Hugging Face's logo

Sim4Rec
/
inter-play-sim-assistant-sft-qwen3-8B

Text Generation
Transformers
TensorBoard
Safetensors
qwen3
Generated from Trainer
trl
sft
conversational
text-generation-inference
Model card Files
xet
Metrics Community
jeromeramos commited on
Commit
df03c6c
·
verified ·
1 Parent(s): e811d7f

End of training

Browse files
Files changed (17) hide show
  1. README.md +1 -1
  2. added_tokens.json +13 -12
  3. all_results.json +8 -3
  4. config.json +1 -1
  5. eval_results.json +8 -0
  6. model-00001-of-00004.safetensors +1 -1
  7. model-00002-of-00004.safetensors +1 -1
  8. model-00003-of-00004.safetensors +1 -1
  9. model-00004-of-00004.safetensors +1 -1
  10. runs/May17_01-35-22_w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-56x5fg8/events.out.tfevents.1747457689.w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-56x5fg8.10629.1 +3 -0
  11. runs/May18_01-21-40_w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-78pqvlh/events.out.tfevents.1747531364.w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-78pqvlh.46539.0 +3 -0
  12. special_tokens_map.json +7 -1
  13. tokenizer.json +2 -2
  14. tokenizer_config.json +20 -12
  15. train_results.json +3 -3
  16. trainer_state.json +588 -588
  17. training_args.bin +1 -1
README.md CHANGED
@@ -27,7 +27,7 @@ print(output["generated_text"])
27
 
28
  ## Training procedure
29
 
30
- [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/jerome-ramos-20/huggingface/runs/uiam4kxx)
31
 
32
 
33
  This model was trained with SFT.
 
27
 
28
  ## Training procedure
29
 
30
+ [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/jerome-ramos-20/huggingface/runs/1p9l3u4m)
31
 
32
 
33
  This model was trained with SFT.
added_tokens.json CHANGED
@@ -1,19 +1,19 @@
1
  {
2
- "</action>": 151672,
3
- "</movie_title>": 151680,
4
- "</response>": 151670,
5
  "</think>": 151668,
6
  "</tool_call>": 151658,
7
  "</tool_response>": 151666,
8
- "<accept>": 151678,
9
- "<action>": 151671,
10
- "<disclose-goal>": 151676,
11
- "<feedback>": 151677,
12
- "<greeting>": 151675,
13
- "<inquire>": 151673,
14
- "<movie_title>": 151679,
15
- "<recommend>": 151674,
16
- "<response>": 151669,
17
  "<think>": 151667,
18
  "<tool_call>": 151657,
19
  "<tool_response>": 151665,
@@ -25,6 +25,7 @@
25
  "<|fim_pad|>": 151662,
26
  "<|fim_prefix|>": 151659,
27
  "<|fim_suffix|>": 151661,
 
28
  "<|im_end|>": 151645,
29
  "<|im_start|>": 151644,
30
  "<|image_pad|>": 151655,
 
1
  {
2
+ "</action>": 151673,
3
+ "</movie_title>": 151681,
4
+ "</response>": 151671,
5
  "</think>": 151668,
6
  "</tool_call>": 151658,
7
  "</tool_response>": 151666,
8
+ "<accept>": 151679,
9
+ "<action>": 151672,
10
+ "<disclose-goal>": 151677,
11
+ "<feedback>": 151678,
12
+ "<greeting>": 151676,
13
+ "<inquire>": 151674,
14
+ "<movie_title>": 151680,
15
+ "<recommend>": 151675,
16
+ "<response>": 151670,
17
  "<think>": 151667,
18
  "<tool_call>": 151657,
19
  "<tool_response>": 151665,
 
25
  "<|fim_pad|>": 151662,
26
  "<|fim_prefix|>": 151659,
27
  "<|fim_suffix|>": 151661,
28
+ "<|finetune_right_pad_id|>": 151669,
29
  "<|im_end|>": 151645,
30
  "<|im_start|>": 151644,
31
  "<|image_pad|>": 151655,
all_results.json CHANGED
@@ -1,9 +1,14 @@
1
  {
2
  "epoch": 1.9982692973347178,
 
 
 
 
 
3
  "total_flos": 4.774796461726499e+18,
4
- "train_loss": 0.7511481684000538,
5
- "train_runtime": 11380.9982,
6
  "train_samples": 46218,
7
- "train_samples_per_second": 8.122,
8
  "train_steps_per_second": 0.127
9
  }
 
1
  {
2
  "epoch": 1.9982692973347178,
3
+ "eval_loss": 0.7263619303703308,
4
+ "eval_runtime": 59.7545,
5
+ "eval_samples": 2069,
6
+ "eval_samples_per_second": 34.625,
7
+ "eval_steps_per_second": 2.176,
8
  "total_flos": 4.774796461726499e+18,
9
+ "train_loss": 0.735994421048838,
10
+ "train_runtime": 11345.701,
11
  "train_samples": 46218,
12
+ "train_samples_per_second": 8.147,
13
  "train_steps_per_second": 0.127
14
  }
config.json CHANGED
@@ -17,7 +17,7 @@
17
  "num_attention_heads": 32,
18
  "num_hidden_layers": 36,
19
  "num_key_value_heads": 8,
20
- "pad_token_id": 128004,
21
  "rms_norm_eps": 1e-06,
22
  "rope_scaling": null,
23
  "rope_theta": 1000000,
 
17
  "num_attention_heads": 32,
18
  "num_hidden_layers": 36,
19
  "num_key_value_heads": 8,
20
+ "pad_token_id": 151669,
21
  "rms_norm_eps": 1e-06,
22
  "rope_scaling": null,
23
  "rope_theta": 1000000,
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.9982692973347178,
3
+ "eval_loss": 0.7263619303703308,
4
+ "eval_runtime": 59.7545,
5
+ "eval_samples": 2069,
6
+ "eval_samples_per_second": 34.625,
7
+ "eval_steps_per_second": 2.176
8
+ }
model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d96a394cf06656eb935d8aca02a770eb51293879dac12830ca28cb3b3db52b73
3
  size 4900684832
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7832cbc16fa62664e3b297c8ec00cc80186da771c1b1d9ca7dcb3c99d2c3eaef
3
  size 4900684832
model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:64e36df7574ed89bb0a7d2333c9c3d1c174fc44e89e5d5c02a24295e815c6a64
3
  size 4915960368
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1c25e1aced354079a8de14c53844bcedce801be8dbf42e67b9cc738507b8a8a
3
  size 4915960368
model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3265e7d720cb19447f8949b971b894cc15d84d77bafebc76b6d4c312650a6585
3
  size 4983068496
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f80218553535f5267c19c3be0290157553a5dd80725c982f63424d5a57abdb76
3
  size 4983068496
model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:566b74e36508f9f8b1f841d007f80c63469a6e2afb6d9afd9f230f5e3f96061b
3
  size 1578657400
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38483bb3eb096cd0f137a08a5f629a7a24a5bc8da06312f7899326f31830a5d7
3
  size 1578657400
runs/May17_01-35-22_w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-56x5fg8/events.out.tfevents.1747457689.w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-56x5fg8.10629.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:165fe2e59b8e226b9ff6201d42d0f1e12b90ade992bfba91a25d581919575180
3
+ size 359
runs/May18_01-21-40_w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-78pqvlh/events.out.tfevents.1747531364.w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-78pqvlh.46539.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b87d27e9a21eb0a3698a9f328c7a41173c8f1e15b6b120309019f5217173906a
3
+ size 67464
special_tokens_map.json CHANGED
@@ -99,5 +99,11 @@
99
  "rstrip": false,
100
  "single_word": false
101
  },
102
- "pad_token": "ÙģØŃ"
 
 
 
 
 
 
103
  }
 
99
  "rstrip": false,
100
  "single_word": false
101
  },
102
+ "pad_token": {
103
+ "content": "<|finetune_right_pad_id|>",
104
+ "lstrip": false,
105
+ "normalized": false,
106
+ "rstrip": false,
107
+ "single_word": false
108
+ }
109
  }
tokenizer.json CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:db976505810ec9dd83bba0026e309fb8472c6e5b43d986b588991da67502f70b
3
- size 11425005
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c649824415514cc027a5b397c4680b0c0b9f6d425fb6a5512dc62813853da0e3
3
+ size 11425207
tokenizer_config.json CHANGED
@@ -211,7 +211,7 @@
211
  "special": false
212
  },
213
  "151669": {
214
- "content": "<response>",
215
  "lstrip": false,
216
  "normalized": false,
217
  "rstrip": false,
@@ -219,7 +219,7 @@
219
  "special": true
220
  },
221
  "151670": {
222
- "content": "</response>",
223
  "lstrip": false,
224
  "normalized": false,
225
  "rstrip": false,
@@ -227,7 +227,7 @@
227
  "special": true
228
  },
229
  "151671": {
230
- "content": "<action>",
231
  "lstrip": false,
232
  "normalized": false,
233
  "rstrip": false,
@@ -235,7 +235,7 @@
235
  "special": true
236
  },
237
  "151672": {
238
- "content": "</action>",
239
  "lstrip": false,
240
  "normalized": false,
241
  "rstrip": false,
@@ -243,7 +243,7 @@
243
  "special": true
244
  },
245
  "151673": {
246
- "content": "<inquire>",
247
  "lstrip": false,
248
  "normalized": false,
249
  "rstrip": false,
@@ -251,7 +251,7 @@
251
  "special": true
252
  },
253
  "151674": {
254
- "content": "<recommend>",
255
  "lstrip": false,
256
  "normalized": false,
257
  "rstrip": false,
@@ -259,7 +259,7 @@
259
  "special": true
260
  },
261
  "151675": {
262
- "content": "<greeting>",
263
  "lstrip": false,
264
  "normalized": false,
265
  "rstrip": false,
@@ -267,7 +267,7 @@
267
  "special": true
268
  },
269
  "151676": {
270
- "content": "<disclose-goal>",
271
  "lstrip": false,
272
  "normalized": false,
273
  "rstrip": false,
@@ -275,7 +275,7 @@
275
  "special": true
276
  },
277
  "151677": {
278
- "content": "<feedback>",
279
  "lstrip": false,
280
  "normalized": false,
281
  "rstrip": false,
@@ -283,7 +283,7 @@
283
  "special": true
284
  },
285
  "151678": {
286
- "content": "<accept>",
287
  "lstrip": false,
288
  "normalized": false,
289
  "rstrip": false,
@@ -291,7 +291,7 @@
291
  "special": true
292
  },
293
  "151679": {
294
- "content": "<movie_title>",
295
  "lstrip": false,
296
  "normalized": false,
297
  "rstrip": false,
@@ -299,6 +299,14 @@
299
  "special": true
300
  },
301
  "151680": {
 
 
 
 
 
 
 
 
302
  "content": "</movie_title>",
303
  "lstrip": false,
304
  "normalized": false,
@@ -328,7 +336,7 @@
328
  "errors": "replace",
329
  "extra_special_tokens": {},
330
  "model_max_length": 131072,
331
- "pad_token": "ÙģØŃ",
332
  "split_special_tokens": false,
333
  "tokenizer_class": "Qwen2Tokenizer",
334
  "unk_token": null
 
211
  "special": false
212
  },
213
  "151669": {
214
+ "content": "<|finetune_right_pad_id|>",
215
  "lstrip": false,
216
  "normalized": false,
217
  "rstrip": false,
 
219
  "special": true
220
  },
221
  "151670": {
222
+ "content": "<response>",
223
  "lstrip": false,
224
  "normalized": false,
225
  "rstrip": false,
 
227
  "special": true
228
  },
229
  "151671": {
230
+ "content": "</response>",
231
  "lstrip": false,
232
  "normalized": false,
233
  "rstrip": false,
 
235
  "special": true
236
  },
237
  "151672": {
238
+ "content": "<action>",
239
  "lstrip": false,
240
  "normalized": false,
241
  "rstrip": false,
 
243
  "special": true
244
  },
245
  "151673": {
246
+ "content": "</action>",
247
  "lstrip": false,
248
  "normalized": false,
249
  "rstrip": false,
 
251
  "special": true
252
  },
253
  "151674": {
254
+ "content": "<inquire>",
255
  "lstrip": false,
256
  "normalized": false,
257
  "rstrip": false,
 
259
  "special": true
260
  },
261
  "151675": {
262
+ "content": "<recommend>",
263
  "lstrip": false,
264
  "normalized": false,
265
  "rstrip": false,
 
267
  "special": true
268
  },
269
  "151676": {
270
+ "content": "<greeting>",
271
  "lstrip": false,
272
  "normalized": false,
273
  "rstrip": false,
 
275
  "special": true
276
  },
277
  "151677": {
278
+ "content": "<disclose-goal>",
279
  "lstrip": false,
280
  "normalized": false,
281
  "rstrip": false,
 
283
  "special": true
284
  },
285
  "151678": {
286
+ "content": "<feedback>",
287
  "lstrip": false,
288
  "normalized": false,
289
  "rstrip": false,
 
291
  "special": true
292
  },
293
  "151679": {
294
+ "content": "<accept>",
295
  "lstrip": false,
296
  "normalized": false,
297
  "rstrip": false,
 
299
  "special": true
300
  },
301
  "151680": {
302
+ "content": "<movie_title>",
303
+ "lstrip": false,
304
+ "normalized": false,
305
+ "rstrip": false,
306
+ "single_word": false,
307
+ "special": true
308
+ },
309
+ "151681": {
310
  "content": "</movie_title>",
311
  "lstrip": false,
312
  "normalized": false,
 
336
  "errors": "replace",
337
  "extra_special_tokens": {},
338
  "model_max_length": 131072,
339
+ "pad_token": "<|finetune_right_pad_id|>",
340
  "split_special_tokens": false,
341
  "tokenizer_class": "Qwen2Tokenizer",
342
  "unk_token": null
train_results.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
  "epoch": 1.9982692973347178,
3
  "total_flos": 4.774796461726499e+18,
4
- "train_loss": 0.7511481684000538,
5
- "train_runtime": 11380.9982,
6
  "train_samples": 46218,
7
- "train_samples_per_second": 8.122,
8
  "train_steps_per_second": 0.127
9
  }
 
1
  {
2
  "epoch": 1.9982692973347178,
3
  "total_flos": 4.774796461726499e+18,
4
+ "train_loss": 0.735994421048838,
5
+ "train_runtime": 11345.701,
6
  "train_samples": 46218,
7
+ "train_samples_per_second": 8.147,
8
  "train_steps_per_second": 0.127
9
  }
trainer_state.json CHANGED
@@ -11,2050 +11,2050 @@
11
  "log_history": [
12
  {
13
  "epoch": 0.0013845621322256837,
14
- "grad_norm": 36.99692916870117,
15
  "learning_rate": 0.0,
16
- "loss": 3.0679,
17
  "step": 1
18
  },
19
  {
20
  "epoch": 0.006922810661128418,
21
- "grad_norm": 15.91442584991455,
22
  "learning_rate": 5.517241379310345e-06,
23
- "loss": 2.9546,
24
  "step": 5
25
  },
26
  {
27
  "epoch": 0.013845621322256836,
28
- "grad_norm": 6.445705413818359,
29
  "learning_rate": 1.2413793103448277e-05,
30
- "loss": 2.3265,
31
  "step": 10
32
  },
33
  {
34
  "epoch": 0.020768431983385256,
35
- "grad_norm": 7.000129699707031,
36
  "learning_rate": 1.9310344827586207e-05,
37
- "loss": 2.0234,
38
  "step": 15
39
  },
40
  {
41
  "epoch": 0.027691242644513673,
42
- "grad_norm": 9.449470520019531,
43
  "learning_rate": 2.620689655172414e-05,
44
- "loss": 1.7883,
45
  "step": 20
46
  },
47
  {
48
  "epoch": 0.034614053305642094,
49
- "grad_norm": 7.50212287902832,
50
  "learning_rate": 3.310344827586207e-05,
51
- "loss": 1.5598,
52
  "step": 25
53
  },
54
  {
55
  "epoch": 0.04153686396677051,
56
- "grad_norm": 3.6690421104431152,
57
  "learning_rate": 4e-05,
58
- "loss": 1.2904,
59
  "step": 30
60
  },
61
  {
62
  "epoch": 0.04845967462789893,
63
- "grad_norm": 3.425504446029663,
64
  "learning_rate": 4.689655172413793e-05,
65
- "loss": 1.2698,
66
  "step": 35
67
  },
68
  {
69
  "epoch": 0.055382485289027346,
70
- "grad_norm": 3.642073392868042,
71
  "learning_rate": 5.379310344827586e-05,
72
- "loss": 1.2002,
73
  "step": 40
74
  },
75
  {
76
  "epoch": 0.06230529595015576,
77
- "grad_norm": 3.359834909439087,
78
  "learning_rate": 6.068965517241379e-05,
79
- "loss": 1.1228,
80
  "step": 45
81
  },
82
  {
83
  "epoch": 0.06922810661128419,
84
- "grad_norm": 2.215944528579712,
85
  "learning_rate": 6.758620689655173e-05,
86
- "loss": 1.0101,
87
  "step": 50
88
  },
89
  {
90
  "epoch": 0.0761509172724126,
91
- "grad_norm": 1.9135385751724243,
92
  "learning_rate": 7.448275862068966e-05,
93
- "loss": 0.9731,
94
  "step": 55
95
  },
96
  {
97
  "epoch": 0.08307372793354102,
98
- "grad_norm": 1.415416955947876,
99
  "learning_rate": 8.137931034482759e-05,
100
- "loss": 0.9475,
101
  "step": 60
102
  },
103
  {
104
  "epoch": 0.08999653859466944,
105
- "grad_norm": 1.5166492462158203,
106
  "learning_rate": 8.827586206896552e-05,
107
- "loss": 0.998,
108
  "step": 65
109
  },
110
  {
111
  "epoch": 0.09691934925579786,
112
- "grad_norm": 1.4899680614471436,
113
  "learning_rate": 9.517241379310345e-05,
114
- "loss": 0.9947,
115
  "step": 70
116
  },
117
  {
118
  "epoch": 0.10384215991692627,
119
- "grad_norm": 1.1434413194656372,
120
  "learning_rate": 0.0001020689655172414,
121
- "loss": 0.9879,
122
  "step": 75
123
  },
124
  {
125
  "epoch": 0.11076497057805469,
126
- "grad_norm": 1.0257912874221802,
127
  "learning_rate": 0.00010896551724137931,
128
- "loss": 1.0003,
129
  "step": 80
130
  },
131
  {
132
  "epoch": 0.11768778123918311,
133
- "grad_norm": 1.187396764755249,
134
  "learning_rate": 0.00011586206896551725,
135
- "loss": 1.0114,
136
  "step": 85
137
  },
138
  {
139
  "epoch": 0.12461059190031153,
140
- "grad_norm": 0.9548481106758118,
141
  "learning_rate": 0.00012275862068965518,
142
- "loss": 1.0146,
143
  "step": 90
144
  },
145
  {
146
  "epoch": 0.13153340256143994,
147
- "grad_norm": 1.1940933465957642,
148
  "learning_rate": 0.0001296551724137931,
149
- "loss": 0.995,
150
  "step": 95
151
  },
152
  {
153
  "epoch": 0.13845621322256838,
154
- "grad_norm": 1.1477693319320679,
155
  "learning_rate": 0.00013655172413793104,
156
- "loss": 0.9914,
157
  "step": 100
158
  },
159
  {
160
  "epoch": 0.14537902388369678,
161
- "grad_norm": 0.9720847606658936,
162
  "learning_rate": 0.00014344827586206896,
163
- "loss": 0.9888,
164
  "step": 105
165
  },
166
  {
167
  "epoch": 0.1523018345448252,
168
- "grad_norm": 0.9881941080093384,
169
  "learning_rate": 0.0001503448275862069,
170
- "loss": 1.0105,
171
  "step": 110
172
  },
173
  {
174
  "epoch": 0.1592246452059536,
175
- "grad_norm": 0.8291202783584595,
176
  "learning_rate": 0.00015724137931034485,
177
- "loss": 1.0039,
178
  "step": 115
179
  },
180
  {
181
  "epoch": 0.16614745586708204,
182
- "grad_norm": 2.3537802696228027,
183
  "learning_rate": 0.00016413793103448276,
184
- "loss": 0.9973,
185
  "step": 120
186
  },
187
  {
188
  "epoch": 0.17307026652821045,
189
- "grad_norm": 10.804582595825195,
190
  "learning_rate": 0.0001710344827586207,
191
- "loss": 1.1009,
192
  "step": 125
193
  },
194
  {
195
  "epoch": 0.17999307718933888,
196
- "grad_norm": 37.236934661865234,
197
  "learning_rate": 0.00017793103448275862,
198
- "loss": 1.7116,
199
  "step": 130
200
  },
201
  {
202
  "epoch": 0.18691588785046728,
203
- "grad_norm": 2.900162935256958,
204
  "learning_rate": 0.00018482758620689654,
205
- "loss": 1.1986,
206
  "step": 135
207
  },
208
  {
209
  "epoch": 0.19383869851159571,
210
- "grad_norm": 1.222854733467102,
211
  "learning_rate": 0.0001917241379310345,
212
- "loss": 1.0546,
213
  "step": 140
214
  },
215
  {
216
  "epoch": 0.20076150917272412,
217
- "grad_norm": 1.0050123929977417,
218
  "learning_rate": 0.00019862068965517243,
219
- "loss": 1.0607,
220
  "step": 145
221
  },
222
  {
223
  "epoch": 0.20768431983385255,
224
- "grad_norm": 1.7881501913070679,
225
  "learning_rate": 0.0001999953208384105,
226
- "loss": 1.1013,
227
  "step": 150
228
  },
229
  {
230
  "epoch": 0.21460713049498095,
231
- "grad_norm": 1.2163009643554688,
232
  "learning_rate": 0.0001999763124949378,
233
- "loss": 1.06,
234
  "step": 155
235
  },
236
  {
237
  "epoch": 0.22152994115610938,
238
- "grad_norm": 0.9521149396896362,
239
  "learning_rate": 0.0001999426852993044,
240
- "loss": 0.9803,
241
  "step": 160
242
  },
243
  {
244
  "epoch": 0.2284527518172378,
245
- "grad_norm": 0.8825615644454956,
246
  "learning_rate": 0.00019989444416858502,
247
- "loss": 0.9539,
248
  "step": 165
249
  },
250
  {
251
  "epoch": 0.23537556247836622,
252
- "grad_norm": 1.0269418954849243,
253
  "learning_rate": 0.0001998315961567502,
254
- "loss": 0.9898,
255
  "step": 170
256
  },
257
  {
258
  "epoch": 0.24229837313949462,
259
- "grad_norm": 1.4703561067581177,
260
  "learning_rate": 0.00019975415045363467,
261
- "loss": 1.0,
262
  "step": 175
263
  },
264
  {
265
  "epoch": 0.24922118380062305,
266
- "grad_norm": 3.445188045501709,
267
  "learning_rate": 0.00019966211838359378,
268
- "loss": 1.0302,
269
  "step": 180
270
  },
271
  {
272
  "epoch": 0.25614399446175146,
273
- "grad_norm": 1.095542311668396,
274
  "learning_rate": 0.00019955551340384743,
275
- "loss": 1.0432,
276
  "step": 185
277
  },
278
  {
279
  "epoch": 0.2630668051228799,
280
- "grad_norm": 3.0603744983673096,
281
  "learning_rate": 0.00019943435110251248,
282
- "loss": 1.0197,
283
  "step": 190
284
  },
285
  {
286
  "epoch": 0.2699896157840083,
287
- "grad_norm": 4.020196914672852,
288
  "learning_rate": 0.0001992986491963232,
289
- "loss": 1.2696,
290
  "step": 195
291
  },
292
  {
293
  "epoch": 0.27691242644513675,
294
- "grad_norm": 2.0897724628448486,
295
  "learning_rate": 0.00019914842752804103,
296
- "loss": 1.7173,
297
  "step": 200
298
  },
299
  {
300
  "epoch": 0.2838352371062651,
301
- "grad_norm": 7.256534576416016,
302
  "learning_rate": 0.00019898370806355263,
303
- "loss": 1.1485,
304
  "step": 205
305
  },
306
  {
307
  "epoch": 0.29075804776739356,
308
- "grad_norm": 1.7438695430755615,
309
  "learning_rate": 0.00019880451488865836,
310
- "loss": 1.1805,
311
  "step": 210
312
  },
313
  {
314
  "epoch": 0.297680858428522,
315
- "grad_norm": 1.0164345502853394,
316
  "learning_rate": 0.00019861087420555018,
317
- "loss": 1.1521,
318
  "step": 215
319
  },
320
  {
321
  "epoch": 0.3046036690896504,
322
- "grad_norm": 0.9694517850875854,
323
  "learning_rate": 0.0001984028143289803,
324
- "loss": 0.9914,
325
  "step": 220
326
  },
327
  {
328
  "epoch": 0.3115264797507788,
329
- "grad_norm": 0.8658626079559326,
330
  "learning_rate": 0.00019818036568212106,
331
- "loss": 0.951,
332
  "step": 225
333
  },
334
  {
335
  "epoch": 0.3184492904119072,
336
- "grad_norm": 0.6972668766975403,
337
  "learning_rate": 0.00019794356079211604,
338
- "loss": 0.9571,
339
  "step": 230
340
  },
341
  {
342
  "epoch": 0.32537210107303566,
343
- "grad_norm": 0.8320409655570984,
344
  "learning_rate": 0.00019769243428532422,
345
- "loss": 0.9898,
346
  "step": 235
347
  },
348
  {
349
  "epoch": 0.3322949117341641,
350
- "grad_norm": 0.7591261267662048,
351
  "learning_rate": 0.00019742702288225652,
352
- "loss": 0.9512,
353
  "step": 240
354
  },
355
  {
356
  "epoch": 0.33921772239529246,
357
- "grad_norm": 1.5846096277236938,
358
  "learning_rate": 0.00019714736539220648,
359
- "loss": 0.9506,
360
  "step": 245
361
  },
362
  {
363
  "epoch": 0.3461405330564209,
364
- "grad_norm": 0.6878282427787781,
365
  "learning_rate": 0.00019685350270757555,
366
- "loss": 0.9655,
367
  "step": 250
368
  },
369
  {
370
  "epoch": 0.3530633437175493,
371
- "grad_norm": 0.6172508597373962,
372
  "learning_rate": 0.0001965454777978936,
373
- "loss": 0.9626,
374
  "step": 255
375
  },
376
  {
377
  "epoch": 0.35998615437867776,
378
- "grad_norm": 0.6852116584777832,
379
  "learning_rate": 0.00019622333570353567,
380
- "loss": 0.9412,
381
  "step": 260
382
  },
383
  {
384
  "epoch": 0.36690896503980613,
385
- "grad_norm": 0.6652204990386963,
386
  "learning_rate": 0.00019588712352913625,
387
- "loss": 0.9313,
388
  "step": 265
389
  },
390
  {
391
  "epoch": 0.37383177570093457,
392
- "grad_norm": 0.638995885848999,
393
  "learning_rate": 0.00019553689043670127,
394
- "loss": 0.9454,
395
  "step": 270
396
  },
397
  {
398
  "epoch": 0.380754586362063,
399
- "grad_norm": 4.199680328369141,
400
  "learning_rate": 0.00019517268763841962,
401
- "loss": 0.9238,
402
  "step": 275
403
  },
404
  {
405
  "epoch": 0.38767739702319143,
406
- "grad_norm": 0.6641167402267456,
407
  "learning_rate": 0.00019479456838917473,
408
- "loss": 0.9295,
409
  "step": 280
410
  },
411
  {
412
  "epoch": 0.39460020768431986,
413
- "grad_norm": 0.6575414538383484,
414
  "learning_rate": 0.0001944025879787574,
415
- "loss": 0.9135,
416
  "step": 285
417
  },
418
  {
419
  "epoch": 0.40152301834544823,
420
- "grad_norm": 0.6238983273506165,
421
  "learning_rate": 0.0001939968037237812,
422
- "loss": 0.9436,
423
  "step": 290
424
  },
425
  {
426
  "epoch": 0.40844582900657667,
427
- "grad_norm": 0.587856650352478,
428
  "learning_rate": 0.00019357727495930146,
429
- "loss": 0.9264,
430
  "step": 295
431
  },
432
  {
433
  "epoch": 0.4153686396677051,
434
- "grad_norm": 0.6809896230697632,
435
  "learning_rate": 0.00019314406303013904,
436
- "loss": 0.918,
437
  "step": 300
438
  },
439
  {
440
  "epoch": 0.42229145032883353,
441
- "grad_norm": 0.6164669394493103,
442
  "learning_rate": 0.00019269723128191048,
443
- "loss": 0.9091,
444
  "step": 305
445
  },
446
  {
447
  "epoch": 0.4292142609899619,
448
- "grad_norm": 0.5151002407073975,
449
  "learning_rate": 0.00019223684505176517,
450
- "loss": 0.9181,
451
  "step": 310
452
  },
453
  {
454
  "epoch": 0.43613707165109034,
455
- "grad_norm": 2.156830310821533,
456
  "learning_rate": 0.00019176297165883165,
457
- "loss": 0.9445,
458
  "step": 315
459
  },
460
  {
461
  "epoch": 0.44305988231221877,
462
- "grad_norm": 0.5134182572364807,
463
  "learning_rate": 0.00019127568039437406,
464
- "loss": 0.9116,
465
  "step": 320
466
  },
467
  {
468
  "epoch": 0.4499826929733472,
469
- "grad_norm": 0.6247857809066772,
470
  "learning_rate": 0.00019077504251166006,
471
- "loss": 0.9073,
472
  "step": 325
473
  },
474
  {
475
  "epoch": 0.4569055036344756,
476
- "grad_norm": 0.6191543340682983,
477
  "learning_rate": 0.00019026113121554204,
478
- "loss": 0.9038,
479
  "step": 330
480
  },
481
  {
482
  "epoch": 0.463828314295604,
483
- "grad_norm": 0.5713596940040588,
484
  "learning_rate": 0.00018973402165175268,
485
- "loss": 0.9082,
486
  "step": 335
487
  },
488
  {
489
  "epoch": 0.47075112495673244,
490
- "grad_norm": 0.524921178817749,
491
  "learning_rate": 0.0001891937908959172,
492
- "loss": 0.8912,
493
  "step": 340
494
  },
495
  {
496
  "epoch": 0.47767393561786087,
497
- "grad_norm": 0.5696610808372498,
498
  "learning_rate": 0.00018864051794228282,
499
- "loss": 0.9034,
500
  "step": 345
501
  },
502
  {
503
  "epoch": 0.48459674627898924,
504
- "grad_norm": 0.6064379215240479,
505
  "learning_rate": 0.00018807428369216822,
506
- "loss": 0.8678,
507
  "step": 350
508
  },
509
  {
510
  "epoch": 0.4915195569401177,
511
- "grad_norm": 0.5061264634132385,
512
  "learning_rate": 0.00018749517094213375,
513
- "loss": 0.8718,
514
  "step": 355
515
  },
516
  {
517
  "epoch": 0.4984423676012461,
518
- "grad_norm": 0.6122836470603943,
519
  "learning_rate": 0.00018690326437187472,
520
- "loss": 0.8837,
521
  "step": 360
522
  },
523
  {
524
  "epoch": 0.5053651782623745,
525
- "grad_norm": 0.5305922031402588,
526
  "learning_rate": 0.00018629865053183911,
527
- "loss": 0.8754,
528
  "step": 365
529
  },
530
  {
531
  "epoch": 0.5122879889235029,
532
- "grad_norm": 0.6255595088005066,
533
  "learning_rate": 0.0001856814178305722,
534
- "loss": 0.86,
535
  "step": 370
536
  },
537
  {
538
  "epoch": 0.5192107995846313,
539
- "grad_norm": 0.49683791399002075,
540
  "learning_rate": 0.00018505165652178893,
541
- "loss": 0.8589,
542
  "step": 375
543
  },
544
  {
545
  "epoch": 0.5261336102457598,
546
- "grad_norm": 0.5160537362098694,
547
  "learning_rate": 0.00018440945869117675,
548
- "loss": 0.8842,
549
  "step": 380
550
  },
551
  {
552
  "epoch": 0.5330564209068882,
553
- "grad_norm": 0.4708365201950073,
554
  "learning_rate": 0.00018375491824293067,
555
- "loss": 0.8795,
556
  "step": 385
557
  },
558
  {
559
  "epoch": 0.5399792315680166,
560
- "grad_norm": 0.47004085779190063,
561
  "learning_rate": 0.00018308813088602227,
562
- "loss": 0.8793,
563
  "step": 390
564
  },
565
  {
566
  "epoch": 0.5469020422291451,
567
- "grad_norm": 0.5247258543968201,
568
  "learning_rate": 0.00018240919412020466,
569
- "loss": 0.859,
570
  "step": 395
571
  },
572
  {
573
  "epoch": 0.5538248528902735,
574
- "grad_norm": 0.5026205778121948,
575
  "learning_rate": 0.00018171820722175604,
576
- "loss": 0.8764,
577
  "step": 400
578
  },
579
  {
580
  "epoch": 0.5607476635514018,
581
- "grad_norm": 0.504045307636261,
582
  "learning_rate": 0.00018101527122896293,
583
- "loss": 0.8728,
584
  "step": 405
585
  },
586
  {
587
  "epoch": 0.5676704742125303,
588
- "grad_norm": 0.5254899859428406,
589
  "learning_rate": 0.0001803004889273463,
590
- "loss": 0.8593,
591
  "step": 410
592
  },
593
  {
594
  "epoch": 0.5745932848736587,
595
- "grad_norm": 0.46688538789749146,
596
  "learning_rate": 0.0001795739648346317,
597
- "loss": 0.8352,
598
  "step": 415
599
  },
600
  {
601
  "epoch": 0.5815160955347871,
602
- "grad_norm": 0.48121586441993713,
603
  "learning_rate": 0.00017883580518546647,
604
- "loss": 0.8342,
605
  "step": 420
606
  },
607
  {
608
  "epoch": 0.5884389061959155,
609
- "grad_norm": 0.5108678936958313,
610
  "learning_rate": 0.00017808611791588584,
611
- "loss": 0.8504,
612
  "step": 425
613
  },
614
  {
615
  "epoch": 0.595361716857044,
616
- "grad_norm": 0.5048635005950928,
617
  "learning_rate": 0.00017732501264752996,
618
- "loss": 0.8389,
619
  "step": 430
620
  },
621
  {
622
  "epoch": 0.6022845275181724,
623
- "grad_norm": 0.4736952781677246,
624
  "learning_rate": 0.00017655260067161503,
625
- "loss": 0.8625,
626
  "step": 435
627
  },
628
  {
629
  "epoch": 0.6092073381793008,
630
- "grad_norm": 0.5057955980300903,
631
  "learning_rate": 0.00017576899493265954,
632
- "loss": 0.847,
633
  "step": 440
634
  },
635
  {
636
  "epoch": 0.6161301488404292,
637
- "grad_norm": 0.4556281864643097,
638
  "learning_rate": 0.00017497431001196943,
639
- "loss": 0.849,
640
  "step": 445
641
  },
642
  {
643
  "epoch": 0.6230529595015576,
644
- "grad_norm": 0.4683106541633606,
645
  "learning_rate": 0.00017416866211088366,
646
- "loss": 0.8554,
647
  "step": 450
648
  },
649
  {
650
  "epoch": 0.629975770162686,
651
- "grad_norm": 0.5124343633651733,
652
  "learning_rate": 0.00017335216903378267,
653
- "loss": 0.8383,
654
  "step": 455
655
  },
656
  {
657
  "epoch": 0.6368985808238145,
658
- "grad_norm": 0.5505478978157043,
659
  "learning_rate": 0.00017252495017086296,
660
- "loss": 0.8315,
661
  "step": 460
662
  },
663
  {
664
  "epoch": 0.6438213914849429,
665
- "grad_norm": 0.486219197511673,
666
  "learning_rate": 0.0001716871264806791,
667
- "loss": 0.8476,
668
  "step": 465
669
  },
670
  {
671
  "epoch": 0.6507442021460713,
672
- "grad_norm": 0.45981094241142273,
673
  "learning_rate": 0.0001708388204724572,
674
- "loss": 0.8453,
675
  "step": 470
676
  },
677
  {
678
  "epoch": 0.6576670128071997,
679
- "grad_norm": 0.4615215063095093,
680
  "learning_rate": 0.00016998015618818087,
681
- "loss": 0.8579,
682
  "step": 475
683
  },
684
  {
685
  "epoch": 0.6645898234683282,
686
- "grad_norm": 0.4511210024356842,
687
  "learning_rate": 0.00016911125918445363,
688
- "loss": 0.8384,
689
  "step": 480
690
  },
691
  {
692
  "epoch": 0.6715126341294566,
693
- "grad_norm": 0.42190468311309814,
694
  "learning_rate": 0.00016823225651413953,
695
- "loss": 0.8265,
696
  "step": 485
697
  },
698
  {
699
  "epoch": 0.6784354447905849,
700
- "grad_norm": 0.479647696018219,
701
  "learning_rate": 0.00016734327670778508,
702
- "loss": 0.8191,
703
  "step": 490
704
  },
705
  {
706
  "epoch": 0.6853582554517134,
707
- "grad_norm": 0.608330249786377,
708
  "learning_rate": 0.00016644444975482523,
709
- "loss": 0.8149,
710
  "step": 495
711
  },
712
  {
713
  "epoch": 0.6922810661128418,
714
- "grad_norm": 0.4981730580329895,
715
  "learning_rate": 0.0001655359070845757,
716
- "loss": 0.8219,
717
  "step": 500
718
  },
719
  {
720
  "epoch": 0.6992038767739702,
721
- "grad_norm": 0.4622678756713867,
722
  "learning_rate": 0.0001646177815470151,
723
- "loss": 0.8246,
724
  "step": 505
725
  },
726
  {
727
  "epoch": 0.7061266874350987,
728
- "grad_norm": 0.4577881693840027,
729
  "learning_rate": 0.00016369020739335924,
730
- "loss": 0.8066,
731
  "step": 510
732
  },
733
  {
734
  "epoch": 0.7130494980962271,
735
- "grad_norm": 0.44852638244628906,
736
  "learning_rate": 0.00016275332025643028,
737
- "loss": 0.8216,
738
  "step": 515
739
  },
740
  {
741
  "epoch": 0.7199723087573555,
742
- "grad_norm": 0.4342646300792694,
743
  "learning_rate": 0.00016180725713082434,
744
- "loss": 0.8273,
745
  "step": 520
746
  },
747
  {
748
  "epoch": 0.726895119418484,
749
- "grad_norm": 0.5106623768806458,
750
  "learning_rate": 0.0001608521563528795,
751
- "loss": 0.8121,
752
  "step": 525
753
  },
754
  {
755
  "epoch": 0.7338179300796123,
756
- "grad_norm": 0.45252540707588196,
757
  "learning_rate": 0.00015988815758044792,
758
- "loss": 0.8056,
759
  "step": 530
760
  },
761
  {
762
  "epoch": 0.7407407407407407,
763
- "grad_norm": 0.4194222092628479,
764
  "learning_rate": 0.00015891540177247467,
765
- "loss": 0.8296,
766
  "step": 535
767
  },
768
  {
769
  "epoch": 0.7476635514018691,
770
- "grad_norm": 0.5437247157096863,
771
  "learning_rate": 0.00015793403116838613,
772
- "loss": 0.8201,
773
  "step": 540
774
  },
775
  {
776
  "epoch": 0.7545863620629976,
777
- "grad_norm": 0.46110814809799194,
778
  "learning_rate": 0.00015694418926729146,
779
- "loss": 0.8049,
780
  "step": 545
781
  },
782
  {
783
  "epoch": 0.761509172724126,
784
- "grad_norm": 0.4320140779018402,
785
  "learning_rate": 0.00015594602080699956,
786
- "loss": 0.8129,
787
  "step": 550
788
  },
789
  {
790
  "epoch": 0.7684319833852544,
791
- "grad_norm": 0.5183108448982239,
792
  "learning_rate": 0.00015493967174285512,
793
- "loss": 0.8029,
794
  "step": 555
795
  },
796
  {
797
  "epoch": 0.7753547940463829,
798
- "grad_norm": 0.40923014283180237,
799
  "learning_rate": 0.00015392528922639662,
800
- "loss": 0.8151,
801
  "step": 560
802
  },
803
  {
804
  "epoch": 0.7822776047075113,
805
- "grad_norm": 0.4213641583919525,
806
  "learning_rate": 0.00015290302158383923,
807
- "loss": 0.8027,
808
  "step": 565
809
  },
810
  {
811
  "epoch": 0.7892004153686397,
812
- "grad_norm": 0.4880472719669342,
813
  "learning_rate": 0.0001518730182943862,
814
- "loss": 0.805,
815
  "step": 570
816
  },
817
  {
818
  "epoch": 0.796123226029768,
819
- "grad_norm": 0.41654378175735474,
820
  "learning_rate": 0.0001508354299683715,
821
- "loss": 0.8012,
822
  "step": 575
823
  },
824
  {
825
  "epoch": 0.8030460366908965,
826
- "grad_norm": 0.46547961235046387,
827
  "learning_rate": 0.0001497904083252371,
828
- "loss": 0.8049,
829
  "step": 580
830
  },
831
  {
832
  "epoch": 0.8099688473520249,
833
- "grad_norm": 0.41686978936195374,
834
  "learning_rate": 0.00014873810617134806,
835
- "loss": 0.7993,
836
  "step": 585
837
  },
838
  {
839
  "epoch": 0.8168916580131533,
840
- "grad_norm": 0.4086816906929016,
841
  "learning_rate": 0.00014767867737764876,
842
- "loss": 0.7949,
843
  "step": 590
844
  },
845
  {
846
  "epoch": 0.8238144686742818,
847
- "grad_norm": 0.43626415729522705,
848
  "learning_rate": 0.0001466122768571633,
849
- "loss": 0.8033,
850
  "step": 595
851
  },
852
  {
853
  "epoch": 0.8307372793354102,
854
- "grad_norm": 0.42999717593193054,
855
  "learning_rate": 0.0001455390605423437,
856
- "loss": 0.8104,
857
  "step": 600
858
  },
859
  {
860
  "epoch": 0.8376600899965386,
861
- "grad_norm": 0.5020118951797485,
862
  "learning_rate": 0.00014445918536226904,
863
- "loss": 0.7927,
864
  "step": 605
865
  },
866
  {
867
  "epoch": 0.8445829006576671,
868
- "grad_norm": 0.48819220066070557,
869
  "learning_rate": 0.00014337280921969858,
870
- "loss": 0.7883,
871
  "step": 610
872
  },
873
  {
874
  "epoch": 0.8515057113187954,
875
- "grad_norm": 0.4125864803791046,
876
  "learning_rate": 0.0001422800909679829,
877
- "loss": 0.7987,
878
  "step": 615
879
  },
880
  {
881
  "epoch": 0.8584285219799238,
882
- "grad_norm": 0.48757144808769226,
883
  "learning_rate": 0.00014118119038783588,
884
- "loss": 0.7902,
885
  "step": 620
886
  },
887
  {
888
  "epoch": 0.8653513326410522,
889
- "grad_norm": 0.47712209820747375,
890
  "learning_rate": 0.00014007626816397087,
891
- "loss": 0.8082,
892
  "step": 625
893
  },
894
  {
895
  "epoch": 0.8722741433021807,
896
- "grad_norm": 0.4102008640766144,
897
  "learning_rate": 0.00013896548586160503,
898
- "loss": 0.7948,
899
  "step": 630
900
  },
901
  {
902
  "epoch": 0.8791969539633091,
903
- "grad_norm": 0.4306066632270813,
904
  "learning_rate": 0.00013784900590283473,
905
- "loss": 0.7952,
906
  "step": 635
907
  },
908
  {
909
  "epoch": 0.8861197646244375,
910
- "grad_norm": 0.42309749126434326,
911
  "learning_rate": 0.00013672699154288569,
912
- "loss": 0.7787,
913
  "step": 640
914
  },
915
  {
916
  "epoch": 0.893042575285566,
917
- "grad_norm": 0.4400608539581299,
918
  "learning_rate": 0.00013559960684624112,
919
- "loss": 0.8045,
920
  "step": 645
921
  },
922
  {
923
  "epoch": 0.8999653859466944,
924
- "grad_norm": 0.50711989402771,
925
  "learning_rate": 0.0001344670166626519,
926
- "loss": 0.7725,
927
  "step": 650
928
  },
929
  {
930
  "epoch": 0.9068881966078228,
931
- "grad_norm": 0.45436131954193115,
932
  "learning_rate": 0.0001333293866030317,
933
- "loss": 0.7773,
934
  "step": 655
935
  },
936
  {
937
  "epoch": 0.9138110072689511,
938
- "grad_norm": 0.40752795338630676,
939
  "learning_rate": 0.0001321868830152407,
940
- "loss": 0.7597,
941
  "step": 660
942
  },
943
  {
944
  "epoch": 0.9207338179300796,
945
- "grad_norm": 0.43623602390289307,
946
  "learning_rate": 0.00013103967295976179,
947
- "loss": 0.7849,
948
  "step": 665
949
  },
950
  {
951
  "epoch": 0.927656628591208,
952
- "grad_norm": 0.4025138318538666,
953
  "learning_rate": 0.0001298879241852724,
954
- "loss": 0.7823,
955
  "step": 670
956
  },
957
  {
958
  "epoch": 0.9345794392523364,
959
- "grad_norm": 0.438235878944397,
960
  "learning_rate": 0.0001287318051041159,
961
- "loss": 0.7812,
962
  "step": 675
963
  },
964
  {
965
  "epoch": 0.9415022499134649,
966
- "grad_norm": 0.4313248097896576,
967
  "learning_rate": 0.00012757148476767553,
968
- "loss": 0.7726,
969
  "step": 680
970
  },
971
  {
972
  "epoch": 0.9484250605745933,
973
- "grad_norm": 0.40096625685691833,
974
  "learning_rate": 0.0001264071328416555,
975
- "loss": 0.7776,
976
  "step": 685
977
  },
978
  {
979
  "epoch": 0.9553478712357217,
980
- "grad_norm": 0.39130938053131104,
981
  "learning_rate": 0.00012523891958127168,
982
- "loss": 0.7729,
983
  "step": 690
984
  },
985
  {
986
  "epoch": 0.9622706818968502,
987
- "grad_norm": 0.4303322434425354,
988
  "learning_rate": 0.0001240670158063565,
989
- "loss": 0.7707,
990
  "step": 695
991
  },
992
  {
993
  "epoch": 0.9691934925579785,
994
- "grad_norm": 0.3983173072338104,
995
  "learning_rate": 0.00012289159287638105,
996
- "loss": 0.7801,
997
  "step": 700
998
  },
999
  {
1000
  "epoch": 0.9761163032191069,
1001
- "grad_norm": 0.40058690309524536,
1002
  "learning_rate": 0.00012171282266539845,
1003
- "loss": 0.7767,
1004
  "step": 705
1005
  },
1006
  {
1007
  "epoch": 0.9830391138802353,
1008
- "grad_norm": 0.4056876003742218,
1009
  "learning_rate": 0.00012053087753691172,
1010
- "loss": 0.7585,
1011
  "step": 710
1012
  },
1013
  {
1014
  "epoch": 0.9899619245413638,
1015
- "grad_norm": 0.4484279155731201,
1016
  "learning_rate": 0.00011934593031867038,
1017
- "loss": 0.7733,
1018
  "step": 715
1019
  },
1020
  {
1021
  "epoch": 0.9968847352024922,
1022
- "grad_norm": 0.38870301842689514,
1023
  "learning_rate": 0.00011815815427739911,
1024
  "loss": 0.7706,
1025
  "step": 720
1026
  },
1027
  {
1028
  "epoch": 1.0,
1029
- "eval_loss": 0.7746492028236389,
1030
- "eval_runtime": 60.0688,
1031
- "eval_samples_per_second": 34.444,
1032
- "eval_steps_per_second": 2.164,
1033
  "step": 723
1034
  },
1035
  {
1036
  "epoch": 1.0027691242644514,
1037
- "grad_norm": 0.41168180108070374,
1038
  "learning_rate": 0.00011696772309346182,
1039
- "loss": 0.6965,
1040
  "step": 725
1041
  },
1042
  {
1043
  "epoch": 1.0096919349255797,
1044
- "grad_norm": 0.4434298574924469,
1045
  "learning_rate": 0.0001157748108354659,
1046
- "loss": 0.5974,
1047
  "step": 730
1048
  },
1049
  {
1050
  "epoch": 1.0166147455867083,
1051
- "grad_norm": 0.3997260332107544,
1052
  "learning_rate": 0.00011457959193480925,
1053
- "loss": 0.5963,
1054
  "step": 735
1055
  },
1056
  {
1057
  "epoch": 1.0235375562478366,
1058
- "grad_norm": 0.4161541163921356,
1059
  "learning_rate": 0.00011338224116017423,
1060
- "loss": 0.5973,
1061
  "step": 740
1062
  },
1063
  {
1064
  "epoch": 1.0304603669089651,
1065
- "grad_norm": 0.6814983487129211,
1066
  "learning_rate": 0.00011218293359197267,
1067
- "loss": 0.5994,
1068
  "step": 745
1069
  },
1070
  {
1071
  "epoch": 1.0373831775700935,
1072
- "grad_norm": 0.4457721710205078,
1073
  "learning_rate": 0.00011098184459674484,
1074
- "loss": 0.5688,
1075
  "step": 750
1076
  },
1077
  {
1078
  "epoch": 1.0443059882312218,
1079
- "grad_norm": 0.41027259826660156,
1080
  "learning_rate": 0.00010977914980151691,
1081
- "loss": 0.5982,
1082
  "step": 755
1083
  },
1084
  {
1085
  "epoch": 1.0512287988923503,
1086
- "grad_norm": 0.41261303424835205,
1087
  "learning_rate": 0.00010857502506812029,
1088
- "loss": 0.5781,
1089
  "step": 760
1090
  },
1091
  {
1092
  "epoch": 1.0581516095534786,
1093
- "grad_norm": 0.3924131393432617,
1094
  "learning_rate": 0.00010736964646747629,
1095
- "loss": 0.578,
1096
  "step": 765
1097
  },
1098
  {
1099
  "epoch": 1.0650744202146072,
1100
- "grad_norm": 0.38322097063064575,
1101
  "learning_rate": 0.00010616319025385089,
1102
- "loss": 0.5979,
1103
  "step": 770
1104
  },
1105
  {
1106
  "epoch": 1.0719972308757355,
1107
- "grad_norm": 0.3903074562549591,
1108
  "learning_rate": 0.00010495583283908177,
1109
- "loss": 0.5887,
1110
  "step": 775
1111
  },
1112
  {
1113
  "epoch": 1.078920041536864,
1114
- "grad_norm": 0.40374380350112915,
1115
  "learning_rate": 0.00010374775076678332,
1116
- "loss": 0.5708,
1117
  "step": 780
1118
  },
1119
  {
1120
  "epoch": 1.0858428521979924,
1121
- "grad_norm": 0.41232338547706604,
1122
  "learning_rate": 0.00010253912068653146,
1123
- "loss": 0.5899,
1124
  "step": 785
1125
  },
1126
  {
1127
  "epoch": 1.092765662859121,
1128
- "grad_norm": 0.37063470482826233,
1129
  "learning_rate": 0.00010133011932803378,
1130
- "loss": 0.5792,
1131
  "step": 790
1132
  },
1133
  {
1134
  "epoch": 1.0996884735202492,
1135
- "grad_norm": 0.42684876918792725,
1136
  "learning_rate": 0.00010012092347528705,
1137
- "loss": 0.589,
1138
  "step": 795
1139
  },
1140
  {
1141
  "epoch": 1.1066112841813776,
1142
- "grad_norm": 0.4422379732131958,
1143
  "learning_rate": 9.891170994072793e-05,
1144
- "loss": 0.5821,
1145
  "step": 800
1146
  },
1147
  {
1148
  "epoch": 1.113534094842506,
1149
- "grad_norm": 0.4900904595851898,
1150
  "learning_rate": 9.770265553937831e-05,
1151
- "loss": 0.5807,
1152
  "step": 805
1153
  },
1154
  {
1155
  "epoch": 1.1204569055036344,
1156
- "grad_norm": 0.43178027868270874,
1157
  "learning_rate": 9.649393706299127e-05,
1158
- "loss": 0.5657,
1159
  "step": 810
1160
  },
1161
  {
1162
  "epoch": 1.127379716164763,
1163
- "grad_norm": 0.43203285336494446,
1164
  "learning_rate": 9.52857312541998e-05,
1165
- "loss": 0.5762,
1166
  "step": 815
1167
  },
1168
  {
1169
  "epoch": 1.1343025268258913,
1170
- "grad_norm": 0.41020694375038147,
1171
  "learning_rate": 9.407821478067304e-05,
1172
- "loss": 0.5682,
1173
  "step": 820
1174
  },
1175
  {
1176
  "epoch": 1.1412253374870198,
1177
- "grad_norm": 0.404141902923584,
1178
  "learning_rate": 9.287156420928342e-05,
1179
- "loss": 0.5701,
1180
  "step": 825
1181
  },
1182
  {
1183
  "epoch": 1.1481481481481481,
1184
- "grad_norm": 0.36725813150405884,
1185
  "learning_rate": 9.166595598028832e-05,
1186
- "loss": 0.5909,
1187
  "step": 830
1188
  },
1189
  {
1190
  "epoch": 1.1550709588092767,
1191
- "grad_norm": 0.40943267941474915,
1192
  "learning_rate": 9.046156638153056e-05,
1193
- "loss": 0.5683,
1194
  "step": 835
1195
  },
1196
  {
1197
  "epoch": 1.161993769470405,
1198
- "grad_norm": 0.3729795515537262,
1199
  "learning_rate": 8.925857152266103e-05,
1200
- "loss": 0.5716,
1201
  "step": 840
1202
  },
1203
  {
1204
  "epoch": 1.1689165801315333,
1205
- "grad_norm": 0.4161226451396942,
1206
  "learning_rate": 8.805714730938728e-05,
1207
- "loss": 0.5769,
1208
  "step": 845
1209
  },
1210
  {
1211
  "epoch": 1.1758393907926619,
1212
- "grad_norm": 0.3843950927257538,
1213
  "learning_rate": 8.685746941775219e-05,
1214
- "loss": 0.5678,
1215
  "step": 850
1216
  },
1217
  {
1218
  "epoch": 1.1827622014537902,
1219
- "grad_norm": 0.40685024857521057,
1220
  "learning_rate": 8.565971326844584e-05,
1221
- "loss": 0.5692,
1222
  "step": 855
1223
  },
1224
  {
1225
  "epoch": 1.1896850121149187,
1226
- "grad_norm": 0.40106266736984253,
1227
  "learning_rate": 8.44640540011553e-05,
1228
- "loss": 0.5738,
1229
  "step": 860
1230
  },
1231
  {
1232
  "epoch": 1.196607822776047,
1233
- "grad_norm": 0.4031878113746643,
1234
  "learning_rate": 8.327066644895469e-05,
1235
- "loss": 0.5712,
1236
  "step": 865
1237
  },
1238
  {
1239
  "epoch": 1.2035306334371756,
1240
- "grad_norm": 0.42471978068351746,
1241
  "learning_rate": 8.207972511274095e-05,
1242
- "loss": 0.5639,
1243
  "step": 870
1244
  },
1245
  {
1246
  "epoch": 1.210453444098304,
1247
- "grad_norm": 0.3819376528263092,
1248
  "learning_rate": 8.089140413571747e-05,
1249
- "loss": 0.5665,
1250
  "step": 875
1251
  },
1252
  {
1253
  "epoch": 1.2173762547594325,
1254
- "grad_norm": 0.43387338519096375,
1255
  "learning_rate": 7.970587727793051e-05,
1256
- "loss": 0.5593,
1257
  "step": 880
1258
  },
1259
  {
1260
  "epoch": 1.2242990654205608,
1261
- "grad_norm": 0.38693514466285706,
1262
  "learning_rate": 7.852331789086135e-05,
1263
- "loss": 0.5644,
1264
  "step": 885
1265
  },
1266
  {
1267
  "epoch": 1.231221876081689,
1268
- "grad_norm": 0.40408262610435486,
1269
  "learning_rate": 7.73438988920784e-05,
1270
- "loss": 0.5692,
1271
  "step": 890
1272
  },
1273
  {
1274
  "epoch": 1.2381446867428176,
1275
- "grad_norm": 0.4097476303577423,
1276
  "learning_rate": 7.616779273995252e-05,
1277
- "loss": 0.5719,
1278
  "step": 895
1279
  },
1280
  {
1281
  "epoch": 1.245067497403946,
1282
- "grad_norm": 0.40725842118263245,
1283
  "learning_rate": 7.499517140843968e-05,
1284
- "loss": 0.5652,
1285
  "step": 900
1286
  },
1287
  {
1288
  "epoch": 1.2519903080650745,
1289
- "grad_norm": 0.41550350189208984,
1290
  "learning_rate": 7.382620636193438e-05,
1291
- "loss": 0.5612,
1292
  "step": 905
1293
  },
1294
  {
1295
  "epoch": 1.2589131187262028,
1296
- "grad_norm": 0.4235980808734894,
1297
  "learning_rate": 7.266106853019759e-05,
1298
- "loss": 0.5597,
1299
  "step": 910
1300
  },
1301
  {
1302
  "epoch": 1.2658359293873311,
1303
- "grad_norm": 0.3927691876888275,
1304
  "learning_rate": 7.149992828336274e-05,
1305
- "loss": 0.5658,
1306
  "step": 915
1307
  },
1308
  {
1309
  "epoch": 1.2727587400484597,
1310
- "grad_norm": 0.4296377897262573,
1311
  "learning_rate": 7.034295540702397e-05,
1312
- "loss": 0.5591,
1313
  "step": 920
1314
  },
1315
  {
1316
  "epoch": 1.2796815507095882,
1317
- "grad_norm": 0.39783212542533875,
1318
  "learning_rate": 6.919031907740922e-05,
1319
- "loss": 0.5779,
1320
  "step": 925
1321
  },
1322
  {
1323
  "epoch": 1.2866043613707165,
1324
- "grad_norm": 0.4201776087284088,
1325
  "learning_rate": 6.804218783664288e-05,
1326
- "loss": 0.5715,
1327
  "step": 930
1328
  },
1329
  {
1330
  "epoch": 1.2935271720318449,
1331
- "grad_norm": 0.3674381673336029,
1332
  "learning_rate": 6.689872956810102e-05,
1333
- "loss": 0.5583,
1334
  "step": 935
1335
  },
1336
  {
1337
  "epoch": 1.3004499826929734,
1338
- "grad_norm": 0.39247825741767883,
1339
  "learning_rate": 6.5760111471863e-05,
1340
- "loss": 0.5574,
1341
  "step": 940
1342
  },
1343
  {
1344
  "epoch": 1.3073727933541017,
1345
- "grad_norm": 0.3951855003833771,
1346
  "learning_rate": 6.46265000402628e-05,
1347
- "loss": 0.5646,
1348
  "step": 945
1349
  },
1350
  {
1351
  "epoch": 1.3142956040152303,
1352
- "grad_norm": 0.37041106820106506,
1353
  "learning_rate": 6.349806103354417e-05,
1354
- "loss": 0.561,
1355
  "step": 950
1356
  },
1357
  {
1358
  "epoch": 1.3212184146763586,
1359
- "grad_norm": 0.4068072736263275,
1360
  "learning_rate": 6.23749594556226e-05,
1361
- "loss": 0.5555,
1362
  "step": 955
1363
  },
1364
  {
1365
  "epoch": 1.328141225337487,
1366
- "grad_norm": 0.3812420666217804,
1367
  "learning_rate": 6.125735952995775e-05,
1368
- "loss": 0.5606,
1369
  "step": 960
1370
  },
1371
  {
1372
  "epoch": 1.3350640359986155,
1373
- "grad_norm": 0.42359232902526855,
1374
  "learning_rate": 6.0145424675540394e-05,
1375
- "loss": 0.5557,
1376
  "step": 965
1377
  },
1378
  {
1379
  "epoch": 1.341986846659744,
1380
- "grad_norm": 0.3885987401008606,
1381
  "learning_rate": 5.9039317482996735e-05,
1382
- "loss": 0.5655,
1383
  "step": 970
1384
  },
1385
  {
1386
  "epoch": 1.3489096573208723,
1387
- "grad_norm": 0.3710838258266449,
1388
  "learning_rate": 5.793919969081374e-05,
1389
- "loss": 0.5445,
1390
  "step": 975
1391
  },
1392
  {
1393
  "epoch": 1.3558324679820006,
1394
- "grad_norm": 0.3927723169326782,
1395
  "learning_rate": 5.68452321616894e-05,
1396
- "loss": 0.5417,
1397
  "step": 980
1398
  },
1399
  {
1400
  "epoch": 1.3627552786431292,
1401
- "grad_norm": 0.39248526096343994,
1402
  "learning_rate": 5.5757574859010886e-05,
1403
- "loss": 0.5584,
1404
  "step": 985
1405
  },
1406
  {
1407
  "epoch": 1.3696780893042575,
1408
- "grad_norm": 0.4080273509025574,
1409
  "learning_rate": 5.467638682346403e-05,
1410
- "loss": 0.5412,
1411
  "step": 990
1412
  },
1413
  {
1414
  "epoch": 1.376600899965386,
1415
- "grad_norm": 0.3992749750614166,
1416
  "learning_rate": 5.3601826149777966e-05,
1417
- "loss": 0.5734,
1418
  "step": 995
1419
  },
1420
  {
1421
  "epoch": 1.3835237106265144,
1422
- "grad_norm": 0.37569689750671387,
1423
  "learning_rate": 5.25340499636082e-05,
1424
- "loss": 0.5538,
1425
  "step": 1000
1426
  },
1427
  {
1428
  "epoch": 1.3904465212876427,
1429
- "grad_norm": 0.40693604946136475,
1430
  "learning_rate": 5.147321439856091e-05,
1431
- "loss": 0.557,
1432
  "step": 1005
1433
  },
1434
  {
1435
  "epoch": 1.3973693319487712,
1436
- "grad_norm": 0.35887908935546875,
1437
  "learning_rate": 5.041947457336274e-05,
1438
- "loss": 0.5589,
1439
  "step": 1010
1440
  },
1441
  {
1442
  "epoch": 1.4042921426098995,
1443
- "grad_norm": 0.36887434124946594,
1444
  "learning_rate": 4.937298456917889e-05,
1445
- "loss": 0.5357,
1446
  "step": 1015
1447
  },
1448
  {
1449
  "epoch": 1.411214953271028,
1450
- "grad_norm": 0.3944549262523651,
1451
  "learning_rate": 4.8333897407082896e-05,
1452
- "loss": 0.5465,
1453
  "step": 1020
1454
  },
1455
  {
1456
  "epoch": 1.4181377639321564,
1457
- "grad_norm": 0.3607221841812134,
1458
  "learning_rate": 4.7302365025681206e-05,
1459
- "loss": 0.5553,
1460
  "step": 1025
1461
  },
1462
  {
1463
  "epoch": 1.425060574593285,
1464
- "grad_norm": 0.3826063275337219,
1465
  "learning_rate": 4.627853825889642e-05,
1466
- "loss": 0.5313,
1467
  "step": 1030
1468
  },
1469
  {
1470
  "epoch": 1.4319833852544133,
1471
- "grad_norm": 0.3818247616291046,
1472
  "learning_rate": 4.526256681391192e-05,
1473
- "loss": 0.5486,
1474
  "step": 1035
1475
  },
1476
  {
1477
  "epoch": 1.4389061959155418,
1478
- "grad_norm": 0.40942633152008057,
1479
  "learning_rate": 4.4254599249281016e-05,
1480
- "loss": 0.5302,
1481
  "step": 1040
1482
  },
1483
  {
1484
  "epoch": 1.4458290065766701,
1485
- "grad_norm": 0.40826553106307983,
1486
  "learning_rate": 4.3254782953204375e-05,
1487
- "loss": 0.5407,
1488
  "step": 1045
1489
  },
1490
  {
1491
  "epoch": 1.4527518172377984,
1492
- "grad_norm": 0.3731902241706848,
1493
  "learning_rate": 4.2263264121978505e-05,
1494
- "loss": 0.5471,
1495
  "step": 1050
1496
  },
1497
  {
1498
  "epoch": 1.459674627898927,
1499
- "grad_norm": 0.40096205472946167,
1500
  "learning_rate": 4.12801877386183e-05,
1501
- "loss": 0.536,
1502
  "step": 1055
1503
  },
1504
  {
1505
  "epoch": 1.4665974385600553,
1506
- "grad_norm": 0.374845415353775,
1507
  "learning_rate": 4.030569755165736e-05,
1508
- "loss": 0.546,
1509
  "step": 1060
1510
  },
1511
  {
1512
  "epoch": 1.4735202492211839,
1513
- "grad_norm": 0.3810202181339264,
1514
  "learning_rate": 3.9339936054128466e-05,
1515
- "loss": 0.5375,
1516
  "step": 1065
1517
  },
1518
  {
1519
  "epoch": 1.4804430598823122,
1520
- "grad_norm": 0.39244675636291504,
1521
  "learning_rate": 3.8383044462727826e-05,
1522
- "loss": 0.5306,
1523
  "step": 1070
1524
  },
1525
  {
1526
  "epoch": 1.4873658705434407,
1527
- "grad_norm": 0.40670761466026306,
1528
  "learning_rate": 3.7435162697166107e-05,
1529
- "loss": 0.5497,
1530
  "step": 1075
1531
  },
1532
  {
1533
  "epoch": 1.494288681204569,
1534
- "grad_norm": 0.4042389392852783,
1535
  "learning_rate": 3.649642935970859e-05,
1536
- "loss": 0.5541,
1537
  "step": 1080
1538
  },
1539
  {
1540
  "epoch": 1.5012114918656976,
1541
- "grad_norm": 0.39131784439086914,
1542
  "learning_rate": 3.556698171490871e-05,
1543
- "loss": 0.5415,
1544
  "step": 1085
1545
  },
1546
  {
1547
  "epoch": 1.508134302526826,
1548
- "grad_norm": 0.3889631927013397,
1549
  "learning_rate": 3.464695566953644e-05,
1550
- "loss": 0.5313,
1551
  "step": 1090
1552
  },
1553
  {
1554
  "epoch": 1.5150571131879542,
1555
- "grad_norm": 0.4033074975013733,
1556
  "learning_rate": 3.373648575270576e-05,
1557
- "loss": 0.538,
1558
  "step": 1095
1559
  },
1560
  {
1561
  "epoch": 1.5219799238490828,
1562
- "grad_norm": 0.36675384640693665,
1563
  "learning_rate": 3.283570509620344e-05,
1564
- "loss": 0.5443,
1565
  "step": 1100
1566
  },
1567
  {
1568
  "epoch": 1.5289027345102113,
1569
- "grad_norm": 0.380545049905777,
1570
  "learning_rate": 3.1944745415021915e-05,
1571
- "loss": 0.5394,
1572
  "step": 1105
1573
  },
1574
  {
1575
  "epoch": 1.5358255451713396,
1576
- "grad_norm": 0.35882043838500977,
1577
  "learning_rate": 3.106373698809958e-05,
1578
- "loss": 0.5445,
1579
  "step": 1110
1580
  },
1581
  {
1582
  "epoch": 1.542748355832468,
1583
- "grad_norm": 0.3889442980289459,
1584
  "learning_rate": 3.0192808639271065e-05,
1585
- "loss": 0.5536,
1586
  "step": 1115
1587
  },
1588
  {
1589
  "epoch": 1.5496711664935963,
1590
- "grad_norm": 0.3797926902770996,
1591
  "learning_rate": 2.933208771843018e-05,
1592
- "loss": 0.5409,
1593
  "step": 1120
1594
  },
1595
  {
1596
  "epoch": 1.5565939771547248,
1597
- "grad_norm": 0.38983410596847534,
1598
  "learning_rate": 2.8481700082908268e-05,
1599
- "loss": 0.5347,
1600
  "step": 1125
1601
  },
1602
  {
1603
  "epoch": 1.5635167878158533,
1604
- "grad_norm": 0.39294636249542236,
1605
  "learning_rate": 2.7641770079071127e-05,
1606
- "loss": 0.5374,
1607
  "step": 1130
1608
  },
1609
  {
1610
  "epoch": 1.5704395984769817,
1611
- "grad_norm": 0.37386760115623474,
1612
  "learning_rate": 2.681242052413656e-05,
1613
- "loss": 0.5198,
1614
  "step": 1135
1615
  },
1616
  {
1617
  "epoch": 1.57736240913811,
1618
- "grad_norm": 0.41262131929397583,
1619
  "learning_rate": 2.59937726882158e-05,
1620
- "loss": 0.5231,
1621
  "step": 1140
1622
  },
1623
  {
1624
  "epoch": 1.5842852197992385,
1625
- "grad_norm": 0.4110383987426758,
1626
  "learning_rate": 2.5185946276580918e-05,
1627
- "loss": 0.5478,
1628
  "step": 1145
1629
  },
1630
  {
1631
  "epoch": 1.591208030460367,
1632
- "grad_norm": 0.3801433742046356,
1633
  "learning_rate": 2.4389059412161087e-05,
1634
- "loss": 0.5395,
1635
  "step": 1150
1636
  },
1637
  {
1638
  "epoch": 1.5981308411214954,
1639
- "grad_norm": 0.37619632482528687,
1640
  "learning_rate": 2.3603228618270356e-05,
1641
- "loss": 0.5355,
1642
  "step": 1155
1643
  },
1644
  {
1645
  "epoch": 1.6050536517826237,
1646
- "grad_norm": 0.3852304220199585,
1647
  "learning_rate": 2.2828568801569283e-05,
1648
- "loss": 0.5133,
1649
  "step": 1160
1650
  },
1651
  {
1652
  "epoch": 1.611976462443752,
1653
- "grad_norm": 0.35891711711883545,
1654
  "learning_rate": 2.2065193235262782e-05,
1655
- "loss": 0.5114,
1656
  "step": 1165
1657
  },
1658
  {
1659
  "epoch": 1.6188992731048806,
1660
- "grad_norm": 0.3851226568222046,
1661
  "learning_rate": 2.1313213542537058e-05,
1662
- "loss": 0.5308,
1663
  "step": 1170
1664
  },
1665
  {
1666
  "epoch": 1.6258220837660091,
1667
- "grad_norm": 0.3977566659450531,
1668
  "learning_rate": 2.0572739680237717e-05,
1669
- "loss": 0.5459,
1670
  "step": 1175
1671
  },
1672
  {
1673
  "epoch": 1.6327448944271374,
1674
- "grad_norm": 0.3800656497478485,
1675
  "learning_rate": 1.984387992279153e-05,
1676
- "loss": 0.5479,
1677
  "step": 1180
1678
  },
1679
  {
1680
  "epoch": 1.6396677050882658,
1681
- "grad_norm": 0.41137266159057617,
1682
  "learning_rate": 1.912674084637409e-05,
1683
- "loss": 0.5429,
1684
  "step": 1185
1685
  },
1686
  {
1687
  "epoch": 1.6465905157493943,
1688
- "grad_norm": 0.3646909296512604,
1689
  "learning_rate": 1.8421427313326046e-05,
1690
- "loss": 0.5212,
1691
  "step": 1190
1692
  },
1693
  {
1694
  "epoch": 1.6535133264105226,
1695
- "grad_norm": 0.38323456048965454,
1696
  "learning_rate": 1.7728042456819704e-05,
1697
- "loss": 0.5419,
1698
  "step": 1195
1699
  },
1700
  {
1701
  "epoch": 1.6604361370716512,
1702
- "grad_norm": 0.3656870126724243,
1703
  "learning_rate": 1.7046687665778717e-05,
1704
- "loss": 0.523,
1705
  "step": 1200
1706
  },
1707
  {
1708
  "epoch": 1.6673589477327795,
1709
- "grad_norm": 0.3620772957801819,
1710
  "learning_rate": 1.6377462570052438e-05,
1711
- "loss": 0.5203,
1712
  "step": 1205
1713
  },
1714
  {
1715
  "epoch": 1.6742817583939078,
1716
- "grad_norm": 0.4021745026111603,
1717
  "learning_rate": 1.5720465025847987e-05,
1718
- "loss": 0.5243,
1719
  "step": 1210
1720
  },
1721
  {
1722
  "epoch": 1.6812045690550363,
1723
- "grad_norm": 0.3713143467903137,
1724
  "learning_rate": 1.5075791101421166e-05,
1725
- "loss": 0.5251,
1726
  "step": 1215
1727
  },
1728
  {
1729
  "epoch": 1.6881273797161649,
1730
- "grad_norm": 0.3876360356807709,
1731
  "learning_rate": 1.4443535063029279e-05,
1732
- "loss": 0.5429,
1733
  "step": 1220
1734
  },
1735
  {
1736
  "epoch": 1.6950501903772932,
1737
- "grad_norm": 0.38543200492858887,
1738
  "learning_rate": 1.3823789361147044e-05,
1739
- "loss": 0.5178,
1740
  "step": 1225
1741
  },
1742
  {
1743
  "epoch": 1.7019730010384215,
1744
- "grad_norm": 0.39921507239341736,
1745
  "learning_rate": 1.3216644616948282e-05,
1746
- "loss": 0.5248,
1747
  "step": 1230
1748
  },
1749
  {
1750
  "epoch": 1.70889581169955,
1751
- "grad_norm": 0.36942583322525024,
1752
  "learning_rate": 1.2622189609054979e-05,
1753
- "loss": 0.5174,
1754
  "step": 1235
1755
  },
1756
  {
1757
  "epoch": 1.7158186223606784,
1758
- "grad_norm": 0.41628918051719666,
1759
  "learning_rate": 1.2040511260555831e-05,
1760
- "loss": 0.5253,
1761
  "step": 1240
1762
  },
1763
  {
1764
  "epoch": 1.722741433021807,
1765
- "grad_norm": 0.3788851797580719,
1766
  "learning_rate": 1.1471694626296049e-05,
1767
- "loss": 0.523,
1768
  "step": 1245
1769
  },
1770
  {
1771
  "epoch": 1.7296642436829353,
1772
- "grad_norm": 0.369893342256546,
1773
  "learning_rate": 1.0915822880440308e-05,
1774
- "loss": 0.5167,
1775
  "step": 1250
1776
  },
1777
  {
1778
  "epoch": 1.7365870543440636,
1779
- "grad_norm": 0.37253957986831665,
1780
  "learning_rate": 1.037297730431085e-05,
1781
- "loss": 0.513,
1782
  "step": 1255
1783
  },
1784
  {
1785
  "epoch": 1.7435098650051921,
1786
- "grad_norm": 0.3707478940486908,
1787
  "learning_rate": 9.843237274502282e-06,
1788
- "loss": 0.5268,
1789
  "step": 1260
1790
  },
1791
  {
1792
  "epoch": 1.7504326756663207,
1793
- "grad_norm": 0.38524284958839417,
1794
  "learning_rate": 9.326680251274777e-06,
1795
- "loss": 0.5357,
1796
  "step": 1265
1797
  },
1798
  {
1799
  "epoch": 1.757355486327449,
1800
- "grad_norm": 0.3852950930595398,
1801
  "learning_rate": 8.823381767227667e-06,
1802
- "loss": 0.5257,
1803
  "step": 1270
1804
  },
1805
  {
1806
  "epoch": 1.7642782969885773,
1807
- "grad_norm": 0.38060450553894043,
1808
  "learning_rate": 8.333415416254831e-06,
1809
- "loss": 0.527,
1810
  "step": 1275
1811
  },
1812
  {
1813
  "epoch": 1.7712011076497056,
1814
- "grad_norm": 0.3773637115955353,
1815
  "learning_rate": 7.856852842783547e-06,
1816
- "loss": 0.5161,
1817
  "step": 1280
1818
  },
1819
  {
1820
  "epoch": 1.7781239183108342,
1821
- "grad_norm": 0.3634498119354248,
1822
  "learning_rate": 7.393763731298387e-06,
1823
- "loss": 0.5116,
1824
  "step": 1285
1825
  },
1826
  {
1827
  "epoch": 1.7850467289719627,
1828
- "grad_norm": 0.3612436056137085,
1829
  "learning_rate": 6.944215796151765e-06,
1830
- "loss": 0.5153,
1831
  "step": 1290
1832
  },
1833
  {
1834
  "epoch": 1.791969539633091,
1835
- "grad_norm": 0.40675196051597595,
1836
  "learning_rate": 6.5082747716625255e-06,
1837
- "loss": 0.5276,
1838
  "step": 1295
1839
  },
1840
  {
1841
  "epoch": 1.7988923502942193,
1842
- "grad_norm": 0.3541720509529114,
1843
  "learning_rate": 6.086004402504098e-06,
1844
- "loss": 0.5096,
1845
  "step": 1300
1846
  },
1847
  {
1848
  "epoch": 1.8058151609553479,
1849
- "grad_norm": 0.38791441917419434,
1850
  "learning_rate": 5.6774664343834495e-06,
1851
- "loss": 0.5346,
1852
  "step": 1305
1853
  },
1854
  {
1855
  "epoch": 1.8127379716164764,
1856
- "grad_norm": 0.37747085094451904,
1857
  "learning_rate": 5.2827206050125585e-06,
1858
- "loss": 0.5218,
1859
  "step": 1310
1860
  },
1861
  {
1862
  "epoch": 1.8196607822776047,
1863
- "grad_norm": 0.3927886486053467,
1864
  "learning_rate": 4.9018246353732755e-06,
1865
- "loss": 0.5352,
1866
  "step": 1315
1867
  },
1868
  {
1869
  "epoch": 1.826583592938733,
1870
- "grad_norm": 0.3697344958782196,
1871
  "learning_rate": 4.534834221277262e-06,
1872
- "loss": 0.5143,
1873
  "step": 1320
1874
  },
1875
  {
1876
  "epoch": 1.8335064035998614,
1877
- "grad_norm": 0.37708741426467896,
1878
  "learning_rate": 4.181803025221898e-06,
1879
- "loss": 0.5243,
1880
  "step": 1325
1881
  },
1882
  {
1883
  "epoch": 1.84042921426099,
1884
- "grad_norm": 0.3722737431526184,
1885
  "learning_rate": 3.842782668543599e-06,
1886
- "loss": 0.5211,
1887
  "step": 1330
1888
  },
1889
  {
1890
  "epoch": 1.8473520249221185,
1891
- "grad_norm": 0.347428560256958,
1892
  "learning_rate": 3.5178227238696347e-06,
1893
- "loss": 0.5123,
1894
  "step": 1335
1895
  },
1896
  {
1897
  "epoch": 1.8542748355832468,
1898
- "grad_norm": 0.3938564658164978,
1899
  "learning_rate": 3.2069707078694057e-06,
1900
- "loss": 0.5079,
1901
  "step": 1340
1902
  },
1903
  {
1904
  "epoch": 1.8611976462443751,
1905
- "grad_norm": 0.37338557839393616,
1906
  "learning_rate": 2.9102720743064527e-06,
1907
- "loss": 0.5287,
1908
  "step": 1345
1909
  },
1910
  {
1911
  "epoch": 1.8681204569055037,
1912
- "grad_norm": 0.3692079484462738,
1913
  "learning_rate": 2.6277702073919997e-06,
1914
- "loss": 0.5069,
1915
  "step": 1350
1916
  },
1917
  {
1918
  "epoch": 1.8750432675666322,
1919
- "grad_norm": 0.3760637938976288,
1920
  "learning_rate": 2.3595064154412374e-06,
1921
- "loss": 0.5129,
1922
  "step": 1355
1923
  },
1924
  {
1925
  "epoch": 1.8819660782277605,
1926
- "grad_norm": 0.38482972979545593,
1927
  "learning_rate": 2.1055199248330546e-06,
1928
- "loss": 0.5163,
1929
  "step": 1360
1930
  },
1931
  {
1932
  "epoch": 1.8888888888888888,
1933
- "grad_norm": 0.3872315585613251,
1934
  "learning_rate": 1.8658478742742625e-06,
1935
- "loss": 0.5157,
1936
  "step": 1365
1937
  },
1938
  {
1939
  "epoch": 1.8958116995500172,
1940
- "grad_norm": 0.3754504919052124,
1941
  "learning_rate": 1.6405253093690343e-06,
1942
- "loss": 0.5132,
1943
  "step": 1370
1944
  },
1945
  {
1946
  "epoch": 1.9027345102111457,
1947
- "grad_norm": 0.3796706199645996,
1948
  "learning_rate": 1.429585177494419e-06,
1949
- "loss": 0.5107,
1950
  "step": 1375
1951
  },
1952
  {
1953
  "epoch": 1.9096573208722742,
1954
- "grad_norm": 0.37216416001319885,
1955
  "learning_rate": 1.2330583229827053e-06,
1956
- "loss": 0.5228,
1957
  "step": 1380
1958
  },
1959
  {
1960
  "epoch": 1.9165801315334026,
1961
- "grad_norm": 0.3697955012321472,
1962
  "learning_rate": 1.0509734826112394e-06,
1963
- "loss": 0.5105,
1964
  "step": 1385
1965
  },
1966
  {
1967
  "epoch": 1.9235029421945309,
1968
- "grad_norm": 0.39311596751213074,
1969
  "learning_rate": 8.833572814004321e-07,
1970
- "loss": 0.5069,
1971
  "step": 1390
1972
  },
1973
  {
1974
  "epoch": 1.9304257528556594,
1975
- "grad_norm": 0.3675869405269623,
1976
  "learning_rate": 7.302342287205943e-07,
1977
- "loss": 0.511,
1978
  "step": 1395
1979
  },
1980
  {
1981
  "epoch": 1.937348563516788,
1982
- "grad_norm": 0.38679271936416626,
1983
  "learning_rate": 5.916267147080934e-07,
1984
- "loss": 0.5175,
1985
  "step": 1400
1986
  },
1987
  {
1988
  "epoch": 1.9442713741779163,
1989
- "grad_norm": 0.3862612247467041,
1990
  "learning_rate": 4.6755500699138253e-07,
1991
- "loss": 0.5075,
1992
  "step": 1405
1993
  },
1994
  {
1995
  "epoch": 1.9511941848390446,
1996
- "grad_norm": 0.3884969651699066,
1997
  "learning_rate": 3.5803724772740523e-07,
1998
- "loss": 0.5132,
1999
  "step": 1410
2000
  },
2001
  {
2002
  "epoch": 1.958116995500173,
2003
- "grad_norm": 0.36823418736457825,
2004
  "learning_rate": 2.630894509488058e-07,
2005
- "loss": 0.5151,
2006
  "step": 1415
2007
  },
2008
  {
2009
  "epoch": 1.9650398061613015,
2010
- "grad_norm": 0.36939066648483276,
2011
  "learning_rate": 1.8272550022230272e-07,
2012
- "loss": 0.5128,
2013
  "step": 1420
2014
  },
2015
  {
2016
  "epoch": 1.97196261682243,
2017
- "grad_norm": 0.38535770773887634,
2018
  "learning_rate": 1.1695714661856816e-07,
2019
- "loss": 0.5019,
2020
  "step": 1425
2021
  },
2022
  {
2023
  "epoch": 1.9788854274835583,
2024
- "grad_norm": 0.3671540915966034,
2025
  "learning_rate": 6.579400699397998e-08,
2026
- "loss": 0.5135,
2027
  "step": 1430
2028
  },
2029
  {
2030
  "epoch": 1.9858082381446867,
2031
- "grad_norm": 0.36039796471595764,
2032
  "learning_rate": 2.9243562584402305e-08,
2033
- "loss": 0.525,
2034
  "step": 1435
2035
  },
2036
  {
2037
  "epoch": 1.9927310488058152,
2038
- "grad_norm": 0.3742762804031372,
2039
  "learning_rate": 7.311157911249478e-09,
2040
- "loss": 0.5266,
2041
  "step": 1440
2042
  },
2043
  {
2044
  "epoch": 1.9982692973347178,
2045
- "eval_loss": 0.7263619303703308,
2046
- "eval_runtime": 60.0642,
2047
- "eval_samples_per_second": 34.446,
2048
- "eval_steps_per_second": 2.164,
2049
  "step": 1444
2050
  },
2051
  {
2052
  "epoch": 1.9982692973347178,
2053
  "step": 1444,
2054
  "total_flos": 4.774796461726499e+18,
2055
- "train_loss": 0.7511481684000538,
2056
- "train_runtime": 11380.9982,
2057
- "train_samples_per_second": 8.122,
2058
  "train_steps_per_second": 0.127
2059
  }
2060
  ],
 
11
  "log_history": [
12
  {
13
  "epoch": 0.0013845621322256837,
14
+ "grad_norm": 49.28921127319336,
15
  "learning_rate": 0.0,
16
+ "loss": 3.0795,
17
  "step": 1
18
  },
19
  {
20
  "epoch": 0.006922810661128418,
21
+ "grad_norm": 18.114784240722656,
22
  "learning_rate": 5.517241379310345e-06,
23
+ "loss": 2.9478,
24
  "step": 5
25
  },
26
  {
27
  "epoch": 0.013845621322256836,
28
+ "grad_norm": 6.451106071472168,
29
  "learning_rate": 1.2413793103448277e-05,
30
+ "loss": 2.3094,
31
  "step": 10
32
  },
33
  {
34
  "epoch": 0.020768431983385256,
35
+ "grad_norm": 7.723776817321777,
36
  "learning_rate": 1.9310344827586207e-05,
37
+ "loss": 1.9825,
38
  "step": 15
39
  },
40
  {
41
  "epoch": 0.027691242644513673,
42
+ "grad_norm": 9.21464729309082,
43
  "learning_rate": 2.620689655172414e-05,
44
+ "loss": 1.7586,
45
  "step": 20
46
  },
47
  {
48
  "epoch": 0.034614053305642094,
49
+ "grad_norm": 7.424780368804932,
50
  "learning_rate": 3.310344827586207e-05,
51
+ "loss": 1.5374,
52
  "step": 25
53
  },
54
  {
55
  "epoch": 0.04153686396677051,
56
+ "grad_norm": 3.526257038116455,
57
  "learning_rate": 4e-05,
58
+ "loss": 1.28,
59
  "step": 30
60
  },
61
  {
62
  "epoch": 0.04845967462789893,
63
+ "grad_norm": 2.5668952465057373,
64
  "learning_rate": 4.689655172413793e-05,
65
+ "loss": 1.2187,
66
  "step": 35
67
  },
68
  {
69
  "epoch": 0.055382485289027346,
70
+ "grad_norm": 3.9412596225738525,
71
  "learning_rate": 5.379310344827586e-05,
72
+ "loss": 1.1574,
73
  "step": 40
74
  },
75
  {
76
  "epoch": 0.06230529595015576,
77
+ "grad_norm": 2.965135097503662,
78
  "learning_rate": 6.068965517241379e-05,
79
+ "loss": 1.089,
80
  "step": 45
81
  },
82
  {
83
  "epoch": 0.06922810661128419,
84
+ "grad_norm": 2.2038655281066895,
85
  "learning_rate": 6.758620689655173e-05,
86
+ "loss": 1.0374,
87
  "step": 50
88
  },
89
  {
90
  "epoch": 0.0761509172724126,
91
+ "grad_norm": 1.856432557106018,
92
  "learning_rate": 7.448275862068966e-05,
93
+ "loss": 1.0215,
94
  "step": 55
95
  },
96
  {
97
  "epoch": 0.08307372793354102,
98
+ "grad_norm": 4.251297473907471,
99
  "learning_rate": 8.137931034482759e-05,
100
+ "loss": 1.1169,
101
  "step": 60
102
  },
103
  {
104
  "epoch": 0.08999653859466944,
105
+ "grad_norm": 4.693065166473389,
106
  "learning_rate": 8.827586206896552e-05,
107
+ "loss": 1.0188,
108
  "step": 65
109
  },
110
  {
111
  "epoch": 0.09691934925579786,
112
+ "grad_norm": 2.24104642868042,
113
  "learning_rate": 9.517241379310345e-05,
114
+ "loss": 1.0059,
115
  "step": 70
116
  },
117
  {
118
  "epoch": 0.10384215991692627,
119
+ "grad_norm": 1.4557360410690308,
120
  "learning_rate": 0.0001020689655172414,
121
+ "loss": 0.9839,
122
  "step": 75
123
  },
124
  {
125
  "epoch": 0.11076497057805469,
126
+ "grad_norm": 1.4282116889953613,
127
  "learning_rate": 0.00010896551724137931,
128
+ "loss": 1.01,
129
  "step": 80
130
  },
131
  {
132
  "epoch": 0.11768778123918311,
133
+ "grad_norm": 1.2591500282287598,
134
  "learning_rate": 0.00011586206896551725,
135
+ "loss": 1.0043,
136
  "step": 85
137
  },
138
  {
139
  "epoch": 0.12461059190031153,
140
+ "grad_norm": 1.0611705780029297,
141
  "learning_rate": 0.00012275862068965518,
142
+ "loss": 1.0127,
143
  "step": 90
144
  },
145
  {
146
  "epoch": 0.13153340256143994,
147
+ "grad_norm": 1.1061965227127075,
148
  "learning_rate": 0.0001296551724137931,
149
+ "loss": 0.9933,
150
  "step": 95
151
  },
152
  {
153
  "epoch": 0.13845621322256838,
154
+ "grad_norm": 1.3926725387573242,
155
  "learning_rate": 0.00013655172413793104,
156
+ "loss": 1.0056,
157
  "step": 100
158
  },
159
  {
160
  "epoch": 0.14537902388369678,
161
+ "grad_norm": 0.986765444278717,
162
  "learning_rate": 0.00014344827586206896,
163
+ "loss": 1.0034,
164
  "step": 105
165
  },
166
  {
167
  "epoch": 0.1523018345448252,
168
+ "grad_norm": 0.9959716796875,
169
  "learning_rate": 0.0001503448275862069,
170
+ "loss": 1.0215,
171
  "step": 110
172
  },
173
  {
174
  "epoch": 0.1592246452059536,
175
+ "grad_norm": 0.9828895926475525,
176
  "learning_rate": 0.00015724137931034485,
177
+ "loss": 1.0114,
178
  "step": 115
179
  },
180
  {
181
  "epoch": 0.16614745586708204,
182
+ "grad_norm": 0.9268749356269836,
183
  "learning_rate": 0.00016413793103448276,
184
+ "loss": 1.004,
185
  "step": 120
186
  },
187
  {
188
  "epoch": 0.17307026652821045,
189
+ "grad_norm": 0.949165403842926,
190
  "learning_rate": 0.0001710344827586207,
191
+ "loss": 1.0372,
192
  "step": 125
193
  },
194
  {
195
  "epoch": 0.17999307718933888,
196
+ "grad_norm": 0.7588331699371338,
197
  "learning_rate": 0.00017793103448275862,
198
+ "loss": 1.0159,
199
  "step": 130
200
  },
201
  {
202
  "epoch": 0.18691588785046728,
203
+ "grad_norm": 0.846207857131958,
204
  "learning_rate": 0.00018482758620689654,
205
+ "loss": 1.0207,
206
  "step": 135
207
  },
208
  {
209
  "epoch": 0.19383869851159571,
210
+ "grad_norm": 0.9587262272834778,
211
  "learning_rate": 0.0001917241379310345,
212
+ "loss": 1.0175,
213
  "step": 140
214
  },
215
  {
216
  "epoch": 0.20076150917272412,
217
+ "grad_norm": 0.8179382681846619,
218
  "learning_rate": 0.00019862068965517243,
219
+ "loss": 1.0181,
220
  "step": 145
221
  },
222
  {
223
  "epoch": 0.20768431983385255,
224
+ "grad_norm": 0.8082414865493774,
225
  "learning_rate": 0.0001999953208384105,
226
+ "loss": 1.0108,
227
  "step": 150
228
  },
229
  {
230
  "epoch": 0.21460713049498095,
231
+ "grad_norm": 0.7612893581390381,
232
  "learning_rate": 0.0001999763124949378,
233
+ "loss": 0.9901,
234
  "step": 155
235
  },
236
  {
237
  "epoch": 0.22152994115610938,
238
+ "grad_norm": 0.8941105008125305,
239
  "learning_rate": 0.0001999426852993044,
240
+ "loss": 0.9983,
241
  "step": 160
242
  },
243
  {
244
  "epoch": 0.2284527518172378,
245
+ "grad_norm": 0.7896144390106201,
246
  "learning_rate": 0.00019989444416858502,
247
+ "loss": 0.9707,
248
  "step": 165
249
  },
250
  {
251
  "epoch": 0.23537556247836622,
252
+ "grad_norm": 0.735113799571991,
253
  "learning_rate": 0.0001998315961567502,
254
+ "loss": 0.9955,
255
  "step": 170
256
  },
257
  {
258
  "epoch": 0.24229837313949462,
259
+ "grad_norm": 0.7563541531562805,
260
  "learning_rate": 0.00019975415045363467,
261
+ "loss": 0.9878,
262
  "step": 175
263
  },
264
  {
265
  "epoch": 0.24922118380062305,
266
+ "grad_norm": 0.8806726932525635,
267
  "learning_rate": 0.00019966211838359378,
268
+ "loss": 0.976,
269
  "step": 180
270
  },
271
  {
272
  "epoch": 0.25614399446175146,
273
+ "grad_norm": 0.7841496467590332,
274
  "learning_rate": 0.00019955551340384743,
275
+ "loss": 0.9606,
276
  "step": 185
277
  },
278
  {
279
  "epoch": 0.2630668051228799,
280
+ "grad_norm": 0.5848103761672974,
281
  "learning_rate": 0.00019943435110251248,
282
+ "loss": 0.9877,
283
  "step": 190
284
  },
285
  {
286
  "epoch": 0.2699896157840083,
287
+ "grad_norm": 0.7116856575012207,
288
  "learning_rate": 0.0001992986491963232,
289
+ "loss": 0.955,
290
  "step": 195
291
  },
292
  {
293
  "epoch": 0.27691242644513675,
294
+ "grad_norm": 0.6569689512252808,
295
  "learning_rate": 0.00019914842752804103,
296
+ "loss": 0.9524,
297
  "step": 200
298
  },
299
  {
300
  "epoch": 0.2838352371062651,
301
+ "grad_norm": 0.7533493638038635,
302
  "learning_rate": 0.00019898370806355263,
303
+ "loss": 0.9232,
304
  "step": 205
305
  },
306
  {
307
  "epoch": 0.29075804776739356,
308
+ "grad_norm": 0.7606753706932068,
309
  "learning_rate": 0.00019880451488865836,
310
+ "loss": 0.9854,
311
  "step": 210
312
  },
313
  {
314
  "epoch": 0.297680858428522,
315
+ "grad_norm": 7.585168361663818,
316
  "learning_rate": 0.00019861087420555018,
317
+ "loss": 1.1045,
318
  "step": 215
319
  },
320
  {
321
  "epoch": 0.3046036690896504,
322
+ "grad_norm": 9.095848083496094,
323
  "learning_rate": 0.0001984028143289803,
324
+ "loss": 1.2125,
325
  "step": 220
326
  },
327
  {
328
  "epoch": 0.3115264797507788,
329
+ "grad_norm": 1.6645023822784424,
330
  "learning_rate": 0.00019818036568212106,
331
+ "loss": 1.0607,
332
  "step": 225
333
  },
334
  {
335
  "epoch": 0.3184492904119072,
336
+ "grad_norm": 1.3035401105880737,
337
  "learning_rate": 0.00019794356079211604,
338
+ "loss": 0.9415,
339
  "step": 230
340
  },
341
  {
342
  "epoch": 0.32537210107303566,
343
+ "grad_norm": 0.7202489376068115,
344
  "learning_rate": 0.00019769243428532422,
345
+ "loss": 0.9706,
346
  "step": 235
347
  },
348
  {
349
  "epoch": 0.3322949117341641,
350
+ "grad_norm": 0.8300184011459351,
351
  "learning_rate": 0.00019742702288225652,
352
+ "loss": 0.926,
353
  "step": 240
354
  },
355
  {
356
  "epoch": 0.33921772239529246,
357
+ "grad_norm": 0.6376895308494568,
358
  "learning_rate": 0.00019714736539220648,
359
+ "loss": 0.9173,
360
  "step": 245
361
  },
362
  {
363
  "epoch": 0.3461405330564209,
364
+ "grad_norm": 0.7818527221679688,
365
  "learning_rate": 0.00019685350270757555,
366
+ "loss": 0.9413,
367
  "step": 250
368
  },
369
  {
370
  "epoch": 0.3530633437175493,
371
+ "grad_norm": 0.6731443405151367,
372
  "learning_rate": 0.0001965454777978936,
373
+ "loss": 0.954,
374
  "step": 255
375
  },
376
  {
377
  "epoch": 0.35998615437867776,
378
+ "grad_norm": 3.000696897506714,
379
  "learning_rate": 0.00019622333570353567,
380
+ "loss": 0.9432,
381
  "step": 260
382
  },
383
  {
384
  "epoch": 0.36690896503980613,
385
+ "grad_norm": 0.7889404892921448,
386
  "learning_rate": 0.00019588712352913625,
387
+ "loss": 0.9471,
388
  "step": 265
389
  },
390
  {
391
  "epoch": 0.37383177570093457,
392
+ "grad_norm": 0.8370787501335144,
393
  "learning_rate": 0.00019553689043670127,
394
+ "loss": 0.9521,
395
  "step": 270
396
  },
397
  {
398
  "epoch": 0.380754586362063,
399
+ "grad_norm": 0.6203432679176331,
400
  "learning_rate": 0.00019517268763841962,
401
+ "loss": 0.9263,
402
  "step": 275
403
  },
404
  {
405
  "epoch": 0.38767739702319143,
406
+ "grad_norm": 0.5970066785812378,
407
  "learning_rate": 0.00019479456838917473,
408
+ "loss": 0.9014,
409
  "step": 280
410
  },
411
  {
412
  "epoch": 0.39460020768431986,
413
+ "grad_norm": 0.6280853748321533,
414
  "learning_rate": 0.0001944025879787574,
415
+ "loss": 0.8933,
416
  "step": 285
417
  },
418
  {
419
  "epoch": 0.40152301834544823,
420
+ "grad_norm": 0.8126713037490845,
421
  "learning_rate": 0.0001939968037237812,
422
+ "loss": 0.9203,
423
  "step": 290
424
  },
425
  {
426
  "epoch": 0.40844582900657667,
427
+ "grad_norm": 0.5656815767288208,
428
  "learning_rate": 0.00019357727495930146,
429
+ "loss": 0.8971,
430
  "step": 295
431
  },
432
  {
433
  "epoch": 0.4153686396677051,
434
+ "grad_norm": 0.6746362447738647,
435
  "learning_rate": 0.00019314406303013904,
436
+ "loss": 0.8974,
437
  "step": 300
438
  },
439
  {
440
  "epoch": 0.42229145032883353,
441
+ "grad_norm": 0.6263439655303955,
442
  "learning_rate": 0.00019269723128191048,
443
+ "loss": 0.891,
444
  "step": 305
445
  },
446
  {
447
  "epoch": 0.4292142609899619,
448
+ "grad_norm": 0.4856240451335907,
449
  "learning_rate": 0.00019223684505176517,
450
+ "loss": 0.8999,
451
  "step": 310
452
  },
453
  {
454
  "epoch": 0.43613707165109034,
455
+ "grad_norm": 0.5870519280433655,
456
  "learning_rate": 0.00019176297165883165,
457
+ "loss": 0.9049,
458
  "step": 315
459
  },
460
  {
461
  "epoch": 0.44305988231221877,
462
+ "grad_norm": 0.5149969458580017,
463
  "learning_rate": 0.00019127568039437406,
464
+ "loss": 0.886,
465
  "step": 320
466
  },
467
  {
468
  "epoch": 0.4499826929733472,
469
+ "grad_norm": 0.5945581793785095,
470
  "learning_rate": 0.00019077504251166006,
471
+ "loss": 0.8818,
472
  "step": 325
473
  },
474
  {
475
  "epoch": 0.4569055036344756,
476
+ "grad_norm": 0.5872354507446289,
477
  "learning_rate": 0.00019026113121554204,
478
+ "loss": 0.8882,
479
  "step": 330
480
  },
481
  {
482
  "epoch": 0.463828314295604,
483
+ "grad_norm": 0.5640285611152649,
484
  "learning_rate": 0.00018973402165175268,
485
+ "loss": 0.8928,
486
  "step": 335
487
  },
488
  {
489
  "epoch": 0.47075112495673244,
490
+ "grad_norm": 0.5472027659416199,
491
  "learning_rate": 0.0001891937908959172,
492
+ "loss": 0.8807,
493
  "step": 340
494
  },
495
  {
496
  "epoch": 0.47767393561786087,
497
+ "grad_norm": 0.5739114284515381,
498
  "learning_rate": 0.00018864051794228282,
499
+ "loss": 0.8869,
500
  "step": 345
501
  },
502
  {
503
  "epoch": 0.48459674627898924,
504
+ "grad_norm": 0.580653190612793,
505
  "learning_rate": 0.00018807428369216822,
506
+ "loss": 0.8543,
507
  "step": 350
508
  },
509
  {
510
  "epoch": 0.4915195569401177,
511
+ "grad_norm": 0.5018455982208252,
512
  "learning_rate": 0.00018749517094213375,
513
+ "loss": 0.8548,
514
  "step": 355
515
  },
516
  {
517
  "epoch": 0.4984423676012461,
518
+ "grad_norm": 0.5900467038154602,
519
  "learning_rate": 0.00018690326437187472,
520
+ "loss": 0.8685,
521
  "step": 360
522
  },
523
  {
524
  "epoch": 0.5053651782623745,
525
+ "grad_norm": 0.5394186973571777,
526
  "learning_rate": 0.00018629865053183911,
527
+ "loss": 0.8626,
528
  "step": 365
529
  },
530
  {
531
  "epoch": 0.5122879889235029,
532
+ "grad_norm": 0.6365518569946289,
533
  "learning_rate": 0.0001856814178305722,
534
+ "loss": 0.8474,
535
  "step": 370
536
  },
537
  {
538
  "epoch": 0.5192107995846313,
539
+ "grad_norm": 0.5121780037879944,
540
  "learning_rate": 0.00018505165652178893,
541
+ "loss": 0.8448,
542
  "step": 375
543
  },
544
  {
545
  "epoch": 0.5261336102457598,
546
+ "grad_norm": 0.5004256367683411,
547
  "learning_rate": 0.00018440945869117675,
548
+ "loss": 0.8681,
549
  "step": 380
550
  },
551
  {
552
  "epoch": 0.5330564209068882,
553
+ "grad_norm": 0.5059666633605957,
554
  "learning_rate": 0.00018375491824293067,
555
+ "loss": 0.8752,
556
  "step": 385
557
  },
558
  {
559
  "epoch": 0.5399792315680166,
560
+ "grad_norm": 0.4677795171737671,
561
  "learning_rate": 0.00018308813088602227,
562
+ "loss": 0.8695,
563
  "step": 390
564
  },
565
  {
566
  "epoch": 0.5469020422291451,
567
+ "grad_norm": 0.5352007746696472,
568
  "learning_rate": 0.00018240919412020466,
569
+ "loss": 0.8468,
570
  "step": 395
571
  },
572
  {
573
  "epoch": 0.5538248528902735,
574
+ "grad_norm": 0.5106504559516907,
575
  "learning_rate": 0.00018171820722175604,
576
+ "loss": 0.8699,
577
  "step": 400
578
  },
579
  {
580
  "epoch": 0.5607476635514018,
581
+ "grad_norm": 0.4898536801338196,
582
  "learning_rate": 0.00018101527122896293,
583
+ "loss": 0.8607,
584
  "step": 405
585
  },
586
  {
587
  "epoch": 0.5676704742125303,
588
+ "grad_norm": 0.49010294675827026,
589
  "learning_rate": 0.0001803004889273463,
590
+ "loss": 0.8462,
591
  "step": 410
592
  },
593
  {
594
  "epoch": 0.5745932848736587,
595
+ "grad_norm": 0.4630247950553894,
596
  "learning_rate": 0.0001795739648346317,
597
+ "loss": 0.8243,
598
  "step": 415
599
  },
600
  {
601
  "epoch": 0.5815160955347871,
602
+ "grad_norm": 0.4984998404979706,
603
  "learning_rate": 0.00017883580518546647,
604
+ "loss": 0.8262,
605
  "step": 420
606
  },
607
  {
608
  "epoch": 0.5884389061959155,
609
+ "grad_norm": 0.5157510638237,
610
  "learning_rate": 0.00017808611791588584,
611
+ "loss": 0.8397,
612
  "step": 425
613
  },
614
  {
615
  "epoch": 0.595361716857044,
616
+ "grad_norm": 0.48661330342292786,
617
  "learning_rate": 0.00017732501264752996,
618
+ "loss": 0.8267,
619
  "step": 430
620
  },
621
  {
622
  "epoch": 0.6022845275181724,
623
+ "grad_norm": 0.47257551550865173,
624
  "learning_rate": 0.00017655260067161503,
625
+ "loss": 0.8519,
626
  "step": 435
627
  },
628
  {
629
  "epoch": 0.6092073381793008,
630
+ "grad_norm": 0.5127261877059937,
631
  "learning_rate": 0.00017576899493265954,
632
+ "loss": 0.8355,
633
  "step": 440
634
  },
635
  {
636
  "epoch": 0.6161301488404292,
637
+ "grad_norm": 0.45928096771240234,
638
  "learning_rate": 0.00017497431001196943,
639
+ "loss": 0.8423,
640
  "step": 445
641
  },
642
  {
643
  "epoch": 0.6230529595015576,
644
+ "grad_norm": 0.45476192235946655,
645
  "learning_rate": 0.00017416866211088366,
646
+ "loss": 0.847,
647
  "step": 450
648
  },
649
  {
650
  "epoch": 0.629975770162686,
651
+ "grad_norm": 0.48962315917015076,
652
  "learning_rate": 0.00017335216903378267,
653
+ "loss": 0.8317,
654
  "step": 455
655
  },
656
  {
657
  "epoch": 0.6368985808238145,
658
+ "grad_norm": 0.5388792753219604,
659
  "learning_rate": 0.00017252495017086296,
660
+ "loss": 0.8223,
661
  "step": 460
662
  },
663
  {
664
  "epoch": 0.6438213914849429,
665
+ "grad_norm": 0.4656524658203125,
666
  "learning_rate": 0.0001716871264806791,
667
+ "loss": 0.8355,
668
  "step": 465
669
  },
670
  {
671
  "epoch": 0.6507442021460713,
672
+ "grad_norm": 0.5120430588722229,
673
  "learning_rate": 0.0001708388204724572,
674
+ "loss": 0.8331,
675
  "step": 470
676
  },
677
  {
678
  "epoch": 0.6576670128071997,
679
+ "grad_norm": 0.48045217990875244,
680
  "learning_rate": 0.00016998015618818087,
681
+ "loss": 0.8478,
682
  "step": 475
683
  },
684
  {
685
  "epoch": 0.6645898234683282,
686
+ "grad_norm": 0.44614362716674805,
687
  "learning_rate": 0.00016911125918445363,
688
+ "loss": 0.8286,
689
  "step": 480
690
  },
691
  {
692
  "epoch": 0.6715126341294566,
693
+ "grad_norm": 0.44099363684654236,
694
  "learning_rate": 0.00016823225651413953,
695
+ "loss": 0.8185,
696
  "step": 485
697
  },
698
  {
699
  "epoch": 0.6784354447905849,
700
+ "grad_norm": 0.45090341567993164,
701
  "learning_rate": 0.00016734327670778508,
702
+ "loss": 0.8107,
703
  "step": 490
704
  },
705
  {
706
  "epoch": 0.6853582554517134,
707
+ "grad_norm": 0.5897481441497803,
708
  "learning_rate": 0.00016644444975482523,
709
+ "loss": 0.8067,
710
  "step": 495
711
  },
712
  {
713
  "epoch": 0.6922810661128418,
714
+ "grad_norm": 0.5060038566589355,
715
  "learning_rate": 0.0001655359070845757,
716
+ "loss": 0.8121,
717
  "step": 500
718
  },
719
  {
720
  "epoch": 0.6992038767739702,
721
+ "grad_norm": 0.4440653324127197,
722
  "learning_rate": 0.0001646177815470151,
723
+ "loss": 0.8181,
724
  "step": 505
725
  },
726
  {
727
  "epoch": 0.7061266874350987,
728
+ "grad_norm": 0.4549269676208496,
729
  "learning_rate": 0.00016369020739335924,
730
+ "loss": 0.7997,
731
  "step": 510
732
  },
733
  {
734
  "epoch": 0.7130494980962271,
735
+ "grad_norm": 0.45471733808517456,
736
  "learning_rate": 0.00016275332025643028,
737
+ "loss": 0.8154,
738
  "step": 515
739
  },
740
  {
741
  "epoch": 0.7199723087573555,
742
+ "grad_norm": 0.4292679727077484,
743
  "learning_rate": 0.00016180725713082434,
744
+ "loss": 0.8234,
745
  "step": 520
746
  },
747
  {
748
  "epoch": 0.726895119418484,
749
+ "grad_norm": 0.5121740698814392,
750
  "learning_rate": 0.0001608521563528795,
751
+ "loss": 0.8044,
752
  "step": 525
753
  },
754
  {
755
  "epoch": 0.7338179300796123,
756
+ "grad_norm": 0.44355785846710205,
757
  "learning_rate": 0.00015988815758044792,
758
+ "loss": 0.7996,
759
  "step": 530
760
  },
761
  {
762
  "epoch": 0.7407407407407407,
763
+ "grad_norm": 0.42263615131378174,
764
  "learning_rate": 0.00015891540177247467,
765
+ "loss": 0.8219,
766
  "step": 535
767
  },
768
  {
769
  "epoch": 0.7476635514018691,
770
+ "grad_norm": 0.5358946919441223,
771
  "learning_rate": 0.00015793403116838613,
772
+ "loss": 0.8089,
773
  "step": 540
774
  },
775
  {
776
  "epoch": 0.7545863620629976,
777
+ "grad_norm": 0.47450023889541626,
778
  "learning_rate": 0.00015694418926729146,
779
+ "loss": 0.7956,
780
  "step": 545
781
  },
782
  {
783
  "epoch": 0.761509172724126,
784
+ "grad_norm": 0.470217227935791,
785
  "learning_rate": 0.00015594602080699956,
786
+ "loss": 0.8058,
787
  "step": 550
788
  },
789
  {
790
  "epoch": 0.7684319833852544,
791
+ "grad_norm": 0.5145338773727417,
792
  "learning_rate": 0.00015493967174285512,
793
+ "loss": 0.7972,
794
  "step": 555
795
  },
796
  {
797
  "epoch": 0.7753547940463829,
798
+ "grad_norm": 0.4233357310295105,
799
  "learning_rate": 0.00015392528922639662,
800
+ "loss": 0.8084,
801
  "step": 560
802
  },
803
  {
804
  "epoch": 0.7822776047075113,
805
+ "grad_norm": 0.41947290301322937,
806
  "learning_rate": 0.00015290302158383923,
807
+ "loss": 0.7981,
808
  "step": 565
809
  },
810
  {
811
  "epoch": 0.7892004153686397,
812
+ "grad_norm": 0.46518903970718384,
813
  "learning_rate": 0.0001518730182943862,
814
+ "loss": 0.7962,
815
  "step": 570
816
  },
817
  {
818
  "epoch": 0.796123226029768,
819
+ "grad_norm": 0.41341090202331543,
820
  "learning_rate": 0.0001508354299683715,
821
+ "loss": 0.7952,
822
  "step": 575
823
  },
824
  {
825
  "epoch": 0.8030460366908965,
826
+ "grad_norm": 0.45617249608039856,
827
  "learning_rate": 0.0001497904083252371,
828
+ "loss": 0.8015,
829
  "step": 580
830
  },
831
  {
832
  "epoch": 0.8099688473520249,
833
+ "grad_norm": 0.41003215312957764,
834
  "learning_rate": 0.00014873810617134806,
835
+ "loss": 0.7894,
836
  "step": 585
837
  },
838
  {
839
  "epoch": 0.8168916580131533,
840
+ "grad_norm": 0.4336972236633301,
841
  "learning_rate": 0.00014767867737764876,
842
+ "loss": 0.7886,
843
  "step": 590
844
  },
845
  {
846
  "epoch": 0.8238144686742818,
847
+ "grad_norm": 0.43226006627082825,
848
  "learning_rate": 0.0001466122768571633,
849
+ "loss": 0.7973,
850
  "step": 595
851
  },
852
  {
853
  "epoch": 0.8307372793354102,
854
+ "grad_norm": 0.4515872597694397,
855
  "learning_rate": 0.0001455390605423437,
856
+ "loss": 0.8025,
857
  "step": 600
858
  },
859
  {
860
  "epoch": 0.8376600899965386,
861
+ "grad_norm": 0.49443331360816956,
862
  "learning_rate": 0.00014445918536226904,
863
+ "loss": 0.7883,
864
  "step": 605
865
  },
866
  {
867
  "epoch": 0.8445829006576671,
868
+ "grad_norm": 0.48309868574142456,
869
  "learning_rate": 0.00014337280921969858,
870
+ "loss": 0.7824,
871
  "step": 610
872
  },
873
  {
874
  "epoch": 0.8515057113187954,
875
+ "grad_norm": 0.4243367314338684,
876
  "learning_rate": 0.0001422800909679829,
877
+ "loss": 0.7928,
878
  "step": 615
879
  },
880
  {
881
  "epoch": 0.8584285219799238,
882
+ "grad_norm": 0.49469149112701416,
883
  "learning_rate": 0.00014118119038783588,
884
+ "loss": 0.785,
885
  "step": 620
886
  },
887
  {
888
  "epoch": 0.8653513326410522,
889
+ "grad_norm": 0.4584200084209442,
890
  "learning_rate": 0.00014007626816397087,
891
+ "loss": 0.804,
892
  "step": 625
893
  },
894
  {
895
  "epoch": 0.8722741433021807,
896
+ "grad_norm": 0.4225427210330963,
897
  "learning_rate": 0.00013896548586160503,
898
+ "loss": 0.7888,
899
  "step": 630
900
  },
901
  {
902
  "epoch": 0.8791969539633091,
903
+ "grad_norm": 0.45131561160087585,
904
  "learning_rate": 0.00013784900590283473,
905
+ "loss": 0.7906,
906
  "step": 635
907
  },
908
  {
909
  "epoch": 0.8861197646244375,
910
+ "grad_norm": 0.4222111999988556,
911
  "learning_rate": 0.00013672699154288569,
912
+ "loss": 0.7748,
913
  "step": 640
914
  },
915
  {
916
  "epoch": 0.893042575285566,
917
+ "grad_norm": 0.4449770748615265,
918
  "learning_rate": 0.00013559960684624112,
919
+ "loss": 0.7995,
920
  "step": 645
921
  },
922
  {
923
  "epoch": 0.8999653859466944,
924
+ "grad_norm": 0.4661027491092682,
925
  "learning_rate": 0.0001344670166626519,
926
+ "loss": 0.7682,
927
  "step": 650
928
  },
929
  {
930
  "epoch": 0.9068881966078228,
931
+ "grad_norm": 0.4581912159919739,
932
  "learning_rate": 0.0001333293866030317,
933
+ "loss": 0.7728,
934
  "step": 655
935
  },
936
  {
937
  "epoch": 0.9138110072689511,
938
+ "grad_norm": 0.41657838225364685,
939
  "learning_rate": 0.0001321868830152407,
940
+ "loss": 0.7557,
941
  "step": 660
942
  },
943
  {
944
  "epoch": 0.9207338179300796,
945
+ "grad_norm": 0.4423748254776001,
946
  "learning_rate": 0.00013103967295976179,
947
+ "loss": 0.7798,
948
  "step": 665
949
  },
950
  {
951
  "epoch": 0.927656628591208,
952
+ "grad_norm": 0.411111980676651,
953
  "learning_rate": 0.0001298879241852724,
954
+ "loss": 0.7777,
955
  "step": 670
956
  },
957
  {
958
  "epoch": 0.9345794392523364,
959
+ "grad_norm": 0.4396058917045593,
960
  "learning_rate": 0.0001287318051041159,
961
+ "loss": 0.778,
962
  "step": 675
963
  },
964
  {
965
  "epoch": 0.9415022499134649,
966
+ "grad_norm": 0.4244561493396759,
967
  "learning_rate": 0.00012757148476767553,
968
+ "loss": 0.7698,
969
  "step": 680
970
  },
971
  {
972
  "epoch": 0.9484250605745933,
973
+ "grad_norm": 0.3860670328140259,
974
  "learning_rate": 0.0001264071328416555,
975
+ "loss": 0.772,
976
  "step": 685
977
  },
978
  {
979
  "epoch": 0.9553478712357217,
980
+ "grad_norm": 0.38453128933906555,
981
  "learning_rate": 0.00012523891958127168,
982
+ "loss": 0.7689,
983
  "step": 690
984
  },
985
  {
986
  "epoch": 0.9622706818968502,
987
+ "grad_norm": 0.41507506370544434,
988
  "learning_rate": 0.0001240670158063565,
989
+ "loss": 0.7662,
990
  "step": 695
991
  },
992
  {
993
  "epoch": 0.9691934925579785,
994
+ "grad_norm": 0.4023298919200897,
995
  "learning_rate": 0.00012289159287638105,
996
+ "loss": 0.7721,
997
  "step": 700
998
  },
999
  {
1000
  "epoch": 0.9761163032191069,
1001
+ "grad_norm": 0.41484305262565613,
1002
  "learning_rate": 0.00012171282266539845,
1003
+ "loss": 0.7707,
1004
  "step": 705
1005
  },
1006
  {
1007
  "epoch": 0.9830391138802353,
1008
+ "grad_norm": 0.4068526327610016,
1009
  "learning_rate": 0.00012053087753691172,
1010
+ "loss": 0.7547,
1011
  "step": 710
1012
  },
1013
  {
1014
  "epoch": 0.9899619245413638,
1015
+ "grad_norm": 0.4563179910182953,
1016
  "learning_rate": 0.00011934593031867038,
1017
+ "loss": 0.7691,
1018
  "step": 715
1019
  },
1020
  {
1021
  "epoch": 0.9968847352024922,
1022
+ "grad_norm": 0.3979925513267517,
1023
  "learning_rate": 0.00011815815427739911,
1024
  "loss": 0.7706,
1025
  "step": 720
1026
  },
1027
  {
1028
  "epoch": 1.0,
1029
+ "eval_loss": 0.7720561027526855,
1030
+ "eval_runtime": 59.9017,
1031
+ "eval_samples_per_second": 34.54,
1032
+ "eval_steps_per_second": 2.17,
1033
  "step": 723
1034
  },
1035
  {
1036
  "epoch": 1.0027691242644514,
1037
+ "grad_norm": 0.42567870020866394,
1038
  "learning_rate": 0.00011696772309346182,
1039
+ "loss": 0.6892,
1040
  "step": 725
1041
  },
1042
  {
1043
  "epoch": 1.0096919349255797,
1044
+ "grad_norm": 0.45581841468811035,
1045
  "learning_rate": 0.0001157748108354659,
1046
+ "loss": 0.5876,
1047
  "step": 730
1048
  },
1049
  {
1050
  "epoch": 1.0166147455867083,
1051
+ "grad_norm": 0.3980908989906311,
1052
  "learning_rate": 0.00011457959193480925,
1053
+ "loss": 0.5892,
1054
  "step": 735
1055
  },
1056
  {
1057
  "epoch": 1.0235375562478366,
1058
+ "grad_norm": 0.41051357984542847,
1059
  "learning_rate": 0.00011338224116017423,
1060
+ "loss": 0.5849,
1061
  "step": 740
1062
  },
1063
  {
1064
  "epoch": 1.0304603669089651,
1065
+ "grad_norm": 0.4377879500389099,
1066
  "learning_rate": 0.00011218293359197267,
1067
+ "loss": 0.5918,
1068
  "step": 745
1069
  },
1070
  {
1071
  "epoch": 1.0373831775700935,
1072
+ "grad_norm": 0.495897114276886,
1073
  "learning_rate": 0.00011098184459674484,
1074
+ "loss": 0.5653,
1075
  "step": 750
1076
  },
1077
  {
1078
  "epoch": 1.0443059882312218,
1079
+ "grad_norm": 0.4108388125896454,
1080
  "learning_rate": 0.00010977914980151691,
1081
+ "loss": 0.5915,
1082
  "step": 755
1083
  },
1084
  {
1085
  "epoch": 1.0512287988923503,
1086
+ "grad_norm": 0.40004926919937134,
1087
  "learning_rate": 0.00010857502506812029,
1088
+ "loss": 0.5689,
1089
  "step": 760
1090
  },
1091
  {
1092
  "epoch": 1.0581516095534786,
1093
+ "grad_norm": 0.4072938561439514,
1094
  "learning_rate": 0.00010736964646747629,
1095
+ "loss": 0.5727,
1096
  "step": 765
1097
  },
1098
  {
1099
  "epoch": 1.0650744202146072,
1100
+ "grad_norm": 0.3785325586795807,
1101
  "learning_rate": 0.00010616319025385089,
1102
+ "loss": 0.5846,
1103
  "step": 770
1104
  },
1105
  {
1106
  "epoch": 1.0719972308757355,
1107
+ "grad_norm": 0.3935575485229492,
1108
  "learning_rate": 0.00010495583283908177,
1109
+ "loss": 0.5753,
1110
  "step": 775
1111
  },
1112
  {
1113
  "epoch": 1.078920041536864,
1114
+ "grad_norm": 0.41394945979118347,
1115
  "learning_rate": 0.00010374775076678332,
1116
+ "loss": 0.5608,
1117
  "step": 780
1118
  },
1119
  {
1120
  "epoch": 1.0858428521979924,
1121
+ "grad_norm": 0.4046129882335663,
1122
  "learning_rate": 0.00010253912068653146,
1123
+ "loss": 0.5848,
1124
  "step": 785
1125
  },
1126
  {
1127
  "epoch": 1.092765662859121,
1128
+ "grad_norm": 0.3733786940574646,
1129
  "learning_rate": 0.00010133011932803378,
1130
+ "loss": 0.5717,
1131
  "step": 790
1132
  },
1133
  {
1134
  "epoch": 1.0996884735202492,
1135
+ "grad_norm": 0.428487092256546,
1136
  "learning_rate": 0.00010012092347528705,
1137
+ "loss": 0.5812,
1138
  "step": 795
1139
  },
1140
  {
1141
  "epoch": 1.1066112841813776,
1142
+ "grad_norm": 0.41964441537857056,
1143
  "learning_rate": 9.891170994072793e-05,
1144
+ "loss": 0.576,
1145
  "step": 800
1146
  },
1147
  {
1148
  "epoch": 1.113534094842506,
1149
+ "grad_norm": 0.46450623869895935,
1150
  "learning_rate": 9.770265553937831e-05,
1151
+ "loss": 0.5728,
1152
  "step": 805
1153
  },
1154
  {
1155
  "epoch": 1.1204569055036344,
1156
+ "grad_norm": 0.4148964285850525,
1157
  "learning_rate": 9.649393706299127e-05,
1158
+ "loss": 0.5569,
1159
  "step": 810
1160
  },
1161
  {
1162
  "epoch": 1.127379716164763,
1163
+ "grad_norm": 0.42906996607780457,
1164
  "learning_rate": 9.52857312541998e-05,
1165
+ "loss": 0.568,
1166
  "step": 815
1167
  },
1168
  {
1169
  "epoch": 1.1343025268258913,
1170
+ "grad_norm": 0.4227571189403534,
1171
  "learning_rate": 9.407821478067304e-05,
1172
+ "loss": 0.565,
1173
  "step": 820
1174
  },
1175
  {
1176
  "epoch": 1.1412253374870198,
1177
+ "grad_norm": 0.40537649393081665,
1178
  "learning_rate": 9.287156420928342e-05,
1179
+ "loss": 0.5624,
1180
  "step": 825
1181
  },
1182
  {
1183
  "epoch": 1.1481481481481481,
1184
+ "grad_norm": 0.3608773648738861,
1185
  "learning_rate": 9.166595598028832e-05,
1186
+ "loss": 0.581,
1187
  "step": 830
1188
  },
1189
  {
1190
  "epoch": 1.1550709588092767,
1191
+ "grad_norm": 0.3971177339553833,
1192
  "learning_rate": 9.046156638153056e-05,
1193
+ "loss": 0.5614,
1194
  "step": 835
1195
  },
1196
  {
1197
  "epoch": 1.161993769470405,
1198
+ "grad_norm": 0.37738874554634094,
1199
  "learning_rate": 8.925857152266103e-05,
1200
+ "loss": 0.5662,
1201
  "step": 840
1202
  },
1203
  {
1204
  "epoch": 1.1689165801315333,
1205
+ "grad_norm": 0.409171462059021,
1206
  "learning_rate": 8.805714730938728e-05,
1207
+ "loss": 0.5739,
1208
  "step": 845
1209
  },
1210
  {
1211
  "epoch": 1.1758393907926619,
1212
+ "grad_norm": 0.3909130394458771,
1213
  "learning_rate": 8.685746941775219e-05,
1214
+ "loss": 0.562,
1215
  "step": 850
1216
  },
1217
  {
1218
  "epoch": 1.1827622014537902,
1219
+ "grad_norm": 0.41327953338623047,
1220
  "learning_rate": 8.565971326844584e-05,
1221
+ "loss": 0.5651,
1222
  "step": 855
1223
  },
1224
  {
1225
  "epoch": 1.1896850121149187,
1226
+ "grad_norm": 0.4001190662384033,
1227
  "learning_rate": 8.44640540011553e-05,
1228
+ "loss": 0.5639,
1229
  "step": 860
1230
  },
1231
  {
1232
  "epoch": 1.196607822776047,
1233
+ "grad_norm": 0.40711870789527893,
1234
  "learning_rate": 8.327066644895469e-05,
1235
+ "loss": 0.5618,
1236
  "step": 865
1237
  },
1238
  {
1239
  "epoch": 1.2035306334371756,
1240
+ "grad_norm": 0.43218299746513367,
1241
  "learning_rate": 8.207972511274095e-05,
1242
+ "loss": 0.5585,
1243
  "step": 870
1244
  },
1245
  {
1246
  "epoch": 1.210453444098304,
1247
+ "grad_norm": 0.38643211126327515,
1248
  "learning_rate": 8.089140413571747e-05,
1249
+ "loss": 0.5578,
1250
  "step": 875
1251
  },
1252
  {
1253
  "epoch": 1.2173762547594325,
1254
+ "grad_norm": 0.41631147265434265,
1255
  "learning_rate": 7.970587727793051e-05,
1256
+ "loss": 0.5576,
1257
  "step": 880
1258
  },
1259
  {
1260
  "epoch": 1.2242990654205608,
1261
+ "grad_norm": 0.3960113227367401,
1262
  "learning_rate": 7.852331789086135e-05,
1263
+ "loss": 0.5586,
1264
  "step": 885
1265
  },
1266
  {
1267
  "epoch": 1.231221876081689,
1268
+ "grad_norm": 0.4174184799194336,
1269
  "learning_rate": 7.73438988920784e-05,
1270
+ "loss": 0.5607,
1271
  "step": 890
1272
  },
1273
  {
1274
  "epoch": 1.2381446867428176,
1275
+ "grad_norm": 0.42617419362068176,
1276
  "learning_rate": 7.616779273995252e-05,
1277
+ "loss": 0.5637,
1278
  "step": 895
1279
  },
1280
  {
1281
  "epoch": 1.245067497403946,
1282
+ "grad_norm": 0.40726974606513977,
1283
  "learning_rate": 7.499517140843968e-05,
1284
+ "loss": 0.5641,
1285
  "step": 900
1286
  },
1287
  {
1288
  "epoch": 1.2519903080650745,
1289
+ "grad_norm": 0.417920857667923,
1290
  "learning_rate": 7.382620636193438e-05,
1291
+ "loss": 0.5589,
1292
  "step": 905
1293
  },
1294
  {
1295
  "epoch": 1.2589131187262028,
1296
+ "grad_norm": 0.43630263209342957,
1297
  "learning_rate": 7.266106853019759e-05,
1298
+ "loss": 0.5577,
1299
  "step": 910
1300
  },
1301
  {
1302
  "epoch": 1.2658359293873311,
1303
+ "grad_norm": 0.40452563762664795,
1304
  "learning_rate": 7.149992828336274e-05,
1305
+ "loss": 0.5546,
1306
  "step": 915
1307
  },
1308
  {
1309
  "epoch": 1.2727587400484597,
1310
+ "grad_norm": 0.44135504961013794,
1311
  "learning_rate": 7.034295540702397e-05,
1312
+ "loss": 0.5525,
1313
  "step": 920
1314
  },
1315
  {
1316
  "epoch": 1.2796815507095882,
1317
+ "grad_norm": 0.40712544322013855,
1318
  "learning_rate": 6.919031907740922e-05,
1319
+ "loss": 0.5706,
1320
  "step": 925
1321
  },
1322
  {
1323
  "epoch": 1.2866043613707165,
1324
+ "grad_norm": 0.42069798707962036,
1325
  "learning_rate": 6.804218783664288e-05,
1326
+ "loss": 0.5657,
1327
  "step": 930
1328
  },
1329
  {
1330
  "epoch": 1.2935271720318449,
1331
+ "grad_norm": 0.3738851547241211,
1332
  "learning_rate": 6.689872956810102e-05,
1333
+ "loss": 0.5503,
1334
  "step": 935
1335
  },
1336
  {
1337
  "epoch": 1.3004499826929734,
1338
+ "grad_norm": 0.4131317436695099,
1339
  "learning_rate": 6.5760111471863e-05,
1340
+ "loss": 0.5502,
1341
  "step": 940
1342
  },
1343
  {
1344
  "epoch": 1.3073727933541017,
1345
+ "grad_norm": 0.4052620530128479,
1346
  "learning_rate": 6.46265000402628e-05,
1347
+ "loss": 0.5559,
1348
  "step": 945
1349
  },
1350
  {
1351
  "epoch": 1.3142956040152303,
1352
+ "grad_norm": 0.3770054876804352,
1353
  "learning_rate": 6.349806103354417e-05,
1354
+ "loss": 0.5505,
1355
  "step": 950
1356
  },
1357
  {
1358
  "epoch": 1.3212184146763586,
1359
+ "grad_norm": 0.413860559463501,
1360
  "learning_rate": 6.23749594556226e-05,
1361
+ "loss": 0.5476,
1362
  "step": 955
1363
  },
1364
  {
1365
  "epoch": 1.328141225337487,
1366
+ "grad_norm": 0.39695823192596436,
1367
  "learning_rate": 6.125735952995775e-05,
1368
+ "loss": 0.5519,
1369
  "step": 960
1370
  },
1371
  {
1372
  "epoch": 1.3350640359986155,
1373
+ "grad_norm": 0.43241792917251587,
1374
  "learning_rate": 6.0145424675540394e-05,
1375
+ "loss": 0.5495,
1376
  "step": 965
1377
  },
1378
  {
1379
  "epoch": 1.341986846659744,
1380
+ "grad_norm": 0.4030551612377167,
1381
  "learning_rate": 5.9039317482996735e-05,
1382
+ "loss": 0.5581,
1383
  "step": 970
1384
  },
1385
  {
1386
  "epoch": 1.3489096573208723,
1387
+ "grad_norm": 0.3783753216266632,
1388
  "learning_rate": 5.793919969081374e-05,
1389
+ "loss": 0.5403,
1390
  "step": 975
1391
  },
1392
  {
1393
  "epoch": 1.3558324679820006,
1394
+ "grad_norm": 0.39069420099258423,
1395
  "learning_rate": 5.68452321616894e-05,
1396
+ "loss": 0.5404,
1397
  "step": 980
1398
  },
1399
  {
1400
  "epoch": 1.3627552786431292,
1401
+ "grad_norm": 0.38791337609291077,
1402
  "learning_rate": 5.5757574859010886e-05,
1403
+ "loss": 0.5534,
1404
  "step": 985
1405
  },
1406
  {
1407
  "epoch": 1.3696780893042575,
1408
+ "grad_norm": 0.42228081822395325,
1409
  "learning_rate": 5.467638682346403e-05,
1410
+ "loss": 0.5383,
1411
  "step": 990
1412
  },
1413
  {
1414
  "epoch": 1.376600899965386,
1415
+ "grad_norm": 0.3954121768474579,
1416
  "learning_rate": 5.3601826149777966e-05,
1417
+ "loss": 0.5653,
1418
  "step": 995
1419
  },
1420
  {
1421
  "epoch": 1.3835237106265144,
1422
+ "grad_norm": 0.3909887671470642,
1423
  "learning_rate": 5.25340499636082e-05,
1424
+ "loss": 0.5446,
1425
  "step": 1000
1426
  },
1427
  {
1428
  "epoch": 1.3904465212876427,
1429
+ "grad_norm": 0.41362643241882324,
1430
  "learning_rate": 5.147321439856091e-05,
1431
+ "loss": 0.5466,
1432
  "step": 1005
1433
  },
1434
  {
1435
  "epoch": 1.3973693319487712,
1436
+ "grad_norm": 0.3651193380355835,
1437
  "learning_rate": 5.041947457336274e-05,
1438
+ "loss": 0.5494,
1439
  "step": 1010
1440
  },
1441
  {
1442
  "epoch": 1.4042921426098995,
1443
+ "grad_norm": 0.36420342326164246,
1444
  "learning_rate": 4.937298456917889e-05,
1445
+ "loss": 0.5333,
1446
  "step": 1015
1447
  },
1448
  {
1449
  "epoch": 1.411214953271028,
1450
+ "grad_norm": 0.3928079605102539,
1451
  "learning_rate": 4.8333897407082896e-05,
1452
+ "loss": 0.5445,
1453
  "step": 1020
1454
  },
1455
  {
1456
  "epoch": 1.4181377639321564,
1457
+ "grad_norm": 0.36815693974494934,
1458
  "learning_rate": 4.7302365025681206e-05,
1459
+ "loss": 0.5457,
1460
  "step": 1025
1461
  },
1462
  {
1463
  "epoch": 1.425060574593285,
1464
+ "grad_norm": 0.3875131607055664,
1465
  "learning_rate": 4.627853825889642e-05,
1466
+ "loss": 0.5296,
1467
  "step": 1030
1468
  },
1469
  {
1470
  "epoch": 1.4319833852544133,
1471
+ "grad_norm": 0.4003137946128845,
1472
  "learning_rate": 4.526256681391192e-05,
1473
+ "loss": 0.5455,
1474
  "step": 1035
1475
  },
1476
  {
1477
  "epoch": 1.4389061959155418,
1478
+ "grad_norm": 0.40907323360443115,
1479
  "learning_rate": 4.4254599249281016e-05,
1480
+ "loss": 0.5277,
1481
  "step": 1040
1482
  },
1483
  {
1484
  "epoch": 1.4458290065766701,
1485
+ "grad_norm": 0.40487226843833923,
1486
  "learning_rate": 4.3254782953204375e-05,
1487
+ "loss": 0.5337,
1488
  "step": 1045
1489
  },
1490
  {
1491
  "epoch": 1.4527518172377984,
1492
+ "grad_norm": 0.3791086971759796,
1493
  "learning_rate": 4.2263264121978505e-05,
1494
+ "loss": 0.5394,
1495
  "step": 1050
1496
  },
1497
  {
1498
  "epoch": 1.459674627898927,
1499
+ "grad_norm": 0.4024101197719574,
1500
  "learning_rate": 4.12801877386183e-05,
1501
+ "loss": 0.529,
1502
  "step": 1055
1503
  },
1504
  {
1505
  "epoch": 1.4665974385600553,
1506
+ "grad_norm": 0.3829149305820465,
1507
  "learning_rate": 4.030569755165736e-05,
1508
+ "loss": 0.5365,
1509
  "step": 1060
1510
  },
1511
  {
1512
  "epoch": 1.4735202492211839,
1513
+ "grad_norm": 0.38531187176704407,
1514
  "learning_rate": 3.9339936054128466e-05,
1515
+ "loss": 0.5367,
1516
  "step": 1065
1517
  },
1518
  {
1519
  "epoch": 1.4804430598823122,
1520
+ "grad_norm": 0.39671218395233154,
1521
  "learning_rate": 3.8383044462727826e-05,
1522
+ "loss": 0.5248,
1523
  "step": 1070
1524
  },
1525
  {
1526
  "epoch": 1.4873658705434407,
1527
+ "grad_norm": 0.3982336223125458,
1528
  "learning_rate": 3.7435162697166107e-05,
1529
+ "loss": 0.5469,
1530
  "step": 1075
1531
  },
1532
  {
1533
  "epoch": 1.494288681204569,
1534
+ "grad_norm": 0.40101921558380127,
1535
  "learning_rate": 3.649642935970859e-05,
1536
+ "loss": 0.5461,
1537
  "step": 1080
1538
  },
1539
  {
1540
  "epoch": 1.5012114918656976,
1541
+ "grad_norm": 0.41018062829971313,
1542
  "learning_rate": 3.556698171490871e-05,
1543
+ "loss": 0.53,
1544
  "step": 1085
1545
  },
1546
  {
1547
  "epoch": 1.508134302526826,
1548
+ "grad_norm": 0.39131489396095276,
1549
  "learning_rate": 3.464695566953644e-05,
1550
+ "loss": 0.5287,
1551
  "step": 1090
1552
  },
1553
  {
1554
  "epoch": 1.5150571131879542,
1555
+ "grad_norm": 0.3957297205924988,
1556
  "learning_rate": 3.373648575270576e-05,
1557
+ "loss": 0.5301,
1558
  "step": 1095
1559
  },
1560
  {
1561
  "epoch": 1.5219799238490828,
1562
+ "grad_norm": 0.3676046133041382,
1563
  "learning_rate": 3.283570509620344e-05,
1564
+ "loss": 0.5332,
1565
  "step": 1100
1566
  },
1567
  {
1568
  "epoch": 1.5289027345102113,
1569
+ "grad_norm": 0.3769654333591461,
1570
  "learning_rate": 3.1944745415021915e-05,
1571
+ "loss": 0.5352,
1572
  "step": 1105
1573
  },
1574
  {
1575
  "epoch": 1.5358255451713396,
1576
+ "grad_norm": 0.3649960160255432,
1577
  "learning_rate": 3.106373698809958e-05,
1578
+ "loss": 0.5449,
1579
  "step": 1110
1580
  },
1581
  {
1582
  "epoch": 1.542748355832468,
1583
+ "grad_norm": 0.3833390772342682,
1584
  "learning_rate": 3.0192808639271065e-05,
1585
+ "loss": 0.5462,
1586
  "step": 1115
1587
  },
1588
  {
1589
  "epoch": 1.5496711664935963,
1590
+ "grad_norm": 0.38520780205726624,
1591
  "learning_rate": 2.933208771843018e-05,
1592
+ "loss": 0.5326,
1593
  "step": 1120
1594
  },
1595
  {
1596
  "epoch": 1.5565939771547248,
1597
+ "grad_norm": 0.39242011308670044,
1598
  "learning_rate": 2.8481700082908268e-05,
1599
+ "loss": 0.5275,
1600
  "step": 1125
1601
  },
1602
  {
1603
  "epoch": 1.5635167878158533,
1604
+ "grad_norm": 0.38760125637054443,
1605
  "learning_rate": 2.7641770079071127e-05,
1606
+ "loss": 0.5294,
1607
  "step": 1130
1608
  },
1609
  {
1610
  "epoch": 1.5704395984769817,
1611
+ "grad_norm": 0.37061071395874023,
1612
  "learning_rate": 2.681242052413656e-05,
1613
+ "loss": 0.5173,
1614
  "step": 1135
1615
  },
1616
  {
1617
  "epoch": 1.57736240913811,
1618
+ "grad_norm": 0.40858691930770874,
1619
  "learning_rate": 2.59937726882158e-05,
1620
+ "loss": 0.5149,
1621
  "step": 1140
1622
  },
1623
  {
1624
  "epoch": 1.5842852197992385,
1625
+ "grad_norm": 0.4115959107875824,
1626
  "learning_rate": 2.5185946276580918e-05,
1627
+ "loss": 0.5395,
1628
  "step": 1145
1629
  },
1630
  {
1631
  "epoch": 1.591208030460367,
1632
+ "grad_norm": 0.3848041296005249,
1633
  "learning_rate": 2.4389059412161087e-05,
1634
+ "loss": 0.5317,
1635
  "step": 1150
1636
  },
1637
  {
1638
  "epoch": 1.5981308411214954,
1639
+ "grad_norm": 0.38582009077072144,
1640
  "learning_rate": 2.3603228618270356e-05,
1641
+ "loss": 0.5278,
1642
  "step": 1155
1643
  },
1644
  {
1645
  "epoch": 1.6050536517826237,
1646
+ "grad_norm": 0.38087934255599976,
1647
  "learning_rate": 2.2828568801569283e-05,
1648
+ "loss": 0.5087,
1649
  "step": 1160
1650
  },
1651
  {
1652
  "epoch": 1.611976462443752,
1653
+ "grad_norm": 0.36079418659210205,
1654
  "learning_rate": 2.2065193235262782e-05,
1655
+ "loss": 0.5052,
1656
  "step": 1165
1657
  },
1658
  {
1659
  "epoch": 1.6188992731048806,
1660
+ "grad_norm": 0.38548552989959717,
1661
  "learning_rate": 2.1313213542537058e-05,
1662
+ "loss": 0.5248,
1663
  "step": 1170
1664
  },
1665
  {
1666
  "epoch": 1.6258220837660091,
1667
+ "grad_norm": 0.41799798607826233,
1668
  "learning_rate": 2.0572739680237717e-05,
1669
+ "loss": 0.5357,
1670
  "step": 1175
1671
  },
1672
  {
1673
  "epoch": 1.6327448944271374,
1674
+ "grad_norm": 0.37560001015663147,
1675
  "learning_rate": 1.984387992279153e-05,
1676
+ "loss": 0.5364,
1677
  "step": 1180
1678
  },
1679
  {
1680
  "epoch": 1.6396677050882658,
1681
+ "grad_norm": 0.39734479784965515,
1682
  "learning_rate": 1.912674084637409e-05,
1683
+ "loss": 0.5323,
1684
  "step": 1185
1685
  },
1686
  {
1687
  "epoch": 1.6465905157493943,
1688
+ "grad_norm": 0.3698159158229828,
1689
  "learning_rate": 1.8421427313326046e-05,
1690
+ "loss": 0.5149,
1691
  "step": 1190
1692
  },
1693
  {
1694
  "epoch": 1.6535133264105226,
1695
+ "grad_norm": 0.3833644986152649,
1696
  "learning_rate": 1.7728042456819704e-05,
1697
+ "loss": 0.53,
1698
  "step": 1195
1699
  },
1700
  {
1701
  "epoch": 1.6604361370716512,
1702
+ "grad_norm": 0.37530168890953064,
1703
  "learning_rate": 1.7046687665778717e-05,
1704
+ "loss": 0.5147,
1705
  "step": 1200
1706
  },
1707
  {
1708
  "epoch": 1.6673589477327795,
1709
+ "grad_norm": 0.37578684091567993,
1710
  "learning_rate": 1.6377462570052438e-05,
1711
+ "loss": 0.521,
1712
  "step": 1205
1713
  },
1714
  {
1715
  "epoch": 1.6742817583939078,
1716
+ "grad_norm": 0.4046749770641327,
1717
  "learning_rate": 1.5720465025847987e-05,
1718
+ "loss": 0.5208,
1719
  "step": 1210
1720
  },
1721
  {
1722
  "epoch": 1.6812045690550363,
1723
+ "grad_norm": 0.3772706091403961,
1724
  "learning_rate": 1.5075791101421166e-05,
1725
+ "loss": 0.5204,
1726
  "step": 1215
1727
  },
1728
  {
1729
  "epoch": 1.6881273797161649,
1730
+ "grad_norm": 0.37441906332969666,
1731
  "learning_rate": 1.4443535063029279e-05,
1732
+ "loss": 0.5341,
1733
  "step": 1220
1734
  },
1735
  {
1736
  "epoch": 1.6950501903772932,
1737
+ "grad_norm": 0.39095550775527954,
1738
  "learning_rate": 1.3823789361147044e-05,
1739
+ "loss": 0.5111,
1740
  "step": 1225
1741
  },
1742
  {
1743
  "epoch": 1.7019730010384215,
1744
+ "grad_norm": 0.4048159122467041,
1745
  "learning_rate": 1.3216644616948282e-05,
1746
+ "loss": 0.5183,
1747
  "step": 1230
1748
  },
1749
  {
1750
  "epoch": 1.70889581169955,
1751
+ "grad_norm": 0.3774990141391754,
1752
  "learning_rate": 1.2622189609054979e-05,
1753
+ "loss": 0.5142,
1754
  "step": 1235
1755
  },
1756
  {
1757
  "epoch": 1.7158186223606784,
1758
+ "grad_norm": 0.4018307030200958,
1759
  "learning_rate": 1.2040511260555831e-05,
1760
+ "loss": 0.5179,
1761
  "step": 1240
1762
  },
1763
  {
1764
  "epoch": 1.722741433021807,
1765
+ "grad_norm": 0.37508299946784973,
1766
  "learning_rate": 1.1471694626296049e-05,
1767
+ "loss": 0.514,
1768
  "step": 1245
1769
  },
1770
  {
1771
  "epoch": 1.7296642436829353,
1772
+ "grad_norm": 0.3805769979953766,
1773
  "learning_rate": 1.0915822880440308e-05,
1774
+ "loss": 0.516,
1775
  "step": 1250
1776
  },
1777
  {
1778
  "epoch": 1.7365870543440636,
1779
+ "grad_norm": 0.37996870279312134,
1780
  "learning_rate": 1.037297730431085e-05,
1781
+ "loss": 0.5105,
1782
  "step": 1255
1783
  },
1784
  {
1785
  "epoch": 1.7435098650051921,
1786
+ "grad_norm": 0.3757014572620392,
1787
  "learning_rate": 9.843237274502282e-06,
1788
+ "loss": 0.5176,
1789
  "step": 1260
1790
  },
1791
  {
1792
  "epoch": 1.7504326756663207,
1793
+ "grad_norm": 0.39681464433670044,
1794
  "learning_rate": 9.326680251274777e-06,
1795
+ "loss": 0.527,
1796
  "step": 1265
1797
  },
1798
  {
1799
  "epoch": 1.757355486327449,
1800
+ "grad_norm": 0.3785605728626251,
1801
  "learning_rate": 8.823381767227667e-06,
1802
+ "loss": 0.5223,
1803
  "step": 1270
1804
  },
1805
  {
1806
  "epoch": 1.7642782969885773,
1807
+ "grad_norm": 0.38478976488113403,
1808
  "learning_rate": 8.333415416254831e-06,
1809
+ "loss": 0.5255,
1810
  "step": 1275
1811
  },
1812
  {
1813
  "epoch": 1.7712011076497056,
1814
+ "grad_norm": 0.37452688813209534,
1815
  "learning_rate": 7.856852842783547e-06,
1816
+ "loss": 0.5141,
1817
  "step": 1280
1818
  },
1819
  {
1820
  "epoch": 1.7781239183108342,
1821
+ "grad_norm": 0.3667319118976593,
1822
  "learning_rate": 7.393763731298387e-06,
1823
+ "loss": 0.5056,
1824
  "step": 1285
1825
  },
1826
  {
1827
  "epoch": 1.7850467289719627,
1828
+ "grad_norm": 0.36586126685142517,
1829
  "learning_rate": 6.944215796151765e-06,
1830
+ "loss": 0.5108,
1831
  "step": 1290
1832
  },
1833
  {
1834
  "epoch": 1.791969539633091,
1835
+ "grad_norm": 0.407480388879776,
1836
  "learning_rate": 6.5082747716625255e-06,
1837
+ "loss": 0.5168,
1838
  "step": 1295
1839
  },
1840
  {
1841
  "epoch": 1.7988923502942193,
1842
+ "grad_norm": 0.3544744551181793,
1843
  "learning_rate": 6.086004402504098e-06,
1844
+ "loss": 0.5044,
1845
  "step": 1300
1846
  },
1847
  {
1848
  "epoch": 1.8058151609553479,
1849
+ "grad_norm": 0.39020371437072754,
1850
  "learning_rate": 5.6774664343834495e-06,
1851
+ "loss": 0.5248,
1852
  "step": 1305
1853
  },
1854
  {
1855
  "epoch": 1.8127379716164764,
1856
+ "grad_norm": 0.38111376762390137,
1857
  "learning_rate": 5.2827206050125585e-06,
1858
+ "loss": 0.5151,
1859
  "step": 1310
1860
  },
1861
  {
1862
  "epoch": 1.8196607822776047,
1863
+ "grad_norm": 0.38476815819740295,
1864
  "learning_rate": 4.9018246353732755e-06,
1865
+ "loss": 0.533,
1866
  "step": 1315
1867
  },
1868
  {
1869
  "epoch": 1.826583592938733,
1870
+ "grad_norm": 0.3769124746322632,
1871
  "learning_rate": 4.534834221277262e-06,
1872
+ "loss": 0.5047,
1873
  "step": 1320
1874
  },
1875
  {
1876
  "epoch": 1.8335064035998614,
1877
+ "grad_norm": 0.3822435140609741,
1878
  "learning_rate": 4.181803025221898e-06,
1879
+ "loss": 0.5212,
1880
  "step": 1325
1881
  },
1882
  {
1883
  "epoch": 1.84042921426099,
1884
+ "grad_norm": 0.3835265338420868,
1885
  "learning_rate": 3.842782668543599e-06,
1886
+ "loss": 0.515,
1887
  "step": 1330
1888
  },
1889
  {
1890
  "epoch": 1.8473520249221185,
1891
+ "grad_norm": 0.34277990460395813,
1892
  "learning_rate": 3.5178227238696347e-06,
1893
+ "loss": 0.5097,
1894
  "step": 1335
1895
  },
1896
  {
1897
  "epoch": 1.8542748355832468,
1898
+ "grad_norm": 0.41254210472106934,
1899
  "learning_rate": 3.2069707078694057e-06,
1900
+ "loss": 0.4997,
1901
  "step": 1340
1902
  },
1903
  {
1904
  "epoch": 1.8611976462443751,
1905
+ "grad_norm": 0.38151687383651733,
1906
  "learning_rate": 2.9102720743064527e-06,
1907
+ "loss": 0.5253,
1908
  "step": 1345
1909
  },
1910
  {
1911
  "epoch": 1.8681204569055037,
1912
+ "grad_norm": 0.3614235520362854,
1913
  "learning_rate": 2.6277702073919997e-06,
1914
+ "loss": 0.5022,
1915
  "step": 1350
1916
  },
1917
  {
1918
  "epoch": 1.8750432675666322,
1919
+ "grad_norm": 0.37769678235054016,
1920
  "learning_rate": 2.3595064154412374e-06,
1921
+ "loss": 0.5083,
1922
  "step": 1355
1923
  },
1924
  {
1925
  "epoch": 1.8819660782277605,
1926
+ "grad_norm": 0.3952181041240692,
1927
  "learning_rate": 2.1055199248330546e-06,
1928
+ "loss": 0.5093,
1929
  "step": 1360
1930
  },
1931
  {
1932
  "epoch": 1.8888888888888888,
1933
+ "grad_norm": 0.3974086344242096,
1934
  "learning_rate": 1.8658478742742625e-06,
1935
+ "loss": 0.5087,
1936
  "step": 1365
1937
  },
1938
  {
1939
  "epoch": 1.8958116995500172,
1940
+ "grad_norm": 0.38178950548171997,
1941
  "learning_rate": 1.6405253093690343e-06,
1942
+ "loss": 0.5072,
1943
  "step": 1370
1944
  },
1945
  {
1946
  "epoch": 1.9027345102111457,
1947
+ "grad_norm": 0.39221471548080444,
1948
  "learning_rate": 1.429585177494419e-06,
1949
+ "loss": 0.504,
1950
  "step": 1375
1951
  },
1952
  {
1953
  "epoch": 1.9096573208722742,
1954
+ "grad_norm": 0.37083545327186584,
1955
  "learning_rate": 1.2330583229827053e-06,
1956
+ "loss": 0.5139,
1957
  "step": 1380
1958
  },
1959
  {
1960
  "epoch": 1.9165801315334026,
1961
+ "grad_norm": 0.3701615631580353,
1962
  "learning_rate": 1.0509734826112394e-06,
1963
+ "loss": 0.5023,
1964
  "step": 1385
1965
  },
1966
  {
1967
  "epoch": 1.9235029421945309,
1968
+ "grad_norm": 0.40432795882225037,
1969
  "learning_rate": 8.833572814004321e-07,
1970
+ "loss": 0.5026,
1971
  "step": 1390
1972
  },
1973
  {
1974
  "epoch": 1.9304257528556594,
1975
+ "grad_norm": 0.3880690634250641,
1976
  "learning_rate": 7.302342287205943e-07,
1977
+ "loss": 0.5042,
1978
  "step": 1395
1979
  },
1980
  {
1981
  "epoch": 1.937348563516788,
1982
+ "grad_norm": 0.38099151849746704,
1983
  "learning_rate": 5.916267147080934e-07,
1984
+ "loss": 0.5098,
1985
  "step": 1400
1986
  },
1987
  {
1988
  "epoch": 1.9442713741779163,
1989
+ "grad_norm": 0.3906520903110504,
1990
  "learning_rate": 4.6755500699138253e-07,
1991
+ "loss": 0.5039,
1992
  "step": 1405
1993
  },
1994
  {
1995
  "epoch": 1.9511941848390446,
1996
+ "grad_norm": 0.39573198556900024,
1997
  "learning_rate": 3.5803724772740523e-07,
1998
+ "loss": 0.505,
1999
  "step": 1410
2000
  },
2001
  {
2002
  "epoch": 1.958116995500173,
2003
+ "grad_norm": 0.3726860284805298,
2004
  "learning_rate": 2.630894509488058e-07,
2005
+ "loss": 0.5088,
2006
  "step": 1415
2007
  },
2008
  {
2009
  "epoch": 1.9650398061613015,
2010
+ "grad_norm": 0.35756033658981323,
2011
  "learning_rate": 1.8272550022230272e-07,
2012
+ "loss": 0.506,
2013
  "step": 1420
2014
  },
2015
  {
2016
  "epoch": 1.97196261682243,
2017
+ "grad_norm": 0.3828778862953186,
2018
  "learning_rate": 1.1695714661856816e-07,
2019
+ "loss": 0.494,
2020
  "step": 1425
2021
  },
2022
  {
2023
  "epoch": 1.9788854274835583,
2024
+ "grad_norm": 0.3731125593185425,
2025
  "learning_rate": 6.579400699397998e-08,
2026
+ "loss": 0.5106,
2027
  "step": 1430
2028
  },
2029
  {
2030
  "epoch": 1.9858082381446867,
2031
+ "grad_norm": 0.3635178804397583,
2032
  "learning_rate": 2.9243562584402305e-08,
2033
+ "loss": 0.517,
2034
  "step": 1435
2035
  },
2036
  {
2037
  "epoch": 1.9927310488058152,
2038
+ "grad_norm": 0.38713252544403076,
2039
  "learning_rate": 7.311157911249478e-09,
2040
+ "loss": 0.5221,
2041
  "step": 1440
2042
  },
2043
  {
2044
  "epoch": 1.9982692973347178,
2045
+ "eval_loss": 0.7301178574562073,
2046
+ "eval_runtime": 60.039,
2047
+ "eval_samples_per_second": 34.461,
2048
+ "eval_steps_per_second": 2.165,
2049
  "step": 1444
2050
  },
2051
  {
2052
  "epoch": 1.9982692973347178,
2053
  "step": 1444,
2054
  "total_flos": 4.774796461726499e+18,
2055
+ "train_loss": 0.735994421048838,
2056
+ "train_runtime": 11345.701,
2057
+ "train_samples_per_second": 8.147,
2058
  "train_steps_per_second": 0.127
2059
  }
2060
  ],
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:36f69b66d816c2099f74805fb3fb5224381d74a293161d0078d22df01687bcaf
3
  size 7096
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad6491fa3a07ec6e3eee41c26481d5fc0ad4b009fdaba9b93fdd42368f1ac3c2
3
  size 7096