Hugging Face's logo

nvan15
/
OminiControlRotation

Safetensors
Model card Files
xet
Community
nvan15 commited on
Commit
70ff38e
·
verified ·
1 Parent(s): a89739c

Batch upload part 11

Browse files
Files changed (45) hide show
  1. nl_tasks/run_all/run_exps5/ft/adapter_config.json +18 -0
  2. nl_tasks/run_all/run_exps5/ft/special_tokens_map.json +30 -0
  3. nl_tasks/run_all/run_exps5/ft/tokenizer.json +0 -0
  4. nl_tasks/run_all/run_exps5/ft/tokenizer_config.json +51 -0
  5. nl_tasks/run_all/run_exps5/ft/training_args.bin +3 -0
  6. nl_tasks/run_all/run_exps5/ft2/adapter_config.json +18 -0
  7. nl_tasks/run_all/run_exps5/ft2/adapter_model.bin +3 -0
  8. nl_tasks/run_all/run_exps6/ft/adapter_config.json +18 -0
  9. nl_tasks/run_all/run_exps6/ft/special_tokens_map.json +30 -0
  10. nl_tasks/run_all/run_exps6/ft/tokenizer.json +0 -0
  11. nl_tasks/run_all/run_exps6/ft/tokenizer_config.json +51 -0
  12. nl_tasks/run_all/run_exps6/ft/training_args.bin +3 -0
  13. nl_tasks/run_all/run_exps6/ft2/adapter_config.json +18 -0
  14. nl_tasks/run_all/run_exps6/ft2/adapter_model.bin +3 -0
  15. nl_tasks/run_all/run_exps6/trainer_state.json +1858 -0
  16. nl_tasks/run_all/run_exps7/ft/adapter_config.json +18 -0
  17. nl_tasks/run_all/run_exps7/ft/added_tokens.json +3 -0
  18. nl_tasks/run_all/run_exps7/ft/special_tokens_map.json +30 -0
  19. nl_tasks/run_all/run_exps7/ft/tokenizer.json +0 -0
  20. nl_tasks/run_all/run_exps7/ft/tokenizer.model +3 -0
  21. nl_tasks/run_all/run_exps7/ft/tokenizer_config.json +51 -0
  22. nl_tasks/run_all/run_exps7/ft/training_args.bin +3 -0
  23. nl_tasks/run_all/run_exps7/ft2/adapter_config.json +18 -0
  24. nl_tasks/run_all/run_exps7/ft2/adapter_model.bin +3 -0
  25. nl_tasks/run_all/run_exps7/trainer_state.json +1274 -0
  26. nl_tasks/run_all/run_exps8/ft/adapter_config.json +18 -0
  27. nl_tasks/run_all/run_exps8/ft/added_tokens.json +3 -0
  28. nl_tasks/run_all/run_exps8/ft/special_tokens_map.json +30 -0
  29. nl_tasks/run_all/run_exps8/ft/tokenizer.json +0 -0
  30. nl_tasks/run_all/run_exps8/ft/tokenizer.model +3 -0
  31. nl_tasks/run_all/run_exps8/ft/tokenizer_config.json +51 -0
  32. nl_tasks/run_all/run_exps8/ft/training_args.bin +3 -0
  33. nl_tasks/run_all/run_exps8/ft2/adapter_config.json +18 -0
  34. nl_tasks/run_all/run_exps8/ft2/adapter_model.bin +3 -0
  35. nl_tasks/run_all/run_exps8/trainer_state.json +1106 -0
  36. nl_tasks/run_all/run_exps9/ft/adapter_config.json +18 -0
  37. nl_tasks/run_all/run_exps9/ft/added_tokens.json +3 -0
  38. nl_tasks/run_all/run_exps9/ft/special_tokens_map.json +30 -0
  39. nl_tasks/run_all/run_exps9/ft/tokenizer.json +0 -0
  40. nl_tasks/run_all/run_exps9/ft/tokenizer.model +3 -0
  41. nl_tasks/run_all/run_exps9/ft/tokenizer_config.json +51 -0
  42. nl_tasks/run_all/run_exps9/ft/training_args.bin +3 -0
  43. nl_tasks/run_all/run_exps9/ft2/adapter_config.json +18 -0
  44. nl_tasks/run_all/run_exps9/ft2/adapter_model.bin +3 -0
  45. nl_tasks/run_all/run_exps9/trainer_state.json +1106 -0
nl_tasks/run_all/run_exps5/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_exps5/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_exps5/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exps5/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_exps5/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:942156a3535529a09f23ee506630236c020403d99c9c7d866183f3d3ab09a6ae
3
+ size 6481
nl_tasks/run_all/run_exps5/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_exps5/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e97074df351e27519f063ae5998a754487e84a132e52aadd307ede2c066c06d3
3
+ size 33602659
nl_tasks/run_all/run_exps6/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_exps6/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_exps6/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exps6/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_exps6/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c849a9ab0a6eb14bba18c87f1ddd1ab593cf4c1265e608220d9cda0874bfa12
3
+ size 6481
nl_tasks/run_all/run_exps6/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_exps6/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4fecac2e30e1b68dd7efde0b8dad375f595bda6a0e9b1fa002b0b396a08e3853
3
+ size 33602659
nl_tasks/run_all/run_exps6/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": 0.21249747276306152,
15
+ "learning_rate": 3.89344262295082e-05,
16
+ "loss": 0.7347,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.016406890894175553,
21
+ "eval_loss": 0.679619312286377,
22
+ "eval_runtime": 19.797,
23
+ "eval_samples_per_second": 50.513,
24
+ "eval_steps_per_second": 0.808,
25
+ "step": 20
26
+ },
27
+ {
28
+ "epoch": 0.03281378178835111,
29
+ "grad_norm": 0.20149445533752441,
30
+ "learning_rate": 7.991803278688526e-05,
31
+ "loss": 0.5489,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.03281378178835111,
36
+ "eval_loss": 0.4365108609199524,
37
+ "eval_runtime": 19.2731,
38
+ "eval_samples_per_second": 51.886,
39
+ "eval_steps_per_second": 0.83,
40
+ "step": 40
41
+ },
42
+ {
43
+ "epoch": 0.04922067268252666,
44
+ "grad_norm": 0.18983662128448486,
45
+ "learning_rate": 0.00012090163934426229,
46
+ "loss": 0.4113,
47
+ "step": 60
48
+ },
49
+ {
50
+ "epoch": 0.04922067268252666,
51
+ "eval_loss": 0.37941107153892517,
52
+ "eval_runtime": 19.2828,
53
+ "eval_samples_per_second": 51.86,
54
+ "eval_steps_per_second": 0.83,
55
+ "step": 60
56
+ },
57
+ {
58
+ "epoch": 0.06562756357670221,
59
+ "grad_norm": 0.18760351836681366,
60
+ "learning_rate": 0.00016188524590163937,
61
+ "loss": 0.3816,
62
+ "step": 80
63
+ },
64
+ {
65
+ "epoch": 0.06562756357670221,
66
+ "eval_loss": 0.3530106842517853,
67
+ "eval_runtime": 19.2828,
68
+ "eval_samples_per_second": 51.86,
69
+ "eval_steps_per_second": 0.83,
70
+ "step": 80
71
+ },
72
+ {
73
+ "epoch": 0.08203445447087777,
74
+ "grad_norm": 0.20841318368911743,
75
+ "learning_rate": 0.0002028688524590164,
76
+ "loss": 0.361,
77
+ "step": 100
78
+ },
79
+ {
80
+ "epoch": 0.08203445447087777,
81
+ "eval_loss": 0.3407621681690216,
82
+ "eval_runtime": 19.2819,
83
+ "eval_samples_per_second": 51.862,
84
+ "eval_steps_per_second": 0.83,
85
+ "step": 100
86
+ },
87
+ {
88
+ "epoch": 0.09844134536505332,
89
+ "grad_norm": 0.1878339797258377,
90
+ "learning_rate": 0.00024385245901639344,
91
+ "loss": 0.3413,
92
+ "step": 120
93
+ },
94
+ {
95
+ "epoch": 0.09844134536505332,
96
+ "eval_loss": 0.3304310441017151,
97
+ "eval_runtime": 19.3445,
98
+ "eval_samples_per_second": 51.694,
99
+ "eval_steps_per_second": 0.827,
100
+ "step": 120
101
+ },
102
+ {
103
+ "epoch": 0.11484823625922888,
104
+ "grad_norm": 0.19155746698379517,
105
+ "learning_rate": 0.0002848360655737705,
106
+ "loss": 0.3227,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.11484823625922888,
111
+ "eval_loss": 0.3268902003765106,
112
+ "eval_runtime": 19.3134,
113
+ "eval_samples_per_second": 51.778,
114
+ "eval_steps_per_second": 0.828,
115
+ "step": 140
116
+ },
117
+ {
118
+ "epoch": 0.13125512715340443,
119
+ "grad_norm": 0.19510044157505035,
120
+ "learning_rate": 0.00032581967213114754,
121
+ "loss": 0.3219,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 0.13125512715340443,
126
+ "eval_loss": 0.31588122248649597,
127
+ "eval_runtime": 19.3227,
128
+ "eval_samples_per_second": 51.753,
129
+ "eval_steps_per_second": 0.828,
130
+ "step": 160
131
+ },
132
+ {
133
+ "epoch": 0.14766201804757997,
134
+ "grad_norm": 0.1838303506374359,
135
+ "learning_rate": 0.0003668032786885246,
136
+ "loss": 0.3216,
137
+ "step": 180
138
+ },
139
+ {
140
+ "epoch": 0.14766201804757997,
141
+ "eval_loss": 0.31225940585136414,
142
+ "eval_runtime": 19.2925,
143
+ "eval_samples_per_second": 51.833,
144
+ "eval_steps_per_second": 0.829,
145
+ "step": 180
146
+ },
147
+ {
148
+ "epoch": 0.16406890894175555,
149
+ "grad_norm": 0.15653817355632782,
150
+ "learning_rate": 0.00040778688524590164,
151
+ "loss": 0.3185,
152
+ "step": 200
153
+ },
154
+ {
155
+ "epoch": 0.16406890894175555,
156
+ "eval_loss": 0.30911117792129517,
157
+ "eval_runtime": 19.3004,
158
+ "eval_samples_per_second": 51.812,
159
+ "eval_steps_per_second": 0.829,
160
+ "step": 200
161
+ },
162
+ {
163
+ "epoch": 0.1804757998359311,
164
+ "grad_norm": 0.19769713282585144,
165
+ "learning_rate": 0.00044877049180327874,
166
+ "loss": 0.3252,
167
+ "step": 220
168
+ },
169
+ {
170
+ "epoch": 0.1804757998359311,
171
+ "eval_loss": 0.30457785725593567,
172
+ "eval_runtime": 19.2851,
173
+ "eval_samples_per_second": 51.853,
174
+ "eval_steps_per_second": 0.83,
175
+ "step": 220
176
+ },
177
+ {
178
+ "epoch": 0.19688269073010664,
179
+ "grad_norm": 0.17321603000164032,
180
+ "learning_rate": 0.0004897540983606558,
181
+ "loss": 0.3146,
182
+ "step": 240
183
+ },
184
+ {
185
+ "epoch": 0.19688269073010664,
186
+ "eval_loss": 0.30066341161727905,
187
+ "eval_runtime": 19.2712,
188
+ "eval_samples_per_second": 51.891,
189
+ "eval_steps_per_second": 0.83,
190
+ "step": 240
191
+ },
192
+ {
193
+ "epoch": 0.2132895816242822,
194
+ "grad_norm": 0.17701204121112823,
195
+ "learning_rate": 0.000499942336322776,
196
+ "loss": 0.3095,
197
+ "step": 260
198
+ },
199
+ {
200
+ "epoch": 0.2132895816242822,
201
+ "eval_loss": 0.29611480236053467,
202
+ "eval_runtime": 19.2684,
203
+ "eval_samples_per_second": 51.898,
204
+ "eval_steps_per_second": 0.83,
205
+ "step": 260
206
+ },
207
+ {
208
+ "epoch": 0.22969647251845776,
209
+ "grad_norm": 0.17420244216918945,
210
+ "learning_rate": 0.0004996861069511679,
211
+ "loss": 0.3121,
212
+ "step": 280
213
+ },
214
+ {
215
+ "epoch": 0.22969647251845776,
216
+ "eval_loss": 0.2946722209453583,
217
+ "eval_runtime": 19.2626,
218
+ "eval_samples_per_second": 51.914,
219
+ "eval_steps_per_second": 0.831,
220
+ "step": 280
221
+ },
222
+ {
223
+ "epoch": 0.2461033634126333,
224
+ "grad_norm": 0.46635693311691284,
225
+ "learning_rate": 0.0004992251147198465,
226
+ "loss": 0.299,
227
+ "step": 300
228
+ },
229
+ {
230
+ "epoch": 0.2461033634126333,
231
+ "eval_loss": 0.2974222004413605,
232
+ "eval_runtime": 19.2718,
233
+ "eval_samples_per_second": 51.889,
234
+ "eval_steps_per_second": 0.83,
235
+ "step": 300
236
+ },
237
+ {
238
+ "epoch": 0.26251025430680885,
239
+ "grad_norm": 0.17965629696846008,
240
+ "learning_rate": 0.0004985597376798325,
241
+ "loss": 0.316,
242
+ "step": 320
243
+ },
244
+ {
245
+ "epoch": 0.26251025430680885,
246
+ "eval_loss": 0.28867945075035095,
247
+ "eval_runtime": 19.2681,
248
+ "eval_samples_per_second": 51.899,
249
+ "eval_steps_per_second": 0.83,
250
+ "step": 320
251
+ },
252
+ {
253
+ "epoch": 0.27891714520098443,
254
+ "grad_norm": 0.18957704305648804,
255
+ "learning_rate": 0.0004976905214942672,
256
+ "loss": 0.3054,
257
+ "step": 340
258
+ },
259
+ {
260
+ "epoch": 0.27891714520098443,
261
+ "eval_loss": 0.2872786819934845,
262
+ "eval_runtime": 19.3097,
263
+ "eval_samples_per_second": 51.788,
264
+ "eval_steps_per_second": 0.829,
265
+ "step": 340
266
+ },
267
+ {
268
+ "epoch": 0.29532403609515995,
269
+ "grad_norm": 0.21701280772686005,
270
+ "learning_rate": 0.0004966181789909249,
271
+ "loss": 0.2846,
272
+ "step": 360
273
+ },
274
+ {
275
+ "epoch": 0.29532403609515995,
276
+ "eval_loss": 0.28569233417510986,
277
+ "eval_runtime": 19.3224,
278
+ "eval_samples_per_second": 51.753,
279
+ "eval_steps_per_second": 0.828,
280
+ "step": 360
281
+ },
282
+ {
283
+ "epoch": 0.3117309269893355,
284
+ "grad_norm": 0.2084118276834488,
285
+ "learning_rate": 0.000495343589577636,
286
+ "loss": 0.2918,
287
+ "step": 380
288
+ },
289
+ {
290
+ "epoch": 0.3117309269893355,
291
+ "eval_loss": 0.281757652759552,
292
+ "eval_runtime": 19.3105,
293
+ "eval_samples_per_second": 51.785,
294
+ "eval_steps_per_second": 0.829,
295
+ "step": 380
296
+ },
297
+ {
298
+ "epoch": 0.3281378178835111,
299
+ "grad_norm": 0.21250280737876892,
300
+ "learning_rate": 0.0004938677985211012,
301
+ "loss": 0.2996,
302
+ "step": 400
303
+ },
304
+ {
305
+ "epoch": 0.3281378178835111,
306
+ "eval_loss": 0.27930182218551636,
307
+ "eval_runtime": 19.3349,
308
+ "eval_samples_per_second": 51.72,
309
+ "eval_steps_per_second": 0.828,
310
+ "step": 400
311
+ },
312
+ {
313
+ "epoch": 0.3445447087776866,
314
+ "grad_norm": 0.21832171082496643,
315
+ "learning_rate": 0.0004921920160896879,
316
+ "loss": 0.2921,
317
+ "step": 420
318
+ },
319
+ {
320
+ "epoch": 0.3445447087776866,
321
+ "eval_loss": 0.2787286043167114,
322
+ "eval_runtime": 19.3144,
323
+ "eval_samples_per_second": 51.775,
324
+ "eval_steps_per_second": 0.828,
325
+ "step": 420
326
+ },
327
+ {
328
+ "epoch": 0.3609515996718622,
329
+ "grad_norm": 0.20000235736370087,
330
+ "learning_rate": 0.0004903176165609127,
331
+ "loss": 0.2934,
332
+ "step": 440
333
+ },
334
+ {
335
+ "epoch": 0.3609515996718622,
336
+ "eval_loss": 0.27436938881874084,
337
+ "eval_runtime": 19.3348,
338
+ "eval_samples_per_second": 51.72,
339
+ "eval_steps_per_second": 0.828,
340
+ "step": 440
341
+ },
342
+ {
343
+ "epoch": 0.37735849056603776,
344
+ "grad_norm": 0.17309211194515228,
345
+ "learning_rate": 0.0004882461370944212,
346
+ "loss": 0.2942,
347
+ "step": 460
348
+ },
349
+ {
350
+ "epoch": 0.37735849056603776,
351
+ "eval_loss": 0.27319249510765076,
352
+ "eval_runtime": 19.322,
353
+ "eval_samples_per_second": 51.754,
354
+ "eval_steps_per_second": 0.828,
355
+ "step": 460
356
+ },
357
+ {
358
+ "epoch": 0.3937653814602133,
359
+ "grad_norm": 0.18647795915603638,
360
+ "learning_rate": 0.00048597927647139373,
361
+ "loss": 0.2807,
362
+ "step": 480
363
+ },
364
+ {
365
+ "epoch": 0.3937653814602133,
366
+ "eval_loss": 0.27322131395339966,
367
+ "eval_runtime": 19.2842,
368
+ "eval_samples_per_second": 51.856,
369
+ "eval_steps_per_second": 0.83,
370
+ "step": 480
371
+ },
372
+ {
373
+ "epoch": 0.41017227235438886,
374
+ "grad_norm": 0.18252436816692352,
375
+ "learning_rate": 0.00048351889370140585,
376
+ "loss": 0.2769,
377
+ "step": 500
378
+ },
379
+ {
380
+ "epoch": 0.41017227235438886,
381
+ "eval_loss": 0.2703263461589813,
382
+ "eval_runtime": 19.3007,
383
+ "eval_samples_per_second": 51.812,
384
+ "eval_steps_per_second": 0.829,
385
+ "step": 500
386
+ },
387
+ {
388
+ "epoch": 0.4265791632485644,
389
+ "grad_norm": 0.2371368557214737,
390
+ "learning_rate": 0.0004808670064978903,
391
+ "loss": 0.2848,
392
+ "step": 520
393
+ },
394
+ {
395
+ "epoch": 0.4265791632485644,
396
+ "eval_loss": 0.26800182461738586,
397
+ "eval_runtime": 19.2719,
398
+ "eval_samples_per_second": 51.889,
399
+ "eval_steps_per_second": 0.83,
400
+ "step": 520
401
+ },
402
+ {
403
+ "epoch": 0.44298605414273995,
404
+ "grad_norm": 0.22401951253414154,
405
+ "learning_rate": 0.0004780257896234483,
406
+ "loss": 0.2786,
407
+ "step": 540
408
+ },
409
+ {
410
+ "epoch": 0.44298605414273995,
411
+ "eval_loss": 0.2664169669151306,
412
+ "eval_runtime": 19.2853,
413
+ "eval_samples_per_second": 51.853,
414
+ "eval_steps_per_second": 0.83,
415
+ "step": 540
416
+ },
417
+ {
418
+ "epoch": 0.4593929450369155,
419
+ "grad_norm": 0.19664160907268524,
420
+ "learning_rate": 0.0004749975731063677,
421
+ "loss": 0.2744,
422
+ "step": 560
423
+ },
424
+ {
425
+ "epoch": 0.4593929450369155,
426
+ "eval_loss": 0.26462265849113464,
427
+ "eval_runtime": 19.2867,
428
+ "eval_samples_per_second": 51.849,
429
+ "eval_steps_per_second": 0.83,
430
+ "step": 560
431
+ },
432
+ {
433
+ "epoch": 0.47579983593109104,
434
+ "grad_norm": 0.23553213477134705,
435
+ "learning_rate": 0.0004717848403298107,
436
+ "loss": 0.2697,
437
+ "step": 580
438
+ },
439
+ {
440
+ "epoch": 0.47579983593109104,
441
+ "eval_loss": 0.2656167149543762,
442
+ "eval_runtime": 19.312,
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": 0.2151021808385849,
450
+ "learning_rate": 0.00046839022599523875,
451
+ "loss": 0.2732,
452
+ "step": 600
453
+ },
454
+ {
455
+ "epoch": 0.4922067268252666,
456
+ "eval_loss": 0.2635062634944916,
457
+ "eval_runtime": 19.2866,
458
+ "eval_samples_per_second": 51.85,
459
+ "eval_steps_per_second": 0.83,
460
+ "step": 600
461
+ },
462
+ {
463
+ "epoch": 0.5086136177194421,
464
+ "grad_norm": 0.19232328236103058,
465
+ "learning_rate": 0.00046481651396174274,
466
+ "loss": 0.2635,
467
+ "step": 620
468
+ },
469
+ {
470
+ "epoch": 0.5086136177194421,
471
+ "eval_loss": 0.26092493534088135,
472
+ "eval_runtime": 19.2867,
473
+ "eval_samples_per_second": 51.849,
474
+ "eval_steps_per_second": 0.83,
475
+ "step": 620
476
+ },
477
+ {
478
+ "epoch": 0.5250205086136177,
479
+ "grad_norm": 0.1972268670797348,
480
+ "learning_rate": 0.0004610666349630537,
481
+ "loss": 0.2638,
482
+ "step": 640
483
+ },
484
+ {
485
+ "epoch": 0.5250205086136177,
486
+ "eval_loss": 0.26019811630249023,
487
+ "eval_runtime": 19.2822,
488
+ "eval_samples_per_second": 51.861,
489
+ "eval_steps_per_second": 0.83,
490
+ "step": 640
491
+ },
492
+ {
493
+ "epoch": 0.5414273995077933,
494
+ "grad_norm": 0.24094261229038239,
495
+ "learning_rate": 0.0004571436642041027,
496
+ "loss": 0.2631,
497
+ "step": 660
498
+ },
499
+ {
500
+ "epoch": 0.5414273995077933,
501
+ "eval_loss": 0.2567290663719177,
502
+ "eval_runtime": 19.2725,
503
+ "eval_samples_per_second": 51.887,
504
+ "eval_steps_per_second": 0.83,
505
+ "step": 660
506
+ },
507
+ {
508
+ "epoch": 0.5578342904019689,
509
+ "grad_norm": 0.1992902010679245,
510
+ "learning_rate": 0.0004530508188391039,
511
+ "loss": 0.2631,
512
+ "step": 680
513
+ },
514
+ {
515
+ "epoch": 0.5578342904019689,
516
+ "eval_loss": 0.2599620223045349,
517
+ "eval_runtime": 19.2735,
518
+ "eval_samples_per_second": 51.885,
519
+ "eval_steps_per_second": 0.83,
520
+ "step": 680
521
+ },
522
+ {
523
+ "epoch": 0.5742411812961444,
524
+ "grad_norm": 0.16953237354755402,
525
+ "learning_rate": 0.00044879145533322694,
526
+ "loss": 0.2576,
527
+ "step": 700
528
+ },
529
+ {
530
+ "epoch": 0.5742411812961444,
531
+ "eval_loss": 0.2572336494922638,
532
+ "eval_runtime": 19.2772,
533
+ "eval_samples_per_second": 51.875,
534
+ "eval_steps_per_second": 0.83,
535
+ "step": 700
536
+ },
537
+ {
538
+ "epoch": 0.5906480721903199,
539
+ "grad_norm": 0.19884386658668518,
540
+ "learning_rate": 0.000444369066710023,
541
+ "loss": 0.2528,
542
+ "step": 720
543
+ },
544
+ {
545
+ "epoch": 0.5906480721903199,
546
+ "eval_loss": 0.2562015950679779,
547
+ "eval_runtime": 19.284,
548
+ "eval_samples_per_second": 51.856,
549
+ "eval_steps_per_second": 0.83,
550
+ "step": 720
551
+ },
552
+ {
553
+ "epoch": 0.6070549630844955,
554
+ "grad_norm": 0.1670694798231125,
555
+ "learning_rate": 0.0004397872796868627,
556
+ "loss": 0.2606,
557
+ "step": 740
558
+ },
559
+ {
560
+ "epoch": 0.6070549630844955,
561
+ "eval_loss": 0.25398388504981995,
562
+ "eval_runtime": 19.331,
563
+ "eval_samples_per_second": 51.73,
564
+ "eval_steps_per_second": 0.828,
565
+ "step": 740
566
+ },
567
+ {
568
+ "epoch": 0.623461853978671,
569
+ "grad_norm": 0.20322060585021973,
570
+ "learning_rate": 0.0004350498517007332,
571
+ "loss": 0.267,
572
+ "step": 760
573
+ },
574
+ {
575
+ "epoch": 0.623461853978671,
576
+ "eval_loss": 0.25414517521858215,
577
+ "eval_runtime": 19.2967,
578
+ "eval_samples_per_second": 51.822,
579
+ "eval_steps_per_second": 0.829,
580
+ "step": 760
581
+ },
582
+ {
583
+ "epoch": 0.6398687448728466,
584
+ "grad_norm": 0.2102508246898651,
585
+ "learning_rate": 0.0004301606678268354,
586
+ "loss": 0.2617,
587
+ "step": 780
588
+ },
589
+ {
590
+ "epoch": 0.6398687448728466,
591
+ "eval_loss": 0.25184592604637146,
592
+ "eval_runtime": 19.3147,
593
+ "eval_samples_per_second": 51.774,
594
+ "eval_steps_per_second": 0.828,
595
+ "step": 780
596
+ },
597
+ {
598
+ "epoch": 0.6562756357670222,
599
+ "grad_norm": 0.19933830201625824,
600
+ "learning_rate": 0.0004251237375925071,
601
+ "loss": 0.2608,
602
+ "step": 800
603
+ },
604
+ {
605
+ "epoch": 0.6562756357670222,
606
+ "eval_loss": 0.25019243359565735,
607
+ "eval_runtime": 19.2698,
608
+ "eval_samples_per_second": 51.895,
609
+ "eval_steps_per_second": 0.83,
610
+ "step": 800
611
+ },
612
+ {
613
+ "epoch": 0.6726825266611977,
614
+ "grad_norm": 0.19473214447498322,
615
+ "learning_rate": 0.00041994319168908533,
616
+ "loss": 0.2494,
617
+ "step": 820
618
+ },
619
+ {
620
+ "epoch": 0.6726825266611977,
621
+ "eval_loss": 0.24847277998924255,
622
+ "eval_runtime": 19.2747,
623
+ "eval_samples_per_second": 51.882,
624
+ "eval_steps_per_second": 0.83,
625
+ "step": 820
626
+ },
627
+ {
628
+ "epoch": 0.6890894175553732,
629
+ "grad_norm": 0.21291324496269226,
630
+ "learning_rate": 0.0004146232785844044,
631
+ "loss": 0.262,
632
+ "step": 840
633
+ },
634
+ {
635
+ "epoch": 0.6890894175553732,
636
+ "eval_loss": 0.2481946051120758,
637
+ "eval_runtime": 19.2813,
638
+ "eval_samples_per_second": 51.864,
639
+ "eval_steps_per_second": 0.83,
640
+ "step": 840
641
+ },
642
+ {
643
+ "epoch": 0.7054963084495488,
644
+ "grad_norm": 0.19832803308963776,
645
+ "learning_rate": 0.0004091683610387078,
646
+ "loss": 0.2665,
647
+ "step": 860
648
+ },
649
+ {
650
+ "epoch": 0.7054963084495488,
651
+ "eval_loss": 0.2467126101255417,
652
+ "eval_runtime": 19.2689,
653
+ "eval_samples_per_second": 51.897,
654
+ "eval_steps_per_second": 0.83,
655
+ "step": 860
656
+ },
657
+ {
658
+ "epoch": 0.7219031993437244,
659
+ "grad_norm": 0.1828453689813614,
660
+ "learning_rate": 0.0004035829125268307,
661
+ "loss": 0.2588,
662
+ "step": 880
663
+ },
664
+ {
665
+ "epoch": 0.7219031993437244,
666
+ "eval_loss": 0.24598461389541626,
667
+ "eval_runtime": 19.2772,
668
+ "eval_samples_per_second": 51.875,
669
+ "eval_steps_per_second": 0.83,
670
+ "step": 880
671
+ },
672
+ {
673
+ "epoch": 0.7383100902379,
674
+ "grad_norm": 0.18557998538017273,
675
+ "learning_rate": 0.00039787151356958806,
676
+ "loss": 0.2498,
677
+ "step": 900
678
+ },
679
+ {
680
+ "epoch": 0.7383100902379,
681
+ "eval_loss": 0.24567732214927673,
682
+ "eval_runtime": 19.2669,
683
+ "eval_samples_per_second": 51.902,
684
+ "eval_steps_per_second": 0.83,
685
+ "step": 900
686
+ },
687
+ {
688
+ "epoch": 0.7547169811320755,
689
+ "grad_norm": 0.18263976275920868,
690
+ "learning_rate": 0.0003920388479773756,
691
+ "loss": 0.2551,
692
+ "step": 920
693
+ },
694
+ {
695
+ "epoch": 0.7547169811320755,
696
+ "eval_loss": 0.24500185251235962,
697
+ "eval_runtime": 19.2808,
698
+ "eval_samples_per_second": 51.865,
699
+ "eval_steps_per_second": 0.83,
700
+ "step": 920
701
+ },
702
+ {
703
+ "epoch": 0.771123872026251,
704
+ "grad_norm": 0.16948311030864716,
705
+ "learning_rate": 0.000386089699009066,
706
+ "loss": 0.2381,
707
+ "step": 940
708
+ },
709
+ {
710
+ "epoch": 0.771123872026251,
711
+ "eval_loss": 0.24239838123321533,
712
+ "eval_runtime": 19.2821,
713
+ "eval_samples_per_second": 51.862,
714
+ "eval_steps_per_second": 0.83,
715
+ "step": 940
716
+ },
717
+ {
718
+ "epoch": 0.7875307629204266,
719
+ "grad_norm": 0.203163743019104,
720
+ "learning_rate": 0.00038002894544934807,
721
+ "loss": 0.2471,
722
+ "step": 960
723
+ },
724
+ {
725
+ "epoch": 0.7875307629204266,
726
+ "eval_loss": 0.24170944094657898,
727
+ "eval_runtime": 19.3092,
728
+ "eval_samples_per_second": 51.789,
729
+ "eval_steps_per_second": 0.829,
730
+ "step": 960
731
+ },
732
+ {
733
+ "epoch": 0.8039376538146021,
734
+ "grad_norm": 0.18333876132965088,
735
+ "learning_rate": 0.00037386155760772786,
736
+ "loss": 0.2529,
737
+ "step": 980
738
+ },
739
+ {
740
+ "epoch": 0.8039376538146021,
741
+ "eval_loss": 0.24172595143318176,
742
+ "eval_runtime": 19.304,
743
+ "eval_samples_per_second": 51.803,
744
+ "eval_steps_per_second": 0.829,
745
+ "step": 980
746
+ },
747
+ {
748
+ "epoch": 0.8203445447087777,
749
+ "grad_norm": 0.19429703056812286,
750
+ "learning_rate": 0.0003675925932424715,
751
+ "loss": 0.2528,
752
+ "step": 1000
753
+ },
754
+ {
755
+ "epoch": 0.8203445447087777,
756
+ "eval_loss": 0.24023687839508057,
757
+ "eval_runtime": 19.2585,
758
+ "eval_samples_per_second": 51.925,
759
+ "eval_steps_per_second": 0.831,
760
+ "step": 1000
761
+ },
762
+ {
763
+ "epoch": 0.8367514356029533,
764
+ "grad_norm": 0.20327186584472656,
765
+ "learning_rate": 0.00036122719341283364,
766
+ "loss": 0.2428,
767
+ "step": 1020
768
+ },
769
+ {
770
+ "epoch": 0.8367514356029533,
771
+ "eval_loss": 0.23908580839633942,
772
+ "eval_runtime": 19.2657,
773
+ "eval_samples_per_second": 51.906,
774
+ "eval_steps_per_second": 0.83,
775
+ "step": 1020
776
+ },
777
+ {
778
+ "epoch": 0.8531583264971287,
779
+ "grad_norm": 0.1825667917728424,
780
+ "learning_rate": 0.00035477057826297165,
781
+ "loss": 0.2434,
782
+ "step": 1040
783
+ },
784
+ {
785
+ "epoch": 0.8531583264971287,
786
+ "eval_loss": 0.23917630314826965,
787
+ "eval_runtime": 19.3315,
788
+ "eval_samples_per_second": 51.729,
789
+ "eval_steps_per_second": 0.828,
790
+ "step": 1040
791
+ },
792
+ {
793
+ "epoch": 0.8695652173913043,
794
+ "grad_norm": 0.19843976199626923,
795
+ "learning_rate": 0.00034822804274100357,
796
+ "loss": 0.2442,
797
+ "step": 1060
798
+ },
799
+ {
800
+ "epoch": 0.8695652173913043,
801
+ "eval_loss": 0.2374887764453888,
802
+ "eval_runtime": 19.2699,
803
+ "eval_samples_per_second": 51.894,
804
+ "eval_steps_per_second": 0.83,
805
+ "step": 1060
806
+ },
807
+ {
808
+ "epoch": 0.8859721082854799,
809
+ "grad_norm": 0.15917833149433136,
810
+ "learning_rate": 0.00034160495225672184,
811
+ "loss": 0.254,
812
+ "step": 1080
813
+ },
814
+ {
815
+ "epoch": 0.8859721082854799,
816
+ "eval_loss": 0.23768678307533264,
817
+ "eval_runtime": 19.2633,
818
+ "eval_samples_per_second": 51.912,
819
+ "eval_steps_per_second": 0.831,
820
+ "step": 1080
821
+ },
822
+ {
823
+ "epoch": 0.9023789991796555,
824
+ "grad_norm": 0.19447456300258636,
825
+ "learning_rate": 0.0003349067382815217,
826
+ "loss": 0.2409,
827
+ "step": 1100
828
+ },
829
+ {
830
+ "epoch": 0.9023789991796555,
831
+ "eval_loss": 0.23581616580486298,
832
+ "eval_runtime": 19.2514,
833
+ "eval_samples_per_second": 51.944,
834
+ "eval_steps_per_second": 0.831,
835
+ "step": 1100
836
+ },
837
+ {
838
+ "epoch": 0.918785890073831,
839
+ "grad_norm": 0.17990189790725708,
840
+ "learning_rate": 0.0003281388938941543,
841
+ "loss": 0.2386,
842
+ "step": 1120
843
+ },
844
+ {
845
+ "epoch": 0.918785890073831,
846
+ "eval_loss": 0.23487408459186554,
847
+ "eval_runtime": 19.2664,
848
+ "eval_samples_per_second": 51.904,
849
+ "eval_steps_per_second": 0.83,
850
+ "step": 1120
851
+ },
852
+ {
853
+ "epoch": 0.9351927809680065,
854
+ "grad_norm": 0.21182408928871155,
855
+ "learning_rate": 0.00032130696927595625,
856
+ "loss": 0.2467,
857
+ "step": 1140
858
+ },
859
+ {
860
+ "epoch": 0.9351927809680065,
861
+ "eval_loss": 0.23543472588062286,
862
+ "eval_runtime": 19.2877,
863
+ "eval_samples_per_second": 51.847,
864
+ "eval_steps_per_second": 0.83,
865
+ "step": 1140
866
+ },
867
+ {
868
+ "epoch": 0.9515996718621821,
869
+ "grad_norm": 0.21312598884105682,
870
+ "learning_rate": 0.0003144165671592514,
871
+ "loss": 0.2579,
872
+ "step": 1160
873
+ },
874
+ {
875
+ "epoch": 0.9515996718621821,
876
+ "eval_loss": 0.234401673078537,
877
+ "eval_runtime": 19.3028,
878
+ "eval_samples_per_second": 51.806,
879
+ "eval_steps_per_second": 0.829,
880
+ "step": 1160
881
+ },
882
+ {
883
+ "epoch": 0.9680065627563577,
884
+ "grad_norm": 0.21717378497123718,
885
+ "learning_rate": 0.000307473338232655,
886
+ "loss": 0.2382,
887
+ "step": 1180
888
+ },
889
+ {
890
+ "epoch": 0.9680065627563577,
891
+ "eval_loss": 0.23203913867473602,
892
+ "eval_runtime": 19.3297,
893
+ "eval_samples_per_second": 51.734,
894
+ "eval_steps_per_second": 0.828,
895
+ "step": 1180
896
+ },
897
+ {
898
+ "epoch": 0.9844134536505332,
899
+ "grad_norm": 0.19299665093421936,
900
+ "learning_rate": 0.00030048297650705157,
901
+ "loss": 0.2466,
902
+ "step": 1200
903
+ },
904
+ {
905
+ "epoch": 0.9844134536505332,
906
+ "eval_loss": 0.2323382943868637,
907
+ "eval_runtime": 19.3138,
908
+ "eval_samples_per_second": 51.776,
909
+ "eval_steps_per_second": 0.828,
910
+ "step": 1200
911
+ },
912
+ {
913
+ "epoch": 1.0008203445447088,
914
+ "grad_norm": 0.1672855168581009,
915
+ "learning_rate": 0.0002934512146460435,
916
+ "loss": 0.2521,
917
+ "step": 1220
918
+ },
919
+ {
920
+ "epoch": 1.0008203445447088,
921
+ "eval_loss": 0.2324332594871521,
922
+ "eval_runtime": 19.31,
923
+ "eval_samples_per_second": 51.787,
924
+ "eval_steps_per_second": 0.829,
925
+ "step": 1220
926
+ },
927
+ {
928
+ "epoch": 1.0172272354388843,
929
+ "grad_norm": 0.16059193015098572,
930
+ "learning_rate": 0.0002863838192647024,
931
+ "loss": 0.2094,
932
+ "step": 1240
933
+ },
934
+ {
935
+ "epoch": 1.0172272354388843,
936
+ "eval_loss": 0.2319324016571045,
937
+ "eval_runtime": 19.5412,
938
+ "eval_samples_per_second": 51.174,
939
+ "eval_steps_per_second": 0.819,
940
+ "step": 1240
941
+ },
942
+ {
943
+ "epoch": 1.03363412633306,
944
+ "grad_norm": 0.2210405170917511,
945
+ "learning_rate": 0.00027928658620047646,
946
+ "loss": 0.2161,
947
+ "step": 1260
948
+ },
949
+ {
950
+ "epoch": 1.03363412633306,
951
+ "eval_loss": 0.23039360344409943,
952
+ "eval_runtime": 19.2609,
953
+ "eval_samples_per_second": 51.919,
954
+ "eval_steps_per_second": 0.831,
955
+ "step": 1260
956
+ },
957
+ {
958
+ "epoch": 1.0500410172272354,
959
+ "grad_norm": 0.19888697564601898,
960
+ "learning_rate": 0.00027216533576013426,
961
+ "loss": 0.2154,
962
+ "step": 1280
963
+ },
964
+ {
965
+ "epoch": 1.0500410172272354,
966
+ "eval_loss": 0.22924970090389252,
967
+ "eval_runtime": 19.2656,
968
+ "eval_samples_per_second": 51.906,
969
+ "eval_steps_per_second": 0.83,
970
+ "step": 1280
971
+ },
972
+ {
973
+ "epoch": 1.066447908121411,
974
+ "grad_norm": 0.16993360221385956,
975
+ "learning_rate": 0.00026502590794664073,
976
+ "loss": 0.2097,
977
+ "step": 1300
978
+ },
979
+ {
980
+ "epoch": 1.066447908121411,
981
+ "eval_loss": 0.22938896715641022,
982
+ "eval_runtime": 19.2714,
983
+ "eval_samples_per_second": 51.89,
984
+ "eval_steps_per_second": 0.83,
985
+ "step": 1300
986
+ },
987
+ {
988
+ "epoch": 1.0828547990155866,
989
+ "grad_norm": 0.21647876501083374,
990
+ "learning_rate": 0.0002578741576698813,
991
+ "loss": 0.2078,
992
+ "step": 1320
993
+ },
994
+ {
995
+ "epoch": 1.0828547990155866,
996
+ "eval_loss": 0.2298092395067215,
997
+ "eval_runtime": 19.2675,
998
+ "eval_samples_per_second": 51.901,
999
+ "eval_steps_per_second": 0.83,
1000
+ "step": 1320
1001
+ },
1002
+ {
1003
+ "epoch": 1.099261689909762,
1004
+ "grad_norm": 0.17802399396896362,
1005
+ "learning_rate": 0.00025071594994516034,
1006
+ "loss": 0.216,
1007
+ "step": 1340
1008
+ },
1009
+ {
1010
+ "epoch": 1.099261689909762,
1011
+ "eval_loss": 0.22916612029075623,
1012
+ "eval_runtime": 19.3406,
1013
+ "eval_samples_per_second": 51.705,
1014
+ "eval_steps_per_second": 0.827,
1015
+ "step": 1340
1016
+ },
1017
+ {
1018
+ "epoch": 1.1156685808039377,
1019
+ "grad_norm": 0.17133191227912903,
1020
+ "learning_rate": 0.0002435571550834138,
1021
+ "loss": 0.2164,
1022
+ "step": 1360
1023
+ },
1024
+ {
1025
+ "epoch": 1.1156685808039377,
1026
+ "eval_loss": 0.22891317307949066,
1027
+ "eval_runtime": 19.275,
1028
+ "eval_samples_per_second": 51.881,
1029
+ "eval_steps_per_second": 0.83,
1030
+ "step": 1360
1031
+ },
1032
+ {
1033
+ "epoch": 1.1320754716981132,
1034
+ "grad_norm": 0.2046143114566803,
1035
+ "learning_rate": 0.00023640364387707725,
1036
+ "loss": 0.2149,
1037
+ "step": 1380
1038
+ },
1039
+ {
1040
+ "epoch": 1.1320754716981132,
1041
+ "eval_loss": 0.22800792753696442,
1042
+ "eval_runtime": 19.2662,
1043
+ "eval_samples_per_second": 51.904,
1044
+ "eval_steps_per_second": 0.83,
1045
+ "step": 1380
1046
+ },
1047
+ {
1048
+ "epoch": 1.1484823625922886,
1049
+ "grad_norm": 0.1857651323080063,
1050
+ "learning_rate": 0.0002292612827855605,
1051
+ "loss": 0.22,
1052
+ "step": 1400
1053
+ },
1054
+ {
1055
+ "epoch": 1.1484823625922886,
1056
+ "eval_loss": 0.22732825577259064,
1057
+ "eval_runtime": 19.3135,
1058
+ "eval_samples_per_second": 51.777,
1059
+ "eval_steps_per_second": 0.828,
1060
+ "step": 1400
1061
+ },
1062
+ {
1063
+ "epoch": 1.1648892534864643,
1064
+ "grad_norm": 0.17223379015922546,
1065
+ "learning_rate": 0.00022213592912427474,
1066
+ "loss": 0.2093,
1067
+ "step": 1420
1068
+ },
1069
+ {
1070
+ "epoch": 1.1648892534864643,
1071
+ "eval_loss": 0.2278290092945099,
1072
+ "eval_runtime": 19.3127,
1073
+ "eval_samples_per_second": 51.779,
1074
+ "eval_steps_per_second": 0.828,
1075
+ "step": 1420
1076
+ },
1077
+ {
1078
+ "epoch": 1.1812961443806398,
1079
+ "grad_norm": 0.19259098172187805,
1080
+ "learning_rate": 0.00021503342626115888,
1081
+ "loss": 0.2053,
1082
+ "step": 1440
1083
+ },
1084
+ {
1085
+ "epoch": 1.1812961443806398,
1086
+ "eval_loss": 0.22685323655605316,
1087
+ "eval_runtime": 19.314,
1088
+ "eval_samples_per_second": 51.776,
1089
+ "eval_steps_per_second": 0.828,
1090
+ "step": 1440
1091
+ },
1092
+ {
1093
+ "epoch": 1.1977030352748155,
1094
+ "grad_norm": 0.17038950324058533,
1095
+ "learning_rate": 0.000207959598824644,
1096
+ "loss": 0.2164,
1097
+ "step": 1460
1098
+ },
1099
+ {
1100
+ "epoch": 1.1977030352748155,
1101
+ "eval_loss": 0.2266845405101776,
1102
+ "eval_runtime": 19.3084,
1103
+ "eval_samples_per_second": 51.791,
1104
+ "eval_steps_per_second": 0.829,
1105
+ "step": 1460
1106
+ },
1107
+ {
1108
+ "epoch": 1.214109926168991,
1109
+ "grad_norm": 0.18405473232269287,
1110
+ "learning_rate": 0.00020092024792698528,
1111
+ "loss": 0.2173,
1112
+ "step": 1480
1113
+ },
1114
+ {
1115
+ "epoch": 1.214109926168991,
1116
+ "eval_loss": 0.22538632154464722,
1117
+ "eval_runtime": 19.3257,
1118
+ "eval_samples_per_second": 51.745,
1119
+ "eval_steps_per_second": 0.828,
1120
+ "step": 1480
1121
+ },
1122
+ {
1123
+ "epoch": 1.2305168170631666,
1124
+ "grad_norm": 0.19711589813232422,
1125
+ "learning_rate": 0.00019392114640687984,
1126
+ "loss": 0.2135,
1127
+ "step": 1500
1128
+ },
1129
+ {
1130
+ "epoch": 1.2305168170631666,
1131
+ "eval_loss": 0.22509323060512543,
1132
+ "eval_runtime": 19.343,
1133
+ "eval_samples_per_second": 51.698,
1134
+ "eval_steps_per_second": 0.827,
1135
+ "step": 1500
1136
+ },
1137
+ {
1138
+ "epoch": 1.246923707957342,
1139
+ "grad_norm": 0.19003860652446747,
1140
+ "learning_rate": 0.0001869680340952706,
1141
+ "loss": 0.2173,
1142
+ "step": 1520
1143
+ },
1144
+ {
1145
+ "epoch": 1.246923707957342,
1146
+ "eval_loss": 0.22505062818527222,
1147
+ "eval_runtime": 19.273,
1148
+ "eval_samples_per_second": 51.886,
1149
+ "eval_steps_per_second": 0.83,
1150
+ "step": 1520
1151
+ },
1152
+ {
1153
+ "epoch": 1.2633305988515175,
1154
+ "grad_norm": 0.1689046174287796,
1155
+ "learning_rate": 0.00018006661310821924,
1156
+ "loss": 0.2078,
1157
+ "step": 1540
1158
+ },
1159
+ {
1160
+ "epoch": 1.2633305988515175,
1161
+ "eval_loss": 0.22405382990837097,
1162
+ "eval_runtime": 19.3151,
1163
+ "eval_samples_per_second": 51.773,
1164
+ "eval_steps_per_second": 0.828,
1165
+ "step": 1540
1166
+ },
1167
+ {
1168
+ "epoch": 1.2797374897456932,
1169
+ "grad_norm": 0.2073843628168106,
1170
+ "learning_rate": 0.00017322254317070877,
1171
+ "loss": 0.2139,
1172
+ "step": 1560
1173
+ },
1174
+ {
1175
+ "epoch": 1.2797374897456932,
1176
+ "eval_loss": 0.2237548530101776,
1177
+ "eval_runtime": 19.3091,
1178
+ "eval_samples_per_second": 51.789,
1179
+ "eval_steps_per_second": 0.829,
1180
+ "step": 1560
1181
+ },
1182
+ {
1183
+ "epoch": 1.2961443806398687,
1184
+ "grad_norm": 0.20461881160736084,
1185
+ "learning_rate": 0.0001664414369752095,
1186
+ "loss": 0.2079,
1187
+ "step": 1580
1188
+ },
1189
+ {
1190
+ "epoch": 1.2961443806398687,
1191
+ "eval_loss": 0.22266964614391327,
1192
+ "eval_runtime": 19.316,
1193
+ "eval_samples_per_second": 51.771,
1194
+ "eval_steps_per_second": 0.828,
1195
+ "step": 1580
1196
+ },
1197
+ {
1198
+ "epoch": 1.3125512715340442,
1199
+ "grad_norm": 0.1744360774755478,
1200
+ "learning_rate": 0.00015972885557881666,
1201
+ "loss": 0.2093,
1202
+ "step": 1600
1203
+ },
1204
+ {
1205
+ "epoch": 1.3125512715340442,
1206
+ "eval_loss": 0.22231735289096832,
1207
+ "eval_runtime": 19.3159,
1208
+ "eval_samples_per_second": 51.771,
1209
+ "eval_steps_per_second": 0.828,
1210
+ "step": 1600
1211
+ },
1212
+ {
1213
+ "epoch": 1.3289581624282198,
1214
+ "grad_norm": 0.18790863454341888,
1215
+ "learning_rate": 0.0001530903038427322,
1216
+ "loss": 0.2051,
1217
+ "step": 1620
1218
+ },
1219
+ {
1220
+ "epoch": 1.3289581624282198,
1221
+ "eval_loss": 0.22288507223129272,
1222
+ "eval_runtime": 19.3315,
1223
+ "eval_samples_per_second": 51.729,
1224
+ "eval_steps_per_second": 0.828,
1225
+ "step": 1620
1226
+ },
1227
+ {
1228
+ "epoch": 1.3453650533223955,
1229
+ "grad_norm": 0.212313711643219,
1230
+ "learning_rate": 0.00014653122591783163,
1231
+ "loss": 0.2082,
1232
+ "step": 1640
1233
+ },
1234
+ {
1235
+ "epoch": 1.3453650533223955,
1236
+ "eval_loss": 0.2212054431438446,
1237
+ "eval_runtime": 19.3411,
1238
+ "eval_samples_per_second": 51.703,
1239
+ "eval_steps_per_second": 0.827,
1240
+ "step": 1640
1241
+ },
1242
+ {
1243
+ "epoch": 1.361771944216571,
1244
+ "grad_norm": 0.1949138194322586,
1245
+ "learning_rate": 0.0001400570007800196,
1246
+ "loss": 0.1985,
1247
+ "step": 1660
1248
+ },
1249
+ {
1250
+ "epoch": 1.361771944216571,
1251
+ "eval_loss": 0.22116906940937042,
1252
+ "eval_runtime": 19.3203,
1253
+ "eval_samples_per_second": 51.759,
1254
+ "eval_steps_per_second": 0.828,
1255
+ "step": 1660
1256
+ },
1257
+ {
1258
+ "epoch": 1.3781788351107465,
1259
+ "grad_norm": 0.17868603765964508,
1260
+ "learning_rate": 0.00013367293781903306,
1261
+ "loss": 0.202,
1262
+ "step": 1680
1263
+ },
1264
+ {
1265
+ "epoch": 1.3781788351107465,
1266
+ "eval_loss": 0.22035303711891174,
1267
+ "eval_runtime": 19.3367,
1268
+ "eval_samples_per_second": 51.715,
1269
+ "eval_steps_per_second": 0.827,
1270
+ "step": 1680
1271
+ },
1272
+ {
1273
+ "epoch": 1.3945857260049221,
1274
+ "grad_norm": 0.18695703148841858,
1275
+ "learning_rate": 0.0001273842724843103,
1276
+ "loss": 0.2081,
1277
+ "step": 1700
1278
+ },
1279
+ {
1280
+ "epoch": 1.3945857260049221,
1281
+ "eval_loss": 0.21980178356170654,
1282
+ "eval_runtime": 19.3238,
1283
+ "eval_samples_per_second": 51.75,
1284
+ "eval_steps_per_second": 0.828,
1285
+ "step": 1700
1286
+ },
1287
+ {
1288
+ "epoch": 1.4109926168990976,
1289
+ "grad_norm": 0.21980951726436615,
1290
+ "learning_rate": 0.00012119616199149901,
1291
+ "loss": 0.2158,
1292
+ "step": 1720
1293
+ },
1294
+ {
1295
+ "epoch": 1.4109926168990976,
1296
+ "eval_loss": 0.2198205292224884,
1297
+ "eval_runtime": 19.3311,
1298
+ "eval_samples_per_second": 51.73,
1299
+ "eval_steps_per_second": 0.828,
1300
+ "step": 1720
1301
+ },
1302
+ {
1303
+ "epoch": 1.427399507793273,
1304
+ "grad_norm": 0.1825166940689087,
1305
+ "learning_rate": 0.00011511368109312024,
1306
+ "loss": 0.2148,
1307
+ "step": 1740
1308
+ },
1309
+ {
1310
+ "epoch": 1.427399507793273,
1311
+ "eval_loss": 0.21902789175510406,
1312
+ "eval_runtime": 19.3223,
1313
+ "eval_samples_per_second": 51.754,
1314
+ "eval_steps_per_second": 0.828,
1315
+ "step": 1740
1316
+ },
1317
+ {
1318
+ "epoch": 1.4438063986874488,
1319
+ "grad_norm": 0.1830449253320694,
1320
+ "learning_rate": 0.00010914181791685987,
1321
+ "loss": 0.2099,
1322
+ "step": 1760
1323
+ },
1324
+ {
1325
+ "epoch": 1.4438063986874488,
1326
+ "eval_loss": 0.21859820187091827,
1327
+ "eval_runtime": 19.3402,
1328
+ "eval_samples_per_second": 51.706,
1329
+ "eval_steps_per_second": 0.827,
1330
+ "step": 1760
1331
+ },
1332
+ {
1333
+ "epoch": 1.4602132895816242,
1334
+ "grad_norm": 0.21329660713672638,
1335
+ "learning_rate": 0.00010328546987489997,
1336
+ "loss": 0.205,
1337
+ "step": 1780
1338
+ },
1339
+ {
1340
+ "epoch": 1.4602132895816242,
1341
+ "eval_loss": 0.21843338012695312,
1342
+ "eval_runtime": 19.3251,
1343
+ "eval_samples_per_second": 51.746,
1344
+ "eval_steps_per_second": 0.828,
1345
+ "step": 1780
1346
+ },
1347
+ {
1348
+ "epoch": 1.4766201804758,
1349
+ "grad_norm": 0.17494617402553558,
1350
+ "learning_rate": 9.754943964764229e-05,
1351
+ "loss": 0.2136,
1352
+ "step": 1800
1353
+ },
1354
+ {
1355
+ "epoch": 1.4766201804758,
1356
+ "eval_loss": 0.21801456809043884,
1357
+ "eval_runtime": 19.3336,
1358
+ "eval_samples_per_second": 51.723,
1359
+ "eval_steps_per_second": 0.828,
1360
+ "step": 1800
1361
+ },
1362
+ {
1363
+ "epoch": 1.4930270713699754,
1364
+ "grad_norm": 0.21063773334026337,
1365
+ "learning_rate": 9.193843124512227e-05,
1366
+ "loss": 0.2016,
1367
+ "step": 1820
1368
+ },
1369
+ {
1370
+ "epoch": 1.4930270713699754,
1371
+ "eval_loss": 0.21776027977466583,
1372
+ "eval_runtime": 19.3298,
1373
+ "eval_samples_per_second": 51.734,
1374
+ "eval_steps_per_second": 0.828,
1375
+ "step": 1820
1376
+ },
1377
+ {
1378
+ "epoch": 1.509433962264151,
1379
+ "grad_norm": 0.18612007796764374,
1380
+ "learning_rate": 8.645704614933819e-05,
1381
+ "loss": 0.2103,
1382
+ "step": 1840
1383
+ },
1384
+ {
1385
+ "epoch": 1.509433962264151,
1386
+ "eval_loss": 0.21697460114955902,
1387
+ "eval_runtime": 19.3496,
1388
+ "eval_samples_per_second": 51.681,
1389
+ "eval_steps_per_second": 0.827,
1390
+ "step": 1840
1391
+ },
1392
+ {
1393
+ "epoch": 1.5258408531583265,
1394
+ "grad_norm": 0.16927142441272736,
1395
+ "learning_rate": 8.110977954066303e-05,
1396
+ "loss": 0.2068,
1397
+ "step": 1860
1398
+ },
1399
+ {
1400
+ "epoch": 1.5258408531583265,
1401
+ "eval_loss": 0.21691332757472992,
1402
+ "eval_runtime": 19.3177,
1403
+ "eval_samples_per_second": 51.766,
1404
+ "eval_steps_per_second": 0.828,
1405
+ "step": 1860
1406
+ },
1407
+ {
1408
+ "epoch": 1.542247744052502,
1409
+ "grad_norm": 0.2074907124042511,
1410
+ "learning_rate": 7.590101661143234e-05,
1411
+ "loss": 0.2013,
1412
+ "step": 1880
1413
+ },
1414
+ {
1415
+ "epoch": 1.542247744052502,
1416
+ "eval_loss": 0.21642935276031494,
1417
+ "eval_runtime": 19.3403,
1418
+ "eval_samples_per_second": 51.706,
1419
+ "eval_steps_per_second": 0.827,
1420
+ "step": 1880
1421
+ },
1422
+ {
1423
+ "epoch": 1.5586546349466777,
1424
+ "grad_norm": 0.19013307988643646,
1425
+ "learning_rate": 7.083502896973017e-05,
1426
+ "loss": 0.2022,
1427
+ "step": 1900
1428
+ },
1429
+ {
1430
+ "epoch": 1.5586546349466777,
1431
+ "eval_loss": 0.2169981449842453,
1432
+ "eval_runtime": 19.3364,
1433
+ "eval_samples_per_second": 51.716,
1434
+ "eval_steps_per_second": 0.827,
1435
+ "step": 1900
1436
+ },
1437
+ {
1438
+ "epoch": 1.5750615258408531,
1439
+ "grad_norm": 0.18733009696006775,
1440
+ "learning_rate": 6.591597113632447e-05,
1441
+ "loss": 0.206,
1442
+ "step": 1920
1443
+ },
1444
+ {
1445
+ "epoch": 1.5750615258408531,
1446
+ "eval_loss": 0.21625757217407227,
1447
+ "eval_runtime": 19.3307,
1448
+ "eval_samples_per_second": 51.731,
1449
+ "eval_steps_per_second": 0.828,
1450
+ "step": 1920
1451
+ },
1452
+ {
1453
+ "epoch": 1.5914684167350286,
1454
+ "grad_norm": 0.2090052366256714,
1455
+ "learning_rate": 6.114787713762287e-05,
1456
+ "loss": 0.2072,
1457
+ "step": 1940
1458
+ },
1459
+ {
1460
+ "epoch": 1.5914684167350286,
1461
+ "eval_loss": 0.21603082120418549,
1462
+ "eval_runtime": 19.3283,
1463
+ "eval_samples_per_second": 51.738,
1464
+ "eval_steps_per_second": 0.828,
1465
+ "step": 1940
1466
+ },
1467
+ {
1468
+ "epoch": 1.6078753076292043,
1469
+ "grad_norm": 0.1998095065355301,
1470
+ "learning_rate": 5.653465719744388e-05,
1471
+ "loss": 0.208,
1472
+ "step": 1960
1473
+ },
1474
+ {
1475
+ "epoch": 1.6078753076292043,
1476
+ "eval_loss": 0.2160104215145111,
1477
+ "eval_runtime": 19.3483,
1478
+ "eval_samples_per_second": 51.684,
1479
+ "eval_steps_per_second": 0.827,
1480
+ "step": 1960
1481
+ },
1482
+ {
1483
+ "epoch": 1.62428219852338,
1484
+ "grad_norm": 0.1849084049463272,
1485
+ "learning_rate": 5.2080094530316174e-05,
1486
+ "loss": 0.2032,
1487
+ "step": 1980
1488
+ },
1489
+ {
1490
+ "epoch": 1.62428219852338,
1491
+ "eval_loss": 0.21572308242321014,
1492
+ "eval_runtime": 19.3051,
1493
+ "eval_samples_per_second": 51.8,
1494
+ "eval_steps_per_second": 0.829,
1495
+ "step": 1980
1496
+ },
1497
+ {
1498
+ "epoch": 1.6406890894175554,
1499
+ "grad_norm": 0.19785718619823456,
1500
+ "learning_rate": 4.7787842238936006e-05,
1501
+ "loss": 0.2046,
1502
+ "step": 2000
1503
+ },
1504
+ {
1505
+ "epoch": 1.6406890894175554,
1506
+ "eval_loss": 0.21561187505722046,
1507
+ "eval_runtime": 19.288,
1508
+ "eval_samples_per_second": 51.846,
1509
+ "eval_steps_per_second": 0.83,
1510
+ "step": 2000
1511
+ },
1512
+ {
1513
+ "epoch": 1.6570959803117309,
1514
+ "grad_norm": 0.20306991040706635,
1515
+ "learning_rate": 4.36614203183264e-05,
1516
+ "loss": 0.1967,
1517
+ "step": 2020
1518
+ },
1519
+ {
1520
+ "epoch": 1.6570959803117309,
1521
+ "eval_loss": 0.21528854966163635,
1522
+ "eval_runtime": 19.2792,
1523
+ "eval_samples_per_second": 51.869,
1524
+ "eval_steps_per_second": 0.83,
1525
+ "step": 2020
1526
+ },
1527
+ {
1528
+ "epoch": 1.6735028712059066,
1529
+ "grad_norm": 0.16694621741771698,
1530
+ "learning_rate": 3.970421276915584e-05,
1531
+ "loss": 0.1987,
1532
+ "step": 2040
1533
+ },
1534
+ {
1535
+ "epoch": 1.6735028712059066,
1536
+ "eval_loss": 0.21499016880989075,
1537
+ "eval_runtime": 19.3113,
1538
+ "eval_samples_per_second": 51.783,
1539
+ "eval_steps_per_second": 0.829,
1540
+ "step": 2040
1541
+ },
1542
+ {
1543
+ "epoch": 1.689909762100082,
1544
+ "grad_norm": 0.21434736251831055,
1545
+ "learning_rate": 3.5919464822583486e-05,
1546
+ "loss": 0.2099,
1547
+ "step": 2060
1548
+ },
1549
+ {
1550
+ "epoch": 1.689909762100082,
1551
+ "eval_loss": 0.21498040854930878,
1552
+ "eval_runtime": 19.2737,
1553
+ "eval_samples_per_second": 51.884,
1554
+ "eval_steps_per_second": 0.83,
1555
+ "step": 2060
1556
+ },
1557
+ {
1558
+ "epoch": 1.7063166529942575,
1559
+ "grad_norm": 0.1937168836593628,
1560
+ "learning_rate": 3.2310280278905784e-05,
1561
+ "loss": 0.1976,
1562
+ "step": 2080
1563
+ },
1564
+ {
1565
+ "epoch": 1.7063166529942575,
1566
+ "eval_loss": 0.21494653820991516,
1567
+ "eval_runtime": 19.2962,
1568
+ "eval_samples_per_second": 51.824,
1569
+ "eval_steps_per_second": 0.829,
1570
+ "step": 2080
1571
+ },
1572
+ {
1573
+ "epoch": 1.7227235438884332,
1574
+ "grad_norm": 0.1854638308286667,
1575
+ "learning_rate": 2.8879618962189326e-05,
1576
+ "loss": 0.1988,
1577
+ "step": 2100
1578
+ },
1579
+ {
1580
+ "epoch": 1.7227235438884332,
1581
+ "eval_loss": 0.21440669894218445,
1582
+ "eval_runtime": 19.2908,
1583
+ "eval_samples_per_second": 51.838,
1584
+ "eval_steps_per_second": 0.829,
1585
+ "step": 2100
1586
+ },
1587
+ {
1588
+ "epoch": 1.7391304347826086,
1589
+ "grad_norm": 0.17554204165935516,
1590
+ "learning_rate": 2.563029429297428e-05,
1591
+ "loss": 0.1994,
1592
+ "step": 2120
1593
+ },
1594
+ {
1595
+ "epoch": 1.7391304347826086,
1596
+ "eval_loss": 0.21440716087818146,
1597
+ "eval_runtime": 19.3334,
1598
+ "eval_samples_per_second": 51.724,
1599
+ "eval_steps_per_second": 0.828,
1600
+ "step": 2120
1601
+ },
1602
+ {
1603
+ "epoch": 1.755537325676784,
1604
+ "grad_norm": 0.19237037003040314,
1605
+ "learning_rate": 2.2564970981041792e-05,
1606
+ "loss": 0.1948,
1607
+ "step": 2140
1608
+ },
1609
+ {
1610
+ "epoch": 1.755537325676784,
1611
+ "eval_loss": 0.2138786017894745,
1612
+ "eval_runtime": 19.3232,
1613
+ "eval_samples_per_second": 51.751,
1614
+ "eval_steps_per_second": 0.828,
1615
+ "step": 2140
1616
+ },
1617
+ {
1618
+ "epoch": 1.7719442165709598,
1619
+ "grad_norm": 0.19103820621967316,
1620
+ "learning_rate": 1.968616284013583e-05,
1621
+ "loss": 0.1943,
1622
+ "step": 2160
1623
+ },
1624
+ {
1625
+ "epoch": 1.7719442165709598,
1626
+ "eval_loss": 0.21409012377262115,
1627
+ "eval_runtime": 19.311,
1628
+ "eval_samples_per_second": 51.784,
1629
+ "eval_steps_per_second": 0.829,
1630
+ "step": 2160
1631
+ },
1632
+ {
1633
+ "epoch": 1.7883511074651355,
1634
+ "grad_norm": 0.1740914285182953,
1635
+ "learning_rate": 1.6996230726431526e-05,
1636
+ "loss": 0.1988,
1637
+ "step": 2180
1638
+ },
1639
+ {
1640
+ "epoch": 1.7883511074651355,
1641
+ "eval_loss": 0.21403931081295013,
1642
+ "eval_runtime": 19.3307,
1643
+ "eval_samples_per_second": 51.731,
1644
+ "eval_steps_per_second": 0.828,
1645
+ "step": 2180
1646
+ },
1647
+ {
1648
+ "epoch": 1.804757998359311,
1649
+ "grad_norm": 0.22716930508613586,
1650
+ "learning_rate": 1.4497380602442378e-05,
1651
+ "loss": 0.2052,
1652
+ "step": 2200
1653
+ },
1654
+ {
1655
+ "epoch": 1.804757998359311,
1656
+ "eval_loss": 0.21394230425357819,
1657
+ "eval_runtime": 19.3327,
1658
+ "eval_samples_per_second": 51.726,
1659
+ "eval_steps_per_second": 0.828,
1660
+ "step": 2200
1661
+ },
1662
+ {
1663
+ "epoch": 1.8211648892534864,
1664
+ "grad_norm": 0.19576476514339447,
1665
+ "learning_rate": 1.2191661727951498e-05,
1666
+ "loss": 0.1926,
1667
+ "step": 2220
1668
+ },
1669
+ {
1670
+ "epoch": 1.8211648892534864,
1671
+ "eval_loss": 0.2139243483543396,
1672
+ "eval_runtime": 19.3379,
1673
+ "eval_samples_per_second": 51.712,
1674
+ "eval_steps_per_second": 0.827,
1675
+ "step": 2220
1676
+ },
1677
+ {
1678
+ "epoch": 1.837571780147662,
1679
+ "grad_norm": 0.1827758103609085,
1680
+ "learning_rate": 1.0080964979452573e-05,
1681
+ "loss": 0.2017,
1682
+ "step": 2240
1683
+ },
1684
+ {
1685
+ "epoch": 1.837571780147662,
1686
+ "eval_loss": 0.21385584771633148,
1687
+ "eval_runtime": 19.3511,
1688
+ "eval_samples_per_second": 51.677,
1689
+ "eval_steps_per_second": 0.827,
1690
+ "step": 2240
1691
+ },
1692
+ {
1693
+ "epoch": 1.8539786710418376,
1694
+ "grad_norm": 0.17563924193382263,
1695
+ "learning_rate": 8.16702129947769e-06,
1696
+ "loss": 0.1962,
1697
+ "step": 2260
1698
+ },
1699
+ {
1700
+ "epoch": 1.8539786710418376,
1701
+ "eval_loss": 0.21377405524253845,
1702
+ "eval_runtime": 19.3058,
1703
+ "eval_samples_per_second": 51.798,
1704
+ "eval_steps_per_second": 0.829,
1705
+ "step": 2260
1706
+ },
1707
+ {
1708
+ "epoch": 1.870385561936013,
1709
+ "grad_norm": 0.18761037290096283,
1710
+ "learning_rate": 6.451400277084324e-06,
1711
+ "loss": 0.2006,
1712
+ "step": 2280
1713
+ },
1714
+ {
1715
+ "epoch": 1.870385561936013,
1716
+ "eval_loss": 0.2137526124715805,
1717
+ "eval_runtime": 19.3037,
1718
+ "eval_samples_per_second": 51.804,
1719
+ "eval_steps_per_second": 0.829,
1720
+ "step": 2280
1721
+ },
1722
+ {
1723
+ "epoch": 1.8867924528301887,
1724
+ "grad_norm": 0.23708121478557587,
1725
+ "learning_rate": 4.935508860664601e-06,
1726
+ "loss": 0.2099,
1727
+ "step": 2300
1728
+ },
1729
+ {
1730
+ "epoch": 1.8867924528301887,
1731
+ "eval_loss": 0.21373669803142548,
1732
+ "eval_runtime": 19.2823,
1733
+ "eval_samples_per_second": 51.861,
1734
+ "eval_steps_per_second": 0.83,
1735
+ "step": 2300
1736
+ },
1737
+ {
1738
+ "epoch": 1.9031993437243644,
1739
+ "grad_norm": 0.16906791925430298,
1740
+ "learning_rate": 3.6205902041337725e-06,
1741
+ "loss": 0.2047,
1742
+ "step": 2320
1743
+ },
1744
+ {
1745
+ "epoch": 1.9031993437243644,
1746
+ "eval_loss": 0.21369938552379608,
1747
+ "eval_runtime": 19.2627,
1748
+ "eval_samples_per_second": 51.914,
1749
+ "eval_steps_per_second": 0.831,
1750
+ "step": 2320
1751
+ },
1752
+ {
1753
+ "epoch": 1.9196062346185396,
1754
+ "grad_norm": 0.16215278208255768,
1755
+ "learning_rate": 2.507722647443117e-06,
1756
+ "loss": 0.1974,
1757
+ "step": 2340
1758
+ },
1759
+ {
1760
+ "epoch": 1.9196062346185396,
1761
+ "eval_loss": 0.2136177122592926,
1762
+ "eval_runtime": 19.2972,
1763
+ "eval_samples_per_second": 51.821,
1764
+ "eval_steps_per_second": 0.829,
1765
+ "step": 2340
1766
+ },
1767
+ {
1768
+ "epoch": 1.9360131255127153,
1769
+ "grad_norm": 0.20447717607021332,
1770
+ "learning_rate": 1.5978188322536835e-06,
1771
+ "loss": 0.2027,
1772
+ "step": 2360
1773
+ },
1774
+ {
1775
+ "epoch": 1.9360131255127153,
1776
+ "eval_loss": 0.21356625854969025,
1777
+ "eval_runtime": 19.293,
1778
+ "eval_samples_per_second": 51.832,
1779
+ "eval_steps_per_second": 0.829,
1780
+ "step": 2360
1781
+ },
1782
+ {
1783
+ "epoch": 1.952420016406891,
1784
+ "grad_norm": 0.2655256986618042,
1785
+ "learning_rate": 8.916249534961507e-07,
1786
+ "loss": 0.2116,
1787
+ "step": 2380
1788
+ },
1789
+ {
1790
+ "epoch": 1.952420016406891,
1791
+ "eval_loss": 0.21358250081539154,
1792
+ "eval_runtime": 19.2917,
1793
+ "eval_samples_per_second": 51.836,
1794
+ "eval_steps_per_second": 0.829,
1795
+ "step": 2380
1796
+ },
1797
+ {
1798
+ "epoch": 1.9688269073010665,
1799
+ "grad_norm": 0.174843430519104,
1800
+ "learning_rate": 3.8972014743038356e-07,
1801
+ "loss": 0.2067,
1802
+ "step": 2400
1803
+ },
1804
+ {
1805
+ "epoch": 1.9688269073010665,
1806
+ "eval_loss": 0.21355371177196503,
1807
+ "eval_runtime": 19.2828,
1808
+ "eval_samples_per_second": 51.86,
1809
+ "eval_steps_per_second": 0.83,
1810
+ "step": 2400
1811
+ },
1812
+ {
1813
+ "epoch": 1.985233798195242,
1814
+ "grad_norm": 0.1434665322303772,
1815
+ "learning_rate": 9.251601670681731e-08,
1816
+ "loss": 0.1925,
1817
+ "step": 2420
1818
+ },
1819
+ {
1820
+ "epoch": 1.985233798195242,
1821
+ "eval_loss": 0.21353401243686676,
1822
+ "eval_runtime": 19.3017,
1823
+ "eval_samples_per_second": 51.809,
1824
+ "eval_steps_per_second": 0.829,
1825
+ "step": 2420
1826
+ },
1827
+ {
1828
+ "epoch": 2.0,
1829
+ "step": 2438,
1830
+ "total_flos": 1.58523627405312e+18,
1831
+ "train_loss": 0.24909442228015277,
1832
+ "train_runtime": 4668.9222,
1833
+ "train_samples_per_second": 16.706,
1834
+ "train_steps_per_second": 0.522
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_exps7/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_exps7/ft/added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[PAD]": 32000
3
+ }
nl_tasks/run_all/run_exps7/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_exps7/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exps7/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_exps7/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_exps7/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58f97845f9ed42042b847edcf5ad27e8d80cf2ab05880b87b43f8b1a0f53c093
3
+ size 6481
nl_tasks/run_all/run_exps7/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_exps7/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a75798e7fadeed90fb5b885dfdf511e5b86e24d986a7bb4dd34eaf95551e6046
3
+ size 33602659
nl_tasks/run_all/run_exps7/trainer_state.json ADDED
@@ -0,0 +1,1274 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 0.016406890894175553,
14
+ "grad_norm": 0.24184416234493256,
15
+ "learning_rate": 6.229508196721313e-05,
16
+ "loss": 0.7209,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.03281378178835111,
21
+ "grad_norm": 0.21366377174854279,
22
+ "learning_rate": 0.0001278688524590164,
23
+ "loss": 0.4876,
24
+ "step": 40
25
+ },
26
+ {
27
+ "epoch": 0.04101722723543889,
28
+ "eval_loss": 0.3812011182308197,
29
+ "eval_runtime": 20.0572,
30
+ "eval_samples_per_second": 49.857,
31
+ "eval_steps_per_second": 0.798,
32
+ "step": 50
33
+ },
34
+ {
35
+ "epoch": 0.04922067268252666,
36
+ "grad_norm": 0.16566194593906403,
37
+ "learning_rate": 0.00019344262295081967,
38
+ "loss": 0.3897,
39
+ "step": 60
40
+ },
41
+ {
42
+ "epoch": 0.06562756357670221,
43
+ "grad_norm": 0.20982611179351807,
44
+ "learning_rate": 0.000259016393442623,
45
+ "loss": 0.368,
46
+ "step": 80
47
+ },
48
+ {
49
+ "epoch": 0.08203445447087777,
50
+ "grad_norm": 0.1964726746082306,
51
+ "learning_rate": 0.00032459016393442624,
52
+ "loss": 0.3518,
53
+ "step": 100
54
+ },
55
+ {
56
+ "epoch": 0.08203445447087777,
57
+ "eval_loss": 0.3327180743217468,
58
+ "eval_runtime": 19.7061,
59
+ "eval_samples_per_second": 50.746,
60
+ "eval_steps_per_second": 0.812,
61
+ "step": 100
62
+ },
63
+ {
64
+ "epoch": 0.09844134536505332,
65
+ "grad_norm": 0.19582737982273102,
66
+ "learning_rate": 0.0003901639344262295,
67
+ "loss": 0.3342,
68
+ "step": 120
69
+ },
70
+ {
71
+ "epoch": 0.11484823625922888,
72
+ "grad_norm": 0.19419531524181366,
73
+ "learning_rate": 0.0004557377049180328,
74
+ "loss": 0.3162,
75
+ "step": 140
76
+ },
77
+ {
78
+ "epoch": 0.12305168170631665,
79
+ "eval_loss": 0.3151061236858368,
80
+ "eval_runtime": 19.6859,
81
+ "eval_samples_per_second": 50.798,
82
+ "eval_steps_per_second": 0.813,
83
+ "step": 150
84
+ },
85
+ {
86
+ "epoch": 0.13125512715340443,
87
+ "grad_norm": 0.18924717605113983,
88
+ "learning_rate": 0.0005213114754098361,
89
+ "loss": 0.3169,
90
+ "step": 160
91
+ },
92
+ {
93
+ "epoch": 0.14766201804757997,
94
+ "grad_norm": 0.28647610545158386,
95
+ "learning_rate": 0.0005868852459016393,
96
+ "loss": 0.3283,
97
+ "step": 180
98
+ },
99
+ {
100
+ "epoch": 0.16406890894175555,
101
+ "grad_norm": 0.20606732368469238,
102
+ "learning_rate": 0.0006524590163934426,
103
+ "loss": 0.319,
104
+ "step": 200
105
+ },
106
+ {
107
+ "epoch": 0.16406890894175555,
108
+ "eval_loss": 0.30781295895576477,
109
+ "eval_runtime": 19.6692,
110
+ "eval_samples_per_second": 50.841,
111
+ "eval_steps_per_second": 0.813,
112
+ "step": 200
113
+ },
114
+ {
115
+ "epoch": 0.1804757998359311,
116
+ "grad_norm": 0.2141265571117401,
117
+ "learning_rate": 0.0007180327868852459,
118
+ "loss": 0.3265,
119
+ "step": 220
120
+ },
121
+ {
122
+ "epoch": 0.19688269073010664,
123
+ "grad_norm": 0.21591129899024963,
124
+ "learning_rate": 0.0007836065573770492,
125
+ "loss": 0.3163,
126
+ "step": 240
127
+ },
128
+ {
129
+ "epoch": 0.20508613617719443,
130
+ "eval_loss": 0.30268049240112305,
131
+ "eval_runtime": 19.688,
132
+ "eval_samples_per_second": 50.792,
133
+ "eval_steps_per_second": 0.813,
134
+ "step": 250
135
+ },
136
+ {
137
+ "epoch": 0.2132895816242822,
138
+ "grad_norm": 0.2008507251739502,
139
+ "learning_rate": 0.0007999077381164415,
140
+ "loss": 0.3139,
141
+ "step": 260
142
+ },
143
+ {
144
+ "epoch": 0.22969647251845776,
145
+ "grad_norm": 0.20609350502490997,
146
+ "learning_rate": 0.0007994977711218688,
147
+ "loss": 0.3174,
148
+ "step": 280
149
+ },
150
+ {
151
+ "epoch": 0.2461033634126333,
152
+ "grad_norm": 0.256902277469635,
153
+ "learning_rate": 0.0007987601835517546,
154
+ "loss": 0.3024,
155
+ "step": 300
156
+ },
157
+ {
158
+ "epoch": 0.2461033634126333,
159
+ "eval_loss": 0.2971021234989166,
160
+ "eval_runtime": 19.7063,
161
+ "eval_samples_per_second": 50.745,
162
+ "eval_steps_per_second": 0.812,
163
+ "step": 300
164
+ },
165
+ {
166
+ "epoch": 0.26251025430680885,
167
+ "grad_norm": 0.2500735819339752,
168
+ "learning_rate": 0.0007976955802877319,
169
+ "loss": 0.2939,
170
+ "step": 320
171
+ },
172
+ {
173
+ "epoch": 0.27891714520098443,
174
+ "grad_norm": 0.28339460492134094,
175
+ "learning_rate": 0.0007963048343908275,
176
+ "loss": 0.309,
177
+ "step": 340
178
+ },
179
+ {
180
+ "epoch": 0.2871205906480722,
181
+ "eval_loss": 0.29064130783081055,
182
+ "eval_runtime": 19.6752,
183
+ "eval_samples_per_second": 50.826,
184
+ "eval_steps_per_second": 0.813,
185
+ "step": 350
186
+ },
187
+ {
188
+ "epoch": 0.29532403609515995,
189
+ "grad_norm": 0.25881996750831604,
190
+ "learning_rate": 0.0007945890863854799,
191
+ "loss": 0.296,
192
+ "step": 360
193
+ },
194
+ {
195
+ "epoch": 0.3117309269893355,
196
+ "grad_norm": 0.24443098902702332,
197
+ "learning_rate": 0.0007925497433242176,
198
+ "loss": 0.2975,
199
+ "step": 380
200
+ },
201
+ {
202
+ "epoch": 0.3281378178835111,
203
+ "grad_norm": 0.32072708010673523,
204
+ "learning_rate": 0.0007901884776337618,
205
+ "loss": 0.3029,
206
+ "step": 400
207
+ },
208
+ {
209
+ "epoch": 0.3281378178835111,
210
+ "eval_loss": 0.283695787191391,
211
+ "eval_runtime": 19.6884,
212
+ "eval_samples_per_second": 50.791,
213
+ "eval_steps_per_second": 0.813,
214
+ "step": 400
215
+ },
216
+ {
217
+ "epoch": 0.3445447087776866,
218
+ "grad_norm": 0.2378903180360794,
219
+ "learning_rate": 0.0007875072257435006,
220
+ "loss": 0.2981,
221
+ "step": 420
222
+ },
223
+ {
224
+ "epoch": 0.3609515996718622,
225
+ "grad_norm": 0.2585814595222473,
226
+ "learning_rate": 0.0007845081864974602,
227
+ "loss": 0.2989,
228
+ "step": 440
229
+ },
230
+ {
231
+ "epoch": 0.36915504511895,
232
+ "eval_loss": 0.2760768532752991,
233
+ "eval_runtime": 19.6994,
234
+ "eval_samples_per_second": 50.763,
235
+ "eval_steps_per_second": 0.812,
236
+ "step": 450
237
+ },
238
+ {
239
+ "epoch": 0.37735849056603776,
240
+ "grad_norm": 0.18312865495681763,
241
+ "learning_rate": 0.000781193819351074,
242
+ "loss": 0.2984,
243
+ "step": 460
244
+ },
245
+ {
246
+ "epoch": 0.3937653814602133,
247
+ "grad_norm": 0.2005549520254135,
248
+ "learning_rate": 0.00077756684235423,
249
+ "loss": 0.2853,
250
+ "step": 480
251
+ },
252
+ {
253
+ "epoch": 0.41017227235438886,
254
+ "grad_norm": 0.20137976109981537,
255
+ "learning_rate": 0.0007736302299222494,
256
+ "loss": 0.2794,
257
+ "step": 500
258
+ },
259
+ {
260
+ "epoch": 0.41017227235438886,
261
+ "eval_loss": 0.27147459983825684,
262
+ "eval_runtime": 19.7024,
263
+ "eval_samples_per_second": 50.755,
264
+ "eval_steps_per_second": 0.812,
265
+ "step": 500
266
+ },
267
+ {
268
+ "epoch": 0.4265791632485644,
269
+ "grad_norm": 0.22666695713996887,
270
+ "learning_rate": 0.0007693872103966245,
271
+ "loss": 0.2866,
272
+ "step": 520
273
+ },
274
+ {
275
+ "epoch": 0.44298605414273995,
276
+ "grad_norm": 0.2422036975622177,
277
+ "learning_rate": 0.0007648412633975174,
278
+ "loss": 0.2819,
279
+ "step": 540
280
+ },
281
+ {
282
+ "epoch": 0.45118949958982774,
283
+ "eval_loss": 0.26572293043136597,
284
+ "eval_runtime": 19.7131,
285
+ "eval_samples_per_second": 50.728,
286
+ "eval_steps_per_second": 0.812,
287
+ "step": 550
288
+ },
289
+ {
290
+ "epoch": 0.4593929450369155,
291
+ "grad_norm": 0.21293747425079346,
292
+ "learning_rate": 0.0007599961169701884,
293
+ "loss": 0.2755,
294
+ "step": 560
295
+ },
296
+ {
297
+ "epoch": 0.47579983593109104,
298
+ "grad_norm": 0.21169118583202362,
299
+ "learning_rate": 0.0007548557445276972,
300
+ "loss": 0.2703,
301
+ "step": 580
302
+ },
303
+ {
304
+ "epoch": 0.4922067268252666,
305
+ "grad_norm": 0.22510677576065063,
306
+ "learning_rate": 0.000749424361592382,
307
+ "loss": 0.2742,
308
+ "step": 600
309
+ },
310
+ {
311
+ "epoch": 0.4922067268252666,
312
+ "eval_loss": 0.26300400495529175,
313
+ "eval_runtime": 19.715,
314
+ "eval_samples_per_second": 50.723,
315
+ "eval_steps_per_second": 0.812,
316
+ "step": 600
317
+ },
318
+ {
319
+ "epoch": 0.5086136177194421,
320
+ "grad_norm": 0.19537706673145294,
321
+ "learning_rate": 0.0007437064223387884,
322
+ "loss": 0.2631,
323
+ "step": 620
324
+ },
325
+ {
326
+ "epoch": 0.5250205086136177,
327
+ "grad_norm": 0.1731526255607605,
328
+ "learning_rate": 0.0007377066159408859,
329
+ "loss": 0.2644,
330
+ "step": 640
331
+ },
332
+ {
333
+ "epoch": 0.5332239540607056,
334
+ "eval_loss": 0.2575521469116211,
335
+ "eval_runtime": 19.702,
336
+ "eval_samples_per_second": 50.756,
337
+ "eval_steps_per_second": 0.812,
338
+ "step": 650
339
+ },
340
+ {
341
+ "epoch": 0.5414273995077933,
342
+ "grad_norm": 0.23598860204219818,
343
+ "learning_rate": 0.0007314298627265644,
344
+ "loss": 0.2611,
345
+ "step": 660
346
+ },
347
+ {
348
+ "epoch": 0.5578342904019689,
349
+ "grad_norm": 0.15940260887145996,
350
+ "learning_rate": 0.0007248813101425663,
351
+ "loss": 0.2604,
352
+ "step": 680
353
+ },
354
+ {
355
+ "epoch": 0.5742411812961444,
356
+ "grad_norm": 0.1660739630460739,
357
+ "learning_rate": 0.0007180663285331632,
358
+ "loss": 0.2545,
359
+ "step": 700
360
+ },
361
+ {
362
+ "epoch": 0.5742411812961444,
363
+ "eval_loss": 0.2548382580280304,
364
+ "eval_runtime": 19.6723,
365
+ "eval_samples_per_second": 50.833,
366
+ "eval_steps_per_second": 0.813,
367
+ "step": 700
368
+ },
369
+ {
370
+ "epoch": 0.5906480721903199,
371
+ "grad_norm": 0.18981686234474182,
372
+ "learning_rate": 0.0007109905067360368,
373
+ "loss": 0.2497,
374
+ "step": 720
375
+ },
376
+ {
377
+ "epoch": 0.6070549630844955,
378
+ "grad_norm": 0.1564050018787384,
379
+ "learning_rate": 0.0007036596474989804,
380
+ "loss": 0.2591,
381
+ "step": 740
382
+ },
383
+ {
384
+ "epoch": 0.6152584085315833,
385
+ "eval_loss": 0.2515643537044525,
386
+ "eval_runtime": 19.6455,
387
+ "eval_samples_per_second": 50.902,
388
+ "eval_steps_per_second": 0.814,
389
+ "step": 750
390
+ },
391
+ {
392
+ "epoch": 0.623461853978671,
393
+ "grad_norm": 0.16761599481105804,
394
+ "learning_rate": 0.0006960797627211732,
395
+ "loss": 0.2651,
396
+ "step": 760
397
+ },
398
+ {
399
+ "epoch": 0.6398687448728466,
400
+ "grad_norm": 0.18286292254924774,
401
+ "learning_rate": 0.0006882570685229366,
402
+ "loss": 0.2592,
403
+ "step": 780
404
+ },
405
+ {
406
+ "epoch": 0.6562756357670222,
407
+ "grad_norm": 0.19651605188846588,
408
+ "learning_rate": 0.0006801979801480114,
409
+ "loss": 0.2589,
410
+ "step": 800
411
+ },
412
+ {
413
+ "epoch": 0.6562756357670222,
414
+ "eval_loss": 0.247966930270195,
415
+ "eval_runtime": 19.6936,
416
+ "eval_samples_per_second": 50.778,
417
+ "eval_steps_per_second": 0.812,
418
+ "step": 800
419
+ },
420
+ {
421
+ "epoch": 0.6726825266611977,
422
+ "grad_norm": 0.17967994511127472,
423
+ "learning_rate": 0.0006719091067025366,
424
+ "loss": 0.2467,
425
+ "step": 820
426
+ },
427
+ {
428
+ "epoch": 0.6890894175553732,
429
+ "grad_norm": 0.1687493771314621,
430
+ "learning_rate": 0.000663397245735047,
431
+ "loss": 0.2586,
432
+ "step": 840
433
+ },
434
+ {
435
+ "epoch": 0.6972928630024611,
436
+ "eval_loss": 0.24560077488422394,
437
+ "eval_runtime": 19.7034,
438
+ "eval_samples_per_second": 50.753,
439
+ "eval_steps_per_second": 0.812,
440
+ "step": 850
441
+ },
442
+ {
443
+ "epoch": 0.7054963084495488,
444
+ "grad_norm": 0.1722402125597,
445
+ "learning_rate": 0.0006546693776619325,
446
+ "loss": 0.2646,
447
+ "step": 860
448
+ },
449
+ {
450
+ "epoch": 0.7219031993437244,
451
+ "grad_norm": 0.1573437601327896,
452
+ "learning_rate": 0.0006457326600429292,
453
+ "loss": 0.2557,
454
+ "step": 880
455
+ },
456
+ {
457
+ "epoch": 0.7383100902379,
458
+ "grad_norm": 0.18602094054222107,
459
+ "learning_rate": 0.000636594421711341,
460
+ "loss": 0.2464,
461
+ "step": 900
462
+ },
463
+ {
464
+ "epoch": 0.7383100902379,
465
+ "eval_loss": 0.24180082976818085,
466
+ "eval_runtime": 19.7002,
467
+ "eval_samples_per_second": 50.761,
468
+ "eval_steps_per_second": 0.812,
469
+ "step": 900
470
+ },
471
+ {
472
+ "epoch": 0.7547169811320755,
473
+ "grad_norm": 0.1683952361345291,
474
+ "learning_rate": 0.000627262156763801,
475
+ "loss": 0.251,
476
+ "step": 920
477
+ },
478
+ {
479
+ "epoch": 0.771123872026251,
480
+ "grad_norm": 0.14314571022987366,
481
+ "learning_rate": 0.0006177435184145056,
482
+ "loss": 0.2347,
483
+ "step": 940
484
+ },
485
+ {
486
+ "epoch": 0.7793273174733388,
487
+ "eval_loss": 0.23846356570720673,
488
+ "eval_runtime": 19.7048,
489
+ "eval_samples_per_second": 50.749,
490
+ "eval_steps_per_second": 0.812,
491
+ "step": 950
492
+ },
493
+ {
494
+ "epoch": 0.7875307629204266,
495
+ "grad_norm": 0.17057938873767853,
496
+ "learning_rate": 0.000608046312718957,
497
+ "loss": 0.2432,
498
+ "step": 960
499
+ },
500
+ {
501
+ "epoch": 0.8039376538146021,
502
+ "grad_norm": 0.15681499242782593,
503
+ "learning_rate": 0.0005981784921723646,
504
+ "loss": 0.2498,
505
+ "step": 980
506
+ },
507
+ {
508
+ "epoch": 0.8203445447087777,
509
+ "grad_norm": 0.15658380091190338,
510
+ "learning_rate": 0.0005881481491879545,
511
+ "loss": 0.249,
512
+ "step": 1000
513
+ },
514
+ {
515
+ "epoch": 0.8203445447087777,
516
+ "eval_loss": 0.23703107237815857,
517
+ "eval_runtime": 19.6522,
518
+ "eval_samples_per_second": 50.885,
519
+ "eval_steps_per_second": 0.814,
520
+ "step": 1000
521
+ },
522
+ {
523
+ "epoch": 0.8367514356029533,
524
+ "grad_norm": 0.15780796110630035,
525
+ "learning_rate": 0.0005779635094605339,
526
+ "loss": 0.239,
527
+ "step": 1020
528
+ },
529
+ {
530
+ "epoch": 0.8531583264971287,
531
+ "grad_norm": 0.15824347734451294,
532
+ "learning_rate": 0.0005676329252207547,
533
+ "loss": 0.2402,
534
+ "step": 1040
535
+ },
536
+ {
537
+ "epoch": 0.8613617719442166,
538
+ "eval_loss": 0.23439089953899384,
539
+ "eval_runtime": 19.6931,
540
+ "eval_samples_per_second": 50.779,
541
+ "eval_steps_per_second": 0.812,
542
+ "step": 1050
543
+ },
544
+ {
545
+ "epoch": 0.8695652173913043,
546
+ "grad_norm": 0.14682568609714508,
547
+ "learning_rate": 0.0005571648683856057,
548
+ "loss": 0.2396,
549
+ "step": 1060
550
+ },
551
+ {
552
+ "epoch": 0.8859721082854799,
553
+ "grad_norm": 0.13325947523117065,
554
+ "learning_rate": 0.0005465679236107549,
555
+ "loss": 0.2501,
556
+ "step": 1080
557
+ },
558
+ {
559
+ "epoch": 0.9023789991796555,
560
+ "grad_norm": 0.19277364015579224,
561
+ "learning_rate": 0.0005358507812504347,
562
+ "loss": 0.2361,
563
+ "step": 1100
564
+ },
565
+ {
566
+ "epoch": 0.9023789991796555,
567
+ "eval_loss": 0.23314428329467773,
568
+ "eval_runtime": 19.6983,
569
+ "eval_samples_per_second": 50.766,
570
+ "eval_steps_per_second": 0.812,
571
+ "step": 1100
572
+ },
573
+ {
574
+ "epoch": 0.918785890073831,
575
+ "grad_norm": 0.14934952557086945,
576
+ "learning_rate": 0.0005250222302306469,
577
+ "loss": 0.2364,
578
+ "step": 1120
579
+ },
580
+ {
581
+ "epoch": 0.9351927809680065,
582
+ "grad_norm": 0.1576302945613861,
583
+ "learning_rate": 0.0005140911508415301,
584
+ "loss": 0.2429,
585
+ "step": 1140
586
+ },
587
+ {
588
+ "epoch": 0.9433962264150944,
589
+ "eval_loss": 0.23043398559093475,
590
+ "eval_runtime": 19.6276,
591
+ "eval_samples_per_second": 50.949,
592
+ "eval_steps_per_second": 0.815,
593
+ "step": 1150
594
+ },
595
+ {
596
+ "epoch": 0.9515996718621821,
597
+ "grad_norm": 0.1528480052947998,
598
+ "learning_rate": 0.0005030665074548022,
599
+ "loss": 0.2527,
600
+ "step": 1160
601
+ },
602
+ {
603
+ "epoch": 0.9680065627563577,
604
+ "grad_norm": 0.15908847749233246,
605
+ "learning_rate": 0.000491957341172248,
606
+ "loss": 0.2345,
607
+ "step": 1180
608
+ },
609
+ {
610
+ "epoch": 0.9844134536505332,
611
+ "grad_norm": 0.1384393572807312,
612
+ "learning_rate": 0.00048077276241128257,
613
+ "loss": 0.2438,
614
+ "step": 1200
615
+ },
616
+ {
617
+ "epoch": 0.9844134536505332,
618
+ "eval_loss": 0.22905392944812775,
619
+ "eval_runtime": 19.6492,
620
+ "eval_samples_per_second": 50.893,
621
+ "eval_steps_per_second": 0.814,
622
+ "step": 1200
623
+ },
624
+ {
625
+ "epoch": 1.0008203445447088,
626
+ "grad_norm": 0.15461111068725586,
627
+ "learning_rate": 0.00046952194343366965,
628
+ "loss": 0.248,
629
+ "step": 1220
630
+ },
631
+ {
632
+ "epoch": 1.0172272354388843,
633
+ "grad_norm": 0.13716405630111694,
634
+ "learning_rate": 0.0004582141108235238,
635
+ "loss": 0.2045,
636
+ "step": 1240
637
+ },
638
+ {
639
+ "epoch": 1.0254306808859721,
640
+ "eval_loss": 0.23012343049049377,
641
+ "eval_runtime": 19.698,
642
+ "eval_samples_per_second": 50.767,
643
+ "eval_steps_per_second": 0.812,
644
+ "step": 1250
645
+ },
646
+ {
647
+ "epoch": 1.03363412633306,
648
+ "grad_norm": 0.13788041472434998,
649
+ "learning_rate": 0.00044685853792076236,
650
+ "loss": 0.2108,
651
+ "step": 1260
652
+ },
653
+ {
654
+ "epoch": 1.0500410172272354,
655
+ "grad_norm": 0.1471407115459442,
656
+ "learning_rate": 0.0004354645372162148,
657
+ "loss": 0.2107,
658
+ "step": 1280
659
+ },
660
+ {
661
+ "epoch": 1.066447908121411,
662
+ "grad_norm": 0.1416047066450119,
663
+ "learning_rate": 0.0004240414527146252,
664
+ "loss": 0.2058,
665
+ "step": 1300
666
+ },
667
+ {
668
+ "epoch": 1.066447908121411,
669
+ "eval_loss": 0.22690238058567047,
670
+ "eval_runtime": 19.706,
671
+ "eval_samples_per_second": 50.746,
672
+ "eval_steps_per_second": 0.812,
673
+ "step": 1300
674
+ },
675
+ {
676
+ "epoch": 1.0828547990155866,
677
+ "grad_norm": 0.1709490567445755,
678
+ "learning_rate": 0.00041259865227181005,
679
+ "loss": 0.2029,
680
+ "step": 1320
681
+ },
682
+ {
683
+ "epoch": 1.099261689909762,
684
+ "grad_norm": 0.13066577911376953,
685
+ "learning_rate": 0.0004011455199122565,
686
+ "loss": 0.2118,
687
+ "step": 1340
688
+ },
689
+ {
690
+ "epoch": 1.1074651353568499,
691
+ "eval_loss": 0.22645984590053558,
692
+ "eval_runtime": 19.6494,
693
+ "eval_samples_per_second": 50.892,
694
+ "eval_steps_per_second": 0.814,
695
+ "step": 1350
696
+ },
697
+ {
698
+ "epoch": 1.1156685808039377,
699
+ "grad_norm": 0.1285369098186493,
700
+ "learning_rate": 0.00038969144813346204,
701
+ "loss": 0.2112,
702
+ "step": 1360
703
+ },
704
+ {
705
+ "epoch": 1.1320754716981132,
706
+ "grad_norm": 0.14131706953048706,
707
+ "learning_rate": 0.0003782458302033236,
708
+ "loss": 0.21,
709
+ "step": 1380
710
+ },
711
+ {
712
+ "epoch": 1.1484823625922886,
713
+ "grad_norm": 0.1357734501361847,
714
+ "learning_rate": 0.00036681805245689683,
715
+ "loss": 0.2151,
716
+ "step": 1400
717
+ },
718
+ {
719
+ "epoch": 1.1484823625922886,
720
+ "eval_loss": 0.2255597859621048,
721
+ "eval_runtime": 19.6666,
722
+ "eval_samples_per_second": 50.848,
723
+ "eval_steps_per_second": 0.814,
724
+ "step": 1400
725
+ },
726
+ {
727
+ "epoch": 1.1648892534864643,
728
+ "grad_norm": 0.13068270683288574,
729
+ "learning_rate": 0.0003554174865988396,
730
+ "loss": 0.2045,
731
+ "step": 1420
732
+ },
733
+ {
734
+ "epoch": 1.1812961443806398,
735
+ "grad_norm": 0.1526036411523819,
736
+ "learning_rate": 0.0003440534820178542,
737
+ "loss": 0.2004,
738
+ "step": 1440
739
+ },
740
+ {
741
+ "epoch": 1.1894995898277276,
742
+ "eval_loss": 0.22486895322799683,
743
+ "eval_runtime": 19.6995,
744
+ "eval_samples_per_second": 50.763,
745
+ "eval_steps_per_second": 0.812,
746
+ "step": 1450
747
+ },
748
+ {
749
+ "epoch": 1.1977030352748155,
750
+ "grad_norm": 0.13882817327976227,
751
+ "learning_rate": 0.0003327353581194304,
752
+ "loss": 0.2115,
753
+ "step": 1460
754
+ },
755
+ {
756
+ "epoch": 1.214109926168991,
757
+ "grad_norm": 0.13503047823905945,
758
+ "learning_rate": 0.00032147239668317643,
759
+ "loss": 0.213,
760
+ "step": 1480
761
+ },
762
+ {
763
+ "epoch": 1.2305168170631666,
764
+ "grad_norm": 0.14577829837799072,
765
+ "learning_rate": 0.00031027383425100776,
766
+ "loss": 0.21,
767
+ "step": 1500
768
+ },
769
+ {
770
+ "epoch": 1.2305168170631666,
771
+ "eval_loss": 0.22363610565662384,
772
+ "eval_runtime": 19.7003,
773
+ "eval_samples_per_second": 50.761,
774
+ "eval_steps_per_second": 0.812,
775
+ "step": 1500
776
+ },
777
+ {
778
+ "epoch": 1.246923707957342,
779
+ "grad_norm": 0.1494022160768509,
780
+ "learning_rate": 0.000299148854552433,
781
+ "loss": 0.2127,
782
+ "step": 1520
783
+ },
784
+ {
785
+ "epoch": 1.2633305988515175,
786
+ "grad_norm": 0.13340817391872406,
787
+ "learning_rate": 0.0002881065809731508,
788
+ "loss": 0.2045,
789
+ "step": 1540
790
+ },
791
+ {
792
+ "epoch": 1.2715340442986054,
793
+ "eval_loss": 0.22328989207744598,
794
+ "eval_runtime": 19.716,
795
+ "eval_samples_per_second": 50.72,
796
+ "eval_steps_per_second": 0.812,
797
+ "step": 1550
798
+ },
799
+ {
800
+ "epoch": 1.2797374897456932,
801
+ "grad_norm": 0.16567204892635345,
802
+ "learning_rate": 0.000277156069073134,
803
+ "loss": 0.2098,
804
+ "step": 1560
805
+ },
806
+ {
807
+ "epoch": 1.2961443806398687,
808
+ "grad_norm": 0.16263996064662933,
809
+ "learning_rate": 0.0002663062991603352,
810
+ "loss": 0.2046,
811
+ "step": 1580
812
+ },
813
+ {
814
+ "epoch": 1.3125512715340442,
815
+ "grad_norm": 0.13025769591331482,
816
+ "learning_rate": 0.00025556616892610665,
817
+ "loss": 0.2058,
818
+ "step": 1600
819
+ },
820
+ {
821
+ "epoch": 1.3125512715340442,
822
+ "eval_loss": 0.22085361182689667,
823
+ "eval_runtime": 19.7047,
824
+ "eval_samples_per_second": 50.749,
825
+ "eval_steps_per_second": 0.812,
826
+ "step": 1600
827
+ },
828
+ {
829
+ "epoch": 1.3289581624282198,
830
+ "grad_norm": 0.1469118446111679,
831
+ "learning_rate": 0.00024494448614837154,
832
+ "loss": 0.2022,
833
+ "step": 1620
834
+ },
835
+ {
836
+ "epoch": 1.3453650533223955,
837
+ "grad_norm": 0.14396725594997406,
838
+ "learning_rate": 0.00023444996146853062,
839
+ "loss": 0.2042,
840
+ "step": 1640
841
+ },
842
+ {
843
+ "epoch": 1.3535684987694832,
844
+ "eval_loss": 0.21919582784175873,
845
+ "eval_runtime": 19.6415,
846
+ "eval_samples_per_second": 50.913,
847
+ "eval_steps_per_second": 0.815,
848
+ "step": 1650
849
+ },
850
+ {
851
+ "epoch": 1.361771944216571,
852
+ "grad_norm": 0.14159901440143585,
853
+ "learning_rate": 0.00022409120124803136,
854
+ "loss": 0.1956,
855
+ "step": 1660
856
+ },
857
+ {
858
+ "epoch": 1.3781788351107465,
859
+ "grad_norm": 0.125336155295372,
860
+ "learning_rate": 0.0002138767005104529,
861
+ "loss": 0.1999,
862
+ "step": 1680
863
+ },
864
+ {
865
+ "epoch": 1.3945857260049221,
866
+ "grad_norm": 0.1390431672334671,
867
+ "learning_rate": 0.00020381483597489645,
868
+ "loss": 0.2038,
869
+ "step": 1700
870
+ },
871
+ {
872
+ "epoch": 1.3945857260049221,
873
+ "eval_loss": 0.21807876229286194,
874
+ "eval_runtime": 19.7306,
875
+ "eval_samples_per_second": 50.683,
876
+ "eval_steps_per_second": 0.811,
877
+ "step": 1700
878
+ },
879
+ {
880
+ "epoch": 1.4109926168990976,
881
+ "grad_norm": 0.17577524483203888,
882
+ "learning_rate": 0.00019391385918639844,
883
+ "loss": 0.2123,
884
+ "step": 1720
885
+ },
886
+ {
887
+ "epoch": 1.427399507793273,
888
+ "grad_norm": 0.13635504245758057,
889
+ "learning_rate": 0.00018418188974899237,
890
+ "loss": 0.2122,
891
+ "step": 1740
892
+ },
893
+ {
894
+ "epoch": 1.435602953240361,
895
+ "eval_loss": 0.21744303405284882,
896
+ "eval_runtime": 19.7173,
897
+ "eval_samples_per_second": 50.717,
898
+ "eval_steps_per_second": 0.811,
899
+ "step": 1750
900
+ },
901
+ {
902
+ "epoch": 1.4438063986874488,
903
+ "grad_norm": 0.15170520544052124,
904
+ "learning_rate": 0.0001746269086669758,
905
+ "loss": 0.2066,
906
+ "step": 1760
907
+ },
908
+ {
909
+ "epoch": 1.4602132895816242,
910
+ "grad_norm": 0.153891921043396,
911
+ "learning_rate": 0.00016525675179983998,
912
+ "loss": 0.2021,
913
+ "step": 1780
914
+ },
915
+ {
916
+ "epoch": 1.4766201804758,
917
+ "grad_norm": 0.13469810783863068,
918
+ "learning_rate": 0.00015607910343622768,
919
+ "loss": 0.2104,
920
+ "step": 1800
921
+ },
922
+ {
923
+ "epoch": 1.4766201804758,
924
+ "eval_loss": 0.21633844077587128,
925
+ "eval_runtime": 19.7043,
926
+ "eval_samples_per_second": 50.75,
927
+ "eval_steps_per_second": 0.812,
928
+ "step": 1800
929
+ },
930
+ {
931
+ "epoch": 1.4930270713699754,
932
+ "grad_norm": 0.14728526771068573,
933
+ "learning_rate": 0.00014710148999219565,
934
+ "loss": 0.1974,
935
+ "step": 1820
936
+ },
937
+ {
938
+ "epoch": 1.509433962264151,
939
+ "grad_norm": 0.15127798914909363,
940
+ "learning_rate": 0.0001383312738389411,
941
+ "loss": 0.207,
942
+ "step": 1840
943
+ },
944
+ {
945
+ "epoch": 1.5176374077112387,
946
+ "eval_loss": 0.2154580056667328,
947
+ "eval_runtime": 19.7094,
948
+ "eval_samples_per_second": 50.737,
949
+ "eval_steps_per_second": 0.812,
950
+ "step": 1850
951
+ },
952
+ {
953
+ "epoch": 1.5258408531583265,
954
+ "grad_norm": 0.1252005696296692,
955
+ "learning_rate": 0.00012977564726506085,
956
+ "loss": 0.2031,
957
+ "step": 1860
958
+ },
959
+ {
960
+ "epoch": 1.542247744052502,
961
+ "grad_norm": 0.151875838637352,
962
+ "learning_rate": 0.00012144162657829174,
963
+ "loss": 0.1978,
964
+ "step": 1880
965
+ },
966
+ {
967
+ "epoch": 1.5586546349466777,
968
+ "grad_norm": 0.13818328082561493,
969
+ "learning_rate": 0.0001133360463515683,
970
+ "loss": 0.1993,
971
+ "step": 1900
972
+ },
973
+ {
974
+ "epoch": 1.5586546349466777,
975
+ "eval_loss": 0.21548284590244293,
976
+ "eval_runtime": 19.7147,
977
+ "eval_samples_per_second": 50.724,
978
+ "eval_steps_per_second": 0.812,
979
+ "step": 1900
980
+ },
981
+ {
982
+ "epoch": 1.5750615258408531,
983
+ "grad_norm": 0.13859805464744568,
984
+ "learning_rate": 0.00010546555381811915,
985
+ "loss": 0.2032,
986
+ "step": 1920
987
+ },
988
+ {
989
+ "epoch": 1.5914684167350286,
990
+ "grad_norm": 0.1499071568250656,
991
+ "learning_rate": 9.78366034201966e-05,
992
+ "loss": 0.2042,
993
+ "step": 1940
994
+ },
995
+ {
996
+ "epoch": 1.5996718621821167,
997
+ "eval_loss": 0.21445508301258087,
998
+ "eval_runtime": 19.6269,
999
+ "eval_samples_per_second": 50.95,
1000
+ "eval_steps_per_second": 0.815,
1001
+ "step": 1950
1002
+ },
1003
+ {
1004
+ "epoch": 1.6078753076292043,
1005
+ "grad_norm": 0.15496855974197388,
1006
+ "learning_rate": 9.045545151591022e-05,
1007
+ "loss": 0.2044,
1008
+ "step": 1960
1009
+ },
1010
+ {
1011
+ "epoch": 1.62428219852338,
1012
+ "grad_norm": 0.13916639983654022,
1013
+ "learning_rate": 8.332815124850588e-05,
1014
+ "loss": 0.2009,
1015
+ "step": 1980
1016
+ },
1017
+ {
1018
+ "epoch": 1.6406890894175554,
1019
+ "grad_norm": 0.15067742764949799,
1020
+ "learning_rate": 7.646054758229762e-05,
1021
+ "loss": 0.2017,
1022
+ "step": 2000
1023
+ },
1024
+ {
1025
+ "epoch": 1.6406890894175554,
1026
+ "eval_loss": 0.21360059082508087,
1027
+ "eval_runtime": 19.7129,
1028
+ "eval_samples_per_second": 50.728,
1029
+ "eval_steps_per_second": 0.812,
1030
+ "step": 2000
1031
+ },
1032
+ {
1033
+ "epoch": 1.6570959803117309,
1034
+ "grad_norm": 0.16083775460720062,
1035
+ "learning_rate": 6.985827250932224e-05,
1036
+ "loss": 0.1928,
1037
+ "step": 2020
1038
+ },
1039
+ {
1040
+ "epoch": 1.6735028712059066,
1041
+ "grad_norm": 0.13639019429683685,
1042
+ "learning_rate": 6.352674043064934e-05,
1043
+ "loss": 0.1939,
1044
+ "step": 2040
1045
+ },
1046
+ {
1047
+ "epoch": 1.6817063166529942,
1048
+ "eval_loss": 0.21323883533477783,
1049
+ "eval_runtime": 19.6499,
1050
+ "eval_samples_per_second": 50.891,
1051
+ "eval_steps_per_second": 0.814,
1052
+ "step": 2050
1053
+ },
1054
+ {
1055
+ "epoch": 1.689909762100082,
1056
+ "grad_norm": 0.16499921679496765,
1057
+ "learning_rate": 5.747114371613358e-05,
1058
+ "loss": 0.2066,
1059
+ "step": 2060
1060
+ },
1061
+ {
1062
+ "epoch": 1.7063166529942575,
1063
+ "grad_norm": 0.14705513417720795,
1064
+ "learning_rate": 5.169644844624926e-05,
1065
+ "loss": 0.194,
1066
+ "step": 2080
1067
+ },
1068
+ {
1069
+ "epoch": 1.7227235438884332,
1070
+ "grad_norm": 0.1494346261024475,
1071
+ "learning_rate": 4.620739033950292e-05,
1072
+ "loss": 0.1948,
1073
+ "step": 2100
1074
+ },
1075
+ {
1076
+ "epoch": 1.7227235438884332,
1077
+ "eval_loss": 0.21250872313976288,
1078
+ "eval_runtime": 19.6415,
1079
+ "eval_samples_per_second": 50.913,
1080
+ "eval_steps_per_second": 0.815,
1081
+ "step": 2100
1082
+ },
1083
+ {
1084
+ "epoch": 1.7391304347826086,
1085
+ "grad_norm": 0.1426304578781128,
1086
+ "learning_rate": 4.100847086875885e-05,
1087
+ "loss": 0.1958,
1088
+ "step": 2120
1089
+ },
1090
+ {
1091
+ "epoch": 1.755537325676784,
1092
+ "grad_norm": 0.16745194792747498,
1093
+ "learning_rate": 3.610395356966687e-05,
1094
+ "loss": 0.1915,
1095
+ "step": 2140
1096
+ },
1097
+ {
1098
+ "epoch": 1.7637407711238722,
1099
+ "eval_loss": 0.21186348795890808,
1100
+ "eval_runtime": 19.6356,
1101
+ "eval_samples_per_second": 50.928,
1102
+ "eval_steps_per_second": 0.815,
1103
+ "step": 2150
1104
+ },
1105
+ {
1106
+ "epoch": 1.7719442165709598,
1107
+ "grad_norm": 0.1456213742494583,
1108
+ "learning_rate": 3.1497860544217326e-05,
1109
+ "loss": 0.1927,
1110
+ "step": 2160
1111
+ },
1112
+ {
1113
+ "epoch": 1.7883511074651355,
1114
+ "grad_norm": 0.13879838585853577,
1115
+ "learning_rate": 2.7193969162290445e-05,
1116
+ "loss": 0.195,
1117
+ "step": 2180
1118
+ },
1119
+ {
1120
+ "epoch": 1.804757998359311,
1121
+ "grad_norm": 0.1695457547903061,
1122
+ "learning_rate": 2.3195808963907805e-05,
1123
+ "loss": 0.2018,
1124
+ "step": 2200
1125
+ },
1126
+ {
1127
+ "epoch": 1.804757998359311,
1128
+ "eval_loss": 0.2118394523859024,
1129
+ "eval_runtime": 19.6812,
1130
+ "eval_samples_per_second": 50.81,
1131
+ "eval_steps_per_second": 0.813,
1132
+ "step": 2200
1133
+ },
1134
+ {
1135
+ "epoch": 1.8211648892534864,
1136
+ "grad_norm": 0.15308940410614014,
1137
+ "learning_rate": 1.9506658764722398e-05,
1138
+ "loss": 0.1896,
1139
+ "step": 2220
1140
+ },
1141
+ {
1142
+ "epoch": 1.837571780147662,
1143
+ "grad_norm": 0.14311152696609497,
1144
+ "learning_rate": 1.6129543967124116e-05,
1145
+ "loss": 0.1982,
1146
+ "step": 2240
1147
+ },
1148
+ {
1149
+ "epoch": 1.8457752255947497,
1150
+ "eval_loss": 0.21171796321868896,
1151
+ "eval_runtime": 19.6471,
1152
+ "eval_samples_per_second": 50.898,
1153
+ "eval_steps_per_second": 0.814,
1154
+ "step": 2250
1155
+ },
1156
+ {
1157
+ "epoch": 1.8539786710418376,
1158
+ "grad_norm": 0.13645325601100922,
1159
+ "learning_rate": 1.3067234079164303e-05,
1160
+ "loss": 0.1931,
1161
+ "step": 2260
1162
+ },
1163
+ {
1164
+ "epoch": 1.870385561936013,
1165
+ "grad_norm": 0.15800131857395172,
1166
+ "learning_rate": 1.0322240443334919e-05,
1167
+ "loss": 0.1971,
1168
+ "step": 2280
1169
+ },
1170
+ {
1171
+ "epoch": 1.8867924528301887,
1172
+ "grad_norm": 0.1702418327331543,
1173
+ "learning_rate": 7.896814177063361e-06,
1174
+ "loss": 0.2055,
1175
+ "step": 2300
1176
+ },
1177
+ {
1178
+ "epoch": 1.8867924528301887,
1179
+ "eval_loss": 0.2115548551082611,
1180
+ "eval_runtime": 19.7008,
1181
+ "eval_samples_per_second": 50.759,
1182
+ "eval_steps_per_second": 0.812,
1183
+ "step": 2300
1184
+ },
1185
+ {
1186
+ "epoch": 1.9031993437243644,
1187
+ "grad_norm": 0.13410726189613342,
1188
+ "learning_rate": 5.792944326614036e-06,
1189
+ "loss": 0.201,
1190
+ "step": 2320
1191
+ },
1192
+ {
1193
+ "epoch": 1.9196062346185396,
1194
+ "grad_norm": 0.1345076858997345,
1195
+ "learning_rate": 4.012356235908987e-06,
1196
+ "loss": 0.1942,
1197
+ "step": 2340
1198
+ },
1199
+ {
1200
+ "epoch": 1.9278096800656277,
1201
+ "eval_loss": 0.21144331991672516,
1202
+ "eval_runtime": 19.6313,
1203
+ "eval_samples_per_second": 50.939,
1204
+ "eval_steps_per_second": 0.815,
1205
+ "step": 2350
1206
+ },
1207
+ {
1208
+ "epoch": 1.9360131255127153,
1209
+ "grad_norm": 0.15890908241271973,
1210
+ "learning_rate": 2.5565101316058937e-06,
1211
+ "loss": 0.199,
1212
+ "step": 2360
1213
+ },
1214
+ {
1215
+ "epoch": 1.952420016406891,
1216
+ "grad_norm": 0.14523178339004517,
1217
+ "learning_rate": 1.4265999255938411e-06,
1218
+ "loss": 0.2078,
1219
+ "step": 2380
1220
+ },
1221
+ {
1222
+ "epoch": 1.9688269073010665,
1223
+ "grad_norm": 0.14125356078147888,
1224
+ "learning_rate": 6.235522358886137e-07,
1225
+ "loss": 0.2023,
1226
+ "step": 2400
1227
+ },
1228
+ {
1229
+ "epoch": 1.9688269073010665,
1230
+ "eval_loss": 0.21136069297790527,
1231
+ "eval_runtime": 19.6237,
1232
+ "eval_samples_per_second": 50.959,
1233
+ "eval_steps_per_second": 0.815,
1234
+ "step": 2400
1235
+ },
1236
+ {
1237
+ "epoch": 1.985233798195242,
1238
+ "grad_norm": 0.10682037472724915,
1239
+ "learning_rate": 1.480256267309077e-07,
1240
+ "loss": 0.189,
1241
+ "step": 2420
1242
+ },
1243
+ {
1244
+ "epoch": 2.0,
1245
+ "step": 2438,
1246
+ "total_flos": 1.58523627405312e+18,
1247
+ "train_loss": 0.24579448089724784,
1248
+ "train_runtime": 3558.3175,
1249
+ "train_samples_per_second": 21.92,
1250
+ "train_steps_per_second": 0.685
1251
+ }
1252
+ ],
1253
+ "logging_steps": 20,
1254
+ "max_steps": 2438,
1255
+ "num_input_tokens_seen": 0,
1256
+ "num_train_epochs": 2,
1257
+ "save_steps": 500,
1258
+ "stateful_callbacks": {
1259
+ "TrainerControl": {
1260
+ "args": {
1261
+ "should_epoch_stop": false,
1262
+ "should_evaluate": false,
1263
+ "should_log": false,
1264
+ "should_save": true,
1265
+ "should_training_stop": true
1266
+ },
1267
+ "attributes": {}
1268
+ }
1269
+ },
1270
+ "total_flos": 1.58523627405312e+18,
1271
+ "train_batch_size": 32,
1272
+ "trial_name": null,
1273
+ "trial_params": null
1274
+ }
nl_tasks/run_all/run_exps8/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_exps8/ft/added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[PAD]": 32000
3
+ }
nl_tasks/run_all/run_exps8/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_exps8/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exps8/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_exps8/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_exps8/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77b4d9e9c6b0d186a8d5a0224bec5f5924ec3bffeeb186960f12d9e88e0dd8ef
3
+ size 6481
nl_tasks/run_all/run_exps8/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_exps8/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3ba44f678af3b8042cadb58314183a9aed4a48d2222dc71fbfae70fa5136ed5
3
+ size 33602659
nl_tasks/run_all/run_exps8/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": 0.22884641587734222,
15
+ "learning_rate": 9.836065573770491e-05,
16
+ "loss": 0.6779,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04101722723543889,
21
+ "grad_norm": 0.19105906784534454,
22
+ "learning_rate": 0.00020081967213114754,
23
+ "loss": 0.4154,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.04101722723543889,
28
+ "eval_loss": 0.3721630871295929,
29
+ "eval_runtime": 20.0434,
30
+ "eval_samples_per_second": 49.892,
31
+ "eval_steps_per_second": 0.798,
32
+ "step": 50
33
+ },
34
+ {
35
+ "epoch": 0.06152584085315833,
36
+ "grad_norm": 0.1771579086780548,
37
+ "learning_rate": 0.0003032786885245902,
38
+ "loss": 0.3702,
39
+ "step": 75
40
+ },
41
+ {
42
+ "epoch": 0.08203445447087777,
43
+ "grad_norm": 0.18256710469722748,
44
+ "learning_rate": 0.0004057377049180328,
45
+ "loss": 0.3502,
46
+ "step": 100
47
+ },
48
+ {
49
+ "epoch": 0.08203445447087777,
50
+ "eval_loss": 0.3291940987110138,
51
+ "eval_runtime": 19.667,
52
+ "eval_samples_per_second": 50.847,
53
+ "eval_steps_per_second": 0.814,
54
+ "step": 100
55
+ },
56
+ {
57
+ "epoch": 0.10254306808859721,
58
+ "grad_norm": 0.15866395831108093,
59
+ "learning_rate": 0.0005081967213114754,
60
+ "loss": 0.3301,
61
+ "step": 125
62
+ },
63
+ {
64
+ "epoch": 0.12305168170631665,
65
+ "grad_norm": 0.19323067367076874,
66
+ "learning_rate": 0.000610655737704918,
67
+ "loss": 0.3149,
68
+ "step": 150
69
+ },
70
+ {
71
+ "epoch": 0.12305168170631665,
72
+ "eval_loss": 0.3124563694000244,
73
+ "eval_runtime": 19.6322,
74
+ "eval_samples_per_second": 50.937,
75
+ "eval_steps_per_second": 0.815,
76
+ "step": 150
77
+ },
78
+ {
79
+ "epoch": 0.1435602953240361,
80
+ "grad_norm": 0.208988755941391,
81
+ "learning_rate": 0.0007131147540983607,
82
+ "loss": 0.3128,
83
+ "step": 175
84
+ },
85
+ {
86
+ "epoch": 0.16406890894175555,
87
+ "grad_norm": 0.23417410254478455,
88
+ "learning_rate": 0.0008155737704918033,
89
+ "loss": 0.3197,
90
+ "step": 200
91
+ },
92
+ {
93
+ "epoch": 0.16406890894175555,
94
+ "eval_loss": 0.30872875452041626,
95
+ "eval_runtime": 19.6957,
96
+ "eval_samples_per_second": 50.772,
97
+ "eval_steps_per_second": 0.812,
98
+ "step": 200
99
+ },
100
+ {
101
+ "epoch": 0.184577522559475,
102
+ "grad_norm": 0.698834240436554,
103
+ "learning_rate": 0.0009180327868852459,
104
+ "loss": 0.3372,
105
+ "step": 225
106
+ },
107
+ {
108
+ "epoch": 0.20508613617719443,
109
+ "grad_norm": 0.2546020746231079,
110
+ "learning_rate": 0.0009999871854116063,
111
+ "loss": 0.3255,
112
+ "step": 250
113
+ },
114
+ {
115
+ "epoch": 0.20508613617719443,
116
+ "eval_loss": 0.3122219145298004,
117
+ "eval_runtime": 19.6534,
118
+ "eval_samples_per_second": 50.882,
119
+ "eval_steps_per_second": 0.814,
120
+ "step": 250
121
+ },
122
+ {
123
+ "epoch": 0.22559474979491387,
124
+ "grad_norm": 0.4460746943950653,
125
+ "learning_rate": 0.0009995387437838027,
126
+ "loss": 0.3219,
127
+ "step": 275
128
+ },
129
+ {
130
+ "epoch": 0.2461033634126333,
131
+ "grad_norm": 0.2903735935688019,
132
+ "learning_rate": 0.000998450229439693,
133
+ "loss": 0.3145,
134
+ "step": 300
135
+ },
136
+ {
137
+ "epoch": 0.2461033634126333,
138
+ "eval_loss": 0.3020172417163849,
139
+ "eval_runtime": 19.7043,
140
+ "eval_samples_per_second": 50.75,
141
+ "eval_steps_per_second": 0.812,
142
+ "step": 300
143
+ },
144
+ {
145
+ "epoch": 0.2666119770303528,
146
+ "grad_norm": 0.2259032428264618,
147
+ "learning_rate": 0.000996723037122612,
148
+ "loss": 0.3018,
149
+ "step": 325
150
+ },
151
+ {
152
+ "epoch": 0.2871205906480722,
153
+ "grad_norm": 0.24915221333503723,
154
+ "learning_rate": 0.0009943593799315263,
155
+ "loss": 0.3087,
156
+ "step": 350
157
+ },
158
+ {
159
+ "epoch": 0.2871205906480722,
160
+ "eval_loss": 0.28998228907585144,
161
+ "eval_runtime": 19.7184,
162
+ "eval_samples_per_second": 50.714,
163
+ "eval_steps_per_second": 0.811,
164
+ "step": 350
165
+ },
166
+ {
167
+ "epoch": 0.30762920426579166,
168
+ "grad_norm": 0.21242815256118774,
169
+ "learning_rate": 0.0009913622864853324,
170
+ "loss": 0.2923,
171
+ "step": 375
172
+ },
173
+ {
174
+ "epoch": 0.3281378178835111,
175
+ "grad_norm": 0.26507169008255005,
176
+ "learning_rate": 0.0009877355970422024,
177
+ "loss": 0.3016,
178
+ "step": 400
179
+ },
180
+ {
181
+ "epoch": 0.3281378178835111,
182
+ "eval_loss": 0.2806154191493988,
183
+ "eval_runtime": 19.6827,
184
+ "eval_samples_per_second": 50.806,
185
+ "eval_steps_per_second": 0.813,
186
+ "step": 400
187
+ },
188
+ {
189
+ "epoch": 0.34864643150123054,
190
+ "grad_norm": 0.1836182177066803,
191
+ "learning_rate": 0.0009834839585789559,
192
+ "loss": 0.2965,
193
+ "step": 425
194
+ },
195
+ {
196
+ "epoch": 0.36915504511895,
197
+ "grad_norm": 0.20331695675849915,
198
+ "learning_rate": 0.000978612818836762,
199
+ "loss": 0.2933,
200
+ "step": 450
201
+ },
202
+ {
203
+ "epoch": 0.36915504511895,
204
+ "eval_loss": 0.2746712565422058,
205
+ "eval_runtime": 19.6966,
206
+ "eval_samples_per_second": 50.77,
207
+ "eval_steps_per_second": 0.812,
208
+ "step": 450
209
+ },
210
+ {
211
+ "epoch": 0.3896636587366694,
212
+ "grad_norm": 0.18985247611999512,
213
+ "learning_rate": 0.0009731284193407981,
214
+ "loss": 0.2892,
215
+ "step": 475
216
+ },
217
+ {
218
+ "epoch": 0.41017227235438886,
219
+ "grad_norm": 0.19039498269557953,
220
+ "learning_rate": 0.0009670377874028117,
221
+ "loss": 0.2804,
222
+ "step": 500
223
+ },
224
+ {
225
+ "epoch": 0.41017227235438886,
226
+ "eval_loss": 0.2680283784866333,
227
+ "eval_runtime": 19.7006,
228
+ "eval_samples_per_second": 50.76,
229
+ "eval_steps_per_second": 0.812,
230
+ "step": 500
231
+ },
232
+ {
233
+ "epoch": 0.4306808859721083,
234
+ "grad_norm": 0.18804749846458435,
235
+ "learning_rate": 0.0009603487271168336,
236
+ "loss": 0.2829,
237
+ "step": 525
238
+ },
239
+ {
240
+ "epoch": 0.45118949958982774,
241
+ "grad_norm": 0.17971737682819366,
242
+ "learning_rate": 0.0009530698093595781,
243
+ "loss": 0.2769,
244
+ "step": 550
245
+ },
246
+ {
247
+ "epoch": 0.45118949958982774,
248
+ "eval_loss": 0.2613440454006195,
249
+ "eval_runtime": 19.7029,
250
+ "eval_samples_per_second": 50.754,
251
+ "eval_steps_per_second": 0.812,
252
+ "step": 550
253
+ },
254
+ {
255
+ "epoch": 0.4716981132075472,
256
+ "grad_norm": 0.1943299025297165,
257
+ "learning_rate": 0.0009452103608083418,
258
+ "loss": 0.2628,
259
+ "step": 575
260
+ },
261
+ {
262
+ "epoch": 0.4922067268252666,
263
+ "grad_norm": 0.18716727197170258,
264
+ "learning_rate": 0.0009367804519904775,
265
+ "loss": 0.2724,
266
+ "step": 600
267
+ },
268
+ {
269
+ "epoch": 0.4922067268252666,
270
+ "eval_loss": 0.25977185368537903,
271
+ "eval_runtime": 19.6942,
272
+ "eval_samples_per_second": 50.776,
273
+ "eval_steps_per_second": 0.812,
274
+ "step": 600
275
+ },
276
+ {
277
+ "epoch": 0.5127153404429861,
278
+ "grad_norm": 0.16847378015518188,
279
+ "learning_rate": 0.0009277908843797492,
280
+ "loss": 0.2596,
281
+ "step": 625
282
+ },
283
+ {
284
+ "epoch": 0.5332239540607056,
285
+ "grad_norm": 0.16402754187583923,
286
+ "learning_rate": 0.0009182531765561084,
287
+ "loss": 0.2591,
288
+ "step": 650
289
+ },
290
+ {
291
+ "epoch": 0.5332239540607056,
292
+ "eval_loss": 0.2526455223560333,
293
+ "eval_runtime": 19.7198,
294
+ "eval_samples_per_second": 50.71,
295
+ "eval_steps_per_second": 0.811,
296
+ "step": 650
297
+ },
298
+ {
299
+ "epoch": 0.5537325676784249,
300
+ "grad_norm": 0.16279703378677368,
301
+ "learning_rate": 0.0009081795494466201,
302
+ "loss": 0.2587,
303
+ "step": 675
304
+ },
305
+ {
306
+ "epoch": 0.5742411812961444,
307
+ "grad_norm": 0.14156077802181244,
308
+ "learning_rate": 0.0008975829106664539,
309
+ "loss": 0.2517,
310
+ "step": 700
311
+ },
312
+ {
313
+ "epoch": 0.5742411812961444,
314
+ "eval_loss": 0.25181302428245544,
315
+ "eval_runtime": 19.6967,
316
+ "eval_samples_per_second": 50.77,
317
+ "eval_steps_per_second": 0.812,
318
+ "step": 700
319
+ },
320
+ {
321
+ "epoch": 0.5947497949138638,
322
+ "grad_norm": 0.14295664429664612,
323
+ "learning_rate": 0.0008864768379800017,
324
+ "loss": 0.2486,
325
+ "step": 725
326
+ },
327
+ {
328
+ "epoch": 0.6152584085315833,
329
+ "grad_norm": 0.13009321689605713,
330
+ "learning_rate": 0.0008748755619033153,
331
+ "loss": 0.2564,
332
+ "step": 750
333
+ },
334
+ {
335
+ "epoch": 0.6152584085315833,
336
+ "eval_loss": 0.24707932770252228,
337
+ "eval_runtime": 19.691,
338
+ "eval_samples_per_second": 50.785,
339
+ "eval_steps_per_second": 0.813,
340
+ "step": 750
341
+ },
342
+ {
343
+ "epoch": 0.6357670221493027,
344
+ "grad_norm": 0.13928845524787903,
345
+ "learning_rate": 0.000862793947470155,
346
+ "loss": 0.263,
347
+ "step": 775
348
+ },
349
+ {
350
+ "epoch": 0.6562756357670222,
351
+ "grad_norm": 0.16057232022285461,
352
+ "learning_rate": 0.0008502474751850142,
353
+ "loss": 0.2514,
354
+ "step": 800
355
+ },
356
+ {
357
+ "epoch": 0.6562756357670222,
358
+ "eval_loss": 0.24361753463745117,
359
+ "eval_runtime": 19.6845,
360
+ "eval_samples_per_second": 50.801,
361
+ "eval_steps_per_second": 0.813,
362
+ "step": 800
363
+ },
364
+ {
365
+ "epoch": 0.6767842493847416,
366
+ "grad_norm": 0.13310407102108002,
367
+ "learning_rate": 0.0008372522211875224,
368
+ "loss": 0.2467,
369
+ "step": 825
370
+ },
371
+ {
372
+ "epoch": 0.6972928630024611,
373
+ "grad_norm": 0.12262047827243805,
374
+ "learning_rate": 0.0008238248366536473,
375
+ "loss": 0.2597,
376
+ "step": 850
377
+ },
378
+ {
379
+ "epoch": 0.6972928630024611,
380
+ "eval_loss": 0.24112360179424286,
381
+ "eval_runtime": 19.6766,
382
+ "eval_samples_per_second": 50.822,
383
+ "eval_steps_per_second": 0.813,
384
+ "step": 850
385
+ },
386
+ {
387
+ "epoch": 0.7178014766201805,
388
+ "grad_norm": 0.11482135206460953,
389
+ "learning_rate": 0.0008099825264600842,
390
+ "loss": 0.2535,
391
+ "step": 875
392
+ },
393
+ {
394
+ "epoch": 0.7383100902379,
395
+ "grad_norm": 0.13044659793376923,
396
+ "learning_rate": 0.0007957430271391761,
397
+ "loss": 0.2461,
398
+ "step": 900
399
+ },
400
+ {
401
+ "epoch": 0.7383100902379,
402
+ "eval_loss": 0.23893289268016815,
403
+ "eval_runtime": 19.6972,
404
+ "eval_samples_per_second": 50.769,
405
+ "eval_steps_per_second": 0.812,
406
+ "step": 900
407
+ },
408
+ {
409
+ "epoch": 0.7588187038556193,
410
+ "grad_norm": 0.13767851889133453,
411
+ "learning_rate": 0.0007811245841526062,
412
+ "loss": 0.2496,
413
+ "step": 925
414
+ },
415
+ {
416
+ "epoch": 0.7793273174733388,
417
+ "grad_norm": 0.13775552809238434,
418
+ "learning_rate": 0.0007661459285129879,
419
+ "loss": 0.2329,
420
+ "step": 950
421
+ },
422
+ {
423
+ "epoch": 0.7793273174733388,
424
+ "eval_loss": 0.23598331212997437,
425
+ "eval_runtime": 19.6899,
426
+ "eval_samples_per_second": 50.787,
427
+ "eval_steps_per_second": 0.813,
428
+ "step": 950
429
+ },
430
+ {
431
+ "epoch": 0.7998359310910582,
432
+ "grad_norm": 0.13417741656303406,
433
+ "learning_rate": 0.0007508262527833029,
434
+ "loss": 0.2457,
435
+ "step": 975
436
+ },
437
+ {
438
+ "epoch": 0.8203445447087777,
439
+ "grad_norm": 0.12178505212068558,
440
+ "learning_rate": 0.000735185186484943,
441
+ "loss": 0.2484,
442
+ "step": 1000
443
+ },
444
+ {
445
+ "epoch": 0.8203445447087777,
446
+ "eval_loss": 0.23527154326438904,
447
+ "eval_runtime": 19.7017,
448
+ "eval_samples_per_second": 50.757,
449
+ "eval_steps_per_second": 0.812,
450
+ "step": 1000
451
+ },
452
+ {
453
+ "epoch": 0.8408531583264971,
454
+ "grad_norm": 0.12044777721166611,
455
+ "learning_rate": 0.0007192427709458656,
456
+ "loss": 0.2352,
457
+ "step": 1025
458
+ },
459
+ {
460
+ "epoch": 0.8613617719442166,
461
+ "grad_norm": 0.1242508813738823,
462
+ "learning_rate": 0.0007030194336210887,
463
+ "loss": 0.2399,
464
+ "step": 1050
465
+ },
466
+ {
467
+ "epoch": 0.8613617719442166,
468
+ "eval_loss": 0.23323005437850952,
469
+ "eval_runtime": 19.7003,
470
+ "eval_samples_per_second": 50.761,
471
+ "eval_steps_per_second": 0.812,
472
+ "step": 1050
473
+ },
474
+ {
475
+ "epoch": 0.881870385561936,
476
+ "grad_norm": 0.12805870175361633,
477
+ "learning_rate": 0.0006865359619184331,
478
+ "loss": 0.2462,
479
+ "step": 1075
480
+ },
481
+ {
482
+ "epoch": 0.9023789991796555,
483
+ "grad_norm": 0.1338253915309906,
484
+ "learning_rate": 0.0006698134765630434,
485
+ "loss": 0.2376,
486
+ "step": 1100
487
+ },
488
+ {
489
+ "epoch": 0.9023789991796555,
490
+ "eval_loss": 0.23185238242149353,
491
+ "eval_runtime": 19.7149,
492
+ "eval_samples_per_second": 50.723,
493
+ "eval_steps_per_second": 0.812,
494
+ "step": 1100
495
+ },
496
+ {
497
+ "epoch": 0.9228876127973749,
498
+ "grad_norm": 0.0990019366145134,
499
+ "learning_rate": 0.0006528734045348248,
500
+ "loss": 0.2355,
501
+ "step": 1125
502
+ },
503
+ {
504
+ "epoch": 0.9433962264150944,
505
+ "grad_norm": 0.11817856878042221,
506
+ "learning_rate": 0.0006357374516134643,
507
+ "loss": 0.2492,
508
+ "step": 1150
509
+ },
510
+ {
511
+ "epoch": 0.9433962264150944,
512
+ "eval_loss": 0.2299334704875946,
513
+ "eval_runtime": 19.6922,
514
+ "eval_samples_per_second": 50.782,
515
+ "eval_steps_per_second": 0.813,
516
+ "step": 1150
517
+ },
518
+ {
519
+ "epoch": 0.9639048400328137,
520
+ "grad_norm": 0.12407154589891434,
521
+ "learning_rate": 0.0006184275745662179,
522
+ "loss": 0.238,
523
+ "step": 1175
524
+ },
525
+ {
526
+ "epoch": 0.9844134536505332,
527
+ "grad_norm": 0.11156418174505234,
528
+ "learning_rate": 0.0006009659530141031,
529
+ "loss": 0.2409,
530
+ "step": 1200
531
+ },
532
+ {
533
+ "epoch": 0.9844134536505332,
534
+ "eval_loss": 0.22886136174201965,
535
+ "eval_runtime": 19.7269,
536
+ "eval_samples_per_second": 50.692,
537
+ "eval_steps_per_second": 0.811,
538
+ "step": 1200
539
+ },
540
+ {
541
+ "epoch": 1.0049220672682526,
542
+ "grad_norm": 0.10731246322393417,
543
+ "learning_rate": 0.0005833749610125402,
544
+ "loss": 0.2383,
545
+ "step": 1225
546
+ },
547
+ {
548
+ "epoch": 1.0254306808859721,
549
+ "grad_norm": 0.11062945425510406,
550
+ "learning_rate": 0.0005656771383828602,
551
+ "loss": 0.2115,
552
+ "step": 1250
553
+ },
554
+ {
555
+ "epoch": 1.0254306808859721,
556
+ "eval_loss": 0.22990986704826355,
557
+ "eval_runtime": 19.6891,
558
+ "eval_samples_per_second": 50.789,
559
+ "eval_steps_per_second": 0.813,
560
+ "step": 1250
561
+ },
562
+ {
563
+ "epoch": 1.0459392945036916,
564
+ "grad_norm": 0.10919920355081558,
565
+ "learning_rate": 0.0005478951618314134,
566
+ "loss": 0.2069,
567
+ "step": 1275
568
+ },
569
+ {
570
+ "epoch": 1.066447908121411,
571
+ "grad_norm": 0.11572052538394928,
572
+ "learning_rate": 0.0005300518158932815,
573
+ "loss": 0.2081,
574
+ "step": 1300
575
+ },
576
+ {
577
+ "epoch": 1.066447908121411,
578
+ "eval_loss": 0.22684714198112488,
579
+ "eval_runtime": 19.7072,
580
+ "eval_samples_per_second": 50.743,
581
+ "eval_steps_per_second": 0.812,
582
+ "step": 1300
583
+ },
584
+ {
585
+ "epoch": 1.0869565217391304,
586
+ "grad_norm": 0.12236768752336502,
587
+ "learning_rate": 0.0005121699637378282,
588
+ "loss": 0.2061,
589
+ "step": 1325
590
+ },
591
+ {
592
+ "epoch": 1.1074651353568499,
593
+ "grad_norm": 0.12757174670696259,
594
+ "learning_rate": 0.0004942725178734903,
595
+ "loss": 0.2116,
596
+ "step": 1350
597
+ },
598
+ {
599
+ "epoch": 1.1074651353568499,
600
+ "eval_loss": 0.22620388865470886,
601
+ "eval_runtime": 19.71,
602
+ "eval_samples_per_second": 50.736,
603
+ "eval_steps_per_second": 0.812,
604
+ "step": 1350
605
+ },
606
+ {
607
+ "epoch": 1.1279737489745694,
608
+ "grad_norm": 0.11296207457780838,
609
+ "learning_rate": 0.00047638241078935324,
610
+ "loss": 0.2093,
611
+ "step": 1375
612
+ },
613
+ {
614
+ "epoch": 1.1484823625922886,
615
+ "grad_norm": 0.12079405784606934,
616
+ "learning_rate": 0.000458522565571121,
617
+ "loss": 0.2172,
618
+ "step": 1400
619
+ },
620
+ {
621
+ "epoch": 1.1484823625922886,
622
+ "eval_loss": 0.22575503587722778,
623
+ "eval_runtime": 19.6961,
624
+ "eval_samples_per_second": 50.772,
625
+ "eval_steps_per_second": 0.812,
626
+ "step": 1400
627
+ },
628
+ {
629
+ "epoch": 1.1689909762100081,
630
+ "grad_norm": 0.125356525182724,
631
+ "learning_rate": 0.00044071586652913767,
632
+ "loss": 0.2013,
633
+ "step": 1425
634
+ },
635
+ {
636
+ "epoch": 1.1894995898277276,
637
+ "grad_norm": 0.1261182576417923,
638
+ "learning_rate": 0.0004229851298760915,
639
+ "loss": 0.2104,
640
+ "step": 1450
641
+ },
642
+ {
643
+ "epoch": 1.1894995898277276,
644
+ "eval_loss": 0.22514581680297852,
645
+ "eval_runtime": 19.7052,
646
+ "eval_samples_per_second": 50.748,
647
+ "eval_steps_per_second": 0.812,
648
+ "step": 1450
649
+ },
650
+ {
651
+ "epoch": 1.2100082034454471,
652
+ "grad_norm": 0.10678336769342422,
653
+ "learning_rate": 0.0004053530744919749,
654
+ "loss": 0.2103,
655
+ "step": 1475
656
+ },
657
+ {
658
+ "epoch": 1.2305168170631666,
659
+ "grad_norm": 0.11951852589845657,
660
+ "learning_rate": 0.0003878422928137597,
661
+ "loss": 0.2129,
662
+ "step": 1500
663
+ },
664
+ {
665
+ "epoch": 1.2305168170631666,
666
+ "eval_loss": 0.223064586520195,
667
+ "eval_runtime": 19.6964,
668
+ "eval_samples_per_second": 50.771,
669
+ "eval_steps_per_second": 0.812,
670
+ "step": 1500
671
+ },
672
+ {
673
+ "epoch": 1.251025430680886,
674
+ "grad_norm": 0.11787497997283936,
675
+ "learning_rate": 0.0003704752218870861,
676
+ "loss": 0.2176,
677
+ "step": 1525
678
+ },
679
+ {
680
+ "epoch": 1.2715340442986054,
681
+ "grad_norm": 0.11035430431365967,
682
+ "learning_rate": 0.00035327411461706025,
683
+ "loss": 0.2002,
684
+ "step": 1550
685
+ },
686
+ {
687
+ "epoch": 1.2715340442986054,
688
+ "eval_loss": 0.2231324166059494,
689
+ "eval_runtime": 19.6957,
690
+ "eval_samples_per_second": 50.772,
691
+ "eval_steps_per_second": 0.812,
692
+ "step": 1550
693
+ },
694
+ {
695
+ "epoch": 1.2920426579163249,
696
+ "grad_norm": 0.13375981152057648,
697
+ "learning_rate": 0.00033626101125499555,
698
+ "loss": 0.2101,
699
+ "step": 1575
700
+ },
701
+ {
702
+ "epoch": 1.3125512715340442,
703
+ "grad_norm": 0.1072443425655365,
704
+ "learning_rate": 0.0003194577111576333,
705
+ "loss": 0.2051,
706
+ "step": 1600
707
+ },
708
+ {
709
+ "epoch": 1.3125512715340442,
710
+ "eval_loss": 0.22061283886432648,
711
+ "eval_runtime": 19.7132,
712
+ "eval_samples_per_second": 50.727,
713
+ "eval_steps_per_second": 0.812,
714
+ "step": 1600
715
+ },
716
+ {
717
+ "epoch": 1.3330598851517639,
718
+ "grad_norm": 0.10793954133987427,
719
+ "learning_rate": 0.00030288574485502756,
720
+ "loss": 0.2045,
721
+ "step": 1625
722
+ },
723
+ {
724
+ "epoch": 1.3535684987694832,
725
+ "grad_norm": 0.11586247384548187,
726
+ "learning_rate": 0.00028656634646288565,
727
+ "loss": 0.2057,
728
+ "step": 1650
729
+ },
730
+ {
731
+ "epoch": 1.3535684987694832,
732
+ "eval_loss": 0.21922053396701813,
733
+ "eval_runtime": 19.7044,
734
+ "eval_samples_per_second": 50.75,
735
+ "eval_steps_per_second": 0.812,
736
+ "step": 1650
737
+ },
738
+ {
739
+ "epoch": 1.3740771123872026,
740
+ "grad_norm": 0.12886735796928406,
741
+ "learning_rate": 0.00027052042647471254,
742
+ "loss": 0.195,
743
+ "step": 1675
744
+ },
745
+ {
746
+ "epoch": 1.3945857260049221,
747
+ "grad_norm": 0.11101047694683075,
748
+ "learning_rate": 0.0002547685449686206,
749
+ "loss": 0.2031,
750
+ "step": 1700
751
+ },
752
+ {
753
+ "epoch": 1.3945857260049221,
754
+ "eval_loss": 0.217972069978714,
755
+ "eval_runtime": 19.6967,
756
+ "eval_samples_per_second": 50.77,
757
+ "eval_steps_per_second": 0.812,
758
+ "step": 1700
759
+ },
760
+ {
761
+ "epoch": 1.4150943396226414,
762
+ "grad_norm": 0.12856581807136536,
763
+ "learning_rate": 0.0002393308852631373,
764
+ "loss": 0.2153,
765
+ "step": 1725
766
+ },
767
+ {
768
+ "epoch": 1.435602953240361,
769
+ "grad_norm": 0.1262427121400833,
770
+ "learning_rate": 0.0002242272280557645,
771
+ "loss": 0.2078,
772
+ "step": 1750
773
+ },
774
+ {
775
+ "epoch": 1.435602953240361,
776
+ "eval_loss": 0.2174151986837387,
777
+ "eval_runtime": 19.7075,
778
+ "eval_samples_per_second": 50.742,
779
+ "eval_steps_per_second": 0.812,
780
+ "step": 1750
781
+ },
782
+ {
783
+ "epoch": 1.4561115668580804,
784
+ "grad_norm": 0.12367235124111176,
785
+ "learning_rate": 0.0002094769260774262,
786
+ "loss": 0.2049,
787
+ "step": 1775
788
+ },
789
+ {
790
+ "epoch": 1.4766201804758,
791
+ "grad_norm": 0.12090994417667389,
792
+ "learning_rate": 0.00019509887929528458,
793
+ "loss": 0.2085,
794
+ "step": 1800
795
+ },
796
+ {
797
+ "epoch": 1.4766201804758,
798
+ "eval_loss": 0.21628864109516144,
799
+ "eval_runtime": 19.7,
800
+ "eval_samples_per_second": 50.761,
801
+ "eval_steps_per_second": 0.812,
802
+ "step": 1800
803
+ },
804
+ {
805
+ "epoch": 1.4971287940935194,
806
+ "grad_norm": 0.13603141903877258,
807
+ "learning_rate": 0.0001811115106956918,
808
+ "loss": 0.1996,
809
+ "step": 1825
810
+ },
811
+ {
812
+ "epoch": 1.5176374077112387,
813
+ "grad_norm": 0.13560324907302856,
814
+ "learning_rate": 0.00016753274267831115,
815
+ "loss": 0.2087,
816
+ "step": 1850
817
+ },
818
+ {
819
+ "epoch": 1.5176374077112387,
820
+ "eval_loss": 0.2152128368616104,
821
+ "eval_runtime": 19.8033,
822
+ "eval_samples_per_second": 50.497,
823
+ "eval_steps_per_second": 0.808,
824
+ "step": 1850
825
+ },
826
+ {
827
+ "epoch": 1.5381460213289582,
828
+ "grad_norm": 0.13079869747161865,
829
+ "learning_rate": 0.00015437997409165478,
830
+ "loss": 0.1994,
831
+ "step": 1875
832
+ },
833
+ {
834
+ "epoch": 1.5586546349466777,
835
+ "grad_norm": 0.10692238062620163,
836
+ "learning_rate": 0.00014167005793946035,
837
+ "loss": 0.1986,
838
+ "step": 1900
839
+ },
840
+ {
841
+ "epoch": 1.5586546349466777,
842
+ "eval_loss": 0.2152107208967209,
843
+ "eval_runtime": 19.6909,
844
+ "eval_samples_per_second": 50.785,
845
+ "eval_steps_per_second": 0.813,
846
+ "step": 1900
847
+ },
848
+ {
849
+ "epoch": 1.579163248564397,
850
+ "grad_norm": 0.12547941505908966,
851
+ "learning_rate": 0.00012941927978647527,
852
+ "loss": 0.2008,
853
+ "step": 1925
854
+ },
855
+ {
856
+ "epoch": 1.5996718621821167,
857
+ "grad_norm": 0.11310374736785889,
858
+ "learning_rate": 0.00011764333689131385,
859
+ "loss": 0.206,
860
+ "step": 1950
861
+ },
862
+ {
863
+ "epoch": 1.5996718621821167,
864
+ "eval_loss": 0.2140214443206787,
865
+ "eval_runtime": 19.6937,
866
+ "eval_samples_per_second": 50.778,
867
+ "eval_steps_per_second": 0.812,
868
+ "step": 1950
869
+ },
870
+ {
871
+ "epoch": 1.620180475799836,
872
+ "grad_norm": 0.103459432721138,
873
+ "learning_rate": 0.00010635731809312993,
874
+ "loss": 0.2024,
875
+ "step": 1975
876
+ },
877
+ {
878
+ "epoch": 1.6406890894175554,
879
+ "grad_norm": 0.13050952553749084,
880
+ "learning_rate": 9.557568447787201e-05,
881
+ "loss": 0.2022,
882
+ "step": 2000
883
+ },
884
+ {
885
+ "epoch": 1.6406890894175554,
886
+ "eval_loss": 0.21337273716926575,
887
+ "eval_runtime": 19.7037,
888
+ "eval_samples_per_second": 50.752,
889
+ "eval_steps_per_second": 0.812,
890
+ "step": 2000
891
+ },
892
+ {
893
+ "epoch": 1.661197703035275,
894
+ "grad_norm": 0.11469099670648575,
895
+ "learning_rate": 8.531225084889654e-05,
896
+ "loss": 0.1957,
897
+ "step": 2025
898
+ },
899
+ {
900
+ "epoch": 1.6817063166529942,
901
+ "grad_norm": 0.13165529072284698,
902
+ "learning_rate": 7.558016802568091e-05,
903
+ "loss": 0.1989,
904
+ "step": 2050
905
+ },
906
+ {
907
+ "epoch": 1.6817063166529942,
908
+ "eval_loss": 0.21270236372947693,
909
+ "eval_runtime": 19.7052,
910
+ "eval_samples_per_second": 50.748,
911
+ "eval_steps_per_second": 0.812,
912
+ "step": 2050
913
+ },
914
+ {
915
+ "epoch": 1.7022149302707137,
916
+ "grad_norm": 0.12024752050638199,
917
+ "learning_rate": 6.639190599331746e-05,
918
+ "loss": 0.2,
919
+ "step": 2075
920
+ },
921
+ {
922
+ "epoch": 1.7227235438884332,
923
+ "grad_norm": 0.12203649431467056,
924
+ "learning_rate": 5.775923792437865e-05,
925
+ "loss": 0.1945,
926
+ "step": 2100
927
+ },
928
+ {
929
+ "epoch": 1.7227235438884332,
930
+ "eval_loss": 0.2121145874261856,
931
+ "eval_runtime": 19.7232,
932
+ "eval_samples_per_second": 50.702,
933
+ "eval_steps_per_second": 0.811,
934
+ "step": 2100
935
+ },
936
+ {
937
+ "epoch": 1.7432321575061525,
938
+ "grad_norm": 0.12066127359867096,
939
+ "learning_rate": 4.9693225093627616e-05,
940
+ "loss": 0.1967,
941
+ "step": 2125
942
+ },
943
+ {
944
+ "epoch": 1.7637407711238722,
945
+ "grad_norm": 0.11252416670322418,
946
+ "learning_rate": 4.220420270490294e-05,
947
+ "loss": 0.1907,
948
+ "step": 2150
949
+ },
950
+ {
951
+ "epoch": 1.7637407711238722,
952
+ "eval_loss": 0.21157300472259521,
953
+ "eval_runtime": 19.6977,
954
+ "eval_samples_per_second": 50.767,
955
+ "eval_steps_per_second": 0.812,
956
+ "step": 2150
957
+ },
958
+ {
959
+ "epoch": 1.7842493847415914,
960
+ "grad_norm": 0.12583795189857483,
961
+ "learning_rate": 3.530176664833834e-05,
962
+ "loss": 0.1926,
963
+ "step": 2175
964
+ },
965
+ {
966
+ "epoch": 1.804757998359311,
967
+ "grad_norm": 0.141811802983284,
968
+ "learning_rate": 2.8994761204884756e-05,
969
+ "loss": 0.201,
970
+ "step": 2200
971
+ },
972
+ {
973
+ "epoch": 1.804757998359311,
974
+ "eval_loss": 0.21168458461761475,
975
+ "eval_runtime": 19.7058,
976
+ "eval_samples_per_second": 50.747,
977
+ "eval_steps_per_second": 0.812,
978
+ "step": 2200
979
+ },
980
+ {
981
+ "epoch": 1.8252666119770304,
982
+ "grad_norm": 0.13142260909080505,
983
+ "learning_rate": 2.329126771388995e-05,
984
+ "loss": 0.193,
985
+ "step": 2225
986
+ },
987
+ {
988
+ "epoch": 1.8457752255947497,
989
+ "grad_norm": 0.11814265698194504,
990
+ "learning_rate": 1.8198594218256815e-05,
991
+ "loss": 0.1939,
992
+ "step": 2250
993
+ },
994
+ {
995
+ "epoch": 1.8457752255947497,
996
+ "eval_loss": 0.21143142879009247,
997
+ "eval_runtime": 19.7079,
998
+ "eval_samples_per_second": 50.741,
999
+ "eval_steps_per_second": 0.812,
1000
+ "step": 2250
1001
+ },
1002
+ {
1003
+ "epoch": 1.8662838392124692,
1004
+ "grad_norm": 0.13810421526432037,
1005
+ "learning_rate": 1.3723266100447052e-05,
1006
+ "loss": 0.1968,
1007
+ "step": 2275
1008
+ },
1009
+ {
1010
+ "epoch": 1.8867924528301887,
1011
+ "grad_norm": 0.14677244424819946,
1012
+ "learning_rate": 9.871017721329201e-06,
1013
+ "loss": 0.2047,
1014
+ "step": 2300
1015
+ },
1016
+ {
1017
+ "epoch": 1.8867924528301887,
1018
+ "eval_loss": 0.21132488548755646,
1019
+ "eval_runtime": 19.6919,
1020
+ "eval_samples_per_second": 50.782,
1021
+ "eval_steps_per_second": 0.813,
1022
+ "step": 2300
1023
+ },
1024
+ {
1025
+ "epoch": 1.907301066447908,
1026
+ "grad_norm": 0.12535341084003448,
1027
+ "learning_rate": 6.646785072584871e-06,
1028
+ "loss": 0.1998,
1029
+ "step": 2325
1030
+ },
1031
+ {
1032
+ "epoch": 1.9278096800656277,
1033
+ "grad_norm": 0.10967744886875153,
1034
+ "learning_rate": 4.054699452086641e-06,
1035
+ "loss": 0.1933,
1036
+ "step": 2350
1037
+ },
1038
+ {
1039
+ "epoch": 1.9278096800656277,
1040
+ "eval_loss": 0.21114005148410797,
1041
+ "eval_runtime": 19.7092,
1042
+ "eval_samples_per_second": 50.738,
1043
+ "eval_steps_per_second": 0.812,
1044
+ "step": 2350
1045
+ },
1046
+ {
1047
+ "epoch": 1.948318293683347,
1048
+ "grad_norm": 0.1192808449268341,
1049
+ "learning_rate": 2.0980821703527886e-06,
1050
+ "loss": 0.2076,
1051
+ "step": 2375
1052
+ },
1053
+ {
1054
+ "epoch": 1.9688269073010665,
1055
+ "grad_norm": 0.12619267404079437,
1056
+ "learning_rate": 7.794402948607671e-07,
1057
+ "loss": 0.2035,
1058
+ "step": 2400
1059
+ },
1060
+ {
1061
+ "epoch": 1.9688269073010665,
1062
+ "eval_loss": 0.21114200353622437,
1063
+ "eval_runtime": 19.688,
1064
+ "eval_samples_per_second": 50.792,
1065
+ "eval_steps_per_second": 0.813,
1066
+ "step": 2400
1067
+ },
1068
+ {
1069
+ "epoch": 1.989335520918786,
1070
+ "grad_norm": 0.14471229910850525,
1071
+ "learning_rate": 1.0046343767294853e-07,
1072
+ "loss": 0.1935,
1073
+ "step": 2425
1074
+ },
1075
+ {
1076
+ "epoch": 2.0,
1077
+ "step": 2438,
1078
+ "total_flos": 1.58523627405312e+18,
1079
+ "train_loss": 0.2451091083568858,
1080
+ "train_runtime": 3430.409,
1081
+ "train_samples_per_second": 22.738,
1082
+ "train_steps_per_second": 0.711
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_exps9/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_exps9/ft/added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[PAD]": 32000
3
+ }
nl_tasks/run_all/run_exps9/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_exps9/ft/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
nl_tasks/run_all/run_exps9/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_exps9/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_exps9/ft/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f2613017d59911903351dd46b970d91458301015e2e28ac6dbbd227e384dc670
3
+ size 6481
nl_tasks/run_all/run_exps9/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_exps9/ft2/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e39790a64172548c39e5817ee2ccf09e14eee1cac7d3e43c14d1ae3a675cec83
3
+ size 33602659
nl_tasks/run_all/run_exps9/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": 0.15631958842277527,
15
+ "learning_rate": 0.00019672131147540983,
16
+ "loss": 0.625,
17
+ "step": 25
18
+ },
19
+ {
20
+ "epoch": 0.04101722723543889,
21
+ "grad_norm": 0.19207292795181274,
22
+ "learning_rate": 0.00040163934426229507,
23
+ "loss": 0.3841,
24
+ "step": 50
25
+ },
26
+ {
27
+ "epoch": 0.04101722723543889,
28
+ "eval_loss": 0.35151487588882446,
29
+ "eval_runtime": 20.0018,
30
+ "eval_samples_per_second": 49.995,
31
+ "eval_steps_per_second": 0.8,
32
+ "step": 50
33
+ },
34
+ {
35
+ "epoch": 0.06152584085315833,
36
+ "grad_norm": 0.16752676665782928,
37
+ "learning_rate": 0.0006065573770491804,
38
+ "loss": 0.3566,
39
+ "step": 75
40
+ },
41
+ {
42
+ "epoch": 0.08203445447087777,
43
+ "grad_norm": 0.20614324510097504,
44
+ "learning_rate": 0.0008114754098360656,
45
+ "loss": 0.3423,
46
+ "step": 100
47
+ },
48
+ {
49
+ "epoch": 0.08203445447087777,
50
+ "eval_loss": 0.32457083463668823,
51
+ "eval_runtime": 19.7076,
52
+ "eval_samples_per_second": 50.742,
53
+ "eval_steps_per_second": 0.812,
54
+ "step": 100
55
+ },
56
+ {
57
+ "epoch": 0.10254306808859721,
58
+ "grad_norm": 0.24570414423942566,
59
+ "learning_rate": 0.0010163934426229509,
60
+ "loss": 0.3296,
61
+ "step": 125
62
+ },
63
+ {
64
+ "epoch": 0.12305168170631665,
65
+ "grad_norm": 0.8766458034515381,
66
+ "learning_rate": 0.001221311475409836,
67
+ "loss": 0.32,
68
+ "step": 150
69
+ },
70
+ {
71
+ "epoch": 0.12305168170631665,
72
+ "eval_loss": 0.32632166147232056,
73
+ "eval_runtime": 19.713,
74
+ "eval_samples_per_second": 50.728,
75
+ "eval_steps_per_second": 0.812,
76
+ "step": 150
77
+ },
78
+ {
79
+ "epoch": 0.1435602953240361,
80
+ "grad_norm": 1.4807082414627075,
81
+ "learning_rate": 0.0014262295081967215,
82
+ "loss": 0.4602,
83
+ "step": 175
84
+ },
85
+ {
86
+ "epoch": 0.16406890894175555,
87
+ "grad_norm": 0.9411895275115967,
88
+ "learning_rate": 0.0016311475409836065,
89
+ "loss": 0.8811,
90
+ "step": 200
91
+ },
92
+ {
93
+ "epoch": 0.16406890894175555,
94
+ "eval_loss": 0.36227503418922424,
95
+ "eval_runtime": 19.7002,
96
+ "eval_samples_per_second": 50.761,
97
+ "eval_steps_per_second": 0.812,
98
+ "step": 200
99
+ },
100
+ {
101
+ "epoch": 0.184577522559475,
102
+ "grad_norm": 1.6836446523666382,
103
+ "learning_rate": 0.0018360655737704918,
104
+ "loss": 0.362,
105
+ "step": 225
106
+ },
107
+ {
108
+ "epoch": 0.20508613617719443,
109
+ "grad_norm": 0.32095012068748474,
110
+ "learning_rate": 0.0019999743708232127,
111
+ "loss": 0.3454,
112
+ "step": 250
113
+ },
114
+ {
115
+ "epoch": 0.20508613617719443,
116
+ "eval_loss": 0.3233221471309662,
117
+ "eval_runtime": 19.6954,
118
+ "eval_samples_per_second": 50.773,
119
+ "eval_steps_per_second": 0.812,
120
+ "step": 250
121
+ },
122
+ {
123
+ "epoch": 0.22559474979491387,
124
+ "grad_norm": 0.17643769085407257,
125
+ "learning_rate": 0.0019990774875676054,
126
+ "loss": 0.3281,
127
+ "step": 275
128
+ },
129
+ {
130
+ "epoch": 0.2461033634126333,
131
+ "grad_norm": 0.1627015918493271,
132
+ "learning_rate": 0.001996900458879386,
133
+ "loss": 0.3125,
134
+ "step": 300
135
+ },
136
+ {
137
+ "epoch": 0.2461033634126333,
138
+ "eval_loss": 0.2965441048145294,
139
+ "eval_runtime": 19.7102,
140
+ "eval_samples_per_second": 50.735,
141
+ "eval_steps_per_second": 0.812,
142
+ "step": 300
143
+ },
144
+ {
145
+ "epoch": 0.2666119770303528,
146
+ "grad_norm": 0.12793457508087158,
147
+ "learning_rate": 0.001993446074245224,
148
+ "loss": 0.2908,
149
+ "step": 325
150
+ },
151
+ {
152
+ "epoch": 0.2871205906480722,
153
+ "grad_norm": 0.11288447678089142,
154
+ "learning_rate": 0.0019887187598630527,
155
+ "loss": 0.2943,
156
+ "step": 350
157
+ },
158
+ {
159
+ "epoch": 0.2871205906480722,
160
+ "eval_loss": 0.27764618396759033,
161
+ "eval_runtime": 19.6774,
162
+ "eval_samples_per_second": 50.82,
163
+ "eval_steps_per_second": 0.813,
164
+ "step": 350
165
+ },
166
+ {
167
+ "epoch": 0.30762920426579166,
168
+ "grad_norm": 0.10496044158935547,
169
+ "learning_rate": 0.0019827245729706648,
170
+ "loss": 0.2786,
171
+ "step": 375
172
+ },
173
+ {
174
+ "epoch": 0.3281378178835111,
175
+ "grad_norm": 0.1280740350484848,
176
+ "learning_rate": 0.0019754711940844047,
177
+ "loss": 0.2883,
178
+ "step": 400
179
+ },
180
+ {
181
+ "epoch": 0.3281378178835111,
182
+ "eval_loss": 0.26895689964294434,
183
+ "eval_runtime": 19.6935,
184
+ "eval_samples_per_second": 50.778,
185
+ "eval_steps_per_second": 0.812,
186
+ "step": 400
187
+ },
188
+ {
189
+ "epoch": 0.34864643150123054,
190
+ "grad_norm": 0.10460188239812851,
191
+ "learning_rate": 0.0019669679171579117,
192
+ "loss": 0.2824,
193
+ "step": 425
194
+ },
195
+ {
196
+ "epoch": 0.36915504511895,
197
+ "grad_norm": 0.09117243438959122,
198
+ "learning_rate": 0.001957225637673524,
199
+ "loss": 0.2783,
200
+ "step": 450
201
+ },
202
+ {
203
+ "epoch": 0.36915504511895,
204
+ "eval_loss": 0.26259535551071167,
205
+ "eval_runtime": 19.6642,
206
+ "eval_samples_per_second": 50.854,
207
+ "eval_steps_per_second": 0.814,
208
+ "step": 450
209
+ },
210
+ {
211
+ "epoch": 0.3896636587366694,
212
+ "grad_norm": 0.0958079993724823,
213
+ "learning_rate": 0.0019462568386815961,
214
+ "loss": 0.277,
215
+ "step": 475
216
+ },
217
+ {
218
+ "epoch": 0.41017227235438886,
219
+ "grad_norm": 0.07896929234266281,
220
+ "learning_rate": 0.0019340755748056234,
221
+ "loss": 0.2691,
222
+ "step": 500
223
+ },
224
+ {
225
+ "epoch": 0.41017227235438886,
226
+ "eval_loss": 0.25900787115097046,
227
+ "eval_runtime": 19.6671,
228
+ "eval_samples_per_second": 50.846,
229
+ "eval_steps_per_second": 0.814,
230
+ "step": 500
231
+ },
232
+ {
233
+ "epoch": 0.4306808859721083,
234
+ "grad_norm": 0.09119442105293274,
235
+ "learning_rate": 0.0019206974542336672,
236
+ "loss": 0.273,
237
+ "step": 525
238
+ },
239
+ {
240
+ "epoch": 0.45118949958982774,
241
+ "grad_norm": 0.08431200683116913,
242
+ "learning_rate": 0.0019061396187191563,
243
+ "loss": 0.2677,
244
+ "step": 550
245
+ },
246
+ {
247
+ "epoch": 0.45118949958982774,
248
+ "eval_loss": 0.25446659326553345,
249
+ "eval_runtime": 19.6469,
250
+ "eval_samples_per_second": 50.899,
251
+ "eval_steps_per_second": 0.814,
252
+ "step": 550
253
+ },
254
+ {
255
+ "epoch": 0.4716981132075472,
256
+ "grad_norm": 0.09090767055749893,
257
+ "learning_rate": 0.0018904207216166836,
258
+ "loss": 0.2561,
259
+ "step": 575
260
+ },
261
+ {
262
+ "epoch": 0.4922067268252666,
263
+ "grad_norm": 0.08792543411254883,
264
+ "learning_rate": 0.001873560903980955,
265
+ "loss": 0.2658,
266
+ "step": 600
267
+ },
268
+ {
269
+ "epoch": 0.4922067268252666,
270
+ "eval_loss": 0.2533581554889679,
271
+ "eval_runtime": 19.6536,
272
+ "eval_samples_per_second": 50.881,
273
+ "eval_steps_per_second": 0.814,
274
+ "step": 600
275
+ },
276
+ {
277
+ "epoch": 0.5127153404429861,
278
+ "grad_norm": 0.08082229644060135,
279
+ "learning_rate": 0.0018555817687594984,
280
+ "loss": 0.2556,
281
+ "step": 625
282
+ },
283
+ {
284
+ "epoch": 0.5332239540607056,
285
+ "grad_norm": 0.07814713567495346,
286
+ "learning_rate": 0.0018365063531122169,
287
+ "loss": 0.255,
288
+ "step": 650
289
+ },
290
+ {
291
+ "epoch": 0.5332239540607056,
292
+ "eval_loss": 0.2496640980243683,
293
+ "eval_runtime": 19.6301,
294
+ "eval_samples_per_second": 50.942,
295
+ "eval_steps_per_second": 0.815,
296
+ "step": 650
297
+ },
298
+ {
299
+ "epoch": 0.5537325676784249,
300
+ "grad_norm": 0.08665795624256134,
301
+ "learning_rate": 0.0018163590988932402,
302
+ "loss": 0.2549,
303
+ "step": 675
304
+ },
305
+ {
306
+ "epoch": 0.5742411812961444,
307
+ "grad_norm": 0.0706089660525322,
308
+ "learning_rate": 0.0017951658213329078,
309
+ "loss": 0.2492,
310
+ "step": 700
311
+ },
312
+ {
313
+ "epoch": 0.5742411812961444,
314
+ "eval_loss": 0.2482205182313919,
315
+ "eval_runtime": 19.6333,
316
+ "eval_samples_per_second": 50.934,
317
+ "eval_steps_per_second": 0.815,
318
+ "step": 700
319
+ },
320
+ {
321
+ "epoch": 0.5947497949138638,
322
+ "grad_norm": 0.06884733587503433,
323
+ "learning_rate": 0.0017729536759600033,
324
+ "loss": 0.2468,
325
+ "step": 725
326
+ },
327
+ {
328
+ "epoch": 0.6152584085315833,
329
+ "grad_norm": 0.07153979688882828,
330
+ "learning_rate": 0.0017497511238066307,
331
+ "loss": 0.2556,
332
+ "step": 750
333
+ },
334
+ {
335
+ "epoch": 0.6152584085315833,
336
+ "eval_loss": 0.24759989976882935,
337
+ "eval_runtime": 19.6852,
338
+ "eval_samples_per_second": 50.8,
339
+ "eval_steps_per_second": 0.813,
340
+ "step": 750
341
+ },
342
+ {
343
+ "epoch": 0.6357670221493027,
344
+ "grad_norm": 0.07271415740251541,
345
+ "learning_rate": 0.00172558789494031,
346
+ "loss": 0.2609,
347
+ "step": 775
348
+ },
349
+ {
350
+ "epoch": 0.6562756357670222,
351
+ "grad_norm": 0.07679114490747452,
352
+ "learning_rate": 0.0017004949503700284,
353
+ "loss": 0.2513,
354
+ "step": 800
355
+ },
356
+ {
357
+ "epoch": 0.6562756357670222,
358
+ "eval_loss": 0.24522894620895386,
359
+ "eval_runtime": 19.7,
360
+ "eval_samples_per_second": 50.761,
361
+ "eval_steps_per_second": 0.812,
362
+ "step": 800
363
+ },
364
+ {
365
+ "epoch": 0.6767842493847416,
366
+ "grad_norm": 0.06669772416353226,
367
+ "learning_rate": 0.0016745044423750449,
368
+ "loss": 0.2476,
369
+ "step": 825
370
+ },
371
+ {
372
+ "epoch": 0.6972928630024611,
373
+ "grad_norm": 0.06657619029283524,
374
+ "learning_rate": 0.0016476496733072946,
375
+ "loss": 0.2602,
376
+ "step": 850
377
+ },
378
+ {
379
+ "epoch": 0.6972928630024611,
380
+ "eval_loss": 0.24357080459594727,
381
+ "eval_runtime": 19.6991,
382
+ "eval_samples_per_second": 50.764,
383
+ "eval_steps_per_second": 0.812,
384
+ "step": 850
385
+ },
386
+ {
387
+ "epoch": 0.7178014766201805,
388
+ "grad_norm": 0.056979499757289886,
389
+ "learning_rate": 0.0016199650529201684,
390
+ "loss": 0.2552,
391
+ "step": 875
392
+ },
393
+ {
394
+ "epoch": 0.7383100902379,
395
+ "grad_norm": 0.0696592777967453,
396
+ "learning_rate": 0.0015914860542783522,
397
+ "loss": 0.2482,
398
+ "step": 900
399
+ },
400
+ {
401
+ "epoch": 0.7383100902379,
402
+ "eval_loss": 0.24042876064777374,
403
+ "eval_runtime": 19.6937,
404
+ "eval_samples_per_second": 50.778,
405
+ "eval_steps_per_second": 0.812,
406
+ "step": 900
407
+ },
408
+ {
409
+ "epoch": 0.7588187038556193,
410
+ "grad_norm": 0.07218654453754425,
411
+ "learning_rate": 0.0015622491683052124,
412
+ "loss": 0.2512,
413
+ "step": 925
414
+ },
415
+ {
416
+ "epoch": 0.7793273174733388,
417
+ "grad_norm": 0.07043641805648804,
418
+ "learning_rate": 0.0015322918570259759,
419
+ "loss": 0.2355,
420
+ "step": 950
421
+ },
422
+ {
423
+ "epoch": 0.7793273174733388,
424
+ "eval_loss": 0.23907436430454254,
425
+ "eval_runtime": 19.7009,
426
+ "eval_samples_per_second": 50.759,
427
+ "eval_steps_per_second": 0.812,
428
+ "step": 950
429
+ },
430
+ {
431
+ "epoch": 0.7998359310910582,
432
+ "grad_norm": 0.06954478472471237,
433
+ "learning_rate": 0.0015016525055666057,
434
+ "loss": 0.2473,
435
+ "step": 975
436
+ },
437
+ {
438
+ "epoch": 0.8203445447087777,
439
+ "grad_norm": 0.07011434435844421,
440
+ "learning_rate": 0.001470370372969886,
441
+ "loss": 0.2512,
442
+ "step": 1000
443
+ },
444
+ {
445
+ "epoch": 0.8203445447087777,
446
+ "eval_loss": 0.2379504293203354,
447
+ "eval_runtime": 19.6277,
448
+ "eval_samples_per_second": 50.948,
449
+ "eval_steps_per_second": 0.815,
450
+ "step": 1000
451
+ },
452
+ {
453
+ "epoch": 0.8408531583264971,
454
+ "grad_norm": 0.05751856788992882,
455
+ "learning_rate": 0.0014384855418917311,
456
+ "loss": 0.2385,
457
+ "step": 1025
458
+ },
459
+ {
460
+ "epoch": 0.8613617719442166,
461
+ "grad_norm": 0.06693562120199203,
462
+ "learning_rate": 0.0014060388672421775,
463
+ "loss": 0.2439,
464
+ "step": 1050
465
+ },
466
+ {
467
+ "epoch": 0.8613617719442166,
468
+ "eval_loss": 0.23606353998184204,
469
+ "eval_runtime": 19.6821,
470
+ "eval_samples_per_second": 50.808,
471
+ "eval_steps_per_second": 0.813,
472
+ "step": 1050
473
+ },
474
+ {
475
+ "epoch": 0.881870385561936,
476
+ "grad_norm": 0.07531697303056717,
477
+ "learning_rate": 0.0013730719238368662,
478
+ "loss": 0.2483,
479
+ "step": 1075
480
+ },
481
+ {
482
+ "epoch": 0.9023789991796555,
483
+ "grad_norm": 0.06496143341064453,
484
+ "learning_rate": 0.0013396269531260867,
485
+ "loss": 0.2408,
486
+ "step": 1100
487
+ },
488
+ {
489
+ "epoch": 0.9023789991796555,
490
+ "eval_loss": 0.23498988151550293,
491
+ "eval_runtime": 19.7172,
492
+ "eval_samples_per_second": 50.717,
493
+ "eval_steps_per_second": 0.811,
494
+ "step": 1100
495
+ },
496
+ {
497
+ "epoch": 0.9228876127973749,
498
+ "grad_norm": 0.056089796125888824,
499
+ "learning_rate": 0.0013057468090696496,
500
+ "loss": 0.2386,
501
+ "step": 1125
502
+ },
503
+ {
504
+ "epoch": 0.9433962264150944,
505
+ "grad_norm": 0.06863794475793839,
506
+ "learning_rate": 0.0012714749032269287,
507
+ "loss": 0.2524,
508
+ "step": 1150
509
+ },
510
+ {
511
+ "epoch": 0.9433962264150944,
512
+ "eval_loss": 0.23351113498210907,
513
+ "eval_runtime": 19.7267,
514
+ "eval_samples_per_second": 50.693,
515
+ "eval_steps_per_second": 0.811,
516
+ "step": 1150
517
+ },
518
+ {
519
+ "epoch": 0.9639048400328137,
520
+ "grad_norm": 0.07869358360767365,
521
+ "learning_rate": 0.0012368551491324358,
522
+ "loss": 0.2418,
523
+ "step": 1175
524
+ },
525
+ {
526
+ "epoch": 0.9844134536505332,
527
+ "grad_norm": 0.06855684518814087,
528
+ "learning_rate": 0.0012019319060282063,
529
+ "loss": 0.245,
530
+ "step": 1200
531
+ },
532
+ {
533
+ "epoch": 0.9844134536505332,
534
+ "eval_loss": 0.23161761462688446,
535
+ "eval_runtime": 19.7326,
536
+ "eval_samples_per_second": 50.677,
537
+ "eval_steps_per_second": 0.811,
538
+ "step": 1200
539
+ },
540
+ {
541
+ "epoch": 1.0049220672682526,
542
+ "grad_norm": 0.0644753947854042,
543
+ "learning_rate": 0.0011667499220250803,
544
+ "loss": 0.2415,
545
+ "step": 1225
546
+ },
547
+ {
548
+ "epoch": 1.0254306808859721,
549
+ "grad_norm": 0.06986167281866074,
550
+ "learning_rate": 0.0011313542767657204,
551
+ "loss": 0.2172,
552
+ "step": 1250
553
+ },
554
+ {
555
+ "epoch": 1.0254306808859721,
556
+ "eval_loss": 0.2324475198984146,
557
+ "eval_runtime": 19.7141,
558
+ "eval_samples_per_second": 50.725,
559
+ "eval_steps_per_second": 0.812,
560
+ "step": 1250
561
+ },
562
+ {
563
+ "epoch": 1.0459392945036916,
564
+ "grad_norm": 0.0747043788433075,
565
+ "learning_rate": 0.0010957903236628267,
566
+ "loss": 0.2115,
567
+ "step": 1275
568
+ },
569
+ {
570
+ "epoch": 1.066447908121411,
571
+ "grad_norm": 0.06977721303701401,
572
+ "learning_rate": 0.001060103631786563,
573
+ "loss": 0.2137,
574
+ "step": 1300
575
+ },
576
+ {
577
+ "epoch": 1.066447908121411,
578
+ "eval_loss": 0.22955697774887085,
579
+ "eval_runtime": 19.7393,
580
+ "eval_samples_per_second": 50.66,
581
+ "eval_steps_per_second": 0.811,
582
+ "step": 1300
583
+ },
584
+ {
585
+ "epoch": 1.0869565217391304,
586
+ "grad_norm": 0.06966695934534073,
587
+ "learning_rate": 0.0010243399274756564,
588
+ "loss": 0.2115,
589
+ "step": 1325
590
+ },
591
+ {
592
+ "epoch": 1.1074651353568499,
593
+ "grad_norm": 0.07589821517467499,
594
+ "learning_rate": 0.0009885450357469806,
595
+ "loss": 0.2169,
596
+ "step": 1350
597
+ },
598
+ {
599
+ "epoch": 1.1074651353568499,
600
+ "eval_loss": 0.2284967601299286,
601
+ "eval_runtime": 19.7087,
602
+ "eval_samples_per_second": 50.739,
603
+ "eval_steps_per_second": 0.812,
604
+ "step": 1350
605
+ },
606
+ {
607
+ "epoch": 1.1279737489745694,
608
+ "grad_norm": 0.07126174867153168,
609
+ "learning_rate": 0.0009527648215787065,
610
+ "loss": 0.2147,
611
+ "step": 1375
612
+ },
613
+ {
614
+ "epoch": 1.1484823625922886,
615
+ "grad_norm": 0.07547113299369812,
616
+ "learning_rate": 0.000917045131142242,
617
+ "loss": 0.2214,
618
+ "step": 1400
619
+ },
620
+ {
621
+ "epoch": 1.1484823625922886,
622
+ "eval_loss": 0.22825054824352264,
623
+ "eval_runtime": 19.7241,
624
+ "eval_samples_per_second": 50.7,
625
+ "eval_steps_per_second": 0.811,
626
+ "step": 1400
627
+ },
628
+ {
629
+ "epoch": 1.1689909762100081,
630
+ "grad_norm": 0.07522042095661163,
631
+ "learning_rate": 0.0008814317330582753,
632
+ "loss": 0.2058,
633
+ "step": 1425
634
+ },
635
+ {
636
+ "epoch": 1.1894995898277276,
637
+ "grad_norm": 0.07902327924966812,
638
+ "learning_rate": 0.000845970259752183,
639
+ "loss": 0.2149,
640
+ "step": 1450
641
+ },
642
+ {
643
+ "epoch": 1.1894995898277276,
644
+ "eval_loss": 0.22894418239593506,
645
+ "eval_runtime": 19.7141,
646
+ "eval_samples_per_second": 50.725,
647
+ "eval_steps_per_second": 0.812,
648
+ "step": 1450
649
+ },
650
+ {
651
+ "epoch": 1.2100082034454471,
652
+ "grad_norm": 0.07127253711223602,
653
+ "learning_rate": 0.0008107061489839498,
654
+ "loss": 0.215,
655
+ "step": 1475
656
+ },
657
+ {
658
+ "epoch": 1.2305168170631666,
659
+ "grad_norm": 0.07822400331497192,
660
+ "learning_rate": 0.0007756845856275194,
661
+ "loss": 0.2191,
662
+ "step": 1500
663
+ },
664
+ {
665
+ "epoch": 1.2305168170631666,
666
+ "eval_loss": 0.22562581300735474,
667
+ "eval_runtime": 19.6991,
668
+ "eval_samples_per_second": 50.764,
669
+ "eval_steps_per_second": 0.812,
670
+ "step": 1500
671
+ },
672
+ {
673
+ "epoch": 1.251025430680886,
674
+ "grad_norm": 0.069583959877491,
675
+ "learning_rate": 0.0007409504437741722,
676
+ "loss": 0.2227,
677
+ "step": 1525
678
+ },
679
+ {
680
+ "epoch": 1.2715340442986054,
681
+ "grad_norm": 0.06680460274219513,
682
+ "learning_rate": 0.0007065482292341205,
683
+ "loss": 0.2049,
684
+ "step": 1550
685
+ },
686
+ {
687
+ "epoch": 1.2715340442986054,
688
+ "eval_loss": 0.22518320381641388,
689
+ "eval_runtime": 19.7091,
690
+ "eval_samples_per_second": 50.738,
691
+ "eval_steps_per_second": 0.812,
692
+ "step": 1550
693
+ },
694
+ {
695
+ "epoch": 1.2920426579163249,
696
+ "grad_norm": 0.08421827852725983,
697
+ "learning_rate": 0.0006725220225099911,
698
+ "loss": 0.2134,
699
+ "step": 1575
700
+ },
701
+ {
702
+ "epoch": 1.3125512715340442,
703
+ "grad_norm": 0.06956930458545685,
704
+ "learning_rate": 0.0006389154223152666,
705
+ "loss": 0.2101,
706
+ "step": 1600
707
+ },
708
+ {
709
+ "epoch": 1.3125512715340442,
710
+ "eval_loss": 0.22292453050613403,
711
+ "eval_runtime": 19.7161,
712
+ "eval_samples_per_second": 50.72,
713
+ "eval_steps_per_second": 0.812,
714
+ "step": 1600
715
+ },
716
+ {
717
+ "epoch": 1.3330598851517639,
718
+ "grad_norm": 0.06830507516860962,
719
+ "learning_rate": 0.0006057714897100551,
720
+ "loss": 0.2091,
721
+ "step": 1625
722
+ },
723
+ {
724
+ "epoch": 1.3535684987694832,
725
+ "grad_norm": 0.07062454521656036,
726
+ "learning_rate": 0.0005731326929257713,
727
+ "loss": 0.2098,
728
+ "step": 1650
729
+ },
730
+ {
731
+ "epoch": 1.3535684987694832,
732
+ "eval_loss": 0.2210480272769928,
733
+ "eval_runtime": 19.7257,
734
+ "eval_samples_per_second": 50.695,
735
+ "eval_steps_per_second": 0.811,
736
+ "step": 1650
737
+ },
738
+ {
739
+ "epoch": 1.3740771123872026,
740
+ "grad_norm": 0.08497516810894012,
741
+ "learning_rate": 0.0005410408529494251,
742
+ "loss": 0.1985,
743
+ "step": 1675
744
+ },
745
+ {
746
+ "epoch": 1.3945857260049221,
747
+ "grad_norm": 0.07322894781827927,
748
+ "learning_rate": 0.0005095370899372412,
749
+ "loss": 0.2063,
750
+ "step": 1700
751
+ },
752
+ {
753
+ "epoch": 1.3945857260049221,
754
+ "eval_loss": 0.22037026286125183,
755
+ "eval_runtime": 19.7099,
756
+ "eval_samples_per_second": 50.736,
757
+ "eval_steps_per_second": 0.812,
758
+ "step": 1700
759
+ },
760
+ {
761
+ "epoch": 1.4150943396226414,
762
+ "grad_norm": 0.08352910727262497,
763
+ "learning_rate": 0.0004786617705262746,
764
+ "loss": 0.2184,
765
+ "step": 1725
766
+ },
767
+ {
768
+ "epoch": 1.435602953240361,
769
+ "grad_norm": 0.0793982669711113,
770
+ "learning_rate": 0.000448454456111529,
771
+ "loss": 0.2115,
772
+ "step": 1750
773
+ },
774
+ {
775
+ "epoch": 1.435602953240361,
776
+ "eval_loss": 0.2191518247127533,
777
+ "eval_runtime": 19.6786,
778
+ "eval_samples_per_second": 50.817,
779
+ "eval_steps_per_second": 0.813,
780
+ "step": 1750
781
+ },
782
+ {
783
+ "epoch": 1.4561115668580804,
784
+ "grad_norm": 0.07494239509105682,
785
+ "learning_rate": 0.0004189538521548524,
786
+ "loss": 0.2069,
787
+ "step": 1775
788
+ },
789
+ {
790
+ "epoch": 1.4766201804758,
791
+ "grad_norm": 0.07422234117984772,
792
+ "learning_rate": 0.00039019775859056916,
793
+ "loss": 0.2118,
794
+ "step": 1800
795
+ },
796
+ {
797
+ "epoch": 1.4766201804758,
798
+ "eval_loss": 0.21833916008472443,
799
+ "eval_runtime": 19.6899,
800
+ "eval_samples_per_second": 50.787,
801
+ "eval_steps_per_second": 0.813,
802
+ "step": 1800
803
+ },
804
+ {
805
+ "epoch": 1.4971287940935194,
806
+ "grad_norm": 0.08856727927923203,
807
+ "learning_rate": 0.0003622230213913836,
808
+ "loss": 0.2024,
809
+ "step": 1825
810
+ },
811
+ {
812
+ "epoch": 1.5176374077112387,
813
+ "grad_norm": 0.07883690297603607,
814
+ "learning_rate": 0.0003350654853566223,
815
+ "loss": 0.2118,
816
+ "step": 1850
817
+ },
818
+ {
819
+ "epoch": 1.5176374077112387,
820
+ "eval_loss": 0.2166438102722168,
821
+ "eval_runtime": 19.6884,
822
+ "eval_samples_per_second": 50.791,
823
+ "eval_steps_per_second": 0.813,
824
+ "step": 1850
825
+ },
826
+ {
827
+ "epoch": 1.5381460213289582,
828
+ "grad_norm": 0.09325719624757767,
829
+ "learning_rate": 0.00030875994818330957,
830
+ "loss": 0.2032,
831
+ "step": 1875
832
+ },
833
+ {
834
+ "epoch": 1.5586546349466777,
835
+ "grad_norm": 0.07193212956190109,
836
+ "learning_rate": 0.0002833401158789207,
837
+ "loss": 0.1999,
838
+ "step": 1900
839
+ },
840
+ {
841
+ "epoch": 1.5586546349466777,
842
+ "eval_loss": 0.21670256555080414,
843
+ "eval_runtime": 19.702,
844
+ "eval_samples_per_second": 50.756,
845
+ "eval_steps_per_second": 0.812,
846
+ "step": 1900
847
+ },
848
+ {
849
+ "epoch": 1.579163248564397,
850
+ "grad_norm": 0.09132040292024612,
851
+ "learning_rate": 0.00025883855957295053,
852
+ "loss": 0.2028,
853
+ "step": 1925
854
+ },
855
+ {
856
+ "epoch": 1.5996718621821167,
857
+ "grad_norm": 0.0698297768831253,
858
+ "learning_rate": 0.0002352866737826277,
859
+ "loss": 0.2086,
860
+ "step": 1950
861
+ },
862
+ {
863
+ "epoch": 1.5996718621821167,
864
+ "eval_loss": 0.21480028331279755,
865
+ "eval_runtime": 19.7213,
866
+ "eval_samples_per_second": 50.707,
867
+ "eval_steps_per_second": 0.811,
868
+ "step": 1950
869
+ },
870
+ {
871
+ "epoch": 1.620180475799836,
872
+ "grad_norm": 0.06829634308815002,
873
+ "learning_rate": 0.00021271463618625986,
874
+ "loss": 0.2038,
875
+ "step": 1975
876
+ },
877
+ {
878
+ "epoch": 1.6406890894175554,
879
+ "grad_norm": 0.08400589972734451,
880
+ "learning_rate": 0.00019115136895574402,
881
+ "loss": 0.2044,
882
+ "step": 2000
883
+ },
884
+ {
885
+ "epoch": 1.6406890894175554,
886
+ "eval_loss": 0.21401509642601013,
887
+ "eval_runtime": 19.6626,
888
+ "eval_samples_per_second": 50.858,
889
+ "eval_steps_per_second": 0.814,
890
+ "step": 2000
891
+ },
892
+ {
893
+ "epoch": 1.661197703035275,
894
+ "grad_norm": 0.07338040322065353,
895
+ "learning_rate": 0.0001706245016977931,
896
+ "loss": 0.1971,
897
+ "step": 2025
898
+ },
899
+ {
900
+ "epoch": 1.6817063166529942,
901
+ "grad_norm": 0.09065773338079453,
902
+ "learning_rate": 0.00015116033605136182,
903
+ "loss": 0.2006,
904
+ "step": 2050
905
+ },
906
+ {
907
+ "epoch": 1.6817063166529942,
908
+ "eval_loss": 0.21327920258045197,
909
+ "eval_runtime": 19.6976,
910
+ "eval_samples_per_second": 50.768,
911
+ "eval_steps_per_second": 0.812,
912
+ "step": 2050
913
+ },
914
+ {
915
+ "epoch": 1.7022149302707137,
916
+ "grad_norm": 0.08500388264656067,
917
+ "learning_rate": 0.00013278381198663492,
918
+ "loss": 0.2007,
919
+ "step": 2075
920
+ },
921
+ {
922
+ "epoch": 1.7227235438884332,
923
+ "grad_norm": 0.08026055246591568,
924
+ "learning_rate": 0.0001155184758487573,
925
+ "loss": 0.196,
926
+ "step": 2100
927
+ },
928
+ {
929
+ "epoch": 1.7227235438884332,
930
+ "eval_loss": 0.21227264404296875,
931
+ "eval_runtime": 19.7037,
932
+ "eval_samples_per_second": 50.752,
933
+ "eval_steps_per_second": 0.812,
934
+ "step": 2100
935
+ },
936
+ {
937
+ "epoch": 1.7432321575061525,
938
+ "grad_norm": 0.07909256964921951,
939
+ "learning_rate": 9.938645018725523e-05,
940
+ "loss": 0.1982,
941
+ "step": 2125
942
+ },
943
+ {
944
+ "epoch": 1.7637407711238722,
945
+ "grad_norm": 0.08352374285459518,
946
+ "learning_rate": 8.440840540980587e-05,
947
+ "loss": 0.1916,
948
+ "step": 2150
949
+ },
950
+ {
951
+ "epoch": 1.7637407711238722,
952
+ "eval_loss": 0.21153756976127625,
953
+ "eval_runtime": 19.7017,
954
+ "eval_samples_per_second": 50.757,
955
+ "eval_steps_per_second": 0.812,
956
+ "step": 2150
957
+ },
958
+ {
959
+ "epoch": 1.7842493847415914,
960
+ "grad_norm": 0.08536770939826965,
961
+ "learning_rate": 7.060353329667668e-05,
962
+ "loss": 0.1931,
963
+ "step": 2175
964
+ },
965
+ {
966
+ "epoch": 1.804757998359311,
967
+ "grad_norm": 0.09643861651420593,
968
+ "learning_rate": 5.798952240976951e-05,
969
+ "loss": 0.2031,
970
+ "step": 2200
971
+ },
972
+ {
973
+ "epoch": 1.804757998359311,
974
+ "eval_loss": 0.21134978532791138,
975
+ "eval_runtime": 19.7003,
976
+ "eval_samples_per_second": 50.761,
977
+ "eval_steps_per_second": 0.812,
978
+ "step": 2200
979
+ },
980
+ {
981
+ "epoch": 1.8252666119770304,
982
+ "grad_norm": 0.08348719030618668,
983
+ "learning_rate": 4.65825354277799e-05,
984
+ "loss": 0.1938,
985
+ "step": 2225
986
+ },
987
+ {
988
+ "epoch": 1.8457752255947497,
989
+ "grad_norm": 0.07448319345712662,
990
+ "learning_rate": 3.639718843651363e-05,
991
+ "loss": 0.1951,
992
+ "step": 2250
993
+ },
994
+ {
995
+ "epoch": 1.8457752255947497,
996
+ "eval_loss": 0.21119001507759094,
997
+ "eval_runtime": 19.682,
998
+ "eval_samples_per_second": 50.808,
999
+ "eval_steps_per_second": 0.813,
1000
+ "step": 2250
1001
+ },
1002
+ {
1003
+ "epoch": 1.8662838392124692,
1004
+ "grad_norm": 0.09347943216562271,
1005
+ "learning_rate": 2.7446532200894104e-05,
1006
+ "loss": 0.1981,
1007
+ "step": 2275
1008
+ },
1009
+ {
1010
+ "epoch": 1.8867924528301887,
1011
+ "grad_norm": 0.09893514215946198,
1012
+ "learning_rate": 1.9742035442658403e-05,
1013
+ "loss": 0.2058,
1014
+ "step": 2300
1015
+ },
1016
+ {
1017
+ "epoch": 1.8867924528301887,
1018
+ "eval_loss": 0.21084289252758026,
1019
+ "eval_runtime": 19.6062,
1020
+ "eval_samples_per_second": 51.004,
1021
+ "eval_steps_per_second": 0.816,
1022
+ "step": 2300
1023
+ },
1024
+ {
1025
+ "epoch": 1.907301066447908,
1026
+ "grad_norm": 0.08711014688014984,
1027
+ "learning_rate": 1.3293570145169742e-05,
1028
+ "loss": 0.1995,
1029
+ "step": 2325
1030
+ },
1031
+ {
1032
+ "epoch": 1.9278096800656277,
1033
+ "grad_norm": 0.07174449414014816,
1034
+ "learning_rate": 8.109398904173282e-06,
1035
+ "loss": 0.1947,
1036
+ "step": 2350
1037
+ },
1038
+ {
1039
+ "epoch": 1.9278096800656277,
1040
+ "eval_loss": 0.2107050120830536,
1041
+ "eval_runtime": 19.6954,
1042
+ "eval_samples_per_second": 50.773,
1043
+ "eval_steps_per_second": 0.812,
1044
+ "step": 2350
1045
+ },
1046
+ {
1047
+ "epoch": 1.948318293683347,
1048
+ "grad_norm": 0.08628223091363907,
1049
+ "learning_rate": 4.196164340705577e-06,
1050
+ "loss": 0.2086,
1051
+ "step": 2375
1052
+ },
1053
+ {
1054
+ "epoch": 1.9688269073010665,
1055
+ "grad_norm": 0.08006993681192398,
1056
+ "learning_rate": 1.5588805897215342e-06,
1057
+ "loss": 0.2041,
1058
+ "step": 2400
1059
+ },
1060
+ {
1061
+ "epoch": 1.9688269073010665,
1062
+ "eval_loss": 0.2106114774942398,
1063
+ "eval_runtime": 19.6968,
1064
+ "eval_samples_per_second": 50.77,
1065
+ "eval_steps_per_second": 0.812,
1066
+ "step": 2400
1067
+ },
1068
+ {
1069
+ "epoch": 1.989335520918786,
1070
+ "grad_norm": 0.09150315076112747,
1071
+ "learning_rate": 2.0092687534589705e-07,
1072
+ "loss": 0.1938,
1073
+ "step": 2425
1074
+ },
1075
+ {
1076
+ "epoch": 2.0,
1077
+ "step": 2438,
1078
+ "total_flos": 1.58523627405312e+18,
1079
+ "train_loss": 0.2521084311362463,
1080
+ "train_runtime": 3435.0992,
1081
+ "train_samples_per_second": 22.707,
1082
+ "train_steps_per_second": 0.71
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
+ }