Hugging Face's logo

fengyao1909
/
ste_nemocode_sl16k_bs64_lr3e-5_ckpt400

Safetensors
qwen3_moe
Model card Files Community
fengyao1909 commited on
Commit
2ed1c0b
·
verified ·
1 Parent(s): 3287ddc

Upload folder using huggingface_hub

Browse files
Files changed (18) hide show
  1. config.json +1 -1
  2. generation_config.json +1 -1
  3. model-00001-of-00013.safetensors +1 -1
  4. model-00002-of-00013.safetensors +1 -1
  5. model-00003-of-00013.safetensors +1 -1
  6. model-00004-of-00013.safetensors +1 -1
  7. model-00005-of-00013.safetensors +1 -1
  8. model-00006-of-00013.safetensors +1 -1
  9. model-00007-of-00013.safetensors +1 -1
  10. model-00008-of-00013.safetensors +1 -1
  11. model-00009-of-00013.safetensors +1 -1
  12. model-00010-of-00013.safetensors +1 -1
  13. model-00011-of-00013.safetensors +1 -1
  14. model-00012-of-00013.safetensors +1 -1
  15. model-00013-of-00013.safetensors +1 -1
  16. tokenizer_config.json +1 -0
  17. trainer_state.json +798 -798
  18. training_args.bin +1 -1
config.json CHANGED
@@ -31,7 +31,7 @@
31
  "sliding_window": null,
32
  "tie_word_embeddings": false,
33
  "torch_dtype": "bfloat16",
34
- "transformers_version": "4.52.0.dev0",
35
  "use_cache": false,
36
  "use_sliding_window": false,
37
  "vocab_size": 151936
 
31
  "sliding_window": null,
32
  "tie_word_embeddings": false,
33
  "torch_dtype": "bfloat16",
34
+ "transformers_version": "4.51.3",
35
  "use_cache": false,
36
  "use_sliding_window": false,
37
  "vocab_size": 151936
generation_config.json CHANGED
@@ -2,5 +2,5 @@
2
  "bos_token_id": 151643,
3
  "eos_token_id": 151643,
4
  "max_new_tokens": 2048,
5
- "transformers_version": "4.52.0.dev0"
6
  }
 
2
  "bos_token_id": 151643,
3
  "eos_token_id": 151643,
4
  "max_new_tokens": 2048,
5
+ "transformers_version": "4.51.3"
6
  }
model-00001-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a46574a2308a287084c392995db17946d83cecc32da3f9e1d919a0b84cda8b85
3
  size 4997184968
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:64b1dd8fd8b113ef8aa508804e2f1e092b78f5eca5b227bc85f3548300bb1b16
3
  size 4997184968
model-00002-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2cdd4c903467c112c05dc3660dd2e35c2156ed118b19041b828e199dc150e063
3
  size 4997741608
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:747c75d6062dc37b38c50fe833c1ac6ba71b7c335021c0f20dc77814a80ad0fd
3
  size 4997741608
model-00003-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3442d0b80a9b46d9c8454739277452d9033c1c9ab9f9aa7bdc24fe9915bdb2ee
3
  size 4997742208
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ef229d423612f779356b9184ee67386551f5ecc8edd7513200412c0b5b5e374
3
  size 4997742208
model-00004-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8a749f26c995f6c2736e67513978dd5d2325fbac354aca30f158f7d87b95f738
3
  size 4997743184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d309918edb035203d53802589f47909688e731b45137d2a04736b4067213d2e8
3
  size 4997743184
model-00005-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:116941e1f2cc212aa3f6695f6d7faadb11e1015d84280768d7e43a1d8641c598
3
  size 4997743184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f35a9336e5673a96111bd764d9e7d2a1886c37729dd67293f48788ba48b6764b
3
  size 4997743184
model-00006-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c51826866c5fbc6bf927b30a9659be304563c248dc041a1de99087beb6249b07
3
  size 4997743184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd4417896aff5991199d40a16f4e2d185291a66b33e6dd6c4ab9ef6f0f1c8526
3
  size 4997743184
model-00007-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f31f16556c57011a1a62d2c980095148f2706ac923ca05a739c15fb2b54e7843
3
  size 4997743184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6067df8cbb18e1df184318224e0c16669999e831eaeb12f36c215894aa70ca8b
3
  size 4997743184
model-00008-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d038fdf77de71f5ba36b9284cf665c4774500c91ef15fbbb3022065ba4e4e80d
3
  size 4997743184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d0980a3b9414a5436889e2a998b1c56f35c923fae8b89e9c3984e615bcffda3
3
  size 4997743184
model-00009-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:903ac5747acf1c1527304e0cdd035306475d6c7398f37fe7993b21fc5220cd80
3
  size 4997743184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bdbc961a54d13bc3e8de9049f7a3a703b83a59fa8a7d32814496b5a96d01c70d
3
  size 4997743184
model-00010-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:32063d6d1242aa9d651500b0fddfb6eb0f6918797b161ea892e2d68b8bd11485
3
  size 4997743184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1fc0d5b2360bc44b3879ba267de1e1f7171d429de988adc2731dd8fc1ab3cfca
3
  size 4997743184
model-00011-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6d03fd7030b7ec318ab04a507335320745cc7108245a1a9199b095697e4e6543
3
  size 4997743184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1049b4d4cc1219bc2a92478b0414913886601f5061a1425721cdc36c82ea584c
3
  size 4997743184
model-00012-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7f76ac2474b664a1a540376eb97e3727283c82fdb3f1757530124d06d1daf7e1
3
  size 4997743184
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a170f7d550807aeb669bf6de177bda23c4e5f6fad61845631a0c347d5881615c
3
  size 4997743184
model-00013-of-00013.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f5835cc1f33c8a297483a91fd1431971c514461a54275041eb039bcd21bfa967
3
  size 1094220288
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2dba6754abb4418b24ff47463b2e03ddfbfa3a213123c9bb8ad62b8430985024
3
  size 1094220288
tokenizer_config.json CHANGED
@@ -227,6 +227,7 @@
227
  "<|video_pad|>"
228
  ],
229
  "bos_token": null,
 
230
  "clean_up_tokenization_spaces": false,
231
  "eos_token": "<|im_end|>",
232
  "errors": "replace",
 
227
  "<|video_pad|>"
228
  ],
229
  "bos_token": null,
230
+ "chat_template": "{%- if tools %}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0].role == 'system' %}\n {{- messages[0].content + '\\n\\n' }}\n {%- endif %}\n {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n {%- if messages[0].role == 'system' %}\n {{- '<|im_start|>system\\n' + messages[0].content + '<|im_end|>\\n' }}\n {%- endif %}\n{%- endif %}\n{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}\n{%- for message in messages[::-1] %}\n {%- set index = (messages|length - 1) - loop.index0 %}\n {%- if ns.multi_step_tool and message.role == \"user\" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}\n {%- set ns.multi_step_tool = false %}\n {%- set ns.last_query_index = index %}\n {%- endif %}\n{%- endfor %}\n{%- for message in messages %}\n {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n {%- elif message.role == \"assistant\" %}\n {%- set content = message.content %}\n {%- set reasoning_content = '' %}\n {%- if message.reasoning_content is defined and message.reasoning_content is not none %}\n {%- set reasoning_content = message.reasoning_content %}\n {%- else %}\n {%- if '</think>' in message.content %}\n {%- set content = message.content.split('</think>')[-1].lstrip('\\n') %}\n {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n {%- endif %}\n {%- endif %}\n {%- if loop.index0 > ns.last_query_index %}\n {%- if loop.last or (not loop.last and reasoning_content) %}\n {{- '<|im_start|>' + message.role + '\\n<think>\\n' + reasoning_content.strip('\\n') + '\\n</think>\\n\\n' + content.lstrip('\\n') }}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- if message.tool_calls %}\n {%- for tool_call in message.tool_calls %}\n {%- if (loop.first and content) or (not loop.first) %}\n {{- '\\n' }}\n {%- endif %}\n {%- if tool_call.function %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {{- '<tool_call>\\n{\"name\": \"' }}\n {{- tool_call.name }}\n {{- '\", \"arguments\": ' }}\n {%- if tool_call.arguments is string %}\n {{- tool_call.arguments }}\n {%- else %}\n {{- tool_call.arguments | tojson }}\n {%- endif %}\n {{- '}\\n</tool_call>' }}\n {%- endfor %}\n {%- endif %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n<tool_response>\\n' }}\n {{- message.content }}\n {{- '\\n</tool_response>' }}\n {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|im_start|>assistant\\n' }}\n {%- if enable_thinking is defined and enable_thinking is false %}\n {{- '<think>\\n\\n</think>\\n\\n' }}\n {%- endif %}\n{%- endif %}",
231
  "clean_up_tokenization_spaces": false,
232
  "eos_token": "<|im_end|>",
233
  "errors": "replace",
trainer_state.json CHANGED
@@ -11,2802 +11,2802 @@
11
  "log_history": [
12
  {
13
  "epoch": 0.0018066847335140017,
14
- "grad_norm": 2.8330458831358536,
15
  "learning_rate": 0.0,
16
  "loss": 0.8251,
17
  "step": 1
18
  },
19
  {
20
  "epoch": 0.0036133694670280035,
21
- "grad_norm": 2.881506507345633,
22
  "learning_rate": 1.0714285714285714e-06,
23
  "loss": 0.8284,
24
  "step": 2
25
  },
26
  {
27
  "epoch": 0.005420054200542005,
28
- "grad_norm": 2.864957143736817,
29
  "learning_rate": 2.1428571428571427e-06,
30
- "loss": 0.8427,
31
  "step": 3
32
  },
33
  {
34
  "epoch": 0.007226738934056007,
35
- "grad_norm": 2.789658157036283,
36
  "learning_rate": 3.2142857142857143e-06,
37
- "loss": 0.8378,
38
  "step": 4
39
  },
40
  {
41
  "epoch": 0.009033423667570008,
42
- "grad_norm": 2.5748296264473147,
43
  "learning_rate": 4.2857142857142855e-06,
44
- "loss": 0.8226,
45
  "step": 5
46
  },
47
  {
48
  "epoch": 0.01084010840108401,
49
- "grad_norm": 2.1226324001162262,
50
  "learning_rate": 5.357142857142857e-06,
51
- "loss": 0.8096,
52
  "step": 6
53
  },
54
  {
55
  "epoch": 0.012646793134598013,
56
- "grad_norm": 1.4591614339485137,
57
  "learning_rate": 6.428571428571429e-06,
58
- "loss": 0.7738,
59
  "step": 7
60
  },
61
  {
62
  "epoch": 0.014453477868112014,
63
- "grad_norm": 1.3910404074529026,
64
  "learning_rate": 7.5e-06,
65
- "loss": 0.7731,
66
  "step": 8
67
  },
68
  {
69
  "epoch": 0.016260162601626018,
70
- "grad_norm": 1.2975941991178286,
71
  "learning_rate": 8.571428571428571e-06,
72
- "loss": 0.7653,
73
  "step": 9
74
  },
75
  {
76
  "epoch": 0.018066847335140017,
77
- "grad_norm": 2.39949931335525,
78
  "learning_rate": 9.642857142857144e-06,
79
- "loss": 0.7503,
80
  "step": 10
81
  },
82
  {
83
  "epoch": 0.01987353206865402,
84
- "grad_norm": 2.339166039208163,
85
  "learning_rate": 1.0714285714285714e-05,
86
- "loss": 0.7353,
87
  "step": 11
88
  },
89
  {
90
  "epoch": 0.02168021680216802,
91
- "grad_norm": 2.2235135720121253,
92
  "learning_rate": 1.1785714285714286e-05,
93
- "loss": 0.7488,
94
  "step": 12
95
  },
96
  {
97
  "epoch": 0.023486901535682024,
98
- "grad_norm": 1.8641369164869521,
99
  "learning_rate": 1.2857142857142857e-05,
100
- "loss": 0.7252,
101
  "step": 13
102
  },
103
  {
104
  "epoch": 0.025293586269196026,
105
- "grad_norm": 1.7375311366152815,
106
  "learning_rate": 1.3928571428571429e-05,
107
- "loss": 0.7282,
108
  "step": 14
109
  },
110
  {
111
  "epoch": 0.02710027100271003,
112
- "grad_norm": 1.4297603171733615,
113
  "learning_rate": 1.5e-05,
114
- "loss": 0.7175,
115
  "step": 15
116
  },
117
  {
118
  "epoch": 0.028906955736224028,
119
- "grad_norm": 1.1151402205703254,
120
  "learning_rate": 1.6071428571428572e-05,
121
- "loss": 0.7016,
122
  "step": 16
123
  },
124
  {
125
  "epoch": 0.03071364046973803,
126
- "grad_norm": 0.8024202349277224,
127
  "learning_rate": 1.7142857142857142e-05,
128
- "loss": 0.6998,
129
  "step": 17
130
  },
131
  {
132
  "epoch": 0.032520325203252036,
133
- "grad_norm": 0.7606508999595812,
134
  "learning_rate": 1.8214285714285712e-05,
135
- "loss": 0.6884,
136
  "step": 18
137
  },
138
  {
139
  "epoch": 0.03432700993676603,
140
- "grad_norm": 0.7625534222641119,
141
  "learning_rate": 1.928571428571429e-05,
142
- "loss": 0.6829,
143
  "step": 19
144
  },
145
  {
146
  "epoch": 0.036133694670280034,
147
- "grad_norm": 0.6694670087645954,
148
  "learning_rate": 2.0357142857142858e-05,
149
- "loss": 0.6597,
150
  "step": 20
151
  },
152
  {
153
  "epoch": 0.037940379403794036,
154
- "grad_norm": 0.5962751724140051,
155
  "learning_rate": 2.1428571428571428e-05,
156
- "loss": 0.6656,
157
  "step": 21
158
  },
159
  {
160
  "epoch": 0.03974706413730804,
161
- "grad_norm": 0.574939617107255,
162
  "learning_rate": 2.25e-05,
163
- "loss": 0.6544,
164
  "step": 22
165
  },
166
  {
167
  "epoch": 0.04155374887082204,
168
- "grad_norm": 0.5740495615931904,
169
  "learning_rate": 2.357142857142857e-05,
170
- "loss": 0.655,
171
  "step": 23
172
  },
173
  {
174
  "epoch": 0.04336043360433604,
175
- "grad_norm": 0.5401386924725768,
176
  "learning_rate": 2.464285714285714e-05,
177
- "loss": 0.6603,
178
  "step": 24
179
  },
180
  {
181
  "epoch": 0.045167118337850046,
182
- "grad_norm": 0.5411734137719719,
183
  "learning_rate": 2.5714285714285714e-05,
184
- "loss": 0.6417,
185
  "step": 25
186
  },
187
  {
188
  "epoch": 0.04697380307136405,
189
- "grad_norm": 0.4707448281296468,
190
  "learning_rate": 2.6785714285714288e-05,
191
- "loss": 0.6565,
192
  "step": 26
193
  },
194
  {
195
  "epoch": 0.04878048780487805,
196
- "grad_norm": 0.4601634980551412,
197
  "learning_rate": 2.7857142857142858e-05,
198
- "loss": 0.6344,
199
  "step": 27
200
  },
201
  {
202
  "epoch": 0.05058717253839205,
203
- "grad_norm": 0.4576924664211419,
204
  "learning_rate": 2.892857142857143e-05,
205
- "loss": 0.637,
206
  "step": 28
207
  },
208
  {
209
  "epoch": 0.052393857271906055,
210
- "grad_norm": 0.44420515620753975,
211
  "learning_rate": 3e-05,
212
- "loss": 0.6484,
213
  "step": 29
214
  },
215
  {
216
  "epoch": 0.05420054200542006,
217
- "grad_norm": 0.41258566428029175,
218
  "learning_rate": 2.9999731440137413e-05,
219
- "loss": 0.6488,
220
  "step": 30
221
  },
222
  {
223
  "epoch": 0.05600722673893405,
224
- "grad_norm": 0.3492774187256118,
225
  "learning_rate": 2.9998925770166232e-05,
226
- "loss": 0.6377,
227
  "step": 31
228
  },
229
  {
230
  "epoch": 0.057813911472448055,
231
- "grad_norm": 0.38161813042837406,
232
  "learning_rate": 2.9997583018935875e-05,
233
- "loss": 0.6262,
234
  "step": 32
235
  },
236
  {
237
  "epoch": 0.05962059620596206,
238
- "grad_norm": 0.3667873345432628,
239
  "learning_rate": 2.9995703234527553e-05,
240
- "loss": 0.6358,
241
  "step": 33
242
  },
243
  {
244
  "epoch": 0.06142728093947606,
245
- "grad_norm": 0.3431907596759456,
246
  "learning_rate": 2.999328648425255e-05,
247
- "loss": 0.6233,
248
  "step": 34
249
  },
250
  {
251
  "epoch": 0.06323396567299007,
252
- "grad_norm": 0.36189091112235033,
253
  "learning_rate": 2.999033285464982e-05,
254
- "loss": 0.6288,
255
  "step": 35
256
  },
257
  {
258
  "epoch": 0.06504065040650407,
259
- "grad_norm": 0.3804643814225747,
260
  "learning_rate": 2.9986842451482876e-05,
261
- "loss": 0.6248,
262
  "step": 36
263
  },
264
  {
265
  "epoch": 0.06684733514001806,
266
- "grad_norm": 0.35830891965133377,
267
  "learning_rate": 2.9982815399736008e-05,
268
- "loss": 0.6297,
269
  "step": 37
270
  },
271
  {
272
  "epoch": 0.06865401987353206,
273
- "grad_norm": 0.349837532228719,
274
  "learning_rate": 2.9978251843609816e-05,
275
- "loss": 0.6219,
276
  "step": 38
277
  },
278
  {
279
  "epoch": 0.07046070460704607,
280
- "grad_norm": 0.40406566853435366,
281
  "learning_rate": 2.9973151946516027e-05,
282
- "loss": 0.6243,
283
  "step": 39
284
  },
285
  {
286
  "epoch": 0.07226738934056007,
287
- "grad_norm": 0.30908023037541016,
288
  "learning_rate": 2.996751589107167e-05,
289
- "loss": 0.6123,
290
  "step": 40
291
  },
292
  {
293
  "epoch": 0.07407407407407407,
294
- "grad_norm": 0.3499175700089795,
295
  "learning_rate": 2.9961343879092512e-05,
296
- "loss": 0.6186,
297
  "step": 41
298
  },
299
  {
300
  "epoch": 0.07588075880758807,
301
- "grad_norm": 0.29453754657609377,
302
  "learning_rate": 2.9954636131585845e-05,
303
- "loss": 0.6117,
304
  "step": 42
305
  },
306
  {
307
  "epoch": 0.07768744354110207,
308
- "grad_norm": 0.31600841861885537,
309
  "learning_rate": 2.9947392888742566e-05,
310
- "loss": 0.6117,
311
  "step": 43
312
  },
313
  {
314
  "epoch": 0.07949412827461608,
315
- "grad_norm": 0.3022084890043788,
316
  "learning_rate": 2.993961440992859e-05,
317
- "loss": 0.6056,
318
  "step": 44
319
  },
320
  {
321
  "epoch": 0.08130081300813008,
322
- "grad_norm": 0.3192692448247662,
323
  "learning_rate": 2.993130097367553e-05,
324
- "loss": 0.613,
325
  "step": 45
326
  },
327
  {
328
  "epoch": 0.08310749774164408,
329
- "grad_norm": 0.2995609365814746,
330
  "learning_rate": 2.9922452877670775e-05,
331
- "loss": 0.6145,
332
  "step": 46
333
  },
334
  {
335
  "epoch": 0.08491418247515808,
336
- "grad_norm": 0.28359568489888487,
337
  "learning_rate": 2.991307043874677e-05,
338
- "loss": 0.6171,
339
  "step": 47
340
  },
341
  {
342
  "epoch": 0.08672086720867209,
343
- "grad_norm": 0.2763705613050705,
344
  "learning_rate": 2.9903153992869734e-05,
345
- "loss": 0.6031,
346
  "step": 48
347
  },
348
  {
349
  "epoch": 0.08852755194218609,
350
- "grad_norm": 0.3024771650885143,
351
  "learning_rate": 2.989270389512756e-05,
352
- "loss": 0.6108,
353
  "step": 49
354
  },
355
  {
356
  "epoch": 0.09033423667570009,
357
- "grad_norm": 0.2961191333250206,
358
  "learning_rate": 2.988172051971717e-05,
359
- "loss": 0.61,
360
  "step": 50
361
  },
362
  {
363
  "epoch": 0.0921409214092141,
364
- "grad_norm": 0.2832954078297925,
365
  "learning_rate": 2.9870204259931062e-05,
366
- "loss": 0.5999,
367
  "step": 51
368
  },
369
  {
370
  "epoch": 0.0939476061427281,
371
- "grad_norm": 0.28528145590927906,
372
  "learning_rate": 2.9858155528143256e-05,
373
- "loss": 0.617,
374
  "step": 52
375
  },
376
  {
377
  "epoch": 0.0957542908762421,
378
- "grad_norm": 0.31958273903448986,
379
  "learning_rate": 2.9845574755794522e-05,
380
- "loss": 0.6027,
381
  "step": 53
382
  },
383
  {
384
  "epoch": 0.0975609756097561,
385
- "grad_norm": 0.2721806575699711,
386
  "learning_rate": 2.9832462393376926e-05,
387
- "loss": 0.5922,
388
  "step": 54
389
  },
390
  {
391
  "epoch": 0.0993676603432701,
392
- "grad_norm": 0.345845783610871,
393
  "learning_rate": 2.9818818910417706e-05,
394
- "loss": 0.6081,
395
  "step": 55
396
  },
397
  {
398
  "epoch": 0.1011743450767841,
399
- "grad_norm": 0.29580862920928147,
400
  "learning_rate": 2.9804644795462437e-05,
401
- "loss": 0.6121,
402
  "step": 56
403
  },
404
  {
405
  "epoch": 0.10298102981029811,
406
- "grad_norm": 0.3175507142787931,
407
  "learning_rate": 2.9789940556057574e-05,
408
- "loss": 0.5976,
409
  "step": 57
410
  },
411
  {
412
  "epoch": 0.10478771454381211,
413
- "grad_norm": 0.3374272189572956,
414
  "learning_rate": 2.9774706718732255e-05,
415
- "loss": 0.6061,
416
  "step": 58
417
  },
418
  {
419
  "epoch": 0.10659439927732611,
420
- "grad_norm": 0.27546053233166856,
421
  "learning_rate": 2.9758943828979444e-05,
422
- "loss": 0.5922,
423
  "step": 59
424
  },
425
  {
426
  "epoch": 0.10840108401084012,
427
- "grad_norm": 0.3657689574287652,
428
  "learning_rate": 2.9742652451236414e-05,
429
- "loss": 0.5934,
430
  "step": 60
431
  },
432
  {
433
  "epoch": 0.1102077687443541,
434
- "grad_norm": 0.2986956716963071,
435
  "learning_rate": 2.972583316886451e-05,
436
- "loss": 0.5971,
437
  "step": 61
438
  },
439
  {
440
  "epoch": 0.1120144534778681,
441
- "grad_norm": 0.3448276185716925,
442
  "learning_rate": 2.9708486584128303e-05,
443
- "loss": 0.6017,
444
  "step": 62
445
  },
446
  {
447
  "epoch": 0.11382113821138211,
448
- "grad_norm": 0.3163562156051723,
449
  "learning_rate": 2.9690613318173966e-05,
450
- "loss": 0.6018,
451
  "step": 63
452
  },
453
  {
454
  "epoch": 0.11562782294489611,
455
- "grad_norm": 0.2891799826775374,
456
  "learning_rate": 2.9672214011007087e-05,
457
- "loss": 0.6001,
458
  "step": 64
459
  },
460
  {
461
  "epoch": 0.11743450767841011,
462
- "grad_norm": 0.3052917388689653,
463
  "learning_rate": 2.9653289321469715e-05,
464
- "loss": 0.5941,
465
  "step": 65
466
  },
467
  {
468
  "epoch": 0.11924119241192412,
469
- "grad_norm": 0.27037779234686826,
470
  "learning_rate": 2.9633839927216793e-05,
471
- "loss": 0.5971,
472
  "step": 66
473
  },
474
  {
475
  "epoch": 0.12104787714543812,
476
- "grad_norm": 0.2837154841373324,
477
  "learning_rate": 2.9613866524691867e-05,
478
- "loss": 0.5873,
479
  "step": 67
480
  },
481
  {
482
  "epoch": 0.12285456187895212,
483
- "grad_norm": 0.29318973292150885,
484
  "learning_rate": 2.9593369829102173e-05,
485
- "loss": 0.6033,
486
  "step": 68
487
  },
488
  {
489
  "epoch": 0.12466124661246612,
490
- "grad_norm": 0.27320922800225206,
491
  "learning_rate": 2.957235057439301e-05,
492
- "loss": 0.6043,
493
  "step": 69
494
  },
495
  {
496
  "epoch": 0.12646793134598014,
497
- "grad_norm": 0.2888792462126571,
498
  "learning_rate": 2.955080951322147e-05,
499
- "loss": 0.609,
500
  "step": 70
501
  },
502
  {
503
  "epoch": 0.12827461607949414,
504
- "grad_norm": 0.3010426601423717,
505
  "learning_rate": 2.9528747416929467e-05,
506
- "loss": 0.5817,
507
  "step": 71
508
  },
509
  {
510
  "epoch": 0.13008130081300814,
511
- "grad_norm": 0.29363653670708806,
512
  "learning_rate": 2.9506165075516148e-05,
513
- "loss": 0.599,
514
  "step": 72
515
  },
516
  {
517
  "epoch": 0.13188798554652212,
518
- "grad_norm": 0.3695495205161594,
519
  "learning_rate": 2.9483063297609577e-05,
520
- "loss": 0.5979,
521
  "step": 73
522
  },
523
  {
524
  "epoch": 0.13369467028003612,
525
- "grad_norm": 0.312283280850037,
526
  "learning_rate": 2.9459442910437798e-05,
527
- "loss": 0.5894,
528
  "step": 74
529
  },
530
  {
531
  "epoch": 0.13550135501355012,
532
- "grad_norm": 0.34410257132188554,
533
  "learning_rate": 2.94353047597992e-05,
534
- "loss": 0.5887,
535
  "step": 75
536
  },
537
  {
538
  "epoch": 0.13730803974706413,
539
- "grad_norm": 0.2913822090159289,
540
  "learning_rate": 2.941064971003224e-05,
541
- "loss": 0.5844,
542
  "step": 76
543
  },
544
  {
545
  "epoch": 0.13911472448057813,
546
- "grad_norm": 0.34659905630109983,
547
  "learning_rate": 2.9385478643984484e-05,
548
- "loss": 0.5796,
549
  "step": 77
550
  },
551
  {
552
  "epoch": 0.14092140921409213,
553
- "grad_norm": 0.3690502006189352,
554
  "learning_rate": 2.9359792462981007e-05,
555
- "loss": 0.591,
556
  "step": 78
557
  },
558
  {
559
  "epoch": 0.14272809394760613,
560
- "grad_norm": 0.3032664893630925,
561
  "learning_rate": 2.9333592086792113e-05,
562
- "loss": 0.5986,
563
  "step": 79
564
  },
565
  {
566
  "epoch": 0.14453477868112014,
567
- "grad_norm": 0.34886668595229536,
568
  "learning_rate": 2.9306878453600382e-05,
569
- "loss": 0.6077,
570
  "step": 80
571
  },
572
  {
573
  "epoch": 0.14634146341463414,
574
- "grad_norm": 0.3308668167904688,
575
  "learning_rate": 2.9279652519967105e-05,
576
- "loss": 0.5927,
577
  "step": 81
578
  },
579
  {
580
  "epoch": 0.14814814814814814,
581
- "grad_norm": 0.34561229364377843,
582
  "learning_rate": 2.9251915260798024e-05,
583
- "loss": 0.585,
584
  "step": 82
585
  },
586
  {
587
  "epoch": 0.14995483288166214,
588
- "grad_norm": 0.3315517031152509,
589
  "learning_rate": 2.9223667669308395e-05,
590
- "loss": 0.5951,
591
  "step": 83
592
  },
593
  {
594
  "epoch": 0.15176151761517614,
595
- "grad_norm": 0.26853290095208654,
596
  "learning_rate": 2.9194910756987464e-05,
597
- "loss": 0.5885,
598
  "step": 84
599
  },
600
  {
601
  "epoch": 0.15356820234869015,
602
- "grad_norm": 0.29448765107140906,
603
  "learning_rate": 2.9165645553562215e-05,
604
- "loss": 0.5914,
605
  "step": 85
606
  },
607
  {
608
  "epoch": 0.15537488708220415,
609
- "grad_norm": 0.2806586567612587,
610
  "learning_rate": 2.9135873106960525e-05,
611
- "loss": 0.5905,
612
  "step": 86
613
  },
614
  {
615
  "epoch": 0.15718157181571815,
616
- "grad_norm": 0.3083804198003563,
617
  "learning_rate": 2.9105594483273603e-05,
618
- "loss": 0.5933,
619
  "step": 87
620
  },
621
  {
622
  "epoch": 0.15898825654923215,
623
- "grad_norm": 0.29100169406971965,
624
  "learning_rate": 2.9074810766717865e-05,
625
- "loss": 0.5971,
626
  "step": 88
627
  },
628
  {
629
  "epoch": 0.16079494128274616,
630
- "grad_norm": 0.3123482151135741,
631
  "learning_rate": 2.904352305959606e-05,
632
- "loss": 0.594,
633
  "step": 89
634
  },
635
  {
636
  "epoch": 0.16260162601626016,
637
- "grad_norm": 0.29086528776110065,
638
  "learning_rate": 2.9011732482257835e-05,
639
- "loss": 0.5858,
640
  "step": 90
641
  },
642
  {
643
  "epoch": 0.16440831074977416,
644
- "grad_norm": 0.32777518300425235,
645
  "learning_rate": 2.89794401730596e-05,
646
- "loss": 0.5838,
647
  "step": 91
648
  },
649
  {
650
  "epoch": 0.16621499548328816,
651
- "grad_norm": 0.31282829577117277,
652
  "learning_rate": 2.894664728832377e-05,
653
- "loss": 0.5878,
654
  "step": 92
655
  },
656
  {
657
  "epoch": 0.16802168021680217,
658
- "grad_norm": 0.30591224241436143,
659
  "learning_rate": 2.8913355002297367e-05,
660
- "loss": 0.5943,
661
  "step": 93
662
  },
663
  {
664
  "epoch": 0.16982836495031617,
665
- "grad_norm": 0.3454363395865153,
666
  "learning_rate": 2.887956450710995e-05,
667
- "loss": 0.5939,
668
  "step": 94
669
  },
670
  {
671
  "epoch": 0.17163504968383017,
672
- "grad_norm": 0.3119824836679083,
673
  "learning_rate": 2.8845277012730963e-05,
674
- "loss": 0.5859,
675
  "step": 95
676
  },
677
  {
678
  "epoch": 0.17344173441734417,
679
- "grad_norm": 0.29451741781860913,
680
  "learning_rate": 2.8810493746926364e-05,
681
- "loss": 0.5853,
682
  "step": 96
683
  },
684
  {
685
  "epoch": 0.17524841915085818,
686
- "grad_norm": 0.3034863549374876,
687
  "learning_rate": 2.87752159552147e-05,
688
- "loss": 0.591,
689
  "step": 97
690
  },
691
  {
692
  "epoch": 0.17705510388437218,
693
- "grad_norm": 0.28746917797897076,
694
  "learning_rate": 2.87394449008225e-05,
695
- "loss": 0.5841,
696
  "step": 98
697
  },
698
  {
699
  "epoch": 0.17886178861788618,
700
- "grad_norm": 0.30996182677103346,
701
  "learning_rate": 2.8703181864639013e-05,
702
- "loss": 0.5969,
703
  "step": 99
704
  },
705
  {
706
  "epoch": 0.18066847335140018,
707
- "grad_norm": 0.29091619146077263,
708
  "learning_rate": 2.8666428145170385e-05,
709
- "loss": 0.5702,
710
  "step": 100
711
  },
712
  {
713
  "epoch": 0.18247515808491419,
714
- "grad_norm": 0.2868056287309143,
715
  "learning_rate": 2.8629185058493116e-05,
716
- "loss": 0.5862,
717
  "step": 101
718
  },
719
  {
720
  "epoch": 0.1842818428184282,
721
- "grad_norm": 0.3505705678502723,
722
  "learning_rate": 2.8591453938206985e-05,
723
- "loss": 0.6047,
724
  "step": 102
725
  },
726
  {
727
  "epoch": 0.1860885275519422,
728
- "grad_norm": 0.27071275031049963,
729
  "learning_rate": 2.8553236135387247e-05,
730
- "loss": 0.5829,
731
  "step": 103
732
  },
733
  {
734
  "epoch": 0.1878952122854562,
735
- "grad_norm": 0.33734329421620535,
736
  "learning_rate": 2.8514533018536286e-05,
737
- "loss": 0.5811,
738
  "step": 104
739
  },
740
  {
741
  "epoch": 0.1897018970189702,
742
- "grad_norm": 0.2770290286284249,
743
  "learning_rate": 2.8475345973534605e-05,
744
- "loss": 0.5732,
745
  "step": 105
746
  },
747
  {
748
  "epoch": 0.1915085817524842,
749
- "grad_norm": 0.27942020628211717,
750
  "learning_rate": 2.8435676403591193e-05,
751
- "loss": 0.576,
752
  "step": 106
753
  },
754
  {
755
  "epoch": 0.1933152664859982,
756
- "grad_norm": 0.28553078573573465,
757
  "learning_rate": 2.8395525729193284e-05,
758
- "loss": 0.5827,
759
  "step": 107
760
  },
761
  {
762
  "epoch": 0.1951219512195122,
763
- "grad_norm": 0.25251226579381214,
764
  "learning_rate": 2.835489538805548e-05,
765
- "loss": 0.5802,
766
  "step": 108
767
  },
768
  {
769
  "epoch": 0.1969286359530262,
770
- "grad_norm": 0.27568353405316703,
771
  "learning_rate": 2.8313786835068314e-05,
772
- "loss": 0.589,
773
  "step": 109
774
  },
775
  {
776
  "epoch": 0.1987353206865402,
777
- "grad_norm": 0.27356824941424274,
778
  "learning_rate": 2.8272201542246077e-05,
779
- "loss": 0.5708,
780
  "step": 110
781
  },
782
  {
783
  "epoch": 0.2005420054200542,
784
- "grad_norm": 0.2930130437286294,
785
  "learning_rate": 2.8230140998674185e-05,
786
- "loss": 0.5906,
787
  "step": 111
788
  },
789
  {
790
  "epoch": 0.2023486901535682,
791
- "grad_norm": 0.27001628683011253,
792
  "learning_rate": 2.8187606710455807e-05,
793
- "loss": 0.584,
794
  "step": 112
795
  },
796
  {
797
  "epoch": 0.2041553748870822,
798
- "grad_norm": 0.30271557970745333,
799
  "learning_rate": 2.8144600200657953e-05,
800
- "loss": 0.5881,
801
  "step": 113
802
  },
803
  {
804
  "epoch": 0.20596205962059622,
805
- "grad_norm": 0.3099699158916623,
806
  "learning_rate": 2.8101123009256946e-05,
807
- "loss": 0.5711,
808
  "step": 114
809
  },
810
  {
811
  "epoch": 0.20776874435411022,
812
- "grad_norm": 0.26752391302167267,
813
  "learning_rate": 2.8057176693083253e-05,
814
- "loss": 0.5871,
815
  "step": 115
816
  },
817
  {
818
  "epoch": 0.20957542908762422,
819
- "grad_norm": 0.30529403725575444,
820
  "learning_rate": 2.8012762825765763e-05,
821
- "loss": 0.5808,
822
  "step": 116
823
  },
824
  {
825
  "epoch": 0.21138211382113822,
826
- "grad_norm": 0.30990442497802834,
827
  "learning_rate": 2.7967882997675424e-05,
828
- "loss": 0.5921,
829
  "step": 117
830
  },
831
  {
832
  "epoch": 0.21318879855465223,
833
- "grad_norm": 0.29406959285002965,
834
  "learning_rate": 2.7922538815868287e-05,
835
- "loss": 0.5792,
836
  "step": 118
837
  },
838
  {
839
  "epoch": 0.21499548328816623,
840
- "grad_norm": 0.26020193379576106,
841
  "learning_rate": 2.7876731904027994e-05,
842
- "loss": 0.5695,
843
  "step": 119
844
  },
845
  {
846
  "epoch": 0.21680216802168023,
847
- "grad_norm": 0.3062499009336232,
848
  "learning_rate": 2.78304639024076e-05,
849
- "loss": 0.5885,
850
  "step": 120
851
  },
852
  {
853
  "epoch": 0.2186088527551942,
854
- "grad_norm": 0.26339824339450146,
855
  "learning_rate": 2.7783736467770863e-05,
856
- "loss": 0.5839,
857
  "step": 121
858
  },
859
  {
860
  "epoch": 0.2204155374887082,
861
- "grad_norm": 0.2997530840298656,
862
  "learning_rate": 2.7736551273332908e-05,
863
- "loss": 0.589,
864
  "step": 122
865
  },
866
  {
867
  "epoch": 0.2222222222222222,
868
- "grad_norm": 0.30497954819719497,
869
  "learning_rate": 2.7688910008700305e-05,
870
- "loss": 0.5787,
871
  "step": 123
872
  },
873
  {
874
  "epoch": 0.2240289069557362,
875
- "grad_norm": 0.25400958693406966,
876
  "learning_rate": 2.764081437981059e-05,
877
- "loss": 0.5651,
878
  "step": 124
879
  },
880
  {
881
  "epoch": 0.22583559168925021,
882
- "grad_norm": 0.27337312352788334,
883
  "learning_rate": 2.7592266108871158e-05,
884
- "loss": 0.5721,
885
  "step": 125
886
  },
887
  {
888
  "epoch": 0.22764227642276422,
889
- "grad_norm": 0.3015648108540393,
890
  "learning_rate": 2.754326693429761e-05,
891
- "loss": 0.5846,
892
  "step": 126
893
  },
894
  {
895
  "epoch": 0.22944896115627822,
896
- "grad_norm": 0.3066830899471946,
897
  "learning_rate": 2.7493818610651493e-05,
898
- "loss": 0.5773,
899
  "step": 127
900
  },
901
  {
902
  "epoch": 0.23125564588979222,
903
- "grad_norm": 0.2557038265672841,
904
  "learning_rate": 2.744392290857747e-05,
905
- "loss": 0.5766,
906
  "step": 128
907
  },
908
  {
909
  "epoch": 0.23306233062330622,
910
- "grad_norm": 0.28287498283890955,
911
  "learning_rate": 2.7393581614739924e-05,
912
- "loss": 0.572,
913
  "step": 129
914
  },
915
  {
916
  "epoch": 0.23486901535682023,
917
- "grad_norm": 0.2760546012980017,
918
  "learning_rate": 2.7342796531758984e-05,
919
- "loss": 0.5902,
920
  "step": 130
921
  },
922
  {
923
  "epoch": 0.23667570009033423,
924
- "grad_norm": 0.29650720341579806,
925
  "learning_rate": 2.729156947814598e-05,
926
- "loss": 0.5849,
927
  "step": 131
928
  },
929
  {
930
  "epoch": 0.23848238482384823,
931
- "grad_norm": 0.2648334919033985,
932
  "learning_rate": 2.7239902288238297e-05,
933
- "loss": 0.5635,
934
  "step": 132
935
  },
936
  {
937
  "epoch": 0.24028906955736223,
938
- "grad_norm": 0.2684479411326874,
939
  "learning_rate": 2.7187796812133733e-05,
940
- "loss": 0.5789,
941
  "step": 133
942
  },
943
  {
944
  "epoch": 0.24209575429087624,
945
- "grad_norm": 0.2903853719359672,
946
  "learning_rate": 2.7135254915624213e-05,
947
- "loss": 0.5795,
948
  "step": 134
949
  },
950
  {
951
  "epoch": 0.24390243902439024,
952
- "grad_norm": 0.2988296028667376,
953
  "learning_rate": 2.708227848012901e-05,
954
- "loss": 0.5771,
955
  "step": 135
956
  },
957
  {
958
  "epoch": 0.24570912375790424,
959
- "grad_norm": 0.278887060643492,
960
  "learning_rate": 2.7028869402627357e-05,
961
- "loss": 0.5805,
962
  "step": 136
963
  },
964
  {
965
  "epoch": 0.24751580849141824,
966
- "grad_norm": 0.3079715134179689,
967
  "learning_rate": 2.6975029595590523e-05,
968
- "loss": 0.5643,
969
  "step": 137
970
  },
971
  {
972
  "epoch": 0.24932249322493225,
973
- "grad_norm": 0.2797620060845942,
974
  "learning_rate": 2.6920760986913332e-05,
975
- "loss": 0.5813,
976
  "step": 138
977
  },
978
  {
979
  "epoch": 0.25112917795844625,
980
- "grad_norm": 0.28689667632739463,
981
  "learning_rate": 2.6866065519845124e-05,
982
- "loss": 0.5841,
983
  "step": 139
984
  },
985
  {
986
  "epoch": 0.2529358626919603,
987
- "grad_norm": 0.2648041926457305,
988
  "learning_rate": 2.681094515292018e-05,
989
- "loss": 0.5837,
990
  "step": 140
991
  },
992
  {
993
  "epoch": 0.25474254742547425,
994
- "grad_norm": 0.25173007019005905,
995
  "learning_rate": 2.6755401859887598e-05,
996
- "loss": 0.5647,
997
  "step": 141
998
  },
999
  {
1000
  "epoch": 0.2565492321589883,
1001
- "grad_norm": 0.31032191815468113,
1002
  "learning_rate": 2.6699437629640595e-05,
1003
- "loss": 0.5843,
1004
  "step": 142
1005
  },
1006
  {
1007
  "epoch": 0.25835591689250226,
1008
- "grad_norm": 0.2888788250337772,
1009
  "learning_rate": 2.6643054466145297e-05,
1010
- "loss": 0.5761,
1011
  "step": 143
1012
  },
1013
  {
1014
  "epoch": 0.2601626016260163,
1015
- "grad_norm": 0.2760034707559814,
1016
  "learning_rate": 2.6586254388368995e-05,
1017
- "loss": 0.5647,
1018
  "step": 144
1019
  },
1020
  {
1021
  "epoch": 0.26196928635953026,
1022
- "grad_norm": 0.32528080890474936,
1023
  "learning_rate": 2.652903943020783e-05,
1024
- "loss": 0.5772,
1025
  "step": 145
1026
  },
1027
  {
1028
  "epoch": 0.26377597109304424,
1029
- "grad_norm": 0.2637573911390966,
1030
  "learning_rate": 2.647141164041398e-05,
1031
- "loss": 0.5821,
1032
  "step": 146
1033
  },
1034
  {
1035
  "epoch": 0.26558265582655827,
1036
- "grad_norm": 0.2761207440107144,
1037
  "learning_rate": 2.641337308252228e-05,
1038
- "loss": 0.5718,
1039
  "step": 147
1040
  },
1041
  {
1042
  "epoch": 0.26738934056007224,
1043
- "grad_norm": 0.2974380721456416,
1044
  "learning_rate": 2.6354925834776346e-05,
1045
- "loss": 0.5733,
1046
  "step": 148
1047
  },
1048
  {
1049
  "epoch": 0.26919602529358627,
1050
- "grad_norm": 0.2875371174946777,
1051
  "learning_rate": 2.6296071990054167e-05,
1052
- "loss": 0.5703,
1053
  "step": 149
1054
  },
1055
  {
1056
  "epoch": 0.27100271002710025,
1057
- "grad_norm": 0.26400215153050155,
1058
  "learning_rate": 2.6236813655793123e-05,
1059
- "loss": 0.5801,
1060
  "step": 150
1061
  },
1062
  {
1063
  "epoch": 0.2728093947606143,
1064
- "grad_norm": 0.2740289239257524,
1065
  "learning_rate": 2.617715295391457e-05,
1066
- "loss": 0.5621,
1067
  "step": 151
1068
  },
1069
  {
1070
  "epoch": 0.27461607949412825,
1071
- "grad_norm": 0.26263956805016714,
1072
  "learning_rate": 2.6117092020747824e-05,
1073
- "loss": 0.5697,
1074
  "step": 152
1075
  },
1076
  {
1077
  "epoch": 0.2764227642276423,
1078
- "grad_norm": 0.27308995986851226,
1079
  "learning_rate": 2.6056633006953677e-05,
1080
- "loss": 0.5697,
1081
  "step": 153
1082
  },
1083
  {
1084
  "epoch": 0.27822944896115626,
1085
- "grad_norm": 0.2682634840950432,
1086
  "learning_rate": 2.5995778077447393e-05,
1087
- "loss": 0.5735,
1088
  "step": 154
1089
  },
1090
  {
1091
  "epoch": 0.2800361336946703,
1092
- "grad_norm": 0.30213654494482944,
1093
  "learning_rate": 2.5934529411321174e-05,
1094
- "loss": 0.5668,
1095
  "step": 155
1096
  },
1097
  {
1098
  "epoch": 0.28184281842818426,
1099
- "grad_norm": 0.24055343886463815,
1100
  "learning_rate": 2.587288920176613e-05,
1101
- "loss": 0.5662,
1102
  "step": 156
1103
  },
1104
  {
1105
  "epoch": 0.2836495031616983,
1106
- "grad_norm": 0.2913187789114745,
1107
  "learning_rate": 2.581085965599375e-05,
1108
- "loss": 0.5795,
1109
  "step": 157
1110
  },
1111
  {
1112
  "epoch": 0.28545618789521227,
1113
- "grad_norm": 0.2576506841389344,
1114
  "learning_rate": 2.5748442995156882e-05,
1115
- "loss": 0.5766,
1116
  "step": 158
1117
  },
1118
  {
1119
  "epoch": 0.2872628726287263,
1120
- "grad_norm": 0.27596403754545956,
1121
  "learning_rate": 2.5685641454270172e-05,
1122
- "loss": 0.5577,
1123
  "step": 159
1124
  },
1125
  {
1126
  "epoch": 0.28906955736224027,
1127
- "grad_norm": 0.2935051396800575,
1128
  "learning_rate": 2.5622457282130046e-05,
1129
- "loss": 0.5668,
1130
  "step": 160
1131
  },
1132
  {
1133
  "epoch": 0.2908762420957543,
1134
- "grad_norm": 0.261323419986792,
1135
  "learning_rate": 2.5558892741234173e-05,
1136
- "loss": 0.5744,
1137
  "step": 161
1138
  },
1139
  {
1140
  "epoch": 0.2926829268292683,
1141
- "grad_norm": 0.29994364267277224,
1142
  "learning_rate": 2.5494950107700482e-05,
1143
- "loss": 0.5606,
1144
  "step": 162
1145
  },
1146
  {
1147
  "epoch": 0.2944896115627823,
1148
- "grad_norm": 0.25057252176689093,
1149
  "learning_rate": 2.5430631671185616e-05,
1150
- "loss": 0.5689,
1151
  "step": 163
1152
  },
1153
  {
1154
  "epoch": 0.2962962962962963,
1155
- "grad_norm": 0.30439652791987415,
1156
  "learning_rate": 2.5365939734802973e-05,
1157
- "loss": 0.5784,
1158
  "step": 164
1159
  },
1160
  {
1161
  "epoch": 0.2981029810298103,
1162
- "grad_norm": 0.2561891058565834,
1163
  "learning_rate": 2.5300876615040223e-05,
1164
- "loss": 0.5693,
1165
  "step": 165
1166
  },
1167
  {
1168
  "epoch": 0.2999096657633243,
1169
- "grad_norm": 0.30789286843223074,
1170
  "learning_rate": 2.523544464167637e-05,
1171
- "loss": 0.5755,
1172
  "step": 166
1173
  },
1174
  {
1175
  "epoch": 0.3017163504968383,
1176
- "grad_norm": 0.2604413140600789,
1177
  "learning_rate": 2.5169646157698313e-05,
1178
- "loss": 0.574,
1179
  "step": 167
1180
  },
1181
  {
1182
  "epoch": 0.3035230352303523,
1183
- "grad_norm": 0.2772614270807846,
1184
  "learning_rate": 2.5103483519216964e-05,
1185
- "loss": 0.5731,
1186
  "step": 168
1187
  },
1188
  {
1189
  "epoch": 0.3053297199638663,
1190
- "grad_norm": 0.26896696271290815,
1191
  "learning_rate": 2.5036959095382875e-05,
1192
- "loss": 0.565,
1193
  "step": 169
1194
  },
1195
  {
1196
  "epoch": 0.3071364046973803,
1197
- "grad_norm": 0.29065186309501,
1198
  "learning_rate": 2.4970075268301388e-05,
1199
- "loss": 0.5783,
1200
  "step": 170
1201
  },
1202
  {
1203
  "epoch": 0.3089430894308943,
1204
- "grad_norm": 0.2466338266655888,
1205
  "learning_rate": 2.4902834432947353e-05,
1206
- "loss": 0.5614,
1207
  "step": 171
1208
  },
1209
  {
1210
  "epoch": 0.3107497741644083,
1211
- "grad_norm": 0.27214096978444124,
1212
  "learning_rate": 2.4835238997079382e-05,
1213
- "loss": 0.5687,
1214
  "step": 172
1215
  },
1216
  {
1217
  "epoch": 0.31255645889792233,
1218
- "grad_norm": 0.2651928040714497,
1219
  "learning_rate": 2.4767291381153603e-05,
1220
- "loss": 0.5719,
1221
  "step": 173
1222
  },
1223
  {
1224
  "epoch": 0.3143631436314363,
1225
- "grad_norm": 0.23672196782852778,
1226
  "learning_rate": 2.4698994018236994e-05,
1227
- "loss": 0.5631,
1228
  "step": 174
1229
  },
1230
  {
1231
  "epoch": 0.31616982836495033,
1232
- "grad_norm": 0.28356502585206267,
1233
  "learning_rate": 2.4630349353920284e-05,
1234
- "loss": 0.5671,
1235
  "step": 175
1236
  },
1237
  {
1238
  "epoch": 0.3179765130984643,
1239
- "grad_norm": 0.28540875579685554,
1240
  "learning_rate": 2.4561359846230346e-05,
1241
- "loss": 0.5818,
1242
  "step": 176
1243
  },
1244
  {
1245
  "epoch": 0.31978319783197834,
1246
- "grad_norm": 0.24190126114290808,
1247
  "learning_rate": 2.4492027965542217e-05,
1248
- "loss": 0.5569,
1249
  "step": 177
1250
  },
1251
  {
1252
  "epoch": 0.3215898825654923,
1253
- "grad_norm": 0.27287155700655424,
1254
  "learning_rate": 2.44223561944906e-05,
1255
- "loss": 0.5643,
1256
  "step": 178
1257
  },
1258
  {
1259
  "epoch": 0.32339656729900634,
1260
- "grad_norm": 0.24924695151585574,
1261
  "learning_rate": 2.4352347027881003e-05,
1262
- "loss": 0.5738,
1263
  "step": 179
1264
  },
1265
  {
1266
  "epoch": 0.3252032520325203,
1267
- "grad_norm": 0.2759090650898249,
1268
  "learning_rate": 2.4282002972600382e-05,
1269
- "loss": 0.5817,
1270
  "step": 180
1271
  },
1272
  {
1273
  "epoch": 0.32700993676603435,
1274
- "grad_norm": 0.22785083526499783,
1275
  "learning_rate": 2.4211326547527377e-05,
1276
- "loss": 0.5601,
1277
  "step": 181
1278
  },
1279
  {
1280
  "epoch": 0.3288166214995483,
1281
- "grad_norm": 0.2691907674221595,
1282
  "learning_rate": 2.4140320283442125e-05,
1283
- "loss": 0.5699,
1284
  "step": 182
1285
  },
1286
  {
1287
  "epoch": 0.33062330623306235,
1288
- "grad_norm": 0.2708392714868768,
1289
  "learning_rate": 2.4068986722935625e-05,
1290
- "loss": 0.5806,
1291
  "step": 183
1292
  },
1293
  {
1294
  "epoch": 0.3324299909665763,
1295
- "grad_norm": 0.27339466736189244,
1296
  "learning_rate": 2.3997328420318705e-05,
1297
- "loss": 0.5769,
1298
  "step": 184
1299
  },
1300
  {
1301
  "epoch": 0.33423667570009036,
1302
- "grad_norm": 0.26435674776037377,
1303
  "learning_rate": 2.3925347941530556e-05,
1304
- "loss": 0.5733,
1305
  "step": 185
1306
  },
1307
  {
1308
  "epoch": 0.33604336043360433,
1309
- "grad_norm": 0.26341316111660884,
1310
  "learning_rate": 2.3853047864046843e-05,
1311
- "loss": 0.5684,
1312
  "step": 186
1313
  },
1314
  {
1315
  "epoch": 0.33785004516711836,
1316
- "grad_norm": 0.27858160542364235,
1317
  "learning_rate": 2.3780430776787413e-05,
1318
- "loss": 0.5658,
1319
  "step": 187
1320
  },
1321
  {
1322
  "epoch": 0.33965672990063234,
1323
- "grad_norm": 0.23833042598843737,
1324
  "learning_rate": 2.3707499280023604e-05,
1325
- "loss": 0.5655,
1326
  "step": 188
1327
  },
1328
  {
1329
  "epoch": 0.34146341463414637,
1330
- "grad_norm": 0.2808631689375272,
1331
  "learning_rate": 2.3634255985285104e-05,
1332
- "loss": 0.5729,
1333
  "step": 189
1334
  },
1335
  {
1336
  "epoch": 0.34327009936766034,
1337
- "grad_norm": 0.25111983735673515,
1338
  "learning_rate": 2.356070351526648e-05,
1339
- "loss": 0.562,
1340
  "step": 190
1341
  },
1342
  {
1343
  "epoch": 0.34507678410117437,
1344
- "grad_norm": 0.2571681206798971,
1345
  "learning_rate": 2.348684450373322e-05,
1346
- "loss": 0.5599,
1347
  "step": 191
1348
  },
1349
  {
1350
  "epoch": 0.34688346883468835,
1351
- "grad_norm": 0.2409025648925046,
1352
  "learning_rate": 2.3412681595427467e-05,
1353
- "loss": 0.5674,
1354
  "step": 192
1355
  },
1356
  {
1357
  "epoch": 0.3486901535682023,
1358
- "grad_norm": 0.29444654953284566,
1359
  "learning_rate": 2.3338217445973268e-05,
1360
- "loss": 0.5687,
1361
  "step": 193
1362
  },
1363
  {
1364
  "epoch": 0.35049683830171635,
1365
- "grad_norm": 0.2496920890877963,
1366
  "learning_rate": 2.3263454721781537e-05,
1367
- "loss": 0.5705,
1368
  "step": 194
1369
  },
1370
  {
1371
  "epoch": 0.3523035230352303,
1372
- "grad_norm": 0.27467425910302257,
1373
  "learning_rate": 2.318839609995453e-05,
1374
- "loss": 0.5536,
1375
  "step": 195
1376
  },
1377
  {
1378
  "epoch": 0.35411020776874436,
1379
- "grad_norm": 0.25067206006192067,
1380
  "learning_rate": 2.3113044268189995e-05,
1381
- "loss": 0.5668,
1382
  "step": 196
1383
  },
1384
  {
1385
  "epoch": 0.35591689250225833,
1386
- "grad_norm": 0.25127765224467463,
1387
  "learning_rate": 2.303740192468495e-05,
1388
- "loss": 0.5671,
1389
  "step": 197
1390
  },
1391
  {
1392
  "epoch": 0.35772357723577236,
1393
- "grad_norm": 0.2742190997822365,
1394
  "learning_rate": 2.2961471778039045e-05,
1395
- "loss": 0.5436,
1396
  "step": 198
1397
  },
1398
  {
1399
  "epoch": 0.35953026196928634,
1400
- "grad_norm": 0.2617420424365515,
1401
  "learning_rate": 2.288525654715757e-05,
1402
- "loss": 0.567,
1403
  "step": 199
1404
  },
1405
  {
1406
  "epoch": 0.36133694670280037,
1407
- "grad_norm": 0.2601868662327404,
1408
  "learning_rate": 2.280875896115413e-05,
1409
- "loss": 0.556,
1410
  "step": 200
1411
  },
1412
  {
1413
  "epoch": 0.36314363143631434,
1414
- "grad_norm": 0.2488467816010916,
1415
  "learning_rate": 2.2731981759252876e-05,
1416
- "loss": 0.5506,
1417
  "step": 201
1418
  },
1419
  {
1420
  "epoch": 0.36495031616982837,
1421
- "grad_norm": 0.24184725633849133,
1422
  "learning_rate": 2.2654927690690445e-05,
1423
- "loss": 0.5773,
1424
  "step": 202
1425
  },
1426
  {
1427
  "epoch": 0.36675700090334235,
1428
- "grad_norm": 0.25373414117124127,
1429
  "learning_rate": 2.257759951461752e-05,
1430
- "loss": 0.5641,
1431
  "step": 203
1432
  },
1433
  {
1434
  "epoch": 0.3685636856368564,
1435
- "grad_norm": 0.26564822413876754,
1436
  "learning_rate": 2.25e-05,
1437
- "loss": 0.5773,
1438
  "step": 204
1439
  },
1440
  {
1441
  "epoch": 0.37037037037037035,
1442
- "grad_norm": 0.22934416047361944,
1443
  "learning_rate": 2.24221319255199e-05,
1444
- "loss": 0.5677,
1445
  "step": 205
1446
  },
1447
  {
1448
  "epoch": 0.3721770551038844,
1449
- "grad_norm": 0.27064508142785815,
1450
  "learning_rate": 2.234399807947579e-05,
1451
- "loss": 0.5548,
1452
  "step": 206
1453
  },
1454
  {
1455
  "epoch": 0.37398373983739835,
1456
- "grad_norm": 0.2371896313671772,
1457
  "learning_rate": 2.2265601259683e-05,
1458
- "loss": 0.5627,
1459
  "step": 207
1460
  },
1461
  {
1462
  "epoch": 0.3757904245709124,
1463
- "grad_norm": 0.24008373009834236,
1464
  "learning_rate": 2.2186944273373426e-05,
1465
- "loss": 0.5499,
1466
  "step": 208
1467
  },
1468
  {
1469
  "epoch": 0.37759710930442636,
1470
- "grad_norm": 0.25685866079731606,
1471
  "learning_rate": 2.210802993709498e-05,
1472
- "loss": 0.5717,
1473
  "step": 209
1474
  },
1475
  {
1476
  "epoch": 0.3794037940379404,
1477
- "grad_norm": 0.2629866117258039,
1478
  "learning_rate": 2.202886107661078e-05,
1479
- "loss": 0.5676,
1480
  "step": 210
1481
  },
1482
  {
1483
  "epoch": 0.38121047877145436,
1484
- "grad_norm": 0.2522800121206238,
1485
  "learning_rate": 2.1949440526797928e-05,
1486
- "loss": 0.5669,
1487
  "step": 211
1488
  },
1489
  {
1490
  "epoch": 0.3830171635049684,
1491
- "grad_norm": 0.2540123635971477,
1492
  "learning_rate": 2.1869771131546015e-05,
1493
- "loss": 0.5746,
1494
  "step": 212
1495
  },
1496
  {
1497
  "epoch": 0.38482384823848237,
1498
- "grad_norm": 0.2673805985713425,
1499
  "learning_rate": 2.178985574365529e-05,
1500
- "loss": 0.5751,
1501
  "step": 213
1502
  },
1503
  {
1504
  "epoch": 0.3866305329719964,
1505
- "grad_norm": 0.26844790033448696,
1506
  "learning_rate": 2.170969722473449e-05,
1507
- "loss": 0.5608,
1508
  "step": 214
1509
  },
1510
  {
1511
  "epoch": 0.3884372177055104,
1512
- "grad_norm": 0.3287228900436167,
1513
  "learning_rate": 2.1629298445098403e-05,
1514
- "loss": 0.5881,
1515
  "step": 215
1516
  },
1517
  {
1518
  "epoch": 0.3902439024390244,
1519
- "grad_norm": 0.24641908304543073,
1520
  "learning_rate": 2.154866228366505e-05,
1521
- "loss": 0.5767,
1522
  "step": 216
1523
  },
1524
  {
1525
  "epoch": 0.3920505871725384,
1526
- "grad_norm": 0.2356470126020688,
1527
  "learning_rate": 2.146779162785263e-05,
1528
- "loss": 0.5768,
1529
  "step": 217
1530
  },
1531
  {
1532
  "epoch": 0.3938572719060524,
1533
- "grad_norm": 0.26263666882985864,
1534
  "learning_rate": 2.138668937347609e-05,
1535
- "loss": 0.5683,
1536
  "step": 218
1537
  },
1538
  {
1539
  "epoch": 0.3956639566395664,
1540
- "grad_norm": 0.23701425043610796,
1541
  "learning_rate": 2.1305358424643484e-05,
1542
- "loss": 0.5634,
1543
  "step": 219
1544
  },
1545
  {
1546
  "epoch": 0.3974706413730804,
1547
- "grad_norm": 0.24138606059569664,
1548
  "learning_rate": 2.1223801693651927e-05,
1549
- "loss": 0.5656,
1550
  "step": 220
1551
  },
1552
  {
1553
  "epoch": 0.3992773261065944,
1554
- "grad_norm": 0.2595567201221021,
1555
  "learning_rate": 2.114202210088336e-05,
1556
- "loss": 0.5546,
1557
  "step": 221
1558
  },
1559
  {
1560
  "epoch": 0.4010840108401084,
1561
- "grad_norm": 0.2491513558591219,
1562
  "learning_rate": 2.106002257469993e-05,
1563
- "loss": 0.5718,
1564
  "step": 222
1565
  },
1566
  {
1567
  "epoch": 0.4028906955736224,
1568
- "grad_norm": 0.2486982800205461,
1569
  "learning_rate": 2.0977806051339172e-05,
1570
- "loss": 0.5551,
1571
  "step": 223
1572
  },
1573
  {
1574
  "epoch": 0.4046973803071364,
1575
- "grad_norm": 0.23727864667012805,
1576
  "learning_rate": 2.0895375474808857e-05,
1577
- "loss": 0.559,
1578
  "step": 224
1579
  },
1580
  {
1581
  "epoch": 0.4065040650406504,
1582
- "grad_norm": 0.287906688283815,
1583
  "learning_rate": 2.0812733796781544e-05,
1584
- "loss": 0.5648,
1585
  "step": 225
1586
  },
1587
  {
1588
  "epoch": 0.4083107497741644,
1589
- "grad_norm": 0.23398485566681798,
1590
  "learning_rate": 2.0729883976488936e-05,
1591
- "loss": 0.5699,
1592
  "step": 226
1593
  },
1594
  {
1595
  "epoch": 0.4101174345076784,
1596
- "grad_norm": 0.2672121559206604,
1597
  "learning_rate": 2.064682898061588e-05,
1598
- "loss": 0.5745,
1599
  "step": 227
1600
  },
1601
  {
1602
  "epoch": 0.41192411924119243,
1603
- "grad_norm": 0.25274044632247594,
1604
  "learning_rate": 2.0563571783194146e-05,
1605
- "loss": 0.5607,
1606
  "step": 228
1607
  },
1608
  {
1609
  "epoch": 0.4137308039747064,
1610
- "grad_norm": 0.23024580592682378,
1611
  "learning_rate": 2.0480115365495928e-05,
1612
- "loss": 0.5633,
1613
  "step": 229
1614
  },
1615
  {
1616
  "epoch": 0.41553748870822044,
1617
- "grad_norm": 0.2643430047659416,
1618
  "learning_rate": 2.0396462715927107e-05,
1619
- "loss": 0.5657,
1620
  "step": 230
1621
  },
1622
  {
1623
  "epoch": 0.4173441734417344,
1624
- "grad_norm": 0.24590645504611103,
1625
  "learning_rate": 2.0312616829920222e-05,
1626
- "loss": 0.5651,
1627
  "step": 231
1628
  },
1629
  {
1630
  "epoch": 0.41915085817524844,
1631
- "grad_norm": 0.232079984180177,
1632
  "learning_rate": 2.022858070982723e-05,
1633
- "loss": 0.5676,
1634
  "step": 232
1635
  },
1636
  {
1637
  "epoch": 0.4209575429087624,
1638
- "grad_norm": 0.23416214443037042,
1639
  "learning_rate": 2.0144357364811973e-05,
1640
- "loss": 0.5501,
1641
  "step": 233
1642
  },
1643
  {
1644
  "epoch": 0.42276422764227645,
1645
- "grad_norm": 0.24394361384059154,
1646
  "learning_rate": 2.0059949810742452e-05,
1647
- "loss": 0.565,
1648
  "step": 234
1649
  },
1650
  {
1651
  "epoch": 0.4245709123757904,
1652
- "grad_norm": 0.22171860398747742,
1653
  "learning_rate": 1.997536107008281e-05,
1654
- "loss": 0.5603,
1655
  "step": 235
1656
  },
1657
  {
1658
  "epoch": 0.42637759710930445,
1659
- "grad_norm": 0.2573737818729415,
1660
  "learning_rate": 1.9890594171785128e-05,
1661
- "loss": 0.5718,
1662
  "step": 236
1663
  },
1664
  {
1665
  "epoch": 0.4281842818428184,
1666
- "grad_norm": 0.24929065928195437,
1667
  "learning_rate": 1.9805652151180945e-05,
1668
- "loss": 0.5716,
1669
  "step": 237
1670
  },
1671
  {
1672
  "epoch": 0.42999096657633246,
1673
- "grad_norm": 0.22521672092224404,
1674
  "learning_rate": 1.972053804987258e-05,
1675
- "loss": 0.5799,
1676
  "step": 238
1677
  },
1678
  {
1679
  "epoch": 0.43179765130984643,
1680
- "grad_norm": 0.2400133312081751,
1681
  "learning_rate": 1.963525491562421e-05,
1682
- "loss": 0.5607,
1683
  "step": 239
1684
  },
1685
  {
1686
  "epoch": 0.43360433604336046,
1687
- "grad_norm": 0.2220251642670775,
1688
  "learning_rate": 1.954980580225275e-05,
1689
- "loss": 0.5613,
1690
  "step": 240
1691
  },
1692
  {
1693
  "epoch": 0.43541102077687444,
1694
- "grad_norm": 0.24242836746273963,
1695
  "learning_rate": 1.946419376951848e-05,
1696
- "loss": 0.5601,
1697
  "step": 241
1698
  },
1699
  {
1700
  "epoch": 0.4372177055103884,
1701
- "grad_norm": 0.23280796509155544,
1702
  "learning_rate": 1.9378421883015505e-05,
1703
- "loss": 0.5608,
1704
  "step": 242
1705
  },
1706
  {
1707
  "epoch": 0.43902439024390244,
1708
- "grad_norm": 0.2424520004437008,
1709
  "learning_rate": 1.9292493214061953e-05,
1710
- "loss": 0.5653,
1711
  "step": 243
1712
  },
1713
  {
1714
  "epoch": 0.4408310749774164,
1715
- "grad_norm": 0.21938548719470424,
1716
  "learning_rate": 1.9206410839590042e-05,
1717
- "loss": 0.5713,
1718
  "step": 244
1719
  },
1720
  {
1721
  "epoch": 0.44263775971093045,
1722
- "grad_norm": 0.2873818775831322,
1723
  "learning_rate": 1.9120177842035853e-05,
1724
- "loss": 0.5635,
1725
  "step": 245
1726
  },
1727
  {
1728
  "epoch": 0.4444444444444444,
1729
- "grad_norm": 0.24359879240501903,
1730
  "learning_rate": 1.9033797309228984e-05,
1731
- "loss": 0.5703,
1732
  "step": 246
1733
  },
1734
  {
1735
  "epoch": 0.44625112917795845,
1736
- "grad_norm": 0.2357242641893439,
1737
  "learning_rate": 1.8947272334281977e-05,
1738
- "loss": 0.5521,
1739
  "step": 247
1740
  },
1741
  {
1742
  "epoch": 0.4480578139114724,
1743
- "grad_norm": 0.2462690537681842,
1744
  "learning_rate": 1.8860606015479537e-05,
1745
- "loss": 0.5706,
1746
  "step": 248
1747
  },
1748
  {
1749
  "epoch": 0.44986449864498645,
1750
- "grad_norm": 0.23970905475638837,
1751
  "learning_rate": 1.877380145616763e-05,
1752
- "loss": 0.5543,
1753
  "step": 249
1754
  },
1755
  {
1756
  "epoch": 0.45167118337850043,
1757
- "grad_norm": 0.25559374336438795,
1758
  "learning_rate": 1.868686176464232e-05,
1759
- "loss": 0.5694,
1760
  "step": 250
1761
  },
1762
  {
1763
  "epoch": 0.45347786811201446,
1764
- "grad_norm": 0.22667022533452194,
1765
  "learning_rate": 1.8599790054038487e-05,
1766
- "loss": 0.5563,
1767
  "step": 251
1768
  },
1769
  {
1770
  "epoch": 0.45528455284552843,
1771
- "grad_norm": 0.2328027982974321,
1772
  "learning_rate": 1.8512589442218358e-05,
1773
- "loss": 0.5384,
1774
  "step": 252
1775
  },
1776
  {
1777
  "epoch": 0.45709123757904246,
1778
- "grad_norm": 0.2490658489853876,
1779
  "learning_rate": 1.8425263051659838e-05,
1780
- "loss": 0.5613,
1781
  "step": 253
1782
  },
1783
  {
1784
  "epoch": 0.45889792231255644,
1785
- "grad_norm": 0.25086634603487334,
1786
  "learning_rate": 1.8337814009344716e-05,
1787
- "loss": 0.5602,
1788
  "step": 254
1789
  },
1790
  {
1791
  "epoch": 0.46070460704607047,
1792
- "grad_norm": 0.25605732223142635,
1793
  "learning_rate": 1.8250245446646707e-05,
1794
- "loss": 0.5535,
1795
  "step": 255
1796
  },
1797
  {
1798
  "epoch": 0.46251129177958444,
1799
- "grad_norm": 0.26469759888755845,
1800
  "learning_rate": 1.8162560499219286e-05,
1801
- "loss": 0.5537,
1802
  "step": 256
1803
  },
1804
  {
1805
  "epoch": 0.4643179765130985,
1806
- "grad_norm": 0.2476106293930312,
1807
  "learning_rate": 1.807476230688346e-05,
1808
- "loss": 0.5644,
1809
  "step": 257
1810
  },
1811
  {
1812
  "epoch": 0.46612466124661245,
1813
- "grad_norm": 0.2611791488235218,
1814
  "learning_rate": 1.7986854013515274e-05,
1815
- "loss": 0.5589,
1816
  "step": 258
1817
  },
1818
  {
1819
  "epoch": 0.4679313459801265,
1820
- "grad_norm": 0.25614740455319657,
1821
  "learning_rate": 1.78988387669333e-05,
1822
- "loss": 0.5626,
1823
  "step": 259
1824
  },
1825
  {
1826
  "epoch": 0.46973803071364045,
1827
- "grad_norm": 0.26413847448296346,
1828
  "learning_rate": 1.781071971878587e-05,
1829
- "loss": 0.5763,
1830
  "step": 260
1831
  },
1832
  {
1833
  "epoch": 0.4715447154471545,
1834
- "grad_norm": 0.2670037728752572,
1835
  "learning_rate": 1.7722500024438244e-05,
1836
- "loss": 0.5671,
1837
  "step": 261
1838
  },
1839
  {
1840
  "epoch": 0.47335140018066846,
1841
- "grad_norm": 0.2655083894599294,
1842
  "learning_rate": 1.7634182842859628e-05,
1843
- "loss": 0.5556,
1844
  "step": 262
1845
  },
1846
  {
1847
  "epoch": 0.4751580849141825,
1848
- "grad_norm": 0.2602266219558274,
1849
  "learning_rate": 1.7545771336510033e-05,
1850
- "loss": 0.5603,
1851
  "step": 263
1852
  },
1853
  {
1854
  "epoch": 0.47696476964769646,
1855
- "grad_norm": 0.23837446328300185,
1856
  "learning_rate": 1.7457268671227067e-05,
1857
- "loss": 0.5661,
1858
  "step": 264
1859
  },
1860
  {
1861
  "epoch": 0.4787714543812105,
1862
- "grad_norm": 0.2855243343978605,
1863
  "learning_rate": 1.736867801611254e-05,
1864
- "loss": 0.5479,
1865
  "step": 265
1866
  },
1867
  {
1868
  "epoch": 0.48057813911472447,
1869
- "grad_norm": 0.23335711495710984,
1870
  "learning_rate": 1.728000254341901e-05,
1871
- "loss": 0.5717,
1872
  "step": 266
1873
  },
1874
  {
1875
  "epoch": 0.4823848238482385,
1876
- "grad_norm": 0.2257992261469873,
1877
  "learning_rate": 1.7191245428436175e-05,
1878
- "loss": 0.5566,
1879
  "step": 267
1880
  },
1881
  {
1882
  "epoch": 0.48419150858175247,
1883
- "grad_norm": 0.23986035421561297,
1884
  "learning_rate": 1.7102409849377188e-05,
1885
- "loss": 0.5563,
1886
  "step": 268
1887
  },
1888
  {
1889
  "epoch": 0.4859981933152665,
1890
- "grad_norm": 0.22957382135670054,
1891
  "learning_rate": 1.7013498987264832e-05,
1892
- "loss": 0.5634,
1893
  "step": 269
1894
  },
1895
  {
1896
  "epoch": 0.4878048780487805,
1897
- "grad_norm": 0.2332081125649847,
1898
  "learning_rate": 1.6924516025817636e-05,
1899
- "loss": 0.5448,
1900
  "step": 270
1901
  },
1902
  {
1903
  "epoch": 0.4896115627822945,
1904
- "grad_norm": 0.2358236013629391,
1905
  "learning_rate": 1.683546415133584e-05,
1906
- "loss": 0.5663,
1907
  "step": 271
1908
  },
1909
  {
1910
  "epoch": 0.4914182475158085,
1911
- "grad_norm": 0.2651609439225248,
1912
  "learning_rate": 1.6746346552587342e-05,
1913
- "loss": 0.561,
1914
  "step": 272
1915
  },
1916
  {
1917
  "epoch": 0.4932249322493225,
1918
- "grad_norm": 0.25056099802020176,
1919
  "learning_rate": 1.665716642069349e-05,
1920
- "loss": 0.5689,
1921
  "step": 273
1922
  },
1923
  {
1924
  "epoch": 0.4950316169828365,
1925
- "grad_norm": 0.2699052892919483,
1926
  "learning_rate": 1.6567926949014805e-05,
1927
- "loss": 0.5567,
1928
  "step": 274
1929
  },
1930
  {
1931
  "epoch": 0.4968383017163505,
1932
- "grad_norm": 0.23647220641096864,
1933
  "learning_rate": 1.6478631333036655e-05,
1934
- "loss": 0.5507,
1935
  "step": 275
1936
  },
1937
  {
1938
  "epoch": 0.4986449864498645,
1939
- "grad_norm": 0.21602219843408765,
1940
  "learning_rate": 1.638928277025482e-05,
1941
- "loss": 0.5432,
1942
  "step": 276
1943
  },
1944
  {
1945
  "epoch": 0.5004516711833785,
1946
- "grad_norm": 0.2354573231695346,
1947
  "learning_rate": 1.6299884460061005e-05,
1948
- "loss": 0.5547,
1949
  "step": 277
1950
  },
1951
  {
1952
  "epoch": 0.5022583559168925,
1953
- "grad_norm": 0.2197114014140093,
1954
  "learning_rate": 1.621043960362826e-05,
1955
- "loss": 0.5418,
1956
  "step": 278
1957
  },
1958
  {
1959
  "epoch": 0.5040650406504065,
1960
- "grad_norm": 0.23257701295408562,
1961
  "learning_rate": 1.6120951403796367e-05,
1962
- "loss": 0.5568,
1963
  "step": 279
1964
  },
1965
  {
1966
  "epoch": 0.5058717253839206,
1967
- "grad_norm": 0.24433486985285227,
1968
  "learning_rate": 1.603142306495714e-05,
1969
- "loss": 0.553,
1970
  "step": 280
1971
  },
1972
  {
1973
  "epoch": 0.5076784101174345,
1974
- "grad_norm": 0.20527397519546922,
1975
  "learning_rate": 1.5941857792939702e-05,
1976
- "loss": 0.554,
1977
  "step": 281
1978
  },
1979
  {
1980
  "epoch": 0.5094850948509485,
1981
- "grad_norm": 0.22929324716898536,
1982
  "learning_rate": 1.585225879489567e-05,
1983
- "loss": 0.5487,
1984
  "step": 282
1985
  },
1986
  {
1987
  "epoch": 0.5112917795844625,
1988
- "grad_norm": 0.2238089603853136,
1989
  "learning_rate": 1.5762629279184326e-05,
1990
- "loss": 0.5634,
1991
  "step": 283
1992
  },
1993
  {
1994
  "epoch": 0.5130984643179766,
1995
- "grad_norm": 0.22412925067398695,
1996
  "learning_rate": 1.5672972455257726e-05,
1997
- "loss": 0.5518,
1998
  "step": 284
1999
  },
2000
  {
2001
  "epoch": 0.5149051490514905,
2002
- "grad_norm": 0.23076070566516943,
2003
  "learning_rate": 1.5583291533545775e-05,
2004
- "loss": 0.5463,
2005
  "step": 285
2006
  },
2007
  {
2008
  "epoch": 0.5167118337850045,
2009
- "grad_norm": 0.23140896156408275,
2010
  "learning_rate": 1.549358972534128e-05,
2011
- "loss": 0.5437,
2012
  "step": 286
2013
  },
2014
  {
2015
  "epoch": 0.5185185185185185,
2016
- "grad_norm": 0.22643479753851717,
2017
  "learning_rate": 1.5403870242684942e-05,
2018
- "loss": 0.5651,
2019
  "step": 287
2020
  },
2021
  {
2022
  "epoch": 0.5203252032520326,
2023
- "grad_norm": 0.24172544315368483,
2024
  "learning_rate": 1.5314136298250355e-05,
2025
- "loss": 0.5625,
2026
  "step": 288
2027
  },
2028
  {
2029
  "epoch": 0.5221318879855466,
2030
- "grad_norm": 0.21456673351384214,
2031
  "learning_rate": 1.5224391105228956e-05,
2032
- "loss": 0.547,
2033
  "step": 289
2034
  },
2035
  {
2036
  "epoch": 0.5239385727190605,
2037
- "grad_norm": 0.20236395606188895,
2038
  "learning_rate": 1.5134637877214968e-05,
2039
- "loss": 0.5588,
2040
  "step": 290
2041
  },
2042
  {
2043
  "epoch": 0.5257452574525745,
2044
- "grad_norm": 0.23554049753605505,
2045
  "learning_rate": 1.5044879828090346e-05,
2046
- "loss": 0.5631,
2047
  "step": 291
2048
  },
2049
  {
2050
  "epoch": 0.5275519421860885,
2051
- "grad_norm": 0.22947805697224488,
2052
  "learning_rate": 1.4955120171909658e-05,
2053
- "loss": 0.5669,
2054
  "step": 292
2055
  },
2056
  {
2057
  "epoch": 0.5293586269196026,
2058
- "grad_norm": 0.1986163494264529,
2059
  "learning_rate": 1.4865362122785031e-05,
2060
- "loss": 0.5595,
2061
  "step": 293
2062
  },
2063
  {
2064
  "epoch": 0.5311653116531165,
2065
- "grad_norm": 0.22657031125938396,
2066
  "learning_rate": 1.4775608894771048e-05,
2067
- "loss": 0.5519,
2068
  "step": 294
2069
  },
2070
  {
2071
  "epoch": 0.5329719963866305,
2072
- "grad_norm": 0.26178909293346697,
2073
  "learning_rate": 1.4685863701749648e-05,
2074
- "loss": 0.5497,
2075
  "step": 295
2076
  },
2077
  {
2078
  "epoch": 0.5347786811201445,
2079
- "grad_norm": 0.24491042948023758,
2080
  "learning_rate": 1.4596129757315062e-05,
2081
- "loss": 0.5537,
2082
  "step": 296
2083
  },
2084
  {
2085
  "epoch": 0.5365853658536586,
2086
- "grad_norm": 0.234622033552508,
2087
  "learning_rate": 1.4506410274658718e-05,
2088
- "loss": 0.5586,
2089
  "step": 297
2090
  },
2091
  {
2092
  "epoch": 0.5383920505871725,
2093
- "grad_norm": 0.23622066572903277,
2094
  "learning_rate": 1.441670846645423e-05,
2095
- "loss": 0.5533,
2096
  "step": 298
2097
  },
2098
  {
2099
  "epoch": 0.5401987353206865,
2100
- "grad_norm": 0.24071443512391985,
2101
  "learning_rate": 1.4327027544742281e-05,
2102
- "loss": 0.5685,
2103
  "step": 299
2104
  },
2105
  {
2106
  "epoch": 0.5420054200542005,
2107
- "grad_norm": 0.22544308961774373,
2108
  "learning_rate": 1.4237370720815675e-05,
2109
- "loss": 0.5566,
2110
  "step": 300
2111
  },
2112
  {
2113
  "epoch": 0.5438121047877146,
2114
- "grad_norm": 0.20755783402347355,
2115
  "learning_rate": 1.4147741205104336e-05,
2116
- "loss": 0.5585,
2117
  "step": 301
2118
  },
2119
  {
2120
  "epoch": 0.5456187895212286,
2121
- "grad_norm": 0.21105544320784098,
2122
  "learning_rate": 1.40581422070603e-05,
2123
- "loss": 0.5704,
2124
  "step": 302
2125
  },
2126
  {
2127
  "epoch": 0.5474254742547425,
2128
- "grad_norm": 0.22637703359544134,
2129
  "learning_rate": 1.3968576935042864e-05,
2130
- "loss": 0.5517,
2131
  "step": 303
2132
  },
2133
  {
2134
  "epoch": 0.5492321589882565,
2135
- "grad_norm": 0.21912871932215822,
2136
  "learning_rate": 1.3879048596203637e-05,
2137
- "loss": 0.5468,
2138
  "step": 304
2139
  },
2140
  {
2141
  "epoch": 0.5510388437217706,
2142
- "grad_norm": 0.23076981598731436,
2143
  "learning_rate": 1.3789560396371742e-05,
2144
- "loss": 0.5536,
2145
  "step": 305
2146
  },
2147
  {
2148
  "epoch": 0.5528455284552846,
2149
- "grad_norm": 0.21396418531369818,
2150
  "learning_rate": 1.3700115539938995e-05,
2151
- "loss": 0.5512,
2152
  "step": 306
2153
  },
2154
  {
2155
  "epoch": 0.5546522131887985,
2156
- "grad_norm": 0.2373863943495516,
2157
  "learning_rate": 1.3610717229745183e-05,
2158
- "loss": 0.568,
2159
  "step": 307
2160
  },
2161
  {
2162
  "epoch": 0.5564588979223125,
2163
- "grad_norm": 0.21450533643732628,
2164
  "learning_rate": 1.3521368666963348e-05,
2165
- "loss": 0.5475,
2166
  "step": 308
2167
  },
2168
  {
2169
  "epoch": 0.5582655826558266,
2170
- "grad_norm": 0.22267671286449942,
2171
  "learning_rate": 1.3432073050985201e-05,
2172
- "loss": 0.5571,
2173
  "step": 309
2174
  },
2175
  {
2176
  "epoch": 0.5600722673893406,
2177
- "grad_norm": 0.21198421182222235,
2178
  "learning_rate": 1.3342833579306511e-05,
2179
- "loss": 0.5396,
2180
  "step": 310
2181
  },
2182
  {
2183
  "epoch": 0.5618789521228545,
2184
- "grad_norm": 0.20922310954929335,
2185
  "learning_rate": 1.3253653447412658e-05,
2186
- "loss": 0.5586,
2187
  "step": 311
2188
  },
2189
  {
2190
  "epoch": 0.5636856368563685,
2191
- "grad_norm": 0.2814394272292605,
2192
  "learning_rate": 1.3164535848664167e-05,
2193
- "loss": 0.5576,
2194
  "step": 312
2195
  },
2196
  {
2197
  "epoch": 0.5654923215898826,
2198
- "grad_norm": 0.20840853463224346,
2199
  "learning_rate": 1.3075483974182368e-05,
2200
- "loss": 0.5587,
2201
  "step": 313
2202
  },
2203
  {
2204
  "epoch": 0.5672990063233966,
2205
- "grad_norm": 0.19543507039118269,
2206
  "learning_rate": 1.2986501012735174e-05,
2207
- "loss": 0.5512,
2208
  "step": 314
2209
  },
2210
  {
2211
  "epoch": 0.5691056910569106,
2212
- "grad_norm": 0.20718793275357836,
2213
  "learning_rate": 1.2897590150622812e-05,
2214
- "loss": 0.5518,
2215
  "step": 315
2216
  },
2217
  {
2218
  "epoch": 0.5709123757904245,
2219
- "grad_norm": 0.20671416767052964,
2220
  "learning_rate": 1.2808754571563827e-05,
2221
- "loss": 0.548,
2222
  "step": 316
2223
  },
2224
  {
2225
  "epoch": 0.5727190605239386,
2226
- "grad_norm": 0.21397949847399053,
2227
  "learning_rate": 1.2719997456580993e-05,
2228
- "loss": 0.5435,
2229
  "step": 317
2230
  },
2231
  {
2232
  "epoch": 0.5745257452574526,
2233
- "grad_norm": 0.21927216118410173,
2234
  "learning_rate": 1.2631321983887464e-05,
2235
- "loss": 0.5558,
2236
  "step": 318
2237
  },
2238
  {
2239
  "epoch": 0.5763324299909666,
2240
- "grad_norm": 0.2259801782500092,
2241
  "learning_rate": 1.2542731328772936e-05,
2242
- "loss": 0.5652,
2243
  "step": 319
2244
  },
2245
  {
2246
  "epoch": 0.5781391147244805,
2247
- "grad_norm": 0.23327651839440386,
2248
  "learning_rate": 1.2454228663489967e-05,
2249
- "loss": 0.5473,
2250
  "step": 320
2251
  },
2252
  {
2253
  "epoch": 0.5799457994579946,
2254
- "grad_norm": 0.21590518161476593,
2255
  "learning_rate": 1.2365817157140373e-05,
2256
- "loss": 0.5443,
2257
  "step": 321
2258
  },
2259
  {
2260
  "epoch": 0.5817524841915086,
2261
- "grad_norm": 0.2160452415449525,
2262
  "learning_rate": 1.2277499975561755e-05,
2263
- "loss": 0.5656,
2264
  "step": 322
2265
  },
2266
  {
2267
  "epoch": 0.5835591689250226,
2268
- "grad_norm": 0.2107872282650075,
2269
  "learning_rate": 1.2189280281214128e-05,
2270
- "loss": 0.5514,
2271
  "step": 323
2272
  },
2273
  {
2274
  "epoch": 0.5853658536585366,
2275
- "grad_norm": 0.2224759639805539,
2276
  "learning_rate": 1.2101161233066704e-05,
2277
- "loss": 0.5323,
2278
  "step": 324
2279
  },
2280
  {
2281
  "epoch": 0.5871725383920506,
2282
- "grad_norm": 0.21533052575699613,
2283
  "learning_rate": 1.201314598648473e-05,
2284
- "loss": 0.5501,
2285
  "step": 325
2286
  },
2287
  {
2288
  "epoch": 0.5889792231255646,
2289
- "grad_norm": 0.2148695807483304,
2290
  "learning_rate": 1.1925237693116546e-05,
2291
- "loss": 0.5591,
2292
  "step": 326
2293
  },
2294
  {
2295
  "epoch": 0.5907859078590786,
2296
- "grad_norm": 0.21593841879832285,
2297
  "learning_rate": 1.1837439500780718e-05,
2298
- "loss": 0.5458,
2299
  "step": 327
2300
  },
2301
  {
2302
  "epoch": 0.5925925925925926,
2303
- "grad_norm": 0.24020104607554493,
2304
  "learning_rate": 1.1749754553353297e-05,
2305
- "loss": 0.5521,
2306
  "step": 328
2307
  },
2308
  {
2309
  "epoch": 0.5943992773261066,
2310
- "grad_norm": 0.21539575179480944,
2311
  "learning_rate": 1.1662185990655285e-05,
2312
- "loss": 0.5421,
2313
  "step": 329
2314
  },
2315
  {
2316
  "epoch": 0.5962059620596206,
2317
- "grad_norm": 0.21296303994341217,
2318
  "learning_rate": 1.1574736948340163e-05,
2319
- "loss": 0.5494,
2320
  "step": 330
2321
  },
2322
  {
2323
  "epoch": 0.5980126467931346,
2324
- "grad_norm": 0.2137280499573947,
2325
  "learning_rate": 1.1487410557781642e-05,
2326
- "loss": 0.5534,
2327
  "step": 331
2328
  },
2329
  {
2330
  "epoch": 0.5998193315266486,
2331
- "grad_norm": 0.21337396479946083,
2332
  "learning_rate": 1.140020994596151e-05,
2333
- "loss": 0.5373,
2334
  "step": 332
2335
  },
2336
  {
2337
  "epoch": 0.6016260162601627,
2338
- "grad_norm": 0.20423590394338298,
2339
  "learning_rate": 1.1313138235357683e-05,
2340
- "loss": 0.5579,
2341
  "step": 333
2342
  },
2343
  {
2344
  "epoch": 0.6034327009936766,
2345
- "grad_norm": 0.2098057492573464,
2346
  "learning_rate": 1.1226198543832373e-05,
2347
- "loss": 0.5702,
2348
  "step": 334
2349
  },
2350
  {
2351
  "epoch": 0.6052393857271906,
2352
- "grad_norm": 0.223601029063386,
2353
  "learning_rate": 1.1139393984520467e-05,
2354
- "loss": 0.5531,
2355
  "step": 335
2356
  },
2357
  {
2358
  "epoch": 0.6070460704607046,
2359
- "grad_norm": 0.20481524169643586,
2360
  "learning_rate": 1.1052727665718027e-05,
2361
- "loss": 0.5501,
2362
  "step": 336
2363
  },
2364
  {
2365
  "epoch": 0.6088527551942186,
2366
- "grad_norm": 0.21503604717197333,
2367
  "learning_rate": 1.0966202690771015e-05,
2368
- "loss": 0.5516,
2369
  "step": 337
2370
  },
2371
  {
2372
  "epoch": 0.6106594399277326,
2373
- "grad_norm": 0.2313465130277773,
2374
  "learning_rate": 1.087982215796415e-05,
2375
- "loss": 0.5477,
2376
  "step": 338
2377
  },
2378
  {
2379
  "epoch": 0.6124661246612466,
2380
- "grad_norm": 0.2110290437582214,
2381
  "learning_rate": 1.079358916040996e-05,
2382
- "loss": 0.5561,
2383
  "step": 339
2384
  },
2385
  {
2386
  "epoch": 0.6142728093947606,
2387
- "grad_norm": 0.20253428502981877,
2388
  "learning_rate": 1.0707506785938052e-05,
2389
- "loss": 0.551,
2390
  "step": 340
2391
  },
2392
  {
2393
  "epoch": 0.6160794941282746,
2394
- "grad_norm": 0.1913919297696134,
2395
  "learning_rate": 1.0621578116984501e-05,
2396
- "loss": 0.55,
2397
  "step": 341
2398
  },
2399
  {
2400
  "epoch": 0.6178861788617886,
2401
- "grad_norm": 0.21195082036495064,
2402
  "learning_rate": 1.0535806230481525e-05,
2403
- "loss": 0.5527,
2404
  "step": 342
2405
  },
2406
  {
2407
  "epoch": 0.6196928635953026,
2408
- "grad_norm": 0.20565633004943384,
2409
  "learning_rate": 1.0450194197747252e-05,
2410
- "loss": 0.5474,
2411
  "step": 343
2412
  },
2413
  {
2414
  "epoch": 0.6214995483288166,
2415
- "grad_norm": 0.19638372021410463,
2416
  "learning_rate": 1.036474508437579e-05,
2417
- "loss": 0.5368,
2418
  "step": 344
2419
  },
2420
  {
2421
  "epoch": 0.6233062330623306,
2422
- "grad_norm": 0.20315514847179497,
2423
  "learning_rate": 1.0279461950127422e-05,
2424
- "loss": 0.5508,
2425
  "step": 345
2426
  },
2427
  {
2428
  "epoch": 0.6251129177958447,
2429
- "grad_norm": 0.2076415773455686,
2430
  "learning_rate": 1.0194347848819055e-05,
2431
- "loss": 0.5526,
2432
  "step": 346
2433
  },
2434
  {
2435
  "epoch": 0.6269196025293586,
2436
- "grad_norm": 0.2011750948359859,
2437
  "learning_rate": 1.0109405828214872e-05,
2438
- "loss": 0.5385,
2439
  "step": 347
2440
  },
2441
  {
2442
  "epoch": 0.6287262872628726,
2443
- "grad_norm": 0.20681479000332276,
2444
  "learning_rate": 1.0024638929917192e-05,
2445
- "loss": 0.5391,
2446
  "step": 348
2447
  },
2448
  {
2449
  "epoch": 0.6305329719963866,
2450
- "grad_norm": 0.20616836763128274,
2451
  "learning_rate": 9.940050189257552e-06,
2452
- "loss": 0.5428,
2453
  "step": 349
2454
  },
2455
  {
2456
  "epoch": 0.6323396567299007,
2457
- "grad_norm": 0.19586568244011807,
2458
  "learning_rate": 9.85564263518803e-06,
2459
- "loss": 0.555,
2460
  "step": 350
2461
  },
2462
  {
2463
  "epoch": 0.6341463414634146,
2464
- "grad_norm": 0.20202959361534048,
2465
  "learning_rate": 9.771419290172776e-06,
2466
- "loss": 0.5444,
2467
  "step": 351
2468
  },
2469
  {
2470
  "epoch": 0.6359530261969286,
2471
- "grad_norm": 0.20451264583805118,
2472
  "learning_rate": 9.68738317007978e-06,
2473
- "loss": 0.5579,
2474
  "step": 352
2475
  },
2476
  {
2477
  "epoch": 0.6377597109304426,
2478
- "grad_norm": 0.21373282488633522,
2479
  "learning_rate": 9.6035372840729e-06,
2480
- "loss": 0.5585,
2481
  "step": 353
2482
  },
2483
  {
2484
  "epoch": 0.6395663956639567,
2485
- "grad_norm": 0.20704885874511708,
2486
  "learning_rate": 9.519884634504074e-06,
2487
- "loss": 0.5535,
2488
  "step": 354
2489
  },
2490
  {
2491
  "epoch": 0.6413730803974707,
2492
- "grad_norm": 0.19730712728411348,
2493
  "learning_rate": 9.436428216805862e-06,
2494
- "loss": 0.5423,
2495
  "step": 355
2496
  },
2497
  {
2498
  "epoch": 0.6431797651309846,
2499
- "grad_norm": 0.1852944339743618,
2500
  "learning_rate": 9.35317101938412e-06,
2501
- "loss": 0.5561,
2502
  "step": 356
2503
  },
2504
  {
2505
  "epoch": 0.6449864498644986,
2506
- "grad_norm": 0.1993128994483299,
2507
  "learning_rate": 9.270116023511063e-06,
2508
- "loss": 0.5574,
2509
  "step": 357
2510
  },
2511
  {
2512
  "epoch": 0.6467931345980127,
2513
- "grad_norm": 0.21880170033439258,
2514
  "learning_rate": 9.187266203218457e-06,
2515
- "loss": 0.5632,
2516
  "step": 358
2517
  },
2518
  {
2519
  "epoch": 0.6485998193315267,
2520
- "grad_norm": 0.2099893002209851,
2521
  "learning_rate": 9.104624525191147e-06,
2522
- "loss": 0.5365,
2523
  "step": 359
2524
  },
2525
  {
2526
  "epoch": 0.6504065040650406,
2527
- "grad_norm": 0.18557648513272196,
2528
  "learning_rate": 9.022193948660825e-06,
2529
- "loss": 0.545,
2530
  "step": 360
2531
  },
2532
  {
2533
  "epoch": 0.6522131887985546,
2534
- "grad_norm": 0.20969372987373008,
2535
  "learning_rate": 8.939977425300072e-06,
2536
- "loss": 0.55,
2537
  "step": 361
2538
  },
2539
  {
2540
  "epoch": 0.6540198735320687,
2541
- "grad_norm": 0.19508780289135563,
2542
  "learning_rate": 8.857977899116645e-06,
2543
- "loss": 0.5616,
2544
  "step": 362
2545
  },
2546
  {
2547
  "epoch": 0.6558265582655827,
2548
- "grad_norm": 0.2307324113893453,
2549
  "learning_rate": 8.776198306348072e-06,
2550
- "loss": 0.5522,
2551
  "step": 363
2552
  },
2553
  {
2554
  "epoch": 0.6576332429990966,
2555
- "grad_norm": 0.20707494383230668,
2556
  "learning_rate": 8.69464157535652e-06,
2557
- "loss": 0.5519,
2558
  "step": 364
2559
  },
2560
  {
2561
  "epoch": 0.6594399277326106,
2562
- "grad_norm": 0.19165357639913966,
2563
  "learning_rate": 8.61331062652391e-06,
2564
- "loss": 0.5435,
2565
  "step": 365
2566
  },
2567
  {
2568
  "epoch": 0.6612466124661247,
2569
- "grad_norm": 0.24674203313821708,
2570
  "learning_rate": 8.532208372147376e-06,
2571
- "loss": 0.5643,
2572
  "step": 366
2573
  },
2574
  {
2575
  "epoch": 0.6630532971996387,
2576
- "grad_norm": 0.19774419729169798,
2577
  "learning_rate": 8.451337716334953e-06,
2578
- "loss": 0.5462,
2579
  "step": 367
2580
  },
2581
  {
2582
  "epoch": 0.6648599819331527,
2583
- "grad_norm": 0.8597063719073577,
2584
  "learning_rate": 8.370701554901603e-06,
2585
- "loss": 0.5577,
2586
  "step": 368
2587
  },
2588
  {
2589
  "epoch": 0.6666666666666666,
2590
- "grad_norm": 0.2040142889228323,
2591
  "learning_rate": 8.290302775265509e-06,
2592
- "loss": 0.5585,
2593
  "step": 369
2594
  },
2595
  {
2596
  "epoch": 0.6684733514001807,
2597
- "grad_norm": 0.19512815474532405,
2598
  "learning_rate": 8.210144256344714e-06,
2599
- "loss": 0.557,
2600
  "step": 370
2601
  },
2602
  {
2603
  "epoch": 0.6702800361336947,
2604
- "grad_norm": 0.21244446703252592,
2605
  "learning_rate": 8.130228868453989e-06,
2606
- "loss": 0.5451,
2607
  "step": 371
2608
  },
2609
  {
2610
  "epoch": 0.6720867208672087,
2611
- "grad_norm": 0.1995166486722308,
2612
  "learning_rate": 8.050559473202078e-06,
2613
- "loss": 0.5623,
2614
  "step": 372
2615
  },
2616
  {
2617
  "epoch": 0.6738934056007226,
2618
- "grad_norm": 0.20175012832244793,
2619
  "learning_rate": 7.971138923389221e-06,
2620
- "loss": 0.5483,
2621
  "step": 373
2622
  },
2623
  {
2624
  "epoch": 0.6757000903342367,
2625
- "grad_norm": 0.2003498800103412,
2626
  "learning_rate": 7.89197006290502e-06,
2627
- "loss": 0.5384,
2628
  "step": 374
2629
  },
2630
  {
2631
  "epoch": 0.6775067750677507,
2632
- "grad_norm": 0.20139295383264785,
2633
  "learning_rate": 7.813055726626579e-06,
2634
- "loss": 0.5525,
2635
  "step": 375
2636
  },
2637
  {
2638
  "epoch": 0.6793134598012647,
2639
- "grad_norm": 0.2240479312770787,
2640
  "learning_rate": 7.734398740316999e-06,
2641
- "loss": 0.5375,
2642
  "step": 376
2643
  },
2644
  {
2645
  "epoch": 0.6811201445347786,
2646
- "grad_norm": 0.20393366594433568,
2647
  "learning_rate": 7.65600192052422e-06,
2648
- "loss": 0.5536,
2649
  "step": 377
2650
  },
2651
  {
2652
  "epoch": 0.6829268292682927,
2653
- "grad_norm": 0.20543533342894374,
2654
  "learning_rate": 7.577868074480107e-06,
2655
- "loss": 0.5543,
2656
  "step": 378
2657
  },
2658
  {
2659
  "epoch": 0.6847335140018067,
2660
- "grad_norm": 0.20996728435530929,
2661
  "learning_rate": 7.500000000000004e-06,
2662
- "loss": 0.5573,
2663
  "step": 379
2664
  },
2665
  {
2666
  "epoch": 0.6865401987353207,
2667
- "grad_norm": 0.19784358219258258,
2668
  "learning_rate": 7.422400485382484e-06,
2669
- "loss": 0.5382,
2670
  "step": 380
2671
  },
2672
  {
2673
  "epoch": 0.6883468834688347,
2674
- "grad_norm": 0.20267397640041715,
2675
  "learning_rate": 7.345072309309557e-06,
2676
- "loss": 0.5544,
2677
  "step": 381
2678
  },
2679
  {
2680
  "epoch": 0.6901535682023487,
2681
- "grad_norm": 0.20633896625792722,
2682
  "learning_rate": 7.268018240747129e-06,
2683
- "loss": 0.5418,
2684
  "step": 382
2685
  },
2686
  {
2687
  "epoch": 0.6919602529358627,
2688
- "grad_norm": 0.2073228668097345,
2689
  "learning_rate": 7.191241038845876e-06,
2690
- "loss": 0.5607,
2691
  "step": 383
2692
  },
2693
  {
2694
  "epoch": 0.6937669376693767,
2695
- "grad_norm": 0.2001643570352802,
2696
  "learning_rate": 7.114743452842428e-06,
2697
- "loss": 0.5581,
2698
  "step": 384
2699
  },
2700
  {
2701
  "epoch": 0.6955736224028907,
2702
- "grad_norm": 0.18173924874412256,
2703
  "learning_rate": 7.038528221960959e-06,
2704
- "loss": 0.5409,
2705
  "step": 385
2706
  },
2707
  {
2708
  "epoch": 0.6973803071364046,
2709
- "grad_norm": 0.1894607594536073,
2710
  "learning_rate": 6.962598075315047e-06,
2711
- "loss": 0.5523,
2712
  "step": 386
2713
  },
2714
  {
2715
  "epoch": 0.6991869918699187,
2716
- "grad_norm": 0.21893519782032744,
2717
  "learning_rate": 6.886955731810009e-06,
2718
- "loss": 0.5526,
2719
  "step": 387
2720
  },
2721
  {
2722
  "epoch": 0.7009936766034327,
2723
- "grad_norm": 0.20060107223389628,
2724
  "learning_rate": 6.811603900045477e-06,
2725
- "loss": 0.5514,
2726
  "step": 388
2727
  },
2728
  {
2729
  "epoch": 0.7028003613369467,
2730
- "grad_norm": 0.1939348717979157,
2731
  "learning_rate": 6.736545278218464e-06,
2732
- "loss": 0.554,
2733
  "step": 389
2734
  },
2735
  {
2736
  "epoch": 0.7046070460704607,
2737
- "grad_norm": 0.18811319599750992,
2738
  "learning_rate": 6.661782554026735e-06,
2739
- "loss": 0.5486,
2740
  "step": 390
2741
  },
2742
  {
2743
  "epoch": 0.7064137308039747,
2744
- "grad_norm": 0.1999069584795676,
2745
  "learning_rate": 6.587318404572537e-06,
2746
- "loss": 0.5481,
2747
  "step": 391
2748
  },
2749
  {
2750
  "epoch": 0.7082204155374887,
2751
- "grad_norm": 0.1943901355351736,
2752
  "learning_rate": 6.513155496266783e-06,
2753
- "loss": 0.5545,
2754
  "step": 392
2755
  },
2756
  {
2757
  "epoch": 0.7100271002710027,
2758
- "grad_norm": 0.19410019462805625,
2759
  "learning_rate": 6.439296484733526e-06,
2760
- "loss": 0.559,
2761
  "step": 393
2762
  },
2763
  {
2764
  "epoch": 0.7118337850045167,
2765
- "grad_norm": 0.19155280625900392,
2766
  "learning_rate": 6.3657440147149e-06,
2767
- "loss": 0.5615,
2768
  "step": 394
2769
  },
2770
  {
2771
  "epoch": 0.7136404697380307,
2772
- "grad_norm": 0.18053182884872418,
2773
  "learning_rate": 6.292500719976397e-06,
2774
- "loss": 0.5559,
2775
  "step": 395
2776
  },
2777
  {
2778
  "epoch": 0.7154471544715447,
2779
- "grad_norm": 0.18049395484657527,
2780
  "learning_rate": 6.219569223212585e-06,
2781
- "loss": 0.5506,
2782
  "step": 396
2783
  },
2784
  {
2785
  "epoch": 0.7172538392050587,
2786
- "grad_norm": 0.19679369568259336,
2787
  "learning_rate": 6.146952135953158e-06,
2788
- "loss": 0.5503,
2789
  "step": 397
2790
  },
2791
  {
2792
  "epoch": 0.7190605239385727,
2793
- "grad_norm": 0.18703524060865126,
2794
  "learning_rate": 6.074652058469447e-06,
2795
- "loss": 0.5514,
2796
  "step": 398
2797
  },
2798
  {
2799
  "epoch": 0.7208672086720868,
2800
- "grad_norm": 0.19600871335070388,
2801
  "learning_rate": 6.0026715796812945e-06,
2802
- "loss": 0.5538,
2803
  "step": 399
2804
  },
2805
  {
2806
  "epoch": 0.7226738934056007,
2807
- "grad_norm": 0.21866086120474482,
2808
  "learning_rate": 5.931013277064377e-06,
2809
- "loss": 0.5513,
2810
  "step": 400
2811
  }
2812
  ],
 
11
  "log_history": [
12
  {
13
  "epoch": 0.0018066847335140017,
14
+ "grad_norm": 2.6597820350894357,
15
  "learning_rate": 0.0,
16
  "loss": 0.8251,
17
  "step": 1
18
  },
19
  {
20
  "epoch": 0.0036133694670280035,
21
+ "grad_norm": 2.7266510344811494,
22
  "learning_rate": 1.0714285714285714e-06,
23
  "loss": 0.8284,
24
  "step": 2
25
  },
26
  {
27
  "epoch": 0.005420054200542005,
28
+ "grad_norm": 2.7200407943823968,
29
  "learning_rate": 2.1428571428571427e-06,
30
+ "loss": 0.8429,
31
  "step": 3
32
  },
33
  {
34
  "epoch": 0.007226738934056007,
35
+ "grad_norm": 2.6398267415203076,
36
  "learning_rate": 3.2142857142857143e-06,
37
+ "loss": 0.8381,
38
  "step": 4
39
  },
40
  {
41
  "epoch": 0.009033423667570008,
42
+ "grad_norm": 2.450058379558589,
43
  "learning_rate": 4.2857142857142855e-06,
44
+ "loss": 0.8245,
45
  "step": 5
46
  },
47
  {
48
  "epoch": 0.01084010840108401,
49
+ "grad_norm": 1.9536840033460428,
50
  "learning_rate": 5.357142857142857e-06,
51
+ "loss": 0.814,
52
  "step": 6
53
  },
54
  {
55
  "epoch": 0.012646793134598013,
56
+ "grad_norm": 1.2180340367347295,
57
  "learning_rate": 6.428571428571429e-06,
58
+ "loss": 0.7845,
59
  "step": 7
60
  },
61
  {
62
  "epoch": 0.014453477868112014,
63
+ "grad_norm": 1.1514973199014618,
64
  "learning_rate": 7.5e-06,
65
+ "loss": 0.7874,
66
  "step": 8
67
  },
68
  {
69
  "epoch": 0.016260162601626018,
70
+ "grad_norm": 1.1427251731602237,
71
  "learning_rate": 8.571428571428571e-06,
72
+ "loss": 0.7793,
73
  "step": 9
74
  },
75
  {
76
  "epoch": 0.018066847335140017,
77
+ "grad_norm": 2.4537001853666673,
78
  "learning_rate": 9.642857142857144e-06,
79
+ "loss": 0.7624,
80
  "step": 10
81
  },
82
  {
83
  "epoch": 0.01987353206865402,
84
+ "grad_norm": 2.474412722550665,
85
  "learning_rate": 1.0714285714285714e-05,
86
+ "loss": 0.7491,
87
  "step": 11
88
  },
89
  {
90
  "epoch": 0.02168021680216802,
91
+ "grad_norm": 2.333456673377541,
92
  "learning_rate": 1.1785714285714286e-05,
93
+ "loss": 0.762,
94
  "step": 12
95
  },
96
  {
97
  "epoch": 0.023486901535682024,
98
+ "grad_norm": 1.6274020508818596,
99
  "learning_rate": 1.2857142857142857e-05,
100
+ "loss": 0.7368,
101
  "step": 13
102
  },
103
  {
104
  "epoch": 0.025293586269196026,
105
+ "grad_norm": 1.563869927563279,
106
  "learning_rate": 1.3928571428571429e-05,
107
+ "loss": 0.7378,
108
  "step": 14
109
  },
110
  {
111
  "epoch": 0.02710027100271003,
112
+ "grad_norm": 1.253626680340628,
113
  "learning_rate": 1.5e-05,
114
+ "loss": 0.7271,
115
  "step": 15
116
  },
117
  {
118
  "epoch": 0.028906955736224028,
119
+ "grad_norm": 0.9176330164857595,
120
  "learning_rate": 1.6071428571428572e-05,
121
+ "loss": 0.7131,
122
  "step": 16
123
  },
124
  {
125
  "epoch": 0.03071364046973803,
126
+ "grad_norm": 0.7185343991261351,
127
  "learning_rate": 1.7142857142857142e-05,
128
+ "loss": 0.7131,
129
  "step": 17
130
  },
131
  {
132
  "epoch": 0.032520325203252036,
133
+ "grad_norm": 0.7884609645105337,
134
  "learning_rate": 1.8214285714285712e-05,
135
+ "loss": 0.7045,
136
  "step": 18
137
  },
138
  {
139
  "epoch": 0.03432700993676603,
140
+ "grad_norm": 0.7399830971167167,
141
  "learning_rate": 1.928571428571429e-05,
142
+ "loss": 0.6995,
143
  "step": 19
144
  },
145
  {
146
  "epoch": 0.036133694670280034,
147
+ "grad_norm": 0.7376328500886337,
148
  "learning_rate": 2.0357142857142858e-05,
149
+ "loss": 0.6779,
150
  "step": 20
151
  },
152
  {
153
  "epoch": 0.037940379403794036,
154
+ "grad_norm": 0.6439873145292982,
155
  "learning_rate": 2.1428571428571428e-05,
156
+ "loss": 0.6834,
157
  "step": 21
158
  },
159
  {
160
  "epoch": 0.03974706413730804,
161
+ "grad_norm": 0.5790542088592184,
162
  "learning_rate": 2.25e-05,
163
+ "loss": 0.6721,
164
  "step": 22
165
  },
166
  {
167
  "epoch": 0.04155374887082204,
168
+ "grad_norm": 0.5191304250088061,
169
  "learning_rate": 2.357142857142857e-05,
170
+ "loss": 0.6707,
171
  "step": 23
172
  },
173
  {
174
  "epoch": 0.04336043360433604,
175
+ "grad_norm": 0.4488946670945292,
176
  "learning_rate": 2.464285714285714e-05,
177
+ "loss": 0.676,
178
  "step": 24
179
  },
180
  {
181
  "epoch": 0.045167118337850046,
182
+ "grad_norm": 0.4798647764154148,
183
  "learning_rate": 2.5714285714285714e-05,
184
+ "loss": 0.6585,
185
  "step": 25
186
  },
187
  {
188
  "epoch": 0.04697380307136405,
189
+ "grad_norm": 0.4204330566493729,
190
  "learning_rate": 2.6785714285714288e-05,
191
+ "loss": 0.673,
192
  "step": 26
193
  },
194
  {
195
  "epoch": 0.04878048780487805,
196
+ "grad_norm": 0.47130112684574876,
197
  "learning_rate": 2.7857142857142858e-05,
198
+ "loss": 0.6516,
199
  "step": 27
200
  },
201
  {
202
  "epoch": 0.05058717253839205,
203
+ "grad_norm": 0.4362815754533529,
204
  "learning_rate": 2.892857142857143e-05,
205
+ "loss": 0.6545,
206
  "step": 28
207
  },
208
  {
209
  "epoch": 0.052393857271906055,
210
+ "grad_norm": 0.43822648560890193,
211
  "learning_rate": 3e-05,
212
+ "loss": 0.6647,
213
  "step": 29
214
  },
215
  {
216
  "epoch": 0.05420054200542006,
217
+ "grad_norm": 0.3853311574389301,
218
  "learning_rate": 2.9999731440137413e-05,
219
+ "loss": 0.6644,
220
  "step": 30
221
  },
222
  {
223
  "epoch": 0.05600722673893405,
224
+ "grad_norm": 0.4221785516753897,
225
  "learning_rate": 2.9998925770166232e-05,
226
+ "loss": 0.6534,
227
  "step": 31
228
  },
229
  {
230
  "epoch": 0.057813911472448055,
231
+ "grad_norm": 0.3980570059327311,
232
  "learning_rate": 2.9997583018935875e-05,
233
+ "loss": 0.6416,
234
  "step": 32
235
  },
236
  {
237
  "epoch": 0.05962059620596206,
238
+ "grad_norm": 0.35571676272289743,
239
  "learning_rate": 2.9995703234527553e-05,
240
+ "loss": 0.6513,
241
  "step": 33
242
  },
243
  {
244
  "epoch": 0.06142728093947606,
245
+ "grad_norm": 0.38024957472516396,
246
  "learning_rate": 2.999328648425255e-05,
247
+ "loss": 0.6385,
248
  "step": 34
249
  },
250
  {
251
  "epoch": 0.06323396567299007,
252
+ "grad_norm": 0.2924858728569485,
253
  "learning_rate": 2.999033285464982e-05,
254
+ "loss": 0.644,
255
  "step": 35
256
  },
257
  {
258
  "epoch": 0.06504065040650407,
259
+ "grad_norm": 0.3482798299572746,
260
  "learning_rate": 2.9986842451482876e-05,
261
+ "loss": 0.6387,
262
  "step": 36
263
  },
264
  {
265
  "epoch": 0.06684733514001806,
266
+ "grad_norm": 0.3705337140725527,
267
  "learning_rate": 2.9982815399736008e-05,
268
+ "loss": 0.6432,
269
  "step": 37
270
  },
271
  {
272
  "epoch": 0.06865401987353206,
273
+ "grad_norm": 0.3364721163164359,
274
  "learning_rate": 2.9978251843609816e-05,
275
+ "loss": 0.6351,
276
  "step": 38
277
  },
278
  {
279
  "epoch": 0.07046070460704607,
280
+ "grad_norm": 0.3272631660096153,
281
  "learning_rate": 2.9973151946516027e-05,
282
+ "loss": 0.6364,
283
  "step": 39
284
  },
285
  {
286
  "epoch": 0.07226738934056007,
287
+ "grad_norm": 0.29708127282462954,
288
  "learning_rate": 2.996751589107167e-05,
289
+ "loss": 0.6252,
290
  "step": 40
291
  },
292
  {
293
  "epoch": 0.07407407407407407,
294
+ "grad_norm": 0.24921077158970448,
295
  "learning_rate": 2.9961343879092512e-05,
296
+ "loss": 0.6308,
297
  "step": 41
298
  },
299
  {
300
  "epoch": 0.07588075880758807,
301
+ "grad_norm": 0.3479569257927118,
302
  "learning_rate": 2.9954636131585845e-05,
303
+ "loss": 0.6234,
304
  "step": 42
305
  },
306
  {
307
  "epoch": 0.07768744354110207,
308
+ "grad_norm": 0.28917511643628596,
309
  "learning_rate": 2.9947392888742566e-05,
310
+ "loss": 0.6233,
311
  "step": 43
312
  },
313
  {
314
  "epoch": 0.07949412827461608,
315
+ "grad_norm": 0.34737822266968965,
316
  "learning_rate": 2.993961440992859e-05,
317
+ "loss": 0.6171,
318
  "step": 44
319
  },
320
  {
321
  "epoch": 0.08130081300813008,
322
+ "grad_norm": 0.3247689228102166,
323
  "learning_rate": 2.993130097367553e-05,
324
+ "loss": 0.6238,
325
  "step": 45
326
  },
327
  {
328
  "epoch": 0.08310749774164408,
329
+ "grad_norm": 0.30719902123424325,
330
  "learning_rate": 2.9922452877670775e-05,
331
+ "loss": 0.6252,
332
  "step": 46
333
  },
334
  {
335
  "epoch": 0.08491418247515808,
336
+ "grad_norm": 0.3175141153740525,
337
  "learning_rate": 2.991307043874677e-05,
338
+ "loss": 0.6281,
339
  "step": 47
340
  },
341
  {
342
  "epoch": 0.08672086720867209,
343
+ "grad_norm": 0.31244488771705314,
344
  "learning_rate": 2.9903153992869734e-05,
345
+ "loss": 0.6148,
346
  "step": 48
347
  },
348
  {
349
  "epoch": 0.08852755194218609,
350
+ "grad_norm": 0.3429131319420423,
351
  "learning_rate": 2.989270389512756e-05,
352
+ "loss": 0.6211,
353
  "step": 49
354
  },
355
  {
356
  "epoch": 0.09033423667570009,
357
+ "grad_norm": 0.3104062918234721,
358
  "learning_rate": 2.988172051971717e-05,
359
+ "loss": 0.6204,
360
  "step": 50
361
  },
362
  {
363
  "epoch": 0.0921409214092141,
364
+ "grad_norm": 0.31461103786584654,
365
  "learning_rate": 2.9870204259931062e-05,
366
+ "loss": 0.6097,
367
  "step": 51
368
  },
369
  {
370
  "epoch": 0.0939476061427281,
371
+ "grad_norm": 0.3151625720082357,
372
  "learning_rate": 2.9858155528143256e-05,
373
+ "loss": 0.6268,
374
  "step": 52
375
  },
376
  {
377
  "epoch": 0.0957542908762421,
378
+ "grad_norm": 0.2860126481493585,
379
  "learning_rate": 2.9845574755794522e-05,
380
+ "loss": 0.6125,
381
  "step": 53
382
  },
383
  {
384
  "epoch": 0.0975609756097561,
385
+ "grad_norm": 0.28841121592431124,
386
  "learning_rate": 2.9832462393376926e-05,
387
+ "loss": 0.6018,
388
  "step": 54
389
  },
390
  {
391
  "epoch": 0.0993676603432701,
392
+ "grad_norm": 0.23534181927847642,
393
  "learning_rate": 2.9818818910417706e-05,
394
+ "loss": 0.6169,
395
  "step": 55
396
  },
397
  {
398
  "epoch": 0.1011743450767841,
399
+ "grad_norm": 0.28661471046317555,
400
  "learning_rate": 2.9804644795462437e-05,
401
+ "loss": 0.6212,
402
  "step": 56
403
  },
404
  {
405
  "epoch": 0.10298102981029811,
406
+ "grad_norm": 0.2443857675436309,
407
  "learning_rate": 2.9789940556057574e-05,
408
+ "loss": 0.6065,
409
  "step": 57
410
  },
411
  {
412
  "epoch": 0.10478771454381211,
413
+ "grad_norm": 0.292418070185767,
414
  "learning_rate": 2.9774706718732255e-05,
415
+ "loss": 0.6152,
416
  "step": 58
417
  },
418
  {
419
  "epoch": 0.10659439927732611,
420
+ "grad_norm": 0.24742182455944323,
421
  "learning_rate": 2.9758943828979444e-05,
422
+ "loss": 0.6009,
423
  "step": 59
424
  },
425
  {
426
  "epoch": 0.10840108401084012,
427
+ "grad_norm": 0.25007211477600044,
428
  "learning_rate": 2.9742652451236414e-05,
429
+ "loss": 0.6023,
430
  "step": 60
431
  },
432
  {
433
  "epoch": 0.1102077687443541,
434
+ "grad_norm": 0.24539117288225054,
435
  "learning_rate": 2.972583316886451e-05,
436
+ "loss": 0.6059,
437
  "step": 61
438
  },
439
  {
440
  "epoch": 0.1120144534778681,
441
+ "grad_norm": 0.2544907417129908,
442
  "learning_rate": 2.9708486584128303e-05,
443
+ "loss": 0.6107,
444
  "step": 62
445
  },
446
  {
447
  "epoch": 0.11382113821138211,
448
+ "grad_norm": 0.25269030392560393,
449
  "learning_rate": 2.9690613318173966e-05,
450
+ "loss": 0.611,
451
  "step": 63
452
  },
453
  {
454
  "epoch": 0.11562782294489611,
455
+ "grad_norm": 0.20651691272156247,
456
  "learning_rate": 2.9672214011007087e-05,
457
+ "loss": 0.6091,
458
  "step": 64
459
  },
460
  {
461
  "epoch": 0.11743450767841011,
462
+ "grad_norm": 0.25759031036445135,
463
  "learning_rate": 2.9653289321469715e-05,
464
+ "loss": 0.6031,
465
  "step": 65
466
  },
467
  {
468
  "epoch": 0.11924119241192412,
469
+ "grad_norm": 0.27004948462624506,
470
  "learning_rate": 2.9633839927216793e-05,
471
+ "loss": 0.6061,
472
  "step": 66
473
  },
474
  {
475
  "epoch": 0.12104787714543812,
476
+ "grad_norm": 0.23899477664304233,
477
  "learning_rate": 2.9613866524691867e-05,
478
+ "loss": 0.5958,
479
  "step": 67
480
  },
481
  {
482
  "epoch": 0.12285456187895212,
483
+ "grad_norm": 0.3109356502387666,
484
  "learning_rate": 2.9593369829102173e-05,
485
+ "loss": 0.6119,
486
  "step": 68
487
  },
488
  {
489
  "epoch": 0.12466124661246612,
490
+ "grad_norm": 0.23973937288759506,
491
  "learning_rate": 2.957235057439301e-05,
492
+ "loss": 0.6135,
493
  "step": 69
494
  },
495
  {
496
  "epoch": 0.12646793134598014,
497
+ "grad_norm": 0.29402451520285133,
498
  "learning_rate": 2.955080951322147e-05,
499
+ "loss": 0.6178,
500
  "step": 70
501
  },
502
  {
503
  "epoch": 0.12827461607949414,
504
+ "grad_norm": 0.29085498534508775,
505
  "learning_rate": 2.9528747416929467e-05,
506
+ "loss": 0.5907,
507
  "step": 71
508
  },
509
  {
510
  "epoch": 0.13008130081300814,
511
+ "grad_norm": 0.25477514255751904,
512
  "learning_rate": 2.9506165075516148e-05,
513
+ "loss": 0.6073,
514
  "step": 72
515
  },
516
  {
517
  "epoch": 0.13188798554652212,
518
+ "grad_norm": 0.28366229733780235,
519
  "learning_rate": 2.9483063297609577e-05,
520
+ "loss": 0.6063,
521
  "step": 73
522
  },
523
  {
524
  "epoch": 0.13369467028003612,
525
+ "grad_norm": 0.26317649422081363,
526
  "learning_rate": 2.9459442910437798e-05,
527
+ "loss": 0.598,
528
  "step": 74
529
  },
530
  {
531
  "epoch": 0.13550135501355012,
532
+ "grad_norm": 0.28039599919104496,
533
  "learning_rate": 2.94353047597992e-05,
534
+ "loss": 0.5971,
535
  "step": 75
536
  },
537
  {
538
  "epoch": 0.13730803974706413,
539
+ "grad_norm": 0.271745198157929,
540
  "learning_rate": 2.941064971003224e-05,
541
+ "loss": 0.593,
542
  "step": 76
543
  },
544
  {
545
  "epoch": 0.13911472448057813,
546
+ "grad_norm": 0.24998502389958926,
547
  "learning_rate": 2.9385478643984484e-05,
548
+ "loss": 0.5878,
549
  "step": 77
550
  },
551
  {
552
  "epoch": 0.14092140921409213,
553
+ "grad_norm": 0.30322559816019845,
554
  "learning_rate": 2.9359792462981007e-05,
555
+ "loss": 0.599,
556
  "step": 78
557
  },
558
  {
559
  "epoch": 0.14272809394760613,
560
+ "grad_norm": 0.29351194680487536,
561
  "learning_rate": 2.9333592086792113e-05,
562
+ "loss": 0.6075,
563
  "step": 79
564
  },
565
  {
566
  "epoch": 0.14453477868112014,
567
+ "grad_norm": 0.2670107767188747,
568
  "learning_rate": 2.9306878453600382e-05,
569
+ "loss": 0.6161,
570
  "step": 80
571
  },
572
  {
573
  "epoch": 0.14634146341463414,
574
+ "grad_norm": 0.2927366904918854,
575
  "learning_rate": 2.9279652519967105e-05,
576
+ "loss": 0.6004,
577
  "step": 81
578
  },
579
  {
580
  "epoch": 0.14814814814814814,
581
+ "grad_norm": 0.28777141201555767,
582
  "learning_rate": 2.9251915260798024e-05,
583
+ "loss": 0.593,
584
  "step": 82
585
  },
586
  {
587
  "epoch": 0.14995483288166214,
588
+ "grad_norm": 0.282262421377073,
589
  "learning_rate": 2.9223667669308395e-05,
590
+ "loss": 0.6038,
591
  "step": 83
592
  },
593
  {
594
  "epoch": 0.15176151761517614,
595
+ "grad_norm": 0.24811154699761442,
596
  "learning_rate": 2.9194910756987464e-05,
597
+ "loss": 0.5976,
598
  "step": 84
599
  },
600
  {
601
  "epoch": 0.15356820234869015,
602
+ "grad_norm": 0.27670190484589513,
603
  "learning_rate": 2.9165645553562215e-05,
604
+ "loss": 0.5997,
605
  "step": 85
606
  },
607
  {
608
  "epoch": 0.15537488708220415,
609
+ "grad_norm": 0.24338487756299684,
610
  "learning_rate": 2.9135873106960525e-05,
611
+ "loss": 0.5994,
612
  "step": 86
613
  },
614
  {
615
  "epoch": 0.15718157181571815,
616
+ "grad_norm": 0.3040191021283788,
617
  "learning_rate": 2.9105594483273603e-05,
618
+ "loss": 0.6012,
619
  "step": 87
620
  },
621
  {
622
  "epoch": 0.15898825654923215,
623
+ "grad_norm": 0.26703924304748566,
624
  "learning_rate": 2.9074810766717865e-05,
625
+ "loss": 0.6058,
626
  "step": 88
627
  },
628
  {
629
  "epoch": 0.16079494128274616,
630
+ "grad_norm": 0.2694312104706441,
631
  "learning_rate": 2.904352305959606e-05,
632
+ "loss": 0.602,
633
  "step": 89
634
  },
635
  {
636
  "epoch": 0.16260162601626016,
637
+ "grad_norm": 0.29517928394037635,
638
  "learning_rate": 2.9011732482257835e-05,
639
+ "loss": 0.5944,
640
  "step": 90
641
  },
642
  {
643
  "epoch": 0.16440831074977416,
644
+ "grad_norm": 0.25118801401605734,
645
  "learning_rate": 2.89794401730596e-05,
646
+ "loss": 0.5923,
647
  "step": 91
648
  },
649
  {
650
  "epoch": 0.16621499548328816,
651
+ "grad_norm": 0.3348680034328015,
652
  "learning_rate": 2.894664728832377e-05,
653
+ "loss": 0.5958,
654
  "step": 92
655
  },
656
  {
657
  "epoch": 0.16802168021680217,
658
+ "grad_norm": 0.25945936645525364,
659
  "learning_rate": 2.8913355002297367e-05,
660
+ "loss": 0.6025,
661
  "step": 93
662
  },
663
  {
664
  "epoch": 0.16982836495031617,
665
+ "grad_norm": 0.31672412557275315,
666
  "learning_rate": 2.887956450710995e-05,
667
+ "loss": 0.602,
668
  "step": 94
669
  },
670
  {
671
  "epoch": 0.17163504968383017,
672
+ "grad_norm": 0.2631935513788724,
673
  "learning_rate": 2.8845277012730963e-05,
674
+ "loss": 0.5939,
675
  "step": 95
676
  },
677
  {
678
  "epoch": 0.17344173441734417,
679
+ "grad_norm": 0.28309967073656406,
680
  "learning_rate": 2.8810493746926364e-05,
681
+ "loss": 0.5932,
682
  "step": 96
683
  },
684
  {
685
  "epoch": 0.17524841915085818,
686
+ "grad_norm": 0.28939178950900774,
687
  "learning_rate": 2.87752159552147e-05,
688
+ "loss": 0.5993,
689
  "step": 97
690
  },
691
  {
692
  "epoch": 0.17705510388437218,
693
+ "grad_norm": 0.3021445789644071,
694
  "learning_rate": 2.87394449008225e-05,
695
+ "loss": 0.5924,
696
  "step": 98
697
  },
698
  {
699
  "epoch": 0.17886178861788618,
700
+ "grad_norm": 0.2648161341085879,
701
  "learning_rate": 2.8703181864639013e-05,
702
+ "loss": 0.6058,
703
  "step": 99
704
  },
705
  {
706
  "epoch": 0.18066847335140018,
707
+ "grad_norm": 0.28484263640616314,
708
  "learning_rate": 2.8666428145170385e-05,
709
+ "loss": 0.5789,
710
  "step": 100
711
  },
712
  {
713
  "epoch": 0.18247515808491419,
714
+ "grad_norm": 0.2611846886724786,
715
  "learning_rate": 2.8629185058493116e-05,
716
+ "loss": 0.5949,
717
  "step": 101
718
  },
719
  {
720
  "epoch": 0.1842818428184282,
721
+ "grad_norm": 0.33149916071305763,
722
  "learning_rate": 2.8591453938206985e-05,
723
+ "loss": 0.6134,
724
  "step": 102
725
  },
726
  {
727
  "epoch": 0.1860885275519422,
728
+ "grad_norm": 0.28402400837841096,
729
  "learning_rate": 2.8553236135387247e-05,
730
+ "loss": 0.5914,
731
  "step": 103
732
  },
733
  {
734
  "epoch": 0.1878952122854562,
735
+ "grad_norm": 0.28825320782588004,
736
  "learning_rate": 2.8514533018536286e-05,
737
+ "loss": 0.5897,
738
  "step": 104
739
  },
740
  {
741
  "epoch": 0.1897018970189702,
742
+ "grad_norm": 0.2901395812522203,
743
  "learning_rate": 2.8475345973534605e-05,
744
+ "loss": 0.5822,
745
  "step": 105
746
  },
747
  {
748
  "epoch": 0.1915085817524842,
749
+ "grad_norm": 0.25034051990006584,
750
  "learning_rate": 2.8435676403591193e-05,
751
+ "loss": 0.5846,
752
  "step": 106
753
  },
754
  {
755
  "epoch": 0.1933152664859982,
756
+ "grad_norm": 0.27868357694305945,
757
  "learning_rate": 2.8395525729193284e-05,
758
+ "loss": 0.5907,
759
  "step": 107
760
  },
761
  {
762
  "epoch": 0.1951219512195122,
763
+ "grad_norm": 0.257742952703987,
764
  "learning_rate": 2.835489538805548e-05,
765
+ "loss": 0.5888,
766
  "step": 108
767
  },
768
  {
769
  "epoch": 0.1969286359530262,
770
+ "grad_norm": 0.26397326166674545,
771
  "learning_rate": 2.8313786835068314e-05,
772
+ "loss": 0.5976,
773
  "step": 109
774
  },
775
  {
776
  "epoch": 0.1987353206865402,
777
+ "grad_norm": 0.2721484510112116,
778
  "learning_rate": 2.8272201542246077e-05,
779
+ "loss": 0.5792,
780
  "step": 110
781
  },
782
  {
783
  "epoch": 0.2005420054200542,
784
+ "grad_norm": 0.2374425057236497,
785
  "learning_rate": 2.8230140998674185e-05,
786
+ "loss": 0.599,
787
  "step": 111
788
  },
789
  {
790
  "epoch": 0.2023486901535682,
791
+ "grad_norm": 0.24242218904343407,
792
  "learning_rate": 2.8187606710455807e-05,
793
+ "loss": 0.5921,
794
  "step": 112
795
  },
796
  {
797
  "epoch": 0.2041553748870822,
798
+ "grad_norm": 0.2742052397132371,
799
  "learning_rate": 2.8144600200657953e-05,
800
+ "loss": 0.5965,
801
  "step": 113
802
  },
803
  {
804
  "epoch": 0.20596205962059622,
805
+ "grad_norm": 0.2240931438132582,
806
  "learning_rate": 2.8101123009256946e-05,
807
+ "loss": 0.5792,
808
  "step": 114
809
  },
810
  {
811
  "epoch": 0.20776874435411022,
812
+ "grad_norm": 0.2763456467888191,
813
  "learning_rate": 2.8057176693083253e-05,
814
+ "loss": 0.5958,
815
  "step": 115
816
  },
817
  {
818
  "epoch": 0.20957542908762422,
819
+ "grad_norm": 0.2435434093260049,
820
  "learning_rate": 2.8012762825765763e-05,
821
+ "loss": 0.5891,
822
  "step": 116
823
  },
824
  {
825
  "epoch": 0.21138211382113822,
826
+ "grad_norm": 0.2769370185736096,
827
  "learning_rate": 2.7967882997675424e-05,
828
+ "loss": 0.6012,
829
  "step": 117
830
  },
831
  {
832
  "epoch": 0.21318879855465223,
833
+ "grad_norm": 0.2556991024086139,
834
  "learning_rate": 2.7922538815868287e-05,
835
+ "loss": 0.5879,
836
  "step": 118
837
  },
838
  {
839
  "epoch": 0.21499548328816623,
840
+ "grad_norm": 0.2913165244476914,
841
  "learning_rate": 2.7876731904027994e-05,
842
+ "loss": 0.5784,
843
  "step": 119
844
  },
845
  {
846
  "epoch": 0.21680216802168023,
847
+ "grad_norm": 0.29037318268167045,
848
  "learning_rate": 2.78304639024076e-05,
849
+ "loss": 0.5968,
850
  "step": 120
851
  },
852
  {
853
  "epoch": 0.2186088527551942,
854
+ "grad_norm": 0.24340049924612678,
855
  "learning_rate": 2.7783736467770863e-05,
856
+ "loss": 0.5927,
857
  "step": 121
858
  },
859
  {
860
  "epoch": 0.2204155374887082,
861
+ "grad_norm": 0.29825692316659325,
862
  "learning_rate": 2.7736551273332908e-05,
863
+ "loss": 0.5972,
864
  "step": 122
865
  },
866
  {
867
  "epoch": 0.2222222222222222,
868
+ "grad_norm": 0.2593832259683122,
869
  "learning_rate": 2.7688910008700305e-05,
870
+ "loss": 0.5876,
871
  "step": 123
872
  },
873
  {
874
  "epoch": 0.2240289069557362,
875
+ "grad_norm": 0.27547771011759425,
876
  "learning_rate": 2.764081437981059e-05,
877
+ "loss": 0.5742,
878
  "step": 124
879
  },
880
  {
881
  "epoch": 0.22583559168925021,
882
+ "grad_norm": 0.27158658907185634,
883
  "learning_rate": 2.7592266108871158e-05,
884
+ "loss": 0.5812,
885
  "step": 125
886
  },
887
  {
888
  "epoch": 0.22764227642276422,
889
+ "grad_norm": 0.2715497634460973,
890
  "learning_rate": 2.754326693429761e-05,
891
+ "loss": 0.5931,
892
  "step": 126
893
  },
894
  {
895
  "epoch": 0.22944896115627822,
896
+ "grad_norm": 0.24142985625388147,
897
  "learning_rate": 2.7493818610651493e-05,
898
+ "loss": 0.5859,
899
  "step": 127
900
  },
901
  {
902
  "epoch": 0.23125564588979222,
903
+ "grad_norm": 0.2530474955317245,
904
  "learning_rate": 2.744392290857747e-05,
905
+ "loss": 0.5857,
906
  "step": 128
907
  },
908
  {
909
  "epoch": 0.23306233062330622,
910
+ "grad_norm": 0.2567731311347638,
911
  "learning_rate": 2.7393581614739924e-05,
912
+ "loss": 0.5803,
913
  "step": 129
914
  },
915
  {
916
  "epoch": 0.23486901535682023,
917
+ "grad_norm": 0.2628073110459229,
918
  "learning_rate": 2.7342796531758984e-05,
919
+ "loss": 0.5987,
920
  "step": 130
921
  },
922
  {
923
  "epoch": 0.23667570009033423,
924
+ "grad_norm": 0.23510376935644756,
925
  "learning_rate": 2.729156947814598e-05,
926
+ "loss": 0.5926,
927
  "step": 131
928
  },
929
  {
930
  "epoch": 0.23848238482384823,
931
+ "grad_norm": 0.23576954058784624,
932
  "learning_rate": 2.7239902288238297e-05,
933
+ "loss": 0.5721,
934
  "step": 132
935
  },
936
  {
937
  "epoch": 0.24028906955736223,
938
+ "grad_norm": 0.24964231366212492,
939
  "learning_rate": 2.7187796812133733e-05,
940
+ "loss": 0.5873,
941
  "step": 133
942
  },
943
  {
944
  "epoch": 0.24209575429087624,
945
+ "grad_norm": 0.23911643013915115,
946
  "learning_rate": 2.7135254915624213e-05,
947
+ "loss": 0.5878,
948
  "step": 134
949
  },
950
  {
951
  "epoch": 0.24390243902439024,
952
+ "grad_norm": 0.23601189225311667,
953
  "learning_rate": 2.708227848012901e-05,
954
+ "loss": 0.585,
955
  "step": 135
956
  },
957
  {
958
  "epoch": 0.24570912375790424,
959
+ "grad_norm": 0.28061149806949875,
960
  "learning_rate": 2.7028869402627357e-05,
961
+ "loss": 0.5889,
962
  "step": 136
963
  },
964
  {
965
  "epoch": 0.24751580849141824,
966
+ "grad_norm": 0.2537605160128264,
967
  "learning_rate": 2.6975029595590523e-05,
968
+ "loss": 0.5726,
969
  "step": 137
970
  },
971
  {
972
  "epoch": 0.24932249322493225,
973
+ "grad_norm": 0.2631279788827005,
974
  "learning_rate": 2.6920760986913332e-05,
975
+ "loss": 0.5901,
976
  "step": 138
977
  },
978
  {
979
  "epoch": 0.25112917795844625,
980
+ "grad_norm": 0.24684546421512338,
981
  "learning_rate": 2.6866065519845124e-05,
982
+ "loss": 0.5932,
983
  "step": 139
984
  },
985
  {
986
  "epoch": 0.2529358626919603,
987
+ "grad_norm": 0.25108587954115535,
988
  "learning_rate": 2.681094515292018e-05,
989
+ "loss": 0.5927,
990
  "step": 140
991
  },
992
  {
993
  "epoch": 0.25474254742547425,
994
+ "grad_norm": 0.2759732073497199,
995
  "learning_rate": 2.6755401859887598e-05,
996
+ "loss": 0.573,
997
  "step": 141
998
  },
999
  {
1000
  "epoch": 0.2565492321589883,
1001
+ "grad_norm": 0.2927986426856055,
1002
  "learning_rate": 2.6699437629640595e-05,
1003
+ "loss": 0.5934,
1004
  "step": 142
1005
  },
1006
  {
1007
  "epoch": 0.25835591689250226,
1008
+ "grad_norm": 0.2915811926744858,
1009
  "learning_rate": 2.6643054466145297e-05,
1010
+ "loss": 0.5846,
1011
  "step": 143
1012
  },
1013
  {
1014
  "epoch": 0.2601626016260163,
1015
+ "grad_norm": 0.25639674407942853,
1016
  "learning_rate": 2.6586254388368995e-05,
1017
+ "loss": 0.5733,
1018
  "step": 144
1019
  },
1020
  {
1021
  "epoch": 0.26196928635953026,
1022
+ "grad_norm": 0.25466281147535325,
1023
  "learning_rate": 2.652903943020783e-05,
1024
+ "loss": 0.5862,
1025
  "step": 145
1026
  },
1027
  {
1028
  "epoch": 0.26377597109304424,
1029
+ "grad_norm": 0.24151461009286013,
1030
  "learning_rate": 2.647141164041398e-05,
1031
+ "loss": 0.5904,
1032
  "step": 146
1033
  },
1034
  {
1035
  "epoch": 0.26558265582655827,
1036
+ "grad_norm": 0.259822033974557,
1037
  "learning_rate": 2.641337308252228e-05,
1038
+ "loss": 0.581,
1039
  "step": 147
1040
  },
1041
  {
1042
  "epoch": 0.26738934056007224,
1043
+ "grad_norm": 0.24532475377494284,
1044
  "learning_rate": 2.6354925834776346e-05,
1045
+ "loss": 0.5817,
1046
  "step": 148
1047
  },
1048
  {
1049
  "epoch": 0.26919602529358627,
1050
+ "grad_norm": 0.2555788655192275,
1051
  "learning_rate": 2.6296071990054167e-05,
1052
+ "loss": 0.579,
1053
  "step": 149
1054
  },
1055
  {
1056
  "epoch": 0.27100271002710025,
1057
+ "grad_norm": 0.2534339164884998,
1058
  "learning_rate": 2.6236813655793123e-05,
1059
+ "loss": 0.5886,
1060
  "step": 150
1061
  },
1062
  {
1063
  "epoch": 0.2728093947606143,
1064
+ "grad_norm": 0.25366511050468843,
1065
  "learning_rate": 2.617715295391457e-05,
1066
+ "loss": 0.5707,
1067
  "step": 151
1068
  },
1069
  {
1070
  "epoch": 0.27461607949412825,
1071
+ "grad_norm": 0.26118391438335414,
1072
  "learning_rate": 2.6117092020747824e-05,
1073
+ "loss": 0.5779,
1074
  "step": 152
1075
  },
1076
  {
1077
  "epoch": 0.2764227642276423,
1078
+ "grad_norm": 0.31049055796662123,
1079
  "learning_rate": 2.6056633006953677e-05,
1080
+ "loss": 0.5783,
1081
  "step": 153
1082
  },
1083
  {
1084
  "epoch": 0.27822944896115626,
1085
+ "grad_norm": 0.2611788471953917,
1086
  "learning_rate": 2.5995778077447393e-05,
1087
+ "loss": 0.5823,
1088
  "step": 154
1089
  },
1090
  {
1091
  "epoch": 0.2800361336946703,
1092
+ "grad_norm": 0.3259190805002546,
1093
  "learning_rate": 2.5934529411321174e-05,
1094
+ "loss": 0.5743,
1095
  "step": 155
1096
  },
1097
  {
1098
  "epoch": 0.28184281842818426,
1099
+ "grad_norm": 0.2222204544644696,
1100
  "learning_rate": 2.587288920176613e-05,
1101
+ "loss": 0.575,
1102
  "step": 156
1103
  },
1104
  {
1105
  "epoch": 0.2836495031616983,
1106
+ "grad_norm": 0.29042132848143276,
1107
  "learning_rate": 2.581085965599375e-05,
1108
+ "loss": 0.5877,
1109
  "step": 157
1110
  },
1111
  {
1112
  "epoch": 0.28545618789521227,
1113
+ "grad_norm": 0.2577824036794116,
1114
  "learning_rate": 2.5748442995156882e-05,
1115
+ "loss": 0.5854,
1116
  "step": 158
1117
  },
1118
  {
1119
  "epoch": 0.2872628726287263,
1120
+ "grad_norm": 0.24041412490024394,
1121
  "learning_rate": 2.5685641454270172e-05,
1122
+ "loss": 0.566,
1123
  "step": 159
1124
  },
1125
  {
1126
  "epoch": 0.28906955736224027,
1127
+ "grad_norm": 0.33137613211077116,
1128
  "learning_rate": 2.5622457282130046e-05,
1129
+ "loss": 0.5753,
1130
  "step": 160
1131
  },
1132
  {
1133
  "epoch": 0.2908762420957543,
1134
+ "grad_norm": 0.2252657452973701,
1135
  "learning_rate": 2.5558892741234173e-05,
1136
+ "loss": 0.5827,
1137
  "step": 161
1138
  },
1139
  {
1140
  "epoch": 0.2926829268292683,
1141
+ "grad_norm": 0.353225035311219,
1142
  "learning_rate": 2.5494950107700482e-05,
1143
+ "loss": 0.5692,
1144
  "step": 162
1145
  },
1146
  {
1147
  "epoch": 0.2944896115627823,
1148
+ "grad_norm": 0.2549710140850951,
1149
  "learning_rate": 2.5430631671185616e-05,
1150
+ "loss": 0.5775,
1151
  "step": 163
1152
  },
1153
  {
1154
  "epoch": 0.2962962962962963,
1155
+ "grad_norm": 0.320108207545845,
1156
  "learning_rate": 2.5365939734802973e-05,
1157
+ "loss": 0.5871,
1158
  "step": 164
1159
  },
1160
  {
1161
  "epoch": 0.2981029810298103,
1162
+ "grad_norm": 0.27118178617195304,
1163
  "learning_rate": 2.5300876615040223e-05,
1164
+ "loss": 0.5777,
1165
  "step": 165
1166
  },
1167
  {
1168
  "epoch": 0.2999096657633243,
1169
+ "grad_norm": 0.2732065296628375,
1170
  "learning_rate": 2.523544464167637e-05,
1171
+ "loss": 0.5842,
1172
  "step": 166
1173
  },
1174
  {
1175
  "epoch": 0.3017163504968383,
1176
+ "grad_norm": 0.2440418264084237,
1177
  "learning_rate": 2.5169646157698313e-05,
1178
+ "loss": 0.583,
1179
  "step": 167
1180
  },
1181
  {
1182
  "epoch": 0.3035230352303523,
1183
+ "grad_norm": 0.2400624724182379,
1184
  "learning_rate": 2.5103483519216964e-05,
1185
+ "loss": 0.5813,
1186
  "step": 168
1187
  },
1188
  {
1189
  "epoch": 0.3053297199638663,
1190
+ "grad_norm": 0.25905283588713374,
1191
  "learning_rate": 2.5036959095382875e-05,
1192
+ "loss": 0.574,
1193
  "step": 169
1194
  },
1195
  {
1196
  "epoch": 0.3071364046973803,
1197
+ "grad_norm": 0.2591290671871167,
1198
  "learning_rate": 2.4970075268301388e-05,
1199
+ "loss": 0.5877,
1200
  "step": 170
1201
  },
1202
  {
1203
  "epoch": 0.3089430894308943,
1204
+ "grad_norm": 0.2235510897395504,
1205
  "learning_rate": 2.4902834432947353e-05,
1206
+ "loss": 0.5706,
1207
  "step": 171
1208
  },
1209
  {
1210
  "epoch": 0.3107497741644083,
1211
+ "grad_norm": 0.23760061910249497,
1212
  "learning_rate": 2.4835238997079382e-05,
1213
+ "loss": 0.5771,
1214
  "step": 172
1215
  },
1216
  {
1217
  "epoch": 0.31255645889792233,
1218
+ "grad_norm": 0.2617951363502666,
1219
  "learning_rate": 2.4767291381153603e-05,
1220
+ "loss": 0.5807,
1221
  "step": 173
1222
  },
1223
  {
1224
  "epoch": 0.3143631436314363,
1225
+ "grad_norm": 0.23091611449448254,
1226
  "learning_rate": 2.4698994018236994e-05,
1227
+ "loss": 0.5721,
1228
  "step": 174
1229
  },
1230
  {
1231
  "epoch": 0.31616982836495033,
1232
+ "grad_norm": 0.25262572503013164,
1233
  "learning_rate": 2.4630349353920284e-05,
1234
+ "loss": 0.5767,
1235
  "step": 175
1236
  },
1237
  {
1238
  "epoch": 0.3179765130984643,
1239
+ "grad_norm": 0.23055115622552902,
1240
  "learning_rate": 2.4561359846230346e-05,
1241
+ "loss": 0.5906,
1242
  "step": 176
1243
  },
1244
  {
1245
  "epoch": 0.31978319783197834,
1246
+ "grad_norm": 0.21968694104114944,
1247
  "learning_rate": 2.4492027965542217e-05,
1248
+ "loss": 0.5657,
1249
  "step": 177
1250
  },
1251
  {
1252
  "epoch": 0.3215898825654923,
1253
+ "grad_norm": 0.2525014182918365,
1254
  "learning_rate": 2.44223561944906e-05,
1255
+ "loss": 0.5724,
1256
  "step": 178
1257
  },
1258
  {
1259
  "epoch": 0.32339656729900634,
1260
+ "grad_norm": 0.2407328065060556,
1261
  "learning_rate": 2.4352347027881003e-05,
1262
+ "loss": 0.5825,
1263
  "step": 179
1264
  },
1265
  {
1266
  "epoch": 0.3252032520325203,
1267
+ "grad_norm": 0.2547387307288275,
1268
  "learning_rate": 2.4282002972600382e-05,
1269
+ "loss": 0.5908,
1270
  "step": 180
1271
  },
1272
  {
1273
  "epoch": 0.32700993676603435,
1274
+ "grad_norm": 0.2263495601766316,
1275
  "learning_rate": 2.4211326547527377e-05,
1276
+ "loss": 0.5688,
1277
  "step": 181
1278
  },
1279
  {
1280
  "epoch": 0.3288166214995483,
1281
+ "grad_norm": 0.2789402543570071,
1282
  "learning_rate": 2.4140320283442125e-05,
1283
+ "loss": 0.579,
1284
  "step": 182
1285
  },
1286
  {
1287
  "epoch": 0.33062330623306235,
1288
+ "grad_norm": 0.2567309540458022,
1289
  "learning_rate": 2.4068986722935625e-05,
1290
+ "loss": 0.5893,
1291
  "step": 183
1292
  },
1293
  {
1294
  "epoch": 0.3324299909665763,
1295
+ "grad_norm": 0.24182513399667138,
1296
  "learning_rate": 2.3997328420318705e-05,
1297
+ "loss": 0.5858,
1298
  "step": 184
1299
  },
1300
  {
1301
  "epoch": 0.33423667570009036,
1302
+ "grad_norm": 0.2675738732389291,
1303
  "learning_rate": 2.3925347941530556e-05,
1304
+ "loss": 0.5822,
1305
  "step": 185
1306
  },
1307
  {
1308
  "epoch": 0.33604336043360433,
1309
+ "grad_norm": 0.23428952659343405,
1310
  "learning_rate": 2.3853047864046843e-05,
1311
+ "loss": 0.5779,
1312
  "step": 186
1313
  },
1314
  {
1315
  "epoch": 0.33785004516711836,
1316
+ "grad_norm": 0.26339154899807604,
1317
  "learning_rate": 2.3780430776787413e-05,
1318
+ "loss": 0.5747,
1319
  "step": 187
1320
  },
1321
  {
1322
  "epoch": 0.33965672990063234,
1323
+ "grad_norm": 0.21349849308750005,
1324
  "learning_rate": 2.3707499280023604e-05,
1325
+ "loss": 0.5747,
1326
  "step": 188
1327
  },
1328
  {
1329
  "epoch": 0.34146341463414637,
1330
+ "grad_norm": 0.26554041110868815,
1331
  "learning_rate": 2.3634255985285104e-05,
1332
+ "loss": 0.5825,
1333
  "step": 189
1334
  },
1335
  {
1336
  "epoch": 0.34327009936766034,
1337
+ "grad_norm": 0.23538164196617103,
1338
  "learning_rate": 2.356070351526648e-05,
1339
+ "loss": 0.5708,
1340
  "step": 190
1341
  },
1342
  {
1343
  "epoch": 0.34507678410117437,
1344
+ "grad_norm": 0.24516795740455555,
1345
  "learning_rate": 2.348684450373322e-05,
1346
+ "loss": 0.5691,
1347
  "step": 191
1348
  },
1349
  {
1350
  "epoch": 0.34688346883468835,
1351
+ "grad_norm": 0.2320939063694949,
1352
  "learning_rate": 2.3412681595427467e-05,
1353
+ "loss": 0.5761,
1354
  "step": 192
1355
  },
1356
  {
1357
  "epoch": 0.3486901535682023,
1358
+ "grad_norm": 0.24367157013907606,
1359
  "learning_rate": 2.3338217445973268e-05,
1360
+ "loss": 0.5776,
1361
  "step": 193
1362
  },
1363
  {
1364
  "epoch": 0.35049683830171635,
1365
+ "grad_norm": 0.2383952654545809,
1366
  "learning_rate": 2.3263454721781537e-05,
1367
+ "loss": 0.5793,
1368
  "step": 194
1369
  },
1370
  {
1371
  "epoch": 0.3523035230352303,
1372
+ "grad_norm": 0.21970860044862534,
1373
  "learning_rate": 2.318839609995453e-05,
1374
+ "loss": 0.5625,
1375
  "step": 195
1376
  },
1377
  {
1378
  "epoch": 0.35411020776874436,
1379
+ "grad_norm": 0.22393266445368756,
1380
  "learning_rate": 2.3113044268189995e-05,
1381
+ "loss": 0.5758,
1382
  "step": 196
1383
  },
1384
  {
1385
  "epoch": 0.35591689250225833,
1386
+ "grad_norm": 0.20965253283535495,
1387
  "learning_rate": 2.303740192468495e-05,
1388
+ "loss": 0.5754,
1389
  "step": 197
1390
  },
1391
  {
1392
  "epoch": 0.35772357723577236,
1393
+ "grad_norm": 0.2134213495454822,
1394
  "learning_rate": 2.2961471778039045e-05,
1395
+ "loss": 0.5521,
1396
  "step": 198
1397
  },
1398
  {
1399
  "epoch": 0.35953026196928634,
1400
+ "grad_norm": 0.2077804399624621,
1401
  "learning_rate": 2.288525654715757e-05,
1402
+ "loss": 0.5758,
1403
  "step": 199
1404
  },
1405
  {
1406
  "epoch": 0.36133694670280037,
1407
+ "grad_norm": 0.21089298206975676,
1408
  "learning_rate": 2.280875896115413e-05,
1409
+ "loss": 0.5645,
1410
  "step": 200
1411
  },
1412
  {
1413
  "epoch": 0.36314363143631434,
1414
+ "grad_norm": 0.22751400705060473,
1415
  "learning_rate": 2.2731981759252876e-05,
1416
+ "loss": 0.5595,
1417
  "step": 201
1418
  },
1419
  {
1420
  "epoch": 0.36495031616982837,
1421
+ "grad_norm": 0.22071355320317226,
1422
  "learning_rate": 2.2654927690690445e-05,
1423
+ "loss": 0.5858,
1424
  "step": 202
1425
  },
1426
  {
1427
  "epoch": 0.36675700090334235,
1428
+ "grad_norm": 0.2269131246517602,
1429
  "learning_rate": 2.257759951461752e-05,
1430
+ "loss": 0.5733,
1431
  "step": 203
1432
  },
1433
  {
1434
  "epoch": 0.3685636856368564,
1435
+ "grad_norm": 0.27195813360203375,
1436
  "learning_rate": 2.25e-05,
1437
+ "loss": 0.5855,
1438
  "step": 204
1439
  },
1440
  {
1441
  "epoch": 0.37037037037037035,
1442
+ "grad_norm": 0.24074584104894936,
1443
  "learning_rate": 2.24221319255199e-05,
1444
+ "loss": 0.5761,
1445
  "step": 205
1446
  },
1447
  {
1448
  "epoch": 0.3721770551038844,
1449
+ "grad_norm": 0.2643169067368919,
1450
  "learning_rate": 2.234399807947579e-05,
1451
+ "loss": 0.5632,
1452
  "step": 206
1453
  },
1454
  {
1455
  "epoch": 0.37398373983739835,
1456
+ "grad_norm": 0.2602452192335381,
1457
  "learning_rate": 2.2265601259683e-05,
1458
+ "loss": 0.5713,
1459
  "step": 207
1460
  },
1461
  {
1462
  "epoch": 0.3757904245709124,
1463
+ "grad_norm": 0.22491768260975722,
1464
  "learning_rate": 2.2186944273373426e-05,
1465
+ "loss": 0.5589,
1466
  "step": 208
1467
  },
1468
  {
1469
  "epoch": 0.37759710930442636,
1470
+ "grad_norm": 0.23899265916988002,
1471
  "learning_rate": 2.210802993709498e-05,
1472
+ "loss": 0.5799,
1473
  "step": 209
1474
  },
1475
  {
1476
  "epoch": 0.3794037940379404,
1477
+ "grad_norm": 0.28185732149789655,
1478
  "learning_rate": 2.202886107661078e-05,
1479
+ "loss": 0.576,
1480
  "step": 210
1481
  },
1482
  {
1483
  "epoch": 0.38121047877145436,
1484
+ "grad_norm": 0.20671343969870537,
1485
  "learning_rate": 2.1949440526797928e-05,
1486
+ "loss": 0.5761,
1487
  "step": 211
1488
  },
1489
  {
1490
  "epoch": 0.3830171635049684,
1491
+ "grad_norm": 0.245945113158813,
1492
  "learning_rate": 2.1869771131546015e-05,
1493
+ "loss": 0.5834,
1494
  "step": 212
1495
  },
1496
  {
1497
  "epoch": 0.38482384823848237,
1498
+ "grad_norm": 0.2051130167210825,
1499
  "learning_rate": 2.178985574365529e-05,
1500
+ "loss": 0.5837,
1501
  "step": 213
1502
  },
1503
  {
1504
  "epoch": 0.3866305329719964,
1505
+ "grad_norm": 0.25952305018947347,
1506
  "learning_rate": 2.170969722473449e-05,
1507
+ "loss": 0.57,
1508
  "step": 214
1509
  },
1510
  {
1511
  "epoch": 0.3884372177055104,
1512
+ "grad_norm": 0.24249808093673167,
1513
  "learning_rate": 2.1629298445098403e-05,
1514
+ "loss": 0.5966,
1515
  "step": 215
1516
  },
1517
  {
1518
  "epoch": 0.3902439024390244,
1519
+ "grad_norm": 0.237872387820411,
1520
  "learning_rate": 2.154866228366505e-05,
1521
+ "loss": 0.5856,
1522
  "step": 216
1523
  },
1524
  {
1525
  "epoch": 0.3920505871725384,
1526
+ "grad_norm": 0.20572539636470738,
1527
  "learning_rate": 2.146779162785263e-05,
1528
+ "loss": 0.5857,
1529
  "step": 217
1530
  },
1531
  {
1532
  "epoch": 0.3938572719060524,
1533
+ "grad_norm": 0.22849509908427684,
1534
  "learning_rate": 2.138668937347609e-05,
1535
+ "loss": 0.5775,
1536
  "step": 218
1537
  },
1538
  {
1539
  "epoch": 0.3956639566395664,
1540
+ "grad_norm": 0.21171703597428448,
1541
  "learning_rate": 2.1305358424643484e-05,
1542
+ "loss": 0.572,
1543
  "step": 219
1544
  },
1545
  {
1546
  "epoch": 0.3974706413730804,
1547
+ "grad_norm": 0.2333174990121186,
1548
  "learning_rate": 2.1223801693651927e-05,
1549
+ "loss": 0.5746,
1550
  "step": 220
1551
  },
1552
  {
1553
  "epoch": 0.3992773261065944,
1554
+ "grad_norm": 0.21728105831361583,
1555
  "learning_rate": 2.114202210088336e-05,
1556
+ "loss": 0.5633,
1557
  "step": 221
1558
  },
1559
  {
1560
  "epoch": 0.4010840108401084,
1561
+ "grad_norm": 0.2232469073657137,
1562
  "learning_rate": 2.106002257469993e-05,
1563
+ "loss": 0.5805,
1564
  "step": 222
1565
  },
1566
  {
1567
  "epoch": 0.4028906955736224,
1568
+ "grad_norm": 0.23039919622981786,
1569
  "learning_rate": 2.0977806051339172e-05,
1570
+ "loss": 0.5631,
1571
  "step": 223
1572
  },
1573
  {
1574
  "epoch": 0.4046973803071364,
1575
+ "grad_norm": 0.24767592985180426,
1576
  "learning_rate": 2.0895375474808857e-05,
1577
+ "loss": 0.5681,
1578
  "step": 224
1579
  },
1580
  {
1581
  "epoch": 0.4065040650406504,
1582
+ "grad_norm": 0.22933111893390376,
1583
  "learning_rate": 2.0812733796781544e-05,
1584
+ "loss": 0.5728,
1585
  "step": 225
1586
  },
1587
  {
1588
  "epoch": 0.4083107497741644,
1589
+ "grad_norm": 0.24263450877516124,
1590
  "learning_rate": 2.0729883976488936e-05,
1591
+ "loss": 0.5792,
1592
  "step": 226
1593
  },
1594
  {
1595
  "epoch": 0.4101174345076784,
1596
+ "grad_norm": 0.24644389391459087,
1597
  "learning_rate": 2.064682898061588e-05,
1598
+ "loss": 0.5836,
1599
  "step": 227
1600
  },
1601
  {
1602
  "epoch": 0.41192411924119243,
1603
+ "grad_norm": 0.23999077805752708,
1604
  "learning_rate": 2.0563571783194146e-05,
1605
+ "loss": 0.5697,
1606
  "step": 228
1607
  },
1608
  {
1609
  "epoch": 0.4137308039747064,
1610
+ "grad_norm": 0.20338232613186608,
1611
  "learning_rate": 2.0480115365495928e-05,
1612
+ "loss": 0.5723,
1613
  "step": 229
1614
  },
1615
  {
1616
  "epoch": 0.41553748870822044,
1617
+ "grad_norm": 0.21451640836421845,
1618
  "learning_rate": 2.0396462715927107e-05,
1619
+ "loss": 0.5748,
1620
  "step": 230
1621
  },
1622
  {
1623
  "epoch": 0.4173441734417344,
1624
+ "grad_norm": 0.23674092702209837,
1625
  "learning_rate": 2.0312616829920222e-05,
1626
+ "loss": 0.5737,
1627
  "step": 231
1628
  },
1629
  {
1630
  "epoch": 0.41915085817524844,
1631
+ "grad_norm": 0.2040271869891673,
1632
  "learning_rate": 2.022858070982723e-05,
1633
+ "loss": 0.5768,
1634
  "step": 232
1635
  },
1636
  {
1637
  "epoch": 0.4209575429087624,
1638
+ "grad_norm": 0.23013110791835603,
1639
  "learning_rate": 2.0144357364811973e-05,
1640
+ "loss": 0.5586,
1641
  "step": 233
1642
  },
1643
  {
1644
  "epoch": 0.42276422764227645,
1645
+ "grad_norm": 0.23251714747465213,
1646
  "learning_rate": 2.0059949810742452e-05,
1647
+ "loss": 0.5744,
1648
  "step": 234
1649
  },
1650
  {
1651
  "epoch": 0.4245709123757904,
1652
+ "grad_norm": 0.24113338282908972,
1653
  "learning_rate": 1.997536107008281e-05,
1654
+ "loss": 0.5696,
1655
  "step": 235
1656
  },
1657
  {
1658
  "epoch": 0.42637759710930445,
1659
+ "grad_norm": 0.217275693069977,
1660
  "learning_rate": 1.9890594171785128e-05,
1661
+ "loss": 0.5807,
1662
  "step": 236
1663
  },
1664
  {
1665
  "epoch": 0.4281842818428184,
1666
+ "grad_norm": 0.20589216064741256,
1667
  "learning_rate": 1.9805652151180945e-05,
1668
+ "loss": 0.5804,
1669
  "step": 237
1670
  },
1671
  {
1672
  "epoch": 0.42999096657633246,
1673
+ "grad_norm": 0.22359306351640823,
1674
  "learning_rate": 1.972053804987258e-05,
1675
+ "loss": 0.5889,
1676
  "step": 238
1677
  },
1678
  {
1679
  "epoch": 0.43179765130984643,
1680
+ "grad_norm": 0.20375725610745427,
1681
  "learning_rate": 1.963525491562421e-05,
1682
+ "loss": 0.5701,
1683
  "step": 239
1684
  },
1685
  {
1686
  "epoch": 0.43360433604336046,
1687
+ "grad_norm": 0.20454609097134233,
1688
  "learning_rate": 1.954980580225275e-05,
1689
+ "loss": 0.5702,
1690
  "step": 240
1691
  },
1692
  {
1693
  "epoch": 0.43541102077687444,
1694
+ "grad_norm": 0.19497163244699564,
1695
  "learning_rate": 1.946419376951848e-05,
1696
+ "loss": 0.5693,
1697
  "step": 241
1698
  },
1699
  {
1700
  "epoch": 0.4372177055103884,
1701
+ "grad_norm": 0.21551459772294188,
1702
  "learning_rate": 1.9378421883015505e-05,
1703
+ "loss": 0.57,
1704
  "step": 242
1705
  },
1706
  {
1707
  "epoch": 0.43902439024390244,
1708
+ "grad_norm": 0.23200634978654613,
1709
  "learning_rate": 1.9292493214061953e-05,
1710
+ "loss": 0.5741,
1711
  "step": 243
1712
  },
1713
  {
1714
  "epoch": 0.4408310749774164,
1715
+ "grad_norm": 0.1884963011054949,
1716
  "learning_rate": 1.9206410839590042e-05,
1717
+ "loss": 0.5804,
1718
  "step": 244
1719
  },
1720
  {
1721
  "epoch": 0.44263775971093045,
1722
+ "grad_norm": 0.25404075829436523,
1723
  "learning_rate": 1.9120177842035853e-05,
1724
+ "loss": 0.5726,
1725
  "step": 245
1726
  },
1727
  {
1728
  "epoch": 0.4444444444444444,
1729
+ "grad_norm": 0.22097319274710464,
1730
  "learning_rate": 1.9033797309228984e-05,
1731
+ "loss": 0.5794,
1732
  "step": 246
1733
  },
1734
  {
1735
  "epoch": 0.44625112917795845,
1736
+ "grad_norm": 0.19246576672187318,
1737
  "learning_rate": 1.8947272334281977e-05,
1738
+ "loss": 0.5616,
1739
  "step": 247
1740
  },
1741
  {
1742
  "epoch": 0.4480578139114724,
1743
+ "grad_norm": 0.22700593999033858,
1744
  "learning_rate": 1.8860606015479537e-05,
1745
+ "loss": 0.5801,
1746
  "step": 248
1747
  },
1748
  {
1749
  "epoch": 0.44986449864498645,
1750
+ "grad_norm": 0.1934681513313933,
1751
  "learning_rate": 1.877380145616763e-05,
1752
+ "loss": 0.5632,
1753
  "step": 249
1754
  },
1755
  {
1756
  "epoch": 0.45167118337850043,
1757
+ "grad_norm": 0.21167453440737605,
1758
  "learning_rate": 1.868686176464232e-05,
1759
+ "loss": 0.5781,
1760
  "step": 250
1761
  },
1762
  {
1763
  "epoch": 0.45347786811201446,
1764
+ "grad_norm": 0.23469761452053498,
1765
  "learning_rate": 1.8599790054038487e-05,
1766
+ "loss": 0.5653,
1767
  "step": 251
1768
  },
1769
  {
1770
  "epoch": 0.45528455284552843,
1771
+ "grad_norm": 0.20579485783023402,
1772
  "learning_rate": 1.8512589442218358e-05,
1773
+ "loss": 0.5472,
1774
  "step": 252
1775
  },
1776
  {
1777
  "epoch": 0.45709123757904246,
1778
+ "grad_norm": 0.20450811714095307,
1779
  "learning_rate": 1.8425263051659838e-05,
1780
+ "loss": 0.5701,
1781
  "step": 253
1782
  },
1783
  {
1784
  "epoch": 0.45889792231255644,
1785
+ "grad_norm": 0.2795902989622695,
1786
  "learning_rate": 1.8337814009344716e-05,
1787
+ "loss": 0.5692,
1788
  "step": 254
1789
  },
1790
  {
1791
  "epoch": 0.46070460704607047,
1792
+ "grad_norm": 0.26043149982302555,
1793
  "learning_rate": 1.8250245446646707e-05,
1794
+ "loss": 0.5627,
1795
  "step": 255
1796
  },
1797
  {
1798
  "epoch": 0.46251129177958444,
1799
+ "grad_norm": 0.2115149201329313,
1800
  "learning_rate": 1.8162560499219286e-05,
1801
+ "loss": 0.563,
1802
  "step": 256
1803
  },
1804
  {
1805
  "epoch": 0.4643179765130985,
1806
+ "grad_norm": 0.2863165296718786,
1807
  "learning_rate": 1.807476230688346e-05,
1808
+ "loss": 0.5726,
1809
  "step": 257
1810
  },
1811
  {
1812
  "epoch": 0.46612466124661245,
1813
+ "grad_norm": 0.22723135404479125,
1814
  "learning_rate": 1.7986854013515274e-05,
1815
+ "loss": 0.5682,
1816
  "step": 258
1817
  },
1818
  {
1819
  "epoch": 0.4679313459801265,
1820
+ "grad_norm": 0.21269365868540174,
1821
  "learning_rate": 1.78988387669333e-05,
1822
+ "loss": 0.5718,
1823
  "step": 259
1824
  },
1825
  {
1826
  "epoch": 0.46973803071364045,
1827
+ "grad_norm": 0.27706961689827647,
1828
  "learning_rate": 1.781071971878587e-05,
1829
+ "loss": 0.5852,
1830
  "step": 260
1831
  },
1832
  {
1833
  "epoch": 0.4715447154471545,
1834
+ "grad_norm": 0.21919053031721478,
1835
  "learning_rate": 1.7722500024438244e-05,
1836
+ "loss": 0.5766,
1837
  "step": 261
1838
  },
1839
  {
1840
  "epoch": 0.47335140018066846,
1841
+ "grad_norm": 0.2458759125768864,
1842
  "learning_rate": 1.7634182842859628e-05,
1843
+ "loss": 0.565,
1844
  "step": 262
1845
  },
1846
  {
1847
  "epoch": 0.4751580849141825,
1848
+ "grad_norm": 0.2489757112909537,
1849
  "learning_rate": 1.7545771336510033e-05,
1850
+ "loss": 0.5695,
1851
  "step": 263
1852
  },
1853
  {
1854
  "epoch": 0.47696476964769646,
1855
+ "grad_norm": 0.2101979687684982,
1856
  "learning_rate": 1.7457268671227067e-05,
1857
+ "loss": 0.5759,
1858
  "step": 264
1859
  },
1860
  {
1861
  "epoch": 0.4787714543812105,
1862
+ "grad_norm": 0.2500578266252668,
1863
  "learning_rate": 1.736867801611254e-05,
1864
+ "loss": 0.5569,
1865
  "step": 265
1866
  },
1867
  {
1868
  "epoch": 0.48057813911472447,
1869
+ "grad_norm": 0.19262415552045406,
1870
  "learning_rate": 1.728000254341901e-05,
1871
+ "loss": 0.5813,
1872
  "step": 266
1873
  },
1874
  {
1875
  "epoch": 0.4823848238482385,
1876
+ "grad_norm": 0.19478984136741012,
1877
  "learning_rate": 1.7191245428436175e-05,
1878
+ "loss": 0.5653,
1879
  "step": 267
1880
  },
1881
  {
1882
  "epoch": 0.48419150858175247,
1883
+ "grad_norm": 0.23871243123226799,
1884
  "learning_rate": 1.7102409849377188e-05,
1885
+ "loss": 0.5661,
1886
  "step": 268
1887
  },
1888
  {
1889
  "epoch": 0.4859981933152665,
1890
+ "grad_norm": 0.17665577180387385,
1891
  "learning_rate": 1.7013498987264832e-05,
1892
+ "loss": 0.5727,
1893
  "step": 269
1894
  },
1895
  {
1896
  "epoch": 0.4878048780487805,
1897
+ "grad_norm": 0.21091637968705407,
1898
  "learning_rate": 1.6924516025817636e-05,
1899
+ "loss": 0.5544,
1900
  "step": 270
1901
  },
1902
  {
1903
  "epoch": 0.4896115627822945,
1904
+ "grad_norm": 0.25253202754252524,
1905
  "learning_rate": 1.683546415133584e-05,
1906
+ "loss": 0.5761,
1907
  "step": 271
1908
  },
1909
  {
1910
  "epoch": 0.4914182475158085,
1911
+ "grad_norm": 0.2150474071339765,
1912
  "learning_rate": 1.6746346552587342e-05,
1913
+ "loss": 0.5704,
1914
  "step": 272
1915
  },
1916
  {
1917
  "epoch": 0.4932249322493225,
1918
+ "grad_norm": 0.24422291161795256,
1919
  "learning_rate": 1.665716642069349e-05,
1920
+ "loss": 0.5785,
1921
  "step": 273
1922
  },
1923
  {
1924
  "epoch": 0.4950316169828365,
1925
+ "grad_norm": 0.19465875800542706,
1926
  "learning_rate": 1.6567926949014805e-05,
1927
+ "loss": 0.5658,
1928
  "step": 274
1929
  },
1930
  {
1931
  "epoch": 0.4968383017163505,
1932
+ "grad_norm": 0.20059784342394038,
1933
  "learning_rate": 1.6478631333036655e-05,
1934
+ "loss": 0.5607,
1935
  "step": 275
1936
  },
1937
  {
1938
  "epoch": 0.4986449864498645,
1939
+ "grad_norm": 0.19428493006954836,
1940
  "learning_rate": 1.638928277025482e-05,
1941
+ "loss": 0.5523,
1942
  "step": 276
1943
  },
1944
  {
1945
  "epoch": 0.5004516711833785,
1946
+ "grad_norm": 0.190112158192132,
1947
  "learning_rate": 1.6299884460061005e-05,
1948
+ "loss": 0.5643,
1949
  "step": 277
1950
  },
1951
  {
1952
  "epoch": 0.5022583559168925,
1953
+ "grad_norm": 0.18317648770164868,
1954
  "learning_rate": 1.621043960362826e-05,
1955
+ "loss": 0.5509,
1956
  "step": 278
1957
  },
1958
  {
1959
  "epoch": 0.5040650406504065,
1960
+ "grad_norm": 0.20777757920037251,
1961
  "learning_rate": 1.6120951403796367e-05,
1962
+ "loss": 0.5659,
1963
  "step": 279
1964
  },
1965
  {
1966
  "epoch": 0.5058717253839206,
1967
+ "grad_norm": 0.24714269404949812,
1968
  "learning_rate": 1.603142306495714e-05,
1969
+ "loss": 0.5625,
1970
  "step": 280
1971
  },
1972
  {
1973
  "epoch": 0.5076784101174345,
1974
+ "grad_norm": 0.1771707758636966,
1975
  "learning_rate": 1.5941857792939702e-05,
1976
+ "loss": 0.5629,
1977
  "step": 281
1978
  },
1979
  {
1980
  "epoch": 0.5094850948509485,
1981
+ "grad_norm": 0.19209684745372535,
1982
  "learning_rate": 1.585225879489567e-05,
1983
+ "loss": 0.5577,
1984
  "step": 282
1985
  },
1986
  {
1987
  "epoch": 0.5112917795844625,
1988
+ "grad_norm": 0.22597515112664288,
1989
  "learning_rate": 1.5762629279184326e-05,
1990
+ "loss": 0.5728,
1991
  "step": 283
1992
  },
1993
  {
1994
  "epoch": 0.5130984643179766,
1995
+ "grad_norm": 0.17250544743705307,
1996
  "learning_rate": 1.5672972455257726e-05,
1997
+ "loss": 0.5611,
1998
  "step": 284
1999
  },
2000
  {
2001
  "epoch": 0.5149051490514905,
2002
+ "grad_norm": 0.20410988336988065,
2003
  "learning_rate": 1.5583291533545775e-05,
2004
+ "loss": 0.5557,
2005
  "step": 285
2006
  },
2007
  {
2008
  "epoch": 0.5167118337850045,
2009
+ "grad_norm": 0.20391700905101098,
2010
  "learning_rate": 1.549358972534128e-05,
2011
+ "loss": 0.5538,
2012
  "step": 286
2013
  },
2014
  {
2015
  "epoch": 0.5185185185185185,
2016
+ "grad_norm": 0.18259503875294242,
2017
  "learning_rate": 1.5403870242684942e-05,
2018
+ "loss": 0.5754,
2019
  "step": 287
2020
  },
2021
  {
2022
  "epoch": 0.5203252032520326,
2023
+ "grad_norm": 0.18696008552172258,
2024
  "learning_rate": 1.5314136298250355e-05,
2025
+ "loss": 0.5723,
2026
  "step": 288
2027
  },
2028
  {
2029
  "epoch": 0.5221318879855466,
2030
+ "grad_norm": 0.19653948540679103,
2031
  "learning_rate": 1.5224391105228956e-05,
2032
+ "loss": 0.5565,
2033
  "step": 289
2034
  },
2035
  {
2036
  "epoch": 0.5239385727190605,
2037
+ "grad_norm": 0.17253613754365116,
2038
  "learning_rate": 1.5134637877214968e-05,
2039
+ "loss": 0.5682,
2040
  "step": 290
2041
  },
2042
  {
2043
  "epoch": 0.5257452574525745,
2044
+ "grad_norm": 0.2056061343083173,
2045
  "learning_rate": 1.5044879828090346e-05,
2046
+ "loss": 0.5727,
2047
  "step": 291
2048
  },
2049
  {
2050
  "epoch": 0.5275519421860885,
2051
+ "grad_norm": 0.19894424925026757,
2052
  "learning_rate": 1.4955120171909658e-05,
2053
+ "loss": 0.5763,
2054
  "step": 292
2055
  },
2056
  {
2057
  "epoch": 0.5293586269196026,
2058
+ "grad_norm": 0.1659275906965013,
2059
  "learning_rate": 1.4865362122785031e-05,
2060
+ "loss": 0.5689,
2061
  "step": 293
2062
  },
2063
  {
2064
  "epoch": 0.5311653116531165,
2065
+ "grad_norm": 0.17986609793129524,
2066
  "learning_rate": 1.4775608894771048e-05,
2067
+ "loss": 0.5612,
2068
  "step": 294
2069
  },
2070
  {
2071
  "epoch": 0.5329719963866305,
2072
+ "grad_norm": 0.232250278692599,
2073
  "learning_rate": 1.4685863701749648e-05,
2074
+ "loss": 0.5589,
2075
  "step": 295
2076
  },
2077
  {
2078
  "epoch": 0.5347786811201445,
2079
+ "grad_norm": 0.22464497097731725,
2080
  "learning_rate": 1.4596129757315062e-05,
2081
+ "loss": 0.5628,
2082
  "step": 296
2083
  },
2084
  {
2085
  "epoch": 0.5365853658536586,
2086
+ "grad_norm": 0.1952420747386217,
2087
  "learning_rate": 1.4506410274658718e-05,
2088
+ "loss": 0.5682,
2089
  "step": 297
2090
  },
2091
  {
2092
  "epoch": 0.5383920505871725,
2093
+ "grad_norm": 0.2106666884744506,
2094
  "learning_rate": 1.441670846645423e-05,
2095
+ "loss": 0.5627,
2096
  "step": 298
2097
  },
2098
  {
2099
  "epoch": 0.5401987353206865,
2100
+ "grad_norm": 0.21061181918065086,
2101
  "learning_rate": 1.4327027544742281e-05,
2102
+ "loss": 0.5783,
2103
  "step": 299
2104
  },
2105
  {
2106
  "epoch": 0.5420054200542005,
2107
+ "grad_norm": 0.1861290798519661,
2108
  "learning_rate": 1.4237370720815675e-05,
2109
+ "loss": 0.5664,
2110
  "step": 300
2111
  },
2112
  {
2113
  "epoch": 0.5438121047877146,
2114
+ "grad_norm": 0.18862693026886343,
2115
  "learning_rate": 1.4147741205104336e-05,
2116
+ "loss": 0.5683,
2117
  "step": 301
2118
  },
2119
  {
2120
  "epoch": 0.5456187895212286,
2121
+ "grad_norm": 0.18676460565472633,
2122
  "learning_rate": 1.40581422070603e-05,
2123
+ "loss": 0.5796,
2124
  "step": 302
2125
  },
2126
  {
2127
  "epoch": 0.5474254742547425,
2128
+ "grad_norm": 0.1752400383590249,
2129
  "learning_rate": 1.3968576935042864e-05,
2130
+ "loss": 0.5608,
2131
  "step": 303
2132
  },
2133
  {
2134
  "epoch": 0.5492321589882565,
2135
+ "grad_norm": 0.18836688433228918,
2136
  "learning_rate": 1.3879048596203637e-05,
2137
+ "loss": 0.5561,
2138
  "step": 304
2139
  },
2140
  {
2141
  "epoch": 0.5510388437217706,
2142
+ "grad_norm": 0.19530506408089982,
2143
  "learning_rate": 1.3789560396371742e-05,
2144
+ "loss": 0.5622,
2145
  "step": 305
2146
  },
2147
  {
2148
  "epoch": 0.5528455284552846,
2149
+ "grad_norm": 0.17965863699921675,
2150
  "learning_rate": 1.3700115539938995e-05,
2151
+ "loss": 0.5605,
2152
  "step": 306
2153
  },
2154
  {
2155
  "epoch": 0.5546522131887985,
2156
+ "grad_norm": 0.21750349150527712,
2157
  "learning_rate": 1.3610717229745183e-05,
2158
+ "loss": 0.578,
2159
  "step": 307
2160
  },
2161
  {
2162
  "epoch": 0.5564588979223125,
2163
+ "grad_norm": 0.19147158863524186,
2164
  "learning_rate": 1.3521368666963348e-05,
2165
+ "loss": 0.5572,
2166
  "step": 308
2167
  },
2168
  {
2169
  "epoch": 0.5582655826558266,
2170
+ "grad_norm": 0.18608236035974007,
2171
  "learning_rate": 1.3432073050985201e-05,
2172
+ "loss": 0.5669,
2173
  "step": 309
2174
  },
2175
  {
2176
  "epoch": 0.5600722673893406,
2177
+ "grad_norm": 0.18440593641158914,
2178
  "learning_rate": 1.3342833579306511e-05,
2179
+ "loss": 0.5492,
2180
  "step": 310
2181
  },
2182
  {
2183
  "epoch": 0.5618789521228545,
2184
+ "grad_norm": 0.1843799787511743,
2185
  "learning_rate": 1.3253653447412658e-05,
2186
+ "loss": 0.5685,
2187
  "step": 311
2188
  },
2189
  {
2190
  "epoch": 0.5636856368563685,
2191
+ "grad_norm": 0.20261725696919525,
2192
  "learning_rate": 1.3164535848664167e-05,
2193
+ "loss": 0.5674,
2194
  "step": 312
2195
  },
2196
  {
2197
  "epoch": 0.5654923215898826,
2198
+ "grad_norm": 0.17558184672847507,
2199
  "learning_rate": 1.3075483974182368e-05,
2200
+ "loss": 0.5684,
2201
  "step": 313
2202
  },
2203
  {
2204
  "epoch": 0.5672990063233966,
2205
+ "grad_norm": 0.17907331587608558,
2206
  "learning_rate": 1.2986501012735174e-05,
2207
+ "loss": 0.5609,
2208
  "step": 314
2209
  },
2210
  {
2211
  "epoch": 0.5691056910569106,
2212
+ "grad_norm": 0.1934736256775333,
2213
  "learning_rate": 1.2897590150622812e-05,
2214
+ "loss": 0.5613,
2215
  "step": 315
2216
  },
2217
  {
2218
  "epoch": 0.5709123757904245,
2219
+ "grad_norm": 0.17567517580021857,
2220
  "learning_rate": 1.2808754571563827e-05,
2221
+ "loss": 0.5573,
2222
  "step": 316
2223
  },
2224
  {
2225
  "epoch": 0.5727190605239386,
2226
+ "grad_norm": 0.19690748080324788,
2227
  "learning_rate": 1.2719997456580993e-05,
2228
+ "loss": 0.5525,
2229
  "step": 317
2230
  },
2231
  {
2232
  "epoch": 0.5745257452574526,
2233
+ "grad_norm": 0.20212907631156332,
2234
  "learning_rate": 1.2631321983887464e-05,
2235
+ "loss": 0.5654,
2236
  "step": 318
2237
  },
2238
  {
2239
  "epoch": 0.5763324299909666,
2240
+ "grad_norm": 0.19608541723414,
2241
  "learning_rate": 1.2542731328772936e-05,
2242
+ "loss": 0.5744,
2243
  "step": 319
2244
  },
2245
  {
2246
  "epoch": 0.5781391147244805,
2247
+ "grad_norm": 0.20857428025225674,
2248
  "learning_rate": 1.2454228663489967e-05,
2249
+ "loss": 0.5569,
2250
  "step": 320
2251
  },
2252
  {
2253
  "epoch": 0.5799457994579946,
2254
+ "grad_norm": 0.17609191697173338,
2255
  "learning_rate": 1.2365817157140373e-05,
2256
+ "loss": 0.5541,
2257
  "step": 321
2258
  },
2259
  {
2260
  "epoch": 0.5817524841915086,
2261
+ "grad_norm": 0.18308421719879198,
2262
  "learning_rate": 1.2277499975561755e-05,
2263
+ "loss": 0.5753,
2264
  "step": 322
2265
  },
2266
  {
2267
  "epoch": 0.5835591689250226,
2268
+ "grad_norm": 0.17484046660309832,
2269
  "learning_rate": 1.2189280281214128e-05,
2270
+ "loss": 0.5609,
2271
  "step": 323
2272
  },
2273
  {
2274
  "epoch": 0.5853658536585366,
2275
+ "grad_norm": 0.17252551757672757,
2276
  "learning_rate": 1.2101161233066704e-05,
2277
+ "loss": 0.5419,
2278
  "step": 324
2279
  },
2280
  {
2281
  "epoch": 0.5871725383920506,
2282
+ "grad_norm": 0.17553463607522976,
2283
  "learning_rate": 1.201314598648473e-05,
2284
+ "loss": 0.5591,
2285
  "step": 325
2286
  },
2287
  {
2288
  "epoch": 0.5889792231255646,
2289
+ "grad_norm": 0.17326333720847636,
2290
  "learning_rate": 1.1925237693116546e-05,
2291
+ "loss": 0.5689,
2292
  "step": 326
2293
  },
2294
  {
2295
  "epoch": 0.5907859078590786,
2296
+ "grad_norm": 0.18687445963344007,
2297
  "learning_rate": 1.1837439500780718e-05,
2298
+ "loss": 0.5553,
2299
  "step": 327
2300
  },
2301
  {
2302
  "epoch": 0.5925925925925926,
2303
+ "grad_norm": 0.19428514170213504,
2304
  "learning_rate": 1.1749754553353297e-05,
2305
+ "loss": 0.562,
2306
  "step": 328
2307
  },
2308
  {
2309
  "epoch": 0.5943992773261066,
2310
+ "grad_norm": 0.17110021647920834,
2311
  "learning_rate": 1.1662185990655285e-05,
2312
+ "loss": 0.5518,
2313
  "step": 329
2314
  },
2315
  {
2316
  "epoch": 0.5962059620596206,
2317
+ "grad_norm": 0.18984185404031334,
2318
  "learning_rate": 1.1574736948340163e-05,
2319
+ "loss": 0.559,
2320
  "step": 330
2321
  },
2322
  {
2323
  "epoch": 0.5980126467931346,
2324
+ "grad_norm": 0.16874712774452827,
2325
  "learning_rate": 1.1487410557781642e-05,
2326
+ "loss": 0.5632,
2327
  "step": 331
2328
  },
2329
  {
2330
  "epoch": 0.5998193315266486,
2331
+ "grad_norm": 0.1806801244410073,
2332
  "learning_rate": 1.140020994596151e-05,
2333
+ "loss": 0.5467,
2334
  "step": 332
2335
  },
2336
  {
2337
  "epoch": 0.6016260162601627,
2338
+ "grad_norm": 0.17626693416571312,
2339
  "learning_rate": 1.1313138235357683e-05,
2340
+ "loss": 0.5674,
2341
  "step": 333
2342
  },
2343
  {
2344
  "epoch": 0.6034327009936766,
2345
+ "grad_norm": 0.1770136401046488,
2346
  "learning_rate": 1.1226198543832373e-05,
2347
+ "loss": 0.5798,
2348
  "step": 334
2349
  },
2350
  {
2351
  "epoch": 0.6052393857271906,
2352
+ "grad_norm": 0.1910726991764081,
2353
  "learning_rate": 1.1139393984520467e-05,
2354
+ "loss": 0.563,
2355
  "step": 335
2356
  },
2357
  {
2358
  "epoch": 0.6070460704607046,
2359
+ "grad_norm": 0.1718176664612323,
2360
  "learning_rate": 1.1052727665718027e-05,
2361
+ "loss": 0.5596,
2362
  "step": 336
2363
  },
2364
  {
2365
  "epoch": 0.6088527551942186,
2366
+ "grad_norm": 0.17250346288280585,
2367
  "learning_rate": 1.0966202690771015e-05,
2368
+ "loss": 0.5611,
2369
  "step": 337
2370
  },
2371
  {
2372
  "epoch": 0.6106594399277326,
2373
+ "grad_norm": 0.19111207535840807,
2374
  "learning_rate": 1.087982215796415e-05,
2375
+ "loss": 0.5572,
2376
  "step": 338
2377
  },
2378
  {
2379
  "epoch": 0.6124661246612466,
2380
+ "grad_norm": 0.1686795998299278,
2381
  "learning_rate": 1.079358916040996e-05,
2382
+ "loss": 0.5657,
2383
  "step": 339
2384
  },
2385
  {
2386
  "epoch": 0.6142728093947606,
2387
+ "grad_norm": 0.16862020455628596,
2388
  "learning_rate": 1.0707506785938052e-05,
2389
+ "loss": 0.5611,
2390
  "step": 340
2391
  },
2392
  {
2393
  "epoch": 0.6160794941282746,
2394
+ "grad_norm": 0.1637656583385122,
2395
  "learning_rate": 1.0621578116984501e-05,
2396
+ "loss": 0.5593,
2397
  "step": 341
2398
  },
2399
  {
2400
  "epoch": 0.6178861788617886,
2401
+ "grad_norm": 0.16170288648566758,
2402
  "learning_rate": 1.0535806230481525e-05,
2403
+ "loss": 0.5628,
2404
  "step": 342
2405
  },
2406
  {
2407
  "epoch": 0.6196928635953026,
2408
+ "grad_norm": 0.1717924247456263,
2409
  "learning_rate": 1.0450194197747252e-05,
2410
+ "loss": 0.5567,
2411
  "step": 343
2412
  },
2413
  {
2414
  "epoch": 0.6214995483288166,
2415
+ "grad_norm": 0.16174500558159677,
2416
  "learning_rate": 1.036474508437579e-05,
2417
+ "loss": 0.5464,
2418
  "step": 344
2419
  },
2420
  {
2421
  "epoch": 0.6233062330623306,
2422
+ "grad_norm": 0.16381172759058904,
2423
  "learning_rate": 1.0279461950127422e-05,
2424
+ "loss": 0.5608,
2425
  "step": 345
2426
  },
2427
  {
2428
  "epoch": 0.6251129177958447,
2429
+ "grad_norm": 0.17001972937010654,
2430
  "learning_rate": 1.0194347848819055e-05,
2431
+ "loss": 0.5623,
2432
  "step": 346
2433
  },
2434
  {
2435
  "epoch": 0.6269196025293586,
2436
+ "grad_norm": 0.17303311757989492,
2437
  "learning_rate": 1.0109405828214872e-05,
2438
+ "loss": 0.5482,
2439
  "step": 347
2440
  },
2441
  {
2442
  "epoch": 0.6287262872628726,
2443
+ "grad_norm": 0.1882537905657933,
2444
  "learning_rate": 1.0024638929917192e-05,
2445
+ "loss": 0.5492,
2446
  "step": 348
2447
  },
2448
  {
2449
  "epoch": 0.6305329719963866,
2450
+ "grad_norm": 0.15457360070696674,
2451
  "learning_rate": 9.940050189257552e-06,
2452
+ "loss": 0.5521,
2453
  "step": 349
2454
  },
2455
  {
2456
  "epoch": 0.6323396567299007,
2457
+ "grad_norm": 0.15615688544922562,
2458
  "learning_rate": 9.85564263518803e-06,
2459
+ "loss": 0.5643,
2460
  "step": 350
2461
  },
2462
  {
2463
  "epoch": 0.6341463414634146,
2464
+ "grad_norm": 0.16319062879774318,
2465
  "learning_rate": 9.771419290172776e-06,
2466
+ "loss": 0.554,
2467
  "step": 351
2468
  },
2469
  {
2470
  "epoch": 0.6359530261969286,
2471
+ "grad_norm": 0.15770060432248675,
2472
  "learning_rate": 9.68738317007978e-06,
2473
+ "loss": 0.5682,
2474
  "step": 352
2475
  },
2476
  {
2477
  "epoch": 0.6377597109304426,
2478
+ "grad_norm": 0.16898427412267347,
2479
  "learning_rate": 9.6035372840729e-06,
2480
+ "loss": 0.568,
2481
  "step": 353
2482
  },
2483
  {
2484
  "epoch": 0.6395663956639567,
2485
+ "grad_norm": 0.15918955141776597,
2486
  "learning_rate": 9.519884634504074e-06,
2487
+ "loss": 0.5631,
2488
  "step": 354
2489
  },
2490
  {
2491
  "epoch": 0.6413730803974707,
2492
+ "grad_norm": 0.1635896927311437,
2493
  "learning_rate": 9.436428216805862e-06,
2494
+ "loss": 0.5524,
2495
  "step": 355
2496
  },
2497
  {
2498
  "epoch": 0.6431797651309846,
2499
+ "grad_norm": 0.15329421947992738,
2500
  "learning_rate": 9.35317101938412e-06,
2501
+ "loss": 0.566,
2502
  "step": 356
2503
  },
2504
  {
2505
  "epoch": 0.6449864498644986,
2506
+ "grad_norm": 0.18290183186305234,
2507
  "learning_rate": 9.270116023511063e-06,
2508
+ "loss": 0.5676,
2509
  "step": 357
2510
  },
2511
  {
2512
  "epoch": 0.6467931345980127,
2513
+ "grad_norm": 0.18253658487556157,
2514
  "learning_rate": 9.187266203218457e-06,
2515
+ "loss": 0.5729,
2516
  "step": 358
2517
  },
2518
  {
2519
  "epoch": 0.6485998193315267,
2520
+ "grad_norm": 0.1791341463101353,
2521
  "learning_rate": 9.104624525191147e-06,
2522
+ "loss": 0.546,
2523
  "step": 359
2524
  },
2525
  {
2526
  "epoch": 0.6504065040650406,
2527
+ "grad_norm": 0.14674620232896526,
2528
  "learning_rate": 9.022193948660825e-06,
2529
+ "loss": 0.5548,
2530
  "step": 360
2531
  },
2532
  {
2533
  "epoch": 0.6522131887985546,
2534
+ "grad_norm": 0.19660949623094373,
2535
  "learning_rate": 8.939977425300072e-06,
2536
+ "loss": 0.56,
2537
  "step": 361
2538
  },
2539
  {
2540
  "epoch": 0.6540198735320687,
2541
+ "grad_norm": 0.1876501178479123,
2542
  "learning_rate": 8.857977899116645e-06,
2543
+ "loss": 0.5717,
2544
  "step": 362
2545
  },
2546
  {
2547
  "epoch": 0.6558265582655827,
2548
+ "grad_norm": 0.21507796351172995,
2549
  "learning_rate": 8.776198306348072e-06,
2550
+ "loss": 0.5618,
2551
  "step": 363
2552
  },
2553
  {
2554
  "epoch": 0.6576332429990966,
2555
+ "grad_norm": 0.18228320215554383,
2556
  "learning_rate": 8.69464157535652e-06,
2557
+ "loss": 0.5612,
2558
  "step": 364
2559
  },
2560
  {
2561
  "epoch": 0.6594399277326106,
2562
+ "grad_norm": 0.1757206345144353,
2563
  "learning_rate": 8.61331062652391e-06,
2564
+ "loss": 0.5534,
2565
  "step": 365
2566
  },
2567
  {
2568
  "epoch": 0.6612466124661247,
2569
+ "grad_norm": 0.19026770352731234,
2570
  "learning_rate": 8.532208372147376e-06,
2571
+ "loss": 0.5746,
2572
  "step": 366
2573
  },
2574
  {
2575
  "epoch": 0.6630532971996387,
2576
+ "grad_norm": 0.1865886995183675,
2577
  "learning_rate": 8.451337716334953e-06,
2578
+ "loss": 0.5556,
2579
  "step": 367
2580
  },
2581
  {
2582
  "epoch": 0.6648599819331527,
2583
+ "grad_norm": 0.1797695927669861,
2584
  "learning_rate": 8.370701554901603e-06,
2585
+ "loss": 0.5675,
2586
  "step": 368
2587
  },
2588
  {
2589
  "epoch": 0.6666666666666666,
2590
+ "grad_norm": 0.17670482413719021,
2591
  "learning_rate": 8.290302775265509e-06,
2592
+ "loss": 0.5686,
2593
  "step": 369
2594
  },
2595
  {
2596
  "epoch": 0.6684733514001807,
2597
+ "grad_norm": 0.1701577263486839,
2598
  "learning_rate": 8.210144256344714e-06,
2599
+ "loss": 0.5668,
2600
  "step": 370
2601
  },
2602
  {
2603
  "epoch": 0.6702800361336947,
2604
+ "grad_norm": 0.2061129978805257,
2605
  "learning_rate": 8.130228868453989e-06,
2606
+ "loss": 0.555,
2607
  "step": 371
2608
  },
2609
  {
2610
  "epoch": 0.6720867208672087,
2611
+ "grad_norm": 0.1969046824239301,
2612
  "learning_rate": 8.050559473202078e-06,
2613
+ "loss": 0.5727,
2614
  "step": 372
2615
  },
2616
  {
2617
  "epoch": 0.6738934056007226,
2618
+ "grad_norm": 0.15311615793545202,
2619
  "learning_rate": 7.971138923389221e-06,
2620
+ "loss": 0.5583,
2621
  "step": 373
2622
  },
2623
  {
2624
  "epoch": 0.6757000903342367,
2625
+ "grad_norm": 0.18976350186526125,
2626
  "learning_rate": 7.89197006290502e-06,
2627
+ "loss": 0.5473,
2628
  "step": 374
2629
  },
2630
  {
2631
  "epoch": 0.6775067750677507,
2632
+ "grad_norm": 0.1879352111528028,
2633
  "learning_rate": 7.813055726626579e-06,
2634
+ "loss": 0.5619,
2635
  "step": 375
2636
  },
2637
  {
2638
  "epoch": 0.6793134598012647,
2639
+ "grad_norm": 0.20499323771840253,
2640
  "learning_rate": 7.734398740316999e-06,
2641
+ "loss": 0.5467,
2642
  "step": 376
2643
  },
2644
  {
2645
  "epoch": 0.6811201445347786,
2646
+ "grad_norm": 0.17137028091933101,
2647
  "learning_rate": 7.65600192052422e-06,
2648
+ "loss": 0.563,
2649
  "step": 377
2650
  },
2651
  {
2652
  "epoch": 0.6829268292682927,
2653
+ "grad_norm": 0.18099761803792097,
2654
  "learning_rate": 7.577868074480107e-06,
2655
+ "loss": 0.5634,
2656
  "step": 378
2657
  },
2658
  {
2659
  "epoch": 0.6847335140018067,
2660
+ "grad_norm": 0.19896866364484633,
2661
  "learning_rate": 7.500000000000004e-06,
2662
+ "loss": 0.5664,
2663
  "step": 379
2664
  },
2665
  {
2666
  "epoch": 0.6865401987353207,
2667
+ "grad_norm": 0.1884102435991128,
2668
  "learning_rate": 7.422400485382484e-06,
2669
+ "loss": 0.5473,
2670
  "step": 380
2671
  },
2672
  {
2673
  "epoch": 0.6883468834688347,
2674
+ "grad_norm": 0.17513345385135934,
2675
  "learning_rate": 7.345072309309557e-06,
2676
+ "loss": 0.5638,
2677
  "step": 381
2678
  },
2679
  {
2680
  "epoch": 0.6901535682023487,
2681
+ "grad_norm": 0.1733806383727063,
2682
  "learning_rate": 7.268018240747129e-06,
2683
+ "loss": 0.5515,
2684
  "step": 382
2685
  },
2686
  {
2687
  "epoch": 0.6919602529358627,
2688
+ "grad_norm": 0.18213237487884423,
2689
  "learning_rate": 7.191241038845876e-06,
2690
+ "loss": 0.5704,
2691
  "step": 383
2692
  },
2693
  {
2694
  "epoch": 0.6937669376693767,
2695
+ "grad_norm": 0.20160376104741037,
2696
  "learning_rate": 7.114743452842428e-06,
2697
+ "loss": 0.5681,
2698
  "step": 384
2699
  },
2700
  {
2701
  "epoch": 0.6955736224028907,
2702
+ "grad_norm": 0.14133588164420194,
2703
  "learning_rate": 7.038528221960959e-06,
2704
+ "loss": 0.5504,
2705
  "step": 385
2706
  },
2707
  {
2708
  "epoch": 0.6973803071364046,
2709
+ "grad_norm": 0.1554959410199801,
2710
  "learning_rate": 6.962598075315047e-06,
2711
+ "loss": 0.5622,
2712
  "step": 386
2713
  },
2714
  {
2715
  "epoch": 0.6991869918699187,
2716
+ "grad_norm": 0.17804707625227728,
2717
  "learning_rate": 6.886955731810009e-06,
2718
+ "loss": 0.5622,
2719
  "step": 387
2720
  },
2721
  {
2722
  "epoch": 0.7009936766034327,
2723
+ "grad_norm": 0.1770700659186734,
2724
  "learning_rate": 6.811603900045477e-06,
2725
+ "loss": 0.5612,
2726
  "step": 388
2727
  },
2728
  {
2729
  "epoch": 0.7028003613369467,
2730
+ "grad_norm": 0.15981371678518486,
2731
  "learning_rate": 6.736545278218464e-06,
2732
+ "loss": 0.5642,
2733
  "step": 389
2734
  },
2735
  {
2736
  "epoch": 0.7046070460704607,
2737
+ "grad_norm": 0.1530064162827602,
2738
  "learning_rate": 6.661782554026735e-06,
2739
+ "loss": 0.5584,
2740
  "step": 390
2741
  },
2742
  {
2743
  "epoch": 0.7064137308039747,
2744
+ "grad_norm": 0.16596305105845469,
2745
  "learning_rate": 6.587318404572537e-06,
2746
+ "loss": 0.5583,
2747
  "step": 391
2748
  },
2749
  {
2750
  "epoch": 0.7082204155374887,
2751
+ "grad_norm": 0.16327470933567317,
2752
  "learning_rate": 6.513155496266783e-06,
2753
+ "loss": 0.5641,
2754
  "step": 392
2755
  },
2756
  {
2757
  "epoch": 0.7100271002710027,
2758
+ "grad_norm": 0.1574337263337296,
2759
  "learning_rate": 6.439296484733526e-06,
2760
+ "loss": 0.5689,
2761
  "step": 393
2762
  },
2763
  {
2764
  "epoch": 0.7118337850045167,
2765
+ "grad_norm": 0.16018951967211098,
2766
  "learning_rate": 6.3657440147149e-06,
2767
+ "loss": 0.5716,
2768
  "step": 394
2769
  },
2770
  {
2771
  "epoch": 0.7136404697380307,
2772
+ "grad_norm": 0.14974112027258443,
2773
  "learning_rate": 6.292500719976397e-06,
2774
+ "loss": 0.5659,
2775
  "step": 395
2776
  },
2777
  {
2778
  "epoch": 0.7154471544715447,
2779
+ "grad_norm": 0.16158709365351387,
2780
  "learning_rate": 6.219569223212585e-06,
2781
+ "loss": 0.56,
2782
  "step": 396
2783
  },
2784
  {
2785
  "epoch": 0.7172538392050587,
2786
+ "grad_norm": 0.15600729659089102,
2787
  "learning_rate": 6.146952135953158e-06,
2788
+ "loss": 0.5598,
2789
  "step": 397
2790
  },
2791
  {
2792
  "epoch": 0.7190605239385727,
2793
+ "grad_norm": 0.15620491253414515,
2794
  "learning_rate": 6.074652058469447e-06,
2795
+ "loss": 0.5616,
2796
  "step": 398
2797
  },
2798
  {
2799
  "epoch": 0.7208672086720868,
2800
+ "grad_norm": 0.16380045583881614,
2801
  "learning_rate": 6.0026715796812945e-06,
2802
+ "loss": 0.5644,
2803
  "step": 399
2804
  },
2805
  {
2806
  "epoch": 0.7226738934056007,
2807
+ "grad_norm": 0.14967907113452944,
2808
  "learning_rate": 5.931013277064377e-06,
2809
+ "loss": 0.5609,
2810
  "step": 400
2811
  }
2812
  ],
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e17e9ef49452dfbb88bdb398539e0416fca0be014460bc3a0231be831d4c7664
3
  size 8017
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b75b4a596e16d5098959216e1c00d4bffb5d4776de6dbede10a6917de84c825
3
  size 8017