Hugging Face's logo

nvan15
/
OminiControlRotation

Safetensors
Model card Files
xet
Community
nvan15 commited on
Commit
a89739c
·
verified ·
1 Parent(s): 1af4760

Batch upload part 10

Browse files
Files changed (50) hide show
  1. nl_tasks/run_all/run_exnr12/ft/adapter_config.json +18 -0
  2. nl_tasks/run_all/run_exnr12/ft/added_tokens.json +3 -0
  3. nl_tasks/run_all/run_exnr12/ft/tokenizer.json +0 -0
  4. nl_tasks/run_all/run_exnr12/ft/tokenizer.model +3 -0
  5. nl_tasks/run_all/run_exnr12/ft/training_args.bin +3 -0
  6. nl_tasks/run_all/run_exnr12/ft2/adapter_config.json +18 -0
  7. nl_tasks/run_all/run_exnr12/ft2/adapter_model.bin +3 -0
  8. nl_tasks/run_all/run_exnr13/ft/adapter_config.json +18 -0
  9. nl_tasks/run_all/run_exnr13/ft/added_tokens.json +3 -0
  10. nl_tasks/run_all/run_exnr13/ft/special_tokens_map.json +30 -0
  11. nl_tasks/run_all/run_exnr13/ft/tokenizer.json +0 -0
  12. nl_tasks/run_all/run_exnr13/ft/tokenizer.model +3 -0
  13. nl_tasks/run_all/run_exnr13/ft/tokenizer_config.json +51 -0
  14. nl_tasks/run_all/run_exnr13/ft/training_args.bin +3 -0
  15. nl_tasks/run_all/run_exnr13/ft2/adapter_config.json +18 -0
  16. nl_tasks/run_all/run_exnr13/ft2/adapter_model.bin +3 -0
  17. nl_tasks/run_all/run_exnr13/trainer_state.json +1106 -0
  18. nl_tasks/run_all/run_exps1/ft/adapter_config.json +18 -0
  19. nl_tasks/run_all/run_exps1/ft/special_tokens_map.json +30 -0
  20. nl_tasks/run_all/run_exps1/ft/tokenizer.json +0 -0
  21. nl_tasks/run_all/run_exps1/ft/tokenizer_config.json +51 -0
  22. nl_tasks/run_all/run_exps1/ft/training_args.bin +3 -0
  23. nl_tasks/run_all/run_exps1/ft2/adapter_config.json +18 -0
  24. nl_tasks/run_all/run_exps1/ft2/adapter_model.bin +3 -0
  25. nl_tasks/run_all/run_exps1/trainer_state.json +1858 -0
  26. nl_tasks/run_all/run_exps2/ft/adapter_config.json +18 -0
  27. nl_tasks/run_all/run_exps2/ft/special_tokens_map.json +30 -0
  28. nl_tasks/run_all/run_exps2/ft/tokenizer.json +0 -0
  29. nl_tasks/run_all/run_exps2/ft/tokenizer_config.json +51 -0
  30. nl_tasks/run_all/run_exps2/ft/training_args.bin +3 -0
  31. nl_tasks/run_all/run_exps2/ft2/adapter_config.json +18 -0
  32. nl_tasks/run_all/run_exps2/ft2/adapter_model.bin +3 -0
  33. nl_tasks/run_all/run_exps2/trainer_state.json +1858 -0
  34. nl_tasks/run_all/run_exps3/ft/adapter_config.json +18 -0
  35. nl_tasks/run_all/run_exps3/ft/special_tokens_map.json +30 -0
  36. nl_tasks/run_all/run_exps3/ft/tokenizer.json +0 -0
  37. nl_tasks/run_all/run_exps3/ft/tokenizer_config.json +51 -0
  38. nl_tasks/run_all/run_exps3/ft/training_args.bin +3 -0
  39. nl_tasks/run_all/run_exps3/ft2/adapter_config.json +18 -0
  40. nl_tasks/run_all/run_exps3/ft2/adapter_model.bin +3 -0
  41. nl_tasks/run_all/run_exps3/trainer_state.json +1858 -0
  42. nl_tasks/run_all/run_exps4/ft/adapter_config.json +18 -0
  43. nl_tasks/run_all/run_exps4/ft/special_tokens_map.json +30 -0
  44. nl_tasks/run_all/run_exps4/ft/tokenizer.json +0 -0
  45. nl_tasks/run_all/run_exps4/ft/tokenizer_config.json +51 -0
  46. nl_tasks/run_all/run_exps4/ft/training_args.bin +3 -0
  47. nl_tasks/run_all/run_exps4/ft2/adapter_config.json +18 -0
  48. nl_tasks/run_all/run_exps4/ft2/adapter_model.bin +3 -0
  49. nl_tasks/run_all/run_exps4/trainer_state.json +1858 -0
  50. nl_tasks/run_all/run_exps5/trainer_state.json +1858 -0
nl_tasks/run_all/run_exnr12/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 16,
9
+ "peft_type": "ROTATION",
10
+ "r": 1,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "v_proj",
14
+ "q_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exnr12/ft/added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[PAD]": 32000
3
+ }
nl_tasks/run_all/run_exnr12/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exnr12/ft/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
nl_tasks/run_all/run_exnr12/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f45572059226f068f7d7099d155791e95f2fa1267bcde04365a44462bc69490
3
+ size 6481
nl_tasks/run_all/run_exnr12/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 16,
9
+ "peft_type": "ROTATION",
10
+ "r": 1,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "v_proj",
14
+ "q_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exnr12/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae187d1551c544a38ed5541fabb49534794ceeff4e4985fa5abe345f1506ac93
3
+ size 33602659
nl_tasks/run_all/run_exnr13/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exnr13/ft/added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[PAD]": 32000
3
+ }
nl_tasks/run_all/run_exnr13/ft/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "</s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "</s>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
nl_tasks/run_all/run_exnr13/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exnr13/ft/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
nl_tasks/run_all/run_exnr13/ft/tokenizer_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ },
30
+ "32000": {
31
+ "content": "[PAD]",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": true
37
+ }
38
+ },
39
+ "bos_token": "</s>",
40
+ "clean_up_tokenization_spaces": false,
41
+ "eos_token": "</s>",
42
+ "extra_special_tokens": {},
43
+ "legacy": false,
44
+ "model_max_length": 512,
45
+ "pad_token": "[PAD]",
46
+ "padding_side": "right",
47
+ "sp_model_kwargs": {},
48
+ "tokenizer_class": "LlamaTokenizer",
49
+ "unk_token": "</s>",
50
+ "use_default_system_prompt": false
51
+ }
nl_tasks/run_all/run_exnr13/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:037f8b17117a8b6b8cbdbb85a67bb2636eb8dffaae6ac4b622ba5e37566621a9
3
+ size 6481
nl_tasks/run_all/run_exnr13/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 1,
9
+ "peft_type": "ROTATION",
10
+ "r": 16,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exnr13/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dcdf9e7560726805241e21c9550b6e4de0f6a48a704b8ee52845cbff82771bc2
3
+ size 33602915
nl_tasks/run_all/run_exnr13/trainer_state.json ADDED
@@ -0,0 +1,1106 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.0,
6
+ "eval_steps": 50,
7
+ "global_step": 2438,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.020508613617719443,
14
+ "grad_norm": 0.22931261360645294,
15
+ "learning_rate": 9.836065573770491e-05,
16
+ "loss": 0.678,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04101722723543889,
21
+ "grad_norm": 0.18445120751857758,
22
+ "learning_rate": 0.00020081967213114754,
23
+ "loss": 0.4156,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.04101722723543889,
28
+ "eval_loss": 0.3720075786113739,
29
+ "eval_runtime": 19.7665,
30
+ "eval_samples_per_second": 50.591,
31
+ "eval_steps_per_second": 0.809,
32
+ "step": 50
33
+ },
34
+ {
35
+ "epoch": 0.06152584085315833,
36
+ "grad_norm": 0.17781366407871246,
37
+ "learning_rate": 0.0003032786885245902,
38
+ "loss": 0.3699,
39
+ "step": 75
40
+ },
41
+ {
42
+ "epoch": 0.08203445447087777,
43
+ "grad_norm": 0.19237235188484192,
44
+ "learning_rate": 0.0004057377049180328,
45
+ "loss": 0.35,
46
+ "step": 100
47
+ },
48
+ {
49
+ "epoch": 0.08203445447087777,
50
+ "eval_loss": 0.3287361264228821,
51
+ "eval_runtime": 19.2993,
52
+ "eval_samples_per_second": 51.815,
53
+ "eval_steps_per_second": 0.829,
54
+ "step": 100
55
+ },
56
+ {
57
+ "epoch": 0.10254306808859721,
58
+ "grad_norm": 0.1524256020784378,
59
+ "learning_rate": 0.0005081967213114754,
60
+ "loss": 0.3296,
61
+ "step": 125
62
+ },
63
+ {
64
+ "epoch": 0.12305168170631665,
65
+ "grad_norm": 0.17991341650485992,
66
+ "learning_rate": 0.000610655737704918,
67
+ "loss": 0.3139,
68
+ "step": 150
69
+ },
70
+ {
71
+ "epoch": 0.12305168170631665,
72
+ "eval_loss": 0.31378939747810364,
73
+ "eval_runtime": 19.3746,
74
+ "eval_samples_per_second": 51.614,
75
+ "eval_steps_per_second": 0.826,
76
+ "step": 150
77
+ },
78
+ {
79
+ "epoch": 0.1435602953240361,
80
+ "grad_norm": 0.21170128881931305,
81
+ "learning_rate": 0.0007131147540983607,
82
+ "loss": 0.3132,
83
+ "step": 175
84
+ },
85
+ {
86
+ "epoch": 0.16406890894175555,
87
+ "grad_norm": 0.25959813594818115,
88
+ "learning_rate": 0.0008155737704918033,
89
+ "loss": 0.3198,
90
+ "step": 200
91
+ },
92
+ {
93
+ "epoch": 0.16406890894175555,
94
+ "eval_loss": 0.30950409173965454,
95
+ "eval_runtime": 19.3652,
96
+ "eval_samples_per_second": 51.639,
97
+ "eval_steps_per_second": 0.826,
98
+ "step": 200
99
+ },
100
+ {
101
+ "epoch": 0.184577522559475,
102
+ "grad_norm": 0.25062230229377747,
103
+ "learning_rate": 0.0009180327868852459,
104
+ "loss": 0.3264,
105
+ "step": 225
106
+ },
107
+ {
108
+ "epoch": 0.20508613617719443,
109
+ "grad_norm": 0.28308653831481934,
110
+ "learning_rate": 0.0009999871854116063,
111
+ "loss": 0.3204,
112
+ "step": 250
113
+ },
114
+ {
115
+ "epoch": 0.20508613617719443,
116
+ "eval_loss": 0.30955788493156433,
117
+ "eval_runtime": 19.2696,
118
+ "eval_samples_per_second": 51.895,
119
+ "eval_steps_per_second": 0.83,
120
+ "step": 250
121
+ },
122
+ {
123
+ "epoch": 0.22559474979491387,
124
+ "grad_norm": 0.9601122140884399,
125
+ "learning_rate": 0.0009995387437838027,
126
+ "loss": 0.3283,
127
+ "step": 275
128
+ },
129
+ {
130
+ "epoch": 0.2461033634126333,
131
+ "grad_norm": 0.306286096572876,
132
+ "learning_rate": 0.000998450229439693,
133
+ "loss": 0.3245,
134
+ "step": 300
135
+ },
136
+ {
137
+ "epoch": 0.2461033634126333,
138
+ "eval_loss": 0.3065921366214752,
139
+ "eval_runtime": 19.3494,
140
+ "eval_samples_per_second": 51.681,
141
+ "eval_steps_per_second": 0.827,
142
+ "step": 300
143
+ },
144
+ {
145
+ "epoch": 0.2666119770303528,
146
+ "grad_norm": 0.2865601181983948,
147
+ "learning_rate": 0.000996723037122612,
148
+ "loss": 0.3068,
149
+ "step": 325
150
+ },
151
+ {
152
+ "epoch": 0.2871205906480722,
153
+ "grad_norm": 0.2877274453639984,
154
+ "learning_rate": 0.0009943593799315263,
155
+ "loss": 0.311,
156
+ "step": 350
157
+ },
158
+ {
159
+ "epoch": 0.2871205906480722,
160
+ "eval_loss": 0.2946481704711914,
161
+ "eval_runtime": 19.3413,
162
+ "eval_samples_per_second": 51.703,
163
+ "eval_steps_per_second": 0.827,
164
+ "step": 350
165
+ },
166
+ {
167
+ "epoch": 0.30762920426579166,
168
+ "grad_norm": 0.28202855587005615,
169
+ "learning_rate": 0.0009913622864853324,
170
+ "loss": 0.2987,
171
+ "step": 375
172
+ },
173
+ {
174
+ "epoch": 0.3281378178835111,
175
+ "grad_norm": 0.2914764881134033,
176
+ "learning_rate": 0.0009877355970422024,
177
+ "loss": 0.3077,
178
+ "step": 400
179
+ },
180
+ {
181
+ "epoch": 0.3281378178835111,
182
+ "eval_loss": 0.284617155790329,
183
+ "eval_runtime": 19.3313,
184
+ "eval_samples_per_second": 51.73,
185
+ "eval_steps_per_second": 0.828,
186
+ "step": 400
187
+ },
188
+ {
189
+ "epoch": 0.34864643150123054,
190
+ "grad_norm": 0.21162962913513184,
191
+ "learning_rate": 0.0009834839585789559,
192
+ "loss": 0.3021,
193
+ "step": 425
194
+ },
195
+ {
196
+ "epoch": 0.36915504511895,
197
+ "grad_norm": 0.2067781537771225,
198
+ "learning_rate": 0.000978612818836762,
199
+ "loss": 0.2991,
200
+ "step": 450
201
+ },
202
+ {
203
+ "epoch": 0.36915504511895,
204
+ "eval_loss": 0.279570609331131,
205
+ "eval_runtime": 19.3677,
206
+ "eval_samples_per_second": 51.632,
207
+ "eval_steps_per_second": 0.826,
208
+ "step": 450
209
+ },
210
+ {
211
+ "epoch": 0.3896636587366694,
212
+ "grad_norm": 0.21456116437911987,
213
+ "learning_rate": 0.0009731284193407981,
214
+ "loss": 0.2948,
215
+ "step": 475
216
+ },
217
+ {
218
+ "epoch": 0.41017227235438886,
219
+ "grad_norm": 0.2070964127779007,
220
+ "learning_rate": 0.0009670377874028117,
221
+ "loss": 0.2846,
222
+ "step": 500
223
+ },
224
+ {
225
+ "epoch": 0.41017227235438886,
226
+ "eval_loss": 0.27043354511260986,
227
+ "eval_runtime": 19.3777,
228
+ "eval_samples_per_second": 51.606,
229
+ "eval_steps_per_second": 0.826,
230
+ "step": 500
231
+ },
232
+ {
233
+ "epoch": 0.4306808859721083,
234
+ "grad_norm": 0.2700050175189972,
235
+ "learning_rate": 0.0009603487271168336,
236
+ "loss": 0.2884,
237
+ "step": 525
238
+ },
239
+ {
240
+ "epoch": 0.45118949958982774,
241
+ "grad_norm": 0.20913958549499512,
242
+ "learning_rate": 0.0009530698093595781,
243
+ "loss": 0.2819,
244
+ "step": 550
245
+ },
246
+ {
247
+ "epoch": 0.45118949958982774,
248
+ "eval_loss": 0.26538342237472534,
249
+ "eval_runtime": 19.3572,
250
+ "eval_samples_per_second": 51.66,
251
+ "eval_steps_per_second": 0.827,
252
+ "step": 550
253
+ },
254
+ {
255
+ "epoch": 0.4716981132075472,
256
+ "grad_norm": 0.2159457504749298,
257
+ "learning_rate": 0.0009452103608083418,
258
+ "loss": 0.2708,
259
+ "step": 575
260
+ },
261
+ {
262
+ "epoch": 0.4922067268252666,
263
+ "grad_norm": 0.21397002041339874,
264
+ "learning_rate": 0.0009367804519904775,
265
+ "loss": 0.2802,
266
+ "step": 600
267
+ },
268
+ {
269
+ "epoch": 0.4922067268252666,
270
+ "eval_loss": 0.26951679587364197,
271
+ "eval_runtime": 19.3658,
272
+ "eval_samples_per_second": 51.637,
273
+ "eval_steps_per_second": 0.826,
274
+ "step": 600
275
+ },
276
+ {
277
+ "epoch": 0.5127153404429861,
278
+ "grad_norm": 0.17700916528701782,
279
+ "learning_rate": 0.0009277908843797492,
280
+ "loss": 0.2683,
281
+ "step": 625
282
+ },
283
+ {
284
+ "epoch": 0.5332239540607056,
285
+ "grad_norm": 0.20436996221542358,
286
+ "learning_rate": 0.0009182531765561084,
287
+ "loss": 0.2657,
288
+ "step": 650
289
+ },
290
+ {
291
+ "epoch": 0.5332239540607056,
292
+ "eval_loss": 0.25960028171539307,
293
+ "eval_runtime": 19.285,
294
+ "eval_samples_per_second": 51.854,
295
+ "eval_steps_per_second": 0.83,
296
+ "step": 650
297
+ },
298
+ {
299
+ "epoch": 0.5537325676784249,
300
+ "grad_norm": 0.1906837671995163,
301
+ "learning_rate": 0.0009081795494466201,
302
+ "loss": 0.2655,
303
+ "step": 675
304
+ },
305
+ {
306
+ "epoch": 0.5742411812961444,
307
+ "grad_norm": 0.17952291667461395,
308
+ "learning_rate": 0.0008975829106664539,
309
+ "loss": 0.257,
310
+ "step": 700
311
+ },
312
+ {
313
+ "epoch": 0.5742411812961444,
314
+ "eval_loss": 0.2571607530117035,
315
+ "eval_runtime": 19.3224,
316
+ "eval_samples_per_second": 51.753,
317
+ "eval_steps_per_second": 0.828,
318
+ "step": 700
319
+ },
320
+ {
321
+ "epoch": 0.5947497949138638,
322
+ "grad_norm": 0.1855345517396927,
323
+ "learning_rate": 0.0008864768379800017,
324
+ "loss": 0.2538,
325
+ "step": 725
326
+ },
327
+ {
328
+ "epoch": 0.6152584085315833,
329
+ "grad_norm": 0.16786430776119232,
330
+ "learning_rate": 0.0008748755619033153,
331
+ "loss": 0.2634,
332
+ "step": 750
333
+ },
334
+ {
335
+ "epoch": 0.6152584085315833,
336
+ "eval_loss": 0.25393542647361755,
337
+ "eval_runtime": 19.2859,
338
+ "eval_samples_per_second": 51.851,
339
+ "eval_steps_per_second": 0.83,
340
+ "step": 750
341
+ },
342
+ {
343
+ "epoch": 0.6357670221493027,
344
+ "grad_norm": 0.2089242935180664,
345
+ "learning_rate": 0.000862793947470155,
346
+ "loss": 0.2666,
347
+ "step": 775
348
+ },
349
+ {
350
+ "epoch": 0.6562756357670222,
351
+ "grad_norm": 0.1806870400905609,
352
+ "learning_rate": 0.0008502474751850142,
353
+ "loss": 0.2557,
354
+ "step": 800
355
+ },
356
+ {
357
+ "epoch": 0.6562756357670222,
358
+ "eval_loss": 0.24840402603149414,
359
+ "eval_runtime": 20.0247,
360
+ "eval_samples_per_second": 49.938,
361
+ "eval_steps_per_second": 0.799,
362
+ "step": 800
363
+ },
364
+ {
365
+ "epoch": 0.6767842493847416,
366
+ "grad_norm": 0.16617436707019806,
367
+ "learning_rate": 0.0008372522211875224,
368
+ "loss": 0.2498,
369
+ "step": 825
370
+ },
371
+ {
372
+ "epoch": 0.6972928630024611,
373
+ "grad_norm": 0.15508487820625305,
374
+ "learning_rate": 0.0008238248366536473,
375
+ "loss": 0.2631,
376
+ "step": 850
377
+ },
378
+ {
379
+ "epoch": 0.6972928630024611,
380
+ "eval_loss": 0.24494421482086182,
381
+ "eval_runtime": 19.2906,
382
+ "eval_samples_per_second": 51.839,
383
+ "eval_steps_per_second": 0.829,
384
+ "step": 850
385
+ },
386
+ {
387
+ "epoch": 0.7178014766201805,
388
+ "grad_norm": 0.13191890716552734,
389
+ "learning_rate": 0.0008099825264600842,
390
+ "loss": 0.2575,
391
+ "step": 875
392
+ },
393
+ {
394
+ "epoch": 0.7383100902379,
395
+ "grad_norm": 0.22435954213142395,
396
+ "learning_rate": 0.0007957430271391761,
397
+ "loss": 0.2492,
398
+ "step": 900
399
+ },
400
+ {
401
+ "epoch": 0.7383100902379,
402
+ "eval_loss": 0.24173128604888916,
403
+ "eval_runtime": 19.2803,
404
+ "eval_samples_per_second": 51.866,
405
+ "eval_steps_per_second": 0.83,
406
+ "step": 900
407
+ },
408
+ {
409
+ "epoch": 0.7588187038556193,
410
+ "grad_norm": 0.16361841559410095,
411
+ "learning_rate": 0.0007811245841526062,
412
+ "loss": 0.252,
413
+ "step": 925
414
+ },
415
+ {
416
+ "epoch": 0.7793273174733388,
417
+ "grad_norm": 0.1659049540758133,
418
+ "learning_rate": 0.0007661459285129879,
419
+ "loss": 0.2352,
420
+ "step": 950
421
+ },
422
+ {
423
+ "epoch": 0.7793273174733388,
424
+ "eval_loss": 0.23935402929782867,
425
+ "eval_runtime": 19.2861,
426
+ "eval_samples_per_second": 51.851,
427
+ "eval_steps_per_second": 0.83,
428
+ "step": 950
429
+ },
430
+ {
431
+ "epoch": 0.7998359310910582,
432
+ "grad_norm": 0.1710589975118637,
433
+ "learning_rate": 0.0007508262527833029,
434
+ "loss": 0.2475,
435
+ "step": 975
436
+ },
437
+ {
438
+ "epoch": 0.8203445447087777,
439
+ "grad_norm": 0.16337169706821442,
440
+ "learning_rate": 0.000735185186484943,
441
+ "loss": 0.2496,
442
+ "step": 1000
443
+ },
444
+ {
445
+ "epoch": 0.8203445447087777,
446
+ "eval_loss": 0.23631958663463593,
447
+ "eval_runtime": 19.2983,
448
+ "eval_samples_per_second": 51.818,
449
+ "eval_steps_per_second": 0.829,
450
+ "step": 1000
451
+ },
452
+ {
453
+ "epoch": 0.8408531583264971,
454
+ "grad_norm": 0.1336127668619156,
455
+ "learning_rate": 0.0007192427709458656,
456
+ "loss": 0.2367,
457
+ "step": 1025
458
+ },
459
+ {
460
+ "epoch": 0.8613617719442166,
461
+ "grad_norm": 0.1590573489665985,
462
+ "learning_rate": 0.0007030194336210887,
463
+ "loss": 0.2415,
464
+ "step": 1050
465
+ },
466
+ {
467
+ "epoch": 0.8613617719442166,
468
+ "eval_loss": 0.23346620798110962,
469
+ "eval_runtime": 19.3105,
470
+ "eval_samples_per_second": 51.785,
471
+ "eval_steps_per_second": 0.829,
472
+ "step": 1050
473
+ },
474
+ {
475
+ "epoch": 0.881870385561936,
476
+ "grad_norm": 0.16918089985847473,
477
+ "learning_rate": 0.0006865359619184331,
478
+ "loss": 0.2452,
479
+ "step": 1075
480
+ },
481
+ {
482
+ "epoch": 0.9023789991796555,
483
+ "grad_norm": 0.14695101976394653,
484
+ "learning_rate": 0.0006698134765630434,
485
+ "loss": 0.2386,
486
+ "step": 1100
487
+ },
488
+ {
489
+ "epoch": 0.9023789991796555,
490
+ "eval_loss": 0.2332371324300766,
491
+ "eval_runtime": 19.3273,
492
+ "eval_samples_per_second": 51.74,
493
+ "eval_steps_per_second": 0.828,
494
+ "step": 1100
495
+ },
496
+ {
497
+ "epoch": 0.9228876127973749,
498
+ "grad_norm": 0.12737445533275604,
499
+ "learning_rate": 0.0006528734045348248,
500
+ "loss": 0.2374,
501
+ "step": 1125
502
+ },
503
+ {
504
+ "epoch": 0.9433962264150944,
505
+ "grad_norm": 0.1528356820344925,
506
+ "learning_rate": 0.0006357374516134643,
507
+ "loss": 0.2519,
508
+ "step": 1150
509
+ },
510
+ {
511
+ "epoch": 0.9433962264150944,
512
+ "eval_loss": 0.23056186735630035,
513
+ "eval_runtime": 19.2927,
514
+ "eval_samples_per_second": 51.833,
515
+ "eval_steps_per_second": 0.829,
516
+ "step": 1150
517
+ },
518
+ {
519
+ "epoch": 0.9639048400328137,
520
+ "grad_norm": 0.1549467146396637,
521
+ "learning_rate": 0.0006184275745662179,
522
+ "loss": 0.2396,
523
+ "step": 1175
524
+ },
525
+ {
526
+ "epoch": 0.9844134536505332,
527
+ "grad_norm": 0.1511608213186264,
528
+ "learning_rate": 0.0006009659530141031,
529
+ "loss": 0.2408,
530
+ "step": 1200
531
+ },
532
+ {
533
+ "epoch": 0.9844134536505332,
534
+ "eval_loss": 0.22885626554489136,
535
+ "eval_runtime": 19.2756,
536
+ "eval_samples_per_second": 51.879,
537
+ "eval_steps_per_second": 0.83,
538
+ "step": 1200
539
+ },
540
+ {
541
+ "epoch": 1.0049220672682526,
542
+ "grad_norm": 0.14819197356700897,
543
+ "learning_rate": 0.0005833749610125402,
544
+ "loss": 0.2362,
545
+ "step": 1225
546
+ },
547
+ {
548
+ "epoch": 1.0254306808859721,
549
+ "grad_norm": 0.13643483817577362,
550
+ "learning_rate": 0.0005656771383828602,
551
+ "loss": 0.2084,
552
+ "step": 1250
553
+ },
554
+ {
555
+ "epoch": 1.0254306808859721,
556
+ "eval_loss": 0.22903910279273987,
557
+ "eval_runtime": 19.3112,
558
+ "eval_samples_per_second": 51.784,
559
+ "eval_steps_per_second": 0.829,
560
+ "step": 1250
561
+ },
562
+ {
563
+ "epoch": 1.0459392945036916,
564
+ "grad_norm": 0.13602110743522644,
565
+ "learning_rate": 0.0005478951618314134,
566
+ "loss": 0.2025,
567
+ "step": 1275
568
+ },
569
+ {
570
+ "epoch": 1.066447908121411,
571
+ "grad_norm": 0.1276039481163025,
572
+ "learning_rate": 0.0005300518158932815,
573
+ "loss": 0.2045,
574
+ "step": 1300
575
+ },
576
+ {
577
+ "epoch": 1.066447908121411,
578
+ "eval_loss": 0.22647729516029358,
579
+ "eval_runtime": 19.3123,
580
+ "eval_samples_per_second": 51.781,
581
+ "eval_steps_per_second": 0.828,
582
+ "step": 1300
583
+ },
584
+ {
585
+ "epoch": 1.0869565217391304,
586
+ "grad_norm": 0.1431492418050766,
587
+ "learning_rate": 0.0005121699637378282,
588
+ "loss": 0.203,
589
+ "step": 1325
590
+ },
591
+ {
592
+ "epoch": 1.1074651353568499,
593
+ "grad_norm": 0.15327255427837372,
594
+ "learning_rate": 0.0004942725178734903,
595
+ "loss": 0.2076,
596
+ "step": 1350
597
+ },
598
+ {
599
+ "epoch": 1.1074651353568499,
600
+ "eval_loss": 0.22494623064994812,
601
+ "eval_runtime": 19.272,
602
+ "eval_samples_per_second": 51.889,
603
+ "eval_steps_per_second": 0.83,
604
+ "step": 1350
605
+ },
606
+ {
607
+ "epoch": 1.1279737489745694,
608
+ "grad_norm": 0.14988966286182404,
609
+ "learning_rate": 0.00047638241078935324,
610
+ "loss": 0.2051,
611
+ "step": 1375
612
+ },
613
+ {
614
+ "epoch": 1.1484823625922886,
615
+ "grad_norm": 0.1339106261730194,
616
+ "learning_rate": 0.000458522565571121,
617
+ "loss": 0.2128,
618
+ "step": 1400
619
+ },
620
+ {
621
+ "epoch": 1.1484823625922886,
622
+ "eval_loss": 0.22338636219501495,
623
+ "eval_runtime": 19.3087,
624
+ "eval_samples_per_second": 51.79,
625
+ "eval_steps_per_second": 0.829,
626
+ "step": 1400
627
+ },
628
+ {
629
+ "epoch": 1.1689909762100081,
630
+ "grad_norm": 0.1396779716014862,
631
+ "learning_rate": 0.00044071586652913767,
632
+ "loss": 0.1965,
633
+ "step": 1425
634
+ },
635
+ {
636
+ "epoch": 1.1894995898277276,
637
+ "grad_norm": 0.15899169445037842,
638
+ "learning_rate": 0.0004229851298760915,
639
+ "loss": 0.2063,
640
+ "step": 1450
641
+ },
642
+ {
643
+ "epoch": 1.1894995898277276,
644
+ "eval_loss": 0.22318963706493378,
645
+ "eval_runtime": 19.3444,
646
+ "eval_samples_per_second": 51.695,
647
+ "eval_steps_per_second": 0.827,
648
+ "step": 1450
649
+ },
650
+ {
651
+ "epoch": 1.2100082034454471,
652
+ "grad_norm": 0.13465991616249084,
653
+ "learning_rate": 0.0004053530744919749,
654
+ "loss": 0.2055,
655
+ "step": 1475
656
+ },
657
+ {
658
+ "epoch": 1.2305168170631666,
659
+ "grad_norm": 0.15314656496047974,
660
+ "learning_rate": 0.0003878422928137597,
661
+ "loss": 0.2091,
662
+ "step": 1500
663
+ },
664
+ {
665
+ "epoch": 1.2305168170631666,
666
+ "eval_loss": 0.22200444340705872,
667
+ "eval_runtime": 19.3449,
668
+ "eval_samples_per_second": 51.693,
669
+ "eval_steps_per_second": 0.827,
670
+ "step": 1500
671
+ },
672
+ {
673
+ "epoch": 1.251025430680886,
674
+ "grad_norm": 0.16617698967456818,
675
+ "learning_rate": 0.0003704752218870861,
676
+ "loss": 0.2132,
677
+ "step": 1525
678
+ },
679
+ {
680
+ "epoch": 1.2715340442986054,
681
+ "grad_norm": 0.1396445631980896,
682
+ "learning_rate": 0.00035327411461706025,
683
+ "loss": 0.1952,
684
+ "step": 1550
685
+ },
686
+ {
687
+ "epoch": 1.2715340442986054,
688
+ "eval_loss": 0.22176627814769745,
689
+ "eval_runtime": 19.3568,
690
+ "eval_samples_per_second": 51.661,
691
+ "eval_steps_per_second": 0.827,
692
+ "step": 1550
693
+ },
694
+ {
695
+ "epoch": 1.2920426579163249,
696
+ "grad_norm": 0.14781972765922546,
697
+ "learning_rate": 0.00033626101125499555,
698
+ "loss": 0.2058,
699
+ "step": 1575
700
+ },
701
+ {
702
+ "epoch": 1.3125512715340442,
703
+ "grad_norm": 0.13822010159492493,
704
+ "learning_rate": 0.0003194577111576333,
705
+ "loss": 0.2002,
706
+ "step": 1600
707
+ },
708
+ {
709
+ "epoch": 1.3125512715340442,
710
+ "eval_loss": 0.21868599951267242,
711
+ "eval_runtime": 19.3967,
712
+ "eval_samples_per_second": 51.555,
713
+ "eval_steps_per_second": 0.825,
714
+ "step": 1600
715
+ },
716
+ {
717
+ "epoch": 1.3330598851517639,
718
+ "grad_norm": 0.12466410547494888,
719
+ "learning_rate": 0.00030288574485502756,
720
+ "loss": 0.1991,
721
+ "step": 1625
722
+ },
723
+ {
724
+ "epoch": 1.3535684987694832,
725
+ "grad_norm": 0.13315178453922272,
726
+ "learning_rate": 0.00028656634646288565,
727
+ "loss": 0.1999,
728
+ "step": 1650
729
+ },
730
+ {
731
+ "epoch": 1.3535684987694832,
732
+ "eval_loss": 0.2176971733570099,
733
+ "eval_runtime": 19.3868,
734
+ "eval_samples_per_second": 51.582,
735
+ "eval_steps_per_second": 0.825,
736
+ "step": 1650
737
+ },
738
+ {
739
+ "epoch": 1.3740771123872026,
740
+ "grad_norm": 0.16215480864048004,
741
+ "learning_rate": 0.00027052042647471254,
742
+ "loss": 0.1899,
743
+ "step": 1675
744
+ },
745
+ {
746
+ "epoch": 1.3945857260049221,
747
+ "grad_norm": 0.1372925043106079,
748
+ "learning_rate": 0.0002547685449686206,
749
+ "loss": 0.1982,
750
+ "step": 1700
751
+ },
752
+ {
753
+ "epoch": 1.3945857260049221,
754
+ "eval_loss": 0.21591512858867645,
755
+ "eval_runtime": 19.3868,
756
+ "eval_samples_per_second": 51.581,
757
+ "eval_steps_per_second": 0.825,
758
+ "step": 1700
759
+ },
760
+ {
761
+ "epoch": 1.4150943396226414,
762
+ "grad_norm": 0.15481165051460266,
763
+ "learning_rate": 0.0002393308852631373,
764
+ "loss": 0.2096,
765
+ "step": 1725
766
+ },
767
+ {
768
+ "epoch": 1.435602953240361,
769
+ "grad_norm": 0.14576007425785065,
770
+ "learning_rate": 0.0002242272280557645,
771
+ "loss": 0.2033,
772
+ "step": 1750
773
+ },
774
+ {
775
+ "epoch": 1.435602953240361,
776
+ "eval_loss": 0.21569040417671204,
777
+ "eval_runtime": 19.3745,
778
+ "eval_samples_per_second": 51.614,
779
+ "eval_steps_per_second": 0.826,
780
+ "step": 1750
781
+ },
782
+ {
783
+ "epoch": 1.4561115668580804,
784
+ "grad_norm": 0.14915944635868073,
785
+ "learning_rate": 0.0002094769260774262,
786
+ "loss": 0.1988,
787
+ "step": 1775
788
+ },
789
+ {
790
+ "epoch": 1.4766201804758,
791
+ "grad_norm": 0.13568426668643951,
792
+ "learning_rate": 0.00019509887929528458,
793
+ "loss": 0.2031,
794
+ "step": 1800
795
+ },
796
+ {
797
+ "epoch": 1.4766201804758,
798
+ "eval_loss": 0.21403798460960388,
799
+ "eval_runtime": 19.3655,
800
+ "eval_samples_per_second": 51.638,
801
+ "eval_steps_per_second": 0.826,
802
+ "step": 1800
803
+ },
804
+ {
805
+ "epoch": 1.4971287940935194,
806
+ "grad_norm": 0.15686564147472382,
807
+ "learning_rate": 0.0001811115106956918,
808
+ "loss": 0.1939,
809
+ "step": 1825
810
+ },
811
+ {
812
+ "epoch": 1.5176374077112387,
813
+ "grad_norm": 0.16230961680412292,
814
+ "learning_rate": 0.00016753274267831115,
815
+ "loss": 0.2037,
816
+ "step": 1850
817
+ },
818
+ {
819
+ "epoch": 1.5176374077112387,
820
+ "eval_loss": 0.21287399530410767,
821
+ "eval_runtime": 19.3395,
822
+ "eval_samples_per_second": 51.708,
823
+ "eval_steps_per_second": 0.827,
824
+ "step": 1850
825
+ },
826
+ {
827
+ "epoch": 1.5381460213289582,
828
+ "grad_norm": 0.1522003710269928,
829
+ "learning_rate": 0.00015437997409165478,
830
+ "loss": 0.1934,
831
+ "step": 1875
832
+ },
833
+ {
834
+ "epoch": 1.5586546349466777,
835
+ "grad_norm": 0.1406460851430893,
836
+ "learning_rate": 0.00014167005793946035,
837
+ "loss": 0.193,
838
+ "step": 1900
839
+ },
840
+ {
841
+ "epoch": 1.5586546349466777,
842
+ "eval_loss": 0.21294145286083221,
843
+ "eval_runtime": 19.3135,
844
+ "eval_samples_per_second": 51.777,
845
+ "eval_steps_per_second": 0.828,
846
+ "step": 1900
847
+ },
848
+ {
849
+ "epoch": 1.579163248564397,
850
+ "grad_norm": 0.150811567902565,
851
+ "learning_rate": 0.00012941927978647527,
852
+ "loss": 0.1945,
853
+ "step": 1925
854
+ },
855
+ {
856
+ "epoch": 1.5996718621821167,
857
+ "grad_norm": 0.12927386164665222,
858
+ "learning_rate": 0.00011764333689131385,
859
+ "loss": 0.2009,
860
+ "step": 1950
861
+ },
862
+ {
863
+ "epoch": 1.5996718621821167,
864
+ "eval_loss": 0.21167391538619995,
865
+ "eval_runtime": 19.3462,
866
+ "eval_samples_per_second": 51.69,
867
+ "eval_steps_per_second": 0.827,
868
+ "step": 1950
869
+ },
870
+ {
871
+ "epoch": 1.620180475799836,
872
+ "grad_norm": 0.11802957952022552,
873
+ "learning_rate": 0.00010635731809312993,
874
+ "loss": 0.1971,
875
+ "step": 1975
876
+ },
877
+ {
878
+ "epoch": 1.6406890894175554,
879
+ "grad_norm": 0.1508980095386505,
880
+ "learning_rate": 9.557568447787201e-05,
881
+ "loss": 0.1964,
882
+ "step": 2000
883
+ },
884
+ {
885
+ "epoch": 1.6406890894175554,
886
+ "eval_loss": 0.21123281121253967,
887
+ "eval_runtime": 19.3419,
888
+ "eval_samples_per_second": 51.701,
889
+ "eval_steps_per_second": 0.827,
890
+ "step": 2000
891
+ },
892
+ {
893
+ "epoch": 1.661197703035275,
894
+ "grad_norm": 0.13394318521022797,
895
+ "learning_rate": 8.531225084889654e-05,
896
+ "loss": 0.1893,
897
+ "step": 2025
898
+ },
899
+ {
900
+ "epoch": 1.6817063166529942,
901
+ "grad_norm": 0.15524154901504517,
902
+ "learning_rate": 7.558016802568091e-05,
903
+ "loss": 0.1938,
904
+ "step": 2050
905
+ },
906
+ {
907
+ "epoch": 1.6817063166529942,
908
+ "eval_loss": 0.2103249877691269,
909
+ "eval_runtime": 19.3209,
910
+ "eval_samples_per_second": 51.758,
911
+ "eval_steps_per_second": 0.828,
912
+ "step": 2050
913
+ },
914
+ {
915
+ "epoch": 1.7022149302707137,
916
+ "grad_norm": 0.14961044490337372,
917
+ "learning_rate": 6.639190599331746e-05,
918
+ "loss": 0.1936,
919
+ "step": 2075
920
+ },
921
+ {
922
+ "epoch": 1.7227235438884332,
923
+ "grad_norm": 0.13738760352134705,
924
+ "learning_rate": 5.775923792437865e-05,
925
+ "loss": 0.1891,
926
+ "step": 2100
927
+ },
928
+ {
929
+ "epoch": 1.7227235438884332,
930
+ "eval_loss": 0.20943871140480042,
931
+ "eval_runtime": 19.3831,
932
+ "eval_samples_per_second": 51.591,
933
+ "eval_steps_per_second": 0.825,
934
+ "step": 2100
935
+ },
936
+ {
937
+ "epoch": 1.7432321575061525,
938
+ "grad_norm": 0.1795763075351715,
939
+ "learning_rate": 4.9693225093627616e-05,
940
+ "loss": 0.1914,
941
+ "step": 2125
942
+ },
943
+ {
944
+ "epoch": 1.7637407711238722,
945
+ "grad_norm": 0.14526569843292236,
946
+ "learning_rate": 4.220420270490294e-05,
947
+ "loss": 0.1845,
948
+ "step": 2150
949
+ },
950
+ {
951
+ "epoch": 1.7637407711238722,
952
+ "eval_loss": 0.20901177823543549,
953
+ "eval_runtime": 19.3815,
954
+ "eval_samples_per_second": 51.596,
955
+ "eval_steps_per_second": 0.826,
956
+ "step": 2150
957
+ },
958
+ {
959
+ "epoch": 1.7842493847415914,
960
+ "grad_norm": 0.1493556648492813,
961
+ "learning_rate": 3.530176664833834e-05,
962
+ "loss": 0.1874,
963
+ "step": 2175
964
+ },
965
+ {
966
+ "epoch": 1.804757998359311,
967
+ "grad_norm": 0.1753922700881958,
968
+ "learning_rate": 2.8994761204884756e-05,
969
+ "loss": 0.1957,
970
+ "step": 2200
971
+ },
972
+ {
973
+ "epoch": 1.804757998359311,
974
+ "eval_loss": 0.20887655019760132,
975
+ "eval_runtime": 19.3659,
976
+ "eval_samples_per_second": 51.637,
977
+ "eval_steps_per_second": 0.826,
978
+ "step": 2200
979
+ },
980
+ {
981
+ "epoch": 1.8252666119770304,
982
+ "grad_norm": 0.15183115005493164,
983
+ "learning_rate": 2.329126771388995e-05,
984
+ "loss": 0.188,
985
+ "step": 2225
986
+ },
987
+ {
988
+ "epoch": 1.8457752255947497,
989
+ "grad_norm": 0.13431748747825623,
990
+ "learning_rate": 1.8198594218256815e-05,
991
+ "loss": 0.1873,
992
+ "step": 2250
993
+ },
994
+ {
995
+ "epoch": 1.8457752255947497,
996
+ "eval_loss": 0.20877276360988617,
997
+ "eval_runtime": 19.3708,
998
+ "eval_samples_per_second": 51.624,
999
+ "eval_steps_per_second": 0.826,
1000
+ "step": 2250
1001
+ },
1002
+ {
1003
+ "epoch": 1.8662838392124692,
1004
+ "grad_norm": 0.14999055862426758,
1005
+ "learning_rate": 1.3723266100447052e-05,
1006
+ "loss": 0.1915,
1007
+ "step": 2275
1008
+ },
1009
+ {
1010
+ "epoch": 1.8867924528301887,
1011
+ "grad_norm": 0.1781681329011917,
1012
+ "learning_rate": 9.871017721329201e-06,
1013
+ "loss": 0.1991,
1014
+ "step": 2300
1015
+ },
1016
+ {
1017
+ "epoch": 1.8867924528301887,
1018
+ "eval_loss": 0.20860543847084045,
1019
+ "eval_runtime": 19.3513,
1020
+ "eval_samples_per_second": 51.676,
1021
+ "eval_steps_per_second": 0.827,
1022
+ "step": 2300
1023
+ },
1024
+ {
1025
+ "epoch": 1.907301066447908,
1026
+ "grad_norm": 0.158808171749115,
1027
+ "learning_rate": 6.646785072584871e-06,
1028
+ "loss": 0.1937,
1029
+ "step": 2325
1030
+ },
1031
+ {
1032
+ "epoch": 1.9278096800656277,
1033
+ "grad_norm": 0.1277729570865631,
1034
+ "learning_rate": 4.054699452086641e-06,
1035
+ "loss": 0.1882,
1036
+ "step": 2350
1037
+ },
1038
+ {
1039
+ "epoch": 1.9278096800656277,
1040
+ "eval_loss": 0.20845511555671692,
1041
+ "eval_runtime": 19.3624,
1042
+ "eval_samples_per_second": 51.646,
1043
+ "eval_steps_per_second": 0.826,
1044
+ "step": 2350
1045
+ },
1046
+ {
1047
+ "epoch": 1.948318293683347,
1048
+ "grad_norm": 0.15287554264068604,
1049
+ "learning_rate": 2.0980821703527886e-06,
1050
+ "loss": 0.2023,
1051
+ "step": 2375
1052
+ },
1053
+ {
1054
+ "epoch": 1.9688269073010665,
1055
+ "grad_norm": 0.143690288066864,
1056
+ "learning_rate": 7.794402948607671e-07,
1057
+ "loss": 0.1975,
1058
+ "step": 2400
1059
+ },
1060
+ {
1061
+ "epoch": 1.9688269073010665,
1062
+ "eval_loss": 0.20845800638198853,
1063
+ "eval_runtime": 19.3666,
1064
+ "eval_samples_per_second": 51.635,
1065
+ "eval_steps_per_second": 0.826,
1066
+ "step": 2400
1067
+ },
1068
+ {
1069
+ "epoch": 1.989335520918786,
1070
+ "grad_norm": 0.1555429846048355,
1071
+ "learning_rate": 1.0046343767294853e-07,
1072
+ "loss": 0.1874,
1073
+ "step": 2425
1074
+ },
1075
+ {
1076
+ "epoch": 2.0,
1077
+ "step": 2438,
1078
+ "total_flos": 1.58523627405312e+18,
1079
+ "train_loss": 0.24400284996376945,
1080
+ "train_runtime": 3493.6883,
1081
+ "train_samples_per_second": 22.326,
1082
+ "train_steps_per_second": 0.698
1083
+ }
1084
+ ],
1085
+ "logging_steps": 25,
1086
+ "max_steps": 2438,
1087
+ "num_input_tokens_seen": 0,
1088
+ "num_train_epochs": 2,
1089
+ "save_steps": 500,
1090
+ "stateful_callbacks": {
1091
+ "TrainerControl": {
1092
+ "args": {
1093
+ "should_epoch_stop": false,
1094
+ "should_evaluate": false,
1095
+ "should_log": false,
1096
+ "should_save": true,
1097
+ "should_training_stop": true
1098
+ },
1099
+ "attributes": {}
1100
+ }
1101
+ },
1102
+ "total_flos": 1.58523627405312e+18,
1103
+ "train_batch_size": 32,
1104
+ "trial_name": null,
1105
+ "trial_params": null
1106
+ }
nl_tasks/run_all/run_exps1/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 4,
9
+ "peft_type": "ROTATION",
10
+ "r": 4,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exps1/ft/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "</s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "</s>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
nl_tasks/run_all/run_exps1/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exps1/ft/tokenizer_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ },
30
+ "32000": {
31
+ "content": "[PAD]",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": true
37
+ }
38
+ },
39
+ "bos_token": "</s>",
40
+ "clean_up_tokenization_spaces": false,
41
+ "eos_token": "</s>",
42
+ "extra_special_tokens": {},
43
+ "legacy": false,
44
+ "model_max_length": 512,
45
+ "pad_token": "[PAD]",
46
+ "padding_side": "right",
47
+ "sp_model_kwargs": {},
48
+ "tokenizer_class": "LlamaTokenizer",
49
+ "unk_token": "</s>",
50
+ "use_default_system_prompt": false
51
+ }
nl_tasks/run_all/run_exps1/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b2dc39220b8bf437e78f744960d44ba676c8fa62c3e03acc06e628cc3c43434c
3
+ size 6481
nl_tasks/run_all/run_exps1/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 4,
9
+ "peft_type": "ROTATION",
10
+ "r": 4,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exps1/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac3cbb71308f2b8b30414200d42cdeb373922169820bf76bf91203c905aec46d
3
+ size 33602659
nl_tasks/run_all/run_exps1/trainer_state.json ADDED
@@ -0,0 +1,1858 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.0,
6
+ "eval_steps": 20,
7
+ "global_step": 2438,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.016406890894175553,
14
+ "grad_norm": 0.045311760157346725,
15
+ "learning_rate": 7.78688524590164e-07,
16
+ "loss": 0.7464,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.016406890894175553,
21
+ "eval_loss": 0.744476854801178,
22
+ "eval_runtime": 19.903,
23
+ "eval_samples_per_second": 50.244,
24
+ "eval_steps_per_second": 0.804,
25
+ "step": 20
26
+ },
27
+ {
28
+ "epoch": 0.03281378178835111,
29
+ "grad_norm": 0.06288333982229233,
30
+ "learning_rate": 1.5983606557377053e-06,
31
+ "loss": 0.7592,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.03281378178835111,
36
+ "eval_loss": 0.7430104613304138,
37
+ "eval_runtime": 19.321,
38
+ "eval_samples_per_second": 51.757,
39
+ "eval_steps_per_second": 0.828,
40
+ "step": 40
41
+ },
42
+ {
43
+ "epoch": 0.04922067268252666,
44
+ "grad_norm": 0.09129688143730164,
45
+ "learning_rate": 2.418032786885246e-06,
46
+ "loss": 0.7476,
47
+ "step": 60
48
+ },
49
+ {
50
+ "epoch": 0.04922067268252666,
51
+ "eval_loss": 0.7391518354415894,
52
+ "eval_runtime": 19.3266,
53
+ "eval_samples_per_second": 51.742,
54
+ "eval_steps_per_second": 0.828,
55
+ "step": 60
56
+ },
57
+ {
58
+ "epoch": 0.06562756357670221,
59
+ "grad_norm": 0.15382955968379974,
60
+ "learning_rate": 3.2377049180327876e-06,
61
+ "loss": 0.7378,
62
+ "step": 80
63
+ },
64
+ {
65
+ "epoch": 0.06562756357670221,
66
+ "eval_loss": 0.7282525897026062,
67
+ "eval_runtime": 19.3273,
68
+ "eval_samples_per_second": 51.74,
69
+ "eval_steps_per_second": 0.828,
70
+ "step": 80
71
+ },
72
+ {
73
+ "epoch": 0.08203445447087777,
74
+ "grad_norm": 0.211552694439888,
75
+ "learning_rate": 4.057377049180329e-06,
76
+ "loss": 0.7309,
77
+ "step": 100
78
+ },
79
+ {
80
+ "epoch": 0.08203445447087777,
81
+ "eval_loss": 0.7028547525405884,
82
+ "eval_runtime": 19.3319,
83
+ "eval_samples_per_second": 51.728,
84
+ "eval_steps_per_second": 0.828,
85
+ "step": 100
86
+ },
87
+ {
88
+ "epoch": 0.09844134536505332,
89
+ "grad_norm": 0.2945902347564697,
90
+ "learning_rate": 4.877049180327869e-06,
91
+ "loss": 0.6809,
92
+ "step": 120
93
+ },
94
+ {
95
+ "epoch": 0.09844134536505332,
96
+ "eval_loss": 0.6537977457046509,
97
+ "eval_runtime": 19.3409,
98
+ "eval_samples_per_second": 51.704,
99
+ "eval_steps_per_second": 0.827,
100
+ "step": 120
101
+ },
102
+ {
103
+ "epoch": 0.11484823625922888,
104
+ "grad_norm": 0.2522474229335785,
105
+ "learning_rate": 5.696721311475411e-06,
106
+ "loss": 0.6215,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.11484823625922888,
111
+ "eval_loss": 0.5743948817253113,
112
+ "eval_runtime": 19.3206,
113
+ "eval_samples_per_second": 51.758,
114
+ "eval_steps_per_second": 0.828,
115
+ "step": 140
116
+ },
117
+ {
118
+ "epoch": 0.13125512715340443,
119
+ "grad_norm": 0.2089417278766632,
120
+ "learning_rate": 6.516393442622952e-06,
121
+ "loss": 0.5423,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.13125512715340443,
126
+ "eval_loss": 0.508235514163971,
127
+ "eval_runtime": 19.3328,
128
+ "eval_samples_per_second": 51.725,
129
+ "eval_steps_per_second": 0.828,
130
+ "step": 160
131
+ },
132
+ {
133
+ "epoch": 0.14766201804757997,
134
+ "grad_norm": 0.15879088640213013,
135
+ "learning_rate": 7.336065573770492e-06,
136
+ "loss": 0.497,
137
+ "step": 180
138
+ },
139
+ {
140
+ "epoch": 0.14766201804757997,
141
+ "eval_loss": 0.46629706025123596,
142
+ "eval_runtime": 19.3303,
143
+ "eval_samples_per_second": 51.732,
144
+ "eval_steps_per_second": 0.828,
145
+ "step": 180
146
+ },
147
+ {
148
+ "epoch": 0.16406890894175555,
149
+ "grad_norm": 0.13250677287578583,
150
+ "learning_rate": 8.155737704918034e-06,
151
+ "loss": 0.4643,
152
+ "step": 200
153
+ },
154
+ {
155
+ "epoch": 0.16406890894175555,
156
+ "eval_loss": 0.44612953066825867,
157
+ "eval_runtime": 19.3338,
158
+ "eval_samples_per_second": 51.723,
159
+ "eval_steps_per_second": 0.828,
160
+ "step": 200
161
+ },
162
+ {
163
+ "epoch": 0.1804757998359311,
164
+ "grad_norm": 0.16223840415477753,
165
+ "learning_rate": 8.975409836065575e-06,
166
+ "loss": 0.4557,
167
+ "step": 220
168
+ },
169
+ {
170
+ "epoch": 0.1804757998359311,
171
+ "eval_loss": 0.4310664236545563,
172
+ "eval_runtime": 19.3246,
173
+ "eval_samples_per_second": 51.748,
174
+ "eval_steps_per_second": 0.828,
175
+ "step": 220
176
+ },
177
+ {
178
+ "epoch": 0.19688269073010664,
179
+ "grad_norm": 0.17253851890563965,
180
+ "learning_rate": 9.795081967213116e-06,
181
+ "loss": 0.4395,
182
+ "step": 240
183
+ },
184
+ {
185
+ "epoch": 0.19688269073010664,
186
+ "eval_loss": 0.4179575741291046,
187
+ "eval_runtime": 19.3165,
188
+ "eval_samples_per_second": 51.769,
189
+ "eval_steps_per_second": 0.828,
190
+ "step": 240
191
+ },
192
+ {
193
+ "epoch": 0.2132895816242822,
194
+ "grad_norm": 0.17709316313266754,
195
+ "learning_rate": 9.99884672645552e-06,
196
+ "loss": 0.4225,
197
+ "step": 260
198
+ },
199
+ {
200
+ "epoch": 0.2132895816242822,
201
+ "eval_loss": 0.40669238567352295,
202
+ "eval_runtime": 19.3402,
203
+ "eval_samples_per_second": 51.706,
204
+ "eval_steps_per_second": 0.827,
205
+ "step": 260
206
+ },
207
+ {
208
+ "epoch": 0.22969647251845776,
209
+ "grad_norm": 0.163985013961792,
210
+ "learning_rate": 9.993722139023359e-06,
211
+ "loss": 0.4171,
212
+ "step": 280
213
+ },
214
+ {
215
+ "epoch": 0.22969647251845776,
216
+ "eval_loss": 0.3979131281375885,
217
+ "eval_runtime": 19.3591,
218
+ "eval_samples_per_second": 51.655,
219
+ "eval_steps_per_second": 0.826,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2461033634126333,
224
+ "grad_norm": 0.21911385655403137,
225
+ "learning_rate": 9.984502294396932e-06,
226
+ "loss": 0.4008,
227
+ "step": 300
228
+ },
229
+ {
230
+ "epoch": 0.2461033634126333,
231
+ "eval_loss": 0.3902350068092346,
232
+ "eval_runtime": 19.3359,
233
+ "eval_samples_per_second": 51.717,
234
+ "eval_steps_per_second": 0.827,
235
+ "step": 300
236
+ },
237
+ {
238
+ "epoch": 0.26251025430680885,
239
+ "grad_norm": 0.18699711561203003,
240
+ "learning_rate": 9.971194753596649e-06,
241
+ "loss": 0.3905,
242
+ "step": 320
243
+ },
244
+ {
245
+ "epoch": 0.26251025430680885,
246
+ "eval_loss": 0.38398095965385437,
247
+ "eval_runtime": 19.3371,
248
+ "eval_samples_per_second": 51.714,
249
+ "eval_steps_per_second": 0.827,
250
+ "step": 320
251
+ },
252
+ {
253
+ "epoch": 0.27891714520098443,
254
+ "grad_norm": 0.20045825839042664,
255
+ "learning_rate": 9.953810429885345e-06,
256
+ "loss": 0.3976,
257
+ "step": 340
258
+ },
259
+ {
260
+ "epoch": 0.27891714520098443,
261
+ "eval_loss": 0.37802764773368835,
262
+ "eval_runtime": 19.3296,
263
+ "eval_samples_per_second": 51.734,
264
+ "eval_steps_per_second": 0.828,
265
+ "step": 340
266
+ },
267
+ {
268
+ "epoch": 0.29532403609515995,
269
+ "grad_norm": 0.21255679428577423,
270
+ "learning_rate": 9.9323635798185e-06,
271
+ "loss": 0.3759,
272
+ "step": 360
273
+ },
274
+ {
275
+ "epoch": 0.29532403609515995,
276
+ "eval_loss": 0.3733063042163849,
277
+ "eval_runtime": 19.3481,
278
+ "eval_samples_per_second": 51.685,
279
+ "eval_steps_per_second": 0.827,
280
+ "step": 360
281
+ },
282
+ {
283
+ "epoch": 0.3117309269893355,
284
+ "grad_norm": 0.24217866361141205,
285
+ "learning_rate": 9.906871791552722e-06,
286
+ "loss": 0.3786,
287
+ "step": 380
288
+ },
289
+ {
290
+ "epoch": 0.3117309269893355,
291
+ "eval_loss": 0.36885637044906616,
292
+ "eval_runtime": 19.3455,
293
+ "eval_samples_per_second": 51.692,
294
+ "eval_steps_per_second": 0.827,
295
+ "step": 380
296
+ },
297
+ {
298
+ "epoch": 0.3281378178835111,
299
+ "grad_norm": 0.22478993237018585,
300
+ "learning_rate": 9.877355970422024e-06,
301
+ "loss": 0.3863,
302
+ "step": 400
303
+ },
304
+ {
305
+ "epoch": 0.3281378178835111,
306
+ "eval_loss": 0.3649672567844391,
307
+ "eval_runtime": 19.3226,
308
+ "eval_samples_per_second": 51.753,
309
+ "eval_steps_per_second": 0.828,
310
+ "step": 400
311
+ },
312
+ {
313
+ "epoch": 0.3445447087776866,
314
+ "grad_norm": 0.22086811065673828,
315
+ "learning_rate": 9.843840321793758e-06,
316
+ "loss": 0.3787,
317
+ "step": 420
318
+ },
319
+ {
320
+ "epoch": 0.3445447087776866,
321
+ "eval_loss": 0.361759215593338,
322
+ "eval_runtime": 19.3534,
323
+ "eval_samples_per_second": 51.671,
324
+ "eval_steps_per_second": 0.827,
325
+ "step": 420
326
+ },
327
+ {
328
+ "epoch": 0.3609515996718622,
329
+ "grad_norm": 0.23850491642951965,
330
+ "learning_rate": 9.806352331218253e-06,
331
+ "loss": 0.3757,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.3609515996718622,
336
+ "eval_loss": 0.35869115591049194,
337
+ "eval_runtime": 19.3164,
338
+ "eval_samples_per_second": 51.769,
339
+ "eval_steps_per_second": 0.828,
340
+ "step": 440
341
+ },
342
+ {
343
+ "epoch": 0.37735849056603776,
344
+ "grad_norm": 0.1960180103778839,
345
+ "learning_rate": 9.764922741888425e-06,
346
+ "loss": 0.3745,
347
+ "step": 460
348
+ },
349
+ {
350
+ "epoch": 0.37735849056603776,
351
+ "eval_loss": 0.3557225465774536,
352
+ "eval_runtime": 19.3265,
353
+ "eval_samples_per_second": 51.742,
354
+ "eval_steps_per_second": 0.828,
355
+ "step": 460
356
+ },
357
+ {
358
+ "epoch": 0.3937653814602133,
359
+ "grad_norm": 0.2612729072570801,
360
+ "learning_rate": 9.719585529427876e-06,
361
+ "loss": 0.3649,
362
+ "step": 480
363
+ },
364
+ {
365
+ "epoch": 0.3937653814602133,
366
+ "eval_loss": 0.3532167375087738,
367
+ "eval_runtime": 19.3447,
368
+ "eval_samples_per_second": 51.694,
369
+ "eval_steps_per_second": 0.827,
370
+ "step": 480
371
+ },
372
+ {
373
+ "epoch": 0.41017227235438886,
374
+ "grad_norm": 0.18914461135864258,
375
+ "learning_rate": 9.670377874028118e-06,
376
+ "loss": 0.3584,
377
+ "step": 500
378
+ },
379
+ {
380
+ "epoch": 0.41017227235438886,
381
+ "eval_loss": 0.3509090840816498,
382
+ "eval_runtime": 19.3391,
383
+ "eval_samples_per_second": 51.709,
384
+ "eval_steps_per_second": 0.827,
385
+ "step": 500
386
+ },
387
+ {
388
+ "epoch": 0.4265791632485644,
389
+ "grad_norm": 0.2655193507671356,
390
+ "learning_rate": 9.617340129957807e-06,
391
+ "loss": 0.3663,
392
+ "step": 520
393
+ },
394
+ {
395
+ "epoch": 0.4265791632485644,
396
+ "eval_loss": 0.34918835759162903,
397
+ "eval_runtime": 19.309,
398
+ "eval_samples_per_second": 51.789,
399
+ "eval_steps_per_second": 0.829,
400
+ "step": 520
401
+ },
402
+ {
403
+ "epoch": 0.44298605414273995,
404
+ "grad_norm": 0.321635365486145,
405
+ "learning_rate": 9.560515792468967e-06,
406
+ "loss": 0.3562,
407
+ "step": 540
408
+ },
409
+ {
410
+ "epoch": 0.44298605414273995,
411
+ "eval_loss": 0.34672680497169495,
412
+ "eval_runtime": 19.3063,
413
+ "eval_samples_per_second": 51.797,
414
+ "eval_steps_per_second": 0.829,
415
+ "step": 540
416
+ },
417
+ {
418
+ "epoch": 0.4593929450369155,
419
+ "grad_norm": 0.25474146008491516,
420
+ "learning_rate": 9.499951462127354e-06,
421
+ "loss": 0.3545,
422
+ "step": 560
423
+ },
424
+ {
425
+ "epoch": 0.4593929450369155,
426
+ "eval_loss": 0.34476643800735474,
427
+ "eval_runtime": 19.3046,
428
+ "eval_samples_per_second": 51.801,
429
+ "eval_steps_per_second": 0.829,
430
+ "step": 560
431
+ },
432
+ {
433
+ "epoch": 0.47579983593109104,
434
+ "grad_norm": 0.2607789933681488,
435
+ "learning_rate": 9.435696806596215e-06,
436
+ "loss": 0.3503,
437
+ "step": 580
438
+ },
439
+ {
440
+ "epoch": 0.47579983593109104,
441
+ "eval_loss": 0.34293389320373535,
442
+ "eval_runtime": 19.3434,
443
+ "eval_samples_per_second": 51.697,
444
+ "eval_steps_per_second": 0.827,
445
+ "step": 580
446
+ },
447
+ {
448
+ "epoch": 0.4922067268252666,
449
+ "grad_norm": 0.26284271478652954,
450
+ "learning_rate": 9.367804519904775e-06,
451
+ "loss": 0.3576,
452
+ "step": 600
453
+ },
454
+ {
455
+ "epoch": 0.4922067268252666,
456
+ "eval_loss": 0.3416861295700073,
457
+ "eval_runtime": 19.3158,
458
+ "eval_samples_per_second": 51.771,
459
+ "eval_steps_per_second": 0.828,
460
+ "step": 600
461
+ },
462
+ {
463
+ "epoch": 0.5086136177194421,
464
+ "grad_norm": 0.2197471708059311,
465
+ "learning_rate": 9.296330279234855e-06,
466
+ "loss": 0.3386,
467
+ "step": 620
468
+ },
469
+ {
470
+ "epoch": 0.5086136177194421,
471
+ "eval_loss": 0.3401975929737091,
472
+ "eval_runtime": 19.3391,
473
+ "eval_samples_per_second": 51.709,
474
+ "eval_steps_per_second": 0.827,
475
+ "step": 620
476
+ },
477
+ {
478
+ "epoch": 0.5250205086136177,
479
+ "grad_norm": 0.27967914938926697,
480
+ "learning_rate": 9.221332699261074e-06,
481
+ "loss": 0.3429,
482
+ "step": 640
483
+ },
484
+ {
485
+ "epoch": 0.5250205086136177,
486
+ "eval_loss": 0.338499516248703,
487
+ "eval_runtime": 19.3308,
488
+ "eval_samples_per_second": 51.731,
489
+ "eval_steps_per_second": 0.828,
490
+ "step": 640
491
+ },
492
+ {
493
+ "epoch": 0.5414273995077933,
494
+ "grad_norm": 0.3297106921672821,
495
+ "learning_rate": 9.142873284082055e-06,
496
+ "loss": 0.3436,
497
+ "step": 660
498
+ },
499
+ {
500
+ "epoch": 0.5414273995077933,
501
+ "eval_loss": 0.3370824158191681,
502
+ "eval_runtime": 19.3326,
503
+ "eval_samples_per_second": 51.726,
504
+ "eval_steps_per_second": 0.828,
505
+ "step": 660
506
+ },
507
+ {
508
+ "epoch": 0.5578342904019689,
509
+ "grad_norm": 0.24821262061595917,
510
+ "learning_rate": 9.061016376782078e-06,
511
+ "loss": 0.3403,
512
+ "step": 680
513
+ },
514
+ {
515
+ "epoch": 0.5578342904019689,
516
+ "eval_loss": 0.3358338177204132,
517
+ "eval_runtime": 19.3236,
518
+ "eval_samples_per_second": 51.75,
519
+ "eval_steps_per_second": 0.828,
520
+ "step": 680
521
+ },
522
+ {
523
+ "epoch": 0.5742411812961444,
524
+ "grad_norm": 0.2711612582206726,
525
+ "learning_rate": 8.97582910666454e-06,
526
+ "loss": 0.3385,
527
+ "step": 700
528
+ },
529
+ {
530
+ "epoch": 0.5742411812961444,
531
+ "eval_loss": 0.33477649092674255,
532
+ "eval_runtime": 19.335,
533
+ "eval_samples_per_second": 51.72,
534
+ "eval_steps_per_second": 0.828,
535
+ "step": 700
536
+ },
537
+ {
538
+ "epoch": 0.5906480721903199,
539
+ "grad_norm": 0.27696484327316284,
540
+ "learning_rate": 8.88738133420046e-06,
541
+ "loss": 0.3349,
542
+ "step": 720
543
+ },
544
+ {
545
+ "epoch": 0.5906480721903199,
546
+ "eval_loss": 0.33330047130584717,
547
+ "eval_runtime": 19.3293,
548
+ "eval_samples_per_second": 51.735,
549
+ "eval_steps_per_second": 0.828,
550
+ "step": 720
551
+ },
552
+ {
553
+ "epoch": 0.6070549630844955,
554
+ "grad_norm": 0.24008512496948242,
555
+ "learning_rate": 8.795745593737254e-06,
556
+ "loss": 0.3363,
557
+ "step": 740
558
+ },
559
+ {
560
+ "epoch": 0.6070549630844955,
561
+ "eval_loss": 0.3324872553348541,
562
+ "eval_runtime": 19.326,
563
+ "eval_samples_per_second": 51.744,
564
+ "eval_steps_per_second": 0.828,
565
+ "step": 740
566
+ },
567
+ {
568
+ "epoch": 0.623461853978671,
569
+ "grad_norm": 0.3409559726715088,
570
+ "learning_rate": 8.700997034014665e-06,
571
+ "loss": 0.3424,
572
+ "step": 760
573
+ },
574
+ {
575
+ "epoch": 0.623461853978671,
576
+ "eval_loss": 0.33104297518730164,
577
+ "eval_runtime": 19.3214,
578
+ "eval_samples_per_second": 51.756,
579
+ "eval_steps_per_second": 0.828,
580
+ "step": 760
581
+ },
582
+ {
583
+ "epoch": 0.6398687448728466,
584
+ "grad_norm": 0.2893770933151245,
585
+ "learning_rate": 8.603213356536708e-06,
586
+ "loss": 0.3394,
587
+ "step": 780
588
+ },
589
+ {
590
+ "epoch": 0.6398687448728466,
591
+ "eval_loss": 0.3302854299545288,
592
+ "eval_runtime": 19.3343,
593
+ "eval_samples_per_second": 51.722,
594
+ "eval_steps_per_second": 0.828,
595
+ "step": 780
596
+ },
597
+ {
598
+ "epoch": 0.6562756357670222,
599
+ "grad_norm": 0.26910558342933655,
600
+ "learning_rate": 8.502474751850143e-06,
601
+ "loss": 0.341,
602
+ "step": 800
603
+ },
604
+ {
605
+ "epoch": 0.6562756357670222,
606
+ "eval_loss": 0.3295200765132904,
607
+ "eval_runtime": 19.322,
608
+ "eval_samples_per_second": 51.755,
609
+ "eval_steps_per_second": 0.828,
610
+ "step": 800
611
+ },
612
+ {
613
+ "epoch": 0.6726825266611977,
614
+ "grad_norm": 0.2739856541156769,
615
+ "learning_rate": 8.398863833781707e-06,
616
+ "loss": 0.3329,
617
+ "step": 820
618
+ },
619
+ {
620
+ "epoch": 0.6726825266611977,
621
+ "eval_loss": 0.32813555002212524,
622
+ "eval_runtime": 19.3474,
623
+ "eval_samples_per_second": 51.687,
624
+ "eval_steps_per_second": 0.827,
625
+ "step": 820
626
+ },
627
+ {
628
+ "epoch": 0.6890894175553732,
629
+ "grad_norm": 0.26377996802330017,
630
+ "learning_rate": 8.29246557168809e-06,
631
+ "loss": 0.345,
632
+ "step": 840
633
+ },
634
+ {
635
+ "epoch": 0.6890894175553732,
636
+ "eval_loss": 0.3274153470993042,
637
+ "eval_runtime": 19.3322,
638
+ "eval_samples_per_second": 51.727,
639
+ "eval_steps_per_second": 0.828,
640
+ "step": 840
641
+ },
642
+ {
643
+ "epoch": 0.7054963084495488,
644
+ "grad_norm": 0.2864283323287964,
645
+ "learning_rate": 8.183367220774157e-06,
646
+ "loss": 0.3496,
647
+ "step": 860
648
+ },
649
+ {
650
+ "epoch": 0.7054963084495488,
651
+ "eval_loss": 0.32650744915008545,
652
+ "eval_runtime": 19.329,
653
+ "eval_samples_per_second": 51.736,
654
+ "eval_steps_per_second": 0.828,
655
+ "step": 860
656
+ },
657
+ {
658
+ "epoch": 0.7219031993437244,
659
+ "grad_norm": 0.25862371921539307,
660
+ "learning_rate": 8.071658250536614e-06,
661
+ "loss": 0.3377,
662
+ "step": 880
663
+ },
664
+ {
665
+ "epoch": 0.7219031993437244,
666
+ "eval_loss": 0.3258334994316101,
667
+ "eval_runtime": 19.3357,
668
+ "eval_samples_per_second": 51.718,
669
+ "eval_steps_per_second": 0.827,
670
+ "step": 880
671
+ },
672
+ {
673
+ "epoch": 0.7383100902379,
674
+ "grad_norm": 0.265072226524353,
675
+ "learning_rate": 7.957430271391762e-06,
676
+ "loss": 0.3316,
677
+ "step": 900
678
+ },
679
+ {
680
+ "epoch": 0.7383100902379,
681
+ "eval_loss": 0.3248139023780823,
682
+ "eval_runtime": 19.3321,
683
+ "eval_samples_per_second": 51.727,
684
+ "eval_steps_per_second": 0.828,
685
+ "step": 900
686
+ },
687
+ {
688
+ "epoch": 0.7547169811320755,
689
+ "grad_norm": 0.29285672307014465,
690
+ "learning_rate": 7.840776959547512e-06,
691
+ "loss": 0.3344,
692
+ "step": 920
693
+ },
694
+ {
695
+ "epoch": 0.7547169811320755,
696
+ "eval_loss": 0.32421526312828064,
697
+ "eval_runtime": 19.3164,
698
+ "eval_samples_per_second": 51.769,
699
+ "eval_steps_per_second": 0.828,
700
+ "step": 920
701
+ },
702
+ {
703
+ "epoch": 0.771123872026251,
704
+ "grad_norm": 0.26662516593933105,
705
+ "learning_rate": 7.72179398018132e-06,
706
+ "loss": 0.3137,
707
+ "step": 940
708
+ },
709
+ {
710
+ "epoch": 0.771123872026251,
711
+ "eval_loss": 0.3238050639629364,
712
+ "eval_runtime": 19.3416,
713
+ "eval_samples_per_second": 51.702,
714
+ "eval_steps_per_second": 0.827,
715
+ "step": 940
716
+ },
717
+ {
718
+ "epoch": 0.7875307629204266,
719
+ "grad_norm": 0.29773077368736267,
720
+ "learning_rate": 7.600578908986962e-06,
721
+ "loss": 0.3241,
722
+ "step": 960
723
+ },
724
+ {
725
+ "epoch": 0.7875307629204266,
726
+ "eval_loss": 0.3228878676891327,
727
+ "eval_runtime": 19.3347,
728
+ "eval_samples_per_second": 51.72,
729
+ "eval_steps_per_second": 0.828,
730
+ "step": 960
731
+ },
732
+ {
733
+ "epoch": 0.8039376538146021,
734
+ "grad_norm": 0.3109775483608246,
735
+ "learning_rate": 7.477231152154557e-06,
736
+ "loss": 0.3352,
737
+ "step": 980
738
+ },
739
+ {
740
+ "epoch": 0.8039376538146021,
741
+ "eval_loss": 0.3221863806247711,
742
+ "eval_runtime": 19.3226,
743
+ "eval_samples_per_second": 51.753,
744
+ "eval_steps_per_second": 0.828,
745
+ "step": 980
746
+ },
747
+ {
748
+ "epoch": 0.8203445447087777,
749
+ "grad_norm": 0.3013681471347809,
750
+ "learning_rate": 7.351851864849431e-06,
751
+ "loss": 0.3338,
752
+ "step": 1000
753
+ },
754
+ {
755
+ "epoch": 0.8203445447087777,
756
+ "eval_loss": 0.32160434126853943,
757
+ "eval_runtime": 19.3388,
758
+ "eval_samples_per_second": 51.709,
759
+ "eval_steps_per_second": 0.827,
760
+ "step": 1000
761
+ },
762
+ {
763
+ "epoch": 0.8367514356029533,
764
+ "grad_norm": 0.32970356941223145,
765
+ "learning_rate": 7.224543868256673e-06,
766
+ "loss": 0.3292,
767
+ "step": 1020
768
+ },
769
+ {
770
+ "epoch": 0.8367514356029533,
771
+ "eval_loss": 0.32090088725090027,
772
+ "eval_runtime": 19.3413,
773
+ "eval_samples_per_second": 51.703,
774
+ "eval_steps_per_second": 0.827,
775
+ "step": 1020
776
+ },
777
+ {
778
+ "epoch": 0.8531583264971287,
779
+ "grad_norm": 0.30748239159584045,
780
+ "learning_rate": 7.095411565259434e-06,
781
+ "loss": 0.3236,
782
+ "step": 1040
783
+ },
784
+ {
785
+ "epoch": 0.8531583264971287,
786
+ "eval_loss": 0.3199499547481537,
787
+ "eval_runtime": 19.3101,
788
+ "eval_samples_per_second": 51.786,
789
+ "eval_steps_per_second": 0.829,
790
+ "step": 1040
791
+ },
792
+ {
793
+ "epoch": 0.8695652173913043,
794
+ "grad_norm": 0.3342198133468628,
795
+ "learning_rate": 6.964560854820072e-06,
796
+ "loss": 0.3258,
797
+ "step": 1060
798
+ },
799
+ {
800
+ "epoch": 0.8695652173913043,
801
+ "eval_loss": 0.3197961449623108,
802
+ "eval_runtime": 19.3347,
803
+ "eval_samples_per_second": 51.72,
804
+ "eval_steps_per_second": 0.828,
805
+ "step": 1060
806
+ },
807
+ {
808
+ "epoch": 0.8859721082854799,
809
+ "grad_norm": 0.2627074420452118,
810
+ "learning_rate": 6.832099045134437e-06,
811
+ "loss": 0.3327,
812
+ "step": 1080
813
+ },
814
+ {
815
+ "epoch": 0.8859721082854799,
816
+ "eval_loss": 0.31915849447250366,
817
+ "eval_runtime": 19.3445,
818
+ "eval_samples_per_second": 51.694,
819
+ "eval_steps_per_second": 0.827,
820
+ "step": 1080
821
+ },
822
+ {
823
+ "epoch": 0.9023789991796555,
824
+ "grad_norm": 0.2862927317619324,
825
+ "learning_rate": 6.698134765630434e-06,
826
+ "loss": 0.3227,
827
+ "step": 1100
828
+ },
829
+ {
830
+ "epoch": 0.9023789991796555,
831
+ "eval_loss": 0.3186987042427063,
832
+ "eval_runtime": 19.3343,
833
+ "eval_samples_per_second": 51.722,
834
+ "eval_steps_per_second": 0.828,
835
+ "step": 1100
836
+ },
837
+ {
838
+ "epoch": 0.918785890073831,
839
+ "grad_norm": 0.30319395661354065,
840
+ "learning_rate": 6.562777877883086e-06,
841
+ "loss": 0.323,
842
+ "step": 1120
843
+ },
844
+ {
845
+ "epoch": 0.918785890073831,
846
+ "eval_loss": 0.31824254989624023,
847
+ "eval_runtime": 19.3189,
848
+ "eval_samples_per_second": 51.763,
849
+ "eval_steps_per_second": 0.828,
850
+ "step": 1120
851
+ },
852
+ {
853
+ "epoch": 0.9351927809680065,
854
+ "grad_norm": 0.32216838002204895,
855
+ "learning_rate": 6.4261393855191255e-06,
856
+ "loss": 0.3327,
857
+ "step": 1140
858
+ },
859
+ {
860
+ "epoch": 0.9351927809680065,
861
+ "eval_loss": 0.31814008951187134,
862
+ "eval_runtime": 19.3285,
863
+ "eval_samples_per_second": 51.737,
864
+ "eval_steps_per_second": 0.828,
865
+ "step": 1140
866
+ },
867
+ {
868
+ "epoch": 0.9515996718621821,
869
+ "grad_norm": 0.2896561622619629,
870
+ "learning_rate": 6.288331343185028e-06,
871
+ "loss": 0.3418,
872
+ "step": 1160
873
+ },
874
+ {
875
+ "epoch": 0.9515996718621821,
876
+ "eval_loss": 0.3173450529575348,
877
+ "eval_runtime": 19.3156,
878
+ "eval_samples_per_second": 51.772,
879
+ "eval_steps_per_second": 0.828,
880
+ "step": 1160
881
+ },
882
+ {
883
+ "epoch": 0.9680065627563577,
884
+ "grad_norm": 0.3283473551273346,
885
+ "learning_rate": 6.1494667646531e-06,
886
+ "loss": 0.3216,
887
+ "step": 1180
888
+ },
889
+ {
890
+ "epoch": 0.9680065627563577,
891
+ "eval_loss": 0.3168336749076843,
892
+ "eval_runtime": 19.3336,
893
+ "eval_samples_per_second": 51.723,
894
+ "eval_steps_per_second": 0.828,
895
+ "step": 1180
896
+ },
897
+ {
898
+ "epoch": 0.9844134536505332,
899
+ "grad_norm": 0.26251649856567383,
900
+ "learning_rate": 6.0096595301410324e-06,
901
+ "loss": 0.3296,
902
+ "step": 1200
903
+ },
904
+ {
905
+ "epoch": 0.9844134536505332,
906
+ "eval_loss": 0.31627118587493896,
907
+ "eval_runtime": 19.3399,
908
+ "eval_samples_per_second": 51.707,
909
+ "eval_steps_per_second": 0.827,
910
+ "step": 1200
911
+ },
912
+ {
913
+ "epoch": 1.0008203445447088,
914
+ "grad_norm": 0.2938898503780365,
915
+ "learning_rate": 5.869024292920871e-06,
916
+ "loss": 0.3348,
917
+ "step": 1220
918
+ },
919
+ {
920
+ "epoch": 1.0008203445447088,
921
+ "eval_loss": 0.3160431385040283,
922
+ "eval_runtime": 19.2726,
923
+ "eval_samples_per_second": 51.887,
924
+ "eval_steps_per_second": 0.83,
925
+ "step": 1220
926
+ },
927
+ {
928
+ "epoch": 1.0172272354388843,
929
+ "grad_norm": 0.27582621574401855,
930
+ "learning_rate": 5.727676385294048e-06,
931
+ "loss": 0.3155,
932
+ "step": 1240
933
+ },
934
+ {
935
+ "epoch": 1.0172272354388843,
936
+ "eval_loss": 0.3154791295528412,
937
+ "eval_runtime": 19.2866,
938
+ "eval_samples_per_second": 51.85,
939
+ "eval_steps_per_second": 0.83,
940
+ "step": 1240
941
+ },
942
+ {
943
+ "epoch": 1.03363412633306,
944
+ "grad_norm": 0.3005710542201996,
945
+ "learning_rate": 5.58573172400953e-06,
946
+ "loss": 0.3243,
947
+ "step": 1260
948
+ },
949
+ {
950
+ "epoch": 1.03363412633306,
951
+ "eval_loss": 0.3151017725467682,
952
+ "eval_runtime": 19.3053,
953
+ "eval_samples_per_second": 51.799,
954
+ "eval_steps_per_second": 0.829,
955
+ "step": 1260
956
+ },
957
+ {
958
+ "epoch": 1.0500410172272354,
959
+ "grad_norm": 0.3207043409347534,
960
+ "learning_rate": 5.443306715202686e-06,
961
+ "loss": 0.3228,
962
+ "step": 1280
963
+ },
964
+ {
965
+ "epoch": 1.0500410172272354,
966
+ "eval_loss": 0.3146088719367981,
967
+ "eval_runtime": 19.317,
968
+ "eval_samples_per_second": 51.768,
969
+ "eval_steps_per_second": 0.828,
970
+ "step": 1280
971
+ },
972
+ {
973
+ "epoch": 1.066447908121411,
974
+ "grad_norm": 0.27647730708122253,
975
+ "learning_rate": 5.300518158932815e-06,
976
+ "loss": 0.3156,
977
+ "step": 1300
978
+ },
979
+ {
980
+ "epoch": 1.066447908121411,
981
+ "eval_loss": 0.31434598565101624,
982
+ "eval_runtime": 19.3417,
983
+ "eval_samples_per_second": 51.702,
984
+ "eval_steps_per_second": 0.827,
985
+ "step": 1300
986
+ },
987
+ {
988
+ "epoch": 1.0828547990155866,
989
+ "grad_norm": 0.34843432903289795,
990
+ "learning_rate": 5.157483153397626e-06,
991
+ "loss": 0.3125,
992
+ "step": 1320
993
+ },
994
+ {
995
+ "epoch": 1.0828547990155866,
996
+ "eval_loss": 0.31402087211608887,
997
+ "eval_runtime": 19.3303,
998
+ "eval_samples_per_second": 51.732,
999
+ "eval_steps_per_second": 0.828,
1000
+ "step": 1320
1001
+ },
1002
+ {
1003
+ "epoch": 1.099261689909762,
1004
+ "grad_norm": 0.29794737696647644,
1005
+ "learning_rate": 5.014318998903207e-06,
1006
+ "loss": 0.3234,
1007
+ "step": 1340
1008
+ },
1009
+ {
1010
+ "epoch": 1.099261689909762,
1011
+ "eval_loss": 0.31365808844566345,
1012
+ "eval_runtime": 19.324,
1013
+ "eval_samples_per_second": 51.749,
1014
+ "eval_steps_per_second": 0.828,
1015
+ "step": 1340
1016
+ },
1017
+ {
1018
+ "epoch": 1.1156685808039377,
1019
+ "grad_norm": 0.3145115077495575,
1020
+ "learning_rate": 4.871143101668276e-06,
1021
+ "loss": 0.3254,
1022
+ "step": 1360
1023
+ },
1024
+ {
1025
+ "epoch": 1.1156685808039377,
1026
+ "eval_loss": 0.3132895827293396,
1027
+ "eval_runtime": 19.3427,
1028
+ "eval_samples_per_second": 51.699,
1029
+ "eval_steps_per_second": 0.827,
1030
+ "step": 1360
1031
+ },
1032
+ {
1033
+ "epoch": 1.1320754716981132,
1034
+ "grad_norm": 0.3209640085697174,
1035
+ "learning_rate": 4.728072877541545e-06,
1036
+ "loss": 0.3162,
1037
+ "step": 1380
1038
+ },
1039
+ {
1040
+ "epoch": 1.1320754716981132,
1041
+ "eval_loss": 0.3130640983581543,
1042
+ "eval_runtime": 19.3451,
1043
+ "eval_samples_per_second": 51.693,
1044
+ "eval_steps_per_second": 0.827,
1045
+ "step": 1380
1046
+ },
1047
+ {
1048
+ "epoch": 1.1484823625922886,
1049
+ "grad_norm": 0.2869022488594055,
1050
+ "learning_rate": 4.5852256557112105e-06,
1051
+ "loss": 0.328,
1052
+ "step": 1400
1053
+ },
1054
+ {
1055
+ "epoch": 1.1484823625922886,
1056
+ "eval_loss": 0.3126784563064575,
1057
+ "eval_runtime": 19.3541,
1058
+ "eval_samples_per_second": 51.669,
1059
+ "eval_steps_per_second": 0.827,
1060
+ "step": 1400
1061
+ },
1062
+ {
1063
+ "epoch": 1.1648892534864643,
1064
+ "grad_norm": 0.27980753779411316,
1065
+ "learning_rate": 4.442718582485495e-06,
1066
+ "loss": 0.3176,
1067
+ "step": 1420
1068
+ },
1069
+ {
1070
+ "epoch": 1.1648892534864643,
1071
+ "eval_loss": 0.3125167191028595,
1072
+ "eval_runtime": 19.3258,
1073
+ "eval_samples_per_second": 51.744,
1074
+ "eval_steps_per_second": 0.828,
1075
+ "step": 1420
1076
+ },
1077
+ {
1078
+ "epoch": 1.1812961443806398,
1079
+ "grad_norm": 0.317171186208725,
1080
+ "learning_rate": 4.300668525223178e-06,
1081
+ "loss": 0.3057,
1082
+ "step": 1440
1083
+ },
1084
+ {
1085
+ "epoch": 1.1812961443806398,
1086
+ "eval_loss": 0.31224748492240906,
1087
+ "eval_runtime": 19.3354,
1088
+ "eval_samples_per_second": 51.719,
1089
+ "eval_steps_per_second": 0.827,
1090
+ "step": 1440
1091
+ },
1092
+ {
1093
+ "epoch": 1.1977030352748155,
1094
+ "grad_norm": 0.29117709398269653,
1095
+ "learning_rate": 4.1591919764928805e-06,
1096
+ "loss": 0.3256,
1097
+ "step": 1460
1098
+ },
1099
+ {
1100
+ "epoch": 1.1977030352748155,
1101
+ "eval_loss": 0.3119552433490753,
1102
+ "eval_runtime": 19.3392,
1103
+ "eval_samples_per_second": 51.708,
1104
+ "eval_steps_per_second": 0.827,
1105
+ "step": 1460
1106
+ },
1107
+ {
1108
+ "epoch": 1.214109926168991,
1109
+ "grad_norm": 0.27612075209617615,
1110
+ "learning_rate": 4.018404958539706e-06,
1111
+ "loss": 0.3235,
1112
+ "step": 1480
1113
+ },
1114
+ {
1115
+ "epoch": 1.214109926168991,
1116
+ "eval_loss": 0.3116450905799866,
1117
+ "eval_runtime": 19.3362,
1118
+ "eval_samples_per_second": 51.716,
1119
+ "eval_steps_per_second": 0.827,
1120
+ "step": 1480
1121
+ },
1122
+ {
1123
+ "epoch": 1.2305168170631666,
1124
+ "grad_norm": 0.3429708778858185,
1125
+ "learning_rate": 3.8784229281375964e-06,
1126
+ "loss": 0.3209,
1127
+ "step": 1500
1128
+ },
1129
+ {
1130
+ "epoch": 1.2305168170631666,
1131
+ "eval_loss": 0.3114907145500183,
1132
+ "eval_runtime": 19.3439,
1133
+ "eval_samples_per_second": 51.696,
1134
+ "eval_steps_per_second": 0.827,
1135
+ "step": 1500
1136
+ },
1137
+ {
1138
+ "epoch": 1.246923707957342,
1139
+ "grad_norm": 0.33991965651512146,
1140
+ "learning_rate": 3.7393606819054123e-06,
1141
+ "loss": 0.3265,
1142
+ "step": 1520
1143
+ },
1144
+ {
1145
+ "epoch": 1.246923707957342,
1146
+ "eval_loss": 0.31118032336235046,
1147
+ "eval_runtime": 19.3512,
1148
+ "eval_samples_per_second": 51.676,
1149
+ "eval_steps_per_second": 0.827,
1150
+ "step": 1520
1151
+ },
1152
+ {
1153
+ "epoch": 1.2633305988515175,
1154
+ "grad_norm": 0.2624008059501648,
1155
+ "learning_rate": 3.6013322621643855e-06,
1156
+ "loss": 0.312,
1157
+ "step": 1540
1158
+ },
1159
+ {
1160
+ "epoch": 1.2633305988515175,
1161
+ "eval_loss": 0.31103944778442383,
1162
+ "eval_runtime": 19.319,
1163
+ "eval_samples_per_second": 51.762,
1164
+ "eval_steps_per_second": 0.828,
1165
+ "step": 1540
1166
+ },
1167
+ {
1168
+ "epoch": 1.2797374897456932,
1169
+ "grad_norm": 0.28790125250816345,
1170
+ "learning_rate": 3.4644508634141756e-06,
1171
+ "loss": 0.3219,
1172
+ "step": 1560
1173
+ },
1174
+ {
1175
+ "epoch": 1.2797374897456932,
1176
+ "eval_loss": 0.3108413815498352,
1177
+ "eval_runtime": 19.3152,
1178
+ "eval_samples_per_second": 51.773,
1179
+ "eval_steps_per_second": 0.828,
1180
+ "step": 1560
1181
+ },
1182
+ {
1183
+ "epoch": 1.2961443806398687,
1184
+ "grad_norm": 0.32741355895996094,
1185
+ "learning_rate": 3.32882873950419e-06,
1186
+ "loss": 0.3185,
1187
+ "step": 1580
1188
+ },
1189
+ {
1190
+ "epoch": 1.2961443806398687,
1191
+ "eval_loss": 0.3106067478656769,
1192
+ "eval_runtime": 19.318,
1193
+ "eval_samples_per_second": 51.765,
1194
+ "eval_steps_per_second": 0.828,
1195
+ "step": 1580
1196
+ },
1197
+ {
1198
+ "epoch": 1.3125512715340442,
1199
+ "grad_norm": 0.2802814245223999,
1200
+ "learning_rate": 3.1945771115763335e-06,
1201
+ "loss": 0.3157,
1202
+ "step": 1600
1203
+ },
1204
+ {
1205
+ "epoch": 1.3125512715340442,
1206
+ "eval_loss": 0.31043756008148193,
1207
+ "eval_runtime": 19.3353,
1208
+ "eval_samples_per_second": 51.719,
1209
+ "eval_steps_per_second": 0.828,
1210
+ "step": 1600
1211
+ },
1212
+ {
1213
+ "epoch": 1.3289581624282198,
1214
+ "grad_norm": 0.30865079164505005,
1215
+ "learning_rate": 3.0618060768546445e-06,
1216
+ "loss": 0.3162,
1217
+ "step": 1620
1218
+ },
1219
+ {
1220
+ "epoch": 1.3289581624282198,
1221
+ "eval_loss": 0.31036943197250366,
1222
+ "eval_runtime": 19.3296,
1223
+ "eval_samples_per_second": 51.734,
1224
+ "eval_steps_per_second": 0.828,
1225
+ "step": 1620
1226
+ },
1227
+ {
1228
+ "epoch": 1.3453650533223955,
1229
+ "grad_norm": 0.3288448452949524,
1230
+ "learning_rate": 2.930624518356633e-06,
1231
+ "loss": 0.3182,
1232
+ "step": 1640
1233
+ },
1234
+ {
1235
+ "epoch": 1.3453650533223955,
1236
+ "eval_loss": 0.3100779056549072,
1237
+ "eval_runtime": 19.3281,
1238
+ "eval_samples_per_second": 51.738,
1239
+ "eval_steps_per_second": 0.828,
1240
+ "step": 1640
1241
+ },
1242
+ {
1243
+ "epoch": 1.361771944216571,
1244
+ "grad_norm": 0.3578881621360779,
1245
+ "learning_rate": 2.801140015600392e-06,
1246
+ "loss": 0.3052,
1247
+ "step": 1660
1248
+ },
1249
+ {
1250
+ "epoch": 1.361771944216571,
1251
+ "eval_loss": 0.310004323720932,
1252
+ "eval_runtime": 19.3388,
1253
+ "eval_samples_per_second": 51.709,
1254
+ "eval_steps_per_second": 0.827,
1255
+ "step": 1660
1256
+ },
1257
+ {
1258
+ "epoch": 1.3781788351107465,
1259
+ "grad_norm": 0.2946149706840515,
1260
+ "learning_rate": 2.6734587563806615e-06,
1261
+ "loss": 0.3126,
1262
+ "step": 1680
1263
+ },
1264
+ {
1265
+ "epoch": 1.3781788351107465,
1266
+ "eval_loss": 0.3097366690635681,
1267
+ "eval_runtime": 19.3245,
1268
+ "eval_samples_per_second": 51.748,
1269
+ "eval_steps_per_second": 0.828,
1270
+ "step": 1680
1271
+ },
1272
+ {
1273
+ "epoch": 1.3945857260049221,
1274
+ "grad_norm": 0.29855024814605713,
1275
+ "learning_rate": 2.5476854496862058e-06,
1276
+ "loss": 0.3168,
1277
+ "step": 1700
1278
+ },
1279
+ {
1280
+ "epoch": 1.3945857260049221,
1281
+ "eval_loss": 0.3095654249191284,
1282
+ "eval_runtime": 19.3409,
1283
+ "eval_samples_per_second": 51.704,
1284
+ "eval_steps_per_second": 0.827,
1285
+ "step": 1700
1286
+ },
1287
+ {
1288
+ "epoch": 1.4109926168990976,
1289
+ "grad_norm": 0.3168609142303467,
1290
+ "learning_rate": 2.4239232398299804e-06,
1291
+ "loss": 0.3304,
1292
+ "step": 1720
1293
+ },
1294
+ {
1295
+ "epoch": 1.4109926168990976,
1296
+ "eval_loss": 0.3095927834510803,
1297
+ "eval_runtime": 19.3354,
1298
+ "eval_samples_per_second": 51.719,
1299
+ "eval_steps_per_second": 0.827,
1300
+ "step": 1720
1301
+ },
1302
+ {
1303
+ "epoch": 1.427399507793273,
1304
+ "grad_norm": 0.32232147455215454,
1305
+ "learning_rate": 2.3022736218624048e-06,
1306
+ "loss": 0.3297,
1307
+ "step": 1740
1308
+ },
1309
+ {
1310
+ "epoch": 1.427399507793273,
1311
+ "eval_loss": 0.3092802166938782,
1312
+ "eval_runtime": 19.338,
1313
+ "eval_samples_per_second": 51.712,
1314
+ "eval_steps_per_second": 0.827,
1315
+ "step": 1740
1316
+ },
1317
+ {
1318
+ "epoch": 1.4438063986874488,
1319
+ "grad_norm": 0.2926480770111084,
1320
+ "learning_rate": 2.1828363583371975e-06,
1321
+ "loss": 0.3219,
1322
+ "step": 1760
1323
+ },
1324
+ {
1325
+ "epoch": 1.4438063986874488,
1326
+ "eval_loss": 0.30907222628593445,
1327
+ "eval_runtime": 19.3488,
1328
+ "eval_samples_per_second": 51.683,
1329
+ "eval_steps_per_second": 0.827,
1330
+ "step": 1760
1331
+ },
1332
+ {
1333
+ "epoch": 1.4602132895816242,
1334
+ "grad_norm": 0.3070685565471649,
1335
+ "learning_rate": 2.0657093974979995e-06,
1336
+ "loss": 0.3129,
1337
+ "step": 1780
1338
+ },
1339
+ {
1340
+ "epoch": 1.4602132895816242,
1341
+ "eval_loss": 0.30910417437553406,
1342
+ "eval_runtime": 19.3537,
1343
+ "eval_samples_per_second": 51.67,
1344
+ "eval_steps_per_second": 0.827,
1345
+ "step": 1780
1346
+ },
1347
+ {
1348
+ "epoch": 1.4766201804758,
1349
+ "grad_norm": 0.3446504771709442,
1350
+ "learning_rate": 1.950988792952846e-06,
1351
+ "loss": 0.3255,
1352
+ "step": 1800
1353
+ },
1354
+ {
1355
+ "epoch": 1.4766201804758,
1356
+ "eval_loss": 0.30904003977775574,
1357
+ "eval_runtime": 19.3161,
1358
+ "eval_samples_per_second": 51.77,
1359
+ "eval_steps_per_second": 0.828,
1360
+ "step": 1800
1361
+ },
1362
+ {
1363
+ "epoch": 1.4930270713699754,
1364
+ "grad_norm": 0.3230670094490051,
1365
+ "learning_rate": 1.8387686249024456e-06,
1366
+ "loss": 0.3065,
1367
+ "step": 1820
1368
+ },
1369
+ {
1370
+ "epoch": 1.4930270713699754,
1371
+ "eval_loss": 0.3087674677371979,
1372
+ "eval_runtime": 19.3247,
1373
+ "eval_samples_per_second": 51.747,
1374
+ "eval_steps_per_second": 0.828,
1375
+ "step": 1820
1376
+ },
1377
+ {
1378
+ "epoch": 1.509433962264151,
1379
+ "grad_norm": 0.30200743675231934,
1380
+ "learning_rate": 1.729140922986764e-06,
1381
+ "loss": 0.3267,
1382
+ "step": 1840
1383
+ },
1384
+ {
1385
+ "epoch": 1.509433962264151,
1386
+ "eval_loss": 0.308708131313324,
1387
+ "eval_runtime": 19.3207,
1388
+ "eval_samples_per_second": 51.758,
1389
+ "eval_steps_per_second": 0.828,
1390
+ "step": 1840
1391
+ },
1392
+ {
1393
+ "epoch": 1.5258408531583265,
1394
+ "grad_norm": 0.28980085253715515,
1395
+ "learning_rate": 1.622195590813261e-06,
1396
+ "loss": 0.3194,
1397
+ "step": 1860
1398
+ },
1399
+ {
1400
+ "epoch": 1.5258408531583265,
1401
+ "eval_loss": 0.3086763918399811,
1402
+ "eval_runtime": 19.3342,
1403
+ "eval_samples_per_second": 51.722,
1404
+ "eval_steps_per_second": 0.828,
1405
+ "step": 1860
1406
+ },
1407
+ {
1408
+ "epoch": 1.542247744052502,
1409
+ "grad_norm": 0.34729528427124023,
1410
+ "learning_rate": 1.5180203322286468e-06,
1411
+ "loss": 0.3124,
1412
+ "step": 1880
1413
+ },
1414
+ {
1415
+ "epoch": 1.542247744052502,
1416
+ "eval_loss": 0.3084729015827179,
1417
+ "eval_runtime": 19.338,
1418
+ "eval_samples_per_second": 51.712,
1419
+ "eval_steps_per_second": 0.827,
1420
+ "step": 1880
1421
+ },
1422
+ {
1423
+ "epoch": 1.5586546349466777,
1424
+ "grad_norm": 0.2892967164516449,
1425
+ "learning_rate": 1.4167005793946036e-06,
1426
+ "loss": 0.3162,
1427
+ "step": 1900
1428
+ },
1429
+ {
1430
+ "epoch": 1.5586546349466777,
1431
+ "eval_loss": 0.30849403142929077,
1432
+ "eval_runtime": 19.3372,
1433
+ "eval_samples_per_second": 51.714,
1434
+ "eval_steps_per_second": 0.827,
1435
+ "step": 1900
1436
+ },
1437
+ {
1438
+ "epoch": 1.5750615258408531,
1439
+ "grad_norm": 0.33552056550979614,
1440
+ "learning_rate": 1.3183194227264894e-06,
1441
+ "loss": 0.3203,
1442
+ "step": 1920
1443
+ },
1444
+ {
1445
+ "epoch": 1.5750615258408531,
1446
+ "eval_loss": 0.3084371089935303,
1447
+ "eval_runtime": 19.326,
1448
+ "eval_samples_per_second": 51.744,
1449
+ "eval_steps_per_second": 0.828,
1450
+ "step": 1920
1451
+ },
1452
+ {
1453
+ "epoch": 1.5914684167350286,
1454
+ "grad_norm": 0.29354310035705566,
1455
+ "learning_rate": 1.2229575427524576e-06,
1456
+ "loss": 0.319,
1457
+ "step": 1940
1458
+ },
1459
+ {
1460
+ "epoch": 1.5914684167350286,
1461
+ "eval_loss": 0.30841460824012756,
1462
+ "eval_runtime": 19.3405,
1463
+ "eval_samples_per_second": 51.705,
1464
+ "eval_steps_per_second": 0.827,
1465
+ "step": 1940
1466
+ },
1467
+ {
1468
+ "epoch": 1.6078753076292043,
1469
+ "grad_norm": 0.3153488039970398,
1470
+ "learning_rate": 1.1306931439488777e-06,
1471
+ "loss": 0.3194,
1472
+ "step": 1960
1473
+ },
1474
+ {
1475
+ "epoch": 1.6078753076292043,
1476
+ "eval_loss": 0.3082321882247925,
1477
+ "eval_runtime": 19.3386,
1478
+ "eval_samples_per_second": 51.71,
1479
+ "eval_steps_per_second": 0.827,
1480
+ "step": 1960
1481
+ },
1482
+ {
1483
+ "epoch": 1.62428219852338,
1484
+ "grad_norm": 0.3409827947616577,
1485
+ "learning_rate": 1.0416018906063235e-06,
1486
+ "loss": 0.3189,
1487
+ "step": 1980
1488
+ },
1489
+ {
1490
+ "epoch": 1.62428219852338,
1491
+ "eval_loss": 0.3081977069377899,
1492
+ "eval_runtime": 19.3239,
1493
+ "eval_samples_per_second": 51.749,
1494
+ "eval_steps_per_second": 0.828,
1495
+ "step": 1980
1496
+ },
1497
+ {
1498
+ "epoch": 1.6406890894175554,
1499
+ "grad_norm": 0.3304727375507355,
1500
+ "learning_rate": 9.557568447787201e-07,
1501
+ "loss": 0.3237,
1502
+ "step": 2000
1503
+ },
1504
+ {
1505
+ "epoch": 1.6406890894175554,
1506
+ "eval_loss": 0.30818596482276917,
1507
+ "eval_runtime": 19.322,
1508
+ "eval_samples_per_second": 51.754,
1509
+ "eval_steps_per_second": 0.828,
1510
+ "step": 2000
1511
+ },
1512
+ {
1513
+ "epoch": 1.6570959803117309,
1514
+ "grad_norm": 0.29918399453163147,
1515
+ "learning_rate": 8.732284063665281e-07,
1516
+ "loss": 0.3077,
1517
+ "step": 2020
1518
+ },
1519
+ {
1520
+ "epoch": 1.6570959803117309,
1521
+ "eval_loss": 0.3081939220428467,
1522
+ "eval_runtime": 19.2847,
1523
+ "eval_samples_per_second": 51.854,
1524
+ "eval_steps_per_second": 0.83,
1525
+ "step": 2020
1526
+ },
1527
+ {
1528
+ "epoch": 1.6735028712059066,
1529
+ "grad_norm": 0.3122715651988983,
1530
+ "learning_rate": 7.940842553831168e-07,
1531
+ "loss": 0.3085,
1532
+ "step": 2040
1533
+ },
1534
+ {
1535
+ "epoch": 1.6735028712059066,
1536
+ "eval_loss": 0.3080688416957855,
1537
+ "eval_runtime": 19.3041,
1538
+ "eval_samples_per_second": 51.803,
1539
+ "eval_steps_per_second": 0.829,
1540
+ "step": 2040
1541
+ },
1542
+ {
1543
+ "epoch": 1.689909762100082,
1544
+ "grad_norm": 0.3880915641784668,
1545
+ "learning_rate": 7.183892964516697e-07,
1546
+ "loss": 0.3234,
1547
+ "step": 2060
1548
+ },
1549
+ {
1550
+ "epoch": 1.689909762100082,
1551
+ "eval_loss": 0.308072566986084,
1552
+ "eval_runtime": 19.3261,
1553
+ "eval_samples_per_second": 51.743,
1554
+ "eval_steps_per_second": 0.828,
1555
+ "step": 2060
1556
+ },
1557
+ {
1558
+ "epoch": 1.7063166529942575,
1559
+ "grad_norm": 0.32489344477653503,
1560
+ "learning_rate": 6.462056055781157e-07,
1561
+ "loss": 0.3119,
1562
+ "step": 2080
1563
+ },
1564
+ {
1565
+ "epoch": 1.7063166529942575,
1566
+ "eval_loss": 0.30802538990974426,
1567
+ "eval_runtime": 19.3351,
1568
+ "eval_samples_per_second": 51.719,
1569
+ "eval_steps_per_second": 0.828,
1570
+ "step": 2080
1571
+ },
1572
+ {
1573
+ "epoch": 1.7227235438884332,
1574
+ "grad_norm": 0.3166012167930603,
1575
+ "learning_rate": 5.775923792437865e-07,
1576
+ "loss": 0.3094,
1577
+ "step": 2100
1578
+ },
1579
+ {
1580
+ "epoch": 1.7227235438884332,
1581
+ "eval_loss": 0.30788859724998474,
1582
+ "eval_runtime": 19.3268,
1583
+ "eval_samples_per_second": 51.742,
1584
+ "eval_steps_per_second": 0.828,
1585
+ "step": 2100
1586
+ },
1587
+ {
1588
+ "epoch": 1.7391304347826086,
1589
+ "grad_norm": 0.3155883252620697,
1590
+ "learning_rate": 5.126058858594856e-07,
1591
+ "loss": 0.3094,
1592
+ "step": 2120
1593
+ },
1594
+ {
1595
+ "epoch": 1.7391304347826086,
1596
+ "eval_loss": 0.3079642355442047,
1597
+ "eval_runtime": 19.3377,
1598
+ "eval_samples_per_second": 51.712,
1599
+ "eval_steps_per_second": 0.827,
1600
+ "step": 2120
1601
+ },
1602
+ {
1603
+ "epoch": 1.755537325676784,
1604
+ "grad_norm": 0.31531599164009094,
1605
+ "learning_rate": 4.512994196208359e-07,
1606
+ "loss": 0.307,
1607
+ "step": 2140
1608
+ },
1609
+ {
1610
+ "epoch": 1.755537325676784,
1611
+ "eval_loss": 0.3078817129135132,
1612
+ "eval_runtime": 19.3385,
1613
+ "eval_samples_per_second": 51.71,
1614
+ "eval_steps_per_second": 0.827,
1615
+ "step": 2140
1616
+ },
1617
+ {
1618
+ "epoch": 1.7719442165709598,
1619
+ "grad_norm": 0.3274330496788025,
1620
+ "learning_rate": 3.937232568027166e-07,
1621
+ "loss": 0.3089,
1622
+ "step": 2160
1623
+ },
1624
+ {
1625
+ "epoch": 1.7719442165709598,
1626
+ "eval_loss": 0.30784398317337036,
1627
+ "eval_runtime": 19.3101,
1628
+ "eval_samples_per_second": 51.786,
1629
+ "eval_steps_per_second": 0.829,
1630
+ "step": 2160
1631
+ },
1632
+ {
1633
+ "epoch": 1.7883511074651355,
1634
+ "grad_norm": 0.2670111358165741,
1635
+ "learning_rate": 3.3992461452863056e-07,
1636
+ "loss": 0.3109,
1637
+ "step": 2180
1638
+ },
1639
+ {
1640
+ "epoch": 1.7883511074651355,
1641
+ "eval_loss": 0.30787619948387146,
1642
+ "eval_runtime": 19.3456,
1643
+ "eval_samples_per_second": 51.691,
1644
+ "eval_steps_per_second": 0.827,
1645
+ "step": 2180
1646
+ },
1647
+ {
1648
+ "epoch": 1.804757998359311,
1649
+ "grad_norm": 0.4188634753227234,
1650
+ "learning_rate": 2.899476120488476e-07,
1651
+ "loss": 0.3188,
1652
+ "step": 2200
1653
+ },
1654
+ {
1655
+ "epoch": 1.804757998359311,
1656
+ "eval_loss": 0.30780228972435,
1657
+ "eval_runtime": 19.3361,
1658
+ "eval_samples_per_second": 51.717,
1659
+ "eval_steps_per_second": 0.827,
1660
+ "step": 2200
1661
+ },
1662
+ {
1663
+ "epoch": 1.8211648892534864,
1664
+ "grad_norm": 0.36053863167762756,
1665
+ "learning_rate": 2.4383323455902995e-07,
1666
+ "loss": 0.3052,
1667
+ "step": 2220
1668
+ },
1669
+ {
1670
+ "epoch": 1.8211648892534864,
1671
+ "eval_loss": 0.30786797404289246,
1672
+ "eval_runtime": 19.3329,
1673
+ "eval_samples_per_second": 51.725,
1674
+ "eval_steps_per_second": 0.828,
1675
+ "step": 2220
1676
+ },
1677
+ {
1678
+ "epoch": 1.837571780147662,
1679
+ "grad_norm": 0.3387124240398407,
1680
+ "learning_rate": 2.0161929958905146e-07,
1681
+ "loss": 0.3166,
1682
+ "step": 2240
1683
+ },
1684
+ {
1685
+ "epoch": 1.837571780147662,
1686
+ "eval_loss": 0.30779576301574707,
1687
+ "eval_runtime": 19.3241,
1688
+ "eval_samples_per_second": 51.749,
1689
+ "eval_steps_per_second": 0.828,
1690
+ "step": 2240
1691
+ },
1692
+ {
1693
+ "epoch": 1.8539786710418376,
1694
+ "grad_norm": 0.29246774315834045,
1695
+ "learning_rate": 1.633404259895538e-07,
1696
+ "loss": 0.308,
1697
+ "step": 2260
1698
+ },
1699
+ {
1700
+ "epoch": 1.8539786710418376,
1701
+ "eval_loss": 0.30778998136520386,
1702
+ "eval_runtime": 19.3245,
1703
+ "eval_samples_per_second": 51.748,
1704
+ "eval_steps_per_second": 0.828,
1705
+ "step": 2260
1706
+ },
1707
+ {
1708
+ "epoch": 1.870385561936013,
1709
+ "grad_norm": 0.3102712333202362,
1710
+ "learning_rate": 1.290280055416865e-07,
1711
+ "loss": 0.3155,
1712
+ "step": 2280
1713
+ },
1714
+ {
1715
+ "epoch": 1.870385561936013,
1716
+ "eval_loss": 0.30781853199005127,
1717
+ "eval_runtime": 19.3385,
1718
+ "eval_samples_per_second": 51.71,
1719
+ "eval_steps_per_second": 0.827,
1720
+ "step": 2280
1721
+ },
1722
+ {
1723
+ "epoch": 1.8867924528301887,
1724
+ "grad_norm": 0.3679887354373932,
1725
+ "learning_rate": 9.871017721329201e-08,
1726
+ "loss": 0.3253,
1727
+ "step": 2300
1728
+ },
1729
+ {
1730
+ "epoch": 1.8867924528301887,
1731
+ "eval_loss": 0.3077756464481354,
1732
+ "eval_runtime": 19.322,
1733
+ "eval_samples_per_second": 51.754,
1734
+ "eval_steps_per_second": 0.828,
1735
+ "step": 2300
1736
+ },
1737
+ {
1738
+ "epoch": 1.9031993437243644,
1739
+ "grad_norm": 0.2970399558544159,
1740
+ "learning_rate": 7.241180408267545e-08,
1741
+ "loss": 0.3181,
1742
+ "step": 2320
1743
+ },
1744
+ {
1745
+ "epoch": 1.9031993437243644,
1746
+ "eval_loss": 0.307772696018219,
1747
+ "eval_runtime": 19.3529,
1748
+ "eval_samples_per_second": 51.672,
1749
+ "eval_steps_per_second": 0.827,
1750
+ "step": 2320
1751
+ },
1752
+ {
1753
+ "epoch": 1.9196062346185396,
1754
+ "grad_norm": 0.30068063735961914,
1755
+ "learning_rate": 5.015445294886234e-08,
1756
+ "loss": 0.3124,
1757
+ "step": 2340
1758
+ },
1759
+ {
1760
+ "epoch": 1.9196062346185396,
1761
+ "eval_loss": 0.30776679515838623,
1762
+ "eval_runtime": 19.3314,
1763
+ "eval_samples_per_second": 51.729,
1764
+ "eval_steps_per_second": 0.828,
1765
+ "step": 2340
1766
+ },
1767
+ {
1768
+ "epoch": 1.9360131255127153,
1769
+ "grad_norm": 0.34388095140457153,
1770
+ "learning_rate": 3.195637664507367e-08,
1771
+ "loss": 0.3147,
1772
+ "step": 2360
1773
+ },
1774
+ {
1775
+ "epoch": 1.9360131255127153,
1776
+ "eval_loss": 0.30780360102653503,
1777
+ "eval_runtime": 19.3202,
1778
+ "eval_samples_per_second": 51.759,
1779
+ "eval_steps_per_second": 0.828,
1780
+ "step": 2360
1781
+ },
1782
+ {
1783
+ "epoch": 1.952420016406891,
1784
+ "grad_norm": 0.3172294497489929,
1785
+ "learning_rate": 1.7832499069923014e-08,
1786
+ "loss": 0.329,
1787
+ "step": 2380
1788
+ },
1789
+ {
1790
+ "epoch": 1.952420016406891,
1791
+ "eval_loss": 0.30772775411605835,
1792
+ "eval_runtime": 19.3329,
1793
+ "eval_samples_per_second": 51.725,
1794
+ "eval_steps_per_second": 0.828,
1795
+ "step": 2380
1796
+ },
1797
+ {
1798
+ "epoch": 1.9688269073010665,
1799
+ "grad_norm": 0.3001149892807007,
1800
+ "learning_rate": 7.794402948607672e-09,
1801
+ "loss": 0.3248,
1802
+ "step": 2400
1803
+ },
1804
+ {
1805
+ "epoch": 1.9688269073010665,
1806
+ "eval_loss": 0.30774176120758057,
1807
+ "eval_runtime": 19.3388,
1808
+ "eval_samples_per_second": 51.71,
1809
+ "eval_steps_per_second": 0.827,
1810
+ "step": 2400
1811
+ },
1812
+ {
1813
+ "epoch": 1.985233798195242,
1814
+ "grad_norm": 0.3115482032299042,
1815
+ "learning_rate": 1.8503203341363463e-09,
1816
+ "loss": 0.3069,
1817
+ "step": 2420
1818
+ },
1819
+ {
1820
+ "epoch": 1.985233798195242,
1821
+ "eval_loss": 0.30776721239089966,
1822
+ "eval_runtime": 19.3383,
1823
+ "eval_samples_per_second": 51.711,
1824
+ "eval_steps_per_second": 0.827,
1825
+ "step": 2420
1826
+ },
1827
+ {
1828
+ "epoch": 2.0,
1829
+ "step": 2438,
1830
+ "total_flos": 1.58523627405312e+18,
1831
+ "train_loss": 0.36021078112088234,
1832
+ "train_runtime": 4858.8972,
1833
+ "train_samples_per_second": 16.053,
1834
+ "train_steps_per_second": 0.502
1835
+ }
1836
+ ],
1837
+ "logging_steps": 20,
1838
+ "max_steps": 2438,
1839
+ "num_input_tokens_seen": 0,
1840
+ "num_train_epochs": 2,
1841
+ "save_steps": 500,
1842
+ "stateful_callbacks": {
1843
+ "TrainerControl": {
1844
+ "args": {
1845
+ "should_epoch_stop": false,
1846
+ "should_evaluate": false,
1847
+ "should_log": false,
1848
+ "should_save": true,
1849
+ "should_training_stop": true
1850
+ },
1851
+ "attributes": {}
1852
+ }
1853
+ },
1854
+ "total_flos": 1.58523627405312e+18,
1855
+ "train_batch_size": 32,
1856
+ "trial_name": null,
1857
+ "trial_params": null
1858
+ }
nl_tasks/run_all/run_exps2/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 4,
9
+ "peft_type": "ROTATION",
10
+ "r": 4,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exps2/ft/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "</s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "</s>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
nl_tasks/run_all/run_exps2/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exps2/ft/tokenizer_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ },
30
+ "32000": {
31
+ "content": "[PAD]",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": true
37
+ }
38
+ },
39
+ "bos_token": "</s>",
40
+ "clean_up_tokenization_spaces": false,
41
+ "eos_token": "</s>",
42
+ "extra_special_tokens": {},
43
+ "legacy": false,
44
+ "model_max_length": 512,
45
+ "pad_token": "[PAD]",
46
+ "padding_side": "right",
47
+ "sp_model_kwargs": {},
48
+ "tokenizer_class": "LlamaTokenizer",
49
+ "unk_token": "</s>",
50
+ "use_default_system_prompt": false
51
+ }
nl_tasks/run_all/run_exps2/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a700b89db6a2e6d387af73dfd853aab66bde44288aa5d8141708f03f9d466f4
3
+ size 6481
nl_tasks/run_all/run_exps2/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 4,
9
+ "peft_type": "ROTATION",
10
+ "r": 4,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exps2/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3384d5f1702c1e7eef4f367342ea9424cb6356dfa35784810e7e8db0eda1aad
3
+ size 33602659
nl_tasks/run_all/run_exps2/trainer_state.json ADDED
@@ -0,0 +1,1858 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.0,
6
+ "eval_steps": 20,
7
+ "global_step": 2438,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.016406890894175553,
14
+ "grad_norm": 0.05100574344396591,
15
+ "learning_rate": 1.557377049180328e-06,
16
+ "loss": 0.7463,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.016406890894175553,
21
+ "eval_loss": 0.7439825534820557,
22
+ "eval_runtime": 19.7893,
23
+ "eval_samples_per_second": 50.532,
24
+ "eval_steps_per_second": 0.809,
25
+ "step": 20
26
+ },
27
+ {
28
+ "epoch": 0.03281378178835111,
29
+ "grad_norm": 0.08355569839477539,
30
+ "learning_rate": 3.1967213114754105e-06,
31
+ "loss": 0.758,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.03281378178835111,
36
+ "eval_loss": 0.740388035774231,
37
+ "eval_runtime": 19.272,
38
+ "eval_samples_per_second": 51.889,
39
+ "eval_steps_per_second": 0.83,
40
+ "step": 40
41
+ },
42
+ {
43
+ "epoch": 0.04922067268252666,
44
+ "grad_norm": 0.15142929553985596,
45
+ "learning_rate": 4.836065573770492e-06,
46
+ "loss": 0.7412,
47
+ "step": 60
48
+ },
49
+ {
50
+ "epoch": 0.04922067268252666,
51
+ "eval_loss": 0.7260333299636841,
52
+ "eval_runtime": 19.3339,
53
+ "eval_samples_per_second": 51.723,
54
+ "eval_steps_per_second": 0.828,
55
+ "step": 60
56
+ },
57
+ {
58
+ "epoch": 0.06562756357670221,
59
+ "grad_norm": 0.2455596625804901,
60
+ "learning_rate": 6.475409836065575e-06,
61
+ "loss": 0.713,
62
+ "step": 80
63
+ },
64
+ {
65
+ "epoch": 0.06562756357670221,
66
+ "eval_loss": 0.6844377517700195,
67
+ "eval_runtime": 19.3182,
68
+ "eval_samples_per_second": 51.765,
69
+ "eval_steps_per_second": 0.828,
70
+ "step": 80
71
+ },
72
+ {
73
+ "epoch": 0.08203445447087777,
74
+ "grad_norm": 0.29648932814598083,
75
+ "learning_rate": 8.114754098360657e-06,
76
+ "loss": 0.6573,
77
+ "step": 100
78
+ },
79
+ {
80
+ "epoch": 0.08203445447087777,
81
+ "eval_loss": 0.5912646651268005,
82
+ "eval_runtime": 19.3104,
83
+ "eval_samples_per_second": 51.786,
84
+ "eval_steps_per_second": 0.829,
85
+ "step": 100
86
+ },
87
+ {
88
+ "epoch": 0.09844134536505332,
89
+ "grad_norm": 0.2266065627336502,
90
+ "learning_rate": 9.754098360655738e-06,
91
+ "loss": 0.5493,
92
+ "step": 120
93
+ },
94
+ {
95
+ "epoch": 0.09844134536505332,
96
+ "eval_loss": 0.5015877485275269,
97
+ "eval_runtime": 19.3245,
98
+ "eval_samples_per_second": 51.748,
99
+ "eval_steps_per_second": 0.828,
100
+ "step": 120
101
+ },
102
+ {
103
+ "epoch": 0.11484823625922888,
104
+ "grad_norm": 0.18106642365455627,
105
+ "learning_rate": 1.1393442622950821e-05,
106
+ "loss": 0.4788,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.11484823625922888,
111
+ "eval_loss": 0.4532758295536041,
112
+ "eval_runtime": 19.327,
113
+ "eval_samples_per_second": 51.741,
114
+ "eval_steps_per_second": 0.828,
115
+ "step": 140
116
+ },
117
+ {
118
+ "epoch": 0.13125512715340443,
119
+ "grad_norm": 0.1644924432039261,
120
+ "learning_rate": 1.3032786885245904e-05,
121
+ "loss": 0.4416,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.13125512715340443,
126
+ "eval_loss": 0.43009692430496216,
127
+ "eval_runtime": 19.3273,
128
+ "eval_samples_per_second": 51.74,
129
+ "eval_steps_per_second": 0.828,
130
+ "step": 160
131
+ },
132
+ {
133
+ "epoch": 0.14766201804757997,
134
+ "grad_norm": 0.16431841254234314,
135
+ "learning_rate": 1.4672131147540984e-05,
136
+ "loss": 0.4309,
137
+ "step": 180
138
+ },
139
+ {
140
+ "epoch": 0.14766201804757997,
141
+ "eval_loss": 0.41236767172813416,
142
+ "eval_runtime": 19.3211,
143
+ "eval_samples_per_second": 51.757,
144
+ "eval_steps_per_second": 0.828,
145
+ "step": 180
146
+ },
147
+ {
148
+ "epoch": 0.16406890894175555,
149
+ "grad_norm": 0.1948602944612503,
150
+ "learning_rate": 1.6311475409836068e-05,
151
+ "loss": 0.4149,
152
+ "step": 200
153
+ },
154
+ {
155
+ "epoch": 0.16406890894175555,
156
+ "eval_loss": 0.39788001775741577,
157
+ "eval_runtime": 19.3156,
158
+ "eval_samples_per_second": 51.772,
159
+ "eval_steps_per_second": 0.828,
160
+ "step": 200
161
+ },
162
+ {
163
+ "epoch": 0.1804757998359311,
164
+ "grad_norm": 0.2017097920179367,
165
+ "learning_rate": 1.795081967213115e-05,
166
+ "loss": 0.4121,
167
+ "step": 220
168
+ },
169
+ {
170
+ "epoch": 0.1804757998359311,
171
+ "eval_loss": 0.38638782501220703,
172
+ "eval_runtime": 19.3233,
173
+ "eval_samples_per_second": 51.751,
174
+ "eval_steps_per_second": 0.828,
175
+ "step": 220
176
+ },
177
+ {
178
+ "epoch": 0.19688269073010664,
179
+ "grad_norm": 0.20454800128936768,
180
+ "learning_rate": 1.9590163934426232e-05,
181
+ "loss": 0.3946,
182
+ "step": 240
183
+ },
184
+ {
185
+ "epoch": 0.19688269073010664,
186
+ "eval_loss": 0.376147985458374,
187
+ "eval_runtime": 19.3187,
188
+ "eval_samples_per_second": 51.763,
189
+ "eval_steps_per_second": 0.828,
190
+ "step": 240
191
+ },
192
+ {
193
+ "epoch": 0.2132895816242822,
194
+ "grad_norm": 0.2339833378791809,
195
+ "learning_rate": 1.999769345291104e-05,
196
+ "loss": 0.3825,
197
+ "step": 260
198
+ },
199
+ {
200
+ "epoch": 0.2132895816242822,
201
+ "eval_loss": 0.3673856854438782,
202
+ "eval_runtime": 19.3273,
203
+ "eval_samples_per_second": 51.74,
204
+ "eval_steps_per_second": 0.828,
205
+ "step": 260
206
+ },
207
+ {
208
+ "epoch": 0.22969647251845776,
209
+ "grad_norm": 0.19959838688373566,
210
+ "learning_rate": 1.9987444278046718e-05,
211
+ "loss": 0.3799,
212
+ "step": 280
213
+ },
214
+ {
215
+ "epoch": 0.22969647251845776,
216
+ "eval_loss": 0.3609837293624878,
217
+ "eval_runtime": 19.3276,
218
+ "eval_samples_per_second": 51.739,
219
+ "eval_steps_per_second": 0.828,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2461033634126333,
224
+ "grad_norm": 0.2362460196018219,
225
+ "learning_rate": 1.9969004588793865e-05,
226
+ "loss": 0.3644,
227
+ "step": 300
228
+ },
229
+ {
230
+ "epoch": 0.2461033634126333,
231
+ "eval_loss": 0.35560137033462524,
232
+ "eval_runtime": 19.3284,
233
+ "eval_samples_per_second": 51.737,
234
+ "eval_steps_per_second": 0.828,
235
+ "step": 300
236
+ },
237
+ {
238
+ "epoch": 0.26251025430680885,
239
+ "grad_norm": 0.21437275409698486,
240
+ "learning_rate": 1.9942389507193297e-05,
241
+ "loss": 0.3543,
242
+ "step": 320
243
+ },
244
+ {
245
+ "epoch": 0.26251025430680885,
246
+ "eval_loss": 0.35107719898223877,
247
+ "eval_runtime": 19.3428,
248
+ "eval_samples_per_second": 51.699,
249
+ "eval_steps_per_second": 0.827,
250
+ "step": 320
251
+ },
252
+ {
253
+ "epoch": 0.27891714520098443,
254
+ "grad_norm": 0.2086835503578186,
255
+ "learning_rate": 1.990762085977069e-05,
256
+ "loss": 0.3661,
257
+ "step": 340
258
+ },
259
+ {
260
+ "epoch": 0.27891714520098443,
261
+ "eval_loss": 0.3464248478412628,
262
+ "eval_runtime": 19.327,
263
+ "eval_samples_per_second": 51.741,
264
+ "eval_steps_per_second": 0.828,
265
+ "step": 340
266
+ },
267
+ {
268
+ "epoch": 0.29532403609515995,
269
+ "grad_norm": 0.2647511959075928,
270
+ "learning_rate": 1.9864727159637e-05,
271
+ "loss": 0.3441,
272
+ "step": 360
273
+ },
274
+ {
275
+ "epoch": 0.29532403609515995,
276
+ "eval_loss": 0.3429518938064575,
277
+ "eval_runtime": 19.3341,
278
+ "eval_samples_per_second": 51.722,
279
+ "eval_steps_per_second": 0.828,
280
+ "step": 360
281
+ },
282
+ {
283
+ "epoch": 0.3117309269893355,
284
+ "grad_norm": 0.2909955382347107,
285
+ "learning_rate": 1.9813743583105444e-05,
286
+ "loss": 0.349,
287
+ "step": 380
288
+ },
289
+ {
290
+ "epoch": 0.3117309269893355,
291
+ "eval_loss": 0.3390914797782898,
292
+ "eval_runtime": 19.3218,
293
+ "eval_samples_per_second": 51.755,
294
+ "eval_steps_per_second": 0.828,
295
+ "step": 380
296
+ },
297
+ {
298
+ "epoch": 0.3281378178835111,
299
+ "grad_norm": 0.3091684579849243,
300
+ "learning_rate": 1.9754711940844048e-05,
301
+ "loss": 0.358,
302
+ "step": 400
303
+ },
304
+ {
305
+ "epoch": 0.3281378178835111,
306
+ "eval_loss": 0.3364439010620117,
307
+ "eval_runtime": 19.2984,
308
+ "eval_samples_per_second": 51.818,
309
+ "eval_steps_per_second": 0.829,
310
+ "step": 400
311
+ },
312
+ {
313
+ "epoch": 0.3445447087776866,
314
+ "grad_norm": 0.2638132870197296,
315
+ "learning_rate": 1.9687680643587515e-05,
316
+ "loss": 0.3504,
317
+ "step": 420
318
+ },
319
+ {
320
+ "epoch": 0.3445447087776866,
321
+ "eval_loss": 0.33499646186828613,
322
+ "eval_runtime": 19.3252,
323
+ "eval_samples_per_second": 51.746,
324
+ "eval_steps_per_second": 0.828,
325
+ "step": 420
326
+ },
327
+ {
328
+ "epoch": 0.3609515996718622,
329
+ "grad_norm": 0.2888047397136688,
330
+ "learning_rate": 1.9612704662436506e-05,
331
+ "loss": 0.3491,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.3609515996718622,
336
+ "eval_loss": 0.33223557472229004,
337
+ "eval_runtime": 19.3423,
338
+ "eval_samples_per_second": 51.7,
339
+ "eval_steps_per_second": 0.827,
340
+ "step": 440
341
+ },
342
+ {
343
+ "epoch": 0.37735849056603776,
344
+ "grad_norm": 0.20659048855304718,
345
+ "learning_rate": 1.952984548377685e-05,
346
+ "loss": 0.3492,
347
+ "step": 460
348
+ },
349
+ {
350
+ "epoch": 0.37735849056603776,
351
+ "eval_loss": 0.33008918166160583,
352
+ "eval_runtime": 19.3021,
353
+ "eval_samples_per_second": 51.808,
354
+ "eval_steps_per_second": 0.829,
355
+ "step": 460
356
+ },
357
+ {
358
+ "epoch": 0.3937653814602133,
359
+ "grad_norm": 0.2820521295070648,
360
+ "learning_rate": 1.943917105885575e-05,
361
+ "loss": 0.339,
362
+ "step": 480
363
+ },
364
+ {
365
+ "epoch": 0.3937653814602133,
366
+ "eval_loss": 0.3275497257709503,
367
+ "eval_runtime": 19.3336,
368
+ "eval_samples_per_second": 51.723,
369
+ "eval_steps_per_second": 0.828,
370
+ "step": 480
371
+ },
372
+ {
373
+ "epoch": 0.41017227235438886,
374
+ "grad_norm": 0.24061936140060425,
375
+ "learning_rate": 1.9340755748056236e-05,
376
+ "loss": 0.3337,
377
+ "step": 500
378
+ },
379
+ {
380
+ "epoch": 0.41017227235438886,
381
+ "eval_loss": 0.3265324831008911,
382
+ "eval_runtime": 19.3321,
383
+ "eval_samples_per_second": 51.728,
384
+ "eval_steps_per_second": 0.828,
385
+ "step": 500
386
+ },
387
+ {
388
+ "epoch": 0.4265791632485644,
389
+ "grad_norm": 0.3097938299179077,
390
+ "learning_rate": 1.9234680259915613e-05,
391
+ "loss": 0.3414,
392
+ "step": 520
393
+ },
394
+ {
395
+ "epoch": 0.4265791632485644,
396
+ "eval_loss": 0.3249492049217224,
397
+ "eval_runtime": 19.3003,
398
+ "eval_samples_per_second": 51.813,
399
+ "eval_steps_per_second": 0.829,
400
+ "step": 520
401
+ },
402
+ {
403
+ "epoch": 0.44298605414273995,
404
+ "grad_norm": 0.3457156717777252,
405
+ "learning_rate": 1.9121031584937935e-05,
406
+ "loss": 0.3324,
407
+ "step": 540
408
+ },
409
+ {
410
+ "epoch": 0.44298605414273995,
411
+ "eval_loss": 0.323433518409729,
412
+ "eval_runtime": 19.3078,
413
+ "eval_samples_per_second": 51.792,
414
+ "eval_steps_per_second": 0.829,
415
+ "step": 540
416
+ },
417
+ {
418
+ "epoch": 0.4593929450369155,
419
+ "grad_norm": 0.2805982232093811,
420
+ "learning_rate": 1.8999902924254708e-05,
421
+ "loss": 0.3307,
422
+ "step": 560
423
+ },
424
+ {
425
+ "epoch": 0.4593929450369155,
426
+ "eval_loss": 0.3207704722881317,
427
+ "eval_runtime": 19.317,
428
+ "eval_samples_per_second": 51.768,
429
+ "eval_steps_per_second": 0.828,
430
+ "step": 560
431
+ },
432
+ {
433
+ "epoch": 0.47579983593109104,
434
+ "grad_norm": 0.28764742612838745,
435
+ "learning_rate": 1.887139361319243e-05,
436
+ "loss": 0.3264,
437
+ "step": 580
438
+ },
439
+ {
440
+ "epoch": 0.47579983593109104,
441
+ "eval_loss": 0.31976696848869324,
442
+ "eval_runtime": 19.3171,
443
+ "eval_samples_per_second": 51.768,
444
+ "eval_steps_per_second": 0.828,
445
+ "step": 580
446
+ },
447
+ {
448
+ "epoch": 0.4922067268252666,
449
+ "grad_norm": 0.29856017231941223,
450
+ "learning_rate": 1.873560903980955e-05,
451
+ "loss": 0.333,
452
+ "step": 600
453
+ },
454
+ {
455
+ "epoch": 0.4922067268252666,
456
+ "eval_loss": 0.3188488483428955,
457
+ "eval_runtime": 19.3294,
458
+ "eval_samples_per_second": 51.735,
459
+ "eval_steps_per_second": 0.828,
460
+ "step": 600
461
+ },
462
+ {
463
+ "epoch": 0.5086136177194421,
464
+ "grad_norm": 0.2475677728652954,
465
+ "learning_rate": 1.859266055846971e-05,
466
+ "loss": 0.3162,
467
+ "step": 620
468
+ },
469
+ {
470
+ "epoch": 0.5086136177194421,
471
+ "eval_loss": 0.3177519142627716,
472
+ "eval_runtime": 19.3138,
473
+ "eval_samples_per_second": 51.777,
474
+ "eval_steps_per_second": 0.828,
475
+ "step": 620
476
+ },
477
+ {
478
+ "epoch": 0.5250205086136177,
479
+ "grad_norm": 0.31477370858192444,
480
+ "learning_rate": 1.8442665398522147e-05,
481
+ "loss": 0.3206,
482
+ "step": 640
483
+ },
484
+ {
485
+ "epoch": 0.5250205086136177,
486
+ "eval_loss": 0.31575125455856323,
487
+ "eval_runtime": 19.3257,
488
+ "eval_samples_per_second": 51.745,
489
+ "eval_steps_per_second": 0.828,
490
+ "step": 640
491
+ },
492
+ {
493
+ "epoch": 0.5414273995077933,
494
+ "grad_norm": 0.3218989074230194,
495
+ "learning_rate": 1.828574656816411e-05,
496
+ "loss": 0.3212,
497
+ "step": 660
498
+ },
499
+ {
500
+ "epoch": 0.5414273995077933,
501
+ "eval_loss": 0.3145419657230377,
502
+ "eval_runtime": 19.3204,
503
+ "eval_samples_per_second": 51.759,
504
+ "eval_steps_per_second": 0.828,
505
+ "step": 660
506
+ },
507
+ {
508
+ "epoch": 0.5578342904019689,
509
+ "grad_norm": 0.2936188280582428,
510
+ "learning_rate": 1.8122032753564157e-05,
511
+ "loss": 0.3181,
512
+ "step": 680
513
+ },
514
+ {
515
+ "epoch": 0.5578342904019689,
516
+ "eval_loss": 0.3138105273246765,
517
+ "eval_runtime": 19.317,
518
+ "eval_samples_per_second": 51.768,
519
+ "eval_steps_per_second": 0.828,
520
+ "step": 680
521
+ },
522
+ {
523
+ "epoch": 0.5742411812961444,
524
+ "grad_norm": 0.3014962077140808,
525
+ "learning_rate": 1.795165821332908e-05,
526
+ "loss": 0.3159,
527
+ "step": 700
528
+ },
529
+ {
530
+ "epoch": 0.5742411812961444,
531
+ "eval_loss": 0.31264302134513855,
532
+ "eval_runtime": 19.3247,
533
+ "eval_samples_per_second": 51.747,
534
+ "eval_steps_per_second": 0.828,
535
+ "step": 700
536
+ },
537
+ {
538
+ "epoch": 0.5906480721903199,
539
+ "grad_norm": 0.31352338194847107,
540
+ "learning_rate": 1.777476266840092e-05,
541
+ "loss": 0.3124,
542
+ "step": 720
543
+ },
544
+ {
545
+ "epoch": 0.5906480721903199,
546
+ "eval_loss": 0.31203383207321167,
547
+ "eval_runtime": 19.321,
548
+ "eval_samples_per_second": 51.757,
549
+ "eval_steps_per_second": 0.828,
550
+ "step": 720
551
+ },
552
+ {
553
+ "epoch": 0.6070549630844955,
554
+ "grad_norm": 0.27813902497291565,
555
+ "learning_rate": 1.7591491187474508e-05,
556
+ "loss": 0.3151,
557
+ "step": 740
558
+ },
559
+ {
560
+ "epoch": 0.6070549630844955,
561
+ "eval_loss": 0.31110283732414246,
562
+ "eval_runtime": 19.307,
563
+ "eval_samples_per_second": 51.795,
564
+ "eval_steps_per_second": 0.829,
565
+ "step": 740
566
+ },
567
+ {
568
+ "epoch": 0.623461853978671,
569
+ "grad_norm": 0.3086440861225128,
570
+ "learning_rate": 1.740199406802933e-05,
571
+ "loss": 0.3222,
572
+ "step": 760
573
+ },
574
+ {
575
+ "epoch": 0.623461853978671,
576
+ "eval_loss": 0.3100873827934265,
577
+ "eval_runtime": 19.3301,
578
+ "eval_samples_per_second": 51.733,
579
+ "eval_steps_per_second": 0.828,
580
+ "step": 760
581
+ },
582
+ {
583
+ "epoch": 0.6398687448728466,
584
+ "grad_norm": 0.32539573311805725,
585
+ "learning_rate": 1.7206426713073416e-05,
586
+ "loss": 0.3194,
587
+ "step": 780
588
+ },
589
+ {
590
+ "epoch": 0.6398687448728466,
591
+ "eval_loss": 0.3095526397228241,
592
+ "eval_runtime": 19.3423,
593
+ "eval_samples_per_second": 51.7,
594
+ "eval_steps_per_second": 0.827,
595
+ "step": 780
596
+ },
597
+ {
598
+ "epoch": 0.6562756357670222,
599
+ "grad_norm": 0.31149205565452576,
600
+ "learning_rate": 1.7004949503700286e-05,
601
+ "loss": 0.3198,
602
+ "step": 800
603
+ },
604
+ {
605
+ "epoch": 0.6562756357670222,
606
+ "eval_loss": 0.30877068638801575,
607
+ "eval_runtime": 19.3218,
608
+ "eval_samples_per_second": 51.755,
609
+ "eval_steps_per_second": 0.828,
610
+ "step": 800
611
+ },
612
+ {
613
+ "epoch": 0.6726825266611977,
614
+ "grad_norm": 0.29149356484413147,
615
+ "learning_rate": 1.6797727667563414e-05,
616
+ "loss": 0.3108,
617
+ "step": 820
618
+ },
619
+ {
620
+ "epoch": 0.6726825266611977,
621
+ "eval_loss": 0.30757883191108704,
622
+ "eval_runtime": 19.3266,
623
+ "eval_samples_per_second": 51.742,
624
+ "eval_steps_per_second": 0.828,
625
+ "step": 820
626
+ },
627
+ {
628
+ "epoch": 0.6890894175553732,
629
+ "grad_norm": 0.2758270800113678,
630
+ "learning_rate": 1.658493114337618e-05,
631
+ "loss": 0.3236,
632
+ "step": 840
633
+ },
634
+ {
635
+ "epoch": 0.6890894175553732,
636
+ "eval_loss": 0.3069859743118286,
637
+ "eval_runtime": 19.3214,
638
+ "eval_samples_per_second": 51.756,
639
+ "eval_steps_per_second": 0.828,
640
+ "step": 840
641
+ },
642
+ {
643
+ "epoch": 0.7054963084495488,
644
+ "grad_norm": 0.30308738350868225,
645
+ "learning_rate": 1.6366734441548314e-05,
646
+ "loss": 0.3286,
647
+ "step": 860
648
+ },
649
+ {
650
+ "epoch": 0.7054963084495488,
651
+ "eval_loss": 0.30621635913848877,
652
+ "eval_runtime": 19.3452,
653
+ "eval_samples_per_second": 51.692,
654
+ "eval_steps_per_second": 0.827,
655
+ "step": 860
656
+ },
657
+ {
658
+ "epoch": 0.7219031993437244,
659
+ "grad_norm": 0.27197226881980896,
660
+ "learning_rate": 1.6143316501073228e-05,
661
+ "loss": 0.318,
662
+ "step": 880
663
+ },
664
+ {
665
+ "epoch": 0.7219031993437244,
666
+ "eval_loss": 0.3056408762931824,
667
+ "eval_runtime": 19.3268,
668
+ "eval_samples_per_second": 51.742,
669
+ "eval_steps_per_second": 0.828,
670
+ "step": 880
671
+ },
672
+ {
673
+ "epoch": 0.7383100902379,
674
+ "grad_norm": 0.3094601035118103,
675
+ "learning_rate": 1.5914860542783523e-05,
676
+ "loss": 0.3109,
677
+ "step": 900
678
+ },
679
+ {
680
+ "epoch": 0.7383100902379,
681
+ "eval_loss": 0.30550602078437805,
682
+ "eval_runtime": 19.3088,
683
+ "eval_samples_per_second": 51.79,
684
+ "eval_steps_per_second": 0.829,
685
+ "step": 900
686
+ },
687
+ {
688
+ "epoch": 0.7547169811320755,
689
+ "grad_norm": 0.29856327176094055,
690
+ "learning_rate": 1.5681553919095025e-05,
691
+ "loss": 0.3149,
692
+ "step": 920
693
+ },
694
+ {
695
+ "epoch": 0.7547169811320755,
696
+ "eval_loss": 0.30412110686302185,
697
+ "eval_runtime": 19.3166,
698
+ "eval_samples_per_second": 51.769,
699
+ "eval_steps_per_second": 0.828,
700
+ "step": 920
701
+ },
702
+ {
703
+ "epoch": 0.771123872026251,
704
+ "grad_norm": 0.3012203276157379,
705
+ "learning_rate": 1.544358796036264e-05,
706
+ "loss": 0.295,
707
+ "step": 940
708
+ },
709
+ {
710
+ "epoch": 0.771123872026251,
711
+ "eval_loss": 0.30357825756073,
712
+ "eval_runtime": 19.339,
713
+ "eval_samples_per_second": 51.709,
714
+ "eval_steps_per_second": 0.827,
715
+ "step": 940
716
+ },
717
+ {
718
+ "epoch": 0.7875307629204266,
719
+ "grad_norm": 0.32613876461982727,
720
+ "learning_rate": 1.5201157817973924e-05,
721
+ "loss": 0.3049,
722
+ "step": 960
723
+ },
724
+ {
725
+ "epoch": 0.7875307629204266,
726
+ "eval_loss": 0.30287426710128784,
727
+ "eval_runtime": 19.3353,
728
+ "eval_samples_per_second": 51.719,
729
+ "eval_steps_per_second": 0.828,
730
+ "step": 960
731
+ },
732
+ {
733
+ "epoch": 0.8039376538146021,
734
+ "grad_norm": 0.30854496359825134,
735
+ "learning_rate": 1.4954462304309114e-05,
736
+ "loss": 0.3154,
737
+ "step": 980
738
+ },
739
+ {
740
+ "epoch": 0.8039376538146021,
741
+ "eval_loss": 0.30233636498451233,
742
+ "eval_runtime": 19.3322,
743
+ "eval_samples_per_second": 51.727,
744
+ "eval_steps_per_second": 0.828,
745
+ "step": 980
746
+ },
747
+ {
748
+ "epoch": 0.8203445447087777,
749
+ "grad_norm": 0.3212120532989502,
750
+ "learning_rate": 1.4703703729698861e-05,
751
+ "loss": 0.314,
752
+ "step": 1000
753
+ },
754
+ {
755
+ "epoch": 0.8203445447087777,
756
+ "eval_loss": 0.3019111454486847,
757
+ "eval_runtime": 19.3219,
758
+ "eval_samples_per_second": 51.755,
759
+ "eval_steps_per_second": 0.828,
760
+ "step": 1000
761
+ },
762
+ {
763
+ "epoch": 0.8367514356029533,
764
+ "grad_norm": 0.3536788821220398,
765
+ "learning_rate": 1.4449087736513347e-05,
766
+ "loss": 0.3083,
767
+ "step": 1020
768
+ },
769
+ {
770
+ "epoch": 0.8367514356029533,
771
+ "eval_loss": 0.30134984850883484,
772
+ "eval_runtime": 19.2784,
773
+ "eval_samples_per_second": 51.872,
774
+ "eval_steps_per_second": 0.83,
775
+ "step": 1020
776
+ },
777
+ {
778
+ "epoch": 0.8531583264971287,
779
+ "grad_norm": 0.2920973300933838,
780
+ "learning_rate": 1.4190823130518867e-05,
781
+ "loss": 0.3031,
782
+ "step": 1040
783
+ },
784
+ {
785
+ "epoch": 0.8531583264971287,
786
+ "eval_loss": 0.30034327507019043,
787
+ "eval_runtime": 19.2824,
788
+ "eval_samples_per_second": 51.861,
789
+ "eval_steps_per_second": 0.83,
790
+ "step": 1040
791
+ },
792
+ {
793
+ "epoch": 0.8695652173913043,
794
+ "grad_norm": 0.3366055190563202,
795
+ "learning_rate": 1.3929121709640144e-05,
796
+ "loss": 0.3061,
797
+ "step": 1060
798
+ },
799
+ {
800
+ "epoch": 0.8695652173913043,
801
+ "eval_loss": 0.3006821274757385,
802
+ "eval_runtime": 19.3095,
803
+ "eval_samples_per_second": 51.788,
804
+ "eval_steps_per_second": 0.829,
805
+ "step": 1060
806
+ },
807
+ {
808
+ "epoch": 0.8859721082854799,
809
+ "grad_norm": 0.278680682182312,
810
+ "learning_rate": 1.3664198090268874e-05,
811
+ "loss": 0.3148,
812
+ "step": 1080
813
+ },
814
+ {
815
+ "epoch": 0.8859721082854799,
816
+ "eval_loss": 0.3001416027545929,
817
+ "eval_runtime": 19.3334,
818
+ "eval_samples_per_second": 51.724,
819
+ "eval_steps_per_second": 0.828,
820
+ "step": 1080
821
+ },
822
+ {
823
+ "epoch": 0.9023789991796555,
824
+ "grad_norm": 0.304365336894989,
825
+ "learning_rate": 1.3396269531260868e-05,
826
+ "loss": 0.3034,
827
+ "step": 1100
828
+ },
829
+ {
830
+ "epoch": 0.9023789991796555,
831
+ "eval_loss": 0.29926252365112305,
832
+ "eval_runtime": 19.3347,
833
+ "eval_samples_per_second": 51.72,
834
+ "eval_steps_per_second": 0.828,
835
+ "step": 1100
836
+ },
837
+ {
838
+ "epoch": 0.918785890073831,
839
+ "grad_norm": 0.36314526200294495,
840
+ "learning_rate": 1.3125555755766173e-05,
841
+ "loss": 0.3033,
842
+ "step": 1120
843
+ },
844
+ {
845
+ "epoch": 0.918785890073831,
846
+ "eval_loss": 0.29901349544525146,
847
+ "eval_runtime": 19.3128,
848
+ "eval_samples_per_second": 51.779,
849
+ "eval_steps_per_second": 0.828,
850
+ "step": 1120
851
+ },
852
+ {
853
+ "epoch": 0.9351927809680065,
854
+ "grad_norm": 0.3262890875339508,
855
+ "learning_rate": 1.2852278771038251e-05,
856
+ "loss": 0.3122,
857
+ "step": 1140
858
+ },
859
+ {
860
+ "epoch": 0.9351927809680065,
861
+ "eval_loss": 0.29868951439857483,
862
+ "eval_runtime": 19.3329,
863
+ "eval_samples_per_second": 51.725,
864
+ "eval_steps_per_second": 0.828,
865
+ "step": 1140
866
+ },
867
+ {
868
+ "epoch": 0.9515996718621821,
869
+ "grad_norm": 0.3285571336746216,
870
+ "learning_rate": 1.2576662686370056e-05,
871
+ "loss": 0.3222,
872
+ "step": 1160
873
+ },
874
+ {
875
+ "epoch": 0.9515996718621821,
876
+ "eval_loss": 0.29814422130584717,
877
+ "eval_runtime": 19.3295,
878
+ "eval_samples_per_second": 51.734,
879
+ "eval_steps_per_second": 0.828,
880
+ "step": 1160
881
+ },
882
+ {
883
+ "epoch": 0.9680065627563577,
884
+ "grad_norm": 0.35593438148498535,
885
+ "learning_rate": 1.22989335293062e-05,
886
+ "loss": 0.3022,
887
+ "step": 1180
888
+ },
889
+ {
890
+ "epoch": 0.9680065627563577,
891
+ "eval_loss": 0.29777586460113525,
892
+ "eval_runtime": 19.3277,
893
+ "eval_samples_per_second": 51.739,
894
+ "eval_steps_per_second": 0.828,
895
+ "step": 1180
896
+ },
897
+ {
898
+ "epoch": 0.9844134536505332,
899
+ "grad_norm": 0.2741255462169647,
900
+ "learning_rate": 1.2019319060282065e-05,
901
+ "loss": 0.3111,
902
+ "step": 1200
903
+ },
904
+ {
905
+ "epoch": 0.9844134536505332,
906
+ "eval_loss": 0.29723304510116577,
907
+ "eval_runtime": 19.3222,
908
+ "eval_samples_per_second": 51.754,
909
+ "eval_steps_per_second": 0.828,
910
+ "step": 1200
911
+ },
912
+ {
913
+ "epoch": 1.0008203445447088,
914
+ "grad_norm": 0.3063313961029053,
915
+ "learning_rate": 1.1738048585841742e-05,
916
+ "loss": 0.316,
917
+ "step": 1220
918
+ },
919
+ {
920
+ "epoch": 1.0008203445447088,
921
+ "eval_loss": 0.2970496714115143,
922
+ "eval_runtime": 19.2701,
923
+ "eval_samples_per_second": 51.894,
924
+ "eval_steps_per_second": 0.83,
925
+ "step": 1220
926
+ },
927
+ {
928
+ "epoch": 1.0172272354388843,
929
+ "grad_norm": 0.2891760766506195,
930
+ "learning_rate": 1.1455352770588096e-05,
931
+ "loss": 0.2946,
932
+ "step": 1240
933
+ },
934
+ {
935
+ "epoch": 1.0172272354388843,
936
+ "eval_loss": 0.2964462637901306,
937
+ "eval_runtime": 19.3209,
938
+ "eval_samples_per_second": 51.757,
939
+ "eval_steps_per_second": 0.828,
940
+ "step": 1240
941
+ },
942
+ {
943
+ "epoch": 1.03363412633306,
944
+ "grad_norm": 0.3218066096305847,
945
+ "learning_rate": 1.117146344801906e-05,
946
+ "loss": 0.3042,
947
+ "step": 1260
948
+ },
949
+ {
950
+ "epoch": 1.03363412633306,
951
+ "eval_loss": 0.29618626832962036,
952
+ "eval_runtime": 19.3158,
953
+ "eval_samples_per_second": 51.771,
954
+ "eval_steps_per_second": 0.828,
955
+ "step": 1260
956
+ },
957
+ {
958
+ "epoch": 1.0500410172272354,
959
+ "grad_norm": 0.3612477481365204,
960
+ "learning_rate": 1.0886613430405371e-05,
961
+ "loss": 0.3016,
962
+ "step": 1280
963
+ },
964
+ {
965
+ "epoch": 1.0500410172272354,
966
+ "eval_loss": 0.2955453395843506,
967
+ "eval_runtime": 19.3306,
968
+ "eval_samples_per_second": 51.731,
969
+ "eval_steps_per_second": 0.828,
970
+ "step": 1280
971
+ },
972
+ {
973
+ "epoch": 1.066447908121411,
974
+ "grad_norm": 0.3261948227882385,
975
+ "learning_rate": 1.060103631786563e-05,
976
+ "loss": 0.2961,
977
+ "step": 1300
978
+ },
979
+ {
980
+ "epoch": 1.066447908121411,
981
+ "eval_loss": 0.2949953079223633,
982
+ "eval_runtime": 19.3501,
983
+ "eval_samples_per_second": 51.679,
984
+ "eval_steps_per_second": 0.827,
985
+ "step": 1300
986
+ },
987
+ {
988
+ "epoch": 1.0828547990155866,
989
+ "grad_norm": 0.3773254454135895,
990
+ "learning_rate": 1.0314966306795251e-05,
991
+ "loss": 0.2926,
992
+ "step": 1320
993
+ },
994
+ {
995
+ "epoch": 1.0828547990155866,
996
+ "eval_loss": 0.29533928632736206,
997
+ "eval_runtime": 19.3223,
998
+ "eval_samples_per_second": 51.754,
999
+ "eval_steps_per_second": 0.828,
1000
+ "step": 1320
1001
+ },
1002
+ {
1003
+ "epoch": 1.099261689909762,
1004
+ "grad_norm": 0.33832401037216187,
1005
+ "learning_rate": 1.0028637997806415e-05,
1006
+ "loss": 0.3034,
1007
+ "step": 1340
1008
+ },
1009
+ {
1010
+ "epoch": 1.099261689909762,
1011
+ "eval_loss": 0.2946665585041046,
1012
+ "eval_runtime": 19.3351,
1013
+ "eval_samples_per_second": 51.719,
1014
+ "eval_steps_per_second": 0.828,
1015
+ "step": 1340
1016
+ },
1017
+ {
1018
+ "epoch": 1.1156685808039377,
1019
+ "grad_norm": 0.33615320920944214,
1020
+ "learning_rate": 9.742286203336552e-06,
1021
+ "loss": 0.3049,
1022
+ "step": 1360
1023
+ },
1024
+ {
1025
+ "epoch": 1.1156685808039377,
1026
+ "eval_loss": 0.294543981552124,
1027
+ "eval_runtime": 19.3314,
1028
+ "eval_samples_per_second": 51.729,
1029
+ "eval_steps_per_second": 0.828,
1030
+ "step": 1360
1031
+ },
1032
+ {
1033
+ "epoch": 1.1320754716981132,
1034
+ "grad_norm": 0.3433164060115814,
1035
+ "learning_rate": 9.45614575508309e-06,
1036
+ "loss": 0.2964,
1037
+ "step": 1380
1038
+ },
1039
+ {
1040
+ "epoch": 1.1320754716981132,
1041
+ "eval_loss": 0.29447877407073975,
1042
+ "eval_runtime": 19.3213,
1043
+ "eval_samples_per_second": 51.756,
1044
+ "eval_steps_per_second": 0.828,
1045
+ "step": 1380
1046
+ },
1047
+ {
1048
+ "epoch": 1.1484823625922886,
1049
+ "grad_norm": 0.2896648645401001,
1050
+ "learning_rate": 9.170451311422421e-06,
1051
+ "loss": 0.3079,
1052
+ "step": 1400
1053
+ },
1054
+ {
1055
+ "epoch": 1.1484823625922886,
1056
+ "eval_loss": 0.29404282569885254,
1057
+ "eval_runtime": 19.3311,
1058
+ "eval_samples_per_second": 51.73,
1059
+ "eval_steps_per_second": 0.828,
1060
+ "step": 1400
1061
+ },
1062
+ {
1063
+ "epoch": 1.1648892534864643,
1064
+ "grad_norm": 0.29077446460723877,
1065
+ "learning_rate": 8.88543716497099e-06,
1066
+ "loss": 0.2964,
1067
+ "step": 1420
1068
+ },
1069
+ {
1070
+ "epoch": 1.1648892534864643,
1071
+ "eval_loss": 0.2939617931842804,
1072
+ "eval_runtime": 19.3333,
1073
+ "eval_samples_per_second": 51.724,
1074
+ "eval_steps_per_second": 0.828,
1075
+ "step": 1420
1076
+ },
1077
+ {
1078
+ "epoch": 1.1812961443806398,
1079
+ "grad_norm": 0.3333583474159241,
1080
+ "learning_rate": 8.601337050446356e-06,
1081
+ "loss": 0.2858,
1082
+ "step": 1440
1083
+ },
1084
+ {
1085
+ "epoch": 1.1812961443806398,
1086
+ "eval_loss": 0.29347431659698486,
1087
+ "eval_runtime": 19.3153,
1088
+ "eval_samples_per_second": 51.772,
1089
+ "eval_steps_per_second": 0.828,
1090
+ "step": 1440
1091
+ },
1092
+ {
1093
+ "epoch": 1.1977030352748155,
1094
+ "grad_norm": 0.3010202646255493,
1095
+ "learning_rate": 8.318383952985761e-06,
1096
+ "loss": 0.3053,
1097
+ "step": 1460
1098
+ },
1099
+ {
1100
+ "epoch": 1.1977030352748155,
1101
+ "eval_loss": 0.293270081281662,
1102
+ "eval_runtime": 19.3493,
1103
+ "eval_samples_per_second": 51.681,
1104
+ "eval_steps_per_second": 0.827,
1105
+ "step": 1460
1106
+ },
1107
+ {
1108
+ "epoch": 1.214109926168991,
1109
+ "grad_norm": 0.3123159408569336,
1110
+ "learning_rate": 8.036809917079411e-06,
1111
+ "loss": 0.3038,
1112
+ "step": 1480
1113
+ },
1114
+ {
1115
+ "epoch": 1.214109926168991,
1116
+ "eval_loss": 0.293189138174057,
1117
+ "eval_runtime": 19.3274,
1118
+ "eval_samples_per_second": 51.74,
1119
+ "eval_steps_per_second": 0.828,
1120
+ "step": 1480
1121
+ },
1122
+ {
1123
+ "epoch": 1.2305168170631666,
1124
+ "grad_norm": 0.36915215849876404,
1125
+ "learning_rate": 7.756845856275193e-06,
1126
+ "loss": 0.3006,
1127
+ "step": 1500
1128
+ },
1129
+ {
1130
+ "epoch": 1.2305168170631666,
1131
+ "eval_loss": 0.2929602265357971,
1132
+ "eval_runtime": 19.332,
1133
+ "eval_samples_per_second": 51.728,
1134
+ "eval_steps_per_second": 0.828,
1135
+ "step": 1500
1136
+ },
1137
+ {
1138
+ "epoch": 1.246923707957342,
1139
+ "grad_norm": 0.38524895906448364,
1140
+ "learning_rate": 7.4787213638108246e-06,
1141
+ "loss": 0.3051,
1142
+ "step": 1520
1143
+ },
1144
+ {
1145
+ "epoch": 1.246923707957342,
1146
+ "eval_loss": 0.29255110025405884,
1147
+ "eval_runtime": 19.3216,
1148
+ "eval_samples_per_second": 51.756,
1149
+ "eval_steps_per_second": 0.828,
1150
+ "step": 1520
1151
+ },
1152
+ {
1153
+ "epoch": 1.2633305988515175,
1154
+ "grad_norm": 0.28708815574645996,
1155
+ "learning_rate": 7.202664524328771e-06,
1156
+ "loss": 0.2923,
1157
+ "step": 1540
1158
+ },
1159
+ {
1160
+ "epoch": 1.2633305988515175,
1161
+ "eval_loss": 0.2923203408718109,
1162
+ "eval_runtime": 19.3047,
1163
+ "eval_samples_per_second": 51.801,
1164
+ "eval_steps_per_second": 0.829,
1165
+ "step": 1540
1166
+ },
1167
+ {
1168
+ "epoch": 1.2797374897456932,
1169
+ "grad_norm": 0.32109561562538147,
1170
+ "learning_rate": 6.928901726828351e-06,
1171
+ "loss": 0.3018,
1172
+ "step": 1560
1173
+ },
1174
+ {
1175
+ "epoch": 1.2797374897456932,
1176
+ "eval_loss": 0.29232391715049744,
1177
+ "eval_runtime": 19.3188,
1178
+ "eval_samples_per_second": 51.763,
1179
+ "eval_steps_per_second": 0.828,
1180
+ "step": 1560
1181
+ },
1182
+ {
1183
+ "epoch": 1.2961443806398687,
1184
+ "grad_norm": 0.34937840700149536,
1185
+ "learning_rate": 6.65765747900838e-06,
1186
+ "loss": 0.2982,
1187
+ "step": 1580
1188
+ },
1189
+ {
1190
+ "epoch": 1.2961443806398687,
1191
+ "eval_loss": 0.29210370779037476,
1192
+ "eval_runtime": 19.33,
1193
+ "eval_samples_per_second": 51.733,
1194
+ "eval_steps_per_second": 0.828,
1195
+ "step": 1580
1196
+ },
1197
+ {
1198
+ "epoch": 1.3125512715340442,
1199
+ "grad_norm": 0.3151737153530121,
1200
+ "learning_rate": 6.389154223152667e-06,
1201
+ "loss": 0.2964,
1202
+ "step": 1600
1203
+ },
1204
+ {
1205
+ "epoch": 1.3125512715340442,
1206
+ "eval_loss": 0.2919021546840668,
1207
+ "eval_runtime": 19.3268,
1208
+ "eval_samples_per_second": 51.742,
1209
+ "eval_steps_per_second": 0.828,
1210
+ "step": 1600
1211
+ },
1212
+ {
1213
+ "epoch": 1.3289581624282198,
1214
+ "grad_norm": 0.34189173579216003,
1215
+ "learning_rate": 6.123612153709289e-06,
1216
+ "loss": 0.2955,
1217
+ "step": 1620
1218
+ },
1219
+ {
1220
+ "epoch": 1.3289581624282198,
1221
+ "eval_loss": 0.29170525074005127,
1222
+ "eval_runtime": 19.3259,
1223
+ "eval_samples_per_second": 51.744,
1224
+ "eval_steps_per_second": 0.828,
1225
+ "step": 1620
1226
+ },
1227
+ {
1228
+ "epoch": 1.3453650533223955,
1229
+ "grad_norm": 0.35429632663726807,
1230
+ "learning_rate": 5.861249036713266e-06,
1231
+ "loss": 0.2976,
1232
+ "step": 1640
1233
+ },
1234
+ {
1235
+ "epoch": 1.3453650533223955,
1236
+ "eval_loss": 0.29148247838020325,
1237
+ "eval_runtime": 19.3225,
1238
+ "eval_samples_per_second": 51.753,
1239
+ "eval_steps_per_second": 0.828,
1240
+ "step": 1640
1241
+ },
1242
+ {
1243
+ "epoch": 1.361771944216571,
1244
+ "grad_norm": 0.3811974823474884,
1245
+ "learning_rate": 5.602280031200784e-06,
1246
+ "loss": 0.2859,
1247
+ "step": 1660
1248
+ },
1249
+ {
1250
+ "epoch": 1.361771944216571,
1251
+ "eval_loss": 0.2912766635417938,
1252
+ "eval_runtime": 19.3358,
1253
+ "eval_samples_per_second": 51.718,
1254
+ "eval_steps_per_second": 0.827,
1255
+ "step": 1660
1256
+ },
1257
+ {
1258
+ "epoch": 1.3781788351107465,
1259
+ "grad_norm": 0.30380526185035706,
1260
+ "learning_rate": 5.346917512761323e-06,
1261
+ "loss": 0.2921,
1262
+ "step": 1680
1263
+ },
1264
+ {
1265
+ "epoch": 1.3781788351107465,
1266
+ "eval_loss": 0.29110220074653625,
1267
+ "eval_runtime": 19.3205,
1268
+ "eval_samples_per_second": 51.759,
1269
+ "eval_steps_per_second": 0.828,
1270
+ "step": 1680
1271
+ },
1272
+ {
1273
+ "epoch": 1.3945857260049221,
1274
+ "grad_norm": 0.316942423582077,
1275
+ "learning_rate": 5.0953708993724115e-06,
1276
+ "loss": 0.2964,
1277
+ "step": 1700
1278
+ },
1279
+ {
1280
+ "epoch": 1.3945857260049221,
1281
+ "eval_loss": 0.2907075881958008,
1282
+ "eval_runtime": 19.3166,
1283
+ "eval_samples_per_second": 51.769,
1284
+ "eval_steps_per_second": 0.828,
1285
+ "step": 1700
1286
+ },
1287
+ {
1288
+ "epoch": 1.4109926168990976,
1289
+ "grad_norm": 0.3441399335861206,
1290
+ "learning_rate": 4.847846479659961e-06,
1291
+ "loss": 0.3098,
1292
+ "step": 1720
1293
+ },
1294
+ {
1295
+ "epoch": 1.4109926168990976,
1296
+ "eval_loss": 0.2906677722930908,
1297
+ "eval_runtime": 19.3333,
1298
+ "eval_samples_per_second": 51.724,
1299
+ "eval_steps_per_second": 0.828,
1300
+ "step": 1720
1301
+ },
1302
+ {
1303
+ "epoch": 1.427399507793273,
1304
+ "grad_norm": 0.33759692311286926,
1305
+ "learning_rate": 4.6045472437248096e-06,
1306
+ "loss": 0.3088,
1307
+ "step": 1740
1308
+ },
1309
+ {
1310
+ "epoch": 1.427399507793273,
1311
+ "eval_loss": 0.2904299795627594,
1312
+ "eval_runtime": 19.3386,
1313
+ "eval_samples_per_second": 51.71,
1314
+ "eval_steps_per_second": 0.827,
1315
+ "step": 1740
1316
+ },
1317
+ {
1318
+ "epoch": 1.4438063986874488,
1319
+ "grad_norm": 0.3211503028869629,
1320
+ "learning_rate": 4.365672716674395e-06,
1321
+ "loss": 0.3014,
1322
+ "step": 1760
1323
+ },
1324
+ {
1325
+ "epoch": 1.4438063986874488,
1326
+ "eval_loss": 0.29022130370140076,
1327
+ "eval_runtime": 19.3252,
1328
+ "eval_samples_per_second": 51.746,
1329
+ "eval_steps_per_second": 0.828,
1330
+ "step": 1760
1331
+ },
1332
+ {
1333
+ "epoch": 1.4602132895816242,
1334
+ "grad_norm": 0.3411578834056854,
1335
+ "learning_rate": 4.131418794995999e-06,
1336
+ "loss": 0.2932,
1337
+ "step": 1780
1338
+ },
1339
+ {
1340
+ "epoch": 1.4602132895816242,
1341
+ "eval_loss": 0.2903834283351898,
1342
+ "eval_runtime": 19.3225,
1343
+ "eval_samples_per_second": 51.753,
1344
+ "eval_steps_per_second": 0.828,
1345
+ "step": 1780
1346
+ },
1347
+ {
1348
+ "epoch": 1.4766201804758,
1349
+ "grad_norm": 0.3540644943714142,
1350
+ "learning_rate": 3.901977585905692e-06,
1351
+ "loss": 0.3052,
1352
+ "step": 1800
1353
+ },
1354
+ {
1355
+ "epoch": 1.4766201804758,
1356
+ "eval_loss": 0.29012495279312134,
1357
+ "eval_runtime": 19.3397,
1358
+ "eval_samples_per_second": 51.707,
1359
+ "eval_steps_per_second": 0.827,
1360
+ "step": 1800
1361
+ },
1362
+ {
1363
+ "epoch": 1.4930270713699754,
1364
+ "grad_norm": 0.34099695086479187,
1365
+ "learning_rate": 3.6775372498048912e-06,
1366
+ "loss": 0.2875,
1367
+ "step": 1820
1368
+ },
1369
+ {
1370
+ "epoch": 1.4930270713699754,
1371
+ "eval_loss": 0.28988543152809143,
1372
+ "eval_runtime": 19.3381,
1373
+ "eval_samples_per_second": 51.711,
1374
+ "eval_steps_per_second": 0.827,
1375
+ "step": 1820
1376
+ },
1377
+ {
1378
+ "epoch": 1.509433962264151,
1379
+ "grad_norm": 0.32933956384658813,
1380
+ "learning_rate": 3.458281845973528e-06,
1381
+ "loss": 0.3053,
1382
+ "step": 1840
1383
+ },
1384
+ {
1385
+ "epoch": 1.509433962264151,
1386
+ "eval_loss": 0.2898189127445221,
1387
+ "eval_runtime": 19.3309,
1388
+ "eval_samples_per_second": 51.731,
1389
+ "eval_steps_per_second": 0.828,
1390
+ "step": 1840
1391
+ },
1392
+ {
1393
+ "epoch": 1.5258408531583265,
1394
+ "grad_norm": 0.30174052715301514,
1395
+ "learning_rate": 3.244391181626522e-06,
1396
+ "loss": 0.2988,
1397
+ "step": 1860
1398
+ },
1399
+ {
1400
+ "epoch": 1.5258408531583265,
1401
+ "eval_loss": 0.28971782326698303,
1402
+ "eval_runtime": 19.3408,
1403
+ "eval_samples_per_second": 51.704,
1404
+ "eval_steps_per_second": 0.827,
1405
+ "step": 1860
1406
+ },
1407
+ {
1408
+ "epoch": 1.542247744052502,
1409
+ "grad_norm": 0.3678150177001953,
1410
+ "learning_rate": 3.0360406644572935e-06,
1411
+ "loss": 0.2923,
1412
+ "step": 1880
1413
+ },
1414
+ {
1415
+ "epoch": 1.542247744052502,
1416
+ "eval_loss": 0.28958600759506226,
1417
+ "eval_runtime": 19.3299,
1418
+ "eval_samples_per_second": 51.733,
1419
+ "eval_steps_per_second": 0.828,
1420
+ "step": 1880
1421
+ },
1422
+ {
1423
+ "epoch": 1.5586546349466777,
1424
+ "grad_norm": 0.31850239634513855,
1425
+ "learning_rate": 2.8334011587892073e-06,
1426
+ "loss": 0.2956,
1427
+ "step": 1900
1428
+ },
1429
+ {
1430
+ "epoch": 1.5586546349466777,
1431
+ "eval_loss": 0.2897399961948395,
1432
+ "eval_runtime": 19.3388,
1433
+ "eval_samples_per_second": 51.71,
1434
+ "eval_steps_per_second": 0.827,
1435
+ "step": 1900
1436
+ },
1437
+ {
1438
+ "epoch": 1.5750615258408531,
1439
+ "grad_norm": 0.3421303927898407,
1440
+ "learning_rate": 2.636638845452979e-06,
1441
+ "loss": 0.3,
1442
+ "step": 1920
1443
+ },
1444
+ {
1445
+ "epoch": 1.5750615258408531,
1446
+ "eval_loss": 0.2895210087299347,
1447
+ "eval_runtime": 19.3352,
1448
+ "eval_samples_per_second": 51.719,
1449
+ "eval_steps_per_second": 0.828,
1450
+ "step": 1920
1451
+ },
1452
+ {
1453
+ "epoch": 1.5914684167350286,
1454
+ "grad_norm": 0.32379043102264404,
1455
+ "learning_rate": 2.4459150855049152e-06,
1456
+ "loss": 0.2991,
1457
+ "step": 1940
1458
+ },
1459
+ {
1460
+ "epoch": 1.5914684167350286,
1461
+ "eval_loss": 0.28945392370224,
1462
+ "eval_runtime": 19.3341,
1463
+ "eval_samples_per_second": 51.722,
1464
+ "eval_steps_per_second": 0.828,
1465
+ "step": 1940
1466
+ },
1467
+ {
1468
+ "epoch": 1.6078753076292043,
1469
+ "grad_norm": 0.3421061038970947,
1470
+ "learning_rate": 2.2613862878977554e-06,
1471
+ "loss": 0.2992,
1472
+ "step": 1960
1473
+ },
1474
+ {
1475
+ "epoch": 1.6078753076292043,
1476
+ "eval_loss": 0.2892807722091675,
1477
+ "eval_runtime": 19.3293,
1478
+ "eval_samples_per_second": 51.735,
1479
+ "eval_steps_per_second": 0.828,
1480
+ "step": 1960
1481
+ },
1482
+ {
1483
+ "epoch": 1.62428219852338,
1484
+ "grad_norm": 0.3523494303226471,
1485
+ "learning_rate": 2.083203781212647e-06,
1486
+ "loss": 0.2987,
1487
+ "step": 1980
1488
+ },
1489
+ {
1490
+ "epoch": 1.62428219852338,
1491
+ "eval_loss": 0.28924793004989624,
1492
+ "eval_runtime": 19.3139,
1493
+ "eval_samples_per_second": 51.776,
1494
+ "eval_steps_per_second": 0.828,
1495
+ "step": 1980
1496
+ },
1497
+ {
1498
+ "epoch": 1.6406890894175554,
1499
+ "grad_norm": 0.3594619333744049,
1500
+ "learning_rate": 1.9115136895574403e-06,
1501
+ "loss": 0.3028,
1502
+ "step": 2000
1503
+ },
1504
+ {
1505
+ "epoch": 1.6406890894175554,
1506
+ "eval_loss": 0.2891727387905121,
1507
+ "eval_runtime": 19.3202,
1508
+ "eval_samples_per_second": 51.759,
1509
+ "eval_steps_per_second": 0.828,
1510
+ "step": 2000
1511
+ },
1512
+ {
1513
+ "epoch": 1.6570959803117309,
1514
+ "grad_norm": 0.32979699969291687,
1515
+ "learning_rate": 1.7464568127330562e-06,
1516
+ "loss": 0.2879,
1517
+ "step": 2020
1518
+ },
1519
+ {
1520
+ "epoch": 1.6570959803117309,
1521
+ "eval_loss": 0.2892407774925232,
1522
+ "eval_runtime": 19.2847,
1523
+ "eval_samples_per_second": 51.855,
1524
+ "eval_steps_per_second": 0.83,
1525
+ "step": 2020
1526
+ },
1527
+ {
1528
+ "epoch": 1.6735028712059066,
1529
+ "grad_norm": 0.32581305503845215,
1530
+ "learning_rate": 1.5881685107662336e-06,
1531
+ "loss": 0.2887,
1532
+ "step": 2040
1533
+ },
1534
+ {
1535
+ "epoch": 1.6735028712059066,
1536
+ "eval_loss": 0.289132684469223,
1537
+ "eval_runtime": 19.3167,
1538
+ "eval_samples_per_second": 51.769,
1539
+ "eval_steps_per_second": 0.828,
1540
+ "step": 2040
1541
+ },
1542
+ {
1543
+ "epoch": 1.689909762100082,
1544
+ "grad_norm": 0.4018111824989319,
1545
+ "learning_rate": 1.4367785929033395e-06,
1546
+ "loss": 0.3037,
1547
+ "step": 2060
1548
+ },
1549
+ {
1550
+ "epoch": 1.689909762100082,
1551
+ "eval_loss": 0.2891572117805481,
1552
+ "eval_runtime": 19.3154,
1553
+ "eval_samples_per_second": 51.772,
1554
+ "eval_steps_per_second": 0.828,
1555
+ "step": 2060
1556
+ },
1557
+ {
1558
+ "epoch": 1.7063166529942575,
1559
+ "grad_norm": 0.34754762053489685,
1560
+ "learning_rate": 1.2924112111562314e-06,
1561
+ "loss": 0.2921,
1562
+ "step": 2080
1563
+ },
1564
+ {
1565
+ "epoch": 1.7063166529942575,
1566
+ "eval_loss": 0.2889350950717926,
1567
+ "eval_runtime": 19.303,
1568
+ "eval_samples_per_second": 51.806,
1569
+ "eval_steps_per_second": 0.829,
1570
+ "step": 2080
1571
+ },
1572
+ {
1573
+ "epoch": 1.7227235438884332,
1574
+ "grad_norm": 0.33574923872947693,
1575
+ "learning_rate": 1.155184758487573e-06,
1576
+ "loss": 0.2894,
1577
+ "step": 2100
1578
+ },
1579
+ {
1580
+ "epoch": 1.7227235438884332,
1581
+ "eval_loss": 0.28887927532196045,
1582
+ "eval_runtime": 19.3488,
1583
+ "eval_samples_per_second": 51.683,
1584
+ "eval_steps_per_second": 0.827,
1585
+ "step": 2100
1586
+ },
1587
+ {
1588
+ "epoch": 1.7391304347826086,
1589
+ "grad_norm": 0.31638723611831665,
1590
+ "learning_rate": 1.0252117717189713e-06,
1591
+ "loss": 0.2892,
1592
+ "step": 2120
1593
+ },
1594
+ {
1595
+ "epoch": 1.7391304347826086,
1596
+ "eval_loss": 0.28898754715919495,
1597
+ "eval_runtime": 19.3213,
1598
+ "eval_samples_per_second": 51.756,
1599
+ "eval_steps_per_second": 0.828,
1600
+ "step": 2120
1601
+ },
1602
+ {
1603
+ "epoch": 1.755537325676784,
1604
+ "grad_norm": 0.3385666012763977,
1605
+ "learning_rate": 9.025988392416718e-07,
1606
+ "loss": 0.2865,
1607
+ "step": 2140
1608
+ },
1609
+ {
1610
+ "epoch": 1.755537325676784,
1611
+ "eval_loss": 0.28879979252815247,
1612
+ "eval_runtime": 19.3292,
1613
+ "eval_samples_per_second": 51.735,
1614
+ "eval_steps_per_second": 0.828,
1615
+ "step": 2140
1616
+ },
1617
+ {
1618
+ "epoch": 1.7719442165709598,
1619
+ "grad_norm": 0.35082417726516724,
1620
+ "learning_rate": 7.874465136054332e-07,
1621
+ "loss": 0.2886,
1622
+ "step": 2160
1623
+ },
1624
+ {
1625
+ "epoch": 1.7719442165709598,
1626
+ "eval_loss": 0.28894442319869995,
1627
+ "eval_runtime": 19.3371,
1628
+ "eval_samples_per_second": 51.714,
1629
+ "eval_steps_per_second": 0.827,
1630
+ "step": 2160
1631
+ },
1632
+ {
1633
+ "epoch": 1.7883511074651355,
1634
+ "grad_norm": 0.28936704993247986,
1635
+ "learning_rate": 6.798492290572611e-07,
1636
+ "loss": 0.2908,
1637
+ "step": 2180
1638
+ },
1639
+ {
1640
+ "epoch": 1.7883511074651355,
1641
+ "eval_loss": 0.288817435503006,
1642
+ "eval_runtime": 19.3316,
1643
+ "eval_samples_per_second": 51.729,
1644
+ "eval_steps_per_second": 0.828,
1645
+ "step": 2180
1646
+ },
1647
+ {
1648
+ "epoch": 1.804757998359311,
1649
+ "grad_norm": 0.3725113272666931,
1650
+ "learning_rate": 5.798952240976952e-07,
1651
+ "loss": 0.2982,
1652
+ "step": 2200
1653
+ },
1654
+ {
1655
+ "epoch": 1.804757998359311,
1656
+ "eval_loss": 0.2887531518936157,
1657
+ "eval_runtime": 19.3355,
1658
+ "eval_samples_per_second": 51.718,
1659
+ "eval_steps_per_second": 0.827,
1660
+ "step": 2200
1661
+ },
1662
+ {
1663
+ "epoch": 1.8211648892534864,
1664
+ "grad_norm": 0.3677207827568054,
1665
+ "learning_rate": 4.876664691180599e-07,
1666
+ "loss": 0.2849,
1667
+ "step": 2220
1668
+ },
1669
+ {
1670
+ "epoch": 1.8211648892534864,
1671
+ "eval_loss": 0.2887623906135559,
1672
+ "eval_runtime": 19.3212,
1673
+ "eval_samples_per_second": 51.757,
1674
+ "eval_steps_per_second": 0.828,
1675
+ "step": 2220
1676
+ },
1677
+ {
1678
+ "epoch": 1.837571780147662,
1679
+ "grad_norm": 0.36475974321365356,
1680
+ "learning_rate": 4.032385991781029e-07,
1681
+ "loss": 0.2962,
1682
+ "step": 2240
1683
+ },
1684
+ {
1685
+ "epoch": 1.837571780147662,
1686
+ "eval_loss": 0.28872615098953247,
1687
+ "eval_runtime": 19.3232,
1688
+ "eval_samples_per_second": 51.751,
1689
+ "eval_steps_per_second": 0.828,
1690
+ "step": 2240
1691
+ },
1692
+ {
1693
+ "epoch": 1.8539786710418376,
1694
+ "grad_norm": 0.32117658853530884,
1695
+ "learning_rate": 3.266808519791076e-07,
1696
+ "loss": 0.2885,
1697
+ "step": 2260
1698
+ },
1699
+ {
1700
+ "epoch": 1.8539786710418376,
1701
+ "eval_loss": 0.28867271542549133,
1702
+ "eval_runtime": 19.3343,
1703
+ "eval_samples_per_second": 51.722,
1704
+ "eval_steps_per_second": 0.828,
1705
+ "step": 2260
1706
+ },
1707
+ {
1708
+ "epoch": 1.870385561936013,
1709
+ "grad_norm": 0.3309364318847656,
1710
+ "learning_rate": 2.58056011083373e-07,
1711
+ "loss": 0.2952,
1712
+ "step": 2280
1713
+ },
1714
+ {
1715
+ "epoch": 1.870385561936013,
1716
+ "eval_loss": 0.28868669271469116,
1717
+ "eval_runtime": 19.3187,
1718
+ "eval_samples_per_second": 51.763,
1719
+ "eval_steps_per_second": 0.828,
1720
+ "step": 2280
1721
+ },
1722
+ {
1723
+ "epoch": 1.8867924528301887,
1724
+ "grad_norm": 0.39098307490348816,
1725
+ "learning_rate": 1.9742035442658402e-07,
1726
+ "loss": 0.3053,
1727
+ "step": 2300
1728
+ },
1729
+ {
1730
+ "epoch": 1.8867924528301887,
1731
+ "eval_loss": 0.28870368003845215,
1732
+ "eval_runtime": 19.3209,
1733
+ "eval_samples_per_second": 51.757,
1734
+ "eval_steps_per_second": 0.828,
1735
+ "step": 2300
1736
+ },
1737
+ {
1738
+ "epoch": 1.9031993437243644,
1739
+ "grad_norm": 0.3185991644859314,
1740
+ "learning_rate": 1.448236081653509e-07,
1741
+ "loss": 0.2979,
1742
+ "step": 2320
1743
+ },
1744
+ {
1745
+ "epoch": 1.9031993437243644,
1746
+ "eval_loss": 0.288663387298584,
1747
+ "eval_runtime": 19.3251,
1748
+ "eval_samples_per_second": 51.746,
1749
+ "eval_steps_per_second": 0.828,
1750
+ "step": 2320
1751
+ },
1752
+ {
1753
+ "epoch": 1.9196062346185396,
1754
+ "grad_norm": 0.3167499899864197,
1755
+ "learning_rate": 1.0030890589772468e-07,
1756
+ "loss": 0.2926,
1757
+ "step": 2340
1758
+ },
1759
+ {
1760
+ "epoch": 1.9196062346185396,
1761
+ "eval_loss": 0.28869006037712097,
1762
+ "eval_runtime": 19.3272,
1763
+ "eval_samples_per_second": 51.741,
1764
+ "eval_steps_per_second": 0.828,
1765
+ "step": 2340
1766
+ },
1767
+ {
1768
+ "epoch": 1.9360131255127153,
1769
+ "grad_norm": 0.35239729285240173,
1770
+ "learning_rate": 6.391275329014734e-08,
1771
+ "loss": 0.2948,
1772
+ "step": 2360
1773
+ },
1774
+ {
1775
+ "epoch": 1.9360131255127153,
1776
+ "eval_loss": 0.28865668177604675,
1777
+ "eval_runtime": 19.3327,
1778
+ "eval_samples_per_second": 51.726,
1779
+ "eval_steps_per_second": 0.828,
1780
+ "step": 2360
1781
+ },
1782
+ {
1783
+ "epoch": 1.952420016406891,
1784
+ "grad_norm": 0.3485466539859772,
1785
+ "learning_rate": 3.566499813984603e-08,
1786
+ "loss": 0.3083,
1787
+ "step": 2380
1788
+ },
1789
+ {
1790
+ "epoch": 1.952420016406891,
1791
+ "eval_loss": 0.2886667251586914,
1792
+ "eval_runtime": 19.336,
1793
+ "eval_samples_per_second": 51.717,
1794
+ "eval_steps_per_second": 0.827,
1795
+ "step": 2380
1796
+ },
1797
+ {
1798
+ "epoch": 1.9688269073010665,
1799
+ "grad_norm": 0.32429105043411255,
1800
+ "learning_rate": 1.5588805897215344e-08,
1801
+ "loss": 0.3034,
1802
+ "step": 2400
1803
+ },
1804
+ {
1805
+ "epoch": 1.9688269073010665,
1806
+ "eval_loss": 0.28866520524024963,
1807
+ "eval_runtime": 19.327,
1808
+ "eval_samples_per_second": 51.741,
1809
+ "eval_steps_per_second": 0.828,
1810
+ "step": 2400
1811
+ },
1812
+ {
1813
+ "epoch": 1.985233798195242,
1814
+ "grad_norm": 0.3192400336265564,
1815
+ "learning_rate": 3.7006406682726927e-09,
1816
+ "loss": 0.2862,
1817
+ "step": 2420
1818
+ },
1819
+ {
1820
+ "epoch": 1.985233798195242,
1821
+ "eval_loss": 0.28866344690322876,
1822
+ "eval_runtime": 19.3203,
1823
+ "eval_samples_per_second": 51.759,
1824
+ "eval_steps_per_second": 0.828,
1825
+ "step": 2420
1826
+ },
1827
+ {
1828
+ "epoch": 2.0,
1829
+ "step": 2438,
1830
+ "total_flos": 1.58523627405312e+18,
1831
+ "train_loss": 0.33499454004633517,
1832
+ "train_runtime": 4698.1162,
1833
+ "train_samples_per_second": 16.602,
1834
+ "train_steps_per_second": 0.519
1835
+ }
1836
+ ],
1837
+ "logging_steps": 20,
1838
+ "max_steps": 2438,
1839
+ "num_input_tokens_seen": 0,
1840
+ "num_train_epochs": 2,
1841
+ "save_steps": 500,
1842
+ "stateful_callbacks": {
1843
+ "TrainerControl": {
1844
+ "args": {
1845
+ "should_epoch_stop": false,
1846
+ "should_evaluate": false,
1847
+ "should_log": false,
1848
+ "should_save": true,
1849
+ "should_training_stop": true
1850
+ },
1851
+ "attributes": {}
1852
+ }
1853
+ },
1854
+ "total_flos": 1.58523627405312e+18,
1855
+ "train_batch_size": 32,
1856
+ "trial_name": null,
1857
+ "trial_params": null
1858
+ }
nl_tasks/run_all/run_exps3/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 4,
9
+ "peft_type": "ROTATION",
10
+ "r": 4,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exps3/ft/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "</s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "</s>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
nl_tasks/run_all/run_exps3/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exps3/ft/tokenizer_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ },
30
+ "32000": {
31
+ "content": "[PAD]",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": true
37
+ }
38
+ },
39
+ "bos_token": "</s>",
40
+ "clean_up_tokenization_spaces": false,
41
+ "eos_token": "</s>",
42
+ "extra_special_tokens": {},
43
+ "legacy": false,
44
+ "model_max_length": 512,
45
+ "pad_token": "[PAD]",
46
+ "padding_side": "right",
47
+ "sp_model_kwargs": {},
48
+ "tokenizer_class": "LlamaTokenizer",
49
+ "unk_token": "</s>",
50
+ "use_default_system_prompt": false
51
+ }
nl_tasks/run_all/run_exps3/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a26eab11acb98f0acd95e0d9119808866cc65ecac779688f2c7419a77826e61
3
+ size 6481
nl_tasks/run_all/run_exps3/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 4,
9
+ "peft_type": "ROTATION",
10
+ "r": 4,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "q_proj",
14
+ "v_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exps3/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0ed46558a5446f8863c7f6599e0b271b3415d50013ffee75934b492e9768e45
3
+ size 33602659
nl_tasks/run_all/run_exps3/trainer_state.json ADDED
@@ -0,0 +1,1858 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.0,
6
+ "eval_steps": 20,
7
+ "global_step": 2438,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.016406890894175553,
14
+ "grad_norm": 0.05860578641295433,
15
+ "learning_rate": 3.89344262295082e-06,
16
+ "loss": 0.746,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.016406890894175553,
21
+ "eval_loss": 0.7427030205726624,
22
+ "eval_runtime": 19.8106,
23
+ "eval_samples_per_second": 50.478,
24
+ "eval_steps_per_second": 0.808,
25
+ "step": 20
26
+ },
27
+ {
28
+ "epoch": 0.03281378178835111,
29
+ "grad_norm": 0.1488150805234909,
30
+ "learning_rate": 7.991803278688526e-06,
31
+ "loss": 0.7525,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.03281378178835111,
36
+ "eval_loss": 0.7261132597923279,
37
+ "eval_runtime": 19.2875,
38
+ "eval_samples_per_second": 51.847,
39
+ "eval_steps_per_second": 0.83,
40
+ "step": 40
41
+ },
42
+ {
43
+ "epoch": 0.04922067268252666,
44
+ "grad_norm": 0.28110650181770325,
45
+ "learning_rate": 1.209016393442623e-05,
46
+ "loss": 0.703,
47
+ "step": 60
48
+ },
49
+ {
50
+ "epoch": 0.04922067268252666,
51
+ "eval_loss": 0.6480915546417236,
52
+ "eval_runtime": 19.3032,
53
+ "eval_samples_per_second": 51.805,
54
+ "eval_steps_per_second": 0.829,
55
+ "step": 60
56
+ },
57
+ {
58
+ "epoch": 0.06562756357670221,
59
+ "grad_norm": 0.20089583098888397,
60
+ "learning_rate": 1.6188524590163936e-05,
61
+ "loss": 0.5856,
62
+ "step": 80
63
+ },
64
+ {
65
+ "epoch": 0.06562756357670221,
66
+ "eval_loss": 0.5077388286590576,
67
+ "eval_runtime": 19.3009,
68
+ "eval_samples_per_second": 51.811,
69
+ "eval_steps_per_second": 0.829,
70
+ "step": 80
71
+ },
72
+ {
73
+ "epoch": 0.08203445447087777,
74
+ "grad_norm": 0.15508203208446503,
75
+ "learning_rate": 2.028688524590164e-05,
76
+ "loss": 0.4828,
77
+ "step": 100
78
+ },
79
+ {
80
+ "epoch": 0.08203445447087777,
81
+ "eval_loss": 0.44249895215034485,
82
+ "eval_runtime": 19.331,
83
+ "eval_samples_per_second": 51.731,
84
+ "eval_steps_per_second": 0.828,
85
+ "step": 100
86
+ },
87
+ {
88
+ "epoch": 0.09844134536505332,
89
+ "grad_norm": 0.16930700838565826,
90
+ "learning_rate": 2.4385245901639343e-05,
91
+ "loss": 0.4341,
92
+ "step": 120
93
+ },
94
+ {
95
+ "epoch": 0.09844134536505332,
96
+ "eval_loss": 0.4118768870830536,
97
+ "eval_runtime": 19.3095,
98
+ "eval_samples_per_second": 51.788,
99
+ "eval_steps_per_second": 0.829,
100
+ "step": 120
101
+ },
102
+ {
103
+ "epoch": 0.11484823625922888,
104
+ "grad_norm": 0.23524542152881622,
105
+ "learning_rate": 2.848360655737705e-05,
106
+ "loss": 0.4019,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.11484823625922888,
111
+ "eval_loss": 0.3899981677532196,
112
+ "eval_runtime": 19.3319,
113
+ "eval_samples_per_second": 51.728,
114
+ "eval_steps_per_second": 0.828,
115
+ "step": 140
116
+ },
117
+ {
118
+ "epoch": 0.13125512715340443,
119
+ "grad_norm": 0.20420299470424652,
120
+ "learning_rate": 3.2581967213114754e-05,
121
+ "loss": 0.3836,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.13125512715340443,
126
+ "eval_loss": 0.3749648332595825,
127
+ "eval_runtime": 19.3132,
128
+ "eval_samples_per_second": 51.778,
129
+ "eval_steps_per_second": 0.828,
130
+ "step": 160
131
+ },
132
+ {
133
+ "epoch": 0.14766201804757997,
134
+ "grad_norm": 0.20580491423606873,
135
+ "learning_rate": 3.668032786885246e-05,
136
+ "loss": 0.3773,
137
+ "step": 180
138
+ },
139
+ {
140
+ "epoch": 0.14766201804757997,
141
+ "eval_loss": 0.3625333905220032,
142
+ "eval_runtime": 19.3238,
143
+ "eval_samples_per_second": 51.75,
144
+ "eval_steps_per_second": 0.828,
145
+ "step": 180
146
+ },
147
+ {
148
+ "epoch": 0.16406890894175555,
149
+ "grad_norm": 0.2624441087245941,
150
+ "learning_rate": 4.077868852459016e-05,
151
+ "loss": 0.3678,
152
+ "step": 200
153
+ },
154
+ {
155
+ "epoch": 0.16406890894175555,
156
+ "eval_loss": 0.35312020778656006,
157
+ "eval_runtime": 19.3256,
158
+ "eval_samples_per_second": 51.745,
159
+ "eval_steps_per_second": 0.828,
160
+ "step": 200
161
+ },
162
+ {
163
+ "epoch": 0.1804757998359311,
164
+ "grad_norm": 0.2771821618080139,
165
+ "learning_rate": 4.487704918032787e-05,
166
+ "loss": 0.371,
167
+ "step": 220
168
+ },
169
+ {
170
+ "epoch": 0.1804757998359311,
171
+ "eval_loss": 0.345496267080307,
172
+ "eval_runtime": 19.2949,
173
+ "eval_samples_per_second": 51.827,
174
+ "eval_steps_per_second": 0.829,
175
+ "step": 220
176
+ },
177
+ {
178
+ "epoch": 0.19688269073010664,
179
+ "grad_norm": 0.2438289374113083,
180
+ "learning_rate": 4.8975409836065575e-05,
181
+ "loss": 0.3552,
182
+ "step": 240
183
+ },
184
+ {
185
+ "epoch": 0.19688269073010664,
186
+ "eval_loss": 0.3394674062728882,
187
+ "eval_runtime": 19.3273,
188
+ "eval_samples_per_second": 51.74,
189
+ "eval_steps_per_second": 0.828,
190
+ "step": 240
191
+ },
192
+ {
193
+ "epoch": 0.2132895816242822,
194
+ "grad_norm": 0.2446252554655075,
195
+ "learning_rate": 4.9994233632277596e-05,
196
+ "loss": 0.3487,
197
+ "step": 260
198
+ },
199
+ {
200
+ "epoch": 0.2132895816242822,
201
+ "eval_loss": 0.3342157006263733,
202
+ "eval_runtime": 19.3079,
203
+ "eval_samples_per_second": 51.792,
204
+ "eval_steps_per_second": 0.829,
205
+ "step": 260
206
+ },
207
+ {
208
+ "epoch": 0.22969647251845776,
209
+ "grad_norm": 0.22413121163845062,
210
+ "learning_rate": 4.99686106951168e-05,
211
+ "loss": 0.3482,
212
+ "step": 280
213
+ },
214
+ {
215
+ "epoch": 0.22969647251845776,
216
+ "eval_loss": 0.3296491205692291,
217
+ "eval_runtime": 19.3306,
218
+ "eval_samples_per_second": 51.732,
219
+ "eval_steps_per_second": 0.828,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2461033634126333,
224
+ "grad_norm": 0.2740541398525238,
225
+ "learning_rate": 4.992251147198466e-05,
226
+ "loss": 0.3344,
227
+ "step": 300
228
+ },
229
+ {
230
+ "epoch": 0.2461033634126333,
231
+ "eval_loss": 0.32672810554504395,
232
+ "eval_runtime": 19.3105,
233
+ "eval_samples_per_second": 51.785,
234
+ "eval_steps_per_second": 0.829,
235
+ "step": 300
236
+ },
237
+ {
238
+ "epoch": 0.26251025430680885,
239
+ "grad_norm": 0.22817863523960114,
240
+ "learning_rate": 4.985597376798324e-05,
241
+ "loss": 0.3252,
242
+ "step": 320
243
+ },
244
+ {
245
+ "epoch": 0.26251025430680885,
246
+ "eval_loss": 0.3222906291484833,
247
+ "eval_runtime": 19.3039,
248
+ "eval_samples_per_second": 51.803,
249
+ "eval_steps_per_second": 0.829,
250
+ "step": 320
251
+ },
252
+ {
253
+ "epoch": 0.27891714520098443,
254
+ "grad_norm": 0.23338018357753754,
255
+ "learning_rate": 4.976905214942672e-05,
256
+ "loss": 0.3389,
257
+ "step": 340
258
+ },
259
+ {
260
+ "epoch": 0.27891714520098443,
261
+ "eval_loss": 0.3193089962005615,
262
+ "eval_runtime": 19.3464,
263
+ "eval_samples_per_second": 51.689,
264
+ "eval_steps_per_second": 0.827,
265
+ "step": 340
266
+ },
267
+ {
268
+ "epoch": 0.29532403609515995,
269
+ "grad_norm": 0.287139892578125,
270
+ "learning_rate": 4.966181789909249e-05,
271
+ "loss": 0.3175,
272
+ "step": 360
273
+ },
274
+ {
275
+ "epoch": 0.29532403609515995,
276
+ "eval_loss": 0.31686943769454956,
277
+ "eval_runtime": 19.3265,
278
+ "eval_samples_per_second": 51.742,
279
+ "eval_steps_per_second": 0.828,
280
+ "step": 360
281
+ },
282
+ {
283
+ "epoch": 0.3117309269893355,
284
+ "grad_norm": 0.32575851678848267,
285
+ "learning_rate": 4.95343589577636e-05,
286
+ "loss": 0.3231,
287
+ "step": 380
288
+ },
289
+ {
290
+ "epoch": 0.3117309269893355,
291
+ "eval_loss": 0.3144107758998871,
292
+ "eval_runtime": 19.298,
293
+ "eval_samples_per_second": 51.819,
294
+ "eval_steps_per_second": 0.829,
295
+ "step": 380
296
+ },
297
+ {
298
+ "epoch": 0.3281378178835111,
299
+ "grad_norm": 0.28842759132385254,
300
+ "learning_rate": 4.938677985211011e-05,
301
+ "loss": 0.3345,
302
+ "step": 400
303
+ },
304
+ {
305
+ "epoch": 0.3281378178835111,
306
+ "eval_loss": 0.3121042549610138,
307
+ "eval_runtime": 19.3325,
308
+ "eval_samples_per_second": 51.726,
309
+ "eval_steps_per_second": 0.828,
310
+ "step": 400
311
+ },
312
+ {
313
+ "epoch": 0.3445447087776866,
314
+ "grad_norm": 0.24486076831817627,
315
+ "learning_rate": 4.921920160896879e-05,
316
+ "loss": 0.3266,
317
+ "step": 420
318
+ },
319
+ {
320
+ "epoch": 0.3445447087776866,
321
+ "eval_loss": 0.3106197118759155,
322
+ "eval_runtime": 19.322,
323
+ "eval_samples_per_second": 51.755,
324
+ "eval_steps_per_second": 0.828,
325
+ "step": 420
326
+ },
327
+ {
328
+ "epoch": 0.3609515996718622,
329
+ "grad_norm": 0.2711906433105469,
330
+ "learning_rate": 4.9031761656091264e-05,
331
+ "loss": 0.326,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.3609515996718622,
336
+ "eval_loss": 0.3084668517112732,
337
+ "eval_runtime": 19.316,
338
+ "eval_samples_per_second": 51.77,
339
+ "eval_steps_per_second": 0.828,
340
+ "step": 440
341
+ },
342
+ {
343
+ "epoch": 0.37735849056603776,
344
+ "grad_norm": 0.21624377369880676,
345
+ "learning_rate": 4.8824613709442124e-05,
346
+ "loss": 0.3268,
347
+ "step": 460
348
+ },
349
+ {
350
+ "epoch": 0.37735849056603776,
351
+ "eval_loss": 0.3072848618030548,
352
+ "eval_runtime": 19.3229,
353
+ "eval_samples_per_second": 51.752,
354
+ "eval_steps_per_second": 0.828,
355
+ "step": 460
356
+ },
357
+ {
358
+ "epoch": 0.3937653814602133,
359
+ "grad_norm": 0.25868281722068787,
360
+ "learning_rate": 4.859792764713938e-05,
361
+ "loss": 0.316,
362
+ "step": 480
363
+ },
364
+ {
365
+ "epoch": 0.3937653814602133,
366
+ "eval_loss": 0.3052523136138916,
367
+ "eval_runtime": 19.3193,
368
+ "eval_samples_per_second": 51.762,
369
+ "eval_steps_per_second": 0.828,
370
+ "step": 480
371
+ },
372
+ {
373
+ "epoch": 0.41017227235438886,
374
+ "grad_norm": 0.22380059957504272,
375
+ "learning_rate": 4.835188937014059e-05,
376
+ "loss": 0.311,
377
+ "step": 500
378
+ },
379
+ {
380
+ "epoch": 0.41017227235438886,
381
+ "eval_loss": 0.3041682839393616,
382
+ "eval_runtime": 19.3139,
383
+ "eval_samples_per_second": 51.776,
384
+ "eval_steps_per_second": 0.828,
385
+ "step": 500
386
+ },
387
+ {
388
+ "epoch": 0.4265791632485644,
389
+ "grad_norm": 0.3189392685890198,
390
+ "learning_rate": 4.808670064978903e-05,
391
+ "loss": 0.3199,
392
+ "step": 520
393
+ },
394
+ {
395
+ "epoch": 0.4265791632485644,
396
+ "eval_loss": 0.3025827407836914,
397
+ "eval_runtime": 19.3135,
398
+ "eval_samples_per_second": 51.777,
399
+ "eval_steps_per_second": 0.828,
400
+ "step": 520
401
+ },
402
+ {
403
+ "epoch": 0.44298605414273995,
404
+ "grad_norm": 0.30145472288131714,
405
+ "learning_rate": 4.7802578962344836e-05,
406
+ "loss": 0.3116,
407
+ "step": 540
408
+ },
409
+ {
410
+ "epoch": 0.44298605414273995,
411
+ "eval_loss": 0.3007499575614929,
412
+ "eval_runtime": 19.2845,
413
+ "eval_samples_per_second": 51.855,
414
+ "eval_steps_per_second": 0.83,
415
+ "step": 540
416
+ },
417
+ {
418
+ "epoch": 0.4593929450369155,
419
+ "grad_norm": 0.24671076238155365,
420
+ "learning_rate": 4.7499757310636775e-05,
421
+ "loss": 0.3079,
422
+ "step": 560
423
+ },
424
+ {
425
+ "epoch": 0.4593929450369155,
426
+ "eval_loss": 0.2989128530025482,
427
+ "eval_runtime": 19.2966,
428
+ "eval_samples_per_second": 51.823,
429
+ "eval_steps_per_second": 0.829,
430
+ "step": 560
431
+ },
432
+ {
433
+ "epoch": 0.47579983593109104,
434
+ "grad_norm": 0.28683486580848694,
435
+ "learning_rate": 4.7178484032981076e-05,
436
+ "loss": 0.3031,
437
+ "step": 580
438
+ },
439
+ {
440
+ "epoch": 0.47579983593109104,
441
+ "eval_loss": 0.299227237701416,
442
+ "eval_runtime": 19.3123,
443
+ "eval_samples_per_second": 51.781,
444
+ "eval_steps_per_second": 0.828,
445
+ "step": 580
446
+ },
447
+ {
448
+ "epoch": 0.4922067268252666,
449
+ "grad_norm": 0.298780232667923,
450
+ "learning_rate": 4.683902259952387e-05,
451
+ "loss": 0.3101,
452
+ "step": 600
453
+ },
454
+ {
455
+ "epoch": 0.4922067268252666,
456
+ "eval_loss": 0.29782021045684814,
457
+ "eval_runtime": 19.3297,
458
+ "eval_samples_per_second": 51.734,
459
+ "eval_steps_per_second": 0.828,
460
+ "step": 600
461
+ },
462
+ {
463
+ "epoch": 0.5086136177194421,
464
+ "grad_norm": 0.24450340867042542,
465
+ "learning_rate": 4.6481651396174276e-05,
466
+ "loss": 0.2957,
467
+ "step": 620
468
+ },
469
+ {
470
+ "epoch": 0.5086136177194421,
471
+ "eval_loss": 0.2967863082885742,
472
+ "eval_runtime": 19.3079,
473
+ "eval_samples_per_second": 51.792,
474
+ "eval_steps_per_second": 0.829,
475
+ "step": 620
476
+ },
477
+ {
478
+ "epoch": 0.5250205086136177,
479
+ "grad_norm": 0.2989029884338379,
480
+ "learning_rate": 4.610666349630537e-05,
481
+ "loss": 0.3002,
482
+ "step": 640
483
+ },
484
+ {
485
+ "epoch": 0.5250205086136177,
486
+ "eval_loss": 0.2944621443748474,
487
+ "eval_runtime": 19.3168,
488
+ "eval_samples_per_second": 51.768,
489
+ "eval_steps_per_second": 0.828,
490
+ "step": 640
491
+ },
492
+ {
493
+ "epoch": 0.5414273995077933,
494
+ "grad_norm": 0.2885053753852844,
495
+ "learning_rate": 4.571436642041027e-05,
496
+ "loss": 0.3003,
497
+ "step": 660
498
+ },
499
+ {
500
+ "epoch": 0.5414273995077933,
501
+ "eval_loss": 0.29386863112449646,
502
+ "eval_runtime": 19.3201,
503
+ "eval_samples_per_second": 51.76,
504
+ "eval_steps_per_second": 0.828,
505
+ "step": 660
506
+ },
507
+ {
508
+ "epoch": 0.5578342904019689,
509
+ "grad_norm": 0.2574157416820526,
510
+ "learning_rate": 4.530508188391039e-05,
511
+ "loss": 0.297,
512
+ "step": 680
513
+ },
514
+ {
515
+ "epoch": 0.5578342904019689,
516
+ "eval_loss": 0.29420995712280273,
517
+ "eval_runtime": 19.2774,
518
+ "eval_samples_per_second": 51.874,
519
+ "eval_steps_per_second": 0.83,
520
+ "step": 680
521
+ },
522
+ {
523
+ "epoch": 0.5742411812961444,
524
+ "grad_norm": 0.2774631977081299,
525
+ "learning_rate": 4.48791455333227e-05,
526
+ "loss": 0.2943,
527
+ "step": 700
528
+ },
529
+ {
530
+ "epoch": 0.5742411812961444,
531
+ "eval_loss": 0.2925710380077362,
532
+ "eval_runtime": 19.2886,
533
+ "eval_samples_per_second": 51.844,
534
+ "eval_steps_per_second": 0.83,
535
+ "step": 700
536
+ },
537
+ {
538
+ "epoch": 0.5906480721903199,
539
+ "grad_norm": 0.2841639518737793,
540
+ "learning_rate": 4.44369066710023e-05,
541
+ "loss": 0.2913,
542
+ "step": 720
543
+ },
544
+ {
545
+ "epoch": 0.5906480721903199,
546
+ "eval_loss": 0.29260045289993286,
547
+ "eval_runtime": 19.3101,
548
+ "eval_samples_per_second": 51.786,
549
+ "eval_steps_per_second": 0.829,
550
+ "step": 720
551
+ },
552
+ {
553
+ "epoch": 0.6070549630844955,
554
+ "grad_norm": 0.25801992416381836,
555
+ "learning_rate": 4.397872796868627e-05,
556
+ "loss": 0.296,
557
+ "step": 740
558
+ },
559
+ {
560
+ "epoch": 0.6070549630844955,
561
+ "eval_loss": 0.29102346301078796,
562
+ "eval_runtime": 19.3041,
563
+ "eval_samples_per_second": 51.803,
564
+ "eval_steps_per_second": 0.829,
565
+ "step": 740
566
+ },
567
+ {
568
+ "epoch": 0.623461853978671,
569
+ "grad_norm": 0.2615717351436615,
570
+ "learning_rate": 4.350498517007332e-05,
571
+ "loss": 0.3025,
572
+ "step": 760
573
+ },
574
+ {
575
+ "epoch": 0.623461853978671,
576
+ "eval_loss": 0.29038307070732117,
577
+ "eval_runtime": 19.3107,
578
+ "eval_samples_per_second": 51.785,
579
+ "eval_steps_per_second": 0.829,
580
+ "step": 760
581
+ },
582
+ {
583
+ "epoch": 0.6398687448728466,
584
+ "grad_norm": 0.32346561551094055,
585
+ "learning_rate": 4.301606678268354e-05,
586
+ "loss": 0.2993,
587
+ "step": 780
588
+ },
589
+ {
590
+ "epoch": 0.6398687448728466,
591
+ "eval_loss": 0.289517343044281,
592
+ "eval_runtime": 19.3376,
593
+ "eval_samples_per_second": 51.713,
594
+ "eval_steps_per_second": 0.827,
595
+ "step": 780
596
+ },
597
+ {
598
+ "epoch": 0.6562756357670222,
599
+ "grad_norm": 0.28268206119537354,
600
+ "learning_rate": 4.251237375925071e-05,
601
+ "loss": 0.2991,
602
+ "step": 800
603
+ },
604
+ {
605
+ "epoch": 0.6562756357670222,
606
+ "eval_loss": 0.28874993324279785,
607
+ "eval_runtime": 19.306,
608
+ "eval_samples_per_second": 51.798,
609
+ "eval_steps_per_second": 0.829,
610
+ "step": 800
611
+ },
612
+ {
613
+ "epoch": 0.6726825266611977,
614
+ "grad_norm": 0.28216883540153503,
615
+ "learning_rate": 4.199431916890854e-05,
616
+ "loss": 0.2896,
617
+ "step": 820
618
+ },
619
+ {
620
+ "epoch": 0.6726825266611977,
621
+ "eval_loss": 0.28805309534072876,
622
+ "eval_runtime": 19.3332,
623
+ "eval_samples_per_second": 51.724,
624
+ "eval_steps_per_second": 0.828,
625
+ "step": 820
626
+ },
627
+ {
628
+ "epoch": 0.6890894175553732,
629
+ "grad_norm": 0.2768368422985077,
630
+ "learning_rate": 4.146232785844044e-05,
631
+ "loss": 0.3025,
632
+ "step": 840
633
+ },
634
+ {
635
+ "epoch": 0.6890894175553732,
636
+ "eval_loss": 0.2869604527950287,
637
+ "eval_runtime": 19.3055,
638
+ "eval_samples_per_second": 51.799,
639
+ "eval_steps_per_second": 0.829,
640
+ "step": 840
641
+ },
642
+ {
643
+ "epoch": 0.7054963084495488,
644
+ "grad_norm": 0.27127188444137573,
645
+ "learning_rate": 4.091683610387078e-05,
646
+ "loss": 0.3069,
647
+ "step": 860
648
+ },
649
+ {
650
+ "epoch": 0.7054963084495488,
651
+ "eval_loss": 0.286374568939209,
652
+ "eval_runtime": 19.3049,
653
+ "eval_samples_per_second": 51.8,
654
+ "eval_steps_per_second": 0.829,
655
+ "step": 860
656
+ },
657
+ {
658
+ "epoch": 0.7219031993437244,
659
+ "grad_norm": 0.23589690029621124,
660
+ "learning_rate": 4.035829125268307e-05,
661
+ "loss": 0.2977,
662
+ "step": 880
663
+ },
664
+ {
665
+ "epoch": 0.7219031993437244,
666
+ "eval_loss": 0.2856175899505615,
667
+ "eval_runtime": 19.3199,
668
+ "eval_samples_per_second": 51.76,
669
+ "eval_steps_per_second": 0.828,
670
+ "step": 880
671
+ },
672
+ {
673
+ "epoch": 0.7383100902379,
674
+ "grad_norm": 0.3201812207698822,
675
+ "learning_rate": 3.978715135695881e-05,
676
+ "loss": 0.2904,
677
+ "step": 900
678
+ },
679
+ {
680
+ "epoch": 0.7383100902379,
681
+ "eval_loss": 0.2852262854576111,
682
+ "eval_runtime": 19.3109,
683
+ "eval_samples_per_second": 51.784,
684
+ "eval_steps_per_second": 0.829,
685
+ "step": 900
686
+ },
687
+ {
688
+ "epoch": 0.7547169811320755,
689
+ "grad_norm": 0.2653954327106476,
690
+ "learning_rate": 3.920388479773756e-05,
691
+ "loss": 0.2952,
692
+ "step": 920
693
+ },
694
+ {
695
+ "epoch": 0.7547169811320755,
696
+ "eval_loss": 0.2845345437526703,
697
+ "eval_runtime": 19.3179,
698
+ "eval_samples_per_second": 51.765,
699
+ "eval_steps_per_second": 0.828,
700
+ "step": 920
701
+ },
702
+ {
703
+ "epoch": 0.771123872026251,
704
+ "grad_norm": 0.2647739350795746,
705
+ "learning_rate": 3.86089699009066e-05,
706
+ "loss": 0.2763,
707
+ "step": 940
708
+ },
709
+ {
710
+ "epoch": 0.771123872026251,
711
+ "eval_loss": 0.28321462869644165,
712
+ "eval_runtime": 19.3187,
713
+ "eval_samples_per_second": 51.763,
714
+ "eval_steps_per_second": 0.828,
715
+ "step": 940
716
+ },
717
+ {
718
+ "epoch": 0.7875307629204266,
719
+ "grad_norm": 0.28675034642219543,
720
+ "learning_rate": 3.800289454493481e-05,
721
+ "loss": 0.2859,
722
+ "step": 960
723
+ },
724
+ {
725
+ "epoch": 0.7875307629204266,
726
+ "eval_loss": 0.282760351896286,
727
+ "eval_runtime": 19.3112,
728
+ "eval_samples_per_second": 51.783,
729
+ "eval_steps_per_second": 0.829,
730
+ "step": 960
731
+ },
732
+ {
733
+ "epoch": 0.8039376538146021,
734
+ "grad_norm": 0.26485058665275574,
735
+ "learning_rate": 3.738615576077279e-05,
736
+ "loss": 0.2951,
737
+ "step": 980
738
+ },
739
+ {
740
+ "epoch": 0.8039376538146021,
741
+ "eval_loss": 0.28242456912994385,
742
+ "eval_runtime": 19.3082,
743
+ "eval_samples_per_second": 51.791,
744
+ "eval_steps_per_second": 0.829,
745
+ "step": 980
746
+ },
747
+ {
748
+ "epoch": 0.8203445447087777,
749
+ "grad_norm": 0.28419816493988037,
750
+ "learning_rate": 3.675925932424715e-05,
751
+ "loss": 0.2939,
752
+ "step": 1000
753
+ },
754
+ {
755
+ "epoch": 0.8203445447087777,
756
+ "eval_loss": 0.28157588839530945,
757
+ "eval_runtime": 19.3181,
758
+ "eval_samples_per_second": 51.765,
759
+ "eval_steps_per_second": 0.828,
760
+ "step": 1000
761
+ },
762
+ {
763
+ "epoch": 0.8367514356029533,
764
+ "grad_norm": 0.2731083929538727,
765
+ "learning_rate": 3.612271934128337e-05,
766
+ "loss": 0.2875,
767
+ "step": 1020
768
+ },
769
+ {
770
+ "epoch": 0.8367514356029533,
771
+ "eval_loss": 0.28066861629486084,
772
+ "eval_runtime": 19.2504,
773
+ "eval_samples_per_second": 51.947,
774
+ "eval_steps_per_second": 0.831,
775
+ "step": 1020
776
+ },
777
+ {
778
+ "epoch": 0.8531583264971287,
779
+ "grad_norm": 0.2515132427215576,
780
+ "learning_rate": 3.547705782629717e-05,
781
+ "loss": 0.2819,
782
+ "step": 1040
783
+ },
784
+ {
785
+ "epoch": 0.8531583264971287,
786
+ "eval_loss": 0.2809597849845886,
787
+ "eval_runtime": 19.3064,
788
+ "eval_samples_per_second": 51.796,
789
+ "eval_steps_per_second": 0.829,
790
+ "step": 1040
791
+ },
792
+ {
793
+ "epoch": 0.8695652173913043,
794
+ "grad_norm": 0.2833218276500702,
795
+ "learning_rate": 3.4822804274100355e-05,
796
+ "loss": 0.2859,
797
+ "step": 1060
798
+ },
799
+ {
800
+ "epoch": 0.8695652173913043,
801
+ "eval_loss": 0.28050196170806885,
802
+ "eval_runtime": 19.3166,
803
+ "eval_samples_per_second": 51.769,
804
+ "eval_steps_per_second": 0.828,
805
+ "step": 1060
806
+ },
807
+ {
808
+ "epoch": 0.8859721082854799,
809
+ "grad_norm": 0.24503149092197418,
810
+ "learning_rate": 3.416049522567218e-05,
811
+ "loss": 0.2953,
812
+ "step": 1080
813
+ },
814
+ {
815
+ "epoch": 0.8859721082854799,
816
+ "eval_loss": 0.2801743745803833,
817
+ "eval_runtime": 19.3198,
818
+ "eval_samples_per_second": 51.76,
819
+ "eval_steps_per_second": 0.828,
820
+ "step": 1080
821
+ },
822
+ {
823
+ "epoch": 0.9023789991796555,
824
+ "grad_norm": 0.26533713936805725,
825
+ "learning_rate": 3.349067382815217e-05,
826
+ "loss": 0.2834,
827
+ "step": 1100
828
+ },
829
+ {
830
+ "epoch": 0.9023789991796555,
831
+ "eval_loss": 0.2789291739463806,
832
+ "eval_runtime": 19.3156,
833
+ "eval_samples_per_second": 51.772,
834
+ "eval_steps_per_second": 0.828,
835
+ "step": 1100
836
+ },
837
+ {
838
+ "epoch": 0.918785890073831,
839
+ "grad_norm": 0.2972276508808136,
840
+ "learning_rate": 3.281388938941543e-05,
841
+ "loss": 0.2828,
842
+ "step": 1120
843
+ },
844
+ {
845
+ "epoch": 0.918785890073831,
846
+ "eval_loss": 0.27879536151885986,
847
+ "eval_runtime": 19.2998,
848
+ "eval_samples_per_second": 51.814,
849
+ "eval_steps_per_second": 0.829,
850
+ "step": 1120
851
+ },
852
+ {
853
+ "epoch": 0.9351927809680065,
854
+ "grad_norm": 0.2797587215900421,
855
+ "learning_rate": 3.213069692759563e-05,
856
+ "loss": 0.2904,
857
+ "step": 1140
858
+ },
859
+ {
860
+ "epoch": 0.9351927809680065,
861
+ "eval_loss": 0.27857306599617004,
862
+ "eval_runtime": 19.3281,
863
+ "eval_samples_per_second": 51.738,
864
+ "eval_steps_per_second": 0.828,
865
+ "step": 1140
866
+ },
867
+ {
868
+ "epoch": 0.9515996718621821,
869
+ "grad_norm": 0.27988341450691223,
870
+ "learning_rate": 3.1441656715925136e-05,
871
+ "loss": 0.302,
872
+ "step": 1160
873
+ },
874
+ {
875
+ "epoch": 0.9515996718621821,
876
+ "eval_loss": 0.27816376090049744,
877
+ "eval_runtime": 19.3057,
878
+ "eval_samples_per_second": 51.798,
879
+ "eval_steps_per_second": 0.829,
880
+ "step": 1160
881
+ },
882
+ {
883
+ "epoch": 0.9680065627563577,
884
+ "grad_norm": 0.3127523362636566,
885
+ "learning_rate": 3.07473338232655e-05,
886
+ "loss": 0.2814,
887
+ "step": 1180
888
+ },
889
+ {
890
+ "epoch": 0.9680065627563577,
891
+ "eval_loss": 0.27769869565963745,
892
+ "eval_runtime": 19.3204,
893
+ "eval_samples_per_second": 51.759,
894
+ "eval_steps_per_second": 0.828,
895
+ "step": 1180
896
+ },
897
+ {
898
+ "epoch": 0.9844134536505332,
899
+ "grad_norm": 0.25187912583351135,
900
+ "learning_rate": 3.004829765070516e-05,
901
+ "loss": 0.2905,
902
+ "step": 1200
903
+ },
904
+ {
905
+ "epoch": 0.9844134536505332,
906
+ "eval_loss": 0.276981383562088,
907
+ "eval_runtime": 19.3121,
908
+ "eval_samples_per_second": 51.781,
909
+ "eval_steps_per_second": 0.828,
910
+ "step": 1200
911
+ },
912
+ {
913
+ "epoch": 1.0008203445447088,
914
+ "grad_norm": 0.27703630924224854,
915
+ "learning_rate": 2.9345121464604353e-05,
916
+ "loss": 0.2958,
917
+ "step": 1220
918
+ },
919
+ {
920
+ "epoch": 1.0008203445447088,
921
+ "eval_loss": 0.2769216299057007,
922
+ "eval_runtime": 19.2849,
923
+ "eval_samples_per_second": 51.854,
924
+ "eval_steps_per_second": 0.83,
925
+ "step": 1220
926
+ },
927
+ {
928
+ "epoch": 1.0172272354388843,
929
+ "grad_norm": 0.2569476068019867,
930
+ "learning_rate": 2.8638381926470236e-05,
931
+ "loss": 0.2704,
932
+ "step": 1240
933
+ },
934
+ {
935
+ "epoch": 1.0172272354388843,
936
+ "eval_loss": 0.27616381645202637,
937
+ "eval_runtime": 19.3362,
938
+ "eval_samples_per_second": 51.716,
939
+ "eval_steps_per_second": 0.827,
940
+ "step": 1240
941
+ },
942
+ {
943
+ "epoch": 1.03363412633306,
944
+ "grad_norm": 0.2954521179199219,
945
+ "learning_rate": 2.7928658620047647e-05,
946
+ "loss": 0.2816,
947
+ "step": 1260
948
+ },
949
+ {
950
+ "epoch": 1.03363412633306,
951
+ "eval_loss": 0.2756293714046478,
952
+ "eval_runtime": 19.2869,
953
+ "eval_samples_per_second": 51.849,
954
+ "eval_steps_per_second": 0.83,
955
+ "step": 1260
956
+ },
957
+ {
958
+ "epoch": 1.0500410172272354,
959
+ "grad_norm": 0.31093183159828186,
960
+ "learning_rate": 2.7216533576013426e-05,
961
+ "loss": 0.278,
962
+ "step": 1280
963
+ },
964
+ {
965
+ "epoch": 1.0500410172272354,
966
+ "eval_loss": 0.2751014530658722,
967
+ "eval_runtime": 19.292,
968
+ "eval_samples_per_second": 51.835,
969
+ "eval_steps_per_second": 0.829,
970
+ "step": 1280
971
+ },
972
+ {
973
+ "epoch": 1.066447908121411,
974
+ "grad_norm": 0.2663869261741638,
975
+ "learning_rate": 2.6502590794664073e-05,
976
+ "loss": 0.2742,
977
+ "step": 1300
978
+ },
979
+ {
980
+ "epoch": 1.066447908121411,
981
+ "eval_loss": 0.27440693974494934,
982
+ "eval_runtime": 19.3285,
983
+ "eval_samples_per_second": 51.737,
984
+ "eval_steps_per_second": 0.828,
985
+ "step": 1300
986
+ },
987
+ {
988
+ "epoch": 1.0828547990155866,
989
+ "grad_norm": 0.3329567015171051,
990
+ "learning_rate": 2.5787415766988128e-05,
991
+ "loss": 0.2691,
992
+ "step": 1320
993
+ },
994
+ {
995
+ "epoch": 1.0828547990155866,
996
+ "eval_loss": 0.27447032928466797,
997
+ "eval_runtime": 19.3182,
998
+ "eval_samples_per_second": 51.765,
999
+ "eval_steps_per_second": 0.828,
1000
+ "step": 1320
1001
+ },
1002
+ {
1003
+ "epoch": 1.099261689909762,
1004
+ "grad_norm": 0.2891082167625427,
1005
+ "learning_rate": 2.5071594994516033e-05,
1006
+ "loss": 0.2804,
1007
+ "step": 1340
1008
+ },
1009
+ {
1010
+ "epoch": 1.099261689909762,
1011
+ "eval_loss": 0.2739960849285126,
1012
+ "eval_runtime": 19.3331,
1013
+ "eval_samples_per_second": 51.725,
1014
+ "eval_steps_per_second": 0.828,
1015
+ "step": 1340
1016
+ },
1017
+ {
1018
+ "epoch": 1.1156685808039377,
1019
+ "grad_norm": 0.308324933052063,
1020
+ "learning_rate": 2.4355715508341378e-05,
1021
+ "loss": 0.2813,
1022
+ "step": 1360
1023
+ },
1024
+ {
1025
+ "epoch": 1.1156685808039377,
1026
+ "eval_loss": 0.27399778366088867,
1027
+ "eval_runtime": 19.3455,
1028
+ "eval_samples_per_second": 51.692,
1029
+ "eval_steps_per_second": 0.827,
1030
+ "step": 1360
1031
+ },
1032
+ {
1033
+ "epoch": 1.1320754716981132,
1034
+ "grad_norm": 0.29660311341285706,
1035
+ "learning_rate": 2.3640364387707725e-05,
1036
+ "loss": 0.2742,
1037
+ "step": 1380
1038
+ },
1039
+ {
1040
+ "epoch": 1.1320754716981132,
1041
+ "eval_loss": 0.27386149764060974,
1042
+ "eval_runtime": 19.326,
1043
+ "eval_samples_per_second": 51.744,
1044
+ "eval_steps_per_second": 0.828,
1045
+ "step": 1380
1046
+ },
1047
+ {
1048
+ "epoch": 1.1484823625922886,
1049
+ "grad_norm": 0.27859780192375183,
1050
+ "learning_rate": 2.2926128278556052e-05,
1051
+ "loss": 0.2843,
1052
+ "step": 1400
1053
+ },
1054
+ {
1055
+ "epoch": 1.1484823625922886,
1056
+ "eval_loss": 0.27325138449668884,
1057
+ "eval_runtime": 19.3447,
1058
+ "eval_samples_per_second": 51.694,
1059
+ "eval_steps_per_second": 0.827,
1060
+ "step": 1400
1061
+ },
1062
+ {
1063
+ "epoch": 1.1648892534864643,
1064
+ "grad_norm": 0.25910985469818115,
1065
+ "learning_rate": 2.2213592912427474e-05,
1066
+ "loss": 0.2722,
1067
+ "step": 1420
1068
+ },
1069
+ {
1070
+ "epoch": 1.1648892534864643,
1071
+ "eval_loss": 0.2735229730606079,
1072
+ "eval_runtime": 19.3303,
1073
+ "eval_samples_per_second": 51.732,
1074
+ "eval_steps_per_second": 0.828,
1075
+ "step": 1420
1076
+ },
1077
+ {
1078
+ "epoch": 1.1812961443806398,
1079
+ "grad_norm": 0.2874094247817993,
1080
+ "learning_rate": 2.150334262611589e-05,
1081
+ "loss": 0.263,
1082
+ "step": 1440
1083
+ },
1084
+ {
1085
+ "epoch": 1.1812961443806398,
1086
+ "eval_loss": 0.2728740870952606,
1087
+ "eval_runtime": 19.3045,
1088
+ "eval_samples_per_second": 51.801,
1089
+ "eval_steps_per_second": 0.829,
1090
+ "step": 1440
1091
+ },
1092
+ {
1093
+ "epoch": 1.1977030352748155,
1094
+ "grad_norm": 0.27428507804870605,
1095
+ "learning_rate": 2.07959598824644e-05,
1096
+ "loss": 0.2819,
1097
+ "step": 1460
1098
+ },
1099
+ {
1100
+ "epoch": 1.1977030352748155,
1101
+ "eval_loss": 0.2725715637207031,
1102
+ "eval_runtime": 19.3247,
1103
+ "eval_samples_per_second": 51.747,
1104
+ "eval_steps_per_second": 0.828,
1105
+ "step": 1460
1106
+ },
1107
+ {
1108
+ "epoch": 1.214109926168991,
1109
+ "grad_norm": 0.2841712236404419,
1110
+ "learning_rate": 2.0092024792698527e-05,
1111
+ "loss": 0.2809,
1112
+ "step": 1480
1113
+ },
1114
+ {
1115
+ "epoch": 1.214109926168991,
1116
+ "eval_loss": 0.27222681045532227,
1117
+ "eval_runtime": 19.3107,
1118
+ "eval_samples_per_second": 51.785,
1119
+ "eval_steps_per_second": 0.829,
1120
+ "step": 1480
1121
+ },
1122
+ {
1123
+ "epoch": 1.2305168170631666,
1124
+ "grad_norm": 0.32081517577171326,
1125
+ "learning_rate": 1.9392114640687985e-05,
1126
+ "loss": 0.2776,
1127
+ "step": 1500
1128
+ },
1129
+ {
1130
+ "epoch": 1.2305168170631666,
1131
+ "eval_loss": 0.27228766679763794,
1132
+ "eval_runtime": 19.3356,
1133
+ "eval_samples_per_second": 51.718,
1134
+ "eval_steps_per_second": 0.827,
1135
+ "step": 1500
1136
+ },
1137
+ {
1138
+ "epoch": 1.246923707957342,
1139
+ "grad_norm": 0.31806743144989014,
1140
+ "learning_rate": 1.8696803409527062e-05,
1141
+ "loss": 0.281,
1142
+ "step": 1520
1143
+ },
1144
+ {
1145
+ "epoch": 1.246923707957342,
1146
+ "eval_loss": 0.2718179225921631,
1147
+ "eval_runtime": 19.3051,
1148
+ "eval_samples_per_second": 51.8,
1149
+ "eval_steps_per_second": 0.829,
1150
+ "step": 1520
1151
+ },
1152
+ {
1153
+ "epoch": 1.2633305988515175,
1154
+ "grad_norm": 0.267499178647995,
1155
+ "learning_rate": 1.8006661310821926e-05,
1156
+ "loss": 0.269,
1157
+ "step": 1540
1158
+ },
1159
+ {
1160
+ "epoch": 1.2633305988515175,
1161
+ "eval_loss": 0.2716826796531677,
1162
+ "eval_runtime": 19.2781,
1163
+ "eval_samples_per_second": 51.872,
1164
+ "eval_steps_per_second": 0.83,
1165
+ "step": 1540
1166
+ },
1167
+ {
1168
+ "epoch": 1.2797374897456932,
1169
+ "grad_norm": 0.2933109700679779,
1170
+ "learning_rate": 1.7322254317070876e-05,
1171
+ "loss": 0.2781,
1172
+ "step": 1560
1173
+ },
1174
+ {
1175
+ "epoch": 1.2797374897456932,
1176
+ "eval_loss": 0.27161359786987305,
1177
+ "eval_runtime": 19.3284,
1178
+ "eval_samples_per_second": 51.737,
1179
+ "eval_steps_per_second": 0.828,
1180
+ "step": 1560
1181
+ },
1182
+ {
1183
+ "epoch": 1.2961443806398687,
1184
+ "grad_norm": 0.3219626247882843,
1185
+ "learning_rate": 1.664414369752095e-05,
1186
+ "loss": 0.2742,
1187
+ "step": 1580
1188
+ },
1189
+ {
1190
+ "epoch": 1.2961443806398687,
1191
+ "eval_loss": 0.2712157070636749,
1192
+ "eval_runtime": 19.3194,
1193
+ "eval_samples_per_second": 51.761,
1194
+ "eval_steps_per_second": 0.828,
1195
+ "step": 1580
1196
+ },
1197
+ {
1198
+ "epoch": 1.3125512715340442,
1199
+ "grad_norm": 0.2988798916339874,
1200
+ "learning_rate": 1.5972885557881666e-05,
1201
+ "loss": 0.2736,
1202
+ "step": 1600
1203
+ },
1204
+ {
1205
+ "epoch": 1.3125512715340442,
1206
+ "eval_loss": 0.27080392837524414,
1207
+ "eval_runtime": 19.2959,
1208
+ "eval_samples_per_second": 51.824,
1209
+ "eval_steps_per_second": 0.829,
1210
+ "step": 1600
1211
+ },
1212
+ {
1213
+ "epoch": 1.3289581624282198,
1214
+ "grad_norm": 0.2971179783344269,
1215
+ "learning_rate": 1.530903038427322e-05,
1216
+ "loss": 0.2709,
1217
+ "step": 1620
1218
+ },
1219
+ {
1220
+ "epoch": 1.3289581624282198,
1221
+ "eval_loss": 0.2705814838409424,
1222
+ "eval_runtime": 19.29,
1223
+ "eval_samples_per_second": 51.84,
1224
+ "eval_steps_per_second": 0.829,
1225
+ "step": 1620
1226
+ },
1227
+ {
1228
+ "epoch": 1.3453650533223955,
1229
+ "grad_norm": 0.3339160680770874,
1230
+ "learning_rate": 1.4653122591783164e-05,
1231
+ "loss": 0.2738,
1232
+ "step": 1640
1233
+ },
1234
+ {
1235
+ "epoch": 1.3453650533223955,
1236
+ "eval_loss": 0.27005913853645325,
1237
+ "eval_runtime": 19.2838,
1238
+ "eval_samples_per_second": 51.857,
1239
+ "eval_steps_per_second": 0.83,
1240
+ "step": 1640
1241
+ },
1242
+ {
1243
+ "epoch": 1.361771944216571,
1244
+ "grad_norm": 0.3250068128108978,
1245
+ "learning_rate": 1.400570007800196e-05,
1246
+ "loss": 0.2621,
1247
+ "step": 1660
1248
+ },
1249
+ {
1250
+ "epoch": 1.361771944216571,
1251
+ "eval_loss": 0.269867479801178,
1252
+ "eval_runtime": 19.3111,
1253
+ "eval_samples_per_second": 51.784,
1254
+ "eval_steps_per_second": 0.829,
1255
+ "step": 1660
1256
+ },
1257
+ {
1258
+ "epoch": 1.3781788351107465,
1259
+ "grad_norm": 0.28547146916389465,
1260
+ "learning_rate": 1.3367293781903306e-05,
1261
+ "loss": 0.2674,
1262
+ "step": 1680
1263
+ },
1264
+ {
1265
+ "epoch": 1.3781788351107465,
1266
+ "eval_loss": 0.2695818245410919,
1267
+ "eval_runtime": 19.308,
1268
+ "eval_samples_per_second": 51.792,
1269
+ "eval_steps_per_second": 0.829,
1270
+ "step": 1680
1271
+ },
1272
+ {
1273
+ "epoch": 1.3945857260049221,
1274
+ "grad_norm": 0.29051050543785095,
1275
+ "learning_rate": 1.2738427248431028e-05,
1276
+ "loss": 0.2729,
1277
+ "step": 1700
1278
+ },
1279
+ {
1280
+ "epoch": 1.3945857260049221,
1281
+ "eval_loss": 0.2692498564720154,
1282
+ "eval_runtime": 19.3168,
1283
+ "eval_samples_per_second": 51.768,
1284
+ "eval_steps_per_second": 0.828,
1285
+ "step": 1700
1286
+ },
1287
+ {
1288
+ "epoch": 1.4109926168990976,
1289
+ "grad_norm": 0.37129437923431396,
1290
+ "learning_rate": 1.2119616199149903e-05,
1291
+ "loss": 0.2848,
1292
+ "step": 1720
1293
+ },
1294
+ {
1295
+ "epoch": 1.4109926168990976,
1296
+ "eval_loss": 0.2692253291606903,
1297
+ "eval_runtime": 19.3149,
1298
+ "eval_samples_per_second": 51.773,
1299
+ "eval_steps_per_second": 0.828,
1300
+ "step": 1720
1301
+ },
1302
+ {
1303
+ "epoch": 1.427399507793273,
1304
+ "grad_norm": 0.29549741744995117,
1305
+ "learning_rate": 1.1511368109312023e-05,
1306
+ "loss": 0.284,
1307
+ "step": 1740
1308
+ },
1309
+ {
1310
+ "epoch": 1.427399507793273,
1311
+ "eval_loss": 0.2688217759132385,
1312
+ "eval_runtime": 19.3124,
1313
+ "eval_samples_per_second": 51.78,
1314
+ "eval_steps_per_second": 0.828,
1315
+ "step": 1740
1316
+ },
1317
+ {
1318
+ "epoch": 1.4438063986874488,
1319
+ "grad_norm": 0.29102256894111633,
1320
+ "learning_rate": 1.0914181791685988e-05,
1321
+ "loss": 0.2766,
1322
+ "step": 1760
1323
+ },
1324
+ {
1325
+ "epoch": 1.4438063986874488,
1326
+ "eval_loss": 0.2683846354484558,
1327
+ "eval_runtime": 19.2865,
1328
+ "eval_samples_per_second": 51.85,
1329
+ "eval_steps_per_second": 0.83,
1330
+ "step": 1760
1331
+ },
1332
+ {
1333
+ "epoch": 1.4602132895816242,
1334
+ "grad_norm": 0.325428307056427,
1335
+ "learning_rate": 1.0328546987489999e-05,
1336
+ "loss": 0.2693,
1337
+ "step": 1780
1338
+ },
1339
+ {
1340
+ "epoch": 1.4602132895816242,
1341
+ "eval_loss": 0.26858580112457275,
1342
+ "eval_runtime": 19.3168,
1343
+ "eval_samples_per_second": 51.768,
1344
+ "eval_steps_per_second": 0.828,
1345
+ "step": 1780
1346
+ },
1347
+ {
1348
+ "epoch": 1.4766201804758,
1349
+ "grad_norm": 0.31247812509536743,
1350
+ "learning_rate": 9.75494396476423e-06,
1351
+ "loss": 0.281,
1352
+ "step": 1800
1353
+ },
1354
+ {
1355
+ "epoch": 1.4766201804758,
1356
+ "eval_loss": 0.2685295641422272,
1357
+ "eval_runtime": 19.3134,
1358
+ "eval_samples_per_second": 51.777,
1359
+ "eval_steps_per_second": 0.828,
1360
+ "step": 1800
1361
+ },
1362
+ {
1363
+ "epoch": 1.4930270713699754,
1364
+ "grad_norm": 0.30578482151031494,
1365
+ "learning_rate": 9.193843124512228e-06,
1366
+ "loss": 0.2652,
1367
+ "step": 1820
1368
+ },
1369
+ {
1370
+ "epoch": 1.4930270713699754,
1371
+ "eval_loss": 0.26810023188591003,
1372
+ "eval_runtime": 19.3179,
1373
+ "eval_samples_per_second": 51.765,
1374
+ "eval_steps_per_second": 0.828,
1375
+ "step": 1820
1376
+ },
1377
+ {
1378
+ "epoch": 1.509433962264151,
1379
+ "grad_norm": 0.29188328981399536,
1380
+ "learning_rate": 8.64570461493382e-06,
1381
+ "loss": 0.2787,
1382
+ "step": 1840
1383
+ },
1384
+ {
1385
+ "epoch": 1.509433962264151,
1386
+ "eval_loss": 0.26806536316871643,
1387
+ "eval_runtime": 19.3129,
1388
+ "eval_samples_per_second": 51.779,
1389
+ "eval_steps_per_second": 0.828,
1390
+ "step": 1840
1391
+ },
1392
+ {
1393
+ "epoch": 1.5258408531583265,
1394
+ "grad_norm": 0.28083163499832153,
1395
+ "learning_rate": 8.110977954066303e-06,
1396
+ "loss": 0.2745,
1397
+ "step": 1860
1398
+ },
1399
+ {
1400
+ "epoch": 1.5258408531583265,
1401
+ "eval_loss": 0.2678322494029999,
1402
+ "eval_runtime": 19.305,
1403
+ "eval_samples_per_second": 51.8,
1404
+ "eval_steps_per_second": 0.829,
1405
+ "step": 1860
1406
+ },
1407
+ {
1408
+ "epoch": 1.542247744052502,
1409
+ "grad_norm": 0.325198769569397,
1410
+ "learning_rate": 7.590101661143234e-06,
1411
+ "loss": 0.2681,
1412
+ "step": 1880
1413
+ },
1414
+ {
1415
+ "epoch": 1.542247744052502,
1416
+ "eval_loss": 0.26767855882644653,
1417
+ "eval_runtime": 19.3108,
1418
+ "eval_samples_per_second": 51.784,
1419
+ "eval_steps_per_second": 0.829,
1420
+ "step": 1880
1421
+ },
1422
+ {
1423
+ "epoch": 1.5586546349466777,
1424
+ "grad_norm": 0.2920818328857422,
1425
+ "learning_rate": 7.0835028969730185e-06,
1426
+ "loss": 0.271,
1427
+ "step": 1900
1428
+ },
1429
+ {
1430
+ "epoch": 1.5586546349466777,
1431
+ "eval_loss": 0.2676710784435272,
1432
+ "eval_runtime": 19.306,
1433
+ "eval_samples_per_second": 51.797,
1434
+ "eval_steps_per_second": 0.829,
1435
+ "step": 1900
1436
+ },
1437
+ {
1438
+ "epoch": 1.5750615258408531,
1439
+ "grad_norm": 0.2953170835971832,
1440
+ "learning_rate": 6.591597113632447e-06,
1441
+ "loss": 0.2751,
1442
+ "step": 1920
1443
+ },
1444
+ {
1445
+ "epoch": 1.5750615258408531,
1446
+ "eval_loss": 0.2673129141330719,
1447
+ "eval_runtime": 19.3224,
1448
+ "eval_samples_per_second": 51.753,
1449
+ "eval_steps_per_second": 0.828,
1450
+ "step": 1920
1451
+ },
1452
+ {
1453
+ "epoch": 1.5914684167350286,
1454
+ "grad_norm": 0.306307315826416,
1455
+ "learning_rate": 6.114787713762287e-06,
1456
+ "loss": 0.2748,
1457
+ "step": 1940
1458
+ },
1459
+ {
1460
+ "epoch": 1.5914684167350286,
1461
+ "eval_loss": 0.26730725169181824,
1462
+ "eval_runtime": 19.3249,
1463
+ "eval_samples_per_second": 51.747,
1464
+ "eval_steps_per_second": 0.828,
1465
+ "step": 1940
1466
+ },
1467
+ {
1468
+ "epoch": 1.6078753076292043,
1469
+ "grad_norm": 0.31414565443992615,
1470
+ "learning_rate": 5.653465719744389e-06,
1471
+ "loss": 0.2746,
1472
+ "step": 1960
1473
+ },
1474
+ {
1475
+ "epoch": 1.6078753076292043,
1476
+ "eval_loss": 0.2670857608318329,
1477
+ "eval_runtime": 19.3144,
1478
+ "eval_samples_per_second": 51.775,
1479
+ "eval_steps_per_second": 0.828,
1480
+ "step": 1960
1481
+ },
1482
+ {
1483
+ "epoch": 1.62428219852338,
1484
+ "grad_norm": 0.3204467296600342,
1485
+ "learning_rate": 5.2080094530316174e-06,
1486
+ "loss": 0.2739,
1487
+ "step": 1980
1488
+ },
1489
+ {
1490
+ "epoch": 1.62428219852338,
1491
+ "eval_loss": 0.26702049374580383,
1492
+ "eval_runtime": 19.2991,
1493
+ "eval_samples_per_second": 51.816,
1494
+ "eval_steps_per_second": 0.829,
1495
+ "step": 1980
1496
+ },
1497
+ {
1498
+ "epoch": 1.6406890894175554,
1499
+ "grad_norm": 0.33073121309280396,
1500
+ "learning_rate": 4.778784223893601e-06,
1501
+ "loss": 0.2767,
1502
+ "step": 2000
1503
+ },
1504
+ {
1505
+ "epoch": 1.6406890894175554,
1506
+ "eval_loss": 0.2670028805732727,
1507
+ "eval_runtime": 19.3233,
1508
+ "eval_samples_per_second": 51.751,
1509
+ "eval_steps_per_second": 0.828,
1510
+ "step": 2000
1511
+ },
1512
+ {
1513
+ "epoch": 1.6570959803117309,
1514
+ "grad_norm": 0.3207637071609497,
1515
+ "learning_rate": 4.36614203183264e-06,
1516
+ "loss": 0.2638,
1517
+ "step": 2020
1518
+ },
1519
+ {
1520
+ "epoch": 1.6570959803117309,
1521
+ "eval_loss": 0.2668777406215668,
1522
+ "eval_runtime": 19.3222,
1523
+ "eval_samples_per_second": 51.754,
1524
+ "eval_steps_per_second": 0.828,
1525
+ "step": 2020
1526
+ },
1527
+ {
1528
+ "epoch": 1.6735028712059066,
1529
+ "grad_norm": 0.2821946442127228,
1530
+ "learning_rate": 3.970421276915584e-06,
1531
+ "loss": 0.2646,
1532
+ "step": 2040
1533
+ },
1534
+ {
1535
+ "epoch": 1.6735028712059066,
1536
+ "eval_loss": 0.2667301893234253,
1537
+ "eval_runtime": 19.2968,
1538
+ "eval_samples_per_second": 51.822,
1539
+ "eval_steps_per_second": 0.829,
1540
+ "step": 2040
1541
+ },
1542
+ {
1543
+ "epoch": 1.689909762100082,
1544
+ "grad_norm": 0.34934374690055847,
1545
+ "learning_rate": 3.5919464822583486e-06,
1546
+ "loss": 0.2797,
1547
+ "step": 2060
1548
+ },
1549
+ {
1550
+ "epoch": 1.689909762100082,
1551
+ "eval_loss": 0.2666707932949066,
1552
+ "eval_runtime": 19.2974,
1553
+ "eval_samples_per_second": 51.82,
1554
+ "eval_steps_per_second": 0.829,
1555
+ "step": 2060
1556
+ },
1557
+ {
1558
+ "epoch": 1.7063166529942575,
1559
+ "grad_norm": 0.31770169734954834,
1560
+ "learning_rate": 3.2310280278905787e-06,
1561
+ "loss": 0.268,
1562
+ "step": 2080
1563
+ },
1564
+ {
1565
+ "epoch": 1.7063166529942575,
1566
+ "eval_loss": 0.266598641872406,
1567
+ "eval_runtime": 19.3129,
1568
+ "eval_samples_per_second": 51.779,
1569
+ "eval_steps_per_second": 0.828,
1570
+ "step": 2080
1571
+ },
1572
+ {
1573
+ "epoch": 1.7227235438884332,
1574
+ "grad_norm": 0.3003862798213959,
1575
+ "learning_rate": 2.8879618962189326e-06,
1576
+ "loss": 0.2659,
1577
+ "step": 2100
1578
+ },
1579
+ {
1580
+ "epoch": 1.7227235438884332,
1581
+ "eval_loss": 0.2664910852909088,
1582
+ "eval_runtime": 19.3078,
1583
+ "eval_samples_per_second": 51.792,
1584
+ "eval_steps_per_second": 0.829,
1585
+ "step": 2100
1586
+ },
1587
+ {
1588
+ "epoch": 1.7391304347826086,
1589
+ "grad_norm": 0.30080583691596985,
1590
+ "learning_rate": 2.563029429297428e-06,
1591
+ "loss": 0.265,
1592
+ "step": 2120
1593
+ },
1594
+ {
1595
+ "epoch": 1.7391304347826086,
1596
+ "eval_loss": 0.26655587553977966,
1597
+ "eval_runtime": 19.3166,
1598
+ "eval_samples_per_second": 51.769,
1599
+ "eval_steps_per_second": 0.828,
1600
+ "step": 2120
1601
+ },
1602
+ {
1603
+ "epoch": 1.755537325676784,
1604
+ "grad_norm": 0.3040551543235779,
1605
+ "learning_rate": 2.2564970981041793e-06,
1606
+ "loss": 0.2621,
1607
+ "step": 2140
1608
+ },
1609
+ {
1610
+ "epoch": 1.755537325676784,
1611
+ "eval_loss": 0.26629871129989624,
1612
+ "eval_runtime": 19.3143,
1613
+ "eval_samples_per_second": 51.775,
1614
+ "eval_steps_per_second": 0.828,
1615
+ "step": 2140
1616
+ },
1617
+ {
1618
+ "epoch": 1.7719442165709598,
1619
+ "grad_norm": 0.31465449929237366,
1620
+ "learning_rate": 1.968616284013583e-06,
1621
+ "loss": 0.2644,
1622
+ "step": 2160
1623
+ },
1624
+ {
1625
+ "epoch": 1.7719442165709598,
1626
+ "eval_loss": 0.26641103625297546,
1627
+ "eval_runtime": 19.3318,
1628
+ "eval_samples_per_second": 51.728,
1629
+ "eval_steps_per_second": 0.828,
1630
+ "step": 2160
1631
+ },
1632
+ {
1633
+ "epoch": 1.7883511074651355,
1634
+ "grad_norm": 0.2679118812084198,
1635
+ "learning_rate": 1.6996230726431528e-06,
1636
+ "loss": 0.2663,
1637
+ "step": 2180
1638
+ },
1639
+ {
1640
+ "epoch": 1.7883511074651355,
1641
+ "eval_loss": 0.266239196062088,
1642
+ "eval_runtime": 19.3186,
1643
+ "eval_samples_per_second": 51.764,
1644
+ "eval_steps_per_second": 0.828,
1645
+ "step": 2180
1646
+ },
1647
+ {
1648
+ "epoch": 1.804757998359311,
1649
+ "grad_norm": 0.33543169498443604,
1650
+ "learning_rate": 1.4497380602442378e-06,
1651
+ "loss": 0.2728,
1652
+ "step": 2200
1653
+ },
1654
+ {
1655
+ "epoch": 1.804757998359311,
1656
+ "eval_loss": 0.2662188708782196,
1657
+ "eval_runtime": 19.3307,
1658
+ "eval_samples_per_second": 51.731,
1659
+ "eval_steps_per_second": 0.828,
1660
+ "step": 2200
1661
+ },
1662
+ {
1663
+ "epoch": 1.8211648892534864,
1664
+ "grad_norm": 0.32849323749542236,
1665
+ "learning_rate": 1.2191661727951499e-06,
1666
+ "loss": 0.2598,
1667
+ "step": 2220
1668
+ },
1669
+ {
1670
+ "epoch": 1.8211648892534864,
1671
+ "eval_loss": 0.26628053188323975,
1672
+ "eval_runtime": 19.3167,
1673
+ "eval_samples_per_second": 51.769,
1674
+ "eval_steps_per_second": 0.828,
1675
+ "step": 2220
1676
+ },
1677
+ {
1678
+ "epoch": 1.837571780147662,
1679
+ "grad_norm": 0.32713988423347473,
1680
+ "learning_rate": 1.0080964979452572e-06,
1681
+ "loss": 0.2714,
1682
+ "step": 2240
1683
+ },
1684
+ {
1685
+ "epoch": 1.837571780147662,
1686
+ "eval_loss": 0.26624441146850586,
1687
+ "eval_runtime": 19.2915,
1688
+ "eval_samples_per_second": 51.836,
1689
+ "eval_steps_per_second": 0.829,
1690
+ "step": 2240
1691
+ },
1692
+ {
1693
+ "epoch": 1.8539786710418376,
1694
+ "grad_norm": 0.29280194640159607,
1695
+ "learning_rate": 8.167021299477689e-07,
1696
+ "loss": 0.2645,
1697
+ "step": 2260
1698
+ },
1699
+ {
1700
+ "epoch": 1.8539786710418376,
1701
+ "eval_loss": 0.2662700116634369,
1702
+ "eval_runtime": 19.3179,
1703
+ "eval_samples_per_second": 51.765,
1704
+ "eval_steps_per_second": 0.828,
1705
+ "step": 2260
1706
+ },
1707
+ {
1708
+ "epoch": 1.870385561936013,
1709
+ "grad_norm": 0.30739790201187134,
1710
+ "learning_rate": 6.451400277084324e-07,
1711
+ "loss": 0.2707,
1712
+ "step": 2280
1713
+ },
1714
+ {
1715
+ "epoch": 1.870385561936013,
1716
+ "eval_loss": 0.2661494314670563,
1717
+ "eval_runtime": 19.3129,
1718
+ "eval_samples_per_second": 51.779,
1719
+ "eval_steps_per_second": 0.828,
1720
+ "step": 2280
1721
+ },
1722
+ {
1723
+ "epoch": 1.8867924528301887,
1724
+ "grad_norm": 0.35433024168014526,
1725
+ "learning_rate": 4.935508860664601e-07,
1726
+ "loss": 0.2803,
1727
+ "step": 2300
1728
+ },
1729
+ {
1730
+ "epoch": 1.8867924528301887,
1731
+ "eval_loss": 0.2661677896976471,
1732
+ "eval_runtime": 19.2985,
1733
+ "eval_samples_per_second": 51.818,
1734
+ "eval_steps_per_second": 0.829,
1735
+ "step": 2300
1736
+ },
1737
+ {
1738
+ "epoch": 1.9031993437243644,
1739
+ "grad_norm": 0.28700560331344604,
1740
+ "learning_rate": 3.6205902041337727e-07,
1741
+ "loss": 0.2739,
1742
+ "step": 2320
1743
+ },
1744
+ {
1745
+ "epoch": 1.9031993437243644,
1746
+ "eval_loss": 0.26615363359451294,
1747
+ "eval_runtime": 19.3259,
1748
+ "eval_samples_per_second": 51.744,
1749
+ "eval_steps_per_second": 0.828,
1750
+ "step": 2320
1751
+ },
1752
+ {
1753
+ "epoch": 1.9196062346185396,
1754
+ "grad_norm": 0.2813996970653534,
1755
+ "learning_rate": 2.507722647443117e-07,
1756
+ "loss": 0.2683,
1757
+ "step": 2340
1758
+ },
1759
+ {
1760
+ "epoch": 1.9196062346185396,
1761
+ "eval_loss": 0.2661629915237427,
1762
+ "eval_runtime": 19.2987,
1763
+ "eval_samples_per_second": 51.817,
1764
+ "eval_steps_per_second": 0.829,
1765
+ "step": 2340
1766
+ },
1767
+ {
1768
+ "epoch": 1.9360131255127153,
1769
+ "grad_norm": 0.33076563477516174,
1770
+ "learning_rate": 1.5978188322536836e-07,
1771
+ "loss": 0.2708,
1772
+ "step": 2360
1773
+ },
1774
+ {
1775
+ "epoch": 1.9360131255127153,
1776
+ "eval_loss": 0.26614558696746826,
1777
+ "eval_runtime": 19.3189,
1778
+ "eval_samples_per_second": 51.763,
1779
+ "eval_steps_per_second": 0.828,
1780
+ "step": 2360
1781
+ },
1782
+ {
1783
+ "epoch": 1.952420016406891,
1784
+ "grad_norm": 0.3153195381164551,
1785
+ "learning_rate": 8.916249534961507e-08,
1786
+ "loss": 0.2824,
1787
+ "step": 2380
1788
+ },
1789
+ {
1790
+ "epoch": 1.952420016406891,
1791
+ "eval_loss": 0.26614320278167725,
1792
+ "eval_runtime": 19.3201,
1793
+ "eval_samples_per_second": 51.76,
1794
+ "eval_steps_per_second": 0.828,
1795
+ "step": 2380
1796
+ },
1797
+ {
1798
+ "epoch": 1.9688269073010665,
1799
+ "grad_norm": 0.30070924758911133,
1800
+ "learning_rate": 3.8972014743038356e-08,
1801
+ "loss": 0.2778,
1802
+ "step": 2400
1803
+ },
1804
+ {
1805
+ "epoch": 1.9688269073010665,
1806
+ "eval_loss": 0.26609641313552856,
1807
+ "eval_runtime": 19.3387,
1808
+ "eval_samples_per_second": 51.71,
1809
+ "eval_steps_per_second": 0.827,
1810
+ "step": 2400
1811
+ },
1812
+ {
1813
+ "epoch": 1.985233798195242,
1814
+ "grad_norm": 0.2762475311756134,
1815
+ "learning_rate": 9.25160167068173e-09,
1816
+ "loss": 0.2607,
1817
+ "step": 2420
1818
+ },
1819
+ {
1820
+ "epoch": 1.985233798195242,
1821
+ "eval_loss": 0.2661292254924774,
1822
+ "eval_runtime": 19.3179,
1823
+ "eval_samples_per_second": 51.765,
1824
+ "eval_steps_per_second": 0.828,
1825
+ "step": 2420
1826
+ },
1827
+ {
1828
+ "epoch": 2.0,
1829
+ "step": 2438,
1830
+ "total_flos": 1.58523627405312e+18,
1831
+ "train_loss": 0.30763408138284537,
1832
+ "train_runtime": 4680.0314,
1833
+ "train_samples_per_second": 16.667,
1834
+ "train_steps_per_second": 0.521
1835
+ }
1836
+ ],
1837
+ "logging_steps": 20,
1838
+ "max_steps": 2438,
1839
+ "num_input_tokens_seen": 0,
1840
+ "num_train_epochs": 2,
1841
+ "save_steps": 500,
1842
+ "stateful_callbacks": {
1843
+ "TrainerControl": {
1844
+ "args": {
1845
+ "should_epoch_stop": false,
1846
+ "should_evaluate": false,
1847
+ "should_log": false,
1848
+ "should_save": true,
1849
+ "should_training_stop": true
1850
+ },
1851
+ "attributes": {}
1852
+ }
1853
+ },
1854
+ "total_flos": 1.58523627405312e+18,
1855
+ "train_batch_size": 32,
1856
+ "trial_name": null,
1857
+ "trial_params": null
1858
+ }
nl_tasks/run_all/run_exps4/ft/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": false,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 4,
9
+ "peft_type": "ROTATION",
10
+ "r": 4,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "v_proj",
14
+ "q_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exps4/ft/special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "</s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "</s>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
nl_tasks/run_all/run_exps4/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exps4/ft/tokenizer_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ },
30
+ "32000": {
31
+ "content": "[PAD]",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": true
37
+ }
38
+ },
39
+ "bos_token": "</s>",
40
+ "clean_up_tokenization_spaces": false,
41
+ "eos_token": "</s>",
42
+ "extra_special_tokens": {},
43
+ "legacy": false,
44
+ "model_max_length": 512,
45
+ "pad_token": "[PAD]",
46
+ "padding_side": "right",
47
+ "sp_model_kwargs": {},
48
+ "tokenizer_class": "LlamaTokenizer",
49
+ "unk_token": "</s>",
50
+ "use_default_system_prompt": false
51
+ }
nl_tasks/run_all/run_exps4/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af2d541e09f63bb2770f195dc44c2dcb64e9f5cfb5e6cefc2f76470b7b11ebb4
3
+ size 6481
nl_tasks/run_all/run_exps4/ft2/adapter_config.json ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "T": 1.0,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
+ "bias": "none",
5
+ "inference_mode": true,
6
+ "layers_to_transform": null,
7
+ "modules_to_save": null,
8
+ "num_rotations": 4,
9
+ "peft_type": "ROTATION",
10
+ "r": 4,
11
+ "revision": null,
12
+ "target_modules": [
13
+ "v_proj",
14
+ "q_proj"
15
+ ],
16
+ "target_modules_to_skip": null,
17
+ "task_type": "CAUSAL_LM"
18
+ }
nl_tasks/run_all/run_exps4/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6c7e302310df8ca95e9b1c8512c14e6e8afc6114dc37de5c28102f04d3f605dd
3
+ size 33602659
nl_tasks/run_all/run_exps4/trainer_state.json ADDED
@@ -0,0 +1,1858 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.0,
6
+ "eval_steps": 20,
7
+ "global_step": 2438,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.016406890894175553,
14
+ "grad_norm": 0.07573527842760086,
15
+ "learning_rate": 7.78688524590164e-06,
16
+ "loss": 0.7453,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.016406890894175553,
21
+ "eval_loss": 0.7396451234817505,
22
+ "eval_runtime": 19.811,
23
+ "eval_samples_per_second": 50.477,
24
+ "eval_steps_per_second": 0.808,
25
+ "step": 20
26
+ },
27
+ {
28
+ "epoch": 0.03281378178835111,
29
+ "grad_norm": 0.23986372351646423,
30
+ "learning_rate": 1.598360655737705e-05,
31
+ "loss": 0.736,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.03281378178835111,
36
+ "eval_loss": 0.6820931434631348,
37
+ "eval_runtime": 19.2756,
38
+ "eval_samples_per_second": 51.879,
39
+ "eval_steps_per_second": 0.83,
40
+ "step": 40
41
+ },
42
+ {
43
+ "epoch": 0.04922067268252666,
44
+ "grad_norm": 0.2244514673948288,
45
+ "learning_rate": 2.418032786885246e-05,
46
+ "loss": 0.6054,
47
+ "step": 60
48
+ },
49
+ {
50
+ "epoch": 0.04922067268252666,
51
+ "eval_loss": 0.5053430795669556,
52
+ "eval_runtime": 19.2941,
53
+ "eval_samples_per_second": 51.829,
54
+ "eval_steps_per_second": 0.829,
55
+ "step": 60
56
+ },
57
+ {
58
+ "epoch": 0.06562756357670221,
59
+ "grad_norm": 0.15720054507255554,
60
+ "learning_rate": 3.237704918032787e-05,
61
+ "loss": 0.4734,
62
+ "step": 80
63
+ },
64
+ {
65
+ "epoch": 0.06562756357670221,
66
+ "eval_loss": 0.43087294697761536,
67
+ "eval_runtime": 19.3034,
68
+ "eval_samples_per_second": 51.804,
69
+ "eval_steps_per_second": 0.829,
70
+ "step": 80
71
+ },
72
+ {
73
+ "epoch": 0.08203445447087777,
74
+ "grad_norm": 0.1743963211774826,
75
+ "learning_rate": 4.057377049180328e-05,
76
+ "loss": 0.4262,
77
+ "step": 100
78
+ },
79
+ {
80
+ "epoch": 0.08203445447087777,
81
+ "eval_loss": 0.39660337567329407,
82
+ "eval_runtime": 19.2766,
83
+ "eval_samples_per_second": 51.876,
84
+ "eval_steps_per_second": 0.83,
85
+ "step": 100
86
+ },
87
+ {
88
+ "epoch": 0.09844134536505332,
89
+ "grad_norm": 0.18875327706336975,
90
+ "learning_rate": 4.8770491803278687e-05,
91
+ "loss": 0.3936,
92
+ "step": 120
93
+ },
94
+ {
95
+ "epoch": 0.09844134536505332,
96
+ "eval_loss": 0.3746892511844635,
97
+ "eval_runtime": 19.305,
98
+ "eval_samples_per_second": 51.8,
99
+ "eval_steps_per_second": 0.829,
100
+ "step": 120
101
+ },
102
+ {
103
+ "epoch": 0.11484823625922888,
104
+ "grad_norm": 0.23781315982341766,
105
+ "learning_rate": 5.69672131147541e-05,
106
+ "loss": 0.3655,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.11484823625922888,
111
+ "eval_loss": 0.35906875133514404,
112
+ "eval_runtime": 19.3235,
113
+ "eval_samples_per_second": 51.75,
114
+ "eval_steps_per_second": 0.828,
115
+ "step": 140
116
+ },
117
+ {
118
+ "epoch": 0.13125512715340443,
119
+ "grad_norm": 0.2293011099100113,
120
+ "learning_rate": 6.516393442622951e-05,
121
+ "loss": 0.3566,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.13125512715340443,
126
+ "eval_loss": 0.34795984625816345,
127
+ "eval_runtime": 19.3394,
128
+ "eval_samples_per_second": 51.708,
129
+ "eval_steps_per_second": 0.827,
130
+ "step": 160
131
+ },
132
+ {
133
+ "epoch": 0.14766201804757997,
134
+ "grad_norm": 0.21834053099155426,
135
+ "learning_rate": 7.336065573770491e-05,
136
+ "loss": 0.3521,
137
+ "step": 180
138
+ },
139
+ {
140
+ "epoch": 0.14766201804757997,
141
+ "eval_loss": 0.34054890275001526,
142
+ "eval_runtime": 19.3222,
143
+ "eval_samples_per_second": 51.754,
144
+ "eval_steps_per_second": 0.828,
145
+ "step": 180
146
+ },
147
+ {
148
+ "epoch": 0.16406890894175555,
149
+ "grad_norm": 0.24652570486068726,
150
+ "learning_rate": 8.155737704918032e-05,
151
+ "loss": 0.3472,
152
+ "step": 200
153
+ },
154
+ {
155
+ "epoch": 0.16406890894175555,
156
+ "eval_loss": 0.33308225870132446,
157
+ "eval_runtime": 19.3179,
158
+ "eval_samples_per_second": 51.765,
159
+ "eval_steps_per_second": 0.828,
160
+ "step": 200
161
+ },
162
+ {
163
+ "epoch": 0.1804757998359311,
164
+ "grad_norm": 0.29633331298828125,
165
+ "learning_rate": 8.975409836065574e-05,
166
+ "loss": 0.3518,
167
+ "step": 220
168
+ },
169
+ {
170
+ "epoch": 0.1804757998359311,
171
+ "eval_loss": 0.3273587226867676,
172
+ "eval_runtime": 19.3044,
173
+ "eval_samples_per_second": 51.802,
174
+ "eval_steps_per_second": 0.829,
175
+ "step": 220
176
+ },
177
+ {
178
+ "epoch": 0.19688269073010664,
179
+ "grad_norm": 0.22757525742053986,
180
+ "learning_rate": 9.795081967213115e-05,
181
+ "loss": 0.3385,
182
+ "step": 240
183
+ },
184
+ {
185
+ "epoch": 0.19688269073010664,
186
+ "eval_loss": 0.32346010208129883,
187
+ "eval_runtime": 19.3032,
188
+ "eval_samples_per_second": 51.805,
189
+ "eval_steps_per_second": 0.829,
190
+ "step": 240
191
+ },
192
+ {
193
+ "epoch": 0.2132895816242822,
194
+ "grad_norm": 0.22391946613788605,
195
+ "learning_rate": 9.998846726455519e-05,
196
+ "loss": 0.3327,
197
+ "step": 260
198
+ },
199
+ {
200
+ "epoch": 0.2132895816242822,
201
+ "eval_loss": 0.31933942437171936,
202
+ "eval_runtime": 19.3136,
203
+ "eval_samples_per_second": 51.777,
204
+ "eval_steps_per_second": 0.828,
205
+ "step": 260
206
+ },
207
+ {
208
+ "epoch": 0.22969647251845776,
209
+ "grad_norm": 0.2190631777048111,
210
+ "learning_rate": 9.99372213902336e-05,
211
+ "loss": 0.3333,
212
+ "step": 280
213
+ },
214
+ {
215
+ "epoch": 0.22969647251845776,
216
+ "eval_loss": 0.31619322299957275,
217
+ "eval_runtime": 19.3126,
218
+ "eval_samples_per_second": 51.78,
219
+ "eval_steps_per_second": 0.828,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2461033634126333,
224
+ "grad_norm": 0.26894959807395935,
225
+ "learning_rate": 9.984502294396932e-05,
226
+ "loss": 0.3206,
227
+ "step": 300
228
+ },
229
+ {
230
+ "epoch": 0.2461033634126333,
231
+ "eval_loss": 0.31452640891075134,
232
+ "eval_runtime": 19.3063,
233
+ "eval_samples_per_second": 51.796,
234
+ "eval_steps_per_second": 0.829,
235
+ "step": 300
236
+ },
237
+ {
238
+ "epoch": 0.26251025430680885,
239
+ "grad_norm": 0.21208558976650238,
240
+ "learning_rate": 9.971194753596649e-05,
241
+ "loss": 0.3118,
242
+ "step": 320
243
+ },
244
+ {
245
+ "epoch": 0.26251025430680885,
246
+ "eval_loss": 0.3100018799304962,
247
+ "eval_runtime": 19.3052,
248
+ "eval_samples_per_second": 51.8,
249
+ "eval_steps_per_second": 0.829,
250
+ "step": 320
251
+ },
252
+ {
253
+ "epoch": 0.27891714520098443,
254
+ "grad_norm": 0.2156500667333603,
255
+ "learning_rate": 9.953810429885344e-05,
256
+ "loss": 0.3254,
257
+ "step": 340
258
+ },
259
+ {
260
+ "epoch": 0.27891714520098443,
261
+ "eval_loss": 0.3061407208442688,
262
+ "eval_runtime": 19.3161,
263
+ "eval_samples_per_second": 51.77,
264
+ "eval_steps_per_second": 0.828,
265
+ "step": 340
266
+ },
267
+ {
268
+ "epoch": 0.29532403609515995,
269
+ "grad_norm": 0.24295839667320251,
270
+ "learning_rate": 9.932363579818499e-05,
271
+ "loss": 0.3042,
272
+ "step": 360
273
+ },
274
+ {
275
+ "epoch": 0.29532403609515995,
276
+ "eval_loss": 0.30404046177864075,
277
+ "eval_runtime": 19.3366,
278
+ "eval_samples_per_second": 51.715,
279
+ "eval_steps_per_second": 0.827,
280
+ "step": 360
281
+ },
282
+ {
283
+ "epoch": 0.3117309269893355,
284
+ "grad_norm": 0.27939218282699585,
285
+ "learning_rate": 9.90687179155272e-05,
286
+ "loss": 0.3111,
287
+ "step": 380
288
+ },
289
+ {
290
+ "epoch": 0.3117309269893355,
291
+ "eval_loss": 0.3018529415130615,
292
+ "eval_runtime": 19.3009,
293
+ "eval_samples_per_second": 51.811,
294
+ "eval_steps_per_second": 0.829,
295
+ "step": 380
296
+ },
297
+ {
298
+ "epoch": 0.3281378178835111,
299
+ "grad_norm": 0.2700340449810028,
300
+ "learning_rate": 9.877355970422022e-05,
301
+ "loss": 0.3218,
302
+ "step": 400
303
+ },
304
+ {
305
+ "epoch": 0.3281378178835111,
306
+ "eval_loss": 0.2997327446937561,
307
+ "eval_runtime": 19.3159,
308
+ "eval_samples_per_second": 51.771,
309
+ "eval_steps_per_second": 0.828,
310
+ "step": 400
311
+ },
312
+ {
313
+ "epoch": 0.3445447087776866,
314
+ "grad_norm": 0.22255544364452362,
315
+ "learning_rate": 9.843840321793758e-05,
316
+ "loss": 0.3138,
317
+ "step": 420
318
+ },
319
+ {
320
+ "epoch": 0.3445447087776866,
321
+ "eval_loss": 0.2985646724700928,
322
+ "eval_runtime": 19.3227,
323
+ "eval_samples_per_second": 51.753,
324
+ "eval_steps_per_second": 0.828,
325
+ "step": 420
326
+ },
327
+ {
328
+ "epoch": 0.3609515996718622,
329
+ "grad_norm": 0.22309252619743347,
330
+ "learning_rate": 9.806352331218253e-05,
331
+ "loss": 0.3133,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.3609515996718622,
336
+ "eval_loss": 0.2964664399623871,
337
+ "eval_runtime": 19.3056,
338
+ "eval_samples_per_second": 51.799,
339
+ "eval_steps_per_second": 0.829,
340
+ "step": 440
341
+ },
342
+ {
343
+ "epoch": 0.37735849056603776,
344
+ "grad_norm": 0.1800866723060608,
345
+ "learning_rate": 9.764922741888425e-05,
346
+ "loss": 0.3146,
347
+ "step": 460
348
+ },
349
+ {
350
+ "epoch": 0.37735849056603776,
351
+ "eval_loss": 0.2952148914337158,
352
+ "eval_runtime": 19.3129,
353
+ "eval_samples_per_second": 51.779,
354
+ "eval_steps_per_second": 0.828,
355
+ "step": 460
356
+ },
357
+ {
358
+ "epoch": 0.3937653814602133,
359
+ "grad_norm": 0.21533121168613434,
360
+ "learning_rate": 9.719585529427875e-05,
361
+ "loss": 0.3031,
362
+ "step": 480
363
+ },
364
+ {
365
+ "epoch": 0.3937653814602133,
366
+ "eval_loss": 0.2931912839412689,
367
+ "eval_runtime": 19.301,
368
+ "eval_samples_per_second": 51.811,
369
+ "eval_steps_per_second": 0.829,
370
+ "step": 480
371
+ },
372
+ {
373
+ "epoch": 0.41017227235438886,
374
+ "grad_norm": 0.18714798986911774,
375
+ "learning_rate": 9.670377874028118e-05,
376
+ "loss": 0.2983,
377
+ "step": 500
378
+ },
379
+ {
380
+ "epoch": 0.41017227235438886,
381
+ "eval_loss": 0.2921317517757416,
382
+ "eval_runtime": 19.3184,
383
+ "eval_samples_per_second": 51.764,
384
+ "eval_steps_per_second": 0.828,
385
+ "step": 500
386
+ },
387
+ {
388
+ "epoch": 0.4265791632485644,
389
+ "grad_norm": 0.25875231623649597,
390
+ "learning_rate": 9.617340129957806e-05,
391
+ "loss": 0.3068,
392
+ "step": 520
393
+ },
394
+ {
395
+ "epoch": 0.4265791632485644,
396
+ "eval_loss": 0.28973841667175293,
397
+ "eval_runtime": 19.2887,
398
+ "eval_samples_per_second": 51.844,
399
+ "eval_steps_per_second": 0.83,
400
+ "step": 520
401
+ },
402
+ {
403
+ "epoch": 0.44298605414273995,
404
+ "grad_norm": 0.24124988913536072,
405
+ "learning_rate": 9.560515792468967e-05,
406
+ "loss": 0.2998,
407
+ "step": 540
408
+ },
409
+ {
410
+ "epoch": 0.44298605414273995,
411
+ "eval_loss": 0.28842461109161377,
412
+ "eval_runtime": 19.303,
413
+ "eval_samples_per_second": 51.806,
414
+ "eval_steps_per_second": 0.829,
415
+ "step": 540
416
+ },
417
+ {
418
+ "epoch": 0.4593929450369155,
419
+ "grad_norm": 0.19618099927902222,
420
+ "learning_rate": 9.499951462127355e-05,
421
+ "loss": 0.2948,
422
+ "step": 560
423
+ },
424
+ {
425
+ "epoch": 0.4593929450369155,
426
+ "eval_loss": 0.2861073613166809,
427
+ "eval_runtime": 19.297,
428
+ "eval_samples_per_second": 51.822,
429
+ "eval_steps_per_second": 0.829,
430
+ "step": 560
431
+ },
432
+ {
433
+ "epoch": 0.47579983593109104,
434
+ "grad_norm": 0.23357711732387543,
435
+ "learning_rate": 9.435696806596215e-05,
436
+ "loss": 0.2899,
437
+ "step": 580
438
+ },
439
+ {
440
+ "epoch": 0.47579983593109104,
441
+ "eval_loss": 0.286967933177948,
442
+ "eval_runtime": 19.3122,
443
+ "eval_samples_per_second": 51.781,
444
+ "eval_steps_per_second": 0.828,
445
+ "step": 580
446
+ },
447
+ {
448
+ "epoch": 0.4922067268252666,
449
+ "grad_norm": 0.25024715065956116,
450
+ "learning_rate": 9.367804519904774e-05,
451
+ "loss": 0.2965,
452
+ "step": 600
453
+ },
454
+ {
455
+ "epoch": 0.4922067268252666,
456
+ "eval_loss": 0.28561797738075256,
457
+ "eval_runtime": 19.3145,
458
+ "eval_samples_per_second": 51.775,
459
+ "eval_steps_per_second": 0.828,
460
+ "step": 600
461
+ },
462
+ {
463
+ "epoch": 0.5086136177194421,
464
+ "grad_norm": 0.1980004608631134,
465
+ "learning_rate": 9.296330279234855e-05,
466
+ "loss": 0.2839,
467
+ "step": 620
468
+ },
469
+ {
470
+ "epoch": 0.5086136177194421,
471
+ "eval_loss": 0.2841115891933441,
472
+ "eval_runtime": 19.3071,
473
+ "eval_samples_per_second": 51.794,
474
+ "eval_steps_per_second": 0.829,
475
+ "step": 620
476
+ },
477
+ {
478
+ "epoch": 0.5250205086136177,
479
+ "grad_norm": 0.23671481013298035,
480
+ "learning_rate": 9.221332699261074e-05,
481
+ "loss": 0.2876,
482
+ "step": 640
483
+ },
484
+ {
485
+ "epoch": 0.5250205086136177,
486
+ "eval_loss": 0.28222420811653137,
487
+ "eval_runtime": 19.3124,
488
+ "eval_samples_per_second": 51.78,
489
+ "eval_steps_per_second": 0.828,
490
+ "step": 640
491
+ },
492
+ {
493
+ "epoch": 0.5414273995077933,
494
+ "grad_norm": 0.2384444773197174,
495
+ "learning_rate": 9.142873284082055e-05,
496
+ "loss": 0.2869,
497
+ "step": 660
498
+ },
499
+ {
500
+ "epoch": 0.5414273995077933,
501
+ "eval_loss": 0.28106391429901123,
502
+ "eval_runtime": 19.3044,
503
+ "eval_samples_per_second": 51.802,
504
+ "eval_steps_per_second": 0.829,
505
+ "step": 660
506
+ },
507
+ {
508
+ "epoch": 0.5578342904019689,
509
+ "grad_norm": 0.20951734483242035,
510
+ "learning_rate": 9.061016376782078e-05,
511
+ "loss": 0.2842,
512
+ "step": 680
513
+ },
514
+ {
515
+ "epoch": 0.5578342904019689,
516
+ "eval_loss": 0.2815721929073334,
517
+ "eval_runtime": 19.3084,
518
+ "eval_samples_per_second": 51.791,
519
+ "eval_steps_per_second": 0.829,
520
+ "step": 680
521
+ },
522
+ {
523
+ "epoch": 0.5742411812961444,
524
+ "grad_norm": 0.219334676861763,
525
+ "learning_rate": 8.97582910666454e-05,
526
+ "loss": 0.2816,
527
+ "step": 700
528
+ },
529
+ {
530
+ "epoch": 0.5742411812961444,
531
+ "eval_loss": 0.27998918294906616,
532
+ "eval_runtime": 19.286,
533
+ "eval_samples_per_second": 51.851,
534
+ "eval_steps_per_second": 0.83,
535
+ "step": 700
536
+ },
537
+ {
538
+ "epoch": 0.5906480721903199,
539
+ "grad_norm": 0.22159552574157715,
540
+ "learning_rate": 8.88738133420046e-05,
541
+ "loss": 0.2779,
542
+ "step": 720
543
+ },
544
+ {
545
+ "epoch": 0.5906480721903199,
546
+ "eval_loss": 0.2796504497528076,
547
+ "eval_runtime": 19.2993,
548
+ "eval_samples_per_second": 51.815,
549
+ "eval_steps_per_second": 0.829,
550
+ "step": 720
551
+ },
552
+ {
553
+ "epoch": 0.6070549630844955,
554
+ "grad_norm": 0.19857856631278992,
555
+ "learning_rate": 8.795745593737254e-05,
556
+ "loss": 0.2833,
557
+ "step": 740
558
+ },
559
+ {
560
+ "epoch": 0.6070549630844955,
561
+ "eval_loss": 0.27776020765304565,
562
+ "eval_runtime": 19.3146,
563
+ "eval_samples_per_second": 51.774,
564
+ "eval_steps_per_second": 0.828,
565
+ "step": 740
566
+ },
567
+ {
568
+ "epoch": 0.623461853978671,
569
+ "grad_norm": 0.2122536450624466,
570
+ "learning_rate": 8.700997034014665e-05,
571
+ "loss": 0.2891,
572
+ "step": 760
573
+ },
574
+ {
575
+ "epoch": 0.623461853978671,
576
+ "eval_loss": 0.27741900086402893,
577
+ "eval_runtime": 19.3038,
578
+ "eval_samples_per_second": 51.803,
579
+ "eval_steps_per_second": 0.829,
580
+ "step": 760
581
+ },
582
+ {
583
+ "epoch": 0.6398687448728466,
584
+ "grad_norm": 0.2515348494052887,
585
+ "learning_rate": 8.603213356536708e-05,
586
+ "loss": 0.2857,
587
+ "step": 780
588
+ },
589
+ {
590
+ "epoch": 0.6398687448728466,
591
+ "eval_loss": 0.2762429118156433,
592
+ "eval_runtime": 19.2907,
593
+ "eval_samples_per_second": 51.838,
594
+ "eval_steps_per_second": 0.829,
595
+ "step": 780
596
+ },
597
+ {
598
+ "epoch": 0.6562756357670222,
599
+ "grad_norm": 0.22974173724651337,
600
+ "learning_rate": 8.502474751850142e-05,
601
+ "loss": 0.2853,
602
+ "step": 800
603
+ },
604
+ {
605
+ "epoch": 0.6562756357670222,
606
+ "eval_loss": 0.2753770649433136,
607
+ "eval_runtime": 19.3213,
608
+ "eval_samples_per_second": 51.756,
609
+ "eval_steps_per_second": 0.828,
610
+ "step": 800
611
+ },
612
+ {
613
+ "epoch": 0.6726825266611977,
614
+ "grad_norm": 0.21828721463680267,
615
+ "learning_rate": 8.398863833781708e-05,
616
+ "loss": 0.2756,
617
+ "step": 820
618
+ },
619
+ {
620
+ "epoch": 0.6726825266611977,
621
+ "eval_loss": 0.2743964195251465,
622
+ "eval_runtime": 19.3088,
623
+ "eval_samples_per_second": 51.79,
624
+ "eval_steps_per_second": 0.829,
625
+ "step": 820
626
+ },
627
+ {
628
+ "epoch": 0.6890894175553732,
629
+ "grad_norm": 0.2040812075138092,
630
+ "learning_rate": 8.292465571688088e-05,
631
+ "loss": 0.2878,
632
+ "step": 840
633
+ },
634
+ {
635
+ "epoch": 0.6890894175553732,
636
+ "eval_loss": 0.27319151163101196,
637
+ "eval_runtime": 19.3286,
638
+ "eval_samples_per_second": 51.737,
639
+ "eval_steps_per_second": 0.828,
640
+ "step": 840
641
+ },
642
+ {
643
+ "epoch": 0.7054963084495488,
644
+ "grad_norm": 0.21715234220027924,
645
+ "learning_rate": 8.183367220774156e-05,
646
+ "loss": 0.2923,
647
+ "step": 860
648
+ },
649
+ {
650
+ "epoch": 0.7054963084495488,
651
+ "eval_loss": 0.2724905014038086,
652
+ "eval_runtime": 19.2993,
653
+ "eval_samples_per_second": 51.815,
654
+ "eval_steps_per_second": 0.829,
655
+ "step": 860
656
+ },
657
+ {
658
+ "epoch": 0.7219031993437244,
659
+ "grad_norm": 0.1924750655889511,
660
+ "learning_rate": 8.071658250536614e-05,
661
+ "loss": 0.2838,
662
+ "step": 880
663
+ },
664
+ {
665
+ "epoch": 0.7219031993437244,
666
+ "eval_loss": 0.2721441090106964,
667
+ "eval_runtime": 19.3119,
668
+ "eval_samples_per_second": 51.781,
669
+ "eval_steps_per_second": 0.829,
670
+ "step": 880
671
+ },
672
+ {
673
+ "epoch": 0.7383100902379,
674
+ "grad_norm": 0.23901104927062988,
675
+ "learning_rate": 7.957430271391762e-05,
676
+ "loss": 0.2768,
677
+ "step": 900
678
+ },
679
+ {
680
+ "epoch": 0.7383100902379,
681
+ "eval_loss": 0.271420955657959,
682
+ "eval_runtime": 19.2914,
683
+ "eval_samples_per_second": 51.837,
684
+ "eval_steps_per_second": 0.829,
685
+ "step": 900
686
+ },
687
+ {
688
+ "epoch": 0.7547169811320755,
689
+ "grad_norm": 0.210943803191185,
690
+ "learning_rate": 7.840776959547513e-05,
691
+ "loss": 0.2808,
692
+ "step": 920
693
+ },
694
+ {
695
+ "epoch": 0.7547169811320755,
696
+ "eval_loss": 0.27048394083976746,
697
+ "eval_runtime": 19.3054,
698
+ "eval_samples_per_second": 51.799,
699
+ "eval_steps_per_second": 0.829,
700
+ "step": 920
701
+ },
702
+ {
703
+ "epoch": 0.771123872026251,
704
+ "grad_norm": 0.2066313624382019,
705
+ "learning_rate": 7.72179398018132e-05,
706
+ "loss": 0.2633,
707
+ "step": 940
708
+ },
709
+ {
710
+ "epoch": 0.771123872026251,
711
+ "eval_loss": 0.2691468596458435,
712
+ "eval_runtime": 19.3075,
713
+ "eval_samples_per_second": 51.793,
714
+ "eval_steps_per_second": 0.829,
715
+ "step": 940
716
+ },
717
+ {
718
+ "epoch": 0.7875307629204266,
719
+ "grad_norm": 0.2224680632352829,
720
+ "learning_rate": 7.600578908986962e-05,
721
+ "loss": 0.2722,
722
+ "step": 960
723
+ },
724
+ {
725
+ "epoch": 0.7875307629204266,
726
+ "eval_loss": 0.26857373118400574,
727
+ "eval_runtime": 19.3038,
728
+ "eval_samples_per_second": 51.803,
729
+ "eval_steps_per_second": 0.829,
730
+ "step": 960
731
+ },
732
+ {
733
+ "epoch": 0.8039376538146021,
734
+ "grad_norm": 0.20683647692203522,
735
+ "learning_rate": 7.477231152154557e-05,
736
+ "loss": 0.2809,
737
+ "step": 980
738
+ },
739
+ {
740
+ "epoch": 0.8039376538146021,
741
+ "eval_loss": 0.2680775821208954,
742
+ "eval_runtime": 19.294,
743
+ "eval_samples_per_second": 51.83,
744
+ "eval_steps_per_second": 0.829,
745
+ "step": 980
746
+ },
747
+ {
748
+ "epoch": 0.8203445447087777,
749
+ "grad_norm": 0.22218932211399078,
750
+ "learning_rate": 7.35185186484943e-05,
751
+ "loss": 0.2802,
752
+ "step": 1000
753
+ },
754
+ {
755
+ "epoch": 0.8203445447087777,
756
+ "eval_loss": 0.26717033982276917,
757
+ "eval_runtime": 19.3196,
758
+ "eval_samples_per_second": 51.761,
759
+ "eval_steps_per_second": 0.828,
760
+ "step": 1000
761
+ },
762
+ {
763
+ "epoch": 0.8367514356029533,
764
+ "grad_norm": 0.234444260597229,
765
+ "learning_rate": 7.224543868256674e-05,
766
+ "loss": 0.2732,
767
+ "step": 1020
768
+ },
769
+ {
770
+ "epoch": 0.8367514356029533,
771
+ "eval_loss": 0.26637884974479675,
772
+ "eval_runtime": 19.2847,
773
+ "eval_samples_per_second": 51.854,
774
+ "eval_steps_per_second": 0.83,
775
+ "step": 1020
776
+ },
777
+ {
778
+ "epoch": 0.8531583264971287,
779
+ "grad_norm": 0.20181532204151154,
780
+ "learning_rate": 7.095411565259434e-05,
781
+ "loss": 0.2684,
782
+ "step": 1040
783
+ },
784
+ {
785
+ "epoch": 0.8531583264971287,
786
+ "eval_loss": 0.2665154039859772,
787
+ "eval_runtime": 19.3002,
788
+ "eval_samples_per_second": 51.813,
789
+ "eval_steps_per_second": 0.829,
790
+ "step": 1040
791
+ },
792
+ {
793
+ "epoch": 0.8695652173913043,
794
+ "grad_norm": 0.2278612107038498,
795
+ "learning_rate": 6.964560854820071e-05,
796
+ "loss": 0.2718,
797
+ "step": 1060
798
+ },
799
+ {
800
+ "epoch": 0.8695652173913043,
801
+ "eval_loss": 0.2659507989883423,
802
+ "eval_runtime": 19.2914,
803
+ "eval_samples_per_second": 51.836,
804
+ "eval_steps_per_second": 0.829,
805
+ "step": 1060
806
+ },
807
+ {
808
+ "epoch": 0.8859721082854799,
809
+ "grad_norm": 0.19423191249370575,
810
+ "learning_rate": 6.832099045134436e-05,
811
+ "loss": 0.2815,
812
+ "step": 1080
813
+ },
814
+ {
815
+ "epoch": 0.8859721082854799,
816
+ "eval_loss": 0.26532241702079773,
817
+ "eval_runtime": 19.3027,
818
+ "eval_samples_per_second": 51.806,
819
+ "eval_steps_per_second": 0.829,
820
+ "step": 1080
821
+ },
822
+ {
823
+ "epoch": 0.9023789991796555,
824
+ "grad_norm": 0.21753184497356415,
825
+ "learning_rate": 6.698134765630434e-05,
826
+ "loss": 0.2688,
827
+ "step": 1100
828
+ },
829
+ {
830
+ "epoch": 0.9023789991796555,
831
+ "eval_loss": 0.26458197832107544,
832
+ "eval_runtime": 19.2826,
833
+ "eval_samples_per_second": 51.86,
834
+ "eval_steps_per_second": 0.83,
835
+ "step": 1100
836
+ },
837
+ {
838
+ "epoch": 0.918785890073831,
839
+ "grad_norm": 0.21422693133354187,
840
+ "learning_rate": 6.562777877883087e-05,
841
+ "loss": 0.268,
842
+ "step": 1120
843
+ },
844
+ {
845
+ "epoch": 0.918785890073831,
846
+ "eval_loss": 0.2640591859817505,
847
+ "eval_runtime": 19.2864,
848
+ "eval_samples_per_second": 51.85,
849
+ "eval_steps_per_second": 0.83,
850
+ "step": 1120
851
+ },
852
+ {
853
+ "epoch": 0.9351927809680065,
854
+ "grad_norm": 0.21295098960399628,
855
+ "learning_rate": 6.426139385519126e-05,
856
+ "loss": 0.2755,
857
+ "step": 1140
858
+ },
859
+ {
860
+ "epoch": 0.9351927809680065,
861
+ "eval_loss": 0.26378899812698364,
862
+ "eval_runtime": 19.2944,
863
+ "eval_samples_per_second": 51.828,
864
+ "eval_steps_per_second": 0.829,
865
+ "step": 1140
866
+ },
867
+ {
868
+ "epoch": 0.9515996718621821,
869
+ "grad_norm": 0.21974827349185944,
870
+ "learning_rate": 6.288331343185027e-05,
871
+ "loss": 0.2869,
872
+ "step": 1160
873
+ },
874
+ {
875
+ "epoch": 0.9515996718621821,
876
+ "eval_loss": 0.2632799446582794,
877
+ "eval_runtime": 19.286,
878
+ "eval_samples_per_second": 51.851,
879
+ "eval_steps_per_second": 0.83,
880
+ "step": 1160
881
+ },
882
+ {
883
+ "epoch": 0.9680065627563577,
884
+ "grad_norm": 0.2579718828201294,
885
+ "learning_rate": 6.1494667646531e-05,
886
+ "loss": 0.2669,
887
+ "step": 1180
888
+ },
889
+ {
890
+ "epoch": 0.9680065627563577,
891
+ "eval_loss": 0.26267749071121216,
892
+ "eval_runtime": 19.3118,
893
+ "eval_samples_per_second": 51.782,
894
+ "eval_steps_per_second": 0.829,
895
+ "step": 1180
896
+ },
897
+ {
898
+ "epoch": 0.9844134536505332,
899
+ "grad_norm": 0.19952872395515442,
900
+ "learning_rate": 6.009659530141032e-05,
901
+ "loss": 0.2757,
902
+ "step": 1200
903
+ },
904
+ {
905
+ "epoch": 0.9844134536505332,
906
+ "eval_loss": 0.2614596486091614,
907
+ "eval_runtime": 19.3043,
908
+ "eval_samples_per_second": 51.802,
909
+ "eval_steps_per_second": 0.829,
910
+ "step": 1200
911
+ },
912
+ {
913
+ "epoch": 1.0008203445447088,
914
+ "grad_norm": 0.2142801284790039,
915
+ "learning_rate": 5.869024292920871e-05,
916
+ "loss": 0.281,
917
+ "step": 1220
918
+ },
919
+ {
920
+ "epoch": 1.0008203445447088,
921
+ "eval_loss": 0.2614995539188385,
922
+ "eval_runtime": 19.2801,
923
+ "eval_samples_per_second": 51.867,
924
+ "eval_steps_per_second": 0.83,
925
+ "step": 1220
926
+ },
927
+ {
928
+ "epoch": 1.0172272354388843,
929
+ "grad_norm": 0.20783104002475739,
930
+ "learning_rate": 5.727676385294047e-05,
931
+ "loss": 0.2521,
932
+ "step": 1240
933
+ },
934
+ {
935
+ "epoch": 1.0172272354388843,
936
+ "eval_loss": 0.2610865533351898,
937
+ "eval_runtime": 19.3018,
938
+ "eval_samples_per_second": 51.809,
939
+ "eval_steps_per_second": 0.829,
940
+ "step": 1240
941
+ },
942
+ {
943
+ "epoch": 1.03363412633306,
944
+ "grad_norm": 0.23703888058662415,
945
+ "learning_rate": 5.5857317240095295e-05,
946
+ "loss": 0.2623,
947
+ "step": 1260
948
+ },
949
+ {
950
+ "epoch": 1.03363412633306,
951
+ "eval_loss": 0.2600626051425934,
952
+ "eval_runtime": 19.3318,
953
+ "eval_samples_per_second": 51.728,
954
+ "eval_steps_per_second": 0.828,
955
+ "step": 1260
956
+ },
957
+ {
958
+ "epoch": 1.0500410172272354,
959
+ "grad_norm": 0.22795985639095306,
960
+ "learning_rate": 5.443306715202685e-05,
961
+ "loss": 0.2605,
962
+ "step": 1280
963
+ },
964
+ {
965
+ "epoch": 1.0500410172272354,
966
+ "eval_loss": 0.2595843970775604,
967
+ "eval_runtime": 19.3059,
968
+ "eval_samples_per_second": 51.798,
969
+ "eval_steps_per_second": 0.829,
970
+ "step": 1280
971
+ },
972
+ {
973
+ "epoch": 1.066447908121411,
974
+ "grad_norm": 0.22183609008789062,
975
+ "learning_rate": 5.300518158932815e-05,
976
+ "loss": 0.2555,
977
+ "step": 1300
978
+ },
979
+ {
980
+ "epoch": 1.066447908121411,
981
+ "eval_loss": 0.25863635540008545,
982
+ "eval_runtime": 19.3004,
983
+ "eval_samples_per_second": 51.812,
984
+ "eval_steps_per_second": 0.829,
985
+ "step": 1300
986
+ },
987
+ {
988
+ "epoch": 1.0828547990155866,
989
+ "grad_norm": 0.2546403408050537,
990
+ "learning_rate": 5.1574831533976256e-05,
991
+ "loss": 0.2507,
992
+ "step": 1320
993
+ },
994
+ {
995
+ "epoch": 1.0828547990155866,
996
+ "eval_loss": 0.25881943106651306,
997
+ "eval_runtime": 19.3151,
998
+ "eval_samples_per_second": 51.773,
999
+ "eval_steps_per_second": 0.828,
1000
+ "step": 1320
1001
+ },
1002
+ {
1003
+ "epoch": 1.099261689909762,
1004
+ "grad_norm": 0.22737851738929749,
1005
+ "learning_rate": 5.0143189989032066e-05,
1006
+ "loss": 0.2615,
1007
+ "step": 1340
1008
+ },
1009
+ {
1010
+ "epoch": 1.099261689909762,
1011
+ "eval_loss": 0.25874659419059753,
1012
+ "eval_runtime": 19.2986,
1013
+ "eval_samples_per_second": 51.817,
1014
+ "eval_steps_per_second": 0.829,
1015
+ "step": 1340
1016
+ },
1017
+ {
1018
+ "epoch": 1.1156685808039377,
1019
+ "grad_norm": 0.2325291633605957,
1020
+ "learning_rate": 4.8711431016682755e-05,
1021
+ "loss": 0.2621,
1022
+ "step": 1360
1023
+ },
1024
+ {
1025
+ "epoch": 1.1156685808039377,
1026
+ "eval_loss": 0.25852787494659424,
1027
+ "eval_runtime": 19.2891,
1028
+ "eval_samples_per_second": 51.843,
1029
+ "eval_steps_per_second": 0.829,
1030
+ "step": 1360
1031
+ },
1032
+ {
1033
+ "epoch": 1.1320754716981132,
1034
+ "grad_norm": 0.23844651877880096,
1035
+ "learning_rate": 4.728072877541545e-05,
1036
+ "loss": 0.2568,
1037
+ "step": 1380
1038
+ },
1039
+ {
1040
+ "epoch": 1.1320754716981132,
1041
+ "eval_loss": 0.25813454389572144,
1042
+ "eval_runtime": 19.3045,
1043
+ "eval_samples_per_second": 51.801,
1044
+ "eval_steps_per_second": 0.829,
1045
+ "step": 1380
1046
+ },
1047
+ {
1048
+ "epoch": 1.1484823625922886,
1049
+ "grad_norm": 0.2242225557565689,
1050
+ "learning_rate": 4.5852256557112104e-05,
1051
+ "loss": 0.265,
1052
+ "step": 1400
1053
+ },
1054
+ {
1055
+ "epoch": 1.1484823625922886,
1056
+ "eval_loss": 0.2577033042907715,
1057
+ "eval_runtime": 19.319,
1058
+ "eval_samples_per_second": 51.763,
1059
+ "eval_steps_per_second": 0.828,
1060
+ "step": 1400
1061
+ },
1062
+ {
1063
+ "epoch": 1.1648892534864643,
1064
+ "grad_norm": 0.2129354625940323,
1065
+ "learning_rate": 4.442718582485495e-05,
1066
+ "loss": 0.253,
1067
+ "step": 1420
1068
+ },
1069
+ {
1070
+ "epoch": 1.1648892534864643,
1071
+ "eval_loss": 0.2583049237728119,
1072
+ "eval_runtime": 19.332,
1073
+ "eval_samples_per_second": 51.728,
1074
+ "eval_steps_per_second": 0.828,
1075
+ "step": 1420
1076
+ },
1077
+ {
1078
+ "epoch": 1.1812961443806398,
1079
+ "grad_norm": 0.24126175045967102,
1080
+ "learning_rate": 4.300668525223178e-05,
1081
+ "loss": 0.2459,
1082
+ "step": 1440
1083
+ },
1084
+ {
1085
+ "epoch": 1.1812961443806398,
1086
+ "eval_loss": 0.25727447867393494,
1087
+ "eval_runtime": 19.3071,
1088
+ "eval_samples_per_second": 51.794,
1089
+ "eval_steps_per_second": 0.829,
1090
+ "step": 1440
1091
+ },
1092
+ {
1093
+ "epoch": 1.1977030352748155,
1094
+ "grad_norm": 0.22602707147598267,
1095
+ "learning_rate": 4.15919197649288e-05,
1096
+ "loss": 0.2634,
1097
+ "step": 1460
1098
+ },
1099
+ {
1100
+ "epoch": 1.1977030352748155,
1101
+ "eval_loss": 0.256967157125473,
1102
+ "eval_runtime": 19.3123,
1103
+ "eval_samples_per_second": 51.781,
1104
+ "eval_steps_per_second": 0.828,
1105
+ "step": 1460
1106
+ },
1107
+ {
1108
+ "epoch": 1.214109926168991,
1109
+ "grad_norm": 0.22526364028453827,
1110
+ "learning_rate": 4.0184049585397054e-05,
1111
+ "loss": 0.2624,
1112
+ "step": 1480
1113
+ },
1114
+ {
1115
+ "epoch": 1.214109926168991,
1116
+ "eval_loss": 0.2564084231853485,
1117
+ "eval_runtime": 19.2954,
1118
+ "eval_samples_per_second": 51.826,
1119
+ "eval_steps_per_second": 0.829,
1120
+ "step": 1480
1121
+ },
1122
+ {
1123
+ "epoch": 1.2305168170631666,
1124
+ "grad_norm": 0.25452113151550293,
1125
+ "learning_rate": 3.878422928137597e-05,
1126
+ "loss": 0.2594,
1127
+ "step": 1500
1128
+ },
1129
+ {
1130
+ "epoch": 1.2305168170631666,
1131
+ "eval_loss": 0.2563713788986206,
1132
+ "eval_runtime": 19.326,
1133
+ "eval_samples_per_second": 51.744,
1134
+ "eval_steps_per_second": 0.828,
1135
+ "step": 1500
1136
+ },
1137
+ {
1138
+ "epoch": 1.246923707957342,
1139
+ "grad_norm": 0.24358659982681274,
1140
+ "learning_rate": 3.7393606819054125e-05,
1141
+ "loss": 0.262,
1142
+ "step": 1520
1143
+ },
1144
+ {
1145
+ "epoch": 1.246923707957342,
1146
+ "eval_loss": 0.25621649622917175,
1147
+ "eval_runtime": 19.3211,
1148
+ "eval_samples_per_second": 51.757,
1149
+ "eval_steps_per_second": 0.828,
1150
+ "step": 1520
1151
+ },
1152
+ {
1153
+ "epoch": 1.2633305988515175,
1154
+ "grad_norm": 0.21525828540325165,
1155
+ "learning_rate": 3.601332262164385e-05,
1156
+ "loss": 0.2506,
1157
+ "step": 1540
1158
+ },
1159
+ {
1160
+ "epoch": 1.2633305988515175,
1161
+ "eval_loss": 0.2554308772087097,
1162
+ "eval_runtime": 19.2963,
1163
+ "eval_samples_per_second": 51.824,
1164
+ "eval_steps_per_second": 0.829,
1165
+ "step": 1540
1166
+ },
1167
+ {
1168
+ "epoch": 1.2797374897456932,
1169
+ "grad_norm": 0.23919980227947235,
1170
+ "learning_rate": 3.464450863414175e-05,
1171
+ "loss": 0.2591,
1172
+ "step": 1560
1173
+ },
1174
+ {
1175
+ "epoch": 1.2797374897456932,
1176
+ "eval_loss": 0.255275696516037,
1177
+ "eval_runtime": 19.3258,
1178
+ "eval_samples_per_second": 51.744,
1179
+ "eval_steps_per_second": 0.828,
1180
+ "step": 1560
1181
+ },
1182
+ {
1183
+ "epoch": 1.2961443806398687,
1184
+ "grad_norm": 0.2609453499317169,
1185
+ "learning_rate": 3.32882873950419e-05,
1186
+ "loss": 0.2537,
1187
+ "step": 1580
1188
+ },
1189
+ {
1190
+ "epoch": 1.2961443806398687,
1191
+ "eval_loss": 0.2548302114009857,
1192
+ "eval_runtime": 19.3128,
1193
+ "eval_samples_per_second": 51.779,
1194
+ "eval_steps_per_second": 0.828,
1195
+ "step": 1580
1196
+ },
1197
+ {
1198
+ "epoch": 1.3125512715340442,
1199
+ "grad_norm": 0.2392556518316269,
1200
+ "learning_rate": 3.194577111576333e-05,
1201
+ "loss": 0.255,
1202
+ "step": 1600
1203
+ },
1204
+ {
1205
+ "epoch": 1.3125512715340442,
1206
+ "eval_loss": 0.2544667720794678,
1207
+ "eval_runtime": 19.3061,
1208
+ "eval_samples_per_second": 51.797,
1209
+ "eval_steps_per_second": 0.829,
1210
+ "step": 1600
1211
+ },
1212
+ {
1213
+ "epoch": 1.3289581624282198,
1214
+ "grad_norm": 0.245096355676651,
1215
+ "learning_rate": 3.061806076854644e-05,
1216
+ "loss": 0.2511,
1217
+ "step": 1620
1218
+ },
1219
+ {
1220
+ "epoch": 1.3289581624282198,
1221
+ "eval_loss": 0.25414392352104187,
1222
+ "eval_runtime": 19.3189,
1223
+ "eval_samples_per_second": 51.763,
1224
+ "eval_steps_per_second": 0.828,
1225
+ "step": 1620
1226
+ },
1227
+ {
1228
+ "epoch": 1.3453650533223955,
1229
+ "grad_norm": 0.2578522861003876,
1230
+ "learning_rate": 2.9306245183566328e-05,
1231
+ "loss": 0.2544,
1232
+ "step": 1640
1233
+ },
1234
+ {
1235
+ "epoch": 1.3453650533223955,
1236
+ "eval_loss": 0.2534123361110687,
1237
+ "eval_runtime": 19.2928,
1238
+ "eval_samples_per_second": 51.833,
1239
+ "eval_steps_per_second": 0.829,
1240
+ "step": 1640
1241
+ },
1242
+ {
1243
+ "epoch": 1.361771944216571,
1244
+ "grad_norm": 0.254104346036911,
1245
+ "learning_rate": 2.801140015600392e-05,
1246
+ "loss": 0.2433,
1247
+ "step": 1660
1248
+ },
1249
+ {
1250
+ "epoch": 1.361771944216571,
1251
+ "eval_loss": 0.2531507909297943,
1252
+ "eval_runtime": 19.3043,
1253
+ "eval_samples_per_second": 51.802,
1254
+ "eval_steps_per_second": 0.829,
1255
+ "step": 1660
1256
+ },
1257
+ {
1258
+ "epoch": 1.3781788351107465,
1259
+ "grad_norm": 0.23047301173210144,
1260
+ "learning_rate": 2.673458756380661e-05,
1261
+ "loss": 0.2481,
1262
+ "step": 1680
1263
+ },
1264
+ {
1265
+ "epoch": 1.3781788351107465,
1266
+ "eval_loss": 0.25285908579826355,
1267
+ "eval_runtime": 19.3071,
1268
+ "eval_samples_per_second": 51.794,
1269
+ "eval_steps_per_second": 0.829,
1270
+ "step": 1680
1271
+ },
1272
+ {
1273
+ "epoch": 1.3945857260049221,
1274
+ "grad_norm": 0.23477159440517426,
1275
+ "learning_rate": 2.5476854496862057e-05,
1276
+ "loss": 0.2537,
1277
+ "step": 1700
1278
+ },
1279
+ {
1280
+ "epoch": 1.3945857260049221,
1281
+ "eval_loss": 0.2525946795940399,
1282
+ "eval_runtime": 19.3203,
1283
+ "eval_samples_per_second": 51.759,
1284
+ "eval_steps_per_second": 0.828,
1285
+ "step": 1700
1286
+ },
1287
+ {
1288
+ "epoch": 1.4109926168990976,
1289
+ "grad_norm": 0.2652125954627991,
1290
+ "learning_rate": 2.4239232398299806e-05,
1291
+ "loss": 0.2641,
1292
+ "step": 1720
1293
+ },
1294
+ {
1295
+ "epoch": 1.4109926168990976,
1296
+ "eval_loss": 0.252573698759079,
1297
+ "eval_runtime": 19.2988,
1298
+ "eval_samples_per_second": 51.817,
1299
+ "eval_steps_per_second": 0.829,
1300
+ "step": 1720
1301
+ },
1302
+ {
1303
+ "epoch": 1.427399507793273,
1304
+ "grad_norm": 0.2332567572593689,
1305
+ "learning_rate": 2.3022736218624046e-05,
1306
+ "loss": 0.2639,
1307
+ "step": 1740
1308
+ },
1309
+ {
1310
+ "epoch": 1.427399507793273,
1311
+ "eval_loss": 0.2521530091762543,
1312
+ "eval_runtime": 19.3012,
1313
+ "eval_samples_per_second": 51.81,
1314
+ "eval_steps_per_second": 0.829,
1315
+ "step": 1740
1316
+ },
1317
+ {
1318
+ "epoch": 1.4438063986874488,
1319
+ "grad_norm": 0.23365820944309235,
1320
+ "learning_rate": 2.1828363583371975e-05,
1321
+ "loss": 0.2566,
1322
+ "step": 1760
1323
+ },
1324
+ {
1325
+ "epoch": 1.4438063986874488,
1326
+ "eval_loss": 0.2516710162162781,
1327
+ "eval_runtime": 19.3143,
1328
+ "eval_samples_per_second": 51.775,
1329
+ "eval_steps_per_second": 0.828,
1330
+ "step": 1760
1331
+ },
1332
+ {
1333
+ "epoch": 1.4602132895816242,
1334
+ "grad_norm": 0.2693063020706177,
1335
+ "learning_rate": 2.0657093974979998e-05,
1336
+ "loss": 0.2505,
1337
+ "step": 1780
1338
+ },
1339
+ {
1340
+ "epoch": 1.4602132895816242,
1341
+ "eval_loss": 0.25175613164901733,
1342
+ "eval_runtime": 19.2942,
1343
+ "eval_samples_per_second": 51.829,
1344
+ "eval_steps_per_second": 0.829,
1345
+ "step": 1780
1346
+ },
1347
+ {
1348
+ "epoch": 1.4766201804758,
1349
+ "grad_norm": 0.23359094560146332,
1350
+ "learning_rate": 1.950988792952846e-05,
1351
+ "loss": 0.2618,
1352
+ "step": 1800
1353
+ },
1354
+ {
1355
+ "epoch": 1.4766201804758,
1356
+ "eval_loss": 0.25175967812538147,
1357
+ "eval_runtime": 19.3127,
1358
+ "eval_samples_per_second": 51.779,
1359
+ "eval_steps_per_second": 0.828,
1360
+ "step": 1800
1361
+ },
1362
+ {
1363
+ "epoch": 1.4930270713699754,
1364
+ "grad_norm": 0.2708717882633209,
1365
+ "learning_rate": 1.8387686249024456e-05,
1366
+ "loss": 0.2468,
1367
+ "step": 1820
1368
+ },
1369
+ {
1370
+ "epoch": 1.4930270713699754,
1371
+ "eval_loss": 0.25105661153793335,
1372
+ "eval_runtime": 19.3107,
1373
+ "eval_samples_per_second": 51.785,
1374
+ "eval_steps_per_second": 0.829,
1375
+ "step": 1820
1376
+ },
1377
+ {
1378
+ "epoch": 1.509433962264151,
1379
+ "grad_norm": 0.23433652520179749,
1380
+ "learning_rate": 1.729140922986764e-05,
1381
+ "loss": 0.2584,
1382
+ "step": 1840
1383
+ },
1384
+ {
1385
+ "epoch": 1.509433962264151,
1386
+ "eval_loss": 0.2507409155368805,
1387
+ "eval_runtime": 19.3083,
1388
+ "eval_samples_per_second": 51.791,
1389
+ "eval_steps_per_second": 0.829,
1390
+ "step": 1840
1391
+ },
1392
+ {
1393
+ "epoch": 1.5258408531583265,
1394
+ "grad_norm": 0.23181897401809692,
1395
+ "learning_rate": 1.6221955908132607e-05,
1396
+ "loss": 0.2547,
1397
+ "step": 1860
1398
+ },
1399
+ {
1400
+ "epoch": 1.5258408531583265,
1401
+ "eval_loss": 0.25062525272369385,
1402
+ "eval_runtime": 19.282,
1403
+ "eval_samples_per_second": 51.862,
1404
+ "eval_steps_per_second": 0.83,
1405
+ "step": 1860
1406
+ },
1407
+ {
1408
+ "epoch": 1.542247744052502,
1409
+ "grad_norm": 0.2638116478919983,
1410
+ "learning_rate": 1.5180203322286467e-05,
1411
+ "loss": 0.2483,
1412
+ "step": 1880
1413
+ },
1414
+ {
1415
+ "epoch": 1.542247744052502,
1416
+ "eval_loss": 0.2504216432571411,
1417
+ "eval_runtime": 19.3305,
1418
+ "eval_samples_per_second": 51.732,
1419
+ "eval_steps_per_second": 0.828,
1420
+ "step": 1880
1421
+ },
1422
+ {
1423
+ "epoch": 1.5586546349466777,
1424
+ "grad_norm": 0.23512820899486542,
1425
+ "learning_rate": 1.4167005793946037e-05,
1426
+ "loss": 0.251,
1427
+ "step": 1900
1428
+ },
1429
+ {
1430
+ "epoch": 1.5586546349466777,
1431
+ "eval_loss": 0.2505086064338684,
1432
+ "eval_runtime": 19.3179,
1433
+ "eval_samples_per_second": 51.765,
1434
+ "eval_steps_per_second": 0.828,
1435
+ "step": 1900
1436
+ },
1437
+ {
1438
+ "epoch": 1.5750615258408531,
1439
+ "grad_norm": 0.24093784391880035,
1440
+ "learning_rate": 1.3183194227264894e-05,
1441
+ "loss": 0.2544,
1442
+ "step": 1920
1443
+ },
1444
+ {
1445
+ "epoch": 1.5750615258408531,
1446
+ "eval_loss": 0.25002437829971313,
1447
+ "eval_runtime": 19.3193,
1448
+ "eval_samples_per_second": 51.762,
1449
+ "eval_steps_per_second": 0.828,
1450
+ "step": 1920
1451
+ },
1452
+ {
1453
+ "epoch": 1.5914684167350286,
1454
+ "grad_norm": 0.24928167462348938,
1455
+ "learning_rate": 1.2229575427524575e-05,
1456
+ "loss": 0.2552,
1457
+ "step": 1940
1458
+ },
1459
+ {
1460
+ "epoch": 1.5914684167350286,
1461
+ "eval_loss": 0.25014835596084595,
1462
+ "eval_runtime": 19.309,
1463
+ "eval_samples_per_second": 51.789,
1464
+ "eval_steps_per_second": 0.829,
1465
+ "step": 1940
1466
+ },
1467
+ {
1468
+ "epoch": 1.6078753076292043,
1469
+ "grad_norm": 0.25821226835250854,
1470
+ "learning_rate": 1.1306931439488777e-05,
1471
+ "loss": 0.2548,
1472
+ "step": 1960
1473
+ },
1474
+ {
1475
+ "epoch": 1.6078753076292043,
1476
+ "eval_loss": 0.2500286400318146,
1477
+ "eval_runtime": 19.3186,
1478
+ "eval_samples_per_second": 51.764,
1479
+ "eval_steps_per_second": 0.828,
1480
+ "step": 1960
1481
+ },
1482
+ {
1483
+ "epoch": 1.62428219852338,
1484
+ "grad_norm": 0.25347480177879333,
1485
+ "learning_rate": 1.0416018906063235e-05,
1486
+ "loss": 0.2529,
1487
+ "step": 1980
1488
+ },
1489
+ {
1490
+ "epoch": 1.62428219852338,
1491
+ "eval_loss": 0.2498283088207245,
1492
+ "eval_runtime": 19.2979,
1493
+ "eval_samples_per_second": 51.819,
1494
+ "eval_steps_per_second": 0.829,
1495
+ "step": 1980
1496
+ },
1497
+ {
1498
+ "epoch": 1.6406890894175554,
1499
+ "grad_norm": 0.2708413004875183,
1500
+ "learning_rate": 9.557568447787202e-06,
1501
+ "loss": 0.2552,
1502
+ "step": 2000
1503
+ },
1504
+ {
1505
+ "epoch": 1.6406890894175554,
1506
+ "eval_loss": 0.2497609555721283,
1507
+ "eval_runtime": 19.2953,
1508
+ "eval_samples_per_second": 51.826,
1509
+ "eval_steps_per_second": 0.829,
1510
+ "step": 2000
1511
+ },
1512
+ {
1513
+ "epoch": 1.6570959803117309,
1514
+ "grad_norm": 0.27011674642562866,
1515
+ "learning_rate": 8.73228406366528e-06,
1516
+ "loss": 0.2438,
1517
+ "step": 2020
1518
+ },
1519
+ {
1520
+ "epoch": 1.6570959803117309,
1521
+ "eval_loss": 0.24960409104824066,
1522
+ "eval_runtime": 19.2529,
1523
+ "eval_samples_per_second": 51.94,
1524
+ "eval_steps_per_second": 0.831,
1525
+ "step": 2020
1526
+ },
1527
+ {
1528
+ "epoch": 1.6735028712059066,
1529
+ "grad_norm": 0.231898233294487,
1530
+ "learning_rate": 7.940842553831168e-06,
1531
+ "loss": 0.2449,
1532
+ "step": 2040
1533
+ },
1534
+ {
1535
+ "epoch": 1.6735028712059066,
1536
+ "eval_loss": 0.24940046668052673,
1537
+ "eval_runtime": 19.2876,
1538
+ "eval_samples_per_second": 51.847,
1539
+ "eval_steps_per_second": 0.83,
1540
+ "step": 2040
1541
+ },
1542
+ {
1543
+ "epoch": 1.689909762100082,
1544
+ "grad_norm": 0.26983538269996643,
1545
+ "learning_rate": 7.183892964516697e-06,
1546
+ "loss": 0.2595,
1547
+ "step": 2060
1548
+ },
1549
+ {
1550
+ "epoch": 1.689909762100082,
1551
+ "eval_loss": 0.24934378266334534,
1552
+ "eval_runtime": 19.3091,
1553
+ "eval_samples_per_second": 51.789,
1554
+ "eval_steps_per_second": 0.829,
1555
+ "step": 2060
1556
+ },
1557
+ {
1558
+ "epoch": 1.7063166529942575,
1559
+ "grad_norm": 0.2576409876346588,
1560
+ "learning_rate": 6.462056055781157e-06,
1561
+ "loss": 0.2477,
1562
+ "step": 2080
1563
+ },
1564
+ {
1565
+ "epoch": 1.7063166529942575,
1566
+ "eval_loss": 0.24922724068164825,
1567
+ "eval_runtime": 19.2951,
1568
+ "eval_samples_per_second": 51.827,
1569
+ "eval_steps_per_second": 0.829,
1570
+ "step": 2080
1571
+ },
1572
+ {
1573
+ "epoch": 1.7227235438884332,
1574
+ "grad_norm": 0.24653151631355286,
1575
+ "learning_rate": 5.775923792437865e-06,
1576
+ "loss": 0.2466,
1577
+ "step": 2100
1578
+ },
1579
+ {
1580
+ "epoch": 1.7227235438884332,
1581
+ "eval_loss": 0.24906830489635468,
1582
+ "eval_runtime": 19.3203,
1583
+ "eval_samples_per_second": 51.759,
1584
+ "eval_steps_per_second": 0.828,
1585
+ "step": 2100
1586
+ },
1587
+ {
1588
+ "epoch": 1.7391304347826086,
1589
+ "grad_norm": 0.2507191300392151,
1590
+ "learning_rate": 5.126058858594856e-06,
1591
+ "loss": 0.2455,
1592
+ "step": 2120
1593
+ },
1594
+ {
1595
+ "epoch": 1.7391304347826086,
1596
+ "eval_loss": 0.24905811250209808,
1597
+ "eval_runtime": 19.3136,
1598
+ "eval_samples_per_second": 51.777,
1599
+ "eval_steps_per_second": 0.828,
1600
+ "step": 2120
1601
+ },
1602
+ {
1603
+ "epoch": 1.755537325676784,
1604
+ "grad_norm": 0.2599128782749176,
1605
+ "learning_rate": 4.512994196208359e-06,
1606
+ "loss": 0.2414,
1607
+ "step": 2140
1608
+ },
1609
+ {
1610
+ "epoch": 1.755537325676784,
1611
+ "eval_loss": 0.24877455830574036,
1612
+ "eval_runtime": 19.3141,
1613
+ "eval_samples_per_second": 51.776,
1614
+ "eval_steps_per_second": 0.828,
1615
+ "step": 2140
1616
+ },
1617
+ {
1618
+ "epoch": 1.7719442165709598,
1619
+ "grad_norm": 0.24992236495018005,
1620
+ "learning_rate": 3.937232568027166e-06,
1621
+ "loss": 0.2445,
1622
+ "step": 2160
1623
+ },
1624
+ {
1625
+ "epoch": 1.7719442165709598,
1626
+ "eval_loss": 0.24876806139945984,
1627
+ "eval_runtime": 19.3211,
1628
+ "eval_samples_per_second": 51.757,
1629
+ "eval_steps_per_second": 0.828,
1630
+ "step": 2160
1631
+ },
1632
+ {
1633
+ "epoch": 1.7883511074651355,
1634
+ "grad_norm": 0.21700158715248108,
1635
+ "learning_rate": 3.3992461452863056e-06,
1636
+ "loss": 0.2464,
1637
+ "step": 2180
1638
+ },
1639
+ {
1640
+ "epoch": 1.7883511074651355,
1641
+ "eval_loss": 0.2487395703792572,
1642
+ "eval_runtime": 19.2849,
1643
+ "eval_samples_per_second": 51.854,
1644
+ "eval_steps_per_second": 0.83,
1645
+ "step": 2180
1646
+ },
1647
+ {
1648
+ "epoch": 1.804757998359311,
1649
+ "grad_norm": 0.27604594826698303,
1650
+ "learning_rate": 2.8994761204884756e-06,
1651
+ "loss": 0.2527,
1652
+ "step": 2200
1653
+ },
1654
+ {
1655
+ "epoch": 1.804757998359311,
1656
+ "eval_loss": 0.2487618327140808,
1657
+ "eval_runtime": 19.3159,
1658
+ "eval_samples_per_second": 51.771,
1659
+ "eval_steps_per_second": 0.828,
1660
+ "step": 2200
1661
+ },
1662
+ {
1663
+ "epoch": 1.8211648892534864,
1664
+ "grad_norm": 0.26943638920783997,
1665
+ "learning_rate": 2.4383323455902997e-06,
1666
+ "loss": 0.24,
1667
+ "step": 2220
1668
+ },
1669
+ {
1670
+ "epoch": 1.8211648892534864,
1671
+ "eval_loss": 0.24872750043869019,
1672
+ "eval_runtime": 19.3062,
1673
+ "eval_samples_per_second": 51.797,
1674
+ "eval_steps_per_second": 0.829,
1675
+ "step": 2220
1676
+ },
1677
+ {
1678
+ "epoch": 1.837571780147662,
1679
+ "grad_norm": 0.2658500373363495,
1680
+ "learning_rate": 2.0161929958905145e-06,
1681
+ "loss": 0.2514,
1682
+ "step": 2240
1683
+ },
1684
+ {
1685
+ "epoch": 1.837571780147662,
1686
+ "eval_loss": 0.24868498742580414,
1687
+ "eval_runtime": 19.3043,
1688
+ "eval_samples_per_second": 51.802,
1689
+ "eval_steps_per_second": 0.829,
1690
+ "step": 2240
1691
+ },
1692
+ {
1693
+ "epoch": 1.8539786710418376,
1694
+ "grad_norm": 0.2387407273054123,
1695
+ "learning_rate": 1.6334042598955379e-06,
1696
+ "loss": 0.2454,
1697
+ "step": 2260
1698
+ },
1699
+ {
1700
+ "epoch": 1.8539786710418376,
1701
+ "eval_loss": 0.24864910542964935,
1702
+ "eval_runtime": 19.3178,
1703
+ "eval_samples_per_second": 51.766,
1704
+ "eval_steps_per_second": 0.828,
1705
+ "step": 2260
1706
+ },
1707
+ {
1708
+ "epoch": 1.870385561936013,
1709
+ "grad_norm": 0.2613567113876343,
1710
+ "learning_rate": 1.2902800554168649e-06,
1711
+ "loss": 0.2503,
1712
+ "step": 2280
1713
+ },
1714
+ {
1715
+ "epoch": 1.870385561936013,
1716
+ "eval_loss": 0.2486574351787567,
1717
+ "eval_runtime": 19.3078,
1718
+ "eval_samples_per_second": 51.793,
1719
+ "eval_steps_per_second": 0.829,
1720
+ "step": 2280
1721
+ },
1722
+ {
1723
+ "epoch": 1.8867924528301887,
1724
+ "grad_norm": 0.285758376121521,
1725
+ "learning_rate": 9.871017721329201e-07,
1726
+ "loss": 0.2598,
1727
+ "step": 2300
1728
+ },
1729
+ {
1730
+ "epoch": 1.8867924528301887,
1731
+ "eval_loss": 0.24857309460639954,
1732
+ "eval_runtime": 19.297,
1733
+ "eval_samples_per_second": 51.821,
1734
+ "eval_steps_per_second": 0.829,
1735
+ "step": 2300
1736
+ },
1737
+ {
1738
+ "epoch": 1.9031993437243644,
1739
+ "grad_norm": 0.23689517378807068,
1740
+ "learning_rate": 7.241180408267545e-07,
1741
+ "loss": 0.2539,
1742
+ "step": 2320
1743
+ },
1744
+ {
1745
+ "epoch": 1.9031993437243644,
1746
+ "eval_loss": 0.24856111407279968,
1747
+ "eval_runtime": 19.3155,
1748
+ "eval_samples_per_second": 51.772,
1749
+ "eval_steps_per_second": 0.828,
1750
+ "step": 2320
1751
+ },
1752
+ {
1753
+ "epoch": 1.9196062346185396,
1754
+ "grad_norm": 0.23107603192329407,
1755
+ "learning_rate": 5.015445294886234e-07,
1756
+ "loss": 0.2479,
1757
+ "step": 2340
1758
+ },
1759
+ {
1760
+ "epoch": 1.9196062346185396,
1761
+ "eval_loss": 0.24852712452411652,
1762
+ "eval_runtime": 19.3164,
1763
+ "eval_samples_per_second": 51.769,
1764
+ "eval_steps_per_second": 0.828,
1765
+ "step": 2340
1766
+ },
1767
+ {
1768
+ "epoch": 1.9360131255127153,
1769
+ "grad_norm": 0.26831313967704773,
1770
+ "learning_rate": 3.195637664507367e-07,
1771
+ "loss": 0.2508,
1772
+ "step": 2360
1773
+ },
1774
+ {
1775
+ "epoch": 1.9360131255127153,
1776
+ "eval_loss": 0.24849362671375275,
1777
+ "eval_runtime": 19.3248,
1778
+ "eval_samples_per_second": 51.747,
1779
+ "eval_steps_per_second": 0.828,
1780
+ "step": 2360
1781
+ },
1782
+ {
1783
+ "epoch": 1.952420016406891,
1784
+ "grad_norm": 0.2591889500617981,
1785
+ "learning_rate": 1.7832499069923014e-07,
1786
+ "loss": 0.262,
1787
+ "step": 2380
1788
+ },
1789
+ {
1790
+ "epoch": 1.952420016406891,
1791
+ "eval_loss": 0.24852906167507172,
1792
+ "eval_runtime": 19.3174,
1793
+ "eval_samples_per_second": 51.767,
1794
+ "eval_steps_per_second": 0.828,
1795
+ "step": 2380
1796
+ },
1797
+ {
1798
+ "epoch": 1.9688269073010665,
1799
+ "grad_norm": 0.25030118227005005,
1800
+ "learning_rate": 7.794402948607671e-08,
1801
+ "loss": 0.257,
1802
+ "step": 2400
1803
+ },
1804
+ {
1805
+ "epoch": 1.9688269073010665,
1806
+ "eval_loss": 0.24847978353500366,
1807
+ "eval_runtime": 19.3044,
1808
+ "eval_samples_per_second": 51.802,
1809
+ "eval_steps_per_second": 0.829,
1810
+ "step": 2400
1811
+ },
1812
+ {
1813
+ "epoch": 1.985233798195242,
1814
+ "grad_norm": 0.2138320505619049,
1815
+ "learning_rate": 1.850320334136346e-08,
1816
+ "loss": 0.2407,
1817
+ "step": 2420
1818
+ },
1819
+ {
1820
+ "epoch": 1.985233798195242,
1821
+ "eval_loss": 0.24846026301383972,
1822
+ "eval_runtime": 19.3185,
1823
+ "eval_samples_per_second": 51.764,
1824
+ "eval_steps_per_second": 0.828,
1825
+ "step": 2420
1826
+ },
1827
+ {
1828
+ "epoch": 2.0,
1829
+ "step": 2438,
1830
+ "total_flos": 1.58523627405312e+18,
1831
+ "train_loss": 0.2885599858062594,
1832
+ "train_runtime": 4675.7614,
1833
+ "train_samples_per_second": 16.682,
1834
+ "train_steps_per_second": 0.521
1835
+ }
1836
+ ],
1837
+ "logging_steps": 20,
1838
+ "max_steps": 2438,
1839
+ "num_input_tokens_seen": 0,
1840
+ "num_train_epochs": 2,
1841
+ "save_steps": 500,
1842
+ "stateful_callbacks": {
1843
+ "TrainerControl": {
1844
+ "args": {
1845
+ "should_epoch_stop": false,
1846
+ "should_evaluate": false,
1847
+ "should_log": false,
1848
+ "should_save": true,
1849
+ "should_training_stop": true
1850
+ },
1851
+ "attributes": {}
1852
+ }
1853
+ },
1854
+ "total_flos": 1.58523627405312e+18,
1855
+ "train_batch_size": 32,
1856
+ "trial_name": null,
1857
+ "trial_params": null
1858
+ }
nl_tasks/run_all/run_exps5/trainer_state.json ADDED
@@ -0,0 +1,1858 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 2.0,
6
+ "eval_steps": 20,
7
+ "global_step": 2438,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.016406890894175553,
14
+ "grad_norm": 0.11380166560411453,
15
+ "learning_rate": 1.557377049180328e-05,
16
+ "loss": 0.7435,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.016406890894175553,
21
+ "eval_loss": 0.7304632067680359,
22
+ "eval_runtime": 19.8202,
23
+ "eval_samples_per_second": 50.453,
24
+ "eval_steps_per_second": 0.807,
25
+ "step": 20
26
+ },
27
+ {
28
+ "epoch": 0.03281378178835111,
29
+ "grad_norm": 0.25147441029548645,
30
+ "learning_rate": 3.19672131147541e-05,
31
+ "loss": 0.6829,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.03281378178835111,
36
+ "eval_loss": 0.5602980852127075,
37
+ "eval_runtime": 19.2612,
38
+ "eval_samples_per_second": 51.918,
39
+ "eval_steps_per_second": 0.831,
40
+ "step": 40
41
+ },
42
+ {
43
+ "epoch": 0.04922067268252666,
44
+ "grad_norm": 0.14670787751674652,
45
+ "learning_rate": 4.836065573770492e-05,
46
+ "loss": 0.4875,
47
+ "step": 60
48
+ },
49
+ {
50
+ "epoch": 0.04922067268252666,
51
+ "eval_loss": 0.4308978021144867,
52
+ "eval_runtime": 19.3049,
53
+ "eval_samples_per_second": 51.8,
54
+ "eval_steps_per_second": 0.829,
55
+ "step": 60
56
+ },
57
+ {
58
+ "epoch": 0.06562756357670221,
59
+ "grad_norm": 0.1724623739719391,
60
+ "learning_rate": 6.475409836065574e-05,
61
+ "loss": 0.4219,
62
+ "step": 80
63
+ },
64
+ {
65
+ "epoch": 0.06562756357670221,
66
+ "eval_loss": 0.3885975778102875,
67
+ "eval_runtime": 19.3057,
68
+ "eval_samples_per_second": 51.798,
69
+ "eval_steps_per_second": 0.829,
70
+ "step": 80
71
+ },
72
+ {
73
+ "epoch": 0.08203445447087777,
74
+ "grad_norm": 0.24202755093574524,
75
+ "learning_rate": 8.114754098360656e-05,
76
+ "loss": 0.3895,
77
+ "step": 100
78
+ },
79
+ {
80
+ "epoch": 0.08203445447087777,
81
+ "eval_loss": 0.365479439496994,
82
+ "eval_runtime": 19.2994,
83
+ "eval_samples_per_second": 51.815,
84
+ "eval_steps_per_second": 0.829,
85
+ "step": 100
86
+ },
87
+ {
88
+ "epoch": 0.09844134536505332,
89
+ "grad_norm": 0.21835799515247345,
90
+ "learning_rate": 9.754098360655737e-05,
91
+ "loss": 0.3638,
92
+ "step": 120
93
+ },
94
+ {
95
+ "epoch": 0.09844134536505332,
96
+ "eval_loss": 0.3488895893096924,
97
+ "eval_runtime": 19.2809,
98
+ "eval_samples_per_second": 51.865,
99
+ "eval_steps_per_second": 0.83,
100
+ "step": 120
101
+ },
102
+ {
103
+ "epoch": 0.11484823625922888,
104
+ "grad_norm": 0.2230556756258011,
105
+ "learning_rate": 0.0001139344262295082,
106
+ "loss": 0.3405,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.11484823625922888,
111
+ "eval_loss": 0.3406926989555359,
112
+ "eval_runtime": 19.2861,
113
+ "eval_samples_per_second": 51.851,
114
+ "eval_steps_per_second": 0.83,
115
+ "step": 140
116
+ },
117
+ {
118
+ "epoch": 0.13125512715340443,
119
+ "grad_norm": 0.24213728308677673,
120
+ "learning_rate": 0.00013032786885245902,
121
+ "loss": 0.3382,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.13125512715340443,
126
+ "eval_loss": 0.3313358426094055,
127
+ "eval_runtime": 19.3454,
128
+ "eval_samples_per_second": 51.692,
129
+ "eval_steps_per_second": 0.827,
130
+ "step": 160
131
+ },
132
+ {
133
+ "epoch": 0.14766201804757997,
134
+ "grad_norm": 0.21106292307376862,
135
+ "learning_rate": 0.00014672131147540983,
136
+ "loss": 0.3362,
137
+ "step": 180
138
+ },
139
+ {
140
+ "epoch": 0.14766201804757997,
141
+ "eval_loss": 0.3270183801651001,
142
+ "eval_runtime": 19.3128,
143
+ "eval_samples_per_second": 51.779,
144
+ "eval_steps_per_second": 0.828,
145
+ "step": 180
146
+ },
147
+ {
148
+ "epoch": 0.16406890894175555,
149
+ "grad_norm": 0.19206179678440094,
150
+ "learning_rate": 0.00016311475409836064,
151
+ "loss": 0.3333,
152
+ "step": 200
153
+ },
154
+ {
155
+ "epoch": 0.16406890894175555,
156
+ "eval_loss": 0.3220973610877991,
157
+ "eval_runtime": 19.3269,
158
+ "eval_samples_per_second": 51.741,
159
+ "eval_steps_per_second": 0.828,
160
+ "step": 200
161
+ },
162
+ {
163
+ "epoch": 0.1804757998359311,
164
+ "grad_norm": 0.23427358269691467,
165
+ "learning_rate": 0.00017950819672131149,
166
+ "loss": 0.3384,
167
+ "step": 220
168
+ },
169
+ {
170
+ "epoch": 0.1804757998359311,
171
+ "eval_loss": 0.315133661031723,
172
+ "eval_runtime": 19.3062,
173
+ "eval_samples_per_second": 51.797,
174
+ "eval_steps_per_second": 0.829,
175
+ "step": 220
176
+ },
177
+ {
178
+ "epoch": 0.19688269073010664,
179
+ "grad_norm": 0.2133898138999939,
180
+ "learning_rate": 0.0001959016393442623,
181
+ "loss": 0.3267,
182
+ "step": 240
183
+ },
184
+ {
185
+ "epoch": 0.19688269073010664,
186
+ "eval_loss": 0.31321343779563904,
187
+ "eval_runtime": 19.3447,
188
+ "eval_samples_per_second": 51.694,
189
+ "eval_steps_per_second": 0.827,
190
+ "step": 240
191
+ },
192
+ {
193
+ "epoch": 0.2132895816242822,
194
+ "grad_norm": 0.17936132848262787,
195
+ "learning_rate": 0.00019997693452911039,
196
+ "loss": 0.3209,
197
+ "step": 260
198
+ },
199
+ {
200
+ "epoch": 0.2132895816242822,
201
+ "eval_loss": 0.30857598781585693,
202
+ "eval_runtime": 19.322,
203
+ "eval_samples_per_second": 51.754,
204
+ "eval_steps_per_second": 0.828,
205
+ "step": 260
206
+ },
207
+ {
208
+ "epoch": 0.22969647251845776,
209
+ "grad_norm": 0.17790672183036804,
210
+ "learning_rate": 0.0001998744427804672,
211
+ "loss": 0.3226,
212
+ "step": 280
213
+ },
214
+ {
215
+ "epoch": 0.22969647251845776,
216
+ "eval_loss": 0.30542299151420593,
217
+ "eval_runtime": 19.3079,
218
+ "eval_samples_per_second": 51.792,
219
+ "eval_steps_per_second": 0.829,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2461033634126333,
224
+ "grad_norm": 0.24569754302501678,
225
+ "learning_rate": 0.00019969004588793864,
226
+ "loss": 0.3095,
227
+ "step": 300
228
+ },
229
+ {
230
+ "epoch": 0.2461033634126333,
231
+ "eval_loss": 0.30299249291419983,
232
+ "eval_runtime": 19.311,
233
+ "eval_samples_per_second": 51.784,
234
+ "eval_steps_per_second": 0.829,
235
+ "step": 300
236
+ },
237
+ {
238
+ "epoch": 0.26251025430680885,
239
+ "grad_norm": 0.18036572635173798,
240
+ "learning_rate": 0.00019942389507193297,
241
+ "loss": 0.3003,
242
+ "step": 320
243
+ },
244
+ {
245
+ "epoch": 0.26251025430680885,
246
+ "eval_loss": 0.29782596230506897,
247
+ "eval_runtime": 19.3098,
248
+ "eval_samples_per_second": 51.787,
249
+ "eval_steps_per_second": 0.829,
250
+ "step": 320
251
+ },
252
+ {
253
+ "epoch": 0.27891714520098443,
254
+ "grad_norm": 0.16709011793136597,
255
+ "learning_rate": 0.0001990762085977069,
256
+ "loss": 0.3133,
257
+ "step": 340
258
+ },
259
+ {
260
+ "epoch": 0.27891714520098443,
261
+ "eval_loss": 0.29489508271217346,
262
+ "eval_runtime": 19.3201,
263
+ "eval_samples_per_second": 51.759,
264
+ "eval_steps_per_second": 0.828,
265
+ "step": 340
266
+ },
267
+ {
268
+ "epoch": 0.29532403609515995,
269
+ "grad_norm": 0.18404647707939148,
270
+ "learning_rate": 0.00019864727159636997,
271
+ "loss": 0.2933,
272
+ "step": 360
273
+ },
274
+ {
275
+ "epoch": 0.29532403609515995,
276
+ "eval_loss": 0.2923511564731598,
277
+ "eval_runtime": 19.3263,
278
+ "eval_samples_per_second": 51.743,
279
+ "eval_steps_per_second": 0.828,
280
+ "step": 360
281
+ },
282
+ {
283
+ "epoch": 0.3117309269893355,
284
+ "grad_norm": 0.212888702750206,
285
+ "learning_rate": 0.0001981374358310544,
286
+ "loss": 0.2996,
287
+ "step": 380
288
+ },
289
+ {
290
+ "epoch": 0.3117309269893355,
291
+ "eval_loss": 0.29031240940093994,
292
+ "eval_runtime": 19.3114,
293
+ "eval_samples_per_second": 51.783,
294
+ "eval_steps_per_second": 0.829,
295
+ "step": 380
296
+ },
297
+ {
298
+ "epoch": 0.3281378178835111,
299
+ "grad_norm": 0.21747799217700958,
300
+ "learning_rate": 0.00019754711940844045,
301
+ "loss": 0.3099,
302
+ "step": 400
303
+ },
304
+ {
305
+ "epoch": 0.3281378178835111,
306
+ "eval_loss": 0.2885620594024658,
307
+ "eval_runtime": 19.315,
308
+ "eval_samples_per_second": 51.773,
309
+ "eval_steps_per_second": 0.828,
310
+ "step": 400
311
+ },
312
+ {
313
+ "epoch": 0.3445447087776866,
314
+ "grad_norm": 0.1852598935365677,
315
+ "learning_rate": 0.00019687680643587515,
316
+ "loss": 0.3021,
317
+ "step": 420
318
+ },
319
+ {
320
+ "epoch": 0.3445447087776866,
321
+ "eval_loss": 0.28671303391456604,
322
+ "eval_runtime": 19.3235,
323
+ "eval_samples_per_second": 51.75,
324
+ "eval_steps_per_second": 0.828,
325
+ "step": 420
326
+ },
327
+ {
328
+ "epoch": 0.3609515996718622,
329
+ "grad_norm": 0.18169638514518738,
330
+ "learning_rate": 0.00019612704662436506,
331
+ "loss": 0.3015,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.3609515996718622,
336
+ "eval_loss": 0.283815860748291,
337
+ "eval_runtime": 19.3303,
338
+ "eval_samples_per_second": 51.732,
339
+ "eval_steps_per_second": 0.828,
340
+ "step": 440
341
+ },
342
+ {
343
+ "epoch": 0.37735849056603776,
344
+ "grad_norm": 0.15063416957855225,
345
+ "learning_rate": 0.0001952984548377685,
346
+ "loss": 0.3021,
347
+ "step": 460
348
+ },
349
+ {
350
+ "epoch": 0.37735849056603776,
351
+ "eval_loss": 0.2832798957824707,
352
+ "eval_runtime": 19.3186,
353
+ "eval_samples_per_second": 51.763,
354
+ "eval_steps_per_second": 0.828,
355
+ "step": 460
356
+ },
357
+ {
358
+ "epoch": 0.3937653814602133,
359
+ "grad_norm": 0.1891593188047409,
360
+ "learning_rate": 0.0001943917105885575,
361
+ "loss": 0.291,
362
+ "step": 480
363
+ },
364
+ {
365
+ "epoch": 0.3937653814602133,
366
+ "eval_loss": 0.28229087591171265,
367
+ "eval_runtime": 19.3112,
368
+ "eval_samples_per_second": 51.783,
369
+ "eval_steps_per_second": 0.829,
370
+ "step": 480
371
+ },
372
+ {
373
+ "epoch": 0.41017227235438886,
374
+ "grad_norm": 0.15703819692134857,
375
+ "learning_rate": 0.00019340755748056235,
376
+ "loss": 0.2864,
377
+ "step": 500
378
+ },
379
+ {
380
+ "epoch": 0.41017227235438886,
381
+ "eval_loss": 0.27941229939460754,
382
+ "eval_runtime": 19.3309,
383
+ "eval_samples_per_second": 51.731,
384
+ "eval_steps_per_second": 0.828,
385
+ "step": 500
386
+ },
387
+ {
388
+ "epoch": 0.4265791632485644,
389
+ "grad_norm": 0.20838674902915955,
390
+ "learning_rate": 0.0001923468025991561,
391
+ "loss": 0.2941,
392
+ "step": 520
393
+ },
394
+ {
395
+ "epoch": 0.4265791632485644,
396
+ "eval_loss": 0.27715691924095154,
397
+ "eval_runtime": 19.2863,
398
+ "eval_samples_per_second": 51.85,
399
+ "eval_steps_per_second": 0.83,
400
+ "step": 520
401
+ },
402
+ {
403
+ "epoch": 0.44298605414273995,
404
+ "grad_norm": 0.20151035487651825,
405
+ "learning_rate": 0.00019121031584937935,
406
+ "loss": 0.288,
407
+ "step": 540
408
+ },
409
+ {
410
+ "epoch": 0.44298605414273995,
411
+ "eval_loss": 0.2766241729259491,
412
+ "eval_runtime": 19.3153,
413
+ "eval_samples_per_second": 51.773,
414
+ "eval_steps_per_second": 0.828,
415
+ "step": 540
416
+ },
417
+ {
418
+ "epoch": 0.4593929450369155,
419
+ "grad_norm": 0.1652340292930603,
420
+ "learning_rate": 0.0001899990292425471,
421
+ "loss": 0.2831,
422
+ "step": 560
423
+ },
424
+ {
425
+ "epoch": 0.4593929450369155,
426
+ "eval_loss": 0.273937463760376,
427
+ "eval_runtime": 19.308,
428
+ "eval_samples_per_second": 51.792,
429
+ "eval_steps_per_second": 0.829,
430
+ "step": 560
431
+ },
432
+ {
433
+ "epoch": 0.47579983593109104,
434
+ "grad_norm": 0.1996350735425949,
435
+ "learning_rate": 0.0001887139361319243,
436
+ "loss": 0.278,
437
+ "step": 580
438
+ },
439
+ {
440
+ "epoch": 0.47579983593109104,
441
+ "eval_loss": 0.27432310581207275,
442
+ "eval_runtime": 19.3126,
443
+ "eval_samples_per_second": 51.78,
444
+ "eval_steps_per_second": 0.828,
445
+ "step": 580
446
+ },
447
+ {
448
+ "epoch": 0.4922067268252666,
449
+ "grad_norm": 0.21539440751075745,
450
+ "learning_rate": 0.0001873560903980955,
451
+ "loss": 0.2835,
452
+ "step": 600
453
+ },
454
+ {
455
+ "epoch": 0.4922067268252666,
456
+ "eval_loss": 0.2731659412384033,
457
+ "eval_runtime": 19.3354,
458
+ "eval_samples_per_second": 51.719,
459
+ "eval_steps_per_second": 0.827,
460
+ "step": 600
461
+ },
462
+ {
463
+ "epoch": 0.5086136177194421,
464
+ "grad_norm": 0.17816652357578278,
465
+ "learning_rate": 0.0001859266055846971,
466
+ "loss": 0.272,
467
+ "step": 620
468
+ },
469
+ {
470
+ "epoch": 0.5086136177194421,
471
+ "eval_loss": 0.27160513401031494,
472
+ "eval_runtime": 19.324,
473
+ "eval_samples_per_second": 51.749,
474
+ "eval_steps_per_second": 0.828,
475
+ "step": 620
476
+ },
477
+ {
478
+ "epoch": 0.5250205086136177,
479
+ "grad_norm": 0.2001989185810089,
480
+ "learning_rate": 0.00018442665398522149,
481
+ "loss": 0.2744,
482
+ "step": 640
483
+ },
484
+ {
485
+ "epoch": 0.5250205086136177,
486
+ "eval_loss": 0.2692422568798065,
487
+ "eval_runtime": 19.3251,
488
+ "eval_samples_per_second": 51.746,
489
+ "eval_steps_per_second": 0.828,
490
+ "step": 640
491
+ },
492
+ {
493
+ "epoch": 0.5414273995077933,
494
+ "grad_norm": 0.21018487215042114,
495
+ "learning_rate": 0.0001828574656816411,
496
+ "loss": 0.2736,
497
+ "step": 660
498
+ },
499
+ {
500
+ "epoch": 0.5414273995077933,
501
+ "eval_loss": 0.2677000164985657,
502
+ "eval_runtime": 19.3297,
503
+ "eval_samples_per_second": 51.734,
504
+ "eval_steps_per_second": 0.828,
505
+ "step": 660
506
+ },
507
+ {
508
+ "epoch": 0.5578342904019689,
509
+ "grad_norm": 0.21279104053974152,
510
+ "learning_rate": 0.00018122032753564157,
511
+ "loss": 0.2718,
512
+ "step": 680
513
+ },
514
+ {
515
+ "epoch": 0.5578342904019689,
516
+ "eval_loss": 0.26908594369888306,
517
+ "eval_runtime": 19.3302,
518
+ "eval_samples_per_second": 51.733,
519
+ "eval_steps_per_second": 0.828,
520
+ "step": 680
521
+ },
522
+ {
523
+ "epoch": 0.5742411812961444,
524
+ "grad_norm": 0.1887700855731964,
525
+ "learning_rate": 0.0001795165821332908,
526
+ "loss": 0.2678,
527
+ "step": 700
528
+ },
529
+ {
530
+ "epoch": 0.5742411812961444,
531
+ "eval_loss": 0.26670950651168823,
532
+ "eval_runtime": 19.3168,
533
+ "eval_samples_per_second": 51.768,
534
+ "eval_steps_per_second": 0.828,
535
+ "step": 700
536
+ },
537
+ {
538
+ "epoch": 0.5906480721903199,
539
+ "grad_norm": 0.19008482992649078,
540
+ "learning_rate": 0.0001777476266840092,
541
+ "loss": 0.2646,
542
+ "step": 720
543
+ },
544
+ {
545
+ "epoch": 0.5906480721903199,
546
+ "eval_loss": 0.265604168176651,
547
+ "eval_runtime": 19.3108,
548
+ "eval_samples_per_second": 51.785,
549
+ "eval_steps_per_second": 0.829,
550
+ "step": 720
551
+ },
552
+ {
553
+ "epoch": 0.6070549630844955,
554
+ "grad_norm": 0.16397179663181305,
555
+ "learning_rate": 0.0001759149118747451,
556
+ "loss": 0.2701,
557
+ "step": 740
558
+ },
559
+ {
560
+ "epoch": 0.6070549630844955,
561
+ "eval_loss": 0.2638825476169586,
562
+ "eval_runtime": 19.3328,
563
+ "eval_samples_per_second": 51.725,
564
+ "eval_steps_per_second": 0.828,
565
+ "step": 740
566
+ },
567
+ {
568
+ "epoch": 0.623461853978671,
569
+ "grad_norm": 0.1907324194908142,
570
+ "learning_rate": 0.0001740199406802933,
571
+ "loss": 0.2761,
572
+ "step": 760
573
+ },
574
+ {
575
+ "epoch": 0.623461853978671,
576
+ "eval_loss": 0.2637205719947815,
577
+ "eval_runtime": 19.3301,
578
+ "eval_samples_per_second": 51.733,
579
+ "eval_steps_per_second": 0.828,
580
+ "step": 760
581
+ },
582
+ {
583
+ "epoch": 0.6398687448728466,
584
+ "grad_norm": 0.21579375863075256,
585
+ "learning_rate": 0.00017206426713073416,
586
+ "loss": 0.2721,
587
+ "step": 780
588
+ },
589
+ {
590
+ "epoch": 0.6398687448728466,
591
+ "eval_loss": 0.26291656494140625,
592
+ "eval_runtime": 19.3095,
593
+ "eval_samples_per_second": 51.788,
594
+ "eval_steps_per_second": 0.829,
595
+ "step": 780
596
+ },
597
+ {
598
+ "epoch": 0.6562756357670222,
599
+ "grad_norm": 0.20866331458091736,
600
+ "learning_rate": 0.00017004949503700284,
601
+ "loss": 0.2729,
602
+ "step": 800
603
+ },
604
+ {
605
+ "epoch": 0.6562756357670222,
606
+ "eval_loss": 0.2615169584751129,
607
+ "eval_runtime": 19.319,
608
+ "eval_samples_per_second": 51.762,
609
+ "eval_steps_per_second": 0.828,
610
+ "step": 800
611
+ },
612
+ {
613
+ "epoch": 0.6726825266611977,
614
+ "grad_norm": 0.18404333293437958,
615
+ "learning_rate": 0.00016797727667563415,
616
+ "loss": 0.2617,
617
+ "step": 820
618
+ },
619
+ {
620
+ "epoch": 0.6726825266611977,
621
+ "eval_loss": 0.25983768701553345,
622
+ "eval_runtime": 19.3291,
623
+ "eval_samples_per_second": 51.736,
624
+ "eval_steps_per_second": 0.828,
625
+ "step": 820
626
+ },
627
+ {
628
+ "epoch": 0.6890894175553732,
629
+ "grad_norm": 0.17301003634929657,
630
+ "learning_rate": 0.00016584931143376176,
631
+ "loss": 0.2737,
632
+ "step": 840
633
+ },
634
+ {
635
+ "epoch": 0.6890894175553732,
636
+ "eval_loss": 0.2589709460735321,
637
+ "eval_runtime": 19.3364,
638
+ "eval_samples_per_second": 51.716,
639
+ "eval_steps_per_second": 0.827,
640
+ "step": 840
641
+ },
642
+ {
643
+ "epoch": 0.7054963084495488,
644
+ "grad_norm": 0.18135105073451996,
645
+ "learning_rate": 0.00016366734441548313,
646
+ "loss": 0.2778,
647
+ "step": 860
648
+ },
649
+ {
650
+ "epoch": 0.7054963084495488,
651
+ "eval_loss": 0.2588275671005249,
652
+ "eval_runtime": 19.3258,
653
+ "eval_samples_per_second": 51.744,
654
+ "eval_steps_per_second": 0.828,
655
+ "step": 860
656
+ },
657
+ {
658
+ "epoch": 0.7219031993437244,
659
+ "grad_norm": 0.1769489347934723,
660
+ "learning_rate": 0.0001614331650107323,
661
+ "loss": 0.2699,
662
+ "step": 880
663
+ },
664
+ {
665
+ "epoch": 0.7219031993437244,
666
+ "eval_loss": 0.2579975426197052,
667
+ "eval_runtime": 19.3349,
668
+ "eval_samples_per_second": 51.72,
669
+ "eval_steps_per_second": 0.828,
670
+ "step": 880
671
+ },
672
+ {
673
+ "epoch": 0.7383100902379,
674
+ "grad_norm": 0.18334627151489258,
675
+ "learning_rate": 0.00015914860542783524,
676
+ "loss": 0.2621,
677
+ "step": 900
678
+ },
679
+ {
680
+ "epoch": 0.7383100902379,
681
+ "eval_loss": 0.2564004957675934,
682
+ "eval_runtime": 19.3217,
683
+ "eval_samples_per_second": 51.755,
684
+ "eval_steps_per_second": 0.828,
685
+ "step": 900
686
+ },
687
+ {
688
+ "epoch": 0.7547169811320755,
689
+ "grad_norm": 0.17185646295547485,
690
+ "learning_rate": 0.00015681553919095025,
691
+ "loss": 0.2661,
692
+ "step": 920
693
+ },
694
+ {
695
+ "epoch": 0.7547169811320755,
696
+ "eval_loss": 0.2558245062828064,
697
+ "eval_runtime": 19.3184,
698
+ "eval_samples_per_second": 51.764,
699
+ "eval_steps_per_second": 0.828,
700
+ "step": 920
701
+ },
702
+ {
703
+ "epoch": 0.771123872026251,
704
+ "grad_norm": 0.17241282761096954,
705
+ "learning_rate": 0.0001544358796036264,
706
+ "loss": 0.25,
707
+ "step": 940
708
+ },
709
+ {
710
+ "epoch": 0.771123872026251,
711
+ "eval_loss": 0.254745215177536,
712
+ "eval_runtime": 19.3361,
713
+ "eval_samples_per_second": 51.717,
714
+ "eval_steps_per_second": 0.827,
715
+ "step": 940
716
+ },
717
+ {
718
+ "epoch": 0.7875307629204266,
719
+ "grad_norm": 0.19575953483581543,
720
+ "learning_rate": 0.00015201157817973924,
721
+ "loss": 0.2583,
722
+ "step": 960
723
+ },
724
+ {
725
+ "epoch": 0.7875307629204266,
726
+ "eval_loss": 0.2538128197193146,
727
+ "eval_runtime": 19.3314,
728
+ "eval_samples_per_second": 51.729,
729
+ "eval_steps_per_second": 0.828,
730
+ "step": 960
731
+ },
732
+ {
733
+ "epoch": 0.8039376538146021,
734
+ "grad_norm": 0.18103253841400146,
735
+ "learning_rate": 0.00014954462304309115,
736
+ "loss": 0.2662,
737
+ "step": 980
738
+ },
739
+ {
740
+ "epoch": 0.8039376538146021,
741
+ "eval_loss": 0.25316470861434937,
742
+ "eval_runtime": 19.3326,
743
+ "eval_samples_per_second": 51.726,
744
+ "eval_steps_per_second": 0.828,
745
+ "step": 980
746
+ },
747
+ {
748
+ "epoch": 0.8203445447087777,
749
+ "grad_norm": 0.20527029037475586,
750
+ "learning_rate": 0.0001470370372969886,
751
+ "loss": 0.2657,
752
+ "step": 1000
753
+ },
754
+ {
755
+ "epoch": 0.8203445447087777,
756
+ "eval_loss": 0.2526797652244568,
757
+ "eval_runtime": 19.3278,
758
+ "eval_samples_per_second": 51.739,
759
+ "eval_steps_per_second": 0.828,
760
+ "step": 1000
761
+ },
762
+ {
763
+ "epoch": 0.8367514356029533,
764
+ "grad_norm": 0.18703380227088928,
765
+ "learning_rate": 0.00014449087736513347,
766
+ "loss": 0.2581,
767
+ "step": 1020
768
+ },
769
+ {
770
+ "epoch": 0.8367514356029533,
771
+ "eval_loss": 0.2514651119709015,
772
+ "eval_runtime": 19.2682,
773
+ "eval_samples_per_second": 51.899,
774
+ "eval_steps_per_second": 0.83,
775
+ "step": 1020
776
+ },
777
+ {
778
+ "epoch": 0.8531583264971287,
779
+ "grad_norm": 0.17955496907234192,
780
+ "learning_rate": 0.00014190823130518868,
781
+ "loss": 0.2546,
782
+ "step": 1040
783
+ },
784
+ {
785
+ "epoch": 0.8531583264971287,
786
+ "eval_loss": 0.25145697593688965,
787
+ "eval_runtime": 19.2884,
788
+ "eval_samples_per_second": 51.845,
789
+ "eval_steps_per_second": 0.83,
790
+ "step": 1040
791
+ },
792
+ {
793
+ "epoch": 0.8695652173913043,
794
+ "grad_norm": 0.19111481308937073,
795
+ "learning_rate": 0.00013929121709640142,
796
+ "loss": 0.2571,
797
+ "step": 1060
798
+ },
799
+ {
800
+ "epoch": 0.8695652173913043,
801
+ "eval_loss": 0.250641793012619,
802
+ "eval_runtime": 19.311,
803
+ "eval_samples_per_second": 51.784,
804
+ "eval_steps_per_second": 0.829,
805
+ "step": 1060
806
+ },
807
+ {
808
+ "epoch": 0.8859721082854799,
809
+ "grad_norm": 0.1683422178030014,
810
+ "learning_rate": 0.00013664198090268873,
811
+ "loss": 0.2668,
812
+ "step": 1080
813
+ },
814
+ {
815
+ "epoch": 0.8859721082854799,
816
+ "eval_loss": 0.2503809928894043,
817
+ "eval_runtime": 19.3082,
818
+ "eval_samples_per_second": 51.791,
819
+ "eval_steps_per_second": 0.829,
820
+ "step": 1080
821
+ },
822
+ {
823
+ "epoch": 0.9023789991796555,
824
+ "grad_norm": 0.1857733279466629,
825
+ "learning_rate": 0.00013396269531260868,
826
+ "loss": 0.254,
827
+ "step": 1100
828
+ },
829
+ {
830
+ "epoch": 0.9023789991796555,
831
+ "eval_loss": 0.2492498904466629,
832
+ "eval_runtime": 19.3165,
833
+ "eval_samples_per_second": 51.769,
834
+ "eval_steps_per_second": 0.828,
835
+ "step": 1100
836
+ },
837
+ {
838
+ "epoch": 0.918785890073831,
839
+ "grad_norm": 0.18497365713119507,
840
+ "learning_rate": 0.00013125555755766173,
841
+ "loss": 0.2526,
842
+ "step": 1120
843
+ },
844
+ {
845
+ "epoch": 0.918785890073831,
846
+ "eval_loss": 0.248539000749588,
847
+ "eval_runtime": 19.3245,
848
+ "eval_samples_per_second": 51.748,
849
+ "eval_steps_per_second": 0.828,
850
+ "step": 1120
851
+ },
852
+ {
853
+ "epoch": 0.9351927809680065,
854
+ "grad_norm": 0.18615785241127014,
855
+ "learning_rate": 0.00012852278771038252,
856
+ "loss": 0.2598,
857
+ "step": 1140
858
+ },
859
+ {
860
+ "epoch": 0.9351927809680065,
861
+ "eval_loss": 0.24817132949829102,
862
+ "eval_runtime": 19.3102,
863
+ "eval_samples_per_second": 51.786,
864
+ "eval_steps_per_second": 0.829,
865
+ "step": 1140
866
+ },
867
+ {
868
+ "epoch": 0.9515996718621821,
869
+ "grad_norm": 0.19807547330856323,
870
+ "learning_rate": 0.00012576662686370054,
871
+ "loss": 0.2714,
872
+ "step": 1160
873
+ },
874
+ {
875
+ "epoch": 0.9515996718621821,
876
+ "eval_loss": 0.24792517721652985,
877
+ "eval_runtime": 19.3192,
878
+ "eval_samples_per_second": 51.762,
879
+ "eval_steps_per_second": 0.828,
880
+ "step": 1160
881
+ },
882
+ {
883
+ "epoch": 0.9680065627563577,
884
+ "grad_norm": 0.2258663922548294,
885
+ "learning_rate": 0.000122989335293062,
886
+ "loss": 0.2518,
887
+ "step": 1180
888
+ },
889
+ {
890
+ "epoch": 0.9680065627563577,
891
+ "eval_loss": 0.2466147094964981,
892
+ "eval_runtime": 19.3075,
893
+ "eval_samples_per_second": 51.793,
894
+ "eval_steps_per_second": 0.829,
895
+ "step": 1180
896
+ },
897
+ {
898
+ "epoch": 0.9844134536505332,
899
+ "grad_norm": 0.1830626130104065,
900
+ "learning_rate": 0.00012019319060282064,
901
+ "loss": 0.2598,
902
+ "step": 1200
903
+ },
904
+ {
905
+ "epoch": 0.9844134536505332,
906
+ "eval_loss": 0.24613501131534576,
907
+ "eval_runtime": 19.3331,
908
+ "eval_samples_per_second": 51.725,
909
+ "eval_steps_per_second": 0.828,
910
+ "step": 1200
911
+ },
912
+ {
913
+ "epoch": 1.0008203445447088,
914
+ "grad_norm": 0.19599111378192902,
915
+ "learning_rate": 0.00011738048585841741,
916
+ "loss": 0.2659,
917
+ "step": 1220
918
+ },
919
+ {
920
+ "epoch": 1.0008203445447088,
921
+ "eval_loss": 0.24604985117912292,
922
+ "eval_runtime": 19.2705,
923
+ "eval_samples_per_second": 51.893,
924
+ "eval_steps_per_second": 0.83,
925
+ "step": 1220
926
+ },
927
+ {
928
+ "epoch": 1.0172272354388843,
929
+ "grad_norm": 0.17713318765163422,
930
+ "learning_rate": 0.00011455352770588094,
931
+ "loss": 0.2321,
932
+ "step": 1240
933
+ },
934
+ {
935
+ "epoch": 1.0172272354388843,
936
+ "eval_loss": 0.24518859386444092,
937
+ "eval_runtime": 19.3371,
938
+ "eval_samples_per_second": 51.714,
939
+ "eval_steps_per_second": 0.827,
940
+ "step": 1240
941
+ },
942
+ {
943
+ "epoch": 1.03363412633306,
944
+ "grad_norm": 0.19879887998104095,
945
+ "learning_rate": 0.00011171463448019059,
946
+ "loss": 0.2403,
947
+ "step": 1260
948
+ },
949
+ {
950
+ "epoch": 1.03363412633306,
951
+ "eval_loss": 0.24396194517612457,
952
+ "eval_runtime": 19.3373,
953
+ "eval_samples_per_second": 51.713,
954
+ "eval_steps_per_second": 0.827,
955
+ "step": 1260
956
+ },
957
+ {
958
+ "epoch": 1.0500410172272354,
959
+ "grad_norm": 0.20000441372394562,
960
+ "learning_rate": 0.0001088661343040537,
961
+ "loss": 0.24,
962
+ "step": 1280
963
+ },
964
+ {
965
+ "epoch": 1.0500410172272354,
966
+ "eval_loss": 0.24362273514270782,
967
+ "eval_runtime": 19.3082,
968
+ "eval_samples_per_second": 51.792,
969
+ "eval_steps_per_second": 0.829,
970
+ "step": 1280
971
+ },
972
+ {
973
+ "epoch": 1.066447908121411,
974
+ "grad_norm": 0.1881808638572693,
975
+ "learning_rate": 0.0001060103631786563,
976
+ "loss": 0.234,
977
+ "step": 1300
978
+ },
979
+ {
980
+ "epoch": 1.066447908121411,
981
+ "eval_loss": 0.2430313378572464,
982
+ "eval_runtime": 19.3284,
983
+ "eval_samples_per_second": 51.737,
984
+ "eval_steps_per_second": 0.828,
985
+ "step": 1300
986
+ },
987
+ {
988
+ "epoch": 1.0828547990155866,
989
+ "grad_norm": 0.22585070133209229,
990
+ "learning_rate": 0.00010314966306795251,
991
+ "loss": 0.2299,
992
+ "step": 1320
993
+ },
994
+ {
995
+ "epoch": 1.0828547990155866,
996
+ "eval_loss": 0.24361206591129303,
997
+ "eval_runtime": 19.3229,
998
+ "eval_samples_per_second": 51.752,
999
+ "eval_steps_per_second": 0.828,
1000
+ "step": 1320
1001
+ },
1002
+ {
1003
+ "epoch": 1.099261689909762,
1004
+ "grad_norm": 0.18805377185344696,
1005
+ "learning_rate": 0.00010028637997806413,
1006
+ "loss": 0.2396,
1007
+ "step": 1340
1008
+ },
1009
+ {
1010
+ "epoch": 1.099261689909762,
1011
+ "eval_loss": 0.24354341626167297,
1012
+ "eval_runtime": 19.3136,
1013
+ "eval_samples_per_second": 51.777,
1014
+ "eval_steps_per_second": 0.828,
1015
+ "step": 1340
1016
+ },
1017
+ {
1018
+ "epoch": 1.1156685808039377,
1019
+ "grad_norm": 0.18671385943889618,
1020
+ "learning_rate": 9.742286203336551e-05,
1021
+ "loss": 0.2404,
1022
+ "step": 1360
1023
+ },
1024
+ {
1025
+ "epoch": 1.1156685808039377,
1026
+ "eval_loss": 0.24287469685077667,
1027
+ "eval_runtime": 19.3223,
1028
+ "eval_samples_per_second": 51.754,
1029
+ "eval_steps_per_second": 0.828,
1030
+ "step": 1360
1031
+ },
1032
+ {
1033
+ "epoch": 1.1320754716981132,
1034
+ "grad_norm": 0.20467591285705566,
1035
+ "learning_rate": 9.45614575508309e-05,
1036
+ "loss": 0.2371,
1037
+ "step": 1380
1038
+ },
1039
+ {
1040
+ "epoch": 1.1320754716981132,
1041
+ "eval_loss": 0.2421674132347107,
1042
+ "eval_runtime": 19.3152,
1043
+ "eval_samples_per_second": 51.773,
1044
+ "eval_steps_per_second": 0.828,
1045
+ "step": 1380
1046
+ },
1047
+ {
1048
+ "epoch": 1.1484823625922886,
1049
+ "grad_norm": 0.1996966153383255,
1050
+ "learning_rate": 9.170451311422421e-05,
1051
+ "loss": 0.2435,
1052
+ "step": 1400
1053
+ },
1054
+ {
1055
+ "epoch": 1.1484823625922886,
1056
+ "eval_loss": 0.24120301008224487,
1057
+ "eval_runtime": 19.3108,
1058
+ "eval_samples_per_second": 51.785,
1059
+ "eval_steps_per_second": 0.829,
1060
+ "step": 1400
1061
+ },
1062
+ {
1063
+ "epoch": 1.1648892534864643,
1064
+ "grad_norm": 0.17993618547916412,
1065
+ "learning_rate": 8.88543716497099e-05,
1066
+ "loss": 0.2319,
1067
+ "step": 1420
1068
+ },
1069
+ {
1070
+ "epoch": 1.1648892534864643,
1071
+ "eval_loss": 0.2416684776544571,
1072
+ "eval_runtime": 19.3207,
1073
+ "eval_samples_per_second": 51.758,
1074
+ "eval_steps_per_second": 0.828,
1075
+ "step": 1420
1076
+ },
1077
+ {
1078
+ "epoch": 1.1812961443806398,
1079
+ "grad_norm": 0.20344999432563782,
1080
+ "learning_rate": 8.601337050446355e-05,
1081
+ "loss": 0.2263,
1082
+ "step": 1440
1083
+ },
1084
+ {
1085
+ "epoch": 1.1812961443806398,
1086
+ "eval_loss": 0.2408376783132553,
1087
+ "eval_runtime": 19.316,
1088
+ "eval_samples_per_second": 51.77,
1089
+ "eval_steps_per_second": 0.828,
1090
+ "step": 1440
1091
+ },
1092
+ {
1093
+ "epoch": 1.1977030352748155,
1094
+ "grad_norm": 0.19167408347129822,
1095
+ "learning_rate": 8.31838395298576e-05,
1096
+ "loss": 0.2415,
1097
+ "step": 1460
1098
+ },
1099
+ {
1100
+ "epoch": 1.1977030352748155,
1101
+ "eval_loss": 0.24088580906391144,
1102
+ "eval_runtime": 19.3158,
1103
+ "eval_samples_per_second": 51.771,
1104
+ "eval_steps_per_second": 0.828,
1105
+ "step": 1460
1106
+ },
1107
+ {
1108
+ "epoch": 1.214109926168991,
1109
+ "grad_norm": 0.1993206888437271,
1110
+ "learning_rate": 8.036809917079411e-05,
1111
+ "loss": 0.2409,
1112
+ "step": 1480
1113
+ },
1114
+ {
1115
+ "epoch": 1.214109926168991,
1116
+ "eval_loss": 0.23997968435287476,
1117
+ "eval_runtime": 19.3306,
1118
+ "eval_samples_per_second": 51.731,
1119
+ "eval_steps_per_second": 0.828,
1120
+ "step": 1480
1121
+ },
1122
+ {
1123
+ "epoch": 1.2305168170631666,
1124
+ "grad_norm": 0.22202400863170624,
1125
+ "learning_rate": 7.756845856275194e-05,
1126
+ "loss": 0.2375,
1127
+ "step": 1500
1128
+ },
1129
+ {
1130
+ "epoch": 1.2305168170631666,
1131
+ "eval_loss": 0.24000239372253418,
1132
+ "eval_runtime": 19.3092,
1133
+ "eval_samples_per_second": 51.789,
1134
+ "eval_steps_per_second": 0.829,
1135
+ "step": 1500
1136
+ },
1137
+ {
1138
+ "epoch": 1.246923707957342,
1139
+ "grad_norm": 0.21304291486740112,
1140
+ "learning_rate": 7.478721363810825e-05,
1141
+ "loss": 0.2405,
1142
+ "step": 1520
1143
+ },
1144
+ {
1145
+ "epoch": 1.246923707957342,
1146
+ "eval_loss": 0.23960961401462555,
1147
+ "eval_runtime": 19.2639,
1148
+ "eval_samples_per_second": 51.911,
1149
+ "eval_steps_per_second": 0.831,
1150
+ "step": 1520
1151
+ },
1152
+ {
1153
+ "epoch": 1.2633305988515175,
1154
+ "grad_norm": 0.19039390981197357,
1155
+ "learning_rate": 7.20266452432877e-05,
1156
+ "loss": 0.2297,
1157
+ "step": 1540
1158
+ },
1159
+ {
1160
+ "epoch": 1.2633305988515175,
1161
+ "eval_loss": 0.23886562883853912,
1162
+ "eval_runtime": 19.3147,
1163
+ "eval_samples_per_second": 51.774,
1164
+ "eval_steps_per_second": 0.828,
1165
+ "step": 1540
1166
+ },
1167
+ {
1168
+ "epoch": 1.2797374897456932,
1169
+ "grad_norm": 0.22386115789413452,
1170
+ "learning_rate": 6.92890172682835e-05,
1171
+ "loss": 0.2369,
1172
+ "step": 1560
1173
+ },
1174
+ {
1175
+ "epoch": 1.2797374897456932,
1176
+ "eval_loss": 0.2387980967760086,
1177
+ "eval_runtime": 19.3174,
1178
+ "eval_samples_per_second": 51.767,
1179
+ "eval_steps_per_second": 0.828,
1180
+ "step": 1560
1181
+ },
1182
+ {
1183
+ "epoch": 1.2961443806398687,
1184
+ "grad_norm": 0.2163139134645462,
1185
+ "learning_rate": 6.65765747900838e-05,
1186
+ "loss": 0.2315,
1187
+ "step": 1580
1188
+ },
1189
+ {
1190
+ "epoch": 1.2961443806398687,
1191
+ "eval_loss": 0.23805563151836395,
1192
+ "eval_runtime": 19.3198,
1193
+ "eval_samples_per_second": 51.76,
1194
+ "eval_steps_per_second": 0.828,
1195
+ "step": 1580
1196
+ },
1197
+ {
1198
+ "epoch": 1.3125512715340442,
1199
+ "grad_norm": 0.19729720056056976,
1200
+ "learning_rate": 6.389154223152666e-05,
1201
+ "loss": 0.2337,
1202
+ "step": 1600
1203
+ },
1204
+ {
1205
+ "epoch": 1.3125512715340442,
1206
+ "eval_loss": 0.23739255964756012,
1207
+ "eval_runtime": 19.2969,
1208
+ "eval_samples_per_second": 51.822,
1209
+ "eval_steps_per_second": 0.829,
1210
+ "step": 1600
1211
+ },
1212
+ {
1213
+ "epoch": 1.3289581624282198,
1214
+ "grad_norm": 0.2069961577653885,
1215
+ "learning_rate": 6.123612153709288e-05,
1216
+ "loss": 0.2284,
1217
+ "step": 1620
1218
+ },
1219
+ {
1220
+ "epoch": 1.3289581624282198,
1221
+ "eval_loss": 0.2374027967453003,
1222
+ "eval_runtime": 19.3195,
1223
+ "eval_samples_per_second": 51.761,
1224
+ "eval_steps_per_second": 0.828,
1225
+ "step": 1620
1226
+ },
1227
+ {
1228
+ "epoch": 1.3453650533223955,
1229
+ "grad_norm": 0.22076573967933655,
1230
+ "learning_rate": 5.8612490367132656e-05,
1231
+ "loss": 0.2319,
1232
+ "step": 1640
1233
+ },
1234
+ {
1235
+ "epoch": 1.3453650533223955,
1236
+ "eval_loss": 0.23642553389072418,
1237
+ "eval_runtime": 19.3294,
1238
+ "eval_samples_per_second": 51.735,
1239
+ "eval_steps_per_second": 0.828,
1240
+ "step": 1640
1241
+ },
1242
+ {
1243
+ "epoch": 1.361771944216571,
1244
+ "grad_norm": 0.21753518283367157,
1245
+ "learning_rate": 5.602280031200784e-05,
1246
+ "loss": 0.2216,
1247
+ "step": 1660
1248
+ },
1249
+ {
1250
+ "epoch": 1.361771944216571,
1251
+ "eval_loss": 0.2362075001001358,
1252
+ "eval_runtime": 19.3212,
1253
+ "eval_samples_per_second": 51.757,
1254
+ "eval_steps_per_second": 0.828,
1255
+ "step": 1660
1256
+ },
1257
+ {
1258
+ "epoch": 1.3781788351107465,
1259
+ "grad_norm": 0.19170460104942322,
1260
+ "learning_rate": 5.346917512761322e-05,
1261
+ "loss": 0.2265,
1262
+ "step": 1680
1263
+ },
1264
+ {
1265
+ "epoch": 1.3781788351107465,
1266
+ "eval_loss": 0.23582229018211365,
1267
+ "eval_runtime": 19.3226,
1268
+ "eval_samples_per_second": 51.753,
1269
+ "eval_steps_per_second": 0.828,
1270
+ "step": 1680
1271
+ },
1272
+ {
1273
+ "epoch": 1.3945857260049221,
1274
+ "grad_norm": 0.20841749012470245,
1275
+ "learning_rate": 5.0953708993724114e-05,
1276
+ "loss": 0.2314,
1277
+ "step": 1700
1278
+ },
1279
+ {
1280
+ "epoch": 1.3945857260049221,
1281
+ "eval_loss": 0.23545880615711212,
1282
+ "eval_runtime": 19.3033,
1283
+ "eval_samples_per_second": 51.805,
1284
+ "eval_steps_per_second": 0.829,
1285
+ "step": 1700
1286
+ },
1287
+ {
1288
+ "epoch": 1.4109926168990976,
1289
+ "grad_norm": 0.23843340575695038,
1290
+ "learning_rate": 4.847846479659961e-05,
1291
+ "loss": 0.2413,
1292
+ "step": 1720
1293
+ },
1294
+ {
1295
+ "epoch": 1.4109926168990976,
1296
+ "eval_loss": 0.23550079762935638,
1297
+ "eval_runtime": 19.3112,
1298
+ "eval_samples_per_second": 51.783,
1299
+ "eval_steps_per_second": 0.829,
1300
+ "step": 1720
1301
+ },
1302
+ {
1303
+ "epoch": 1.427399507793273,
1304
+ "grad_norm": 0.20170091092586517,
1305
+ "learning_rate": 4.604547243724809e-05,
1306
+ "loss": 0.2412,
1307
+ "step": 1740
1308
+ },
1309
+ {
1310
+ "epoch": 1.427399507793273,
1311
+ "eval_loss": 0.23475277423858643,
1312
+ "eval_runtime": 19.3177,
1313
+ "eval_samples_per_second": 51.766,
1314
+ "eval_steps_per_second": 0.828,
1315
+ "step": 1740
1316
+ },
1317
+ {
1318
+ "epoch": 1.4438063986874488,
1319
+ "grad_norm": 0.20738761126995087,
1320
+ "learning_rate": 4.365672716674395e-05,
1321
+ "loss": 0.2346,
1322
+ "step": 1760
1323
+ },
1324
+ {
1325
+ "epoch": 1.4438063986874488,
1326
+ "eval_loss": 0.2345096319913864,
1327
+ "eval_runtime": 19.3258,
1328
+ "eval_samples_per_second": 51.744,
1329
+ "eval_steps_per_second": 0.828,
1330
+ "step": 1760
1331
+ },
1332
+ {
1333
+ "epoch": 1.4602132895816242,
1334
+ "grad_norm": 0.23127228021621704,
1335
+ "learning_rate": 4.1314187949959996e-05,
1336
+ "loss": 0.229,
1337
+ "step": 1780
1338
+ },
1339
+ {
1340
+ "epoch": 1.4602132895816242,
1341
+ "eval_loss": 0.23429767787456512,
1342
+ "eval_runtime": 19.3185,
1343
+ "eval_samples_per_second": 51.764,
1344
+ "eval_steps_per_second": 0.828,
1345
+ "step": 1780
1346
+ },
1347
+ {
1348
+ "epoch": 1.4766201804758,
1349
+ "grad_norm": 0.20418289303779602,
1350
+ "learning_rate": 3.901977585905692e-05,
1351
+ "loss": 0.239,
1352
+ "step": 1800
1353
+ },
1354
+ {
1355
+ "epoch": 1.4766201804758,
1356
+ "eval_loss": 0.23406723141670227,
1357
+ "eval_runtime": 19.3265,
1358
+ "eval_samples_per_second": 51.742,
1359
+ "eval_steps_per_second": 0.828,
1360
+ "step": 1800
1361
+ },
1362
+ {
1363
+ "epoch": 1.4930270713699754,
1364
+ "grad_norm": 0.2218182235956192,
1365
+ "learning_rate": 3.677537249804891e-05,
1366
+ "loss": 0.2253,
1367
+ "step": 1820
1368
+ },
1369
+ {
1370
+ "epoch": 1.4930270713699754,
1371
+ "eval_loss": 0.23370110988616943,
1372
+ "eval_runtime": 19.3173,
1373
+ "eval_samples_per_second": 51.767,
1374
+ "eval_steps_per_second": 0.828,
1375
+ "step": 1820
1376
+ },
1377
+ {
1378
+ "epoch": 1.509433962264151,
1379
+ "grad_norm": 0.2059129923582077,
1380
+ "learning_rate": 3.458281845973528e-05,
1381
+ "loss": 0.2355,
1382
+ "step": 1840
1383
+ },
1384
+ {
1385
+ "epoch": 1.509433962264151,
1386
+ "eval_loss": 0.23322786390781403,
1387
+ "eval_runtime": 19.3204,
1388
+ "eval_samples_per_second": 51.759,
1389
+ "eval_steps_per_second": 0.828,
1390
+ "step": 1840
1391
+ },
1392
+ {
1393
+ "epoch": 1.5258408531583265,
1394
+ "grad_norm": 0.19780805706977844,
1395
+ "learning_rate": 3.2443911816265214e-05,
1396
+ "loss": 0.2327,
1397
+ "step": 1860
1398
+ },
1399
+ {
1400
+ "epoch": 1.5258408531583265,
1401
+ "eval_loss": 0.23293595016002655,
1402
+ "eval_runtime": 19.3265,
1403
+ "eval_samples_per_second": 51.742,
1404
+ "eval_steps_per_second": 0.828,
1405
+ "step": 1860
1406
+ },
1407
+ {
1408
+ "epoch": 1.542247744052502,
1409
+ "grad_norm": 0.22608228027820587,
1410
+ "learning_rate": 3.0360406644572935e-05,
1411
+ "loss": 0.2259,
1412
+ "step": 1880
1413
+ },
1414
+ {
1415
+ "epoch": 1.542247744052502,
1416
+ "eval_loss": 0.23265497386455536,
1417
+ "eval_runtime": 19.3159,
1418
+ "eval_samples_per_second": 51.771,
1419
+ "eval_steps_per_second": 0.828,
1420
+ "step": 1880
1421
+ },
1422
+ {
1423
+ "epoch": 1.5586546349466777,
1424
+ "grad_norm": 0.20242735743522644,
1425
+ "learning_rate": 2.8334011587892074e-05,
1426
+ "loss": 0.2277,
1427
+ "step": 1900
1428
+ },
1429
+ {
1430
+ "epoch": 1.5586546349466777,
1431
+ "eval_loss": 0.2327563762664795,
1432
+ "eval_runtime": 19.3493,
1433
+ "eval_samples_per_second": 51.681,
1434
+ "eval_steps_per_second": 0.827,
1435
+ "step": 1900
1436
+ },
1437
+ {
1438
+ "epoch": 1.5750615258408531,
1439
+ "grad_norm": 0.2113717496395111,
1440
+ "learning_rate": 2.6366388454529788e-05,
1441
+ "loss": 0.2312,
1442
+ "step": 1920
1443
+ },
1444
+ {
1445
+ "epoch": 1.5750615258408531,
1446
+ "eval_loss": 0.23235367238521576,
1447
+ "eval_runtime": 19.3377,
1448
+ "eval_samples_per_second": 51.712,
1449
+ "eval_steps_per_second": 0.827,
1450
+ "step": 1920
1451
+ },
1452
+ {
1453
+ "epoch": 1.5914684167350286,
1454
+ "grad_norm": 0.21388964354991913,
1455
+ "learning_rate": 2.445915085504915e-05,
1456
+ "loss": 0.2332,
1457
+ "step": 1940
1458
+ },
1459
+ {
1460
+ "epoch": 1.5914684167350286,
1461
+ "eval_loss": 0.23235748708248138,
1462
+ "eval_runtime": 19.3154,
1463
+ "eval_samples_per_second": 51.772,
1464
+ "eval_steps_per_second": 0.828,
1465
+ "step": 1940
1466
+ },
1467
+ {
1468
+ "epoch": 1.6078753076292043,
1469
+ "grad_norm": 0.21950435638427734,
1470
+ "learning_rate": 2.2613862878977555e-05,
1471
+ "loss": 0.2329,
1472
+ "step": 1960
1473
+ },
1474
+ {
1475
+ "epoch": 1.6078753076292043,
1476
+ "eval_loss": 0.2320047914981842,
1477
+ "eval_runtime": 19.3131,
1478
+ "eval_samples_per_second": 51.778,
1479
+ "eval_steps_per_second": 0.828,
1480
+ "step": 1960
1481
+ },
1482
+ {
1483
+ "epoch": 1.62428219852338,
1484
+ "grad_norm": 0.21539780497550964,
1485
+ "learning_rate": 2.083203781212647e-05,
1486
+ "loss": 0.2299,
1487
+ "step": 1980
1488
+ },
1489
+ {
1490
+ "epoch": 1.62428219852338,
1491
+ "eval_loss": 0.23193301260471344,
1492
+ "eval_runtime": 19.2995,
1493
+ "eval_samples_per_second": 51.815,
1494
+ "eval_steps_per_second": 0.829,
1495
+ "step": 1980
1496
+ },
1497
+ {
1498
+ "epoch": 1.6406890894175554,
1499
+ "grad_norm": 0.2272278517484665,
1500
+ "learning_rate": 1.9115136895574404e-05,
1501
+ "loss": 0.2311,
1502
+ "step": 2000
1503
+ },
1504
+ {
1505
+ "epoch": 1.6406890894175554,
1506
+ "eval_loss": 0.23183076083660126,
1507
+ "eval_runtime": 19.3108,
1508
+ "eval_samples_per_second": 51.784,
1509
+ "eval_steps_per_second": 0.829,
1510
+ "step": 2000
1511
+ },
1512
+ {
1513
+ "epoch": 1.6570959803117309,
1514
+ "grad_norm": 0.24351078271865845,
1515
+ "learning_rate": 1.746456812733056e-05,
1516
+ "loss": 0.2215,
1517
+ "step": 2020
1518
+ },
1519
+ {
1520
+ "epoch": 1.6570959803117309,
1521
+ "eval_loss": 0.23151956498622894,
1522
+ "eval_runtime": 19.3258,
1523
+ "eval_samples_per_second": 51.744,
1524
+ "eval_steps_per_second": 0.828,
1525
+ "step": 2020
1526
+ },
1527
+ {
1528
+ "epoch": 1.6735028712059066,
1529
+ "grad_norm": 0.19060897827148438,
1530
+ "learning_rate": 1.5881685107662336e-05,
1531
+ "loss": 0.2221,
1532
+ "step": 2040
1533
+ },
1534
+ {
1535
+ "epoch": 1.6735028712059066,
1536
+ "eval_loss": 0.23138311505317688,
1537
+ "eval_runtime": 19.2846,
1538
+ "eval_samples_per_second": 51.855,
1539
+ "eval_steps_per_second": 0.83,
1540
+ "step": 2040
1541
+ },
1542
+ {
1543
+ "epoch": 1.689909762100082,
1544
+ "grad_norm": 0.23018501698970795,
1545
+ "learning_rate": 1.4367785929033394e-05,
1546
+ "loss": 0.2355,
1547
+ "step": 2060
1548
+ },
1549
+ {
1550
+ "epoch": 1.689909762100082,
1551
+ "eval_loss": 0.2313683182001114,
1552
+ "eval_runtime": 19.3234,
1553
+ "eval_samples_per_second": 51.751,
1554
+ "eval_steps_per_second": 0.828,
1555
+ "step": 2060
1556
+ },
1557
+ {
1558
+ "epoch": 1.7063166529942575,
1559
+ "grad_norm": 0.23208950459957123,
1560
+ "learning_rate": 1.2924112111562315e-05,
1561
+ "loss": 0.2242,
1562
+ "step": 2080
1563
+ },
1564
+ {
1565
+ "epoch": 1.7063166529942575,
1566
+ "eval_loss": 0.23121151328086853,
1567
+ "eval_runtime": 19.3229,
1568
+ "eval_samples_per_second": 51.752,
1569
+ "eval_steps_per_second": 0.828,
1570
+ "step": 2080
1571
+ },
1572
+ {
1573
+ "epoch": 1.7227235438884332,
1574
+ "grad_norm": 0.20929782092571259,
1575
+ "learning_rate": 1.155184758487573e-05,
1576
+ "loss": 0.2236,
1577
+ "step": 2100
1578
+ },
1579
+ {
1580
+ "epoch": 1.7227235438884332,
1581
+ "eval_loss": 0.2309342622756958,
1582
+ "eval_runtime": 19.3276,
1583
+ "eval_samples_per_second": 51.74,
1584
+ "eval_steps_per_second": 0.828,
1585
+ "step": 2100
1586
+ },
1587
+ {
1588
+ "epoch": 1.7391304347826086,
1589
+ "grad_norm": 0.2145085632801056,
1590
+ "learning_rate": 1.0252117717189713e-05,
1591
+ "loss": 0.2231,
1592
+ "step": 2120
1593
+ },
1594
+ {
1595
+ "epoch": 1.7391304347826086,
1596
+ "eval_loss": 0.2309054434299469,
1597
+ "eval_runtime": 19.3244,
1598
+ "eval_samples_per_second": 51.748,
1599
+ "eval_steps_per_second": 0.828,
1600
+ "step": 2120
1601
+ },
1602
+ {
1603
+ "epoch": 1.755537325676784,
1604
+ "grad_norm": 0.22081588208675385,
1605
+ "learning_rate": 9.025988392416717e-06,
1606
+ "loss": 0.2194,
1607
+ "step": 2140
1608
+ },
1609
+ {
1610
+ "epoch": 1.755537325676784,
1611
+ "eval_loss": 0.23062872886657715,
1612
+ "eval_runtime": 19.3231,
1613
+ "eval_samples_per_second": 51.752,
1614
+ "eval_steps_per_second": 0.828,
1615
+ "step": 2140
1616
+ },
1617
+ {
1618
+ "epoch": 1.7719442165709598,
1619
+ "grad_norm": 0.21650107204914093,
1620
+ "learning_rate": 7.874465136054332e-06,
1621
+ "loss": 0.2208,
1622
+ "step": 2160
1623
+ },
1624
+ {
1625
+ "epoch": 1.7719442165709598,
1626
+ "eval_loss": 0.2306279093027115,
1627
+ "eval_runtime": 19.3223,
1628
+ "eval_samples_per_second": 51.754,
1629
+ "eval_steps_per_second": 0.828,
1630
+ "step": 2160
1631
+ },
1632
+ {
1633
+ "epoch": 1.7883511074651355,
1634
+ "grad_norm": 0.1942278891801834,
1635
+ "learning_rate": 6.798492290572611e-06,
1636
+ "loss": 0.2241,
1637
+ "step": 2180
1638
+ },
1639
+ {
1640
+ "epoch": 1.7883511074651355,
1641
+ "eval_loss": 0.2306579053401947,
1642
+ "eval_runtime": 19.333,
1643
+ "eval_samples_per_second": 51.725,
1644
+ "eval_steps_per_second": 0.828,
1645
+ "step": 2180
1646
+ },
1647
+ {
1648
+ "epoch": 1.804757998359311,
1649
+ "grad_norm": 0.2526184916496277,
1650
+ "learning_rate": 5.798952240976951e-06,
1651
+ "loss": 0.2305,
1652
+ "step": 2200
1653
+ },
1654
+ {
1655
+ "epoch": 1.804757998359311,
1656
+ "eval_loss": 0.23054717481136322,
1657
+ "eval_runtime": 19.3308,
1658
+ "eval_samples_per_second": 51.731,
1659
+ "eval_steps_per_second": 0.828,
1660
+ "step": 2200
1661
+ },
1662
+ {
1663
+ "epoch": 1.8211648892534864,
1664
+ "grad_norm": 0.23644739389419556,
1665
+ "learning_rate": 4.8766646911805995e-06,
1666
+ "loss": 0.2182,
1667
+ "step": 2220
1668
+ },
1669
+ {
1670
+ "epoch": 1.8211648892534864,
1671
+ "eval_loss": 0.23052726686000824,
1672
+ "eval_runtime": 19.3237,
1673
+ "eval_samples_per_second": 51.75,
1674
+ "eval_steps_per_second": 0.828,
1675
+ "step": 2220
1676
+ },
1677
+ {
1678
+ "epoch": 1.837571780147662,
1679
+ "grad_norm": 0.24224869906902313,
1680
+ "learning_rate": 4.032385991781029e-06,
1681
+ "loss": 0.2281,
1682
+ "step": 2240
1683
+ },
1684
+ {
1685
+ "epoch": 1.837571780147662,
1686
+ "eval_loss": 0.23048415780067444,
1687
+ "eval_runtime": 19.3078,
1688
+ "eval_samples_per_second": 51.793,
1689
+ "eval_steps_per_second": 0.829,
1690
+ "step": 2240
1691
+ },
1692
+ {
1693
+ "epoch": 1.8539786710418376,
1694
+ "grad_norm": 0.20227192342281342,
1695
+ "learning_rate": 3.2668085197910757e-06,
1696
+ "loss": 0.2221,
1697
+ "step": 2260
1698
+ },
1699
+ {
1700
+ "epoch": 1.8539786710418376,
1701
+ "eval_loss": 0.2304566353559494,
1702
+ "eval_runtime": 19.3092,
1703
+ "eval_samples_per_second": 51.789,
1704
+ "eval_steps_per_second": 0.829,
1705
+ "step": 2260
1706
+ },
1707
+ {
1708
+ "epoch": 1.870385561936013,
1709
+ "grad_norm": 0.21336953341960907,
1710
+ "learning_rate": 2.5805601108337297e-06,
1711
+ "loss": 0.2269,
1712
+ "step": 2280
1713
+ },
1714
+ {
1715
+ "epoch": 1.870385561936013,
1716
+ "eval_loss": 0.2304052710533142,
1717
+ "eval_runtime": 19.346,
1718
+ "eval_samples_per_second": 51.69,
1719
+ "eval_steps_per_second": 0.827,
1720
+ "step": 2280
1721
+ },
1722
+ {
1723
+ "epoch": 1.8867924528301887,
1724
+ "grad_norm": 0.2497740089893341,
1725
+ "learning_rate": 1.9742035442658403e-06,
1726
+ "loss": 0.2359,
1727
+ "step": 2300
1728
+ },
1729
+ {
1730
+ "epoch": 1.8867924528301887,
1731
+ "eval_loss": 0.23032720386981964,
1732
+ "eval_runtime": 19.3272,
1733
+ "eval_samples_per_second": 51.74,
1734
+ "eval_steps_per_second": 0.828,
1735
+ "step": 2300
1736
+ },
1737
+ {
1738
+ "epoch": 1.9031993437243644,
1739
+ "grad_norm": 0.19928687810897827,
1740
+ "learning_rate": 1.448236081653509e-06,
1741
+ "loss": 0.2305,
1742
+ "step": 2320
1743
+ },
1744
+ {
1745
+ "epoch": 1.9031993437243644,
1746
+ "eval_loss": 0.23034411668777466,
1747
+ "eval_runtime": 19.3378,
1748
+ "eval_samples_per_second": 51.712,
1749
+ "eval_steps_per_second": 0.827,
1750
+ "step": 2320
1751
+ },
1752
+ {
1753
+ "epoch": 1.9196062346185396,
1754
+ "grad_norm": 0.19296622276306152,
1755
+ "learning_rate": 1.0030890589772468e-06,
1756
+ "loss": 0.2239,
1757
+ "step": 2340
1758
+ },
1759
+ {
1760
+ "epoch": 1.9196062346185396,
1761
+ "eval_loss": 0.23032715916633606,
1762
+ "eval_runtime": 19.3474,
1763
+ "eval_samples_per_second": 51.687,
1764
+ "eval_steps_per_second": 0.827,
1765
+ "step": 2340
1766
+ },
1767
+ {
1768
+ "epoch": 1.9360131255127153,
1769
+ "grad_norm": 0.22955958545207977,
1770
+ "learning_rate": 6.391275329014734e-07,
1771
+ "loss": 0.2283,
1772
+ "step": 2360
1773
+ },
1774
+ {
1775
+ "epoch": 1.9360131255127153,
1776
+ "eval_loss": 0.23028050363063812,
1777
+ "eval_runtime": 19.3344,
1778
+ "eval_samples_per_second": 51.721,
1779
+ "eval_steps_per_second": 0.828,
1780
+ "step": 2360
1781
+ },
1782
+ {
1783
+ "epoch": 1.952420016406891,
1784
+ "grad_norm": 0.21419735252857208,
1785
+ "learning_rate": 3.566499813984603e-07,
1786
+ "loss": 0.2382,
1787
+ "step": 2380
1788
+ },
1789
+ {
1790
+ "epoch": 1.952420016406891,
1791
+ "eval_loss": 0.23030652105808258,
1792
+ "eval_runtime": 19.3163,
1793
+ "eval_samples_per_second": 51.77,
1794
+ "eval_steps_per_second": 0.828,
1795
+ "step": 2380
1796
+ },
1797
+ {
1798
+ "epoch": 1.9688269073010665,
1799
+ "grad_norm": 0.21221666038036346,
1800
+ "learning_rate": 1.5588805897215342e-07,
1801
+ "loss": 0.2326,
1802
+ "step": 2400
1803
+ },
1804
+ {
1805
+ "epoch": 1.9688269073010665,
1806
+ "eval_loss": 0.2303100824356079,
1807
+ "eval_runtime": 19.3186,
1808
+ "eval_samples_per_second": 51.764,
1809
+ "eval_steps_per_second": 0.828,
1810
+ "step": 2400
1811
+ },
1812
+ {
1813
+ "epoch": 1.985233798195242,
1814
+ "grad_norm": 0.185977041721344,
1815
+ "learning_rate": 3.700640668272692e-08,
1816
+ "loss": 0.2186,
1817
+ "step": 2420
1818
+ },
1819
+ {
1820
+ "epoch": 1.985233798195242,
1821
+ "eval_loss": 0.23031620681285858,
1822
+ "eval_runtime": 19.3172,
1823
+ "eval_samples_per_second": 51.767,
1824
+ "eval_steps_per_second": 0.828,
1825
+ "step": 2420
1826
+ },
1827
+ {
1828
+ "epoch": 2.0,
1829
+ "step": 2438,
1830
+ "total_flos": 1.58523627405312e+18,
1831
+ "train_loss": 0.2688869493138702,
1832
+ "train_runtime": 4683.6469,
1833
+ "train_samples_per_second": 16.654,
1834
+ "train_steps_per_second": 0.521
1835
+ }
1836
+ ],
1837
+ "logging_steps": 20,
1838
+ "max_steps": 2438,
1839
+ "num_input_tokens_seen": 0,
1840
+ "num_train_epochs": 2,
1841
+ "save_steps": 500,
1842
+ "stateful_callbacks": {
1843
+ "TrainerControl": {
1844
+ "args": {
1845
+ "should_epoch_stop": false,
1846
+ "should_evaluate": false,
1847
+ "should_log": false,
1848
+ "should_save": true,
1849
+ "should_training_stop": true
1850
+ },
1851
+ "attributes": {}
1852
+ }
1853
+ },
1854
+ "total_flos": 1.58523627405312e+18,
1855
+ "train_batch_size": 32,
1856
+ "trial_name": null,
1857
+ "trial_params": null
1858
+ }