w-ahmad commited on
Commit
29c0a17
·
verified ·
1 Parent(s): e5fdcfe

Auto upload zain 2026-08-17T22:11:49.260029

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +2 -0
  2. zain/Activation/__pycache__/exp.cpython-311.pyc +0 -0
  3. zain/Activation/exp.py +2 -2
  4. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/config.json +37 -0
  5. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/model.safetensors +3 -0
  6. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/optimizer.pt +3 -0
  7. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/rng_state.pth +3 -0
  8. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/scheduler.pt +3 -0
  9. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/tokenizer.json +0 -0
  10. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/tokenizer_config.json +13 -0
  11. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/trainer_state.json +960 -0
  12. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/training_args.bin +3 -0
  13. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/config.json +37 -0
  14. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/model.safetensors +3 -0
  15. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/optimizer.pt +3 -0
  16. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/rng_state.pth +3 -0
  17. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/scheduler.pt +3 -0
  18. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/tokenizer.json +0 -0
  19. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/tokenizer_config.json +13 -0
  20. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/trainer_state.json +995 -0
  21. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/training_args.bin +3 -0
  22. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/config.json +37 -0
  23. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/model.safetensors +3 -0
  24. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/optimizer.pt +3 -0
  25. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/rng_state.pth +3 -0
  26. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/scheduler.pt +3 -0
  27. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/tokenizer.json +0 -0
  28. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/tokenizer_config.json +13 -0
  29. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/trainer_state.json +1030 -0
  30. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/training_args.bin +3 -0
  31. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/config.json +37 -0
  32. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/model.safetensors +3 -0
  33. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/optimizer.pt +3 -0
  34. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/rng_state.pth +3 -0
  35. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/scheduler.pt +3 -0
  36. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/tokenizer.json +0 -0
  37. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/tokenizer_config.json +13 -0
  38. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/trainer_state.json +1065 -0
  39. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/training_args.bin +3 -0
  40. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/config.json +37 -0
  41. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/model.safetensors +3 -0
  42. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/optimizer.pt +3 -0
  43. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/rng_state.pth +3 -0
  44. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/scheduler.pt +3 -0
  45. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/tokenizer.json +0 -0
  46. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/tokenizer_config.json +13 -0
  47. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/trainer_state.json +1100 -0
  48. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/training_args.bin +3 -0
  49. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3100/config.json +37 -0
  50. zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3100/model.safetensors +3 -0
.gitattributes CHANGED
@@ -37,3 +37,5 @@ zain/Activation/wandb/run-20260817_205627-jwz3vvv2/run-jwz3vvv2.wandb filter=lfs
37
  zain/Activation/wandb/run-20260817_210947-oa1s6mjf/run-oa1s6mjf.wandb filter=lfs diff=lfs merge=lfs -text
38
  zain/Activation/wandb/run-20260817_212143-2al1ijxg/run-2al1ijxg.wandb filter=lfs diff=lfs merge=lfs -text
39
  zain/Activation/wandb/run-20260817_214248-bayjxr06/run-bayjxr06.wandb filter=lfs diff=lfs merge=lfs -text
 
 
 
37
  zain/Activation/wandb/run-20260817_210947-oa1s6mjf/run-oa1s6mjf.wandb filter=lfs diff=lfs merge=lfs -text
38
  zain/Activation/wandb/run-20260817_212143-2al1ijxg/run-2al1ijxg.wandb filter=lfs diff=lfs merge=lfs -text
39
  zain/Activation/wandb/run-20260817_214248-bayjxr06/run-bayjxr06.wandb filter=lfs diff=lfs merge=lfs -text
40
+ zain/Activation/out/glu-powlu-100L_trash_run/training_log.jsonl filter=lfs diff=lfs merge=lfs -text
41
+ zain/Activation/wandb/run-20260817_220435-nrrgbtk5/run-nrrgbtk5.wandb filter=lfs diff=lfs merge=lfs -text
zain/Activation/__pycache__/exp.cpython-311.pyc CHANGED
Binary files a/zain/Activation/__pycache__/exp.cpython-311.pyc and b/zain/Activation/__pycache__/exp.cpython-311.pyc differ
 
zain/Activation/exp.py CHANGED
@@ -1,5 +1,5 @@
1
  # =====================================================================
2
- # exp.py – FULL FILE WITH PowLU INTEGRATION (NaN-gradient fix applied)
3
  # =====================================================================
4
 
5
  import math
@@ -950,4 +950,4 @@ def create_trainer(
950
  callbacks=callbacks,
951
  )
952
 
953
- return trainer
 
1
  # =====================================================================
2
+ # exp.py
3
  # =====================================================================
4
 
5
  import math
 
950
  callbacks=callbacks,
951
  )
952
 
953
+ return trainer
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/config.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation": "powlu",
3
+ "architectures": [
4
+ "TinyLlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 1,
9
+ "dtype": "bfloat16",
10
+ "eos_token_id": 2,
11
+ "head_dim": 32,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 128,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 256,
16
+ "max_position_embeddings": 512,
17
+ "mlp_bias": false,
18
+ "mlp_type": "glu",
19
+ "model_type": "tiny_llama",
20
+ "num_attention_heads": 4,
21
+ "num_hidden_layers": 100,
22
+ "num_key_value_heads": 4,
23
+ "pad_token_id": 0,
24
+ "powlu_m": 3.0,
25
+ "pretraining_tp": 1,
26
+ "rms_norm_eps": 1e-06,
27
+ "rope_parameters": {
28
+ "rope_theta": 10000.0,
29
+ "rope_type": "default"
30
+ },
31
+ "tie_word_embeddings": true,
32
+ "tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
33
+ "transformers_version": "5.16.0.dev0",
34
+ "use_cache": false,
35
+ "vocab_size": 4096,
36
+ "waleed_beta": 10
37
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8818b1a179cb3a5d727fcef986db41b57f7c50fada803ab9448e8413a49b9ffd
3
+ size 33967272
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59226b43f096d4843dafa1d5f464ac6dbcb928d06370e29e91ec9aea13cb6b42
3
+ size 39573050
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cf9097d4513154245c48236b6ec5137b7ee2a21c9f58f2cba798ea275c6026f
3
+ size 14244
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:36621a4ac2b8691c903e4eb5274dfa5ddcba796f686a593c9d0bb266866d9324
3
+ size 1064
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/tokenizer_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "backend": "tokenizers",
4
+ "bos_token": "<|endoftext|>",
5
+ "eos_token": "<|endoftext|>",
6
+ "errors": "replace",
7
+ "is_local": false,
8
+ "local_files_only": false,
9
+ "model_max_length": 1024,
10
+ "pad_token": "<|endoftext|>",
11
+ "tokenizer_class": "GPT2Tokenizer",
12
+ "unk_token": "<|endoftext|>"
13
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/trainer_state.json ADDED
@@ -0,0 +1,960 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.17526120660599934,
6
+ "eval_steps": 2498,
7
+ "global_step": 2600,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.0013481631277384564,
14
+ "grad_norm": 2.09375,
15
+ "learning_rate": 2.66e-05,
16
+ "loss": 8.300566864013671,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.002696326255476913,
21
+ "grad_norm": 1.375,
22
+ "learning_rate": 5.46e-05,
23
+ "loss": 8.056343841552735,
24
+ "step": 40
25
+ },
26
+ {
27
+ "epoch": 0.004044489383215369,
28
+ "grad_norm": 1.2734375,
29
+ "learning_rate": 8.259999999999999e-05,
30
+ "loss": 7.841481781005859,
31
+ "step": 60
32
+ },
33
+ {
34
+ "epoch": 0.005392652510953826,
35
+ "grad_norm": 1.265625,
36
+ "learning_rate": 0.0001106,
37
+ "loss": 7.57496337890625,
38
+ "step": 80
39
+ },
40
+ {
41
+ "epoch": 0.006740815638692282,
42
+ "grad_norm": 1.1953125,
43
+ "learning_rate": 0.0001386,
44
+ "loss": 7.233026123046875,
45
+ "step": 100
46
+ },
47
+ {
48
+ "epoch": 0.008088978766430738,
49
+ "grad_norm": 1.1328125,
50
+ "learning_rate": 0.00016659999999999998,
51
+ "loss": 6.859487152099609,
52
+ "step": 120
53
+ },
54
+ {
55
+ "epoch": 0.009437141894169195,
56
+ "grad_norm": 0.94921875,
57
+ "learning_rate": 0.00019460000000000001,
58
+ "loss": 6.5162200927734375,
59
+ "step": 140
60
+ },
61
+ {
62
+ "epoch": 0.010785305021907651,
63
+ "grad_norm": 0.796875,
64
+ "learning_rate": 0.0002226,
65
+ "loss": 6.237432861328125,
66
+ "step": 160
67
+ },
68
+ {
69
+ "epoch": 0.012133468149646108,
70
+ "grad_norm": 0.71875,
71
+ "learning_rate": 0.00025059999999999997,
72
+ "loss": 6.039522552490235,
73
+ "step": 180
74
+ },
75
+ {
76
+ "epoch": 0.013481631277384564,
77
+ "grad_norm": 0.57421875,
78
+ "learning_rate": 0.0002786,
79
+ "loss": 5.8798988342285154,
80
+ "step": 200
81
+ },
82
+ {
83
+ "epoch": 0.01482979440512302,
84
+ "grad_norm": 1.90625,
85
+ "learning_rate": 0.00030659999999999997,
86
+ "loss": 5.723947143554687,
87
+ "step": 220
88
+ },
89
+ {
90
+ "epoch": 0.016177957532861477,
91
+ "grad_norm": 0.6796875,
92
+ "learning_rate": 0.0003346,
93
+ "loss": 5.551593399047851,
94
+ "step": 240
95
+ },
96
+ {
97
+ "epoch": 0.01752612066059993,
98
+ "grad_norm": 1.6875,
99
+ "learning_rate": 0.00036260000000000003,
100
+ "loss": 5.3516895294189455,
101
+ "step": 260
102
+ },
103
+ {
104
+ "epoch": 0.01887428378833839,
105
+ "grad_norm": 0.84765625,
106
+ "learning_rate": 0.0003906,
107
+ "loss": 5.130373764038086,
108
+ "step": 280
109
+ },
110
+ {
111
+ "epoch": 0.020222446916076844,
112
+ "grad_norm": 0.55859375,
113
+ "learning_rate": 0.0004186,
114
+ "loss": 4.882920455932617,
115
+ "step": 300
116
+ },
117
+ {
118
+ "epoch": 0.021570610043815303,
119
+ "grad_norm": 1.15625,
120
+ "learning_rate": 0.0004466,
121
+ "loss": 4.673929977416992,
122
+ "step": 320
123
+ },
124
+ {
125
+ "epoch": 0.022918773171553757,
126
+ "grad_norm": 1.0,
127
+ "learning_rate": 0.00047460000000000004,
128
+ "loss": 4.511159515380859,
129
+ "step": 340
130
+ },
131
+ {
132
+ "epoch": 0.024266936299292215,
133
+ "grad_norm": 0.96875,
134
+ "learning_rate": 0.0005026,
135
+ "loss": 4.360583114624023,
136
+ "step": 360
137
+ },
138
+ {
139
+ "epoch": 0.02561509942703067,
140
+ "grad_norm": 0.80078125,
141
+ "learning_rate": 0.0005306,
142
+ "loss": 4.208797836303711,
143
+ "step": 380
144
+ },
145
+ {
146
+ "epoch": 0.026963262554769128,
147
+ "grad_norm": 0.703125,
148
+ "learning_rate": 0.0005586,
149
+ "loss": 4.059257507324219,
150
+ "step": 400
151
+ },
152
+ {
153
+ "epoch": 0.028311425682507583,
154
+ "grad_norm": 1.390625,
155
+ "learning_rate": 0.0005866,
156
+ "loss": 3.945643997192383,
157
+ "step": 420
158
+ },
159
+ {
160
+ "epoch": 0.02965958881024604,
161
+ "grad_norm": 0.8046875,
162
+ "learning_rate": 0.0006146,
163
+ "loss": 3.878635787963867,
164
+ "step": 440
165
+ },
166
+ {
167
+ "epoch": 0.031007751937984496,
168
+ "grad_norm": 0.7578125,
169
+ "learning_rate": 0.0006426,
170
+ "loss": 3.763169860839844,
171
+ "step": 460
172
+ },
173
+ {
174
+ "epoch": 0.032355915065722954,
175
+ "grad_norm": 0.578125,
176
+ "learning_rate": 0.0006705999999999999,
177
+ "loss": 3.680609130859375,
178
+ "step": 480
179
+ },
180
+ {
181
+ "epoch": 0.03370407819346141,
182
+ "grad_norm": 0.6484375,
183
+ "learning_rate": 0.0006986,
184
+ "loss": 3.6085578918457033,
185
+ "step": 500
186
+ },
187
+ {
188
+ "epoch": 0.03505224132119986,
189
+ "grad_norm": 0.431640625,
190
+ "learning_rate": 0.0007,
191
+ "loss": 3.5324871063232424,
192
+ "step": 520
193
+ },
194
+ {
195
+ "epoch": 0.03640040444893832,
196
+ "grad_norm": 0.4453125,
197
+ "learning_rate": 0.0007,
198
+ "loss": 3.454126739501953,
199
+ "step": 540
200
+ },
201
+ {
202
+ "epoch": 0.03774856757667678,
203
+ "grad_norm": 0.546875,
204
+ "learning_rate": 0.0007,
205
+ "loss": 3.4041587829589846,
206
+ "step": 560
207
+ },
208
+ {
209
+ "epoch": 0.03909673070441524,
210
+ "grad_norm": 0.46875,
211
+ "learning_rate": 0.0007,
212
+ "loss": 3.347498321533203,
213
+ "step": 580
214
+ },
215
+ {
216
+ "epoch": 0.04044489383215369,
217
+ "grad_norm": 0.546875,
218
+ "learning_rate": 0.0007,
219
+ "loss": 3.313167190551758,
220
+ "step": 600
221
+ },
222
+ {
223
+ "epoch": 0.04179305695989215,
224
+ "grad_norm": 0.51953125,
225
+ "learning_rate": 0.0007,
226
+ "loss": 3.2490795135498045,
227
+ "step": 620
228
+ },
229
+ {
230
+ "epoch": 0.043141220087630605,
231
+ "grad_norm": 0.453125,
232
+ "learning_rate": 0.0007,
233
+ "loss": 3.221387481689453,
234
+ "step": 640
235
+ },
236
+ {
237
+ "epoch": 0.044489383215369056,
238
+ "grad_norm": 0.5234375,
239
+ "learning_rate": 0.0007,
240
+ "loss": 3.1599220275878905,
241
+ "step": 660
242
+ },
243
+ {
244
+ "epoch": 0.045837546343107514,
245
+ "grad_norm": 0.4375,
246
+ "learning_rate": 0.0007,
247
+ "loss": 3.133222198486328,
248
+ "step": 680
249
+ },
250
+ {
251
+ "epoch": 0.04718570947084597,
252
+ "grad_norm": 0.5703125,
253
+ "learning_rate": 0.0007,
254
+ "loss": 3.0884355545043944,
255
+ "step": 700
256
+ },
257
+ {
258
+ "epoch": 0.04853387259858443,
259
+ "grad_norm": 0.431640625,
260
+ "learning_rate": 0.0007,
261
+ "loss": 3.0560993194580077,
262
+ "step": 720
263
+ },
264
+ {
265
+ "epoch": 0.04988203572632288,
266
+ "grad_norm": 0.462890625,
267
+ "learning_rate": 0.0007,
268
+ "loss": 3.0169742584228514,
269
+ "step": 740
270
+ },
271
+ {
272
+ "epoch": 0.05123019885406134,
273
+ "grad_norm": 0.390625,
274
+ "learning_rate": 0.0007,
275
+ "loss": 2.9988849639892576,
276
+ "step": 760
277
+ },
278
+ {
279
+ "epoch": 0.0525783619817998,
280
+ "grad_norm": 0.466796875,
281
+ "learning_rate": 0.0007,
282
+ "loss": 3.004904937744141,
283
+ "step": 780
284
+ },
285
+ {
286
+ "epoch": 0.053926525109538256,
287
+ "grad_norm": 0.443359375,
288
+ "learning_rate": 0.0007,
289
+ "loss": 2.9500053405761717,
290
+ "step": 800
291
+ },
292
+ {
293
+ "epoch": 0.05527468823727671,
294
+ "grad_norm": 0.453125,
295
+ "learning_rate": 0.0007,
296
+ "loss": 2.93444938659668,
297
+ "step": 820
298
+ },
299
+ {
300
+ "epoch": 0.056622851365015166,
301
+ "grad_norm": 0.421875,
302
+ "learning_rate": 0.0007,
303
+ "loss": 2.9036865234375,
304
+ "step": 840
305
+ },
306
+ {
307
+ "epoch": 0.057971014492753624,
308
+ "grad_norm": 0.392578125,
309
+ "learning_rate": 0.0007,
310
+ "loss": 2.859415054321289,
311
+ "step": 860
312
+ },
313
+ {
314
+ "epoch": 0.05931917762049208,
315
+ "grad_norm": 0.42578125,
316
+ "learning_rate": 0.0007,
317
+ "loss": 2.8376497268676757,
318
+ "step": 880
319
+ },
320
+ {
321
+ "epoch": 0.06066734074823053,
322
+ "grad_norm": 0.46484375,
323
+ "learning_rate": 0.0007,
324
+ "loss": 2.821657180786133,
325
+ "step": 900
326
+ },
327
+ {
328
+ "epoch": 0.06201550387596899,
329
+ "grad_norm": 0.46875,
330
+ "learning_rate": 0.0007,
331
+ "loss": 2.7936939239501952,
332
+ "step": 920
333
+ },
334
+ {
335
+ "epoch": 0.06336366700370745,
336
+ "grad_norm": 0.40234375,
337
+ "learning_rate": 0.0007,
338
+ "loss": 2.779148292541504,
339
+ "step": 940
340
+ },
341
+ {
342
+ "epoch": 0.06471183013144591,
343
+ "grad_norm": 0.40234375,
344
+ "learning_rate": 0.0007,
345
+ "loss": 2.7523166656494142,
346
+ "step": 960
347
+ },
348
+ {
349
+ "epoch": 0.06605999325918437,
350
+ "grad_norm": 0.408203125,
351
+ "learning_rate": 0.0007,
352
+ "loss": 2.7326622009277344,
353
+ "step": 980
354
+ },
355
+ {
356
+ "epoch": 0.06740815638692282,
357
+ "grad_norm": 0.396484375,
358
+ "learning_rate": 0.0007,
359
+ "loss": 2.730000305175781,
360
+ "step": 1000
361
+ },
362
+ {
363
+ "epoch": 0.06875631951466127,
364
+ "grad_norm": 0.3828125,
365
+ "learning_rate": 0.0007,
366
+ "loss": 2.6984880447387694,
367
+ "step": 1020
368
+ },
369
+ {
370
+ "epoch": 0.07010448264239973,
371
+ "grad_norm": 0.423828125,
372
+ "learning_rate": 0.0007,
373
+ "loss": 2.6830041885375975,
374
+ "step": 1040
375
+ },
376
+ {
377
+ "epoch": 0.07145264577013818,
378
+ "grad_norm": 0.404296875,
379
+ "learning_rate": 0.0007,
380
+ "loss": 2.6666488647460938,
381
+ "step": 1060
382
+ },
383
+ {
384
+ "epoch": 0.07280080889787664,
385
+ "grad_norm": 0.404296875,
386
+ "learning_rate": 0.0007,
387
+ "loss": 2.6461860656738283,
388
+ "step": 1080
389
+ },
390
+ {
391
+ "epoch": 0.0741489720256151,
392
+ "grad_norm": 0.427734375,
393
+ "learning_rate": 0.0007,
394
+ "loss": 2.638811492919922,
395
+ "step": 1100
396
+ },
397
+ {
398
+ "epoch": 0.07549713515335356,
399
+ "grad_norm": 0.38671875,
400
+ "learning_rate": 0.0007,
401
+ "loss": 2.604878807067871,
402
+ "step": 1120
403
+ },
404
+ {
405
+ "epoch": 0.07684529828109202,
406
+ "grad_norm": 0.423828125,
407
+ "learning_rate": 0.0007,
408
+ "loss": 2.5887590408325196,
409
+ "step": 1140
410
+ },
411
+ {
412
+ "epoch": 0.07819346140883048,
413
+ "grad_norm": 0.416015625,
414
+ "learning_rate": 0.0007,
415
+ "loss": 2.5864500045776366,
416
+ "step": 1160
417
+ },
418
+ {
419
+ "epoch": 0.07954162453656892,
420
+ "grad_norm": 0.40234375,
421
+ "learning_rate": 0.0007,
422
+ "loss": 2.5788290023803713,
423
+ "step": 1180
424
+ },
425
+ {
426
+ "epoch": 0.08088978766430738,
427
+ "grad_norm": 0.404296875,
428
+ "learning_rate": 0.0007,
429
+ "loss": 2.570602607727051,
430
+ "step": 1200
431
+ },
432
+ {
433
+ "epoch": 0.08223795079204584,
434
+ "grad_norm": 0.390625,
435
+ "learning_rate": 0.0007,
436
+ "loss": 2.5528938293457033,
437
+ "step": 1220
438
+ },
439
+ {
440
+ "epoch": 0.0835861139197843,
441
+ "grad_norm": 0.373046875,
442
+ "learning_rate": 0.0007,
443
+ "loss": 2.53753662109375,
444
+ "step": 1240
445
+ },
446
+ {
447
+ "epoch": 0.08493427704752275,
448
+ "grad_norm": 0.384765625,
449
+ "learning_rate": 0.0007,
450
+ "loss": 2.523891830444336,
451
+ "step": 1260
452
+ },
453
+ {
454
+ "epoch": 0.08628244017526121,
455
+ "grad_norm": 0.3671875,
456
+ "learning_rate": 0.0007,
457
+ "loss": 2.5049261093139648,
458
+ "step": 1280
459
+ },
460
+ {
461
+ "epoch": 0.08763060330299967,
462
+ "grad_norm": 0.392578125,
463
+ "learning_rate": 0.0007,
464
+ "loss": 2.4961641311645506,
465
+ "step": 1300
466
+ },
467
+ {
468
+ "epoch": 0.08897876643073811,
469
+ "grad_norm": 0.353515625,
470
+ "learning_rate": 0.0007,
471
+ "loss": 2.487112045288086,
472
+ "step": 1320
473
+ },
474
+ {
475
+ "epoch": 0.09032692955847657,
476
+ "grad_norm": 0.38671875,
477
+ "learning_rate": 0.0007,
478
+ "loss": 2.471820259094238,
479
+ "step": 1340
480
+ },
481
+ {
482
+ "epoch": 0.09167509268621503,
483
+ "grad_norm": 0.392578125,
484
+ "learning_rate": 0.0007,
485
+ "loss": 2.4697391510009767,
486
+ "step": 1360
487
+ },
488
+ {
489
+ "epoch": 0.09302325581395349,
490
+ "grad_norm": 0.369140625,
491
+ "learning_rate": 0.0007,
492
+ "loss": 2.4492652893066404,
493
+ "step": 1380
494
+ },
495
+ {
496
+ "epoch": 0.09437141894169195,
497
+ "grad_norm": 0.384765625,
498
+ "learning_rate": 0.0007,
499
+ "loss": 2.4387527465820313,
500
+ "step": 1400
501
+ },
502
+ {
503
+ "epoch": 0.0957195820694304,
504
+ "grad_norm": 0.380859375,
505
+ "learning_rate": 0.0007,
506
+ "loss": 2.4479516983032226,
507
+ "step": 1420
508
+ },
509
+ {
510
+ "epoch": 0.09706774519716886,
511
+ "grad_norm": 0.359375,
512
+ "learning_rate": 0.0007,
513
+ "loss": 2.4246395111083983,
514
+ "step": 1440
515
+ },
516
+ {
517
+ "epoch": 0.09841590832490732,
518
+ "grad_norm": 0.373046875,
519
+ "learning_rate": 0.0007,
520
+ "loss": 2.4114765167236327,
521
+ "step": 1460
522
+ },
523
+ {
524
+ "epoch": 0.09976407145264576,
525
+ "grad_norm": 0.375,
526
+ "learning_rate": 0.0007,
527
+ "loss": 2.399285888671875,
528
+ "step": 1480
529
+ },
530
+ {
531
+ "epoch": 0.10111223458038422,
532
+ "grad_norm": 0.37890625,
533
+ "learning_rate": 0.0007,
534
+ "loss": 2.3839179992675783,
535
+ "step": 1500
536
+ },
537
+ {
538
+ "epoch": 0.10246039770812268,
539
+ "grad_norm": 0.41796875,
540
+ "learning_rate": 0.0007,
541
+ "loss": 2.3861133575439455,
542
+ "step": 1520
543
+ },
544
+ {
545
+ "epoch": 0.10380856083586114,
546
+ "grad_norm": 0.40625,
547
+ "learning_rate": 0.0007,
548
+ "loss": 2.3732574462890623,
549
+ "step": 1540
550
+ },
551
+ {
552
+ "epoch": 0.1051567239635996,
553
+ "grad_norm": 0.36328125,
554
+ "learning_rate": 0.0007,
555
+ "loss": 2.3624162673950195,
556
+ "step": 1560
557
+ },
558
+ {
559
+ "epoch": 0.10650488709133805,
560
+ "grad_norm": 0.359375,
561
+ "learning_rate": 0.0007,
562
+ "loss": 2.3779842376708986,
563
+ "step": 1580
564
+ },
565
+ {
566
+ "epoch": 0.10785305021907651,
567
+ "grad_norm": 0.369140625,
568
+ "learning_rate": 0.0007,
569
+ "loss": 2.3441116333007814,
570
+ "step": 1600
571
+ },
572
+ {
573
+ "epoch": 0.10920121334681497,
574
+ "grad_norm": 0.376953125,
575
+ "learning_rate": 0.0007,
576
+ "loss": 2.3418384552001954,
577
+ "step": 1620
578
+ },
579
+ {
580
+ "epoch": 0.11054937647455342,
581
+ "grad_norm": 0.373046875,
582
+ "learning_rate": 0.0007,
583
+ "loss": 2.3539493560791014,
584
+ "step": 1640
585
+ },
586
+ {
587
+ "epoch": 0.11189753960229187,
588
+ "grad_norm": 0.359375,
589
+ "learning_rate": 0.0007,
590
+ "loss": 2.332906150817871,
591
+ "step": 1660
592
+ },
593
+ {
594
+ "epoch": 0.11324570273003033,
595
+ "grad_norm": 0.408203125,
596
+ "learning_rate": 0.0007,
597
+ "loss": 2.3174556732177733,
598
+ "step": 1680
599
+ },
600
+ {
601
+ "epoch": 0.11459386585776879,
602
+ "grad_norm": 0.361328125,
603
+ "learning_rate": 0.0007,
604
+ "loss": 2.330640411376953,
605
+ "step": 1700
606
+ },
607
+ {
608
+ "epoch": 0.11594202898550725,
609
+ "grad_norm": 0.423828125,
610
+ "learning_rate": 0.0007,
611
+ "loss": 2.306385803222656,
612
+ "step": 1720
613
+ },
614
+ {
615
+ "epoch": 0.1172901921132457,
616
+ "grad_norm": 0.365234375,
617
+ "learning_rate": 0.0007,
618
+ "loss": 2.2937660217285156,
619
+ "step": 1740
620
+ },
621
+ {
622
+ "epoch": 0.11863835524098416,
623
+ "grad_norm": 0.3671875,
624
+ "learning_rate": 0.0007,
625
+ "loss": 2.297491264343262,
626
+ "step": 1760
627
+ },
628
+ {
629
+ "epoch": 0.11998651836872262,
630
+ "grad_norm": 0.337890625,
631
+ "learning_rate": 0.0007,
632
+ "loss": 2.312083053588867,
633
+ "step": 1780
634
+ },
635
+ {
636
+ "epoch": 0.12133468149646107,
637
+ "grad_norm": 0.357421875,
638
+ "learning_rate": 0.0007,
639
+ "loss": 2.307674789428711,
640
+ "step": 1800
641
+ },
642
+ {
643
+ "epoch": 0.12268284462419952,
644
+ "grad_norm": 0.375,
645
+ "learning_rate": 0.0007,
646
+ "loss": 2.296724319458008,
647
+ "step": 1820
648
+ },
649
+ {
650
+ "epoch": 0.12403100775193798,
651
+ "grad_norm": 0.376953125,
652
+ "learning_rate": 0.0007,
653
+ "loss": 2.2644861221313475,
654
+ "step": 1840
655
+ },
656
+ {
657
+ "epoch": 0.12537917087967643,
658
+ "grad_norm": 0.353515625,
659
+ "learning_rate": 0.0007,
660
+ "loss": 2.2662803649902346,
661
+ "step": 1860
662
+ },
663
+ {
664
+ "epoch": 0.1267273340074149,
665
+ "grad_norm": 0.33984375,
666
+ "learning_rate": 0.0007,
667
+ "loss": 2.2558313369750977,
668
+ "step": 1880
669
+ },
670
+ {
671
+ "epoch": 0.12807549713515334,
672
+ "grad_norm": 0.365234375,
673
+ "learning_rate": 0.0007,
674
+ "loss": 2.262881851196289,
675
+ "step": 1900
676
+ },
677
+ {
678
+ "epoch": 0.12942366026289182,
679
+ "grad_norm": 0.349609375,
680
+ "learning_rate": 0.0007,
681
+ "loss": 2.2384838104248046,
682
+ "step": 1920
683
+ },
684
+ {
685
+ "epoch": 0.13077182339063026,
686
+ "grad_norm": 0.361328125,
687
+ "learning_rate": 0.0007,
688
+ "loss": 2.247391700744629,
689
+ "step": 1940
690
+ },
691
+ {
692
+ "epoch": 0.13211998651836873,
693
+ "grad_norm": 0.34765625,
694
+ "learning_rate": 0.0007,
695
+ "loss": 2.2225168228149412,
696
+ "step": 1960
697
+ },
698
+ {
699
+ "epoch": 0.13346814964610718,
700
+ "grad_norm": 0.380859375,
701
+ "learning_rate": 0.0007,
702
+ "loss": 2.2470256805419924,
703
+ "step": 1980
704
+ },
705
+ {
706
+ "epoch": 0.13481631277384565,
707
+ "grad_norm": 0.3359375,
708
+ "learning_rate": 0.0007,
709
+ "loss": 2.218878173828125,
710
+ "step": 2000
711
+ },
712
+ {
713
+ "epoch": 0.1361644759015841,
714
+ "grad_norm": 0.357421875,
715
+ "learning_rate": 0.0007,
716
+ "loss": 2.227268409729004,
717
+ "step": 2020
718
+ },
719
+ {
720
+ "epoch": 0.13751263902932254,
721
+ "grad_norm": 0.380859375,
722
+ "learning_rate": 0.0007,
723
+ "loss": 2.2079793930053713,
724
+ "step": 2040
725
+ },
726
+ {
727
+ "epoch": 0.138860802157061,
728
+ "grad_norm": 0.380859375,
729
+ "learning_rate": 0.0007,
730
+ "loss": 2.2262474060058595,
731
+ "step": 2060
732
+ },
733
+ {
734
+ "epoch": 0.14020896528479945,
735
+ "grad_norm": 0.349609375,
736
+ "learning_rate": 0.0007,
737
+ "loss": 2.2017698287963867,
738
+ "step": 2080
739
+ },
740
+ {
741
+ "epoch": 0.14155712841253792,
742
+ "grad_norm": 0.333984375,
743
+ "learning_rate": 0.0007,
744
+ "loss": 2.19287223815918,
745
+ "step": 2100
746
+ },
747
+ {
748
+ "epoch": 0.14290529154027637,
749
+ "grad_norm": 0.349609375,
750
+ "learning_rate": 0.0007,
751
+ "loss": 2.201098823547363,
752
+ "step": 2120
753
+ },
754
+ {
755
+ "epoch": 0.14425345466801484,
756
+ "grad_norm": 0.333984375,
757
+ "learning_rate": 0.0007,
758
+ "loss": 2.2028568267822264,
759
+ "step": 2140
760
+ },
761
+ {
762
+ "epoch": 0.14560161779575329,
763
+ "grad_norm": 0.353515625,
764
+ "learning_rate": 0.0007,
765
+ "loss": 2.1869789123535157,
766
+ "step": 2160
767
+ },
768
+ {
769
+ "epoch": 0.14694978092349173,
770
+ "grad_norm": 0.33984375,
771
+ "learning_rate": 0.0007,
772
+ "loss": 2.194090461730957,
773
+ "step": 2180
774
+ },
775
+ {
776
+ "epoch": 0.1482979440512302,
777
+ "grad_norm": 0.333984375,
778
+ "learning_rate": 0.0007,
779
+ "loss": 2.169504165649414,
780
+ "step": 2200
781
+ },
782
+ {
783
+ "epoch": 0.14964610717896865,
784
+ "grad_norm": 0.359375,
785
+ "learning_rate": 0.0007,
786
+ "loss": 2.17819766998291,
787
+ "step": 2220
788
+ },
789
+ {
790
+ "epoch": 0.15099427030670712,
791
+ "grad_norm": 0.361328125,
792
+ "learning_rate": 0.0007,
793
+ "loss": 2.1870670318603516,
794
+ "step": 2240
795
+ },
796
+ {
797
+ "epoch": 0.15234243343444556,
798
+ "grad_norm": 0.359375,
799
+ "learning_rate": 0.0007,
800
+ "loss": 2.1656898498535155,
801
+ "step": 2260
802
+ },
803
+ {
804
+ "epoch": 0.15369059656218403,
805
+ "grad_norm": 0.369140625,
806
+ "learning_rate": 0.0007,
807
+ "loss": 2.174662971496582,
808
+ "step": 2280
809
+ },
810
+ {
811
+ "epoch": 0.15503875968992248,
812
+ "grad_norm": 0.3515625,
813
+ "learning_rate": 0.0007,
814
+ "loss": 2.1561347961425783,
815
+ "step": 2300
816
+ },
817
+ {
818
+ "epoch": 0.15638692281766095,
819
+ "grad_norm": 0.33203125,
820
+ "learning_rate": 0.0007,
821
+ "loss": 2.16455020904541,
822
+ "step": 2320
823
+ },
824
+ {
825
+ "epoch": 0.1577350859453994,
826
+ "grad_norm": 0.330078125,
827
+ "learning_rate": 0.0007,
828
+ "loss": 2.1551021575927733,
829
+ "step": 2340
830
+ },
831
+ {
832
+ "epoch": 0.15908324907313784,
833
+ "grad_norm": 0.341796875,
834
+ "learning_rate": 0.0007,
835
+ "loss": 2.157507133483887,
836
+ "step": 2360
837
+ },
838
+ {
839
+ "epoch": 0.1604314122008763,
840
+ "grad_norm": 0.33984375,
841
+ "learning_rate": 0.0007,
842
+ "loss": 2.1596670150756836,
843
+ "step": 2380
844
+ },
845
+ {
846
+ "epoch": 0.16177957532861476,
847
+ "grad_norm": 0.357421875,
848
+ "learning_rate": 0.0007,
849
+ "loss": 2.1417112350463867,
850
+ "step": 2400
851
+ },
852
+ {
853
+ "epoch": 0.16312773845635323,
854
+ "grad_norm": 0.349609375,
855
+ "learning_rate": 0.0007,
856
+ "loss": 2.1463193893432617,
857
+ "step": 2420
858
+ },
859
+ {
860
+ "epoch": 0.16447590158409167,
861
+ "grad_norm": 0.3359375,
862
+ "learning_rate": 0.0007,
863
+ "loss": 2.133936882019043,
864
+ "step": 2440
865
+ },
866
+ {
867
+ "epoch": 0.16582406471183014,
868
+ "grad_norm": 0.328125,
869
+ "learning_rate": 0.0007,
870
+ "loss": 2.1435129165649416,
871
+ "step": 2460
872
+ },
873
+ {
874
+ "epoch": 0.1671722278395686,
875
+ "grad_norm": 0.341796875,
876
+ "learning_rate": 0.0007,
877
+ "loss": 2.1443099975585938,
878
+ "step": 2480
879
+ },
880
+ {
881
+ "epoch": 0.1683855746545332,
882
+ "eval_loss": 2.138087511062622,
883
+ "eval_runtime": 20.1123,
884
+ "eval_samples_per_second": 473.691,
885
+ "eval_steps_per_second": 0.497,
886
+ "step": 2498
887
+ },
888
+ {
889
+ "epoch": 0.16852039096730703,
890
+ "grad_norm": 0.33984375,
891
+ "learning_rate": 0.0007,
892
+ "loss": 2.126860427856445,
893
+ "step": 2500
894
+ },
895
+ {
896
+ "epoch": 0.16852039096730703,
897
+ "eval_loss": 2.1375532150268555,
898
+ "eval_runtime": 20.4047,
899
+ "eval_samples_per_second": 466.903,
900
+ "eval_steps_per_second": 0.49,
901
+ "step": 2500
902
+ },
903
+ {
904
+ "epoch": 0.1698685540950455,
905
+ "grad_norm": 24903680.0,
906
+ "learning_rate": 3.4200000000000005e-05,
907
+ "loss": 2.128565788269043,
908
+ "step": 2520
909
+ },
910
+ {
911
+ "epoch": 0.17121671722278395,
912
+ "grad_norm": 27000832.0,
913
+ "learning_rate": 7.020000000000001e-05,
914
+ "loss": 2.1492637634277343,
915
+ "step": 2540
916
+ },
917
+ {
918
+ "epoch": 0.17256488035052242,
919
+ "grad_norm": 119537664.0,
920
+ "learning_rate": 9e-05,
921
+ "loss": 3.623552703857422,
922
+ "step": 2560
923
+ },
924
+ {
925
+ "epoch": 0.17391304347826086,
926
+ "grad_norm": 301989888.0,
927
+ "learning_rate": 9e-05,
928
+ "loss": 6.8911598205566404,
929
+ "step": 2580
930
+ },
931
+ {
932
+ "epoch": 0.17526120660599934,
933
+ "grad_norm": 658505728.0,
934
+ "learning_rate": 9e-05,
935
+ "loss": 7.545132446289062,
936
+ "step": 2600
937
+ }
938
+ ],
939
+ "logging_steps": 20,
940
+ "max_steps": 3500,
941
+ "num_input_tokens_seen": 0,
942
+ "num_train_epochs": 1,
943
+ "save_steps": 100,
944
+ "stateful_callbacks": {
945
+ "TrainerControl": {
946
+ "args": {
947
+ "should_epoch_stop": false,
948
+ "should_evaluate": false,
949
+ "should_log": false,
950
+ "should_save": true,
951
+ "should_training_stop": false
952
+ },
953
+ "attributes": {}
954
+ }
955
+ },
956
+ "total_flos": 8388337886822400.0,
957
+ "train_batch_size": 64,
958
+ "trial_name": null,
959
+ "trial_params": null
960
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2600/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f62f8186f489a47bc68f80ec9f826cd67dc6a0232e67ff9d099ed1157b6e25f
3
+ size 4920
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/config.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation": "powlu",
3
+ "architectures": [
4
+ "TinyLlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 1,
9
+ "dtype": "bfloat16",
10
+ "eos_token_id": 2,
11
+ "head_dim": 32,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 128,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 256,
16
+ "max_position_embeddings": 512,
17
+ "mlp_bias": false,
18
+ "mlp_type": "glu",
19
+ "model_type": "tiny_llama",
20
+ "num_attention_heads": 4,
21
+ "num_hidden_layers": 100,
22
+ "num_key_value_heads": 4,
23
+ "pad_token_id": 0,
24
+ "powlu_m": 3.0,
25
+ "pretraining_tp": 1,
26
+ "rms_norm_eps": 1e-06,
27
+ "rope_parameters": {
28
+ "rope_theta": 10000.0,
29
+ "rope_type": "default"
30
+ },
31
+ "tie_word_embeddings": true,
32
+ "tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
33
+ "transformers_version": "5.16.0.dev0",
34
+ "use_cache": false,
35
+ "vocab_size": 4096,
36
+ "waleed_beta": 10
37
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ecda5829fae7ce9053907c35dbd0c36d9d38f3db4ad9df7340319a033304d428
3
+ size 33967272
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b5e01c554dc2064cbbb6c418dc5eed61ad084930d0fb49072eb29ffe70f150b7
3
+ size 39573050
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cf9097d4513154245c48236b6ec5137b7ee2a21c9f58f2cba798ea275c6026f
3
+ size 14244
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6734940bf02a09ae82e3bc3a7a4079331a195a053fcffc27fad5719bee088bc4
3
+ size 1064
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/tokenizer_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "backend": "tokenizers",
4
+ "bos_token": "<|endoftext|>",
5
+ "eos_token": "<|endoftext|>",
6
+ "errors": "replace",
7
+ "is_local": false,
8
+ "local_files_only": false,
9
+ "model_max_length": 1024,
10
+ "pad_token": "<|endoftext|>",
11
+ "tokenizer_class": "GPT2Tokenizer",
12
+ "unk_token": "<|endoftext|>"
13
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/trainer_state.json ADDED
@@ -0,0 +1,995 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.1820020222446916,
6
+ "eval_steps": 2498,
7
+ "global_step": 2700,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.0013481631277384564,
14
+ "grad_norm": 2.09375,
15
+ "learning_rate": 2.66e-05,
16
+ "loss": 8.300566864013671,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.002696326255476913,
21
+ "grad_norm": 1.375,
22
+ "learning_rate": 5.46e-05,
23
+ "loss": 8.056343841552735,
24
+ "step": 40
25
+ },
26
+ {
27
+ "epoch": 0.004044489383215369,
28
+ "grad_norm": 1.2734375,
29
+ "learning_rate": 8.259999999999999e-05,
30
+ "loss": 7.841481781005859,
31
+ "step": 60
32
+ },
33
+ {
34
+ "epoch": 0.005392652510953826,
35
+ "grad_norm": 1.265625,
36
+ "learning_rate": 0.0001106,
37
+ "loss": 7.57496337890625,
38
+ "step": 80
39
+ },
40
+ {
41
+ "epoch": 0.006740815638692282,
42
+ "grad_norm": 1.1953125,
43
+ "learning_rate": 0.0001386,
44
+ "loss": 7.233026123046875,
45
+ "step": 100
46
+ },
47
+ {
48
+ "epoch": 0.008088978766430738,
49
+ "grad_norm": 1.1328125,
50
+ "learning_rate": 0.00016659999999999998,
51
+ "loss": 6.859487152099609,
52
+ "step": 120
53
+ },
54
+ {
55
+ "epoch": 0.009437141894169195,
56
+ "grad_norm": 0.94921875,
57
+ "learning_rate": 0.00019460000000000001,
58
+ "loss": 6.5162200927734375,
59
+ "step": 140
60
+ },
61
+ {
62
+ "epoch": 0.010785305021907651,
63
+ "grad_norm": 0.796875,
64
+ "learning_rate": 0.0002226,
65
+ "loss": 6.237432861328125,
66
+ "step": 160
67
+ },
68
+ {
69
+ "epoch": 0.012133468149646108,
70
+ "grad_norm": 0.71875,
71
+ "learning_rate": 0.00025059999999999997,
72
+ "loss": 6.039522552490235,
73
+ "step": 180
74
+ },
75
+ {
76
+ "epoch": 0.013481631277384564,
77
+ "grad_norm": 0.57421875,
78
+ "learning_rate": 0.0002786,
79
+ "loss": 5.8798988342285154,
80
+ "step": 200
81
+ },
82
+ {
83
+ "epoch": 0.01482979440512302,
84
+ "grad_norm": 1.90625,
85
+ "learning_rate": 0.00030659999999999997,
86
+ "loss": 5.723947143554687,
87
+ "step": 220
88
+ },
89
+ {
90
+ "epoch": 0.016177957532861477,
91
+ "grad_norm": 0.6796875,
92
+ "learning_rate": 0.0003346,
93
+ "loss": 5.551593399047851,
94
+ "step": 240
95
+ },
96
+ {
97
+ "epoch": 0.01752612066059993,
98
+ "grad_norm": 1.6875,
99
+ "learning_rate": 0.00036260000000000003,
100
+ "loss": 5.3516895294189455,
101
+ "step": 260
102
+ },
103
+ {
104
+ "epoch": 0.01887428378833839,
105
+ "grad_norm": 0.84765625,
106
+ "learning_rate": 0.0003906,
107
+ "loss": 5.130373764038086,
108
+ "step": 280
109
+ },
110
+ {
111
+ "epoch": 0.020222446916076844,
112
+ "grad_norm": 0.55859375,
113
+ "learning_rate": 0.0004186,
114
+ "loss": 4.882920455932617,
115
+ "step": 300
116
+ },
117
+ {
118
+ "epoch": 0.021570610043815303,
119
+ "grad_norm": 1.15625,
120
+ "learning_rate": 0.0004466,
121
+ "loss": 4.673929977416992,
122
+ "step": 320
123
+ },
124
+ {
125
+ "epoch": 0.022918773171553757,
126
+ "grad_norm": 1.0,
127
+ "learning_rate": 0.00047460000000000004,
128
+ "loss": 4.511159515380859,
129
+ "step": 340
130
+ },
131
+ {
132
+ "epoch": 0.024266936299292215,
133
+ "grad_norm": 0.96875,
134
+ "learning_rate": 0.0005026,
135
+ "loss": 4.360583114624023,
136
+ "step": 360
137
+ },
138
+ {
139
+ "epoch": 0.02561509942703067,
140
+ "grad_norm": 0.80078125,
141
+ "learning_rate": 0.0005306,
142
+ "loss": 4.208797836303711,
143
+ "step": 380
144
+ },
145
+ {
146
+ "epoch": 0.026963262554769128,
147
+ "grad_norm": 0.703125,
148
+ "learning_rate": 0.0005586,
149
+ "loss": 4.059257507324219,
150
+ "step": 400
151
+ },
152
+ {
153
+ "epoch": 0.028311425682507583,
154
+ "grad_norm": 1.390625,
155
+ "learning_rate": 0.0005866,
156
+ "loss": 3.945643997192383,
157
+ "step": 420
158
+ },
159
+ {
160
+ "epoch": 0.02965958881024604,
161
+ "grad_norm": 0.8046875,
162
+ "learning_rate": 0.0006146,
163
+ "loss": 3.878635787963867,
164
+ "step": 440
165
+ },
166
+ {
167
+ "epoch": 0.031007751937984496,
168
+ "grad_norm": 0.7578125,
169
+ "learning_rate": 0.0006426,
170
+ "loss": 3.763169860839844,
171
+ "step": 460
172
+ },
173
+ {
174
+ "epoch": 0.032355915065722954,
175
+ "grad_norm": 0.578125,
176
+ "learning_rate": 0.0006705999999999999,
177
+ "loss": 3.680609130859375,
178
+ "step": 480
179
+ },
180
+ {
181
+ "epoch": 0.03370407819346141,
182
+ "grad_norm": 0.6484375,
183
+ "learning_rate": 0.0006986,
184
+ "loss": 3.6085578918457033,
185
+ "step": 500
186
+ },
187
+ {
188
+ "epoch": 0.03505224132119986,
189
+ "grad_norm": 0.431640625,
190
+ "learning_rate": 0.0007,
191
+ "loss": 3.5324871063232424,
192
+ "step": 520
193
+ },
194
+ {
195
+ "epoch": 0.03640040444893832,
196
+ "grad_norm": 0.4453125,
197
+ "learning_rate": 0.0007,
198
+ "loss": 3.454126739501953,
199
+ "step": 540
200
+ },
201
+ {
202
+ "epoch": 0.03774856757667678,
203
+ "grad_norm": 0.546875,
204
+ "learning_rate": 0.0007,
205
+ "loss": 3.4041587829589846,
206
+ "step": 560
207
+ },
208
+ {
209
+ "epoch": 0.03909673070441524,
210
+ "grad_norm": 0.46875,
211
+ "learning_rate": 0.0007,
212
+ "loss": 3.347498321533203,
213
+ "step": 580
214
+ },
215
+ {
216
+ "epoch": 0.04044489383215369,
217
+ "grad_norm": 0.546875,
218
+ "learning_rate": 0.0007,
219
+ "loss": 3.313167190551758,
220
+ "step": 600
221
+ },
222
+ {
223
+ "epoch": 0.04179305695989215,
224
+ "grad_norm": 0.51953125,
225
+ "learning_rate": 0.0007,
226
+ "loss": 3.2490795135498045,
227
+ "step": 620
228
+ },
229
+ {
230
+ "epoch": 0.043141220087630605,
231
+ "grad_norm": 0.453125,
232
+ "learning_rate": 0.0007,
233
+ "loss": 3.221387481689453,
234
+ "step": 640
235
+ },
236
+ {
237
+ "epoch": 0.044489383215369056,
238
+ "grad_norm": 0.5234375,
239
+ "learning_rate": 0.0007,
240
+ "loss": 3.1599220275878905,
241
+ "step": 660
242
+ },
243
+ {
244
+ "epoch": 0.045837546343107514,
245
+ "grad_norm": 0.4375,
246
+ "learning_rate": 0.0007,
247
+ "loss": 3.133222198486328,
248
+ "step": 680
249
+ },
250
+ {
251
+ "epoch": 0.04718570947084597,
252
+ "grad_norm": 0.5703125,
253
+ "learning_rate": 0.0007,
254
+ "loss": 3.0884355545043944,
255
+ "step": 700
256
+ },
257
+ {
258
+ "epoch": 0.04853387259858443,
259
+ "grad_norm": 0.431640625,
260
+ "learning_rate": 0.0007,
261
+ "loss": 3.0560993194580077,
262
+ "step": 720
263
+ },
264
+ {
265
+ "epoch": 0.04988203572632288,
266
+ "grad_norm": 0.462890625,
267
+ "learning_rate": 0.0007,
268
+ "loss": 3.0169742584228514,
269
+ "step": 740
270
+ },
271
+ {
272
+ "epoch": 0.05123019885406134,
273
+ "grad_norm": 0.390625,
274
+ "learning_rate": 0.0007,
275
+ "loss": 2.9988849639892576,
276
+ "step": 760
277
+ },
278
+ {
279
+ "epoch": 0.0525783619817998,
280
+ "grad_norm": 0.466796875,
281
+ "learning_rate": 0.0007,
282
+ "loss": 3.004904937744141,
283
+ "step": 780
284
+ },
285
+ {
286
+ "epoch": 0.053926525109538256,
287
+ "grad_norm": 0.443359375,
288
+ "learning_rate": 0.0007,
289
+ "loss": 2.9500053405761717,
290
+ "step": 800
291
+ },
292
+ {
293
+ "epoch": 0.05527468823727671,
294
+ "grad_norm": 0.453125,
295
+ "learning_rate": 0.0007,
296
+ "loss": 2.93444938659668,
297
+ "step": 820
298
+ },
299
+ {
300
+ "epoch": 0.056622851365015166,
301
+ "grad_norm": 0.421875,
302
+ "learning_rate": 0.0007,
303
+ "loss": 2.9036865234375,
304
+ "step": 840
305
+ },
306
+ {
307
+ "epoch": 0.057971014492753624,
308
+ "grad_norm": 0.392578125,
309
+ "learning_rate": 0.0007,
310
+ "loss": 2.859415054321289,
311
+ "step": 860
312
+ },
313
+ {
314
+ "epoch": 0.05931917762049208,
315
+ "grad_norm": 0.42578125,
316
+ "learning_rate": 0.0007,
317
+ "loss": 2.8376497268676757,
318
+ "step": 880
319
+ },
320
+ {
321
+ "epoch": 0.06066734074823053,
322
+ "grad_norm": 0.46484375,
323
+ "learning_rate": 0.0007,
324
+ "loss": 2.821657180786133,
325
+ "step": 900
326
+ },
327
+ {
328
+ "epoch": 0.06201550387596899,
329
+ "grad_norm": 0.46875,
330
+ "learning_rate": 0.0007,
331
+ "loss": 2.7936939239501952,
332
+ "step": 920
333
+ },
334
+ {
335
+ "epoch": 0.06336366700370745,
336
+ "grad_norm": 0.40234375,
337
+ "learning_rate": 0.0007,
338
+ "loss": 2.779148292541504,
339
+ "step": 940
340
+ },
341
+ {
342
+ "epoch": 0.06471183013144591,
343
+ "grad_norm": 0.40234375,
344
+ "learning_rate": 0.0007,
345
+ "loss": 2.7523166656494142,
346
+ "step": 960
347
+ },
348
+ {
349
+ "epoch": 0.06605999325918437,
350
+ "grad_norm": 0.408203125,
351
+ "learning_rate": 0.0007,
352
+ "loss": 2.7326622009277344,
353
+ "step": 980
354
+ },
355
+ {
356
+ "epoch": 0.06740815638692282,
357
+ "grad_norm": 0.396484375,
358
+ "learning_rate": 0.0007,
359
+ "loss": 2.730000305175781,
360
+ "step": 1000
361
+ },
362
+ {
363
+ "epoch": 0.06875631951466127,
364
+ "grad_norm": 0.3828125,
365
+ "learning_rate": 0.0007,
366
+ "loss": 2.6984880447387694,
367
+ "step": 1020
368
+ },
369
+ {
370
+ "epoch": 0.07010448264239973,
371
+ "grad_norm": 0.423828125,
372
+ "learning_rate": 0.0007,
373
+ "loss": 2.6830041885375975,
374
+ "step": 1040
375
+ },
376
+ {
377
+ "epoch": 0.07145264577013818,
378
+ "grad_norm": 0.404296875,
379
+ "learning_rate": 0.0007,
380
+ "loss": 2.6666488647460938,
381
+ "step": 1060
382
+ },
383
+ {
384
+ "epoch": 0.07280080889787664,
385
+ "grad_norm": 0.404296875,
386
+ "learning_rate": 0.0007,
387
+ "loss": 2.6461860656738283,
388
+ "step": 1080
389
+ },
390
+ {
391
+ "epoch": 0.0741489720256151,
392
+ "grad_norm": 0.427734375,
393
+ "learning_rate": 0.0007,
394
+ "loss": 2.638811492919922,
395
+ "step": 1100
396
+ },
397
+ {
398
+ "epoch": 0.07549713515335356,
399
+ "grad_norm": 0.38671875,
400
+ "learning_rate": 0.0007,
401
+ "loss": 2.604878807067871,
402
+ "step": 1120
403
+ },
404
+ {
405
+ "epoch": 0.07684529828109202,
406
+ "grad_norm": 0.423828125,
407
+ "learning_rate": 0.0007,
408
+ "loss": 2.5887590408325196,
409
+ "step": 1140
410
+ },
411
+ {
412
+ "epoch": 0.07819346140883048,
413
+ "grad_norm": 0.416015625,
414
+ "learning_rate": 0.0007,
415
+ "loss": 2.5864500045776366,
416
+ "step": 1160
417
+ },
418
+ {
419
+ "epoch": 0.07954162453656892,
420
+ "grad_norm": 0.40234375,
421
+ "learning_rate": 0.0007,
422
+ "loss": 2.5788290023803713,
423
+ "step": 1180
424
+ },
425
+ {
426
+ "epoch": 0.08088978766430738,
427
+ "grad_norm": 0.404296875,
428
+ "learning_rate": 0.0007,
429
+ "loss": 2.570602607727051,
430
+ "step": 1200
431
+ },
432
+ {
433
+ "epoch": 0.08223795079204584,
434
+ "grad_norm": 0.390625,
435
+ "learning_rate": 0.0007,
436
+ "loss": 2.5528938293457033,
437
+ "step": 1220
438
+ },
439
+ {
440
+ "epoch": 0.0835861139197843,
441
+ "grad_norm": 0.373046875,
442
+ "learning_rate": 0.0007,
443
+ "loss": 2.53753662109375,
444
+ "step": 1240
445
+ },
446
+ {
447
+ "epoch": 0.08493427704752275,
448
+ "grad_norm": 0.384765625,
449
+ "learning_rate": 0.0007,
450
+ "loss": 2.523891830444336,
451
+ "step": 1260
452
+ },
453
+ {
454
+ "epoch": 0.08628244017526121,
455
+ "grad_norm": 0.3671875,
456
+ "learning_rate": 0.0007,
457
+ "loss": 2.5049261093139648,
458
+ "step": 1280
459
+ },
460
+ {
461
+ "epoch": 0.08763060330299967,
462
+ "grad_norm": 0.392578125,
463
+ "learning_rate": 0.0007,
464
+ "loss": 2.4961641311645506,
465
+ "step": 1300
466
+ },
467
+ {
468
+ "epoch": 0.08897876643073811,
469
+ "grad_norm": 0.353515625,
470
+ "learning_rate": 0.0007,
471
+ "loss": 2.487112045288086,
472
+ "step": 1320
473
+ },
474
+ {
475
+ "epoch": 0.09032692955847657,
476
+ "grad_norm": 0.38671875,
477
+ "learning_rate": 0.0007,
478
+ "loss": 2.471820259094238,
479
+ "step": 1340
480
+ },
481
+ {
482
+ "epoch": 0.09167509268621503,
483
+ "grad_norm": 0.392578125,
484
+ "learning_rate": 0.0007,
485
+ "loss": 2.4697391510009767,
486
+ "step": 1360
487
+ },
488
+ {
489
+ "epoch": 0.09302325581395349,
490
+ "grad_norm": 0.369140625,
491
+ "learning_rate": 0.0007,
492
+ "loss": 2.4492652893066404,
493
+ "step": 1380
494
+ },
495
+ {
496
+ "epoch": 0.09437141894169195,
497
+ "grad_norm": 0.384765625,
498
+ "learning_rate": 0.0007,
499
+ "loss": 2.4387527465820313,
500
+ "step": 1400
501
+ },
502
+ {
503
+ "epoch": 0.0957195820694304,
504
+ "grad_norm": 0.380859375,
505
+ "learning_rate": 0.0007,
506
+ "loss": 2.4479516983032226,
507
+ "step": 1420
508
+ },
509
+ {
510
+ "epoch": 0.09706774519716886,
511
+ "grad_norm": 0.359375,
512
+ "learning_rate": 0.0007,
513
+ "loss": 2.4246395111083983,
514
+ "step": 1440
515
+ },
516
+ {
517
+ "epoch": 0.09841590832490732,
518
+ "grad_norm": 0.373046875,
519
+ "learning_rate": 0.0007,
520
+ "loss": 2.4114765167236327,
521
+ "step": 1460
522
+ },
523
+ {
524
+ "epoch": 0.09976407145264576,
525
+ "grad_norm": 0.375,
526
+ "learning_rate": 0.0007,
527
+ "loss": 2.399285888671875,
528
+ "step": 1480
529
+ },
530
+ {
531
+ "epoch": 0.10111223458038422,
532
+ "grad_norm": 0.37890625,
533
+ "learning_rate": 0.0007,
534
+ "loss": 2.3839179992675783,
535
+ "step": 1500
536
+ },
537
+ {
538
+ "epoch": 0.10246039770812268,
539
+ "grad_norm": 0.41796875,
540
+ "learning_rate": 0.0007,
541
+ "loss": 2.3861133575439455,
542
+ "step": 1520
543
+ },
544
+ {
545
+ "epoch": 0.10380856083586114,
546
+ "grad_norm": 0.40625,
547
+ "learning_rate": 0.0007,
548
+ "loss": 2.3732574462890623,
549
+ "step": 1540
550
+ },
551
+ {
552
+ "epoch": 0.1051567239635996,
553
+ "grad_norm": 0.36328125,
554
+ "learning_rate": 0.0007,
555
+ "loss": 2.3624162673950195,
556
+ "step": 1560
557
+ },
558
+ {
559
+ "epoch": 0.10650488709133805,
560
+ "grad_norm": 0.359375,
561
+ "learning_rate": 0.0007,
562
+ "loss": 2.3779842376708986,
563
+ "step": 1580
564
+ },
565
+ {
566
+ "epoch": 0.10785305021907651,
567
+ "grad_norm": 0.369140625,
568
+ "learning_rate": 0.0007,
569
+ "loss": 2.3441116333007814,
570
+ "step": 1600
571
+ },
572
+ {
573
+ "epoch": 0.10920121334681497,
574
+ "grad_norm": 0.376953125,
575
+ "learning_rate": 0.0007,
576
+ "loss": 2.3418384552001954,
577
+ "step": 1620
578
+ },
579
+ {
580
+ "epoch": 0.11054937647455342,
581
+ "grad_norm": 0.373046875,
582
+ "learning_rate": 0.0007,
583
+ "loss": 2.3539493560791014,
584
+ "step": 1640
585
+ },
586
+ {
587
+ "epoch": 0.11189753960229187,
588
+ "grad_norm": 0.359375,
589
+ "learning_rate": 0.0007,
590
+ "loss": 2.332906150817871,
591
+ "step": 1660
592
+ },
593
+ {
594
+ "epoch": 0.11324570273003033,
595
+ "grad_norm": 0.408203125,
596
+ "learning_rate": 0.0007,
597
+ "loss": 2.3174556732177733,
598
+ "step": 1680
599
+ },
600
+ {
601
+ "epoch": 0.11459386585776879,
602
+ "grad_norm": 0.361328125,
603
+ "learning_rate": 0.0007,
604
+ "loss": 2.330640411376953,
605
+ "step": 1700
606
+ },
607
+ {
608
+ "epoch": 0.11594202898550725,
609
+ "grad_norm": 0.423828125,
610
+ "learning_rate": 0.0007,
611
+ "loss": 2.306385803222656,
612
+ "step": 1720
613
+ },
614
+ {
615
+ "epoch": 0.1172901921132457,
616
+ "grad_norm": 0.365234375,
617
+ "learning_rate": 0.0007,
618
+ "loss": 2.2937660217285156,
619
+ "step": 1740
620
+ },
621
+ {
622
+ "epoch": 0.11863835524098416,
623
+ "grad_norm": 0.3671875,
624
+ "learning_rate": 0.0007,
625
+ "loss": 2.297491264343262,
626
+ "step": 1760
627
+ },
628
+ {
629
+ "epoch": 0.11998651836872262,
630
+ "grad_norm": 0.337890625,
631
+ "learning_rate": 0.0007,
632
+ "loss": 2.312083053588867,
633
+ "step": 1780
634
+ },
635
+ {
636
+ "epoch": 0.12133468149646107,
637
+ "grad_norm": 0.357421875,
638
+ "learning_rate": 0.0007,
639
+ "loss": 2.307674789428711,
640
+ "step": 1800
641
+ },
642
+ {
643
+ "epoch": 0.12268284462419952,
644
+ "grad_norm": 0.375,
645
+ "learning_rate": 0.0007,
646
+ "loss": 2.296724319458008,
647
+ "step": 1820
648
+ },
649
+ {
650
+ "epoch": 0.12403100775193798,
651
+ "grad_norm": 0.376953125,
652
+ "learning_rate": 0.0007,
653
+ "loss": 2.2644861221313475,
654
+ "step": 1840
655
+ },
656
+ {
657
+ "epoch": 0.12537917087967643,
658
+ "grad_norm": 0.353515625,
659
+ "learning_rate": 0.0007,
660
+ "loss": 2.2662803649902346,
661
+ "step": 1860
662
+ },
663
+ {
664
+ "epoch": 0.1267273340074149,
665
+ "grad_norm": 0.33984375,
666
+ "learning_rate": 0.0007,
667
+ "loss": 2.2558313369750977,
668
+ "step": 1880
669
+ },
670
+ {
671
+ "epoch": 0.12807549713515334,
672
+ "grad_norm": 0.365234375,
673
+ "learning_rate": 0.0007,
674
+ "loss": 2.262881851196289,
675
+ "step": 1900
676
+ },
677
+ {
678
+ "epoch": 0.12942366026289182,
679
+ "grad_norm": 0.349609375,
680
+ "learning_rate": 0.0007,
681
+ "loss": 2.2384838104248046,
682
+ "step": 1920
683
+ },
684
+ {
685
+ "epoch": 0.13077182339063026,
686
+ "grad_norm": 0.361328125,
687
+ "learning_rate": 0.0007,
688
+ "loss": 2.247391700744629,
689
+ "step": 1940
690
+ },
691
+ {
692
+ "epoch": 0.13211998651836873,
693
+ "grad_norm": 0.34765625,
694
+ "learning_rate": 0.0007,
695
+ "loss": 2.2225168228149412,
696
+ "step": 1960
697
+ },
698
+ {
699
+ "epoch": 0.13346814964610718,
700
+ "grad_norm": 0.380859375,
701
+ "learning_rate": 0.0007,
702
+ "loss": 2.2470256805419924,
703
+ "step": 1980
704
+ },
705
+ {
706
+ "epoch": 0.13481631277384565,
707
+ "grad_norm": 0.3359375,
708
+ "learning_rate": 0.0007,
709
+ "loss": 2.218878173828125,
710
+ "step": 2000
711
+ },
712
+ {
713
+ "epoch": 0.1361644759015841,
714
+ "grad_norm": 0.357421875,
715
+ "learning_rate": 0.0007,
716
+ "loss": 2.227268409729004,
717
+ "step": 2020
718
+ },
719
+ {
720
+ "epoch": 0.13751263902932254,
721
+ "grad_norm": 0.380859375,
722
+ "learning_rate": 0.0007,
723
+ "loss": 2.2079793930053713,
724
+ "step": 2040
725
+ },
726
+ {
727
+ "epoch": 0.138860802157061,
728
+ "grad_norm": 0.380859375,
729
+ "learning_rate": 0.0007,
730
+ "loss": 2.2262474060058595,
731
+ "step": 2060
732
+ },
733
+ {
734
+ "epoch": 0.14020896528479945,
735
+ "grad_norm": 0.349609375,
736
+ "learning_rate": 0.0007,
737
+ "loss": 2.2017698287963867,
738
+ "step": 2080
739
+ },
740
+ {
741
+ "epoch": 0.14155712841253792,
742
+ "grad_norm": 0.333984375,
743
+ "learning_rate": 0.0007,
744
+ "loss": 2.19287223815918,
745
+ "step": 2100
746
+ },
747
+ {
748
+ "epoch": 0.14290529154027637,
749
+ "grad_norm": 0.349609375,
750
+ "learning_rate": 0.0007,
751
+ "loss": 2.201098823547363,
752
+ "step": 2120
753
+ },
754
+ {
755
+ "epoch": 0.14425345466801484,
756
+ "grad_norm": 0.333984375,
757
+ "learning_rate": 0.0007,
758
+ "loss": 2.2028568267822264,
759
+ "step": 2140
760
+ },
761
+ {
762
+ "epoch": 0.14560161779575329,
763
+ "grad_norm": 0.353515625,
764
+ "learning_rate": 0.0007,
765
+ "loss": 2.1869789123535157,
766
+ "step": 2160
767
+ },
768
+ {
769
+ "epoch": 0.14694978092349173,
770
+ "grad_norm": 0.33984375,
771
+ "learning_rate": 0.0007,
772
+ "loss": 2.194090461730957,
773
+ "step": 2180
774
+ },
775
+ {
776
+ "epoch": 0.1482979440512302,
777
+ "grad_norm": 0.333984375,
778
+ "learning_rate": 0.0007,
779
+ "loss": 2.169504165649414,
780
+ "step": 2200
781
+ },
782
+ {
783
+ "epoch": 0.14964610717896865,
784
+ "grad_norm": 0.359375,
785
+ "learning_rate": 0.0007,
786
+ "loss": 2.17819766998291,
787
+ "step": 2220
788
+ },
789
+ {
790
+ "epoch": 0.15099427030670712,
791
+ "grad_norm": 0.361328125,
792
+ "learning_rate": 0.0007,
793
+ "loss": 2.1870670318603516,
794
+ "step": 2240
795
+ },
796
+ {
797
+ "epoch": 0.15234243343444556,
798
+ "grad_norm": 0.359375,
799
+ "learning_rate": 0.0007,
800
+ "loss": 2.1656898498535155,
801
+ "step": 2260
802
+ },
803
+ {
804
+ "epoch": 0.15369059656218403,
805
+ "grad_norm": 0.369140625,
806
+ "learning_rate": 0.0007,
807
+ "loss": 2.174662971496582,
808
+ "step": 2280
809
+ },
810
+ {
811
+ "epoch": 0.15503875968992248,
812
+ "grad_norm": 0.3515625,
813
+ "learning_rate": 0.0007,
814
+ "loss": 2.1561347961425783,
815
+ "step": 2300
816
+ },
817
+ {
818
+ "epoch": 0.15638692281766095,
819
+ "grad_norm": 0.33203125,
820
+ "learning_rate": 0.0007,
821
+ "loss": 2.16455020904541,
822
+ "step": 2320
823
+ },
824
+ {
825
+ "epoch": 0.1577350859453994,
826
+ "grad_norm": 0.330078125,
827
+ "learning_rate": 0.0007,
828
+ "loss": 2.1551021575927733,
829
+ "step": 2340
830
+ },
831
+ {
832
+ "epoch": 0.15908324907313784,
833
+ "grad_norm": 0.341796875,
834
+ "learning_rate": 0.0007,
835
+ "loss": 2.157507133483887,
836
+ "step": 2360
837
+ },
838
+ {
839
+ "epoch": 0.1604314122008763,
840
+ "grad_norm": 0.33984375,
841
+ "learning_rate": 0.0007,
842
+ "loss": 2.1596670150756836,
843
+ "step": 2380
844
+ },
845
+ {
846
+ "epoch": 0.16177957532861476,
847
+ "grad_norm": 0.357421875,
848
+ "learning_rate": 0.0007,
849
+ "loss": 2.1417112350463867,
850
+ "step": 2400
851
+ },
852
+ {
853
+ "epoch": 0.16312773845635323,
854
+ "grad_norm": 0.349609375,
855
+ "learning_rate": 0.0007,
856
+ "loss": 2.1463193893432617,
857
+ "step": 2420
858
+ },
859
+ {
860
+ "epoch": 0.16447590158409167,
861
+ "grad_norm": 0.3359375,
862
+ "learning_rate": 0.0007,
863
+ "loss": 2.133936882019043,
864
+ "step": 2440
865
+ },
866
+ {
867
+ "epoch": 0.16582406471183014,
868
+ "grad_norm": 0.328125,
869
+ "learning_rate": 0.0007,
870
+ "loss": 2.1435129165649416,
871
+ "step": 2460
872
+ },
873
+ {
874
+ "epoch": 0.1671722278395686,
875
+ "grad_norm": 0.341796875,
876
+ "learning_rate": 0.0007,
877
+ "loss": 2.1443099975585938,
878
+ "step": 2480
879
+ },
880
+ {
881
+ "epoch": 0.1683855746545332,
882
+ "eval_loss": 2.138087511062622,
883
+ "eval_runtime": 20.1123,
884
+ "eval_samples_per_second": 473.691,
885
+ "eval_steps_per_second": 0.497,
886
+ "step": 2498
887
+ },
888
+ {
889
+ "epoch": 0.16852039096730703,
890
+ "grad_norm": 0.33984375,
891
+ "learning_rate": 0.0007,
892
+ "loss": 2.126860427856445,
893
+ "step": 2500
894
+ },
895
+ {
896
+ "epoch": 0.16852039096730703,
897
+ "eval_loss": 2.1375532150268555,
898
+ "eval_runtime": 20.4047,
899
+ "eval_samples_per_second": 466.903,
900
+ "eval_steps_per_second": 0.49,
901
+ "step": 2500
902
+ },
903
+ {
904
+ "epoch": 0.1698685540950455,
905
+ "grad_norm": 24903680.0,
906
+ "learning_rate": 3.4200000000000005e-05,
907
+ "loss": 2.128565788269043,
908
+ "step": 2520
909
+ },
910
+ {
911
+ "epoch": 0.17121671722278395,
912
+ "grad_norm": 27000832.0,
913
+ "learning_rate": 7.020000000000001e-05,
914
+ "loss": 2.1492637634277343,
915
+ "step": 2540
916
+ },
917
+ {
918
+ "epoch": 0.17256488035052242,
919
+ "grad_norm": 119537664.0,
920
+ "learning_rate": 9e-05,
921
+ "loss": 3.623552703857422,
922
+ "step": 2560
923
+ },
924
+ {
925
+ "epoch": 0.17391304347826086,
926
+ "grad_norm": 301989888.0,
927
+ "learning_rate": 9e-05,
928
+ "loss": 6.8911598205566404,
929
+ "step": 2580
930
+ },
931
+ {
932
+ "epoch": 0.17526120660599934,
933
+ "grad_norm": 658505728.0,
934
+ "learning_rate": 9e-05,
935
+ "loss": 7.545132446289062,
936
+ "step": 2600
937
+ },
938
+ {
939
+ "epoch": 0.17660936973373778,
940
+ "grad_norm": 1166016512.0,
941
+ "learning_rate": 9e-05,
942
+ "loss": 7.784053039550781,
943
+ "step": 2620
944
+ },
945
+ {
946
+ "epoch": 0.17795753286147623,
947
+ "grad_norm": 1769996288.0,
948
+ "learning_rate": 9e-05,
949
+ "loss": 7.88036880493164,
950
+ "step": 2640
951
+ },
952
+ {
953
+ "epoch": 0.1793056959892147,
954
+ "grad_norm": 2415919104.0,
955
+ "learning_rate": 9e-05,
956
+ "loss": 7.926078033447266,
957
+ "step": 2660
958
+ },
959
+ {
960
+ "epoch": 0.18065385911695314,
961
+ "grad_norm": 3053453312.0,
962
+ "learning_rate": 9e-05,
963
+ "loss": 7.947563934326172,
964
+ "step": 2680
965
+ },
966
+ {
967
+ "epoch": 0.1820020222446916,
968
+ "grad_norm": 3674210304.0,
969
+ "learning_rate": 9e-05,
970
+ "loss": 7.972737121582031,
971
+ "step": 2700
972
+ }
973
+ ],
974
+ "logging_steps": 20,
975
+ "max_steps": 3500,
976
+ "num_input_tokens_seen": 0,
977
+ "num_train_epochs": 1,
978
+ "save_steps": 100,
979
+ "stateful_callbacks": {
980
+ "TrainerControl": {
981
+ "args": {
982
+ "should_epoch_stop": false,
983
+ "should_evaluate": false,
984
+ "should_log": false,
985
+ "should_save": true,
986
+ "should_training_stop": false
987
+ },
988
+ "attributes": {}
989
+ }
990
+ },
991
+ "total_flos": 8710966267084800.0,
992
+ "train_batch_size": 64,
993
+ "trial_name": null,
994
+ "trial_params": null
995
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2700/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f62f8186f489a47bc68f80ec9f826cd67dc6a0232e67ff9d099ed1157b6e25f
3
+ size 4920
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/config.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation": "powlu",
3
+ "architectures": [
4
+ "TinyLlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 1,
9
+ "dtype": "bfloat16",
10
+ "eos_token_id": 2,
11
+ "head_dim": 32,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 128,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 256,
16
+ "max_position_embeddings": 512,
17
+ "mlp_bias": false,
18
+ "mlp_type": "glu",
19
+ "model_type": "tiny_llama",
20
+ "num_attention_heads": 4,
21
+ "num_hidden_layers": 100,
22
+ "num_key_value_heads": 4,
23
+ "pad_token_id": 0,
24
+ "powlu_m": 3.0,
25
+ "pretraining_tp": 1,
26
+ "rms_norm_eps": 1e-06,
27
+ "rope_parameters": {
28
+ "rope_theta": 10000.0,
29
+ "rope_type": "default"
30
+ },
31
+ "tie_word_embeddings": true,
32
+ "tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
33
+ "transformers_version": "5.16.0.dev0",
34
+ "use_cache": false,
35
+ "vocab_size": 4096,
36
+ "waleed_beta": 10
37
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30014e9d74722ab208e3c1b699d30c6487ec73bfdce2029823038595f23ad48f
3
+ size 33967272
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0bd284233d7ca1f7758977504f5f085bb7c2305fa9934946c690ddcc158d48b0
3
+ size 39573050
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cf9097d4513154245c48236b6ec5137b7ee2a21c9f58f2cba798ea275c6026f
3
+ size 14244
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3524231f4a5ff6c2e972dce51861a8bc8472013528f64695abe2c7eea8f6783e
3
+ size 1064
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/tokenizer_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "backend": "tokenizers",
4
+ "bos_token": "<|endoftext|>",
5
+ "eos_token": "<|endoftext|>",
6
+ "errors": "replace",
7
+ "is_local": false,
8
+ "local_files_only": false,
9
+ "model_max_length": 1024,
10
+ "pad_token": "<|endoftext|>",
11
+ "tokenizer_class": "GPT2Tokenizer",
12
+ "unk_token": "<|endoftext|>"
13
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/trainer_state.json ADDED
@@ -0,0 +1,1030 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.1887428378833839,
6
+ "eval_steps": 2498,
7
+ "global_step": 2800,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.0013481631277384564,
14
+ "grad_norm": 2.09375,
15
+ "learning_rate": 2.66e-05,
16
+ "loss": 8.300566864013671,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.002696326255476913,
21
+ "grad_norm": 1.375,
22
+ "learning_rate": 5.46e-05,
23
+ "loss": 8.056343841552735,
24
+ "step": 40
25
+ },
26
+ {
27
+ "epoch": 0.004044489383215369,
28
+ "grad_norm": 1.2734375,
29
+ "learning_rate": 8.259999999999999e-05,
30
+ "loss": 7.841481781005859,
31
+ "step": 60
32
+ },
33
+ {
34
+ "epoch": 0.005392652510953826,
35
+ "grad_norm": 1.265625,
36
+ "learning_rate": 0.0001106,
37
+ "loss": 7.57496337890625,
38
+ "step": 80
39
+ },
40
+ {
41
+ "epoch": 0.006740815638692282,
42
+ "grad_norm": 1.1953125,
43
+ "learning_rate": 0.0001386,
44
+ "loss": 7.233026123046875,
45
+ "step": 100
46
+ },
47
+ {
48
+ "epoch": 0.008088978766430738,
49
+ "grad_norm": 1.1328125,
50
+ "learning_rate": 0.00016659999999999998,
51
+ "loss": 6.859487152099609,
52
+ "step": 120
53
+ },
54
+ {
55
+ "epoch": 0.009437141894169195,
56
+ "grad_norm": 0.94921875,
57
+ "learning_rate": 0.00019460000000000001,
58
+ "loss": 6.5162200927734375,
59
+ "step": 140
60
+ },
61
+ {
62
+ "epoch": 0.010785305021907651,
63
+ "grad_norm": 0.796875,
64
+ "learning_rate": 0.0002226,
65
+ "loss": 6.237432861328125,
66
+ "step": 160
67
+ },
68
+ {
69
+ "epoch": 0.012133468149646108,
70
+ "grad_norm": 0.71875,
71
+ "learning_rate": 0.00025059999999999997,
72
+ "loss": 6.039522552490235,
73
+ "step": 180
74
+ },
75
+ {
76
+ "epoch": 0.013481631277384564,
77
+ "grad_norm": 0.57421875,
78
+ "learning_rate": 0.0002786,
79
+ "loss": 5.8798988342285154,
80
+ "step": 200
81
+ },
82
+ {
83
+ "epoch": 0.01482979440512302,
84
+ "grad_norm": 1.90625,
85
+ "learning_rate": 0.00030659999999999997,
86
+ "loss": 5.723947143554687,
87
+ "step": 220
88
+ },
89
+ {
90
+ "epoch": 0.016177957532861477,
91
+ "grad_norm": 0.6796875,
92
+ "learning_rate": 0.0003346,
93
+ "loss": 5.551593399047851,
94
+ "step": 240
95
+ },
96
+ {
97
+ "epoch": 0.01752612066059993,
98
+ "grad_norm": 1.6875,
99
+ "learning_rate": 0.00036260000000000003,
100
+ "loss": 5.3516895294189455,
101
+ "step": 260
102
+ },
103
+ {
104
+ "epoch": 0.01887428378833839,
105
+ "grad_norm": 0.84765625,
106
+ "learning_rate": 0.0003906,
107
+ "loss": 5.130373764038086,
108
+ "step": 280
109
+ },
110
+ {
111
+ "epoch": 0.020222446916076844,
112
+ "grad_norm": 0.55859375,
113
+ "learning_rate": 0.0004186,
114
+ "loss": 4.882920455932617,
115
+ "step": 300
116
+ },
117
+ {
118
+ "epoch": 0.021570610043815303,
119
+ "grad_norm": 1.15625,
120
+ "learning_rate": 0.0004466,
121
+ "loss": 4.673929977416992,
122
+ "step": 320
123
+ },
124
+ {
125
+ "epoch": 0.022918773171553757,
126
+ "grad_norm": 1.0,
127
+ "learning_rate": 0.00047460000000000004,
128
+ "loss": 4.511159515380859,
129
+ "step": 340
130
+ },
131
+ {
132
+ "epoch": 0.024266936299292215,
133
+ "grad_norm": 0.96875,
134
+ "learning_rate": 0.0005026,
135
+ "loss": 4.360583114624023,
136
+ "step": 360
137
+ },
138
+ {
139
+ "epoch": 0.02561509942703067,
140
+ "grad_norm": 0.80078125,
141
+ "learning_rate": 0.0005306,
142
+ "loss": 4.208797836303711,
143
+ "step": 380
144
+ },
145
+ {
146
+ "epoch": 0.026963262554769128,
147
+ "grad_norm": 0.703125,
148
+ "learning_rate": 0.0005586,
149
+ "loss": 4.059257507324219,
150
+ "step": 400
151
+ },
152
+ {
153
+ "epoch": 0.028311425682507583,
154
+ "grad_norm": 1.390625,
155
+ "learning_rate": 0.0005866,
156
+ "loss": 3.945643997192383,
157
+ "step": 420
158
+ },
159
+ {
160
+ "epoch": 0.02965958881024604,
161
+ "grad_norm": 0.8046875,
162
+ "learning_rate": 0.0006146,
163
+ "loss": 3.878635787963867,
164
+ "step": 440
165
+ },
166
+ {
167
+ "epoch": 0.031007751937984496,
168
+ "grad_norm": 0.7578125,
169
+ "learning_rate": 0.0006426,
170
+ "loss": 3.763169860839844,
171
+ "step": 460
172
+ },
173
+ {
174
+ "epoch": 0.032355915065722954,
175
+ "grad_norm": 0.578125,
176
+ "learning_rate": 0.0006705999999999999,
177
+ "loss": 3.680609130859375,
178
+ "step": 480
179
+ },
180
+ {
181
+ "epoch": 0.03370407819346141,
182
+ "grad_norm": 0.6484375,
183
+ "learning_rate": 0.0006986,
184
+ "loss": 3.6085578918457033,
185
+ "step": 500
186
+ },
187
+ {
188
+ "epoch": 0.03505224132119986,
189
+ "grad_norm": 0.431640625,
190
+ "learning_rate": 0.0007,
191
+ "loss": 3.5324871063232424,
192
+ "step": 520
193
+ },
194
+ {
195
+ "epoch": 0.03640040444893832,
196
+ "grad_norm": 0.4453125,
197
+ "learning_rate": 0.0007,
198
+ "loss": 3.454126739501953,
199
+ "step": 540
200
+ },
201
+ {
202
+ "epoch": 0.03774856757667678,
203
+ "grad_norm": 0.546875,
204
+ "learning_rate": 0.0007,
205
+ "loss": 3.4041587829589846,
206
+ "step": 560
207
+ },
208
+ {
209
+ "epoch": 0.03909673070441524,
210
+ "grad_norm": 0.46875,
211
+ "learning_rate": 0.0007,
212
+ "loss": 3.347498321533203,
213
+ "step": 580
214
+ },
215
+ {
216
+ "epoch": 0.04044489383215369,
217
+ "grad_norm": 0.546875,
218
+ "learning_rate": 0.0007,
219
+ "loss": 3.313167190551758,
220
+ "step": 600
221
+ },
222
+ {
223
+ "epoch": 0.04179305695989215,
224
+ "grad_norm": 0.51953125,
225
+ "learning_rate": 0.0007,
226
+ "loss": 3.2490795135498045,
227
+ "step": 620
228
+ },
229
+ {
230
+ "epoch": 0.043141220087630605,
231
+ "grad_norm": 0.453125,
232
+ "learning_rate": 0.0007,
233
+ "loss": 3.221387481689453,
234
+ "step": 640
235
+ },
236
+ {
237
+ "epoch": 0.044489383215369056,
238
+ "grad_norm": 0.5234375,
239
+ "learning_rate": 0.0007,
240
+ "loss": 3.1599220275878905,
241
+ "step": 660
242
+ },
243
+ {
244
+ "epoch": 0.045837546343107514,
245
+ "grad_norm": 0.4375,
246
+ "learning_rate": 0.0007,
247
+ "loss": 3.133222198486328,
248
+ "step": 680
249
+ },
250
+ {
251
+ "epoch": 0.04718570947084597,
252
+ "grad_norm": 0.5703125,
253
+ "learning_rate": 0.0007,
254
+ "loss": 3.0884355545043944,
255
+ "step": 700
256
+ },
257
+ {
258
+ "epoch": 0.04853387259858443,
259
+ "grad_norm": 0.431640625,
260
+ "learning_rate": 0.0007,
261
+ "loss": 3.0560993194580077,
262
+ "step": 720
263
+ },
264
+ {
265
+ "epoch": 0.04988203572632288,
266
+ "grad_norm": 0.462890625,
267
+ "learning_rate": 0.0007,
268
+ "loss": 3.0169742584228514,
269
+ "step": 740
270
+ },
271
+ {
272
+ "epoch": 0.05123019885406134,
273
+ "grad_norm": 0.390625,
274
+ "learning_rate": 0.0007,
275
+ "loss": 2.9988849639892576,
276
+ "step": 760
277
+ },
278
+ {
279
+ "epoch": 0.0525783619817998,
280
+ "grad_norm": 0.466796875,
281
+ "learning_rate": 0.0007,
282
+ "loss": 3.004904937744141,
283
+ "step": 780
284
+ },
285
+ {
286
+ "epoch": 0.053926525109538256,
287
+ "grad_norm": 0.443359375,
288
+ "learning_rate": 0.0007,
289
+ "loss": 2.9500053405761717,
290
+ "step": 800
291
+ },
292
+ {
293
+ "epoch": 0.05527468823727671,
294
+ "grad_norm": 0.453125,
295
+ "learning_rate": 0.0007,
296
+ "loss": 2.93444938659668,
297
+ "step": 820
298
+ },
299
+ {
300
+ "epoch": 0.056622851365015166,
301
+ "grad_norm": 0.421875,
302
+ "learning_rate": 0.0007,
303
+ "loss": 2.9036865234375,
304
+ "step": 840
305
+ },
306
+ {
307
+ "epoch": 0.057971014492753624,
308
+ "grad_norm": 0.392578125,
309
+ "learning_rate": 0.0007,
310
+ "loss": 2.859415054321289,
311
+ "step": 860
312
+ },
313
+ {
314
+ "epoch": 0.05931917762049208,
315
+ "grad_norm": 0.42578125,
316
+ "learning_rate": 0.0007,
317
+ "loss": 2.8376497268676757,
318
+ "step": 880
319
+ },
320
+ {
321
+ "epoch": 0.06066734074823053,
322
+ "grad_norm": 0.46484375,
323
+ "learning_rate": 0.0007,
324
+ "loss": 2.821657180786133,
325
+ "step": 900
326
+ },
327
+ {
328
+ "epoch": 0.06201550387596899,
329
+ "grad_norm": 0.46875,
330
+ "learning_rate": 0.0007,
331
+ "loss": 2.7936939239501952,
332
+ "step": 920
333
+ },
334
+ {
335
+ "epoch": 0.06336366700370745,
336
+ "grad_norm": 0.40234375,
337
+ "learning_rate": 0.0007,
338
+ "loss": 2.779148292541504,
339
+ "step": 940
340
+ },
341
+ {
342
+ "epoch": 0.06471183013144591,
343
+ "grad_norm": 0.40234375,
344
+ "learning_rate": 0.0007,
345
+ "loss": 2.7523166656494142,
346
+ "step": 960
347
+ },
348
+ {
349
+ "epoch": 0.06605999325918437,
350
+ "grad_norm": 0.408203125,
351
+ "learning_rate": 0.0007,
352
+ "loss": 2.7326622009277344,
353
+ "step": 980
354
+ },
355
+ {
356
+ "epoch": 0.06740815638692282,
357
+ "grad_norm": 0.396484375,
358
+ "learning_rate": 0.0007,
359
+ "loss": 2.730000305175781,
360
+ "step": 1000
361
+ },
362
+ {
363
+ "epoch": 0.06875631951466127,
364
+ "grad_norm": 0.3828125,
365
+ "learning_rate": 0.0007,
366
+ "loss": 2.6984880447387694,
367
+ "step": 1020
368
+ },
369
+ {
370
+ "epoch": 0.07010448264239973,
371
+ "grad_norm": 0.423828125,
372
+ "learning_rate": 0.0007,
373
+ "loss": 2.6830041885375975,
374
+ "step": 1040
375
+ },
376
+ {
377
+ "epoch": 0.07145264577013818,
378
+ "grad_norm": 0.404296875,
379
+ "learning_rate": 0.0007,
380
+ "loss": 2.6666488647460938,
381
+ "step": 1060
382
+ },
383
+ {
384
+ "epoch": 0.07280080889787664,
385
+ "grad_norm": 0.404296875,
386
+ "learning_rate": 0.0007,
387
+ "loss": 2.6461860656738283,
388
+ "step": 1080
389
+ },
390
+ {
391
+ "epoch": 0.0741489720256151,
392
+ "grad_norm": 0.427734375,
393
+ "learning_rate": 0.0007,
394
+ "loss": 2.638811492919922,
395
+ "step": 1100
396
+ },
397
+ {
398
+ "epoch": 0.07549713515335356,
399
+ "grad_norm": 0.38671875,
400
+ "learning_rate": 0.0007,
401
+ "loss": 2.604878807067871,
402
+ "step": 1120
403
+ },
404
+ {
405
+ "epoch": 0.07684529828109202,
406
+ "grad_norm": 0.423828125,
407
+ "learning_rate": 0.0007,
408
+ "loss": 2.5887590408325196,
409
+ "step": 1140
410
+ },
411
+ {
412
+ "epoch": 0.07819346140883048,
413
+ "grad_norm": 0.416015625,
414
+ "learning_rate": 0.0007,
415
+ "loss": 2.5864500045776366,
416
+ "step": 1160
417
+ },
418
+ {
419
+ "epoch": 0.07954162453656892,
420
+ "grad_norm": 0.40234375,
421
+ "learning_rate": 0.0007,
422
+ "loss": 2.5788290023803713,
423
+ "step": 1180
424
+ },
425
+ {
426
+ "epoch": 0.08088978766430738,
427
+ "grad_norm": 0.404296875,
428
+ "learning_rate": 0.0007,
429
+ "loss": 2.570602607727051,
430
+ "step": 1200
431
+ },
432
+ {
433
+ "epoch": 0.08223795079204584,
434
+ "grad_norm": 0.390625,
435
+ "learning_rate": 0.0007,
436
+ "loss": 2.5528938293457033,
437
+ "step": 1220
438
+ },
439
+ {
440
+ "epoch": 0.0835861139197843,
441
+ "grad_norm": 0.373046875,
442
+ "learning_rate": 0.0007,
443
+ "loss": 2.53753662109375,
444
+ "step": 1240
445
+ },
446
+ {
447
+ "epoch": 0.08493427704752275,
448
+ "grad_norm": 0.384765625,
449
+ "learning_rate": 0.0007,
450
+ "loss": 2.523891830444336,
451
+ "step": 1260
452
+ },
453
+ {
454
+ "epoch": 0.08628244017526121,
455
+ "grad_norm": 0.3671875,
456
+ "learning_rate": 0.0007,
457
+ "loss": 2.5049261093139648,
458
+ "step": 1280
459
+ },
460
+ {
461
+ "epoch": 0.08763060330299967,
462
+ "grad_norm": 0.392578125,
463
+ "learning_rate": 0.0007,
464
+ "loss": 2.4961641311645506,
465
+ "step": 1300
466
+ },
467
+ {
468
+ "epoch": 0.08897876643073811,
469
+ "grad_norm": 0.353515625,
470
+ "learning_rate": 0.0007,
471
+ "loss": 2.487112045288086,
472
+ "step": 1320
473
+ },
474
+ {
475
+ "epoch": 0.09032692955847657,
476
+ "grad_norm": 0.38671875,
477
+ "learning_rate": 0.0007,
478
+ "loss": 2.471820259094238,
479
+ "step": 1340
480
+ },
481
+ {
482
+ "epoch": 0.09167509268621503,
483
+ "grad_norm": 0.392578125,
484
+ "learning_rate": 0.0007,
485
+ "loss": 2.4697391510009767,
486
+ "step": 1360
487
+ },
488
+ {
489
+ "epoch": 0.09302325581395349,
490
+ "grad_norm": 0.369140625,
491
+ "learning_rate": 0.0007,
492
+ "loss": 2.4492652893066404,
493
+ "step": 1380
494
+ },
495
+ {
496
+ "epoch": 0.09437141894169195,
497
+ "grad_norm": 0.384765625,
498
+ "learning_rate": 0.0007,
499
+ "loss": 2.4387527465820313,
500
+ "step": 1400
501
+ },
502
+ {
503
+ "epoch": 0.0957195820694304,
504
+ "grad_norm": 0.380859375,
505
+ "learning_rate": 0.0007,
506
+ "loss": 2.4479516983032226,
507
+ "step": 1420
508
+ },
509
+ {
510
+ "epoch": 0.09706774519716886,
511
+ "grad_norm": 0.359375,
512
+ "learning_rate": 0.0007,
513
+ "loss": 2.4246395111083983,
514
+ "step": 1440
515
+ },
516
+ {
517
+ "epoch": 0.09841590832490732,
518
+ "grad_norm": 0.373046875,
519
+ "learning_rate": 0.0007,
520
+ "loss": 2.4114765167236327,
521
+ "step": 1460
522
+ },
523
+ {
524
+ "epoch": 0.09976407145264576,
525
+ "grad_norm": 0.375,
526
+ "learning_rate": 0.0007,
527
+ "loss": 2.399285888671875,
528
+ "step": 1480
529
+ },
530
+ {
531
+ "epoch": 0.10111223458038422,
532
+ "grad_norm": 0.37890625,
533
+ "learning_rate": 0.0007,
534
+ "loss": 2.3839179992675783,
535
+ "step": 1500
536
+ },
537
+ {
538
+ "epoch": 0.10246039770812268,
539
+ "grad_norm": 0.41796875,
540
+ "learning_rate": 0.0007,
541
+ "loss": 2.3861133575439455,
542
+ "step": 1520
543
+ },
544
+ {
545
+ "epoch": 0.10380856083586114,
546
+ "grad_norm": 0.40625,
547
+ "learning_rate": 0.0007,
548
+ "loss": 2.3732574462890623,
549
+ "step": 1540
550
+ },
551
+ {
552
+ "epoch": 0.1051567239635996,
553
+ "grad_norm": 0.36328125,
554
+ "learning_rate": 0.0007,
555
+ "loss": 2.3624162673950195,
556
+ "step": 1560
557
+ },
558
+ {
559
+ "epoch": 0.10650488709133805,
560
+ "grad_norm": 0.359375,
561
+ "learning_rate": 0.0007,
562
+ "loss": 2.3779842376708986,
563
+ "step": 1580
564
+ },
565
+ {
566
+ "epoch": 0.10785305021907651,
567
+ "grad_norm": 0.369140625,
568
+ "learning_rate": 0.0007,
569
+ "loss": 2.3441116333007814,
570
+ "step": 1600
571
+ },
572
+ {
573
+ "epoch": 0.10920121334681497,
574
+ "grad_norm": 0.376953125,
575
+ "learning_rate": 0.0007,
576
+ "loss": 2.3418384552001954,
577
+ "step": 1620
578
+ },
579
+ {
580
+ "epoch": 0.11054937647455342,
581
+ "grad_norm": 0.373046875,
582
+ "learning_rate": 0.0007,
583
+ "loss": 2.3539493560791014,
584
+ "step": 1640
585
+ },
586
+ {
587
+ "epoch": 0.11189753960229187,
588
+ "grad_norm": 0.359375,
589
+ "learning_rate": 0.0007,
590
+ "loss": 2.332906150817871,
591
+ "step": 1660
592
+ },
593
+ {
594
+ "epoch": 0.11324570273003033,
595
+ "grad_norm": 0.408203125,
596
+ "learning_rate": 0.0007,
597
+ "loss": 2.3174556732177733,
598
+ "step": 1680
599
+ },
600
+ {
601
+ "epoch": 0.11459386585776879,
602
+ "grad_norm": 0.361328125,
603
+ "learning_rate": 0.0007,
604
+ "loss": 2.330640411376953,
605
+ "step": 1700
606
+ },
607
+ {
608
+ "epoch": 0.11594202898550725,
609
+ "grad_norm": 0.423828125,
610
+ "learning_rate": 0.0007,
611
+ "loss": 2.306385803222656,
612
+ "step": 1720
613
+ },
614
+ {
615
+ "epoch": 0.1172901921132457,
616
+ "grad_norm": 0.365234375,
617
+ "learning_rate": 0.0007,
618
+ "loss": 2.2937660217285156,
619
+ "step": 1740
620
+ },
621
+ {
622
+ "epoch": 0.11863835524098416,
623
+ "grad_norm": 0.3671875,
624
+ "learning_rate": 0.0007,
625
+ "loss": 2.297491264343262,
626
+ "step": 1760
627
+ },
628
+ {
629
+ "epoch": 0.11998651836872262,
630
+ "grad_norm": 0.337890625,
631
+ "learning_rate": 0.0007,
632
+ "loss": 2.312083053588867,
633
+ "step": 1780
634
+ },
635
+ {
636
+ "epoch": 0.12133468149646107,
637
+ "grad_norm": 0.357421875,
638
+ "learning_rate": 0.0007,
639
+ "loss": 2.307674789428711,
640
+ "step": 1800
641
+ },
642
+ {
643
+ "epoch": 0.12268284462419952,
644
+ "grad_norm": 0.375,
645
+ "learning_rate": 0.0007,
646
+ "loss": 2.296724319458008,
647
+ "step": 1820
648
+ },
649
+ {
650
+ "epoch": 0.12403100775193798,
651
+ "grad_norm": 0.376953125,
652
+ "learning_rate": 0.0007,
653
+ "loss": 2.2644861221313475,
654
+ "step": 1840
655
+ },
656
+ {
657
+ "epoch": 0.12537917087967643,
658
+ "grad_norm": 0.353515625,
659
+ "learning_rate": 0.0007,
660
+ "loss": 2.2662803649902346,
661
+ "step": 1860
662
+ },
663
+ {
664
+ "epoch": 0.1267273340074149,
665
+ "grad_norm": 0.33984375,
666
+ "learning_rate": 0.0007,
667
+ "loss": 2.2558313369750977,
668
+ "step": 1880
669
+ },
670
+ {
671
+ "epoch": 0.12807549713515334,
672
+ "grad_norm": 0.365234375,
673
+ "learning_rate": 0.0007,
674
+ "loss": 2.262881851196289,
675
+ "step": 1900
676
+ },
677
+ {
678
+ "epoch": 0.12942366026289182,
679
+ "grad_norm": 0.349609375,
680
+ "learning_rate": 0.0007,
681
+ "loss": 2.2384838104248046,
682
+ "step": 1920
683
+ },
684
+ {
685
+ "epoch": 0.13077182339063026,
686
+ "grad_norm": 0.361328125,
687
+ "learning_rate": 0.0007,
688
+ "loss": 2.247391700744629,
689
+ "step": 1940
690
+ },
691
+ {
692
+ "epoch": 0.13211998651836873,
693
+ "grad_norm": 0.34765625,
694
+ "learning_rate": 0.0007,
695
+ "loss": 2.2225168228149412,
696
+ "step": 1960
697
+ },
698
+ {
699
+ "epoch": 0.13346814964610718,
700
+ "grad_norm": 0.380859375,
701
+ "learning_rate": 0.0007,
702
+ "loss": 2.2470256805419924,
703
+ "step": 1980
704
+ },
705
+ {
706
+ "epoch": 0.13481631277384565,
707
+ "grad_norm": 0.3359375,
708
+ "learning_rate": 0.0007,
709
+ "loss": 2.218878173828125,
710
+ "step": 2000
711
+ },
712
+ {
713
+ "epoch": 0.1361644759015841,
714
+ "grad_norm": 0.357421875,
715
+ "learning_rate": 0.0007,
716
+ "loss": 2.227268409729004,
717
+ "step": 2020
718
+ },
719
+ {
720
+ "epoch": 0.13751263902932254,
721
+ "grad_norm": 0.380859375,
722
+ "learning_rate": 0.0007,
723
+ "loss": 2.2079793930053713,
724
+ "step": 2040
725
+ },
726
+ {
727
+ "epoch": 0.138860802157061,
728
+ "grad_norm": 0.380859375,
729
+ "learning_rate": 0.0007,
730
+ "loss": 2.2262474060058595,
731
+ "step": 2060
732
+ },
733
+ {
734
+ "epoch": 0.14020896528479945,
735
+ "grad_norm": 0.349609375,
736
+ "learning_rate": 0.0007,
737
+ "loss": 2.2017698287963867,
738
+ "step": 2080
739
+ },
740
+ {
741
+ "epoch": 0.14155712841253792,
742
+ "grad_norm": 0.333984375,
743
+ "learning_rate": 0.0007,
744
+ "loss": 2.19287223815918,
745
+ "step": 2100
746
+ },
747
+ {
748
+ "epoch": 0.14290529154027637,
749
+ "grad_norm": 0.349609375,
750
+ "learning_rate": 0.0007,
751
+ "loss": 2.201098823547363,
752
+ "step": 2120
753
+ },
754
+ {
755
+ "epoch": 0.14425345466801484,
756
+ "grad_norm": 0.333984375,
757
+ "learning_rate": 0.0007,
758
+ "loss": 2.2028568267822264,
759
+ "step": 2140
760
+ },
761
+ {
762
+ "epoch": 0.14560161779575329,
763
+ "grad_norm": 0.353515625,
764
+ "learning_rate": 0.0007,
765
+ "loss": 2.1869789123535157,
766
+ "step": 2160
767
+ },
768
+ {
769
+ "epoch": 0.14694978092349173,
770
+ "grad_norm": 0.33984375,
771
+ "learning_rate": 0.0007,
772
+ "loss": 2.194090461730957,
773
+ "step": 2180
774
+ },
775
+ {
776
+ "epoch": 0.1482979440512302,
777
+ "grad_norm": 0.333984375,
778
+ "learning_rate": 0.0007,
779
+ "loss": 2.169504165649414,
780
+ "step": 2200
781
+ },
782
+ {
783
+ "epoch": 0.14964610717896865,
784
+ "grad_norm": 0.359375,
785
+ "learning_rate": 0.0007,
786
+ "loss": 2.17819766998291,
787
+ "step": 2220
788
+ },
789
+ {
790
+ "epoch": 0.15099427030670712,
791
+ "grad_norm": 0.361328125,
792
+ "learning_rate": 0.0007,
793
+ "loss": 2.1870670318603516,
794
+ "step": 2240
795
+ },
796
+ {
797
+ "epoch": 0.15234243343444556,
798
+ "grad_norm": 0.359375,
799
+ "learning_rate": 0.0007,
800
+ "loss": 2.1656898498535155,
801
+ "step": 2260
802
+ },
803
+ {
804
+ "epoch": 0.15369059656218403,
805
+ "grad_norm": 0.369140625,
806
+ "learning_rate": 0.0007,
807
+ "loss": 2.174662971496582,
808
+ "step": 2280
809
+ },
810
+ {
811
+ "epoch": 0.15503875968992248,
812
+ "grad_norm": 0.3515625,
813
+ "learning_rate": 0.0007,
814
+ "loss": 2.1561347961425783,
815
+ "step": 2300
816
+ },
817
+ {
818
+ "epoch": 0.15638692281766095,
819
+ "grad_norm": 0.33203125,
820
+ "learning_rate": 0.0007,
821
+ "loss": 2.16455020904541,
822
+ "step": 2320
823
+ },
824
+ {
825
+ "epoch": 0.1577350859453994,
826
+ "grad_norm": 0.330078125,
827
+ "learning_rate": 0.0007,
828
+ "loss": 2.1551021575927733,
829
+ "step": 2340
830
+ },
831
+ {
832
+ "epoch": 0.15908324907313784,
833
+ "grad_norm": 0.341796875,
834
+ "learning_rate": 0.0007,
835
+ "loss": 2.157507133483887,
836
+ "step": 2360
837
+ },
838
+ {
839
+ "epoch": 0.1604314122008763,
840
+ "grad_norm": 0.33984375,
841
+ "learning_rate": 0.0007,
842
+ "loss": 2.1596670150756836,
843
+ "step": 2380
844
+ },
845
+ {
846
+ "epoch": 0.16177957532861476,
847
+ "grad_norm": 0.357421875,
848
+ "learning_rate": 0.0007,
849
+ "loss": 2.1417112350463867,
850
+ "step": 2400
851
+ },
852
+ {
853
+ "epoch": 0.16312773845635323,
854
+ "grad_norm": 0.349609375,
855
+ "learning_rate": 0.0007,
856
+ "loss": 2.1463193893432617,
857
+ "step": 2420
858
+ },
859
+ {
860
+ "epoch": 0.16447590158409167,
861
+ "grad_norm": 0.3359375,
862
+ "learning_rate": 0.0007,
863
+ "loss": 2.133936882019043,
864
+ "step": 2440
865
+ },
866
+ {
867
+ "epoch": 0.16582406471183014,
868
+ "grad_norm": 0.328125,
869
+ "learning_rate": 0.0007,
870
+ "loss": 2.1435129165649416,
871
+ "step": 2460
872
+ },
873
+ {
874
+ "epoch": 0.1671722278395686,
875
+ "grad_norm": 0.341796875,
876
+ "learning_rate": 0.0007,
877
+ "loss": 2.1443099975585938,
878
+ "step": 2480
879
+ },
880
+ {
881
+ "epoch": 0.1683855746545332,
882
+ "eval_loss": 2.138087511062622,
883
+ "eval_runtime": 20.1123,
884
+ "eval_samples_per_second": 473.691,
885
+ "eval_steps_per_second": 0.497,
886
+ "step": 2498
887
+ },
888
+ {
889
+ "epoch": 0.16852039096730703,
890
+ "grad_norm": 0.33984375,
891
+ "learning_rate": 0.0007,
892
+ "loss": 2.126860427856445,
893
+ "step": 2500
894
+ },
895
+ {
896
+ "epoch": 0.16852039096730703,
897
+ "eval_loss": 2.1375532150268555,
898
+ "eval_runtime": 20.4047,
899
+ "eval_samples_per_second": 466.903,
900
+ "eval_steps_per_second": 0.49,
901
+ "step": 2500
902
+ },
903
+ {
904
+ "epoch": 0.1698685540950455,
905
+ "grad_norm": 24903680.0,
906
+ "learning_rate": 3.4200000000000005e-05,
907
+ "loss": 2.128565788269043,
908
+ "step": 2520
909
+ },
910
+ {
911
+ "epoch": 0.17121671722278395,
912
+ "grad_norm": 27000832.0,
913
+ "learning_rate": 7.020000000000001e-05,
914
+ "loss": 2.1492637634277343,
915
+ "step": 2540
916
+ },
917
+ {
918
+ "epoch": 0.17256488035052242,
919
+ "grad_norm": 119537664.0,
920
+ "learning_rate": 9e-05,
921
+ "loss": 3.623552703857422,
922
+ "step": 2560
923
+ },
924
+ {
925
+ "epoch": 0.17391304347826086,
926
+ "grad_norm": 301989888.0,
927
+ "learning_rate": 9e-05,
928
+ "loss": 6.8911598205566404,
929
+ "step": 2580
930
+ },
931
+ {
932
+ "epoch": 0.17526120660599934,
933
+ "grad_norm": 658505728.0,
934
+ "learning_rate": 9e-05,
935
+ "loss": 7.545132446289062,
936
+ "step": 2600
937
+ },
938
+ {
939
+ "epoch": 0.17660936973373778,
940
+ "grad_norm": 1166016512.0,
941
+ "learning_rate": 9e-05,
942
+ "loss": 7.784053039550781,
943
+ "step": 2620
944
+ },
945
+ {
946
+ "epoch": 0.17795753286147623,
947
+ "grad_norm": 1769996288.0,
948
+ "learning_rate": 9e-05,
949
+ "loss": 7.88036880493164,
950
+ "step": 2640
951
+ },
952
+ {
953
+ "epoch": 0.1793056959892147,
954
+ "grad_norm": 2415919104.0,
955
+ "learning_rate": 9e-05,
956
+ "loss": 7.926078033447266,
957
+ "step": 2660
958
+ },
959
+ {
960
+ "epoch": 0.18065385911695314,
961
+ "grad_norm": 3053453312.0,
962
+ "learning_rate": 9e-05,
963
+ "loss": 7.947563934326172,
964
+ "step": 2680
965
+ },
966
+ {
967
+ "epoch": 0.1820020222446916,
968
+ "grad_norm": 3674210304.0,
969
+ "learning_rate": 9e-05,
970
+ "loss": 7.972737121582031,
971
+ "step": 2700
972
+ },
973
+ {
974
+ "epoch": 0.18335018537243006,
975
+ "grad_norm": 4278190080.0,
976
+ "learning_rate": 9e-05,
977
+ "loss": 7.9804527282714846,
978
+ "step": 2720
979
+ },
980
+ {
981
+ "epoch": 0.18469834850016853,
982
+ "grad_norm": 4831838208.0,
983
+ "learning_rate": 9e-05,
984
+ "loss": 7.980913543701172,
985
+ "step": 2740
986
+ },
987
+ {
988
+ "epoch": 0.18604651162790697,
989
+ "grad_norm": 5335154688.0,
990
+ "learning_rate": 9e-05,
991
+ "loss": 7.987354278564453,
992
+ "step": 2760
993
+ },
994
+ {
995
+ "epoch": 0.18739467475564545,
996
+ "grad_norm": 5804916736.0,
997
+ "learning_rate": 9e-05,
998
+ "loss": 8.005199432373047,
999
+ "step": 2780
1000
+ },
1001
+ {
1002
+ "epoch": 0.1887428378833839,
1003
+ "grad_norm": 6207569920.0,
1004
+ "learning_rate": 9e-05,
1005
+ "loss": 8.00537338256836,
1006
+ "step": 2800
1007
+ }
1008
+ ],
1009
+ "logging_steps": 20,
1010
+ "max_steps": 3500,
1011
+ "num_input_tokens_seen": 0,
1012
+ "num_train_epochs": 1,
1013
+ "save_steps": 100,
1014
+ "stateful_callbacks": {
1015
+ "TrainerControl": {
1016
+ "args": {
1017
+ "should_epoch_stop": false,
1018
+ "should_evaluate": false,
1019
+ "should_log": false,
1020
+ "should_save": true,
1021
+ "should_training_stop": false
1022
+ },
1023
+ "attributes": {}
1024
+ }
1025
+ },
1026
+ "total_flos": 9033594647347200.0,
1027
+ "train_batch_size": 64,
1028
+ "trial_name": null,
1029
+ "trial_params": null
1030
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2800/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f62f8186f489a47bc68f80ec9f826cd67dc6a0232e67ff9d099ed1157b6e25f
3
+ size 4920
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/config.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation": "powlu",
3
+ "architectures": [
4
+ "TinyLlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 1,
9
+ "dtype": "bfloat16",
10
+ "eos_token_id": 2,
11
+ "head_dim": 32,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 128,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 256,
16
+ "max_position_embeddings": 512,
17
+ "mlp_bias": false,
18
+ "mlp_type": "glu",
19
+ "model_type": "tiny_llama",
20
+ "num_attention_heads": 4,
21
+ "num_hidden_layers": 100,
22
+ "num_key_value_heads": 4,
23
+ "pad_token_id": 0,
24
+ "powlu_m": 3.0,
25
+ "pretraining_tp": 1,
26
+ "rms_norm_eps": 1e-06,
27
+ "rope_parameters": {
28
+ "rope_theta": 10000.0,
29
+ "rope_type": "default"
30
+ },
31
+ "tie_word_embeddings": true,
32
+ "tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
33
+ "transformers_version": "5.16.0.dev0",
34
+ "use_cache": false,
35
+ "vocab_size": 4096,
36
+ "waleed_beta": 10
37
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cfd33196415e67b0f9ffad93c56e1cfbe68e1c80dea84cba6daeec1f93c8ce07
3
+ size 33967272
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b16889ff79fac6df770d9b8d5071505e1f50b80e8e7c1d4d1ddc1597c1f316f
3
+ size 39573050
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cf9097d4513154245c48236b6ec5137b7ee2a21c9f58f2cba798ea275c6026f
3
+ size 14244
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:66f5b68450b0c18ded7ac53967162b3ffa6171c24c7dd24167f487c9909bda68
3
+ size 1064
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/tokenizer_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "backend": "tokenizers",
4
+ "bos_token": "<|endoftext|>",
5
+ "eos_token": "<|endoftext|>",
6
+ "errors": "replace",
7
+ "is_local": false,
8
+ "local_files_only": false,
9
+ "model_max_length": 1024,
10
+ "pad_token": "<|endoftext|>",
11
+ "tokenizer_class": "GPT2Tokenizer",
12
+ "unk_token": "<|endoftext|>"
13
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/trainer_state.json ADDED
@@ -0,0 +1,1065 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.19548365352207617,
6
+ "eval_steps": 2498,
7
+ "global_step": 2900,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.0013481631277384564,
14
+ "grad_norm": 2.09375,
15
+ "learning_rate": 2.66e-05,
16
+ "loss": 8.300566864013671,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.002696326255476913,
21
+ "grad_norm": 1.375,
22
+ "learning_rate": 5.46e-05,
23
+ "loss": 8.056343841552735,
24
+ "step": 40
25
+ },
26
+ {
27
+ "epoch": 0.004044489383215369,
28
+ "grad_norm": 1.2734375,
29
+ "learning_rate": 8.259999999999999e-05,
30
+ "loss": 7.841481781005859,
31
+ "step": 60
32
+ },
33
+ {
34
+ "epoch": 0.005392652510953826,
35
+ "grad_norm": 1.265625,
36
+ "learning_rate": 0.0001106,
37
+ "loss": 7.57496337890625,
38
+ "step": 80
39
+ },
40
+ {
41
+ "epoch": 0.006740815638692282,
42
+ "grad_norm": 1.1953125,
43
+ "learning_rate": 0.0001386,
44
+ "loss": 7.233026123046875,
45
+ "step": 100
46
+ },
47
+ {
48
+ "epoch": 0.008088978766430738,
49
+ "grad_norm": 1.1328125,
50
+ "learning_rate": 0.00016659999999999998,
51
+ "loss": 6.859487152099609,
52
+ "step": 120
53
+ },
54
+ {
55
+ "epoch": 0.009437141894169195,
56
+ "grad_norm": 0.94921875,
57
+ "learning_rate": 0.00019460000000000001,
58
+ "loss": 6.5162200927734375,
59
+ "step": 140
60
+ },
61
+ {
62
+ "epoch": 0.010785305021907651,
63
+ "grad_norm": 0.796875,
64
+ "learning_rate": 0.0002226,
65
+ "loss": 6.237432861328125,
66
+ "step": 160
67
+ },
68
+ {
69
+ "epoch": 0.012133468149646108,
70
+ "grad_norm": 0.71875,
71
+ "learning_rate": 0.00025059999999999997,
72
+ "loss": 6.039522552490235,
73
+ "step": 180
74
+ },
75
+ {
76
+ "epoch": 0.013481631277384564,
77
+ "grad_norm": 0.57421875,
78
+ "learning_rate": 0.0002786,
79
+ "loss": 5.8798988342285154,
80
+ "step": 200
81
+ },
82
+ {
83
+ "epoch": 0.01482979440512302,
84
+ "grad_norm": 1.90625,
85
+ "learning_rate": 0.00030659999999999997,
86
+ "loss": 5.723947143554687,
87
+ "step": 220
88
+ },
89
+ {
90
+ "epoch": 0.016177957532861477,
91
+ "grad_norm": 0.6796875,
92
+ "learning_rate": 0.0003346,
93
+ "loss": 5.551593399047851,
94
+ "step": 240
95
+ },
96
+ {
97
+ "epoch": 0.01752612066059993,
98
+ "grad_norm": 1.6875,
99
+ "learning_rate": 0.00036260000000000003,
100
+ "loss": 5.3516895294189455,
101
+ "step": 260
102
+ },
103
+ {
104
+ "epoch": 0.01887428378833839,
105
+ "grad_norm": 0.84765625,
106
+ "learning_rate": 0.0003906,
107
+ "loss": 5.130373764038086,
108
+ "step": 280
109
+ },
110
+ {
111
+ "epoch": 0.020222446916076844,
112
+ "grad_norm": 0.55859375,
113
+ "learning_rate": 0.0004186,
114
+ "loss": 4.882920455932617,
115
+ "step": 300
116
+ },
117
+ {
118
+ "epoch": 0.021570610043815303,
119
+ "grad_norm": 1.15625,
120
+ "learning_rate": 0.0004466,
121
+ "loss": 4.673929977416992,
122
+ "step": 320
123
+ },
124
+ {
125
+ "epoch": 0.022918773171553757,
126
+ "grad_norm": 1.0,
127
+ "learning_rate": 0.00047460000000000004,
128
+ "loss": 4.511159515380859,
129
+ "step": 340
130
+ },
131
+ {
132
+ "epoch": 0.024266936299292215,
133
+ "grad_norm": 0.96875,
134
+ "learning_rate": 0.0005026,
135
+ "loss": 4.360583114624023,
136
+ "step": 360
137
+ },
138
+ {
139
+ "epoch": 0.02561509942703067,
140
+ "grad_norm": 0.80078125,
141
+ "learning_rate": 0.0005306,
142
+ "loss": 4.208797836303711,
143
+ "step": 380
144
+ },
145
+ {
146
+ "epoch": 0.026963262554769128,
147
+ "grad_norm": 0.703125,
148
+ "learning_rate": 0.0005586,
149
+ "loss": 4.059257507324219,
150
+ "step": 400
151
+ },
152
+ {
153
+ "epoch": 0.028311425682507583,
154
+ "grad_norm": 1.390625,
155
+ "learning_rate": 0.0005866,
156
+ "loss": 3.945643997192383,
157
+ "step": 420
158
+ },
159
+ {
160
+ "epoch": 0.02965958881024604,
161
+ "grad_norm": 0.8046875,
162
+ "learning_rate": 0.0006146,
163
+ "loss": 3.878635787963867,
164
+ "step": 440
165
+ },
166
+ {
167
+ "epoch": 0.031007751937984496,
168
+ "grad_norm": 0.7578125,
169
+ "learning_rate": 0.0006426,
170
+ "loss": 3.763169860839844,
171
+ "step": 460
172
+ },
173
+ {
174
+ "epoch": 0.032355915065722954,
175
+ "grad_norm": 0.578125,
176
+ "learning_rate": 0.0006705999999999999,
177
+ "loss": 3.680609130859375,
178
+ "step": 480
179
+ },
180
+ {
181
+ "epoch": 0.03370407819346141,
182
+ "grad_norm": 0.6484375,
183
+ "learning_rate": 0.0006986,
184
+ "loss": 3.6085578918457033,
185
+ "step": 500
186
+ },
187
+ {
188
+ "epoch": 0.03505224132119986,
189
+ "grad_norm": 0.431640625,
190
+ "learning_rate": 0.0007,
191
+ "loss": 3.5324871063232424,
192
+ "step": 520
193
+ },
194
+ {
195
+ "epoch": 0.03640040444893832,
196
+ "grad_norm": 0.4453125,
197
+ "learning_rate": 0.0007,
198
+ "loss": 3.454126739501953,
199
+ "step": 540
200
+ },
201
+ {
202
+ "epoch": 0.03774856757667678,
203
+ "grad_norm": 0.546875,
204
+ "learning_rate": 0.0007,
205
+ "loss": 3.4041587829589846,
206
+ "step": 560
207
+ },
208
+ {
209
+ "epoch": 0.03909673070441524,
210
+ "grad_norm": 0.46875,
211
+ "learning_rate": 0.0007,
212
+ "loss": 3.347498321533203,
213
+ "step": 580
214
+ },
215
+ {
216
+ "epoch": 0.04044489383215369,
217
+ "grad_norm": 0.546875,
218
+ "learning_rate": 0.0007,
219
+ "loss": 3.313167190551758,
220
+ "step": 600
221
+ },
222
+ {
223
+ "epoch": 0.04179305695989215,
224
+ "grad_norm": 0.51953125,
225
+ "learning_rate": 0.0007,
226
+ "loss": 3.2490795135498045,
227
+ "step": 620
228
+ },
229
+ {
230
+ "epoch": 0.043141220087630605,
231
+ "grad_norm": 0.453125,
232
+ "learning_rate": 0.0007,
233
+ "loss": 3.221387481689453,
234
+ "step": 640
235
+ },
236
+ {
237
+ "epoch": 0.044489383215369056,
238
+ "grad_norm": 0.5234375,
239
+ "learning_rate": 0.0007,
240
+ "loss": 3.1599220275878905,
241
+ "step": 660
242
+ },
243
+ {
244
+ "epoch": 0.045837546343107514,
245
+ "grad_norm": 0.4375,
246
+ "learning_rate": 0.0007,
247
+ "loss": 3.133222198486328,
248
+ "step": 680
249
+ },
250
+ {
251
+ "epoch": 0.04718570947084597,
252
+ "grad_norm": 0.5703125,
253
+ "learning_rate": 0.0007,
254
+ "loss": 3.0884355545043944,
255
+ "step": 700
256
+ },
257
+ {
258
+ "epoch": 0.04853387259858443,
259
+ "grad_norm": 0.431640625,
260
+ "learning_rate": 0.0007,
261
+ "loss": 3.0560993194580077,
262
+ "step": 720
263
+ },
264
+ {
265
+ "epoch": 0.04988203572632288,
266
+ "grad_norm": 0.462890625,
267
+ "learning_rate": 0.0007,
268
+ "loss": 3.0169742584228514,
269
+ "step": 740
270
+ },
271
+ {
272
+ "epoch": 0.05123019885406134,
273
+ "grad_norm": 0.390625,
274
+ "learning_rate": 0.0007,
275
+ "loss": 2.9988849639892576,
276
+ "step": 760
277
+ },
278
+ {
279
+ "epoch": 0.0525783619817998,
280
+ "grad_norm": 0.466796875,
281
+ "learning_rate": 0.0007,
282
+ "loss": 3.004904937744141,
283
+ "step": 780
284
+ },
285
+ {
286
+ "epoch": 0.053926525109538256,
287
+ "grad_norm": 0.443359375,
288
+ "learning_rate": 0.0007,
289
+ "loss": 2.9500053405761717,
290
+ "step": 800
291
+ },
292
+ {
293
+ "epoch": 0.05527468823727671,
294
+ "grad_norm": 0.453125,
295
+ "learning_rate": 0.0007,
296
+ "loss": 2.93444938659668,
297
+ "step": 820
298
+ },
299
+ {
300
+ "epoch": 0.056622851365015166,
301
+ "grad_norm": 0.421875,
302
+ "learning_rate": 0.0007,
303
+ "loss": 2.9036865234375,
304
+ "step": 840
305
+ },
306
+ {
307
+ "epoch": 0.057971014492753624,
308
+ "grad_norm": 0.392578125,
309
+ "learning_rate": 0.0007,
310
+ "loss": 2.859415054321289,
311
+ "step": 860
312
+ },
313
+ {
314
+ "epoch": 0.05931917762049208,
315
+ "grad_norm": 0.42578125,
316
+ "learning_rate": 0.0007,
317
+ "loss": 2.8376497268676757,
318
+ "step": 880
319
+ },
320
+ {
321
+ "epoch": 0.06066734074823053,
322
+ "grad_norm": 0.46484375,
323
+ "learning_rate": 0.0007,
324
+ "loss": 2.821657180786133,
325
+ "step": 900
326
+ },
327
+ {
328
+ "epoch": 0.06201550387596899,
329
+ "grad_norm": 0.46875,
330
+ "learning_rate": 0.0007,
331
+ "loss": 2.7936939239501952,
332
+ "step": 920
333
+ },
334
+ {
335
+ "epoch": 0.06336366700370745,
336
+ "grad_norm": 0.40234375,
337
+ "learning_rate": 0.0007,
338
+ "loss": 2.779148292541504,
339
+ "step": 940
340
+ },
341
+ {
342
+ "epoch": 0.06471183013144591,
343
+ "grad_norm": 0.40234375,
344
+ "learning_rate": 0.0007,
345
+ "loss": 2.7523166656494142,
346
+ "step": 960
347
+ },
348
+ {
349
+ "epoch": 0.06605999325918437,
350
+ "grad_norm": 0.408203125,
351
+ "learning_rate": 0.0007,
352
+ "loss": 2.7326622009277344,
353
+ "step": 980
354
+ },
355
+ {
356
+ "epoch": 0.06740815638692282,
357
+ "grad_norm": 0.396484375,
358
+ "learning_rate": 0.0007,
359
+ "loss": 2.730000305175781,
360
+ "step": 1000
361
+ },
362
+ {
363
+ "epoch": 0.06875631951466127,
364
+ "grad_norm": 0.3828125,
365
+ "learning_rate": 0.0007,
366
+ "loss": 2.6984880447387694,
367
+ "step": 1020
368
+ },
369
+ {
370
+ "epoch": 0.07010448264239973,
371
+ "grad_norm": 0.423828125,
372
+ "learning_rate": 0.0007,
373
+ "loss": 2.6830041885375975,
374
+ "step": 1040
375
+ },
376
+ {
377
+ "epoch": 0.07145264577013818,
378
+ "grad_norm": 0.404296875,
379
+ "learning_rate": 0.0007,
380
+ "loss": 2.6666488647460938,
381
+ "step": 1060
382
+ },
383
+ {
384
+ "epoch": 0.07280080889787664,
385
+ "grad_norm": 0.404296875,
386
+ "learning_rate": 0.0007,
387
+ "loss": 2.6461860656738283,
388
+ "step": 1080
389
+ },
390
+ {
391
+ "epoch": 0.0741489720256151,
392
+ "grad_norm": 0.427734375,
393
+ "learning_rate": 0.0007,
394
+ "loss": 2.638811492919922,
395
+ "step": 1100
396
+ },
397
+ {
398
+ "epoch": 0.07549713515335356,
399
+ "grad_norm": 0.38671875,
400
+ "learning_rate": 0.0007,
401
+ "loss": 2.604878807067871,
402
+ "step": 1120
403
+ },
404
+ {
405
+ "epoch": 0.07684529828109202,
406
+ "grad_norm": 0.423828125,
407
+ "learning_rate": 0.0007,
408
+ "loss": 2.5887590408325196,
409
+ "step": 1140
410
+ },
411
+ {
412
+ "epoch": 0.07819346140883048,
413
+ "grad_norm": 0.416015625,
414
+ "learning_rate": 0.0007,
415
+ "loss": 2.5864500045776366,
416
+ "step": 1160
417
+ },
418
+ {
419
+ "epoch": 0.07954162453656892,
420
+ "grad_norm": 0.40234375,
421
+ "learning_rate": 0.0007,
422
+ "loss": 2.5788290023803713,
423
+ "step": 1180
424
+ },
425
+ {
426
+ "epoch": 0.08088978766430738,
427
+ "grad_norm": 0.404296875,
428
+ "learning_rate": 0.0007,
429
+ "loss": 2.570602607727051,
430
+ "step": 1200
431
+ },
432
+ {
433
+ "epoch": 0.08223795079204584,
434
+ "grad_norm": 0.390625,
435
+ "learning_rate": 0.0007,
436
+ "loss": 2.5528938293457033,
437
+ "step": 1220
438
+ },
439
+ {
440
+ "epoch": 0.0835861139197843,
441
+ "grad_norm": 0.373046875,
442
+ "learning_rate": 0.0007,
443
+ "loss": 2.53753662109375,
444
+ "step": 1240
445
+ },
446
+ {
447
+ "epoch": 0.08493427704752275,
448
+ "grad_norm": 0.384765625,
449
+ "learning_rate": 0.0007,
450
+ "loss": 2.523891830444336,
451
+ "step": 1260
452
+ },
453
+ {
454
+ "epoch": 0.08628244017526121,
455
+ "grad_norm": 0.3671875,
456
+ "learning_rate": 0.0007,
457
+ "loss": 2.5049261093139648,
458
+ "step": 1280
459
+ },
460
+ {
461
+ "epoch": 0.08763060330299967,
462
+ "grad_norm": 0.392578125,
463
+ "learning_rate": 0.0007,
464
+ "loss": 2.4961641311645506,
465
+ "step": 1300
466
+ },
467
+ {
468
+ "epoch": 0.08897876643073811,
469
+ "grad_norm": 0.353515625,
470
+ "learning_rate": 0.0007,
471
+ "loss": 2.487112045288086,
472
+ "step": 1320
473
+ },
474
+ {
475
+ "epoch": 0.09032692955847657,
476
+ "grad_norm": 0.38671875,
477
+ "learning_rate": 0.0007,
478
+ "loss": 2.471820259094238,
479
+ "step": 1340
480
+ },
481
+ {
482
+ "epoch": 0.09167509268621503,
483
+ "grad_norm": 0.392578125,
484
+ "learning_rate": 0.0007,
485
+ "loss": 2.4697391510009767,
486
+ "step": 1360
487
+ },
488
+ {
489
+ "epoch": 0.09302325581395349,
490
+ "grad_norm": 0.369140625,
491
+ "learning_rate": 0.0007,
492
+ "loss": 2.4492652893066404,
493
+ "step": 1380
494
+ },
495
+ {
496
+ "epoch": 0.09437141894169195,
497
+ "grad_norm": 0.384765625,
498
+ "learning_rate": 0.0007,
499
+ "loss": 2.4387527465820313,
500
+ "step": 1400
501
+ },
502
+ {
503
+ "epoch": 0.0957195820694304,
504
+ "grad_norm": 0.380859375,
505
+ "learning_rate": 0.0007,
506
+ "loss": 2.4479516983032226,
507
+ "step": 1420
508
+ },
509
+ {
510
+ "epoch": 0.09706774519716886,
511
+ "grad_norm": 0.359375,
512
+ "learning_rate": 0.0007,
513
+ "loss": 2.4246395111083983,
514
+ "step": 1440
515
+ },
516
+ {
517
+ "epoch": 0.09841590832490732,
518
+ "grad_norm": 0.373046875,
519
+ "learning_rate": 0.0007,
520
+ "loss": 2.4114765167236327,
521
+ "step": 1460
522
+ },
523
+ {
524
+ "epoch": 0.09976407145264576,
525
+ "grad_norm": 0.375,
526
+ "learning_rate": 0.0007,
527
+ "loss": 2.399285888671875,
528
+ "step": 1480
529
+ },
530
+ {
531
+ "epoch": 0.10111223458038422,
532
+ "grad_norm": 0.37890625,
533
+ "learning_rate": 0.0007,
534
+ "loss": 2.3839179992675783,
535
+ "step": 1500
536
+ },
537
+ {
538
+ "epoch": 0.10246039770812268,
539
+ "grad_norm": 0.41796875,
540
+ "learning_rate": 0.0007,
541
+ "loss": 2.3861133575439455,
542
+ "step": 1520
543
+ },
544
+ {
545
+ "epoch": 0.10380856083586114,
546
+ "grad_norm": 0.40625,
547
+ "learning_rate": 0.0007,
548
+ "loss": 2.3732574462890623,
549
+ "step": 1540
550
+ },
551
+ {
552
+ "epoch": 0.1051567239635996,
553
+ "grad_norm": 0.36328125,
554
+ "learning_rate": 0.0007,
555
+ "loss": 2.3624162673950195,
556
+ "step": 1560
557
+ },
558
+ {
559
+ "epoch": 0.10650488709133805,
560
+ "grad_norm": 0.359375,
561
+ "learning_rate": 0.0007,
562
+ "loss": 2.3779842376708986,
563
+ "step": 1580
564
+ },
565
+ {
566
+ "epoch": 0.10785305021907651,
567
+ "grad_norm": 0.369140625,
568
+ "learning_rate": 0.0007,
569
+ "loss": 2.3441116333007814,
570
+ "step": 1600
571
+ },
572
+ {
573
+ "epoch": 0.10920121334681497,
574
+ "grad_norm": 0.376953125,
575
+ "learning_rate": 0.0007,
576
+ "loss": 2.3418384552001954,
577
+ "step": 1620
578
+ },
579
+ {
580
+ "epoch": 0.11054937647455342,
581
+ "grad_norm": 0.373046875,
582
+ "learning_rate": 0.0007,
583
+ "loss": 2.3539493560791014,
584
+ "step": 1640
585
+ },
586
+ {
587
+ "epoch": 0.11189753960229187,
588
+ "grad_norm": 0.359375,
589
+ "learning_rate": 0.0007,
590
+ "loss": 2.332906150817871,
591
+ "step": 1660
592
+ },
593
+ {
594
+ "epoch": 0.11324570273003033,
595
+ "grad_norm": 0.408203125,
596
+ "learning_rate": 0.0007,
597
+ "loss": 2.3174556732177733,
598
+ "step": 1680
599
+ },
600
+ {
601
+ "epoch": 0.11459386585776879,
602
+ "grad_norm": 0.361328125,
603
+ "learning_rate": 0.0007,
604
+ "loss": 2.330640411376953,
605
+ "step": 1700
606
+ },
607
+ {
608
+ "epoch": 0.11594202898550725,
609
+ "grad_norm": 0.423828125,
610
+ "learning_rate": 0.0007,
611
+ "loss": 2.306385803222656,
612
+ "step": 1720
613
+ },
614
+ {
615
+ "epoch": 0.1172901921132457,
616
+ "grad_norm": 0.365234375,
617
+ "learning_rate": 0.0007,
618
+ "loss": 2.2937660217285156,
619
+ "step": 1740
620
+ },
621
+ {
622
+ "epoch": 0.11863835524098416,
623
+ "grad_norm": 0.3671875,
624
+ "learning_rate": 0.0007,
625
+ "loss": 2.297491264343262,
626
+ "step": 1760
627
+ },
628
+ {
629
+ "epoch": 0.11998651836872262,
630
+ "grad_norm": 0.337890625,
631
+ "learning_rate": 0.0007,
632
+ "loss": 2.312083053588867,
633
+ "step": 1780
634
+ },
635
+ {
636
+ "epoch": 0.12133468149646107,
637
+ "grad_norm": 0.357421875,
638
+ "learning_rate": 0.0007,
639
+ "loss": 2.307674789428711,
640
+ "step": 1800
641
+ },
642
+ {
643
+ "epoch": 0.12268284462419952,
644
+ "grad_norm": 0.375,
645
+ "learning_rate": 0.0007,
646
+ "loss": 2.296724319458008,
647
+ "step": 1820
648
+ },
649
+ {
650
+ "epoch": 0.12403100775193798,
651
+ "grad_norm": 0.376953125,
652
+ "learning_rate": 0.0007,
653
+ "loss": 2.2644861221313475,
654
+ "step": 1840
655
+ },
656
+ {
657
+ "epoch": 0.12537917087967643,
658
+ "grad_norm": 0.353515625,
659
+ "learning_rate": 0.0007,
660
+ "loss": 2.2662803649902346,
661
+ "step": 1860
662
+ },
663
+ {
664
+ "epoch": 0.1267273340074149,
665
+ "grad_norm": 0.33984375,
666
+ "learning_rate": 0.0007,
667
+ "loss": 2.2558313369750977,
668
+ "step": 1880
669
+ },
670
+ {
671
+ "epoch": 0.12807549713515334,
672
+ "grad_norm": 0.365234375,
673
+ "learning_rate": 0.0007,
674
+ "loss": 2.262881851196289,
675
+ "step": 1900
676
+ },
677
+ {
678
+ "epoch": 0.12942366026289182,
679
+ "grad_norm": 0.349609375,
680
+ "learning_rate": 0.0007,
681
+ "loss": 2.2384838104248046,
682
+ "step": 1920
683
+ },
684
+ {
685
+ "epoch": 0.13077182339063026,
686
+ "grad_norm": 0.361328125,
687
+ "learning_rate": 0.0007,
688
+ "loss": 2.247391700744629,
689
+ "step": 1940
690
+ },
691
+ {
692
+ "epoch": 0.13211998651836873,
693
+ "grad_norm": 0.34765625,
694
+ "learning_rate": 0.0007,
695
+ "loss": 2.2225168228149412,
696
+ "step": 1960
697
+ },
698
+ {
699
+ "epoch": 0.13346814964610718,
700
+ "grad_norm": 0.380859375,
701
+ "learning_rate": 0.0007,
702
+ "loss": 2.2470256805419924,
703
+ "step": 1980
704
+ },
705
+ {
706
+ "epoch": 0.13481631277384565,
707
+ "grad_norm": 0.3359375,
708
+ "learning_rate": 0.0007,
709
+ "loss": 2.218878173828125,
710
+ "step": 2000
711
+ },
712
+ {
713
+ "epoch": 0.1361644759015841,
714
+ "grad_norm": 0.357421875,
715
+ "learning_rate": 0.0007,
716
+ "loss": 2.227268409729004,
717
+ "step": 2020
718
+ },
719
+ {
720
+ "epoch": 0.13751263902932254,
721
+ "grad_norm": 0.380859375,
722
+ "learning_rate": 0.0007,
723
+ "loss": 2.2079793930053713,
724
+ "step": 2040
725
+ },
726
+ {
727
+ "epoch": 0.138860802157061,
728
+ "grad_norm": 0.380859375,
729
+ "learning_rate": 0.0007,
730
+ "loss": 2.2262474060058595,
731
+ "step": 2060
732
+ },
733
+ {
734
+ "epoch": 0.14020896528479945,
735
+ "grad_norm": 0.349609375,
736
+ "learning_rate": 0.0007,
737
+ "loss": 2.2017698287963867,
738
+ "step": 2080
739
+ },
740
+ {
741
+ "epoch": 0.14155712841253792,
742
+ "grad_norm": 0.333984375,
743
+ "learning_rate": 0.0007,
744
+ "loss": 2.19287223815918,
745
+ "step": 2100
746
+ },
747
+ {
748
+ "epoch": 0.14290529154027637,
749
+ "grad_norm": 0.349609375,
750
+ "learning_rate": 0.0007,
751
+ "loss": 2.201098823547363,
752
+ "step": 2120
753
+ },
754
+ {
755
+ "epoch": 0.14425345466801484,
756
+ "grad_norm": 0.333984375,
757
+ "learning_rate": 0.0007,
758
+ "loss": 2.2028568267822264,
759
+ "step": 2140
760
+ },
761
+ {
762
+ "epoch": 0.14560161779575329,
763
+ "grad_norm": 0.353515625,
764
+ "learning_rate": 0.0007,
765
+ "loss": 2.1869789123535157,
766
+ "step": 2160
767
+ },
768
+ {
769
+ "epoch": 0.14694978092349173,
770
+ "grad_norm": 0.33984375,
771
+ "learning_rate": 0.0007,
772
+ "loss": 2.194090461730957,
773
+ "step": 2180
774
+ },
775
+ {
776
+ "epoch": 0.1482979440512302,
777
+ "grad_norm": 0.333984375,
778
+ "learning_rate": 0.0007,
779
+ "loss": 2.169504165649414,
780
+ "step": 2200
781
+ },
782
+ {
783
+ "epoch": 0.14964610717896865,
784
+ "grad_norm": 0.359375,
785
+ "learning_rate": 0.0007,
786
+ "loss": 2.17819766998291,
787
+ "step": 2220
788
+ },
789
+ {
790
+ "epoch": 0.15099427030670712,
791
+ "grad_norm": 0.361328125,
792
+ "learning_rate": 0.0007,
793
+ "loss": 2.1870670318603516,
794
+ "step": 2240
795
+ },
796
+ {
797
+ "epoch": 0.15234243343444556,
798
+ "grad_norm": 0.359375,
799
+ "learning_rate": 0.0007,
800
+ "loss": 2.1656898498535155,
801
+ "step": 2260
802
+ },
803
+ {
804
+ "epoch": 0.15369059656218403,
805
+ "grad_norm": 0.369140625,
806
+ "learning_rate": 0.0007,
807
+ "loss": 2.174662971496582,
808
+ "step": 2280
809
+ },
810
+ {
811
+ "epoch": 0.15503875968992248,
812
+ "grad_norm": 0.3515625,
813
+ "learning_rate": 0.0007,
814
+ "loss": 2.1561347961425783,
815
+ "step": 2300
816
+ },
817
+ {
818
+ "epoch": 0.15638692281766095,
819
+ "grad_norm": 0.33203125,
820
+ "learning_rate": 0.0007,
821
+ "loss": 2.16455020904541,
822
+ "step": 2320
823
+ },
824
+ {
825
+ "epoch": 0.1577350859453994,
826
+ "grad_norm": 0.330078125,
827
+ "learning_rate": 0.0007,
828
+ "loss": 2.1551021575927733,
829
+ "step": 2340
830
+ },
831
+ {
832
+ "epoch": 0.15908324907313784,
833
+ "grad_norm": 0.341796875,
834
+ "learning_rate": 0.0007,
835
+ "loss": 2.157507133483887,
836
+ "step": 2360
837
+ },
838
+ {
839
+ "epoch": 0.1604314122008763,
840
+ "grad_norm": 0.33984375,
841
+ "learning_rate": 0.0007,
842
+ "loss": 2.1596670150756836,
843
+ "step": 2380
844
+ },
845
+ {
846
+ "epoch": 0.16177957532861476,
847
+ "grad_norm": 0.357421875,
848
+ "learning_rate": 0.0007,
849
+ "loss": 2.1417112350463867,
850
+ "step": 2400
851
+ },
852
+ {
853
+ "epoch": 0.16312773845635323,
854
+ "grad_norm": 0.349609375,
855
+ "learning_rate": 0.0007,
856
+ "loss": 2.1463193893432617,
857
+ "step": 2420
858
+ },
859
+ {
860
+ "epoch": 0.16447590158409167,
861
+ "grad_norm": 0.3359375,
862
+ "learning_rate": 0.0007,
863
+ "loss": 2.133936882019043,
864
+ "step": 2440
865
+ },
866
+ {
867
+ "epoch": 0.16582406471183014,
868
+ "grad_norm": 0.328125,
869
+ "learning_rate": 0.0007,
870
+ "loss": 2.1435129165649416,
871
+ "step": 2460
872
+ },
873
+ {
874
+ "epoch": 0.1671722278395686,
875
+ "grad_norm": 0.341796875,
876
+ "learning_rate": 0.0007,
877
+ "loss": 2.1443099975585938,
878
+ "step": 2480
879
+ },
880
+ {
881
+ "epoch": 0.1683855746545332,
882
+ "eval_loss": 2.138087511062622,
883
+ "eval_runtime": 20.1123,
884
+ "eval_samples_per_second": 473.691,
885
+ "eval_steps_per_second": 0.497,
886
+ "step": 2498
887
+ },
888
+ {
889
+ "epoch": 0.16852039096730703,
890
+ "grad_norm": 0.33984375,
891
+ "learning_rate": 0.0007,
892
+ "loss": 2.126860427856445,
893
+ "step": 2500
894
+ },
895
+ {
896
+ "epoch": 0.16852039096730703,
897
+ "eval_loss": 2.1375532150268555,
898
+ "eval_runtime": 20.4047,
899
+ "eval_samples_per_second": 466.903,
900
+ "eval_steps_per_second": 0.49,
901
+ "step": 2500
902
+ },
903
+ {
904
+ "epoch": 0.1698685540950455,
905
+ "grad_norm": 24903680.0,
906
+ "learning_rate": 3.4200000000000005e-05,
907
+ "loss": 2.128565788269043,
908
+ "step": 2520
909
+ },
910
+ {
911
+ "epoch": 0.17121671722278395,
912
+ "grad_norm": 27000832.0,
913
+ "learning_rate": 7.020000000000001e-05,
914
+ "loss": 2.1492637634277343,
915
+ "step": 2540
916
+ },
917
+ {
918
+ "epoch": 0.17256488035052242,
919
+ "grad_norm": 119537664.0,
920
+ "learning_rate": 9e-05,
921
+ "loss": 3.623552703857422,
922
+ "step": 2560
923
+ },
924
+ {
925
+ "epoch": 0.17391304347826086,
926
+ "grad_norm": 301989888.0,
927
+ "learning_rate": 9e-05,
928
+ "loss": 6.8911598205566404,
929
+ "step": 2580
930
+ },
931
+ {
932
+ "epoch": 0.17526120660599934,
933
+ "grad_norm": 658505728.0,
934
+ "learning_rate": 9e-05,
935
+ "loss": 7.545132446289062,
936
+ "step": 2600
937
+ },
938
+ {
939
+ "epoch": 0.17660936973373778,
940
+ "grad_norm": 1166016512.0,
941
+ "learning_rate": 9e-05,
942
+ "loss": 7.784053039550781,
943
+ "step": 2620
944
+ },
945
+ {
946
+ "epoch": 0.17795753286147623,
947
+ "grad_norm": 1769996288.0,
948
+ "learning_rate": 9e-05,
949
+ "loss": 7.88036880493164,
950
+ "step": 2640
951
+ },
952
+ {
953
+ "epoch": 0.1793056959892147,
954
+ "grad_norm": 2415919104.0,
955
+ "learning_rate": 9e-05,
956
+ "loss": 7.926078033447266,
957
+ "step": 2660
958
+ },
959
+ {
960
+ "epoch": 0.18065385911695314,
961
+ "grad_norm": 3053453312.0,
962
+ "learning_rate": 9e-05,
963
+ "loss": 7.947563934326172,
964
+ "step": 2680
965
+ },
966
+ {
967
+ "epoch": 0.1820020222446916,
968
+ "grad_norm": 3674210304.0,
969
+ "learning_rate": 9e-05,
970
+ "loss": 7.972737121582031,
971
+ "step": 2700
972
+ },
973
+ {
974
+ "epoch": 0.18335018537243006,
975
+ "grad_norm": 4278190080.0,
976
+ "learning_rate": 9e-05,
977
+ "loss": 7.9804527282714846,
978
+ "step": 2720
979
+ },
980
+ {
981
+ "epoch": 0.18469834850016853,
982
+ "grad_norm": 4831838208.0,
983
+ "learning_rate": 9e-05,
984
+ "loss": 7.980913543701172,
985
+ "step": 2740
986
+ },
987
+ {
988
+ "epoch": 0.18604651162790697,
989
+ "grad_norm": 5335154688.0,
990
+ "learning_rate": 9e-05,
991
+ "loss": 7.987354278564453,
992
+ "step": 2760
993
+ },
994
+ {
995
+ "epoch": 0.18739467475564545,
996
+ "grad_norm": 5804916736.0,
997
+ "learning_rate": 9e-05,
998
+ "loss": 8.005199432373047,
999
+ "step": 2780
1000
+ },
1001
+ {
1002
+ "epoch": 0.1887428378833839,
1003
+ "grad_norm": 6207569920.0,
1004
+ "learning_rate": 9e-05,
1005
+ "loss": 8.00537338256836,
1006
+ "step": 2800
1007
+ },
1008
+ {
1009
+ "epoch": 0.19009100101112233,
1010
+ "grad_norm": 6610223104.0,
1011
+ "learning_rate": 9e-05,
1012
+ "loss": 8.006069946289063,
1013
+ "step": 2820
1014
+ },
1015
+ {
1016
+ "epoch": 0.1914391641388608,
1017
+ "grad_norm": 6945767424.0,
1018
+ "learning_rate": 9e-05,
1019
+ "loss": 8.004747009277343,
1020
+ "step": 2840
1021
+ },
1022
+ {
1023
+ "epoch": 0.19278732726659925,
1024
+ "grad_norm": 7281311744.0,
1025
+ "learning_rate": 9e-05,
1026
+ "loss": 7.9950706481933596,
1027
+ "step": 2860
1028
+ },
1029
+ {
1030
+ "epoch": 0.19413549039433772,
1031
+ "grad_norm": 7549747200.0,
1032
+ "learning_rate": 9e-05,
1033
+ "loss": 8.006790161132812,
1034
+ "step": 2880
1035
+ },
1036
+ {
1037
+ "epoch": 0.19548365352207617,
1038
+ "grad_norm": 7818182656.0,
1039
+ "learning_rate": 9e-05,
1040
+ "loss": 8.008560180664062,
1041
+ "step": 2900
1042
+ }
1043
+ ],
1044
+ "logging_steps": 20,
1045
+ "max_steps": 3500,
1046
+ "num_input_tokens_seen": 0,
1047
+ "num_train_epochs": 1,
1048
+ "save_steps": 100,
1049
+ "stateful_callbacks": {
1050
+ "TrainerControl": {
1051
+ "args": {
1052
+ "should_epoch_stop": false,
1053
+ "should_evaluate": false,
1054
+ "should_log": false,
1055
+ "should_save": true,
1056
+ "should_training_stop": false
1057
+ },
1058
+ "attributes": {}
1059
+ }
1060
+ },
1061
+ "total_flos": 9356223027609600.0,
1062
+ "train_batch_size": 64,
1063
+ "trial_name": null,
1064
+ "trial_params": null
1065
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-2900/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f62f8186f489a47bc68f80ec9f826cd67dc6a0232e67ff9d099ed1157b6e25f
3
+ size 4920
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/config.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation": "powlu",
3
+ "architectures": [
4
+ "TinyLlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 1,
9
+ "dtype": "bfloat16",
10
+ "eos_token_id": 2,
11
+ "head_dim": 32,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 128,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 256,
16
+ "max_position_embeddings": 512,
17
+ "mlp_bias": false,
18
+ "mlp_type": "glu",
19
+ "model_type": "tiny_llama",
20
+ "num_attention_heads": 4,
21
+ "num_hidden_layers": 100,
22
+ "num_key_value_heads": 4,
23
+ "pad_token_id": 0,
24
+ "powlu_m": 3.0,
25
+ "pretraining_tp": 1,
26
+ "rms_norm_eps": 1e-06,
27
+ "rope_parameters": {
28
+ "rope_theta": 10000.0,
29
+ "rope_type": "default"
30
+ },
31
+ "tie_word_embeddings": true,
32
+ "tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
33
+ "transformers_version": "5.16.0.dev0",
34
+ "use_cache": false,
35
+ "vocab_size": 4096,
36
+ "waleed_beta": 10
37
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:03bf0879b58e6638eb7e6c60d73aa46cf51a4c625da4c601b36bcdbb70a85eca
3
+ size 33967272
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8961bcc410cc54217bb5f44aea069ebdb7a76b0a8cf0e81e7b5026da61fc3271
3
+ size 39573050
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3cf9097d4513154245c48236b6ec5137b7ee2a21c9f58f2cba798ea275c6026f
3
+ size 14244
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5138b027fc3060283bbcf135873185cb5a8f65583d5c7597645469d2c1529ff7
3
+ size 1064
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/tokenizer_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "backend": "tokenizers",
4
+ "bos_token": "<|endoftext|>",
5
+ "eos_token": "<|endoftext|>",
6
+ "errors": "replace",
7
+ "is_local": false,
8
+ "local_files_only": false,
9
+ "model_max_length": 1024,
10
+ "pad_token": "<|endoftext|>",
11
+ "tokenizer_class": "GPT2Tokenizer",
12
+ "unk_token": "<|endoftext|>"
13
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/trainer_state.json ADDED
@@ -0,0 +1,1100 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.20222446916076844,
6
+ "eval_steps": 2498,
7
+ "global_step": 3000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.0013481631277384564,
14
+ "grad_norm": 2.09375,
15
+ "learning_rate": 2.66e-05,
16
+ "loss": 8.300566864013671,
17
+ "step": 20
18
+ },
19
+ {
20
+ "epoch": 0.002696326255476913,
21
+ "grad_norm": 1.375,
22
+ "learning_rate": 5.46e-05,
23
+ "loss": 8.056343841552735,
24
+ "step": 40
25
+ },
26
+ {
27
+ "epoch": 0.004044489383215369,
28
+ "grad_norm": 1.2734375,
29
+ "learning_rate": 8.259999999999999e-05,
30
+ "loss": 7.841481781005859,
31
+ "step": 60
32
+ },
33
+ {
34
+ "epoch": 0.005392652510953826,
35
+ "grad_norm": 1.265625,
36
+ "learning_rate": 0.0001106,
37
+ "loss": 7.57496337890625,
38
+ "step": 80
39
+ },
40
+ {
41
+ "epoch": 0.006740815638692282,
42
+ "grad_norm": 1.1953125,
43
+ "learning_rate": 0.0001386,
44
+ "loss": 7.233026123046875,
45
+ "step": 100
46
+ },
47
+ {
48
+ "epoch": 0.008088978766430738,
49
+ "grad_norm": 1.1328125,
50
+ "learning_rate": 0.00016659999999999998,
51
+ "loss": 6.859487152099609,
52
+ "step": 120
53
+ },
54
+ {
55
+ "epoch": 0.009437141894169195,
56
+ "grad_norm": 0.94921875,
57
+ "learning_rate": 0.00019460000000000001,
58
+ "loss": 6.5162200927734375,
59
+ "step": 140
60
+ },
61
+ {
62
+ "epoch": 0.010785305021907651,
63
+ "grad_norm": 0.796875,
64
+ "learning_rate": 0.0002226,
65
+ "loss": 6.237432861328125,
66
+ "step": 160
67
+ },
68
+ {
69
+ "epoch": 0.012133468149646108,
70
+ "grad_norm": 0.71875,
71
+ "learning_rate": 0.00025059999999999997,
72
+ "loss": 6.039522552490235,
73
+ "step": 180
74
+ },
75
+ {
76
+ "epoch": 0.013481631277384564,
77
+ "grad_norm": 0.57421875,
78
+ "learning_rate": 0.0002786,
79
+ "loss": 5.8798988342285154,
80
+ "step": 200
81
+ },
82
+ {
83
+ "epoch": 0.01482979440512302,
84
+ "grad_norm": 1.90625,
85
+ "learning_rate": 0.00030659999999999997,
86
+ "loss": 5.723947143554687,
87
+ "step": 220
88
+ },
89
+ {
90
+ "epoch": 0.016177957532861477,
91
+ "grad_norm": 0.6796875,
92
+ "learning_rate": 0.0003346,
93
+ "loss": 5.551593399047851,
94
+ "step": 240
95
+ },
96
+ {
97
+ "epoch": 0.01752612066059993,
98
+ "grad_norm": 1.6875,
99
+ "learning_rate": 0.00036260000000000003,
100
+ "loss": 5.3516895294189455,
101
+ "step": 260
102
+ },
103
+ {
104
+ "epoch": 0.01887428378833839,
105
+ "grad_norm": 0.84765625,
106
+ "learning_rate": 0.0003906,
107
+ "loss": 5.130373764038086,
108
+ "step": 280
109
+ },
110
+ {
111
+ "epoch": 0.020222446916076844,
112
+ "grad_norm": 0.55859375,
113
+ "learning_rate": 0.0004186,
114
+ "loss": 4.882920455932617,
115
+ "step": 300
116
+ },
117
+ {
118
+ "epoch": 0.021570610043815303,
119
+ "grad_norm": 1.15625,
120
+ "learning_rate": 0.0004466,
121
+ "loss": 4.673929977416992,
122
+ "step": 320
123
+ },
124
+ {
125
+ "epoch": 0.022918773171553757,
126
+ "grad_norm": 1.0,
127
+ "learning_rate": 0.00047460000000000004,
128
+ "loss": 4.511159515380859,
129
+ "step": 340
130
+ },
131
+ {
132
+ "epoch": 0.024266936299292215,
133
+ "grad_norm": 0.96875,
134
+ "learning_rate": 0.0005026,
135
+ "loss": 4.360583114624023,
136
+ "step": 360
137
+ },
138
+ {
139
+ "epoch": 0.02561509942703067,
140
+ "grad_norm": 0.80078125,
141
+ "learning_rate": 0.0005306,
142
+ "loss": 4.208797836303711,
143
+ "step": 380
144
+ },
145
+ {
146
+ "epoch": 0.026963262554769128,
147
+ "grad_norm": 0.703125,
148
+ "learning_rate": 0.0005586,
149
+ "loss": 4.059257507324219,
150
+ "step": 400
151
+ },
152
+ {
153
+ "epoch": 0.028311425682507583,
154
+ "grad_norm": 1.390625,
155
+ "learning_rate": 0.0005866,
156
+ "loss": 3.945643997192383,
157
+ "step": 420
158
+ },
159
+ {
160
+ "epoch": 0.02965958881024604,
161
+ "grad_norm": 0.8046875,
162
+ "learning_rate": 0.0006146,
163
+ "loss": 3.878635787963867,
164
+ "step": 440
165
+ },
166
+ {
167
+ "epoch": 0.031007751937984496,
168
+ "grad_norm": 0.7578125,
169
+ "learning_rate": 0.0006426,
170
+ "loss": 3.763169860839844,
171
+ "step": 460
172
+ },
173
+ {
174
+ "epoch": 0.032355915065722954,
175
+ "grad_norm": 0.578125,
176
+ "learning_rate": 0.0006705999999999999,
177
+ "loss": 3.680609130859375,
178
+ "step": 480
179
+ },
180
+ {
181
+ "epoch": 0.03370407819346141,
182
+ "grad_norm": 0.6484375,
183
+ "learning_rate": 0.0006986,
184
+ "loss": 3.6085578918457033,
185
+ "step": 500
186
+ },
187
+ {
188
+ "epoch": 0.03505224132119986,
189
+ "grad_norm": 0.431640625,
190
+ "learning_rate": 0.0007,
191
+ "loss": 3.5324871063232424,
192
+ "step": 520
193
+ },
194
+ {
195
+ "epoch": 0.03640040444893832,
196
+ "grad_norm": 0.4453125,
197
+ "learning_rate": 0.0007,
198
+ "loss": 3.454126739501953,
199
+ "step": 540
200
+ },
201
+ {
202
+ "epoch": 0.03774856757667678,
203
+ "grad_norm": 0.546875,
204
+ "learning_rate": 0.0007,
205
+ "loss": 3.4041587829589846,
206
+ "step": 560
207
+ },
208
+ {
209
+ "epoch": 0.03909673070441524,
210
+ "grad_norm": 0.46875,
211
+ "learning_rate": 0.0007,
212
+ "loss": 3.347498321533203,
213
+ "step": 580
214
+ },
215
+ {
216
+ "epoch": 0.04044489383215369,
217
+ "grad_norm": 0.546875,
218
+ "learning_rate": 0.0007,
219
+ "loss": 3.313167190551758,
220
+ "step": 600
221
+ },
222
+ {
223
+ "epoch": 0.04179305695989215,
224
+ "grad_norm": 0.51953125,
225
+ "learning_rate": 0.0007,
226
+ "loss": 3.2490795135498045,
227
+ "step": 620
228
+ },
229
+ {
230
+ "epoch": 0.043141220087630605,
231
+ "grad_norm": 0.453125,
232
+ "learning_rate": 0.0007,
233
+ "loss": 3.221387481689453,
234
+ "step": 640
235
+ },
236
+ {
237
+ "epoch": 0.044489383215369056,
238
+ "grad_norm": 0.5234375,
239
+ "learning_rate": 0.0007,
240
+ "loss": 3.1599220275878905,
241
+ "step": 660
242
+ },
243
+ {
244
+ "epoch": 0.045837546343107514,
245
+ "grad_norm": 0.4375,
246
+ "learning_rate": 0.0007,
247
+ "loss": 3.133222198486328,
248
+ "step": 680
249
+ },
250
+ {
251
+ "epoch": 0.04718570947084597,
252
+ "grad_norm": 0.5703125,
253
+ "learning_rate": 0.0007,
254
+ "loss": 3.0884355545043944,
255
+ "step": 700
256
+ },
257
+ {
258
+ "epoch": 0.04853387259858443,
259
+ "grad_norm": 0.431640625,
260
+ "learning_rate": 0.0007,
261
+ "loss": 3.0560993194580077,
262
+ "step": 720
263
+ },
264
+ {
265
+ "epoch": 0.04988203572632288,
266
+ "grad_norm": 0.462890625,
267
+ "learning_rate": 0.0007,
268
+ "loss": 3.0169742584228514,
269
+ "step": 740
270
+ },
271
+ {
272
+ "epoch": 0.05123019885406134,
273
+ "grad_norm": 0.390625,
274
+ "learning_rate": 0.0007,
275
+ "loss": 2.9988849639892576,
276
+ "step": 760
277
+ },
278
+ {
279
+ "epoch": 0.0525783619817998,
280
+ "grad_norm": 0.466796875,
281
+ "learning_rate": 0.0007,
282
+ "loss": 3.004904937744141,
283
+ "step": 780
284
+ },
285
+ {
286
+ "epoch": 0.053926525109538256,
287
+ "grad_norm": 0.443359375,
288
+ "learning_rate": 0.0007,
289
+ "loss": 2.9500053405761717,
290
+ "step": 800
291
+ },
292
+ {
293
+ "epoch": 0.05527468823727671,
294
+ "grad_norm": 0.453125,
295
+ "learning_rate": 0.0007,
296
+ "loss": 2.93444938659668,
297
+ "step": 820
298
+ },
299
+ {
300
+ "epoch": 0.056622851365015166,
301
+ "grad_norm": 0.421875,
302
+ "learning_rate": 0.0007,
303
+ "loss": 2.9036865234375,
304
+ "step": 840
305
+ },
306
+ {
307
+ "epoch": 0.057971014492753624,
308
+ "grad_norm": 0.392578125,
309
+ "learning_rate": 0.0007,
310
+ "loss": 2.859415054321289,
311
+ "step": 860
312
+ },
313
+ {
314
+ "epoch": 0.05931917762049208,
315
+ "grad_norm": 0.42578125,
316
+ "learning_rate": 0.0007,
317
+ "loss": 2.8376497268676757,
318
+ "step": 880
319
+ },
320
+ {
321
+ "epoch": 0.06066734074823053,
322
+ "grad_norm": 0.46484375,
323
+ "learning_rate": 0.0007,
324
+ "loss": 2.821657180786133,
325
+ "step": 900
326
+ },
327
+ {
328
+ "epoch": 0.06201550387596899,
329
+ "grad_norm": 0.46875,
330
+ "learning_rate": 0.0007,
331
+ "loss": 2.7936939239501952,
332
+ "step": 920
333
+ },
334
+ {
335
+ "epoch": 0.06336366700370745,
336
+ "grad_norm": 0.40234375,
337
+ "learning_rate": 0.0007,
338
+ "loss": 2.779148292541504,
339
+ "step": 940
340
+ },
341
+ {
342
+ "epoch": 0.06471183013144591,
343
+ "grad_norm": 0.40234375,
344
+ "learning_rate": 0.0007,
345
+ "loss": 2.7523166656494142,
346
+ "step": 960
347
+ },
348
+ {
349
+ "epoch": 0.06605999325918437,
350
+ "grad_norm": 0.408203125,
351
+ "learning_rate": 0.0007,
352
+ "loss": 2.7326622009277344,
353
+ "step": 980
354
+ },
355
+ {
356
+ "epoch": 0.06740815638692282,
357
+ "grad_norm": 0.396484375,
358
+ "learning_rate": 0.0007,
359
+ "loss": 2.730000305175781,
360
+ "step": 1000
361
+ },
362
+ {
363
+ "epoch": 0.06875631951466127,
364
+ "grad_norm": 0.3828125,
365
+ "learning_rate": 0.0007,
366
+ "loss": 2.6984880447387694,
367
+ "step": 1020
368
+ },
369
+ {
370
+ "epoch": 0.07010448264239973,
371
+ "grad_norm": 0.423828125,
372
+ "learning_rate": 0.0007,
373
+ "loss": 2.6830041885375975,
374
+ "step": 1040
375
+ },
376
+ {
377
+ "epoch": 0.07145264577013818,
378
+ "grad_norm": 0.404296875,
379
+ "learning_rate": 0.0007,
380
+ "loss": 2.6666488647460938,
381
+ "step": 1060
382
+ },
383
+ {
384
+ "epoch": 0.07280080889787664,
385
+ "grad_norm": 0.404296875,
386
+ "learning_rate": 0.0007,
387
+ "loss": 2.6461860656738283,
388
+ "step": 1080
389
+ },
390
+ {
391
+ "epoch": 0.0741489720256151,
392
+ "grad_norm": 0.427734375,
393
+ "learning_rate": 0.0007,
394
+ "loss": 2.638811492919922,
395
+ "step": 1100
396
+ },
397
+ {
398
+ "epoch": 0.07549713515335356,
399
+ "grad_norm": 0.38671875,
400
+ "learning_rate": 0.0007,
401
+ "loss": 2.604878807067871,
402
+ "step": 1120
403
+ },
404
+ {
405
+ "epoch": 0.07684529828109202,
406
+ "grad_norm": 0.423828125,
407
+ "learning_rate": 0.0007,
408
+ "loss": 2.5887590408325196,
409
+ "step": 1140
410
+ },
411
+ {
412
+ "epoch": 0.07819346140883048,
413
+ "grad_norm": 0.416015625,
414
+ "learning_rate": 0.0007,
415
+ "loss": 2.5864500045776366,
416
+ "step": 1160
417
+ },
418
+ {
419
+ "epoch": 0.07954162453656892,
420
+ "grad_norm": 0.40234375,
421
+ "learning_rate": 0.0007,
422
+ "loss": 2.5788290023803713,
423
+ "step": 1180
424
+ },
425
+ {
426
+ "epoch": 0.08088978766430738,
427
+ "grad_norm": 0.404296875,
428
+ "learning_rate": 0.0007,
429
+ "loss": 2.570602607727051,
430
+ "step": 1200
431
+ },
432
+ {
433
+ "epoch": 0.08223795079204584,
434
+ "grad_norm": 0.390625,
435
+ "learning_rate": 0.0007,
436
+ "loss": 2.5528938293457033,
437
+ "step": 1220
438
+ },
439
+ {
440
+ "epoch": 0.0835861139197843,
441
+ "grad_norm": 0.373046875,
442
+ "learning_rate": 0.0007,
443
+ "loss": 2.53753662109375,
444
+ "step": 1240
445
+ },
446
+ {
447
+ "epoch": 0.08493427704752275,
448
+ "grad_norm": 0.384765625,
449
+ "learning_rate": 0.0007,
450
+ "loss": 2.523891830444336,
451
+ "step": 1260
452
+ },
453
+ {
454
+ "epoch": 0.08628244017526121,
455
+ "grad_norm": 0.3671875,
456
+ "learning_rate": 0.0007,
457
+ "loss": 2.5049261093139648,
458
+ "step": 1280
459
+ },
460
+ {
461
+ "epoch": 0.08763060330299967,
462
+ "grad_norm": 0.392578125,
463
+ "learning_rate": 0.0007,
464
+ "loss": 2.4961641311645506,
465
+ "step": 1300
466
+ },
467
+ {
468
+ "epoch": 0.08897876643073811,
469
+ "grad_norm": 0.353515625,
470
+ "learning_rate": 0.0007,
471
+ "loss": 2.487112045288086,
472
+ "step": 1320
473
+ },
474
+ {
475
+ "epoch": 0.09032692955847657,
476
+ "grad_norm": 0.38671875,
477
+ "learning_rate": 0.0007,
478
+ "loss": 2.471820259094238,
479
+ "step": 1340
480
+ },
481
+ {
482
+ "epoch": 0.09167509268621503,
483
+ "grad_norm": 0.392578125,
484
+ "learning_rate": 0.0007,
485
+ "loss": 2.4697391510009767,
486
+ "step": 1360
487
+ },
488
+ {
489
+ "epoch": 0.09302325581395349,
490
+ "grad_norm": 0.369140625,
491
+ "learning_rate": 0.0007,
492
+ "loss": 2.4492652893066404,
493
+ "step": 1380
494
+ },
495
+ {
496
+ "epoch": 0.09437141894169195,
497
+ "grad_norm": 0.384765625,
498
+ "learning_rate": 0.0007,
499
+ "loss": 2.4387527465820313,
500
+ "step": 1400
501
+ },
502
+ {
503
+ "epoch": 0.0957195820694304,
504
+ "grad_norm": 0.380859375,
505
+ "learning_rate": 0.0007,
506
+ "loss": 2.4479516983032226,
507
+ "step": 1420
508
+ },
509
+ {
510
+ "epoch": 0.09706774519716886,
511
+ "grad_norm": 0.359375,
512
+ "learning_rate": 0.0007,
513
+ "loss": 2.4246395111083983,
514
+ "step": 1440
515
+ },
516
+ {
517
+ "epoch": 0.09841590832490732,
518
+ "grad_norm": 0.373046875,
519
+ "learning_rate": 0.0007,
520
+ "loss": 2.4114765167236327,
521
+ "step": 1460
522
+ },
523
+ {
524
+ "epoch": 0.09976407145264576,
525
+ "grad_norm": 0.375,
526
+ "learning_rate": 0.0007,
527
+ "loss": 2.399285888671875,
528
+ "step": 1480
529
+ },
530
+ {
531
+ "epoch": 0.10111223458038422,
532
+ "grad_norm": 0.37890625,
533
+ "learning_rate": 0.0007,
534
+ "loss": 2.3839179992675783,
535
+ "step": 1500
536
+ },
537
+ {
538
+ "epoch": 0.10246039770812268,
539
+ "grad_norm": 0.41796875,
540
+ "learning_rate": 0.0007,
541
+ "loss": 2.3861133575439455,
542
+ "step": 1520
543
+ },
544
+ {
545
+ "epoch": 0.10380856083586114,
546
+ "grad_norm": 0.40625,
547
+ "learning_rate": 0.0007,
548
+ "loss": 2.3732574462890623,
549
+ "step": 1540
550
+ },
551
+ {
552
+ "epoch": 0.1051567239635996,
553
+ "grad_norm": 0.36328125,
554
+ "learning_rate": 0.0007,
555
+ "loss": 2.3624162673950195,
556
+ "step": 1560
557
+ },
558
+ {
559
+ "epoch": 0.10650488709133805,
560
+ "grad_norm": 0.359375,
561
+ "learning_rate": 0.0007,
562
+ "loss": 2.3779842376708986,
563
+ "step": 1580
564
+ },
565
+ {
566
+ "epoch": 0.10785305021907651,
567
+ "grad_norm": 0.369140625,
568
+ "learning_rate": 0.0007,
569
+ "loss": 2.3441116333007814,
570
+ "step": 1600
571
+ },
572
+ {
573
+ "epoch": 0.10920121334681497,
574
+ "grad_norm": 0.376953125,
575
+ "learning_rate": 0.0007,
576
+ "loss": 2.3418384552001954,
577
+ "step": 1620
578
+ },
579
+ {
580
+ "epoch": 0.11054937647455342,
581
+ "grad_norm": 0.373046875,
582
+ "learning_rate": 0.0007,
583
+ "loss": 2.3539493560791014,
584
+ "step": 1640
585
+ },
586
+ {
587
+ "epoch": 0.11189753960229187,
588
+ "grad_norm": 0.359375,
589
+ "learning_rate": 0.0007,
590
+ "loss": 2.332906150817871,
591
+ "step": 1660
592
+ },
593
+ {
594
+ "epoch": 0.11324570273003033,
595
+ "grad_norm": 0.408203125,
596
+ "learning_rate": 0.0007,
597
+ "loss": 2.3174556732177733,
598
+ "step": 1680
599
+ },
600
+ {
601
+ "epoch": 0.11459386585776879,
602
+ "grad_norm": 0.361328125,
603
+ "learning_rate": 0.0007,
604
+ "loss": 2.330640411376953,
605
+ "step": 1700
606
+ },
607
+ {
608
+ "epoch": 0.11594202898550725,
609
+ "grad_norm": 0.423828125,
610
+ "learning_rate": 0.0007,
611
+ "loss": 2.306385803222656,
612
+ "step": 1720
613
+ },
614
+ {
615
+ "epoch": 0.1172901921132457,
616
+ "grad_norm": 0.365234375,
617
+ "learning_rate": 0.0007,
618
+ "loss": 2.2937660217285156,
619
+ "step": 1740
620
+ },
621
+ {
622
+ "epoch": 0.11863835524098416,
623
+ "grad_norm": 0.3671875,
624
+ "learning_rate": 0.0007,
625
+ "loss": 2.297491264343262,
626
+ "step": 1760
627
+ },
628
+ {
629
+ "epoch": 0.11998651836872262,
630
+ "grad_norm": 0.337890625,
631
+ "learning_rate": 0.0007,
632
+ "loss": 2.312083053588867,
633
+ "step": 1780
634
+ },
635
+ {
636
+ "epoch": 0.12133468149646107,
637
+ "grad_norm": 0.357421875,
638
+ "learning_rate": 0.0007,
639
+ "loss": 2.307674789428711,
640
+ "step": 1800
641
+ },
642
+ {
643
+ "epoch": 0.12268284462419952,
644
+ "grad_norm": 0.375,
645
+ "learning_rate": 0.0007,
646
+ "loss": 2.296724319458008,
647
+ "step": 1820
648
+ },
649
+ {
650
+ "epoch": 0.12403100775193798,
651
+ "grad_norm": 0.376953125,
652
+ "learning_rate": 0.0007,
653
+ "loss": 2.2644861221313475,
654
+ "step": 1840
655
+ },
656
+ {
657
+ "epoch": 0.12537917087967643,
658
+ "grad_norm": 0.353515625,
659
+ "learning_rate": 0.0007,
660
+ "loss": 2.2662803649902346,
661
+ "step": 1860
662
+ },
663
+ {
664
+ "epoch": 0.1267273340074149,
665
+ "grad_norm": 0.33984375,
666
+ "learning_rate": 0.0007,
667
+ "loss": 2.2558313369750977,
668
+ "step": 1880
669
+ },
670
+ {
671
+ "epoch": 0.12807549713515334,
672
+ "grad_norm": 0.365234375,
673
+ "learning_rate": 0.0007,
674
+ "loss": 2.262881851196289,
675
+ "step": 1900
676
+ },
677
+ {
678
+ "epoch": 0.12942366026289182,
679
+ "grad_norm": 0.349609375,
680
+ "learning_rate": 0.0007,
681
+ "loss": 2.2384838104248046,
682
+ "step": 1920
683
+ },
684
+ {
685
+ "epoch": 0.13077182339063026,
686
+ "grad_norm": 0.361328125,
687
+ "learning_rate": 0.0007,
688
+ "loss": 2.247391700744629,
689
+ "step": 1940
690
+ },
691
+ {
692
+ "epoch": 0.13211998651836873,
693
+ "grad_norm": 0.34765625,
694
+ "learning_rate": 0.0007,
695
+ "loss": 2.2225168228149412,
696
+ "step": 1960
697
+ },
698
+ {
699
+ "epoch": 0.13346814964610718,
700
+ "grad_norm": 0.380859375,
701
+ "learning_rate": 0.0007,
702
+ "loss": 2.2470256805419924,
703
+ "step": 1980
704
+ },
705
+ {
706
+ "epoch": 0.13481631277384565,
707
+ "grad_norm": 0.3359375,
708
+ "learning_rate": 0.0007,
709
+ "loss": 2.218878173828125,
710
+ "step": 2000
711
+ },
712
+ {
713
+ "epoch": 0.1361644759015841,
714
+ "grad_norm": 0.357421875,
715
+ "learning_rate": 0.0007,
716
+ "loss": 2.227268409729004,
717
+ "step": 2020
718
+ },
719
+ {
720
+ "epoch": 0.13751263902932254,
721
+ "grad_norm": 0.380859375,
722
+ "learning_rate": 0.0007,
723
+ "loss": 2.2079793930053713,
724
+ "step": 2040
725
+ },
726
+ {
727
+ "epoch": 0.138860802157061,
728
+ "grad_norm": 0.380859375,
729
+ "learning_rate": 0.0007,
730
+ "loss": 2.2262474060058595,
731
+ "step": 2060
732
+ },
733
+ {
734
+ "epoch": 0.14020896528479945,
735
+ "grad_norm": 0.349609375,
736
+ "learning_rate": 0.0007,
737
+ "loss": 2.2017698287963867,
738
+ "step": 2080
739
+ },
740
+ {
741
+ "epoch": 0.14155712841253792,
742
+ "grad_norm": 0.333984375,
743
+ "learning_rate": 0.0007,
744
+ "loss": 2.19287223815918,
745
+ "step": 2100
746
+ },
747
+ {
748
+ "epoch": 0.14290529154027637,
749
+ "grad_norm": 0.349609375,
750
+ "learning_rate": 0.0007,
751
+ "loss": 2.201098823547363,
752
+ "step": 2120
753
+ },
754
+ {
755
+ "epoch": 0.14425345466801484,
756
+ "grad_norm": 0.333984375,
757
+ "learning_rate": 0.0007,
758
+ "loss": 2.2028568267822264,
759
+ "step": 2140
760
+ },
761
+ {
762
+ "epoch": 0.14560161779575329,
763
+ "grad_norm": 0.353515625,
764
+ "learning_rate": 0.0007,
765
+ "loss": 2.1869789123535157,
766
+ "step": 2160
767
+ },
768
+ {
769
+ "epoch": 0.14694978092349173,
770
+ "grad_norm": 0.33984375,
771
+ "learning_rate": 0.0007,
772
+ "loss": 2.194090461730957,
773
+ "step": 2180
774
+ },
775
+ {
776
+ "epoch": 0.1482979440512302,
777
+ "grad_norm": 0.333984375,
778
+ "learning_rate": 0.0007,
779
+ "loss": 2.169504165649414,
780
+ "step": 2200
781
+ },
782
+ {
783
+ "epoch": 0.14964610717896865,
784
+ "grad_norm": 0.359375,
785
+ "learning_rate": 0.0007,
786
+ "loss": 2.17819766998291,
787
+ "step": 2220
788
+ },
789
+ {
790
+ "epoch": 0.15099427030670712,
791
+ "grad_norm": 0.361328125,
792
+ "learning_rate": 0.0007,
793
+ "loss": 2.1870670318603516,
794
+ "step": 2240
795
+ },
796
+ {
797
+ "epoch": 0.15234243343444556,
798
+ "grad_norm": 0.359375,
799
+ "learning_rate": 0.0007,
800
+ "loss": 2.1656898498535155,
801
+ "step": 2260
802
+ },
803
+ {
804
+ "epoch": 0.15369059656218403,
805
+ "grad_norm": 0.369140625,
806
+ "learning_rate": 0.0007,
807
+ "loss": 2.174662971496582,
808
+ "step": 2280
809
+ },
810
+ {
811
+ "epoch": 0.15503875968992248,
812
+ "grad_norm": 0.3515625,
813
+ "learning_rate": 0.0007,
814
+ "loss": 2.1561347961425783,
815
+ "step": 2300
816
+ },
817
+ {
818
+ "epoch": 0.15638692281766095,
819
+ "grad_norm": 0.33203125,
820
+ "learning_rate": 0.0007,
821
+ "loss": 2.16455020904541,
822
+ "step": 2320
823
+ },
824
+ {
825
+ "epoch": 0.1577350859453994,
826
+ "grad_norm": 0.330078125,
827
+ "learning_rate": 0.0007,
828
+ "loss": 2.1551021575927733,
829
+ "step": 2340
830
+ },
831
+ {
832
+ "epoch": 0.15908324907313784,
833
+ "grad_norm": 0.341796875,
834
+ "learning_rate": 0.0007,
835
+ "loss": 2.157507133483887,
836
+ "step": 2360
837
+ },
838
+ {
839
+ "epoch": 0.1604314122008763,
840
+ "grad_norm": 0.33984375,
841
+ "learning_rate": 0.0007,
842
+ "loss": 2.1596670150756836,
843
+ "step": 2380
844
+ },
845
+ {
846
+ "epoch": 0.16177957532861476,
847
+ "grad_norm": 0.357421875,
848
+ "learning_rate": 0.0007,
849
+ "loss": 2.1417112350463867,
850
+ "step": 2400
851
+ },
852
+ {
853
+ "epoch": 0.16312773845635323,
854
+ "grad_norm": 0.349609375,
855
+ "learning_rate": 0.0007,
856
+ "loss": 2.1463193893432617,
857
+ "step": 2420
858
+ },
859
+ {
860
+ "epoch": 0.16447590158409167,
861
+ "grad_norm": 0.3359375,
862
+ "learning_rate": 0.0007,
863
+ "loss": 2.133936882019043,
864
+ "step": 2440
865
+ },
866
+ {
867
+ "epoch": 0.16582406471183014,
868
+ "grad_norm": 0.328125,
869
+ "learning_rate": 0.0007,
870
+ "loss": 2.1435129165649416,
871
+ "step": 2460
872
+ },
873
+ {
874
+ "epoch": 0.1671722278395686,
875
+ "grad_norm": 0.341796875,
876
+ "learning_rate": 0.0007,
877
+ "loss": 2.1443099975585938,
878
+ "step": 2480
879
+ },
880
+ {
881
+ "epoch": 0.1683855746545332,
882
+ "eval_loss": 2.138087511062622,
883
+ "eval_runtime": 20.1123,
884
+ "eval_samples_per_second": 473.691,
885
+ "eval_steps_per_second": 0.497,
886
+ "step": 2498
887
+ },
888
+ {
889
+ "epoch": 0.16852039096730703,
890
+ "grad_norm": 0.33984375,
891
+ "learning_rate": 0.0007,
892
+ "loss": 2.126860427856445,
893
+ "step": 2500
894
+ },
895
+ {
896
+ "epoch": 0.16852039096730703,
897
+ "eval_loss": 2.1375532150268555,
898
+ "eval_runtime": 20.4047,
899
+ "eval_samples_per_second": 466.903,
900
+ "eval_steps_per_second": 0.49,
901
+ "step": 2500
902
+ },
903
+ {
904
+ "epoch": 0.1698685540950455,
905
+ "grad_norm": 24903680.0,
906
+ "learning_rate": 3.4200000000000005e-05,
907
+ "loss": 2.128565788269043,
908
+ "step": 2520
909
+ },
910
+ {
911
+ "epoch": 0.17121671722278395,
912
+ "grad_norm": 27000832.0,
913
+ "learning_rate": 7.020000000000001e-05,
914
+ "loss": 2.1492637634277343,
915
+ "step": 2540
916
+ },
917
+ {
918
+ "epoch": 0.17256488035052242,
919
+ "grad_norm": 119537664.0,
920
+ "learning_rate": 9e-05,
921
+ "loss": 3.623552703857422,
922
+ "step": 2560
923
+ },
924
+ {
925
+ "epoch": 0.17391304347826086,
926
+ "grad_norm": 301989888.0,
927
+ "learning_rate": 9e-05,
928
+ "loss": 6.8911598205566404,
929
+ "step": 2580
930
+ },
931
+ {
932
+ "epoch": 0.17526120660599934,
933
+ "grad_norm": 658505728.0,
934
+ "learning_rate": 9e-05,
935
+ "loss": 7.545132446289062,
936
+ "step": 2600
937
+ },
938
+ {
939
+ "epoch": 0.17660936973373778,
940
+ "grad_norm": 1166016512.0,
941
+ "learning_rate": 9e-05,
942
+ "loss": 7.784053039550781,
943
+ "step": 2620
944
+ },
945
+ {
946
+ "epoch": 0.17795753286147623,
947
+ "grad_norm": 1769996288.0,
948
+ "learning_rate": 9e-05,
949
+ "loss": 7.88036880493164,
950
+ "step": 2640
951
+ },
952
+ {
953
+ "epoch": 0.1793056959892147,
954
+ "grad_norm": 2415919104.0,
955
+ "learning_rate": 9e-05,
956
+ "loss": 7.926078033447266,
957
+ "step": 2660
958
+ },
959
+ {
960
+ "epoch": 0.18065385911695314,
961
+ "grad_norm": 3053453312.0,
962
+ "learning_rate": 9e-05,
963
+ "loss": 7.947563934326172,
964
+ "step": 2680
965
+ },
966
+ {
967
+ "epoch": 0.1820020222446916,
968
+ "grad_norm": 3674210304.0,
969
+ "learning_rate": 9e-05,
970
+ "loss": 7.972737121582031,
971
+ "step": 2700
972
+ },
973
+ {
974
+ "epoch": 0.18335018537243006,
975
+ "grad_norm": 4278190080.0,
976
+ "learning_rate": 9e-05,
977
+ "loss": 7.9804527282714846,
978
+ "step": 2720
979
+ },
980
+ {
981
+ "epoch": 0.18469834850016853,
982
+ "grad_norm": 4831838208.0,
983
+ "learning_rate": 9e-05,
984
+ "loss": 7.980913543701172,
985
+ "step": 2740
986
+ },
987
+ {
988
+ "epoch": 0.18604651162790697,
989
+ "grad_norm": 5335154688.0,
990
+ "learning_rate": 9e-05,
991
+ "loss": 7.987354278564453,
992
+ "step": 2760
993
+ },
994
+ {
995
+ "epoch": 0.18739467475564545,
996
+ "grad_norm": 5804916736.0,
997
+ "learning_rate": 9e-05,
998
+ "loss": 8.005199432373047,
999
+ "step": 2780
1000
+ },
1001
+ {
1002
+ "epoch": 0.1887428378833839,
1003
+ "grad_norm": 6207569920.0,
1004
+ "learning_rate": 9e-05,
1005
+ "loss": 8.00537338256836,
1006
+ "step": 2800
1007
+ },
1008
+ {
1009
+ "epoch": 0.19009100101112233,
1010
+ "grad_norm": 6610223104.0,
1011
+ "learning_rate": 9e-05,
1012
+ "loss": 8.006069946289063,
1013
+ "step": 2820
1014
+ },
1015
+ {
1016
+ "epoch": 0.1914391641388608,
1017
+ "grad_norm": 6945767424.0,
1018
+ "learning_rate": 9e-05,
1019
+ "loss": 8.004747009277343,
1020
+ "step": 2840
1021
+ },
1022
+ {
1023
+ "epoch": 0.19278732726659925,
1024
+ "grad_norm": 7281311744.0,
1025
+ "learning_rate": 9e-05,
1026
+ "loss": 7.9950706481933596,
1027
+ "step": 2860
1028
+ },
1029
+ {
1030
+ "epoch": 0.19413549039433772,
1031
+ "grad_norm": 7549747200.0,
1032
+ "learning_rate": 9e-05,
1033
+ "loss": 8.006790161132812,
1034
+ "step": 2880
1035
+ },
1036
+ {
1037
+ "epoch": 0.19548365352207617,
1038
+ "grad_norm": 7818182656.0,
1039
+ "learning_rate": 9e-05,
1040
+ "loss": 8.008560180664062,
1041
+ "step": 2900
1042
+ },
1043
+ {
1044
+ "epoch": 0.19683181664981464,
1045
+ "grad_norm": 8086618112.0,
1046
+ "learning_rate": 9e-05,
1047
+ "loss": 8.012400054931641,
1048
+ "step": 2920
1049
+ },
1050
+ {
1051
+ "epoch": 0.19817997977755308,
1052
+ "grad_norm": 8287944704.0,
1053
+ "learning_rate": 9e-05,
1054
+ "loss": 8.018167114257812,
1055
+ "step": 2940
1056
+ },
1057
+ {
1058
+ "epoch": 0.19952814290529153,
1059
+ "grad_norm": 8522825728.0,
1060
+ "learning_rate": 9e-05,
1061
+ "loss": 8.022463989257812,
1062
+ "step": 2960
1063
+ },
1064
+ {
1065
+ "epoch": 0.20087630603303,
1066
+ "grad_norm": 8724152320.0,
1067
+ "learning_rate": 9e-05,
1068
+ "loss": 8.007749938964844,
1069
+ "step": 2980
1070
+ },
1071
+ {
1072
+ "epoch": 0.20222446916076844,
1073
+ "grad_norm": 8858370048.0,
1074
+ "learning_rate": 9e-05,
1075
+ "loss": 8.007595825195313,
1076
+ "step": 3000
1077
+ }
1078
+ ],
1079
+ "logging_steps": 20,
1080
+ "max_steps": 3500,
1081
+ "num_input_tokens_seen": 0,
1082
+ "num_train_epochs": 1,
1083
+ "save_steps": 100,
1084
+ "stateful_callbacks": {
1085
+ "TrainerControl": {
1086
+ "args": {
1087
+ "should_epoch_stop": false,
1088
+ "should_evaluate": false,
1089
+ "should_log": false,
1090
+ "should_save": true,
1091
+ "should_training_stop": false
1092
+ },
1093
+ "attributes": {}
1094
+ }
1095
+ },
1096
+ "total_flos": 9678851407872000.0,
1097
+ "train_batch_size": 64,
1098
+ "trial_name": null,
1099
+ "trial_params": null
1100
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f62f8186f489a47bc68f80ec9f826cd67dc6a0232e67ff9d099ed1157b6e25f
3
+ size 4920
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3100/config.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation": "powlu",
3
+ "architectures": [
4
+ "TinyLlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 1,
9
+ "dtype": "bfloat16",
10
+ "eos_token_id": 2,
11
+ "head_dim": 32,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 128,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 256,
16
+ "max_position_embeddings": 512,
17
+ "mlp_bias": false,
18
+ "mlp_type": "glu",
19
+ "model_type": "tiny_llama",
20
+ "num_attention_heads": 4,
21
+ "num_hidden_layers": 100,
22
+ "num_key_value_heads": 4,
23
+ "pad_token_id": 0,
24
+ "powlu_m": 3.0,
25
+ "pretraining_tp": 1,
26
+ "rms_norm_eps": 1e-06,
27
+ "rope_parameters": {
28
+ "rope_theta": 10000.0,
29
+ "rope_type": "default"
30
+ },
31
+ "tie_word_embeddings": true,
32
+ "tokenizer_name": "w-ahmad/tiny-stories-tokenizer",
33
+ "transformers_version": "5.16.0.dev0",
34
+ "use_cache": false,
35
+ "vocab_size": 4096,
36
+ "waleed_beta": 10
37
+ }
zain/Activation/out/glu-powlu-100L_trash_run/checkpoint-3100/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cdebc899d8a3c3c58a38b03b5070828f856d75411c5ffa2d49598202a89ad4eb
3
+ size 33967272