7669416820281
156
+ },
157
+ {
158
+ "step": 2250,
159
+ "valid accuracy": 0.0,
160
+ "train loss": 1.2245776391029357,
161
+ "train samples": 9000,
162
+ "train time": 27.69884210533928,
163
+ "eval time": 11.05135836795671,
164
+ "tokens / sec": 7760.179980901304,
165
+ "mem allocated avg": 6785839908.864,
166
+ "mem reserved avg": 14071948640.256,
167
+ "elapsed time": 390.5765162970056
168
+ },
169
+ {
170
+ "step": 2500,
171
+ "valid accuracy": 0.0,
172
+ "train loss": 1.2369521839618682,
173
+ "train samples": 10000,
174
+ "train time": 27.172044244071003,
175
+ "eval time": 11.106051086972002,
176
+ "tokens / sec": 7580.10689773341,
177
+ "mem allocated avg": 6772880177.152,
178
+ "mem reserved avg": 13577423421.44,
179
+ "elapsed time": 431.9964616240468
180
+ },
181
+ {
182
+ "step": 2750,
183
+ "valid accuracy": 0.0,
184
+ "train loss": 1.2313981459140777,
185
+ "train samples": 11000,
186
+ "train time": 27.588061350223143,
187
+ "eval time": 11.063589482975658,
188
+ "tokens / sec": 7680.1699586725845,
189
+ "mem allocated avg": 6782132961.28,
190
+ "mem reserved avg": 13968802316.288,
191
+ "elapsed time": 473.7305173620116
192
+ },
193
+ {
194
+ "step": 3000,
195
+ "valid accuracy": 0.0,
196
+ "train loss": 1.2322822377681732,
197
+ "train samples": 12000,
198
+ "train time": 27.48076619656058,
199
+ "eval time": 11.10096250998322,
200
+ "tokens / sec": 7595.53057971594,
201
+ "mem allocated avg": 6777935728.64,
202
+ "mem reserved avg": 13851705737.216,
203
+ "elapsed time": 515.4748477530084
204
+ },
205
+ {
206
+ "step": 3250,
207
+ "valid accuracy": 0.0,
208
+ "train loss": 1.231231040239334,
209
+ "train samples": 13000,
210
+ "train time": 27.788235317741055,
211
+ "eval time": 11.049090766988229,
212
+ "tokens / sec": 7589.578740372652,
213
+ "mem allocated avg": 6780943562.752,
214
+ "mem reserved avg": 13802355556.352,
215
+ "elapsed time": 557.366414040036
216
+ },
217
+ {
218
+ "step": 3500,
219
+ "valid accuracy": 0.0,
220
+ "train loss": 1.2321618583202363,
221
+ "train samples": 14000,
222
+ "train time": 27.022960201604292,
223
+ "eval time": 11.094650793995243,
224
+ "tokens / sec": 7761.917955515015,
225
+ "mem allocated avg": 6777875570.688,
226
+ "mem reserved avg": 13817966755.84,
227
+ "elapsed time": 598.6570834580343
228
+ },
229
+ {
230
+ "step": 3750,
231
+ "valid accuracy": 0.0,
232
+ "train loss": 1.215895681142807,
233
+ "train samples": 15000,
234
+ "train time": 27.39095462608384,
235
+ "eval time": 11.043718303029891,
236
+ "tokens / sec": 7911.4803758478065,
237
+ "mem allocated avg": 6788866940.928,
238
+ "mem reserved avg": 14072342904.832,
239
+ "elapsed time": 640.2053242530092
240
+ },
241
+ {
242
+ "step": 4000,
243
+ "valid accuracy": 0.0,
244
+ "train loss": 1.2472150621414184,
245
+ "train samples": 16000,
246
+ "train time": 26.27257864776766,
247
+ "eval time": 11.109217509045266,
248
+ "tokens / sec": 7778.947119732583,
249
+ "mem allocated avg": 6769699028.992,
250
+ "mem reserved avg": 13815643111.424,
251
+ "elapsed time": 680.7284473860054
252
+ },
253
+ {
254
+ "step": 4250,
255
+ "valid accuracy": 0.0,
256
+ "train loss": 1.21587810254097,
257
+ "train samples": 17000,
258
+ "train time": 26.962625220243353,
259
+ "eval time": 11.047069397987798,
260
+ "tokens / sec": 7840.074854480067,
261
+ "mem allocated avg": 6782053310.464,
262
+ "mem reserved avg": 13778800345.088,
263
+ "elapsed time": 721.68619730504
264
+ },
265
+ {
266
+ "step": 4500,
267
+ "valid accuracy": 0.0,
268
+ "train loss": 1.241484440803528,
269
+ "train samples": 18000,
270
+ "train time": 26.840785892680287,
271
+ "eval time": 11.098841341969091,
272
+ "tokens / sec": 7742.61978881452,
273
+ "mem allocated avg": 6775733297.152,
274
+ "mem reserved avg": 13672147582.976,
275
+ "elapsed time": 762.8068605409935
276
+ },
277
+ {
278
+ "step": 4750,
279
+ "valid accuracy": 0.0,
280
+ "train loss": 1.2253197972774506,
281
+ "train samples": 19000,
282
+ "train time": 27.470003902097233,
283
+ "eval time": 11.055242774018552,
284
+ "tokens / sec": 7642.481622799185,
285
+ "mem allocated avg": 6778600820.736,
286
+ "mem reserved avg": 13782675881.984,
287
+ "elapsed time": 804.4074719570344
288
+ },
289
+ {
290
+ "step": 5000,
291
+ "valid accuracy": 0.0,
292
+ "train loss": 1.2366444897651672,
293
+ "train samples": 20000,
294
+ "train time": 26.569939231791068,
295
+ "eval time": 11.095670820039231,
296
+ "tokens / sec": 7838.93399917121,
297
+ "mem allocated avg": 6775143475.2,
298
+ "mem reserved avg": 13464646975.488,
299
+ "elapsed time": 845.1059082580032
300
+ },
301
+ {
302
+ "step": 5000,
303
+ "test accuracy": 0.0,
304
+ "train loss": 1.2366444897651672,
305
+ "train samples": 20000,
306
+ "train total tokens": 4198051,
307
+ "forgetting": -0.032855987548828125
308
+ }
309
+ ]
310
+ },
311
+ "meta_info": {
312
+ "model_info": {
313
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
314
+ "created_at": "2024-09-18T15:23:48+00:00"
315
+ },
316
+ "dataset_info": {
317
+ "metamath": {
318
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
319
+ "created_at": "2023-09-21T17:22:46+00:00"
320
+ },
321
+ "gsm8k": {
322
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
323
+ "created_at": "2022-04-12T10:22:10+00:00"
324
+ }
325
+ },
326
+ "package_info": {
327
+ "transformers-version": "4.57.1",
328
+ "transformers-commit-hash": null,
329
+ "peft-version": "0.18.1.dev0",
330
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
331
+ "datasets-version": "4.2.0",
332
+ "datasets-commit-hash": null,
333
+ "bitsandbytes-version": "0.46.0",
334
+ "bitsandbytes-commit-hash": null,
335
+ "torch-version": "2.9.0+cu128",
336
+ "torch-commit-hash": null
337
+ },
338
+ "system_info": {
339
+ "system": "Linux",
340
+ "release": "6.14.0-1016-aws",
341
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
342
+ "machine": "x86_64",
343
+ "processor": "x86_64",
344
+ "accelerator": "NVIDIA L40S"
345
+ },
346
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
347
+ }
348
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/loha--llama-3.2-3B-rank32.json ADDED
@@ -0,0 +1,357 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-10T01:21:37+00:00",
4
+ "total_time": 1723.7877959029865,
5
+ "experiment_name": "loha/llama-3.2-3B-rank32",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "LOHA",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "rank_pattern": {},
42
+ "alpha_pattern": {},
43
+ "r": 32,
44
+ "alpha": 64,
45
+ "rank_dropout": 0.0,
46
+ "module_dropout": 0.0,
47
+ "use_effective_conv2d": false,
48
+ "target_modules": [
49
+ "v_proj",
50
+ "q_proj"
51
+ ],
52
+ "exclude_modules": null,
53
+ "init_weights": true,
54
+ "layers_to_transform": null,
55
+ "layers_pattern": null,
56
+ "modules_to_save": null
57
+ },
58
+ "error_msg": ""
59
+ },
60
+ "train_info": {
61
+ "accelerator_memory_reserved_avg": 16069747225,
62
+ "accelerator_memory_max": 23907532800,
63
+ "accelerator_memory_reserved_99th": 21843935232,
64
+ "train_time": 1444.0684821169707,
65
+ "file_size": 73429560,
66
+ "num_trainable_params": 18350080,
67
+ "num_total_params": 3231099904,
68
+ "status": "success",
69
+ "metrics": [
70
+ {
71
+ "step": 250,
72
+ "valid accuracy": 0.0,
73
+ "train loss": 1.2914833612442016,
74
+ "train samples": 1000,
75
+ "train time": 48.783409356023185,
76
+ "eval time": 15.048279424023349,
77
+ "tokens / sec": 4339.979570818157,
78
+ "mem allocated avg": 7075144325.12,
79
+ "mem reserved avg": 16180198768.64,
80
+ "elapsed time": 89.06116562400712
81
+ },
82
+ {
83
+ "step": 500,
84
+ "valid accuracy": 0.36,
85
+ "train loss": 0.9051123208999634,
86
+ "train samples": 2000,
87
+ "train time": 49.313777334056795,
88
+ "eval time": 14.961491579946596,
89
+ "tokens / sec": 4217.786818296632,
90
+ "mem allocated avg": 7068229634.048,
91
+ "mem reserved avg": 15816938487.808,
92
+ "elapsed time": 157.51611204899382
93
+ },
94
+ {
95
+ "step": 750,
96
+ "valid accuracy": 0.32,
97
+ "train loss": 0.7515957379341125,
98
+ "train samples": 3000,
99
+ "train time": 49.29726693738485,
100
+ "eval time": 10.4420769319986,
101
+ "tokens / sec": 4349.1457705418525,
102
+ "mem allocated avg": 7078224801.792,
103
+ "mem reserved avg": 16001924071.424,
104
+ "elapsed time": 221.30589306500042
105
+ },
106
+ {
107
+ "step": 1000,
108
+ "valid accuracy": 0.44,
109
+ "train loss": 0.7083757585287094,
110
+ "train samples": 4000,
111
+ "train time": 48.992843332118355,
112
+ "eval time": 15.014163143991027,
113
+ "tokens / sec": 4252.376180490441,
114
+ "mem allocated avg": 7069774227.456,
115
+ "mem reserved avg": 16085097119.744,
116
+ "elapsed time": 289.58009328099433
117
+ },
118
+ {
119
+ "step": 1250,
120
+ "valid accuracy": 0.36,
121
+ "train loss": 0.6994501632452012,
122
+ "train samples": 5000,
123
+ "train time": 48.87461844785139,
124
+ "eval time": 15.003678467997815,
125
+ "tokens / sec": 4266.795457902294,
126
+ "mem allocated avg": 7069489561.6,
127
+ "mem reserved avg": 16102310543.36,
128
+ "elapsed time": 357.60247106896713
129
+ },
130
+ {
131
+ "step": 1500,
132
+ "valid accuracy": 0.34,
133
+ "train loss": 0.688896736741066,
134
+ "train samples": 6000,
135
+ "train time": 48.57590564060956,
136
+ "eval time": 10.066342948994134,
137
+ "tokens / sec": 4309.358667417182,
138
+ "mem allocated avg": 7071214790.656,
139
+ "mem reserved avg": 16085852094.464,
140
+ "elapsed time": 420.47844035195885
141
+ },
142
+ {
143
+ "step": 1750,
144
+ "valid accuracy": 0.4,
145
+ "train loss": 0.6796333414316177,
146
+ "train samples": 7000,
147
+ "train time": 49.9453335700091,
148
+ "eval time": 11.371492591046263,
149
+ "tokens / sec": 4191.682886781486,
150
+ "mem allocated avg": 7073070399.488,
151
+ "mem reserved avg": 16412395438.08,
152
+ "elapsed time": 485.994160362985
153
+ },
154
+ {
155
+ "step": 2000,
156
+ "valid accuracy": 0.42,
157
+ "train loss": 0.6801165891885758,
158
+ "train samples": 8000,
159
+ "train time": 48.455957927857526,
160
+ "eval time": 15.107590888044797,
161
+ "tokens / sec": 4286.2840583860325,
162
+ "mem allocated avg": 7068561985.536,
163
+ "mem reserved avg": 16088888770.56,
164
+ "elapsed time": 553.8159226219868
165
+ },
166
+ {
167
+ "step": 2250,
168
+ "valid accuracy": 0.38,
169
+ "train loss": 0.6730777739286423,
170
+ "train samples": 9000,
171
+ "train time": 50.08038655488053,
172
+ "eval time": 15.016238837968558,
173
+ "tokens / sec": 4292.059522432988,
174
+ "mem allocated avg": 7080170047.488,
175
+ "mem reserved avg": 16375955324.928,
176
+ "elapsed time": 623.1283458310063
177
+ },
178
+ {
179
+ "step": 2500,
180
+ "valid accuracy": 0.44,
181
+ "train loss": 0.6711871980428695,
182
+ "train samples": 10000,
183
+ "train time": 48.26023929938674,
184
+ "eval time": 8.185648262035102,
185
+ "tokens / sec": 4267.840420812362,
186
+ "mem allocated avg": 7066054148.096,
187
+ "mem reserved avg": 15799381131.264,
188
+ "elapsed time": 683.6707712119678
189
+ },
190
+ {
191
+ "step": 2750,
192
+ "valid accuracy": 0.44,
193
+ "train loss": 0.6645230972766876,
194
+ "train samples": 11000,
195
+ "train time": 50.41757245978806,
196
+ "eval time": 11.489728196000215,
197
+ "tokens / sec": 4202.522843974521,
198
+ "mem allocated avg": 7075331424.256,
199
+ "mem reserved avg": 16211597328.384,
200
+ "elapsed time": 749.7390187569545
201
+ },
202
+ {
203
+ "step": 3000,
204
+ "valid accuracy": 0.4,
205
+ "train loss": 0.6566754117012024,
206
+ "train samples": 12000,
207
+ "train time": 48.205230054212734,
208
+ "eval time": 14.944865885016043,
209
+ "tokens / sec": 4330.048830080392,
210
+ "mem allocated avg": 7070783477.76,
211
+ "mem reserved avg": 16134405357.568,
212
+ "elapsed time": 817.0802311009611
213
+ },
214
+ {
215
+ "step": 3250,
216
+ "valid accuracy": 0.42,
217
+ "train loss": 0.6666845008134842,
218
+ "train samples": 13000,
219
+ "train time": 48.50475694506895,
220
+ "eval time": 14.985318146995269,
221
+ "tokens / sec": 4348.047764445925,
222
+ "mem allocated avg": 7073913442.304,
223
+ "mem reserved avg": 16018088919.04,
224
+ "elapsed time": 884.7171556460089
225
+ },
226
+ {
227
+ "step": 3500,
228
+ "valid accuracy": 0.44,
229
+ "train loss": 0.6538506431579589,
230
+ "train samples": 14000,
231
+ "train time": 48.627094919618685,
232
+ "eval time": 14.961251400003675,
233
+ "tokens / sec": 4313.438842001972,
234
+ "mem allocated avg": 7070698022.912,
235
+ "mem reserved avg": 16014808973.312,
236
+ "elapsed time": 952.4853859379655
237
+ },
238
+ {
239
+ "step": 3750,
240
+ "valid accuracy": 0.44,
241
+ "train loss": 0.6510264467000961,
242
+ "train samples": 15000,
243
+ "train time": 50.97888733900618,
244
+ "eval time": 15.012431181967258,
245
+ "tokens / sec": 4250.838166767737,
246
+ "mem allocated avg": 7083094827.008,
247
+ "mem reserved avg": 16354790866.944,
248
+ "elapsed time": 1022.6562416799716
249
+ },
250
+ {
251
+ "step": 4000,
252
+ "valid accuracy": 0.46,
253
+ "train loss": 0.6674902982711792,
254
+ "train samples": 16000,
255
+ "train time": 48.719796552322805,
256
+ "eval time": 15.030813960998785,
257
+ "tokens / sec": 4194.865628810926,
258
+ "mem allocated avg": 7063056650.24,
259
+ "mem reserved avg": 16042550099.968,
260
+ "elapsed time": 1090.629200581985
261
+ },
262
+ {
263
+ "step": 4250,
264
+ "valid accuracy": 0.48,
265
+ "train loss": 0.6494669053554535,
266
+ "train samples": 17000,
267
+ "train time": 50.06373492471175,
268
+ "eval time": 15.01842499698978,
269
+ "tokens / sec": 4222.397715989367,
270
+ "mem allocated avg": 7074535757.824,
271
+ "mem reserved avg": 15991052435.456,
272
+ "elapsed time": 1159.7775298539782
273
+ },
274
+ {
275
+ "step": 4500,
276
+ "valid accuracy": 0.5,
277
+ "train loss": 0.658131462097168,
278
+ "train samples": 18000,
279
+ "train time": 49.55814389558509,
280
+ "eval time": 14.973931148997508,
281
+ "tokens / sec": 4193.417744576054,
282
+ "mem allocated avg": 7069541146.624,
283
+ "mem reserved avg": 15961180602.368,
284
+ "elapsed time": 1228.5485400629696
285
+ },
286
+ {
287
+ "step": 4750,
288
+ "valid accuracy": 0.48,
289
+ "train loss": 0.6511208745241165,
290
+ "train samples": 19000,
291
+ "train time": 50.09447881835513,
292
+ "eval time": 14.996958026022185,
293
+ "tokens / sec": 4190.861048005878,
294
+ "mem allocated avg": 7071928455.168,
295
+ "mem reserved avg": 16052733870.08,
296
+ "elapsed time": 1297.7279399959953
297
+ },
298
+ {
299
+ "step": 5000,
300
+ "valid accuracy": 0.44,
301
+ "train loss": 0.6570172024965286,
302
+ "train samples": 20000,
303
+ "train time": 48.30597296968335,
304
+ "eval time": 14.95854680496268,
305
+ "tokens / sec": 4311.682121188528,
306
+ "mem allocated avg": 7068810864.64,
307
+ "mem reserved avg": 15664794304.512,
308
+ "elapsed time": 1365.0393339970033
309
+ },
310
+ {
311
+ "step": 5000,
312
+ "test accuracy": 0.4184988627748294,
313
+ "train loss": 0.6570172024965286,
314
+ "train samples": 20000,
315
+ "train total tokens": 4198051,
316
+ "forgetting": 0.07485389709472656
317
+ }
318
+ ]
319
+ },
320
+ "meta_info": {
321
+ "model_info": {
322
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
323
+ "created_at": "2024-09-18T15:23:48+00:00"
324
+ },
325
+ "dataset_info": {
326
+ "metamath": {
327
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
328
+ "created_at": "2023-09-21T17:22:46+00:00"
329
+ },
330
+ "gsm8k": {
331
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
332
+ "created_at": "2022-04-12T10:22:10+00:00"
333
+ }
334
+ },
335
+ "package_info": {
336
+ "transformers-version": "4.57.1",
337
+ "transformers-commit-hash": null,
338
+ "peft-version": "0.18.1.dev0",
339
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
340
+ "datasets-version": "4.2.0",
341
+ "datasets-commit-hash": null,
342
+ "bitsandbytes-version": "0.46.0",
343
+ "bitsandbytes-commit-hash": null,
344
+ "torch-version": "2.9.0+cu128",
345
+ "torch-commit-hash": null
346
+ },
347
+ "system_info": {
348
+ "system": "Linux",
349
+ "release": "6.14.0-1016-aws",
350
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
351
+ "machine": "x86_64",
352
+ "processor": "x86_64",
353
+ "accelerator": "NVIDIA L40S"
354
+ },
355
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
356
+ }
357
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/lokr--llama-3.2-3B-rank32.json ADDED
@@ -0,0 +1,360 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-10T01:50:25+00:00",
4
+ "total_time": 1535.4195199839887,
5
+ "experiment_name": "lokr/llama-3.2-3B-rank32",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": null,
35
+ "peft_type": "LOKR",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "rank_pattern": {},
42
+ "alpha_pattern": {},
43
+ "r": 32,
44
+ "alpha": 64,
45
+ "rank_dropout": 0.0,
46
+ "module_dropout": 0.0,
47
+ "use_effective_conv2d": false,
48
+ "decompose_both": false,
49
+ "decompose_factor": -1,
50
+ "rank_dropout_scale": false,
51
+ "target_modules": [
52
+ "q_proj",
53
+ "v_proj"
54
+ ],
55
+ "exclude_modules": null,
56
+ "init_weights": true,
57
+ "layers_to_transform": null,
58
+ "layers_pattern": null,
59
+ "modules_to_save": null
60
+ },
61
+ "error_msg": ""
62
+ },
63
+ "train_info": {
64
+ "accelerator_memory_reserved_avg": 15792361544,
65
+ "accelerator_memory_max": 23571988480,
66
+ "accelerator_memory_reserved_99th": 21346910208,
67
+ "train_time": 1317.615543427004,
68
+ "file_size": 1131984,
69
+ "num_trainable_params": 279552,
70
+ "num_total_params": 3213029376,
71
+ "status": "success",
72
+ "metrics": [
73
+ {
74
+ "step": 250,
75
+ "valid accuracy": 0.0,
76
+ "train loss": 1.2610098366737366,
77
+ "train samples": 1000,
78
+ "train time": 44.917080010229256,
79
+ "eval time": 13.230089190998115,
80
+ "tokens / sec": 4713.552171062408,
81
+ "mem allocated avg": 6785872599.04,
82
+ "mem reserved avg": 15919304671.232,
83
+ "elapsed time": 85.93055745301535
84
+ },
85
+ {
86
+ "step": 500,
87
+ "valid accuracy": 0.36,
88
+ "train loss": 0.9418223905563354,
89
+ "train samples": 2000,
90
+ "train time": 44.39176498691086,
91
+ "eval time": 13.134529082977679,
92
+ "tokens / sec": 4685.441096143134,
93
+ "mem allocated avg": 6778738747.392,
94
+ "mem reserved avg": 15563761909.76,
95
+ "elapsed time": 147.61094936099835
96
+ },
97
+ {
98
+ "step": 750,
99
+ "valid accuracy": 0.4,
100
+ "train loss": 0.7931550116539001,
101
+ "train samples": 3000,
102
+ "train time": 45.546711406088434,
103
+ "eval time": 13.15151729201898,
104
+ "tokens / sec": 4707.2772848171,
105
+ "mem allocated avg": 6788886910.976,
106
+ "mem reserved avg": 15770750812.16,
107
+ "elapsed time": 210.37875085603446
108
+ },
109
+ {
110
+ "step": 1000,
111
+ "valid accuracy": 0.38,
112
+ "train loss": 0.7486971822977067,
113
+ "train samples": 4000,
114
+ "train time": 44.15514894755324,
115
+ "eval time": 13.2610618310282,
116
+ "tokens / sec": 4718.271933528252,
117
+ "mem allocated avg": 6779705090.048,
118
+ "mem reserved avg": 15782125764.608,
119
+ "elapsed time": 272.00523140100995
120
+ },
121
+ {
122
+ "step": 1250,
123
+ "valid accuracy": 0.34,
124
+ "train loss": 0.7329185563325882,
125
+ "train samples": 5000,
126
+ "train time": 44.73619283729931,
127
+ "eval time": 13.166668922000099,
128
+ "tokens / sec": 4661.505299711804,
129
+ "mem allocated avg": 6780468045.824,
130
+ "mem reserved avg": 15823901032.448,
131
+ "elapsed time": 333.9785253200098
132
+ },
133
+ {
134
+ "step": 1500,
135
+ "valid accuracy": 0.32,
136
+ "train loss": 0.7207690758705139,
137
+ "train samples": 6000,
138
+ "train time": 44.514215660397895,
139
+ "eval time": 13.139724303036928,
140
+ "tokens / sec": 4702.565166979489,
141
+ "mem allocated avg": 6781829281.792,
142
+ "mem reserved avg": 15757807190.016,
143
+ "elapsed time": 395.81570798699977
144
+ },
145
+ {
146
+ "step": 1750,
147
+ "valid accuracy": 0.3,
148
+ "train loss": 0.7103254743814469,
149
+ "train samples": 7000,
150
+ "train time": 45.324679988552816,
151
+ "eval time": 13.146416853996925,
152
+ "tokens / sec": 4619.006688031214,
153
+ "mem allocated avg": 6783828615.168,
154
+ "mem reserved avg": 16158497439.744,
155
+ "elapsed time": 458.46547377400566
156
+ },
157
+ {
158
+ "step": 2000,
159
+ "valid accuracy": 0.36,
160
+ "train loss": 0.7092406690120697,
161
+ "train samples": 8000,
162
+ "train time": 43.866437830845825,
163
+ "eval time": 13.131515079992823,
164
+ "tokens / sec": 4734.735945528569,
165
+ "mem allocated avg": 6779261771.776,
166
+ "mem reserved avg": 15762227986.432,
167
+ "elapsed time": 519.6737852150109
168
+ },
169
+ {
170
+ "step": 2250,
171
+ "valid accuracy": 0.4,
172
+ "train loss": 0.7017901630401612,
173
+ "train samples": 9000,
174
+ "train time": 46.34161411313107,
175
+ "eval time": 9.606278705003206,
176
+ "tokens / sec": 4638.336495471652,
177
+ "mem allocated avg": 6790719053.824,
178
+ "mem reserved avg": 16063865552.896,
179
+ "elapsed time": 579.7241231160006
180
+ },
181
+ {
182
+ "step": 2500,
183
+ "valid accuracy": 0.34,
184
+ "train loss": 0.7000315390825271,
185
+ "train samples": 10000,
186
+ "train time": 43.55846684699645,
187
+ "eval time": 13.126351199985947,
188
+ "tokens / sec": 4728.5181253849005,
189
+ "mem allocated avg": 6776950829.056,
190
+ "mem reserved avg": 15558267371.52,
191
+ "elapsed time": 640.5660474800388
192
+ },
193
+ {
194
+ "step": 2750,
195
+ "valid accuracy": 0.36,
196
+ "train loss": 0.6941497979164124,
197
+ "train samples": 11000,
198
+ "train time": 45.54542878107168,
199
+ "eval time": 7.833579002006445,
200
+ "tokens / sec": 4652.080475923768,
201
+ "mem allocated avg": 6786488795.136,
202
+ "mem reserved avg": 15941710643.2,
203
+ "elapsed time": 698.0084676230326
204
+ },
205
+ {
206
+ "step": 3000,
207
+ "valid accuracy": 0.38,
208
+ "train loss": 0.6868689198493958,
209
+ "train samples": 12000,
210
+ "train time": 44.477772765094414,
211
+ "eval time": 13.114933934004512,
212
+ "tokens / sec": 4692.928333043902,
213
+ "mem allocated avg": 6781943875.584,
214
+ "mem reserved avg": 15848840364.032,
215
+ "elapsed time": 759.7957223660196
216
+ },
217
+ {
218
+ "step": 3250,
219
+ "valid accuracy": 0.36,
220
+ "train loss": 0.6960364075899124,
221
+ "train samples": 13000,
222
+ "train time": 45.38897958409507,
223
+ "eval time": 8.616416077013128,
224
+ "tokens / sec": 4646.524375134943,
225
+ "mem allocated avg": 6784858740.736,
226
+ "mem reserved avg": 15725200670.72,
227
+ "elapsed time": 817.8723878910532
228
+ },
229
+ {
230
+ "step": 3500,
231
+ "valid accuracy": 0.4,
232
+ "train loss": 0.6833818035125733,
233
+ "train samples": 14000,
234
+ "train time": 43.60606637201272,
235
+ "eval time": 7.672805098001845,
236
+ "tokens / sec": 4810.110552292832,
237
+ "mem allocated avg": 6782173394.944,
238
+ "mem reserved avg": 15725863370.752,
239
+ "elapsed time": 873.291599542019
240
+ },
241
+ {
242
+ "step": 3750,
243
+ "valid accuracy": 0.36,
244
+ "train loss": 0.6804632662534714,
245
+ "train samples": 15000,
246
+ "train time": 44.50026800244814,
247
+ "eval time": 8.241878828965127,
248
+ "tokens / sec": 4869.70100917321,
249
+ "mem allocated avg": 6793631014.912,
250
+ "mem reserved avg": 16037810536.448,
251
+ "elapsed time": 930.1443349370384
252
+ },
253
+ {
254
+ "step": 4000,
255
+ "valid accuracy": 0.36,
256
+ "train loss": 0.6991109738349914,
257
+ "train samples": 16000,
258
+ "train time": 42.83764784556115,
259
+ "eval time": 7.653670433035586,
260
+ "tokens / sec": 4770.8735254747935,
261
+ "mem allocated avg": 6774149611.52,
262
+ "mem reserved avg": 15730233835.52,
263
+ "elapsed time": 984.7777602920542
264
+ },
265
+ {
266
+ "step": 4250,
267
+ "valid accuracy": 0.38,
268
+ "train loss": 0.6788504232168198,
269
+ "train samples": 17000,
270
+ "train time": 44.51924567105016,
271
+ "eval time": 7.663433300971519,
272
+ "tokens / sec": 4748.261045614738,
273
+ "mem allocated avg": 6785154426.88,
274
+ "mem reserved avg": 15703524507.648,
275
+ "elapsed time": 1040.9056276420015
276
+ },
277
+ {
278
+ "step": 4500,
279
+ "valid accuracy": 0.38,
280
+ "train loss": 0.6892099533081054,
281
+ "train samples": 18000,
282
+ "train time": 44.30832453438779,
283
+ "eval time": 9.274187314964365,
284
+ "tokens / sec": 4690.269880069872,
285
+ "mem allocated avg": 6780451182.592,
286
+ "mem reserved avg": 15700773044.224,
287
+ "elapsed time": 1098.7072390830144
288
+ },
289
+ {
290
+ "step": 4750,
291
+ "valid accuracy": 0.4,
292
+ "train loss": 0.6815461959838868,
293
+ "train samples": 19000,
294
+ "train time": 43.495211453642696,
295
+ "eval time": 7.511790357995778,
296
+ "tokens / sec": 4826.715240222559,
297
+ "mem allocated avg": 6782852112.384,
298
+ "mem reserved avg": 15813943754.752,
299
+ "elapsed time": 1153.775285233045
300
+ },
301
+ {
302
+ "step": 5000,
303
+ "valid accuracy": 0.36,
304
+ "train loss": 0.6875366779565811,
305
+ "train samples": 20000,
306
+ "train time": 43.26518003502861,
307
+ "eval time": 9.198009185027331,
308
+ "tokens / sec": 4814.032897387025,
309
+ "mem allocated avg": 6779266990.08,
310
+ "mem reserved avg": 15458820423.68,
311
+ "elapsed time": 1210.3138140790397
312
+ },
313
+ {
314
+ "step": 5000,
315
+ "test accuracy": 0.379833206974981,
316
+ "train loss": 0.6875366779565811,
317
+ "train samples": 20000,
318
+ "train total tokens": 4198051,
319
+ "forgetting": 0.3052024841308594
320
+ }
321
+ ]
322
+ },
323
+ "meta_info": {
324
+ "model_info": {
325
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
326
+ "created_at": "2024-09-18T15:23:48+00:00"
327
+ },
328
+ "dataset_info": {
329
+ "metamath": {
330
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
331
+ "created_at": "2023-09-21T17:22:46+00:00"
332
+ },
333
+ "gsm8k": {
334
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
335
+ "created_at": "2022-04-12T10:22:10+00:00"
336
+ }
337
+ },
338
+ "package_info": {
339
+ "transformers-version": "4.57.1",
340
+ "transformers-commit-hash": null,
341
+ "peft-version": "0.18.1.dev0",
342
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
343
+ "datasets-version": "4.2.0",
344
+ "datasets-commit-hash": null,
345
+ "bitsandbytes-version": "0.46.0",
346
+ "bitsandbytes-commit-hash": null,
347
+ "torch-version": "2.9.0+cu128",
348
+ "torch-commit-hash": null
349
+ },
350
+ "system_info": {
351
+ "system": "Linux",
352
+ "release": "6.14.0-1016-aws",
353
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
354
+ "machine": "x86_64",
355
+ "processor": "x86_64",
356
+ "accelerator": "NVIDIA L40S"
357
+ },
358
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
359
+ }
360
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/lora--llama-3.2-3B-rank10-target-mlp.json ADDED
@@ -0,0 +1,375 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-10T02:16:05+00:00",
4
+ "total_time": 1280.0792806840036,
5
+ "experiment_name": "lora/llama-3.2-3B-rank10-target-mlp",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": "CAUSAL_LM",
35
+ "peft_type": "LORA",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "r": 10,
42
+ "target_modules": [
43
+ "up_proj",
44
+ "gate_proj",
45
+ "down_proj"
46
+ ],
47
+ "exclude_modules": null,
48
+ "lora_alpha": 20,
49
+ "lora_dropout": 0.0,
50
+ "fan_in_fan_out": false,
51
+ "bias": "none",
52
+ "use_rslora": false,
53
+ "modules_to_save": null,
54
+ "init_lora_weights": true,
55
+ "layers_to_transform": null,
56
+ "layers_pattern": null,
57
+ "rank_pattern": {},
58
+ "alpha_pattern": {},
59
+ "megatron_config": null,
60
+ "megatron_core": "megatron.core",
61
+ "trainable_token_indices": null,
62
+ "loftq_config": {},
63
+ "eva_config": null,
64
+ "corda_config": null,
65
+ "use_dora": false,
66
+ "alora_invocation_tokens": null,
67
+ "use_qalora": false,
68
+ "qalora_group_size": 16,
69
+ "layer_replication": null,
70
+ "lora_bias": false,
71
+ "target_parameters": null,
72
+ "use_bdlora": null,
73
+ "arrow_config": null,
74
+ "ensure_weight_tying": false
75
+ },
76
+ "error_msg": ""
77
+ },
78
+ "train_info": {
79
+ "accelerator_memory_reserved_avg": 15498352577,
80
+ "accelerator_memory_max": 24693964800,
81
+ "accelerator_memory_reserved_99th": 22066233344,
82
+ "train_time": 1084.286043549073,
83
+ "file_size": 37868984,
84
+ "num_trainable_params": 9461760,
85
+ "num_total_params": 3222211584,
86
+ "status": "success",
87
+ "metrics": [
88
+ {
89
+ "step": 250,
90
+ "valid accuracy": 0.28,
91
+ "train loss": 0.956905131816864,
92
+ "train samples": 1000,
93
+ "train time": 34.45501961914124,
94
+ "eval time": 8.187887417967431,
95
+ "tokens / sec": 6144.794063108907,
96
+ "mem allocated avg": 6931928963.072,
97
+ "mem reserved avg": 15621727191.04,
98
+ "elapsed time": 66.55439633503556
99
+ },
100
+ {
101
+ "step": 500,
102
+ "valid accuracy": 0.34,
103
+ "train loss": 0.6960876834392548,
104
+ "train samples": 2000,
105
+ "train time": 34.311293395468965,
106
+ "eval time": 12.610760554962326,
107
+ "tokens / sec": 6061.998234886334,
108
+ "mem allocated avg": 6925043042.304,
109
+ "mem reserved avg": 15157191245.824,
110
+ "elapsed time": 116.78655780700501
111
+ },
112
+ {
113
+ "step": 750,
114
+ "valid accuracy": 0.4,
115
+ "train loss": 0.6628184849023819,
116
+ "train samples": 3000,
117
+ "train time": 34.70101865468314,
118
+ "eval time": 7.451941114966758,
119
+ "tokens / sec": 6178.5217930789795,
120
+ "mem allocated avg": 6935014987.776,
121
+ "mem reserved avg": 15434812227.584,
122
+ "elapsed time": 162.2516898770118
123
+ },
124
+ {
125
+ "step": 1000,
126
+ "valid accuracy": 0.42,
127
+ "train loss": 0.6416039798259735,
128
+ "train samples": 4000,
129
+ "train time": 34.48607983236434,
130
+ "eval time": 6.998231331992429,
131
+ "tokens / sec": 6041.162144630941,
132
+ "mem allocated avg": 6926214045.696,
133
+ "mem reserved avg": 15482434355.2,
134
+ "elapsed time": 207.0800661740359
135
+ },
136
+ {
137
+ "step": 1250,
138
+ "valid accuracy": 0.38,
139
+ "train loss": 0.6370971938371658,
140
+ "train samples": 5000,
141
+ "train time": 34.169997635355685,
142
+ "eval time": 7.873161940020509,
143
+ "tokens / sec": 6102.956231528264,
144
+ "mem allocated avg": 6926345869.312,
145
+ "mem reserved avg": 15533814579.2,
146
+ "elapsed time": 252.415388560039
147
+ },
148
+ {
149
+ "step": 1500,
150
+ "valid accuracy": 0.5,
151
+ "train loss": 0.6280006822347641,
152
+ "train samples": 6000,
153
+ "train time": 34.276999980502296,
154
+ "eval time": 8.087436428992078,
155
+ "tokens / sec": 6107.039709399109,
156
+ "mem allocated avg": 6927777204.224,
157
+ "mem reserved avg": 15449374851.072,
158
+ "elapsed time": 298.10151039704215
159
+ },
160
+ {
161
+ "step": 1750,
162
+ "valid accuracy": 0.42,
163
+ "train loss": 0.6198004070520401,
164
+ "train samples": 7000,
165
+ "train time": 34.87267060915474,
166
+ "eval time": 12.61192423501052,
167
+ "tokens / sec": 6003.411735981022,
168
+ "mem allocated avg": 6928959924.224,
169
+ "mem reserved avg": 15878183714.816,
170
+ "elapsed time": 348.9116879070061
171
+ },
172
+ {
173
+ "step": 2000,
174
+ "valid accuracy": 0.36,
175
+ "train loss": 0.62178213763237,
176
+ "train samples": 8000,
177
+ "train time": 34.02151767432224,
178
+ "eval time": 7.202711585967336,
179
+ "tokens / sec": 6104.842293874464,
180
+ "mem allocated avg": 6924543111.168,
181
+ "mem reserved avg": 15504924213.248,
182
+ "elapsed time": 393.50626342202304
183
+ },
184
+ {
185
+ "step": 2250,
186
+ "valid accuracy": 0.52,
187
+ "train loss": 0.6131408748626709,
188
+ "train samples": 9000,
189
+ "train time": 34.81766810180852,
190
+ "eval time": 10.382788593007717,
191
+ "tokens / sec": 6173.532339141203,
192
+ "mem allocated avg": 6936508715.008,
193
+ "mem reserved avg": 15821845823.488,
194
+ "elapsed time": 441.9828996449942
195
+ },
196
+ {
197
+ "step": 2500,
198
+ "valid accuracy": 0.48,
199
+ "train loss": 0.6092700402736664,
200
+ "train samples": 10000,
201
+ "train time": 33.81983969331486,
202
+ "eval time": 12.668855660012923,
203
+ "tokens / sec": 6090.123485733533,
204
+ "mem allocated avg": 6922254147.584,
205
+ "mem reserved avg": 15216221880.32,
206
+ "elapsed time": 491.83312584704254
207
+ },
208
+ {
209
+ "step": 2750,
210
+ "valid accuracy": 0.52,
211
+ "train loss": 0.599810087800026,
212
+ "train samples": 11000,
213
+ "train time": 34.98453600978246,
214
+ "eval time": 12.589807912008837,
215
+ "tokens / sec": 6056.418754296279,
216
+ "mem allocated avg": 6932940275.712,
217
+ "mem reserved avg": 15688190132.224,
218
+ "elapsed time": 542.6733375560143
219
+ },
220
+ {
221
+ "step": 3000,
222
+ "valid accuracy": 0.38,
223
+ "train loss": 0.5911425656080246,
224
+ "train samples": 12000,
225
+ "train time": 34.17171706445515,
226
+ "eval time": 10.20739641803084,
227
+ "tokens / sec": 6108.297092776719,
228
+ "mem allocated avg": 6927596806.144,
229
+ "mem reserved avg": 15592476114.944,
230
+ "elapsed time": 590.4343665740453
231
+ },
232
+ {
233
+ "step": 3250,
234
+ "valid accuracy": 0.48,
235
+ "train loss": 0.5990802057981491,
236
+ "train samples": 13000,
237
+ "train time": 34.13041735702427,
238
+ "eval time": 9.451952209987212,
239
+ "tokens / sec": 6179.268123030882,
240
+ "mem allocated avg": 6930062843.904,
241
+ "mem reserved avg": 15432622800.896,
242
+ "elapsed time": 637.2517090940382
243
+ },
244
+ {
245
+ "step": 3500,
246
+ "valid accuracy": 0.54,
247
+ "train loss": 0.5822520101070404,
248
+ "train samples": 14000,
249
+ "train time": 34.94589354854543,
250
+ "eval time": 12.59894744300982,
251
+ "tokens / sec": 6002.135836321477,
252
+ "mem allocated avg": 6928036900.864,
253
+ "mem reserved avg": 15451773992.96,
254
+ "elapsed time": 688.1364196360228
255
+ },
256
+ {
257
+ "step": 3750,
258
+ "valid accuracy": 0.62,
259
+ "train loss": 0.5812049109935761,
260
+ "train samples": 15000,
261
+ "train time": 35.20199201494688,
262
+ "eval time": 12.605403171968646,
263
+ "tokens / sec": 6155.986851766434,
264
+ "mem allocated avg": 6939365515.264,
265
+ "mem reserved avg": 15827197755.392,
266
+ "elapsed time": 739.258546972007
267
+ },
268
+ {
269
+ "step": 4000,
270
+ "valid accuracy": 0.48,
271
+ "train loss": 0.5900991085767746,
272
+ "train samples": 16000,
273
+ "train time": 33.90330129902577,
274
+ "eval time": 8.236152897006832,
275
+ "tokens / sec": 6028.115026245918,
276
+ "mem allocated avg": 6919923267.584,
277
+ "mem reserved avg": 15453300719.616,
278
+ "elapsed time": 784.7485570590361
279
+ },
280
+ {
281
+ "step": 4250,
282
+ "valid accuracy": 0.52,
283
+ "train loss": 0.5774198567867279,
284
+ "train samples": 17000,
285
+ "train time": 34.68033379589906,
286
+ "eval time": 7.178619690006599,
287
+ "tokens / sec": 6095.356556948615,
288
+ "mem allocated avg": 6930677893.12,
289
+ "mem reserved avg": 15433386164.224,
290
+ "elapsed time": 829.8555172120105
291
+ },
292
+ {
293
+ "step": 4500,
294
+ "valid accuracy": 0.5,
295
+ "train loss": 0.5839589636325836,
296
+ "train samples": 18000,
297
+ "train time": 33.97633226803737,
298
+ "eval time": 7.210191433026921,
299
+ "tokens / sec": 6116.551909150627,
300
+ "mem allocated avg": 6925906124.8,
301
+ "mem reserved avg": 15420786475.008,
302
+ "elapsed time": 874.4303943190025
303
+ },
304
+ {
305
+ "step": 4750,
306
+ "valid accuracy": 0.52,
307
+ "train loss": 0.575189305305481,
308
+ "train samples": 19000,
309
+ "train time": 34.245364292524755,
310
+ "eval time": 10.291229791007936,
311
+ "tokens / sec": 6130.435588498806,
312
+ "mem allocated avg": 6928215662.592,
313
+ "mem reserved avg": 15476939816.96,
314
+ "elapsed time": 922.2494909689995
315
+ },
316
+ {
317
+ "step": 5000,
318
+ "valid accuracy": 0.5,
319
+ "train loss": 0.5811728711128235,
320
+ "train samples": 20000,
321
+ "train time": 34.228812442102935,
322
+ "eval time": 8.269593706994783,
323
+ "tokens / sec": 6084.932112450577,
324
+ "mem allocated avg": 6926067277.824,
325
+ "mem reserved avg": 15089847500.8,
326
+ "elapsed time": 968.0756093530217
327
+ },
328
+ {
329
+ "step": 5000,
330
+ "test accuracy": 0.5109931766489765,
331
+ "train loss": 0.5811728711128235,
332
+ "train samples": 20000,
333
+ "train total tokens": 4198051,
334
+ "forgetting": 0.4244680404663086
335
+ }
336
+ ]
337
+ },
338
+ "meta_info": {
339
+ "model_info": {
340
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
341
+ "created_at": "2024-09-18T15:23:48+00:00"
342
+ },
343
+ "dataset_info": {
344
+ "metamath": {
345
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
346
+ "created_at": "2023-09-21T17:22:46+00:00"
347
+ },
348
+ "gsm8k": {
349
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
350
+ "created_at": "2022-04-12T10:22:10+00:00"
351
+ }
352
+ },
353
+ "package_info": {
354
+ "transformers-version": "4.57.1",
355
+ "transformers-commit-hash": null,
356
+ "peft-version": "0.18.1.dev0",
357
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
358
+ "datasets-version": "4.2.0",
359
+ "datasets-commit-hash": null,
360
+ "bitsandbytes-version": "0.46.0",
361
+ "bitsandbytes-commit-hash": null,
362
+ "torch-version": "2.9.0+cu128",
363
+ "torch-commit-hash": null
364
+ },
365
+ "system_info": {
366
+ "system": "Linux",
367
+ "release": "6.14.0-1016-aws",
368
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
369
+ "machine": "x86_64",
370
+ "processor": "x86_64",
371
+ "accelerator": "NVIDIA L40S"
372
+ },
373
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
374
+ }
375
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/lora--llama-3.2-3B-rank14-target-mlp-bdlora.json ADDED
@@ -0,0 +1,388 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-10T02:37:29+00:00",
4
+ "total_time": 1449.8443840030232,
5
+ "experiment_name": "lora/llama-3.2-3B-rank14-target-mlp-bdlora",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": "CAUSAL_LM",
35
+ "peft_type": "LORA",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "r": 14,
42
+ "target_modules": [
43
+ "gate_proj",
44
+ "up_proj",
45
+ "down_proj"
46
+ ],
47
+ "exclude_modules": null,
48
+ "lora_alpha": 28,
49
+ "lora_dropout": 0.0,
50
+ "fan_in_fan_out": false,
51
+ "bias": "none",
52
+ "use_rslora": false,
53
+ "modules_to_save": null,
54
+ "init_lora_weights": true,
55
+ "layers_to_transform": null,
56
+ "layers_pattern": null,
57
+ "rank_pattern": {},
58
+ "alpha_pattern": {},
59
+ "megatron_config": null,
60
+ "megatron_core": "megatron.core",
61
+ "trainable_token_indices": null,
62
+ "loftq_config": {},
63
+ "eva_config": null,
64
+ "corda_config": null,
65
+ "use_dora": false,
66
+ "alora_invocation_tokens": null,
67
+ "use_qalora": false,
68
+ "qalora_group_size": 16,
69
+ "layer_replication": null,
70
+ "lora_bias": false,
71
+ "target_parameters": null,
72
+ "use_bdlora": {
73
+ "target_modules_bd_a": [
74
+ "o_proj",
75
+ "down_proj"
76
+ ],
77
+ "target_modules_bd_b": [
78
+ "q_proj",
79
+ "v_proj",
80
+ "k_proj",
81
+ "up_proj",
82
+ "gate_proj"
83
+ ],
84
+ "nblocks": 2
85
+ },
86
+ "arrow_config": null,
87
+ "ensure_weight_tying": false
88
+ },
89
+ "error_msg": ""
90
+ },
91
+ "train_info": {
92
+ "accelerator_memory_reserved_avg": 15486387067,
93
+ "accelerator_memory_max": 24685576192,
94
+ "accelerator_memory_reserved_99th": 22066233344,
95
+ "train_time": 1247.0251060129958,
96
+ "file_size": 33740152,
97
+ "num_trainable_params": 8429568,
98
+ "num_total_params": 3221179392,
99
+ "status": "success",
100
+ "metrics": [
101
+ {
102
+ "step": 250,
103
+ "valid accuracy": 0.38,
104
+ "train loss": 0.9585020205974579,
105
+ "train samples": 1000,
106
+ "train time": 40.947455714514945,
107
+ "eval time": 9.724019955960102,
108
+ "tokens / sec": 5170.504401448084,
109
+ "mem allocated avg": 6915241809.92,
110
+ "mem reserved avg": 15587350675.456,
111
+ "elapsed time": 76.15033562999452
112
+ },
113
+ {
114
+ "step": 500,
115
+ "valid accuracy": 0.34,
116
+ "train loss": 0.6961374168395996,
117
+ "train samples": 2000,
118
+ "train time": 40.97218767303275,
119
+ "eval time": 12.660414524958469,
120
+ "tokens / sec": 5076.492416266536,
121
+ "mem allocated avg": 6908625721.344,
122
+ "mem reserved avg": 15151772205.056,
123
+ "elapsed time": 133.12352299899794
124
+ },
125
+ {
126
+ "step": 750,
127
+ "valid accuracy": 0.32,
128
+ "train loss": 0.6627170724868774,
129
+ "train samples": 3000,
130
+ "train time": 41.68421668821247,
131
+ "eval time": 8.807691658032127,
132
+ "tokens / sec": 5143.45757301057,
133
+ "mem allocated avg": 6919140038.656,
134
+ "mem reserved avg": 15427161817.088,
135
+ "elapsed time": 186.93488279398298
136
+ },
137
+ {
138
+ "step": 1000,
139
+ "valid accuracy": 0.44,
140
+ "train loss": 0.6412999538183213,
141
+ "train samples": 4000,
142
+ "train time": 40.972820968716405,
143
+ "eval time": 15.062458469008561,
144
+ "tokens / sec": 5084.736541793616,
145
+ "mem allocated avg": 6909692874.752,
146
+ "mem reserved avg": 15471554330.624,
147
+ "elapsed time": 246.33008374396013
148
+ },
149
+ {
150
+ "step": 1250,
151
+ "valid accuracy": 0.4,
152
+ "train loss": 0.6367096021175385,
153
+ "train samples": 5000,
154
+ "train time": 40.94994170789141,
155
+ "eval time": 14.975786323018838,
156
+ "tokens / sec": 5092.510301664554,
157
+ "mem allocated avg": 6909957285.888,
158
+ "mem reserved avg": 15508665532.416,
159
+ "elapsed time": 305.5784399929689
160
+ },
161
+ {
162
+ "step": 1500,
163
+ "valid accuracy": 0.5,
164
+ "train loss": 0.6278209384679795,
165
+ "train samples": 6000,
166
+ "train time": 40.83828021842055,
167
+ "eval time": 9.454851284972392,
168
+ "tokens / sec": 5125.852481554279,
169
+ "mem allocated avg": 6911538085.888,
170
+ "mem reserved avg": 15441019797.504,
171
+ "elapsed time": 359.22269572195364
172
+ },
173
+ {
174
+ "step": 1750,
175
+ "valid accuracy": 0.42,
176
+ "train loss": 0.6189806928634644,
177
+ "train samples": 7000,
178
+ "train time": 41.167949416907504,
179
+ "eval time": 8.857150142022874,
180
+ "tokens / sec": 5085.388098393329,
181
+ "mem allocated avg": 6912511498.24,
182
+ "mem reserved avg": 15868285157.376,
183
+ "elapsed time": 412.59175090299686
184
+ },
185
+ {
186
+ "step": 2000,
187
+ "valid accuracy": 0.4,
188
+ "train loss": 0.6215936172008515,
189
+ "train samples": 8000,
190
+ "train time": 39.99366526392987,
191
+ "eval time": 8.554796635988168,
192
+ "tokens / sec": 5193.222442338142,
193
+ "mem allocated avg": 6908099178.496,
194
+ "mem reserved avg": 15486578327.552,
195
+ "elapsed time": 464.5223481360008
196
+ },
197
+ {
198
+ "step": 2250,
199
+ "valid accuracy": 0.44,
200
+ "train loss": 0.6134832800626755,
201
+ "train samples": 9000,
202
+ "train time": 40.83385715901386,
203
+ "eval time": 9.429859316034708,
204
+ "tokens / sec": 5263.965124895172,
205
+ "mem allocated avg": 6920264648.704,
206
+ "mem reserved avg": 15807736184.832,
207
+ "elapsed time": 518.1040934349876
208
+ },
209
+ {
210
+ "step": 2500,
211
+ "valid accuracy": 0.42,
212
+ "train loss": 0.6091812270879745,
213
+ "train samples": 10000,
214
+ "train time": 39.71863999107154,
215
+ "eval time": 9.170394071028568,
216
+ "tokens / sec": 5185.650869372664,
217
+ "mem allocated avg": 6906021558.272,
218
+ "mem reserved avg": 15211985633.28,
219
+ "elapsed time": 570.3839986599633
220
+ },
221
+ {
222
+ "step": 2750,
223
+ "valid accuracy": 0.52,
224
+ "train loss": 0.6002322329282761,
225
+ "train samples": 11000,
226
+ "train time": 40.94364706269698,
227
+ "eval time": 9.519503901014104,
228
+ "tokens / sec": 5174.942028870724,
229
+ "mem allocated avg": 6916685307.904,
230
+ "mem reserved avg": 15683366682.624,
231
+ "elapsed time": 624.1543187929783
232
+ },
233
+ {
234
+ "step": 3000,
235
+ "valid accuracy": 0.44,
236
+ "train loss": 0.5916472736597062,
237
+ "train samples": 12000,
238
+ "train time": 41.149513348413166,
239
+ "eval time": 9.368373601988424,
240
+ "tokens / sec": 5072.50227317814,
241
+ "mem allocated avg": 6910862186.496,
242
+ "mem reserved avg": 15584163004.416,
243
+ "elapsed time": 678.0904719049577
244
+ },
245
+ {
246
+ "step": 3250,
247
+ "valid accuracy": 0.52,
248
+ "train loss": 0.6005176463127136,
249
+ "train samples": 13000,
250
+ "train time": 41.157697485759854,
251
+ "eval time": 9.39710839802865,
252
+ "tokens / sec": 5124.217652675289,
253
+ "mem allocated avg": 6913147236.352,
254
+ "mem reserved avg": 15416206295.04,
255
+ "elapsed time": 731.8966049049632
256
+ },
257
+ {
258
+ "step": 3500,
259
+ "valid accuracy": 0.54,
260
+ "train loss": 0.5828959219455719,
261
+ "train samples": 14000,
262
+ "train time": 40.61929137952393,
263
+ "eval time": 8.772685625997838,
264
+ "tokens / sec": 5163.802540034817,
265
+ "mem allocated avg": 6912162123.776,
266
+ "mem reserved avg": 15441204346.88,
267
+ "elapsed time": 784.6333682870027
268
+ },
269
+ {
270
+ "step": 3750,
271
+ "valid accuracy": 0.54,
272
+ "train loss": 0.5827477180957794,
273
+ "train samples": 15000,
274
+ "train time": 41.78944715799298,
275
+ "eval time": 8.540672906965483,
276
+ "tokens / sec": 5185.591452806565,
277
+ "mem allocated avg": 6922672640.0,
278
+ "mem reserved avg": 15820361039.872,
279
+ "elapsed time": 838.2750829259749
280
+ },
281
+ {
282
+ "step": 4000,
283
+ "valid accuracy": 0.56,
284
+ "train loss": 0.5919854927062989,
285
+ "train samples": 16000,
286
+ "train time": 40.357097996922676,
287
+ "eval time": 8.853416519996244,
288
+ "tokens / sec": 5064.115363686059,
289
+ "mem allocated avg": 6903590129.664,
290
+ "mem reserved avg": 15445197324.288,
291
+ "elapsed time": 890.848429523001
292
+ },
293
+ {
294
+ "step": 4250,
295
+ "valid accuracy": 0.56,
296
+ "train loss": 0.5791787240505218,
297
+ "train samples": 17000,
298
+ "train time": 40.23014218133176,
299
+ "eval time": 8.847462174016982,
300
+ "tokens / sec": 5254.492988048452,
301
+ "mem allocated avg": 6914294579.2,
302
+ "mem reserved avg": 15425500872.704,
303
+ "elapsed time": 943.1942599099712
304
+ },
305
+ {
306
+ "step": 4500,
307
+ "valid accuracy": 0.5,
308
+ "train loss": 0.585926919221878,
309
+ "train samples": 18000,
310
+ "train time": 39.97698019299423,
311
+ "eval time": 8.84911245398689,
312
+ "tokens / sec": 5198.441678104017,
313
+ "mem allocated avg": 6909213696.0,
314
+ "mem reserved avg": 15410191663.104,
315
+ "elapsed time": 995.4117041879799
316
+ },
317
+ {
318
+ "step": 4750,
319
+ "valid accuracy": 0.52,
320
+ "train loss": 0.5774390150308609,
321
+ "train samples": 19000,
322
+ "train time": 40.493067349656485,
323
+ "eval time": 8.984108597971499,
324
+ "tokens / sec": 5184.566488559207,
325
+ "mem allocated avg": 6911982841.856,
326
+ "mem reserved avg": 15462503022.592,
327
+ "elapsed time": 1048.1989943929948
328
+ },
329
+ {
330
+ "step": 5000,
331
+ "valid accuracy": 0.52,
332
+ "train loss": 0.5829407849311828,
333
+ "train samples": 20000,
334
+ "train time": 40.788713948684745,
335
+ "eval time": 11.74522749998141,
336
+ "tokens / sec": 5106.314463898809,
337
+ "mem allocated avg": 6909793161.216,
338
+ "mem reserved avg": 15076937433.088,
339
+ "elapsed time": 1104.0808472170029
340
+ },
341
+ {
342
+ "step": 5000,
343
+ "test accuracy": 0.513267626990144,
344
+ "train loss": 0.5829407849311828,
345
+ "train samples": 20000,
346
+ "train total tokens": 4198051,
347
+ "forgetting": 0.38550758361816406
348
+ }
349
+ ]
350
+ },
351
+ "meta_info": {
352
+ "model_info": {
353
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
354
+ "created_at": "2024-09-18T15:23:48+00:00"
355
+ },
356
+ "dataset_info": {
357
+ "metamath": {
358
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
359
+ "created_at": "2023-09-21T17:22:46+00:00"
360
+ },
361
+ "gsm8k": {
362
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
363
+ "created_at": "2022-04-12T10:22:10+00:00"
364
+ }
365
+ },
366
+ "package_info": {
367
+ "transformers-version": "4.57.1",
368
+ "transformers-commit-hash": null,
369
+ "peft-version": "0.18.1.dev0",
370
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
371
+ "datasets-version": "4.2.0",
372
+ "datasets-commit-hash": null,
373
+ "bitsandbytes-version": "0.46.0",
374
+ "bitsandbytes-commit-hash": null,
375
+ "torch-version": "2.9.0+cu128",
376
+ "torch-commit-hash": null
377
+ },
378
+ "system_info": {
379
+ "system": "Linux",
380
+ "release": "6.14.0-1016-aws",
381
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
382
+ "machine": "x86_64",
383
+ "processor": "x86_64",
384
+ "accelerator": "NVIDIA L40S"
385
+ },
386
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
387
+ }
388
+ }
tasks/tasksmith-f71b564d1bc2/tests/source/method_comparison/MetaMathQA/results/lora--llama-3.2-3B-rank32-dora.json ADDED
@@ -0,0 +1,374 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_info": {
3
+ "created_at": "2026-01-10T03:21:22+00:00",
4
+ "total_time": 1503.135829265986,
5
+ "experiment_name": "lora/llama-3.2-3B-rank32-dora",
6
+ "peft_branch": "main",
7
+ "train_config": {
8
+ "model_id": "meta-llama/Llama-3.2-3B",
9
+ "dtype": "bfloat16",
10
+ "max_seq_length": 768,
11
+ "batch_size": 4,
12
+ "batch_size_eval": 50,
13
+ "max_steps": 5000,
14
+ "eval_steps": 250,
15
+ "compile": false,
16
+ "query_template": "Question: {query} Think step by step.\nAnswer:",
17
+ "seed": 0,
18
+ "grad_norm_clip": 1.0,
19
+ "optimizer_type": "AdamW",
20
+ "optimizer_kwargs": {
21
+ "lr": 0.0001,
22
+ "weight_decay": 0.1
23
+ },
24
+ "lr_scheduler": "cosine",
25
+ "use_amp": false,
26
+ "autocast_adapter_dtype": true,
27
+ "generation_kwargs": {
28
+ "max_length": 800,
29
+ "max_new_tokens": 300
30
+ },
31
+ "attn_implementation": null
32
+ },
33
+ "peft_config": {
34
+ "task_type": "CAUSAL_LM",
35
+ "peft_type": "LORA",
36
+ "auto_mapping": null,
37
+ "peft_version": "0.18.1.dev0@UNKNOWN",
38
+ "base_model_name_or_path": "meta-llama/Llama-3.2-3B",
39
+ "revision": null,
40
+ "inference_mode": false,
41
+ "r": 32,
42
+ "target_modules": [
43
+ "q_proj",
44
+ "v_proj"
45
+ ],
46
+ "exclude_modules": null,
47
+ "lora_alpha": 64,
48
+ "lora_dropout": 0.0,
49
+ "fan_in_fan_out": false,
50
+ "bias": "none",
51
+ "use_rslora": false,
52
+ "modules_to_save": null,
53
+ "init_lora_weights": true,
54
+ "layers_to_transform": null,
55
+ "layers_pattern": null,
56
+ "rank_pattern": {},
57
+ "alpha_pattern": {},
58
+ "megatron_config": null,
59
+ "megatron_core": "megatron.core",
60
+ "trainable_token_indices": null,
61
+ "loftq_config": {},
62
+ "eva_config": null,
63
+ "corda_config": null,
64
+ "use_dora": true,
65
+ "alora_invocation_tokens": null,
66
+ "use_qalora": false,
67
+ "qalora_group_size": 16,
68
+ "layer_replication": null,
69
+ "lora_bias": false,
70
+ "target_parameters": null,
71
+ "use_bdlora": null,
72
+ "arrow_config": null,
73
+ "ensure_weight_tying": false
74
+ },
75
+ "error_msg": ""
76
+ },
77
+ "train_info": {
78
+ "accelerator_memory_reserved_avg": 15412689371,
79
+ "accelerator_memory_max": 24547164160,
80
+ "accelerator_memory_reserved_99th": 22162702336,
81
+ "train_time": 1216.6712898398982,
82
+ "file_size": 37181760,
83
+ "num_trainable_params": 9289728,
84
+ "num_total_params": 3222039552,
85
+ "status": "success",
86
+ "metrics": [
87
+ {
88
+ "step": 250,
89
+ "valid accuracy": 0.26,
90
+ "train loss": 0.9799610931873322,
91
+ "train samples": 1000,
92
+ "train time": 35.417795918125194,
93
+ "eval time": 17.352634231036063,
94
+ "tokens / sec": 5977.757636003882,
95
+ "mem allocated avg": 6925407713.28,
96
+ "mem reserved avg": 15541154611.2,
97
+ "elapsed time": 75.89598628797103
98
+ },
99
+ {
100
+ "step": 500,
101
+ "valid accuracy": 0.44,
102
+ "train loss": 0.7162615118026734,
103
+ "train samples": 2000,
104
+ "train time": 35.72266378562199,
105
+ "eval time": 17.35569141904125,
106
+ "tokens / sec": 5822.494124408378,
107
+ "mem allocated avg": 6918067554.304,
108
+ "mem reserved avg": 15118301659.136,
109
+ "elapsed time": 132.76655074994778
110
+ },
111
+ {
112
+ "step": 750,
113
+ "valid accuracy": 0.46,
114
+ "train loss": 0.6789851016998291,
115
+ "train samples": 3000,
116
+ "train time": 36.546013267710805,
117
+ "eval time": 11.214822898968123,
118
+ "tokens / sec": 5866.60433873995,
119
+ "mem allocated avg": 6928052195.328,
120
+ "mem reserved avg": 15356437463.04,
121
+ "elapsed time": 184.19187769596465
122
+ },
123
+ {
124
+ "step": 1000,
125
+ "valid accuracy": 0.4,
126
+ "train loss": 0.6588249838352204,
127
+ "train samples": 4000,
128
+ "train time": 36.061139613215346,
129
+ "eval time": 13.717305009020492,
130
+ "tokens / sec": 5777.299393046663,
131
+ "mem allocated avg": 6919732948.992,
132
+ "mem reserved avg": 15453527212.032,
133
+ "elapsed time": 237.84873885498382
134
+ },
135
+ {
136
+ "step": 1250,
137
+ "valid accuracy": 0.36,
138
+ "train loss": 0.6543591912984849,
139
+ "train samples": 5000,
140
+ "train time": 35.93075305066304,
141
+ "eval time": 17.338175220997073,
142
+ "tokens / sec": 5803.886150283504,
143
+ "mem allocated avg": 6919911497.728,
144
+ "mem reserved avg": 15469197131.776,
145
+ "elapsed time": 294.79154046898475
146
+ },
147
+ {
148
+ "step": 1500,
149
+ "valid accuracy": 0.4,
150
+ "train loss": 0.6470256053209305,
151
+ "train samples": 6000,
152
+ "train time": 36.02197089209221,
153
+ "eval time": 17.308061218995135,
154
+ "tokens / sec": 5811.203407694546,
155
+ "mem allocated avg": 6921449252.864,
156
+ "mem reserved avg": 15339878350.848,
157
+ "elapsed time": 351.9409700029646
158
+ },
159
+ {
160
+ "step": 1750,
161
+ "valid accuracy": 0.46,
162
+ "train loss": 0.6376578369140625,
163
+ "train samples": 7000,
164
+ "train time": 36.077694381529,
165
+ "eval time": 12.42961401498178,
166
+ "tokens / sec": 5802.892994935542,
167
+ "mem allocated avg": 6922675496.96,
168
+ "mem reserved avg": 15837507354.624,
169
+ "elapsed time": 404.2594750869903
170
+ },
171
+ {
172
+ "step": 2000,
173
+ "valid accuracy": 0.38,
174
+ "train loss": 0.6405653357505798,
175
+ "train samples": 8000,
176
+ "train time": 36.009058863681275,
177
+ "eval time": 11.470333072997164,
178
+ "tokens / sec": 5767.881931773622,
179
+ "mem allocated avg": 6919334260.736,
180
+ "mem reserved avg": 15421482729.472,
181
+ "elapsed time": 455.6199949949514
182
+ },
183
+ {
184
+ "step": 2250,
185
+ "valid accuracy": 0.46,
186
+ "train loss": 0.6328598493337632,
187
+ "train samples": 9000,
188
+ "train time": 36.205427291512024,
189
+ "eval time": 17.303009946015663,
190
+ "tokens / sec": 5936.899964453458,
191
+ "mem allocated avg": 6930608842.752,
192
+ "mem reserved avg": 15744427360.256,
193
+ "elapsed time": 512.8717058099573
194
+ },
195
+ {
196
+ "step": 2500,
197
+ "valid accuracy": 0.42,
198
+ "train loss": 0.629319528579712,
199
+ "train samples": 10000,
200
+ "train time": 34.97963203344261,
201
+ "eval time": 17.334575710003264,
202
+ "tokens / sec": 5888.1980177230935,
203
+ "mem allocated avg": 6915153993.728,
204
+ "mem reserved avg": 15108914806.784,
205
+ "elapsed time": 568.9176407279447
206
+ },
207
+ {
208
+ "step": 2750,
209
+ "valid accuracy": 0.34,
210
+ "train loss": 0.6210904417037963,
211
+ "train samples": 11000,
212
+ "train time": 35.76373431546381,
213
+ "eval time": 14.790576178987976,
214
+ "tokens / sec": 5924.4652175034535,
215
+ "mem allocated avg": 6926034331.648,
216
+ "mem reserved avg": 15580086140.928,
217
+ "elapsed time": 623.1745745909866
218
+ },
219
+ {
220
+ "step": 3000,
221
+ "valid accuracy": 0.42,
222
+ "train loss": 0.6132080693244935,
223
+ "train samples": 12000,
224
+ "train time": 35.911868128867354,
225
+ "eval time": 14.77954756503459,
226
+ "tokens / sec": 5812.312499338176,
227
+ "mem allocated avg": 6920991168.512,
228
+ "mem reserved avg": 15530618519.552,
229
+ "elapsed time": 677.7459287579986
230
+ },
231
+ {
232
+ "step": 3250,
233
+ "valid accuracy": 0.48,
234
+ "train loss": 0.6221653089523316,
235
+ "train samples": 13000,
236
+ "train time": 35.66093143681064,
237
+ "eval time": 17.444324638985563,
238
+ "tokens / sec": 5914.063135835525,
239
+ "mem allocated avg": 6923086110.72,
240
+ "mem reserved avg": 15357922246.656,
241
+ "elapsed time": 734.5032512069447
242
+ },
243
+ {
244
+ "step": 3500,
245
+ "valid accuracy": 0.5,
246
+ "train loss": 0.6054375174045563,
247
+ "train samples": 14000,
248
+ "train time": 35.69942569779232,
249
+ "eval time": 10.496662761026528,
250
+ "tokens / sec": 5875.444657726555,
251
+ "mem allocated avg": 6920997902.336,
252
+ "mem reserved avg": 15300158291.968,
253
+ "elapsed time": 784.4919563499861
254
+ },
255
+ {
256
+ "step": 3750,
257
+ "valid accuracy": 0.54,
258
+ "train loss": 0.6029420564174652,
259
+ "train samples": 15000,
260
+ "train time": 36.43003757862607,
261
+ "eval time": 12.152991220995318,
262
+ "tokens / sec": 5948.47039430841,
263
+ "mem allocated avg": 6933233727.488,
264
+ "mem reserved avg": 15711619514.368,
265
+ "elapsed time": 836.8070459629525
266
+ },
267
+ {
268
+ "step": 4000,
269
+ "valid accuracy": 0.46,
270
+ "train loss": 0.616380462884903,
271
+ "train samples": 16000,
272
+ "train time": 35.82399892428657,
273
+ "eval time": 10.243308471050113,
274
+ "tokens / sec": 5704.918661703261,
275
+ "mem allocated avg": 6913596911.616,
276
+ "mem reserved avg": 15358928879.616,
277
+ "elapsed time": 886.7407023639535
278
+ },
279
+ {
280
+ "step": 4250,
281
+ "valid accuracy": 0.5,
282
+ "train loss": 0.6010915513038635,
283
+ "train samples": 17000,
284
+ "train time": 36.03616644133581,
285
+ "eval time": 17.357441558968276,
286
+ "tokens / sec": 5866.023522344573,
287
+ "mem allocated avg": 6924514760.704,
288
+ "mem reserved avg": 15282240225.28,
289
+ "elapsed time": 943.718220997951
290
+ },
291
+ {
292
+ "step": 4500,
293
+ "valid accuracy": 0.48,
294
+ "train loss": 0.6073519963026047,
295
+ "train samples": 18000,
296
+ "train time": 36.01844966417411,
297
+ "eval time": 10.97284383297665,
298
+ "tokens / sec": 5769.765271343896,
299
+ "mem allocated avg": 6919068428.288,
300
+ "mem reserved avg": 15324342648.832,
301
+ "elapsed time": 994.5949146979838
302
+ },
303
+ {
304
+ "step": 4750,
305
+ "valid accuracy": 0.48,
306
+ "train loss": 0.59943410551548,
307
+ "train samples": 19000,
308
+ "train time": 36.32360130321467,
309
+ "eval time": 11.35271465400001,
310
+ "tokens / sec": 5779.685726850554,
311
+ "mem allocated avg": 6921889050.624,
312
+ "mem reserved avg": 15411919716.352,
313
+ "elapsed time": 1045.9306690069498
314
+ },
315
+ {
316
+ "step": 5000,
317
+ "valid accuracy": 0.5,
318
+ "train loss": 0.6067105796337128,
319
+ "train samples": 20000,
320
+ "train time": 35.769867127528414,
321
+ "eval time": 10.729407122998964,
322
+ "tokens / sec": 5822.778129352014,
323
+ "mem allocated avg": 6918440036.352,
324
+ "mem reserved avg": 15005122560.0,
325
+ "elapsed time": 1096.1181151779601
326
+ },
327
+ {
328
+ "step": 5000,
329
+ "test accuracy": 0.4799090219863533,
330
+ "train loss": 0.6067105796337128,
331
+ "train samples": 20000,
332
+ "train total tokens": 4198051,
333
+ "forgetting": 0.42719078063964844
334
+ }
335
+ ]
336
+ },
337
+ "meta_info": {
338
+ "model_info": {
339
+ "sha": "13afe5124825b4f3751f836b40dafda64c1ed062",
340
+ "created_at": "2024-09-18T15:23:48+00:00"
341
+ },
342
+ "dataset_info": {
343
+ "metamath": {
344
+ "sha": "aa4f34d3d2d3231299b5b03d9b3e5a20da45aa18",
345
+ "created_at": "2023-09-21T17:22:46+00:00"
346
+ },
347
+ "gsm8k": {
348
+ "sha": "cc7b047b6e5bb11b4f1af84efc572db110a51b3c",
349
+ "created_at": "2022-04-12T10:22:10+00:00"
350
+ }
351
+ },
352
+ "package_info": {
353
+ "transformers-version": "4.57.1",
354
+ "transformers-commit-hash": null,
355
+ "peft-version": "0.18.1.dev0",
356
+ "peft-commit-hash": "8be1a16f5e06ca5e197d2af74bdfc5b3c8072d26",
357
+ "datasets-version": "4.2.0",
358
+ "datasets-commit-hash": null,
359
+ "bitsandbytes-version": "0.46.0",
360
+ "bitsandbytes-commit-hash": null,
361
+ "torch-version": "2.9.0+cu128",
362
+ "torch-commit-hash": null
363
+ },
364
+ "system_info": {
365
+ "system": "Linux",
366
+ "release": "6.14.0-1016-aws",
367
+ "version": "#16~24.04.1-Ubuntu SMP Tue Oct 14 02:15:09 UTC 2025",
368
+ "machine": "x86_64",
369
+ "processor": "x86_64",
370
+ "accelerator": "NVIDIA L40S"
371
+ },
372
+ "pytorch_info": "PyTorch built with:\n - GCC 13.3\n - C++ Version: 201703\n - Intel(R) oneAPI Math Kernel Library Version 2024.2-Product Build 20240605 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v3.7.1 (Git Hash 8d263e693366ef8db40acc569cc7d8edf644556d)\n - OpenMP 201511 (a.k.a. OpenMP 4.5)\n - LAPACK is enabled (usually provided by MKL)\n - NNPACK is enabled\n - CPU capability usage: AVX2\n - CUDA Runtime 12.8\n - NVCC architecture flags: -gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_90,code=sm_90;-gencode;arch=compute_100,code=sm_100;-gencode;arch=compute_120,code=sm_120\n - CuDNN 90.7.1\n - Built with CuDNN 90.8\n - Magma 2.6.1\n - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, COMMIT_SHA=0fabc3ba44823f257e70ce397d989c8de5e362c1, CUDA_VERSION=12.8, CUDNN_VERSION=9.8.0, CXX_COMPILER=/opt/rh/gcc-toolset-13/root/usr/bin/c++, CXX_FLAGS= -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOROCTRACER -DLIBKINETO_NOXPUPTI=ON -DUSE_FBGEMM -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -DC10_NODEPRECATED -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-unknown-pragmas -Wno-unused-parameter -Wno-strict-overflow -Wno-strict-aliasing -Wno-stringop-overflow -Wsuggest-override -Wno-psabi -Wno-error=old-style-cast -faligned-new -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-dangling-reference -Wno-error=dangling-reference -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, TORCH_VERSION=2.9.0, USE_CUDA=ON, USE_CUDNN=ON, USE_CUSPARSELT=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_GLOO=ON, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=1, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF, USE_ROCM_KERNEL_ASSERT=OFF, USE_XCCL=OFF, USE_XPU=OFF, \n"
373
+ }
374
+ }