jeromeramos commited on
Commit
6130b47
·
verified ·
1 Parent(s): 906f965

Training in progress, epoch 0

Browse files
all_results.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
- "epoch": 2.0,
3
- "total_flos": 6.011530730704732e+18,
4
- "train_loss": 0.6795949197244182,
5
- "train_runtime": 13742.7497,
6
- "train_samples": 46269,
7
- "train_samples_per_second": 6.734,
8
- "train_steps_per_second": 0.105
9
  }
 
1
  {
2
+ "epoch": 1.9982692973347178,
3
+ "total_flos": 7.113001462715646e+18,
4
+ "train_loss": 0.6430129308119374,
5
+ "train_runtime": 14278.4866,
6
+ "train_samples": 46218,
7
+ "train_samples_per_second": 6.474,
8
+ "train_steps_per_second": 0.101
9
  }
model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8b15c6381825c63b914fc3ae491fa0b051dcde0f5dd41ee0c94916bb984a8c41
3
  size 4977222960
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7653580215afbcfe93ff934d3744298cf021d9eced1ee1bb0c28c4512aa26033
3
  size 4977222960
model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5465c66718008ceab01e65b8a7ff760284b4208c0250328f95ad2c6ee0cfe763
3
  size 4999802720
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f822b28a20a52d126f52f097f7ed9bb1bf624b3b0353fa3c4e8c190a99313ed7
3
  size 4999802720
model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f0c46bb5f8beb33abb2e43677b1cbdef9638a043142e739954efc8fe50663c8a
3
  size 4915916176
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e49f5ac92dcf91c19476489e033570a7d9f241b73f3dbd6d87cb6c27e5a42168
3
  size 4915916176
model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:794c20aa3d6e5f53ff1c9a7a5325c99a9e8fcdc3847f938bcee06fb2b328c0b9
3
  size 1168663096
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c3d207013b8bc5b2513767fe60cf1f1a37747843e9b11812c9e2b2688c1e52f
3
  size 1168663096
runs/May03_06-39-05_w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-86twgc7/events.out.tfevents.1746254525.w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-86twgc7.95801.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95730cfc6acb91562d931b19dea96b2281568b3803fc46061297b246df6aafeb
3
+ size 67519
runs/May03_10-50-27_w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-86twgc7/events.out.tfevents.1746269595.w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-86twgc7.134588.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1be665195a5b4101aeb11630e120900b29ce2d065b50aba3c7459da579c4c990
3
+ size 6638
runs/May03_11-38-46_w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-86twgc7/events.out.tfevents.1746272645.w-jerom-inter-play-sim-94c6890b9ccf44ea86f033a3db8a5dbd-86twgc7.144281.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a95d8fb5bf567517c4428ed226c8ea095bab6d1329a4a1f9acf5f5eddea957f4
3
+ size 6986
special_tokens_map.json CHANGED
@@ -15,14 +15,14 @@
15
  "single_word": false
16
  },
17
  {
18
- "content": "<answer>",
19
  "lstrip": false,
20
  "normalized": false,
21
  "rstrip": false,
22
  "single_word": false
23
  },
24
  {
25
- "content": "</answer>",
26
  "lstrip": false,
27
  "normalized": false,
28
  "rstrip": false,
@@ -34,6 +34,41 @@
34
  "normalized": false,
35
  "rstrip": false,
36
  "single_word": false
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
37
  }
38
  ],
39
  "bos_token": {
 
15
  "single_word": false
16
  },
17
  {
18
+ "content": "<action>",
19
  "lstrip": false,
20
  "normalized": false,
21
  "rstrip": false,
22
  "single_word": false
23
  },
24
  {
25
+ "content": "</action>",
26
  "lstrip": false,
27
  "normalized": false,
28
  "rstrip": false,
 
34
  "normalized": false,
35
  "rstrip": false,
36
  "single_word": false
37
+ },
38
+ {
39
+ "content": "<recommend>",
40
+ "lstrip": false,
41
+ "normalized": false,
42
+ "rstrip": false,
43
+ "single_word": false
44
+ },
45
+ {
46
+ "content": "<feedback>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false
51
+ },
52
+ {
53
+ "content": "<accept>",
54
+ "lstrip": false,
55
+ "normalized": false,
56
+ "rstrip": false,
57
+ "single_word": false
58
+ },
59
+ {
60
+ "content": "<movie_title>",
61
+ "lstrip": false,
62
+ "normalized": false,
63
+ "rstrip": false,
64
+ "single_word": false
65
+ },
66
+ {
67
+ "content": "</movie_title>",
68
+ "lstrip": false,
69
+ "normalized": false,
70
+ "rstrip": false,
71
+ "single_word": false
72
  }
73
  ],
74
  "bos_token": {
tokenizer.json CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:399c8422d1d7d647447d5522424c38669eb8b89394346f20c9ca2795800eeacb
3
- size 17211327
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b97b77eee18a756e51f2d1b8f463f76706b46ec12fdc443751f99a8ce81bffc0
3
+ size 17212268
tokenizer_config.json CHANGED
@@ -2081,7 +2081,7 @@
2081
  "special": true
2082
  },
2083
  "128260": {
2084
- "content": "<answer>",
2085
  "lstrip": false,
2086
  "normalized": false,
2087
  "rstrip": false,
@@ -2089,7 +2089,7 @@
2089
  "special": true
2090
  },
2091
  "128261": {
2092
- "content": "</answer>",
2093
  "lstrip": false,
2094
  "normalized": false,
2095
  "rstrip": false,
@@ -2103,14 +2103,59 @@
2103
  "rstrip": false,
2104
  "single_word": false,
2105
  "special": true
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2106
  }
2107
  },
2108
  "additional_special_tokens": [
2109
  "<response>",
2110
  "</response>",
2111
- "<answer>",
2112
- "</answer>",
2113
- "<inquire>"
 
 
 
 
 
2114
  ],
2115
  "bos_token": "<|im_start|>",
2116
  "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>user\n' }}{% endif %}",
 
2081
  "special": true
2082
  },
2083
  "128260": {
2084
+ "content": "<action>",
2085
  "lstrip": false,
2086
  "normalized": false,
2087
  "rstrip": false,
 
2089
  "special": true
2090
  },
2091
  "128261": {
2092
+ "content": "</action>",
2093
  "lstrip": false,
2094
  "normalized": false,
2095
  "rstrip": false,
 
2103
  "rstrip": false,
2104
  "single_word": false,
2105
  "special": true
2106
+ },
2107
+ "128263": {
2108
+ "content": "<recommend>",
2109
+ "lstrip": false,
2110
+ "normalized": false,
2111
+ "rstrip": false,
2112
+ "single_word": false,
2113
+ "special": true
2114
+ },
2115
+ "128264": {
2116
+ "content": "<feedback>",
2117
+ "lstrip": false,
2118
+ "normalized": false,
2119
+ "rstrip": false,
2120
+ "single_word": false,
2121
+ "special": true
2122
+ },
2123
+ "128265": {
2124
+ "content": "<accept>",
2125
+ "lstrip": false,
2126
+ "normalized": false,
2127
+ "rstrip": false,
2128
+ "single_word": false,
2129
+ "special": true
2130
+ },
2131
+ "128266": {
2132
+ "content": "<movie_title>",
2133
+ "lstrip": false,
2134
+ "normalized": false,
2135
+ "rstrip": false,
2136
+ "single_word": false,
2137
+ "special": true
2138
+ },
2139
+ "128267": {
2140
+ "content": "</movie_title>",
2141
+ "lstrip": false,
2142
+ "normalized": false,
2143
+ "rstrip": false,
2144
+ "single_word": false,
2145
+ "special": true
2146
  }
2147
  },
2148
  "additional_special_tokens": [
2149
  "<response>",
2150
  "</response>",
2151
+ "<action>",
2152
+ "</action>",
2153
+ "<inquire>",
2154
+ "<recommend>",
2155
+ "<feedback>",
2156
+ "<accept>",
2157
+ "<movie_title>",
2158
+ "</movie_title>"
2159
  ],
2160
  "bos_token": "<|im_start|>",
2161
  "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>user\n' }}{% endif %}",
train_results.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
- "epoch": 2.0,
3
- "total_flos": 6.011530730704732e+18,
4
- "train_loss": 0.6795949197244182,
5
- "train_runtime": 13742.7497,
6
- "train_samples": 46269,
7
- "train_samples_per_second": 6.734,
8
- "train_steps_per_second": 0.105
9
  }
 
1
  {
2
+ "epoch": 1.9982692973347178,
3
+ "total_flos": 7.113001462715646e+18,
4
+ "train_loss": 0.6430129308119374,
5
+ "train_runtime": 14278.4866,
6
+ "train_samples": 46218,
7
+ "train_samples_per_second": 6.474,
8
+ "train_steps_per_second": 0.101
9
  }
trainer_state.json CHANGED
@@ -2,2055 +2,2064 @@
2
  "best_global_step": null,
3
  "best_metric": null,
4
  "best_model_checkpoint": null,
5
- "epoch": 2.0,
6
  "eval_steps": 500,
7
- "global_step": 1446,
8
  "is_hyper_param_search": false,
9
  "is_local_process_zero": true,
10
  "is_world_process_zero": true,
11
  "log_history": [
12
  {
13
- "epoch": 0.0013831258644536654,
14
- "grad_norm": 133.08836364746094,
15
  "learning_rate": 0.0,
16
- "loss": 1.6747,
17
  "step": 1
18
  },
19
  {
20
- "epoch": 0.006915629322268326,
21
- "grad_norm": 10.979754447937012,
22
  "learning_rate": 5.517241379310345e-06,
23
- "loss": 1.5287,
24
  "step": 5
25
  },
26
  {
27
- "epoch": 0.013831258644536652,
28
- "grad_norm": 5.598673343658447,
29
  "learning_rate": 1.2413793103448277e-05,
30
- "loss": 1.1371,
31
  "step": 10
32
  },
33
  {
34
- "epoch": 0.02074688796680498,
35
- "grad_norm": 4.170475959777832,
36
  "learning_rate": 1.9310344827586207e-05,
37
- "loss": 1.0279,
38
  "step": 15
39
  },
40
  {
41
- "epoch": 0.027662517289073305,
42
- "grad_norm": 3.909233331680298,
43
  "learning_rate": 2.620689655172414e-05,
44
- "loss": 0.9947,
45
  "step": 20
46
  },
47
  {
48
- "epoch": 0.034578146611341634,
49
- "grad_norm": 3.94997501373291,
50
  "learning_rate": 3.310344827586207e-05,
51
- "loss": 0.9337,
52
  "step": 25
53
  },
54
  {
55
- "epoch": 0.04149377593360996,
56
- "grad_norm": 3.896550178527832,
57
  "learning_rate": 4e-05,
58
- "loss": 0.8761,
59
  "step": 30
60
  },
61
  {
62
- "epoch": 0.048409405255878286,
63
- "grad_norm": 3.581073522567749,
64
  "learning_rate": 4.689655172413793e-05,
65
- "loss": 0.8054,
66
  "step": 35
67
  },
68
  {
69
- "epoch": 0.05532503457814661,
70
- "grad_norm": 2.294735908508301,
71
  "learning_rate": 5.379310344827586e-05,
72
- "loss": 0.7511,
73
  "step": 40
74
  },
75
  {
76
- "epoch": 0.06224066390041494,
77
- "grad_norm": 1.9514353275299072,
78
  "learning_rate": 6.068965517241379e-05,
79
- "loss": 0.7493,
80
  "step": 45
81
  },
82
  {
83
- "epoch": 0.06915629322268327,
84
- "grad_norm": 2.1982648372650146,
85
  "learning_rate": 6.758620689655173e-05,
86
- "loss": 0.8087,
87
  "step": 50
88
  },
89
  {
90
- "epoch": 0.07607192254495158,
91
- "grad_norm": 2.2902512550354004,
92
  "learning_rate": 7.448275862068966e-05,
93
- "loss": 0.7906,
94
  "step": 55
95
  },
96
  {
97
- "epoch": 0.08298755186721991,
98
- "grad_norm": 1.9472178220748901,
99
  "learning_rate": 8.137931034482759e-05,
100
- "loss": 0.7872,
101
  "step": 60
102
  },
103
  {
104
- "epoch": 0.08990318118948824,
105
- "grad_norm": 1.7574316263198853,
106
  "learning_rate": 8.827586206896552e-05,
107
- "loss": 0.7898,
108
  "step": 65
109
  },
110
  {
111
- "epoch": 0.09681881051175657,
112
- "grad_norm": 1.8785020112991333,
113
  "learning_rate": 9.517241379310345e-05,
114
- "loss": 0.8166,
115
  "step": 70
116
  },
117
  {
118
- "epoch": 0.1037344398340249,
119
- "grad_norm": 1.5703641176223755,
120
  "learning_rate": 0.0001020689655172414,
121
- "loss": 0.7987,
122
  "step": 75
123
  },
124
  {
125
- "epoch": 0.11065006915629322,
126
- "grad_norm": 1.4895148277282715,
127
  "learning_rate": 0.00010896551724137931,
128
- "loss": 0.8054,
129
  "step": 80
130
  },
131
  {
132
- "epoch": 0.11756569847856155,
133
- "grad_norm": 1.7236794233322144,
134
  "learning_rate": 0.00011586206896551725,
135
- "loss": 0.8111,
136
  "step": 85
137
  },
138
  {
139
- "epoch": 0.12448132780082988,
140
- "grad_norm": 1.788547396659851,
141
  "learning_rate": 0.00012275862068965518,
142
- "loss": 0.8067,
143
  "step": 90
144
  },
145
  {
146
- "epoch": 0.1313969571230982,
147
- "grad_norm": 1.6218682527542114,
148
  "learning_rate": 0.0001296551724137931,
149
- "loss": 0.8175,
150
  "step": 95
151
  },
152
  {
153
- "epoch": 0.13831258644536654,
154
- "grad_norm": 1.438413381576538,
155
  "learning_rate": 0.00013655172413793104,
156
- "loss": 0.8485,
157
  "step": 100
158
  },
159
  {
160
- "epoch": 0.14522821576763487,
161
- "grad_norm": 1.6937086582183838,
162
  "learning_rate": 0.00014344827586206896,
163
- "loss": 0.8562,
164
  "step": 105
165
  },
166
  {
167
- "epoch": 0.15214384508990317,
168
- "grad_norm": 1.4097000360488892,
169
  "learning_rate": 0.0001503448275862069,
170
- "loss": 0.851,
171
  "step": 110
172
  },
173
  {
174
- "epoch": 0.1590594744121715,
175
- "grad_norm": 2.706479072570801,
176
  "learning_rate": 0.00015724137931034485,
177
- "loss": 0.8461,
178
  "step": 115
179
  },
180
  {
181
- "epoch": 0.16597510373443983,
182
- "grad_norm": 2.0034353733062744,
183
  "learning_rate": 0.00016413793103448276,
184
- "loss": 0.8548,
185
  "step": 120
186
  },
187
  {
188
- "epoch": 0.17289073305670816,
189
- "grad_norm": 1.3242019414901733,
190
  "learning_rate": 0.0001710344827586207,
191
- "loss": 0.8553,
192
  "step": 125
193
  },
194
  {
195
- "epoch": 0.1798063623789765,
196
- "grad_norm": 2.6091411113739014,
197
  "learning_rate": 0.00017793103448275862,
198
- "loss": 0.848,
199
  "step": 130
200
  },
201
  {
202
- "epoch": 0.18672199170124482,
203
- "grad_norm": 1.3679989576339722,
204
  "learning_rate": 0.00018482758620689654,
205
- "loss": 0.8716,
206
  "step": 135
207
  },
208
  {
209
- "epoch": 0.19363762102351315,
210
- "grad_norm": 1.323454737663269,
211
  "learning_rate": 0.0001917241379310345,
212
- "loss": 0.8488,
213
  "step": 140
214
  },
215
  {
216
- "epoch": 0.20055325034578148,
217
- "grad_norm": 2.7075045108795166,
218
  "learning_rate": 0.00019862068965517243,
219
- "loss": 0.9631,
220
  "step": 145
221
  },
222
  {
223
- "epoch": 0.2074688796680498,
224
- "grad_norm": 2.2683520317077637,
225
- "learning_rate": 0.0001999953352135947,
226
- "loss": 0.9498,
227
  "step": 150
228
  },
229
  {
230
- "epoch": 0.2143845089903181,
231
- "grad_norm": 8.38696002960205,
232
- "learning_rate": 0.0001999763852647035,
233
- "loss": 0.891,
234
  "step": 155
235
  },
236
  {
237
- "epoch": 0.22130013831258644,
238
- "grad_norm": 24.307083129882812,
239
- "learning_rate": 0.00019994286136445976,
240
- "loss": 0.9308,
241
  "step": 160
242
  },
243
  {
244
- "epoch": 0.22821576763485477,
245
- "grad_norm": 2.0212178230285645,
246
- "learning_rate": 0.0001998947683997744,
247
- "loss": 0.8901,
248
  "step": 165
249
  },
250
  {
251
- "epoch": 0.2351313969571231,
252
- "grad_norm": 1.2989600896835327,
253
- "learning_rate": 0.00019983211338134828,
254
- "loss": 0.8546,
255
  "step": 170
256
  },
257
  {
258
- "epoch": 0.24204702627939143,
259
- "grad_norm": 1.2097629308700562,
260
- "learning_rate": 0.00019975490544265012,
261
- "loss": 0.8543,
262
  "step": 175
263
  },
264
  {
265
- "epoch": 0.24896265560165975,
266
- "grad_norm": 30.946725845336914,
267
- "learning_rate": 0.00019966315583858516,
268
- "loss": 1.6915,
269
  "step": 180
270
  },
271
  {
272
- "epoch": 0.25587828492392806,
273
- "grad_norm": 2.118316173553467,
274
- "learning_rate": 0.0001995568779438545,
275
- "loss": 1.0665,
276
  "step": 185
277
  },
278
  {
279
- "epoch": 0.2627939142461964,
280
- "grad_norm": 3.086249589920044,
281
- "learning_rate": 0.00019943608725100532,
282
- "loss": 0.9988,
283
  "step": 190
284
  },
285
  {
286
- "epoch": 0.2697095435684647,
287
- "grad_norm": 1.1766245365142822,
288
- "learning_rate": 0.00019930080136817255,
289
- "loss": 0.9133,
290
  "step": 195
291
  },
292
  {
293
- "epoch": 0.2766251728907331,
294
- "grad_norm": 1.33064603805542,
295
- "learning_rate": 0.00019915104001651203,
296
- "loss": 0.8809,
297
  "step": 200
298
  },
299
  {
300
- "epoch": 0.2835408022130014,
301
- "grad_norm": 1.6439151763916016,
302
- "learning_rate": 0.00019898682502732568,
303
- "loss": 0.8967,
304
  "step": 205
305
  },
306
  {
307
- "epoch": 0.29045643153526973,
308
- "grad_norm": 1.3271840810775757,
309
- "learning_rate": 0.00019880818033887916,
310
- "loss": 0.8866,
311
  "step": 210
312
  },
313
  {
314
- "epoch": 0.29737206085753803,
315
- "grad_norm": 1.1745598316192627,
316
- "learning_rate": 0.0001986151319929121,
317
- "loss": 0.8598,
318
  "step": 215
319
  },
320
  {
321
- "epoch": 0.30428769017980634,
322
- "grad_norm": 1.0390416383743286,
323
- "learning_rate": 0.00019840770813084205,
324
- "loss": 0.8434,
325
  "step": 220
326
  },
327
  {
328
- "epoch": 0.3112033195020747,
329
- "grad_norm": 1.0289931297302246,
330
- "learning_rate": 0.00019818593898966212,
331
- "loss": 0.8765,
332
  "step": 225
333
  },
334
  {
335
- "epoch": 0.318118948824343,
336
- "grad_norm": 1.1883145570755005,
337
- "learning_rate": 0.00019794985689753337,
338
- "loss": 0.8859,
339
  "step": 230
340
  },
341
  {
342
- "epoch": 0.32503457814661135,
343
- "grad_norm": 1.2715511322021484,
344
- "learning_rate": 0.00019769949626907186,
345
- "loss": 0.8626,
346
  "step": 235
347
  },
348
  {
349
- "epoch": 0.33195020746887965,
350
- "grad_norm": 1.287985920906067,
351
- "learning_rate": 0.00019743489360033231,
352
- "loss": 0.8805,
353
  "step": 240
354
  },
355
  {
356
- "epoch": 0.338865836791148,
357
- "grad_norm": 1.1524231433868408,
358
- "learning_rate": 0.00019715608746348763,
359
- "loss": 0.8588,
360
  "step": 245
361
  },
362
  {
363
- "epoch": 0.3457814661134163,
364
- "grad_norm": 1.0479854345321655,
365
- "learning_rate": 0.00019686311850120625,
366
- "loss": 0.858,
367
  "step": 250
368
  },
369
  {
370
- "epoch": 0.35269709543568467,
371
- "grad_norm": 0.9751342535018921,
372
- "learning_rate": 0.0001965560294207274,
373
- "loss": 0.8558,
374
  "step": 255
375
  },
376
  {
377
- "epoch": 0.359612724757953,
378
- "grad_norm": 0.9835847020149231,
379
- "learning_rate": 0.00019623486498763555,
380
- "loss": 0.8755,
381
  "step": 260
382
  },
383
  {
384
- "epoch": 0.3665283540802213,
385
- "grad_norm": 3.7135465145111084,
386
- "learning_rate": 0.00019589967201933471,
387
- "loss": 0.8584,
388
  "step": 265
389
  },
390
  {
391
- "epoch": 0.37344398340248963,
392
- "grad_norm": 1.0380736589431763,
393
- "learning_rate": 0.00019555049937822384,
394
- "loss": 0.8544,
395
  "step": 270
396
  },
397
  {
398
- "epoch": 0.38035961272475793,
399
- "grad_norm": 1.0324746370315552,
400
- "learning_rate": 0.00019518739796457366,
401
- "loss": 0.8673,
402
  "step": 275
403
  },
404
  {
405
- "epoch": 0.3872752420470263,
406
- "grad_norm": 1.0108332633972168,
407
- "learning_rate": 0.00019481042070910705,
408
- "loss": 0.8443,
409
  "step": 280
410
  },
411
  {
412
- "epoch": 0.3941908713692946,
413
- "grad_norm": 0.8845415711402893,
414
- "learning_rate": 0.00019441962256528292,
415
- "loss": 0.857,
416
  "step": 285
417
  },
418
  {
419
- "epoch": 0.40110650069156295,
420
- "grad_norm": 0.8270080089569092,
421
- "learning_rate": 0.00019401506050128556,
422
- "loss": 0.8583,
423
  "step": 290
424
  },
425
  {
426
- "epoch": 0.40802213001383125,
427
- "grad_norm": 0.8553124070167542,
428
- "learning_rate": 0.00019359679349172004,
429
- "loss": 0.8291,
430
  "step": 295
431
  },
432
  {
433
- "epoch": 0.4149377593360996,
434
- "grad_norm": 0.9521012306213379,
435
- "learning_rate": 0.00019316488250901534,
436
- "loss": 0.8486,
437
  "step": 300
438
  },
439
  {
440
- "epoch": 0.4218533886583679,
441
- "grad_norm": 0.8840038776397705,
442
- "learning_rate": 0.00019271939051453612,
443
- "loss": 0.8258,
444
  "step": 305
445
  },
446
  {
447
- "epoch": 0.4287690179806362,
448
- "grad_norm": 1.223691463470459,
449
- "learning_rate": 0.00019226038244940464,
450
- "loss": 0.8142,
451
  "step": 310
452
  },
453
  {
454
- "epoch": 0.43568464730290457,
455
- "grad_norm": 0.8988921642303467,
456
- "learning_rate": 0.00019178792522503394,
457
- "loss": 0.8611,
458
  "step": 315
459
  },
460
  {
461
- "epoch": 0.4426002766251729,
462
- "grad_norm": 0.813004732131958,
463
- "learning_rate": 0.000191302087713374,
464
- "loss": 0.8208,
465
  "step": 320
466
  },
467
  {
468
- "epoch": 0.44951590594744123,
469
- "grad_norm": 0.8524804711341858,
470
- "learning_rate": 0.00019080294073687193,
471
- "loss": 0.8393,
472
  "step": 325
473
  },
474
  {
475
- "epoch": 0.45643153526970953,
476
- "grad_norm": 0.9394051432609558,
477
- "learning_rate": 0.000190290557058148,
478
- "loss": 0.8567,
479
  "step": 330
480
  },
481
  {
482
- "epoch": 0.4633471645919779,
483
- "grad_norm": 0.9042031168937683,
484
- "learning_rate": 0.00018976501136938864,
485
- "loss": 0.8387,
486
  "step": 335
487
  },
488
  {
489
- "epoch": 0.4702627939142462,
490
- "grad_norm": 0.7205588221549988,
491
- "learning_rate": 0.00018922638028145828,
492
- "loss": 0.8231,
493
  "step": 340
494
  },
495
  {
496
- "epoch": 0.47717842323651455,
497
- "grad_norm": 0.798876941204071,
498
- "learning_rate": 0.0001886747423127316,
499
- "loss": 0.8254,
500
  "step": 345
501
  },
502
  {
503
- "epoch": 0.48409405255878285,
504
- "grad_norm": 0.7704493403434753,
505
- "learning_rate": 0.00018811017787764747,
506
- "loss": 0.8244,
507
  "step": 350
508
  },
509
  {
510
- "epoch": 0.49100968188105115,
511
- "grad_norm": 0.8552532196044922,
512
- "learning_rate": 0.00018753276927498659,
513
- "loss": 0.8347,
514
  "step": 355
515
  },
516
  {
517
- "epoch": 0.4979253112033195,
518
- "grad_norm": 0.7718790173530579,
519
- "learning_rate": 0.00018694260067587463,
520
- "loss": 0.7962,
521
  "step": 360
522
  },
523
  {
524
- "epoch": 0.5048409405255878,
525
- "grad_norm": 0.888421356678009,
526
- "learning_rate": 0.00018633975811151223,
527
- "loss": 0.8284,
528
  "step": 365
529
  },
530
  {
531
- "epoch": 0.5117565698478561,
532
- "grad_norm": 0.8525059819221497,
533
- "learning_rate": 0.00018572432946063367,
534
- "loss": 0.8241,
535
  "step": 370
536
  },
537
  {
538
- "epoch": 0.5186721991701245,
539
- "grad_norm": 0.6882660388946533,
540
- "learning_rate": 0.00018509640443669682,
541
- "loss": 0.8001,
542
  "step": 375
543
  },
544
  {
545
- "epoch": 0.5255878284923928,
546
- "grad_norm": 0.7356276512145996,
547
- "learning_rate": 0.00018445607457480493,
548
- "loss": 0.8177,
549
  "step": 380
550
  },
551
  {
552
- "epoch": 0.5325034578146611,
553
- "grad_norm": 0.7074828147888184,
554
- "learning_rate": 0.0001838034332183634,
555
- "loss": 0.837,
556
  "step": 385
557
  },
558
  {
559
- "epoch": 0.5394190871369294,
560
- "grad_norm": 0.7650839686393738,
561
- "learning_rate": 0.0001831385755054726,
562
- "loss": 0.8219,
563
  "step": 390
564
  },
565
  {
566
- "epoch": 0.5463347164591977,
567
- "grad_norm": 0.6335296630859375,
568
- "learning_rate": 0.00018246159835505932,
569
- "loss": 0.8146,
570
  "step": 395
571
  },
572
  {
573
- "epoch": 0.5532503457814661,
574
- "grad_norm": 0.8082166314125061,
575
- "learning_rate": 0.0001817726004527485,
576
- "loss": 0.8083,
577
  "step": 400
578
  },
579
  {
580
- "epoch": 0.5601659751037344,
581
- "grad_norm": 0.6119678616523743,
582
- "learning_rate": 0.0001810716822364774,
583
- "loss": 0.7853,
584
  "step": 405
585
  },
586
  {
587
- "epoch": 0.5670816044260027,
588
- "grad_norm": 0.7896009683609009,
589
- "learning_rate": 0.00018035894588185438,
590
- "loss": 0.7868,
591
  "step": 410
592
  },
593
  {
594
- "epoch": 0.573997233748271,
595
- "grad_norm": 0.7086299657821655,
596
- "learning_rate": 0.0001796344952872643,
597
- "loss": 0.8234,
598
  "step": 415
599
  },
600
  {
601
- "epoch": 0.5809128630705395,
602
- "grad_norm": 0.689249575138092,
603
- "learning_rate": 0.00017889843605872305,
604
- "loss": 0.7917,
605
  "step": 420
606
  },
607
  {
608
- "epoch": 0.5878284923928078,
609
- "grad_norm": 0.6687580943107605,
610
- "learning_rate": 0.0001781508754944827,
611
- "loss": 0.7956,
612
  "step": 425
613
  },
614
  {
615
- "epoch": 0.5947441217150761,
616
- "grad_norm": 0.7054949998855591,
617
- "learning_rate": 0.0001773919225693903,
618
- "loss": 0.7979,
619
  "step": 430
620
  },
621
  {
622
- "epoch": 0.6016597510373444,
623
- "grad_norm": 0.8244301676750183,
624
- "learning_rate": 0.00017662168791900232,
625
- "loss": 0.7949,
626
  "step": 435
627
  },
628
  {
629
- "epoch": 0.6085753803596127,
630
- "grad_norm": 0.7373748421669006,
631
- "learning_rate": 0.00017584028382345654,
632
- "loss": 0.7742,
633
  "step": 440
634
  },
635
  {
636
- "epoch": 0.6154910096818811,
637
- "grad_norm": 0.716790497303009,
638
- "learning_rate": 0.00017504782419110497,
639
- "loss": 0.8082,
640
  "step": 445
641
  },
642
  {
643
- "epoch": 0.6224066390041494,
644
- "grad_norm": 0.7063632011413574,
645
- "learning_rate": 0.00017424442454190862,
646
- "loss": 0.7859,
647
  "step": 450
648
  },
649
  {
650
- "epoch": 0.6293222683264177,
651
- "grad_norm": 0.8756738901138306,
652
- "learning_rate": 0.00017343020199059783,
653
- "loss": 0.791,
654
  "step": 455
655
  },
656
  {
657
- "epoch": 0.636237897648686,
658
- "grad_norm": 0.667121946811676,
659
- "learning_rate": 0.0001726052752296001,
660
- "loss": 0.8044,
661
  "step": 460
662
  },
663
  {
664
- "epoch": 0.6431535269709544,
665
- "grad_norm": 0.7018240690231323,
666
- "learning_rate": 0.00017176976451173758,
667
- "loss": 0.7829,
668
  "step": 465
669
  },
670
  {
671
- "epoch": 0.6500691562932227,
672
- "grad_norm": 0.6935915350914001,
673
- "learning_rate": 0.00017092379163269764,
674
- "loss": 0.7975,
675
  "step": 470
676
  },
677
  {
678
- "epoch": 0.656984785615491,
679
- "grad_norm": 0.6710845232009888,
680
- "learning_rate": 0.00017006747991327796,
681
- "loss": 0.7777,
682
  "step": 475
683
  },
684
  {
685
- "epoch": 0.6639004149377593,
686
- "grad_norm": 0.7099134922027588,
687
- "learning_rate": 0.00016920095418140977,
688
- "loss": 0.7755,
689
  "step": 480
690
  },
691
  {
692
- "epoch": 0.6708160442600276,
693
- "grad_norm": 0.652553915977478,
694
- "learning_rate": 0.00016832434075396101,
695
- "loss": 0.7802,
696
  "step": 485
697
  },
698
  {
699
- "epoch": 0.677731673582296,
700
- "grad_norm": 0.6347463130950928,
701
- "learning_rate": 0.00016743776741832292,
702
- "loss": 0.7814,
703
  "step": 490
704
  },
705
  {
706
- "epoch": 0.6846473029045643,
707
- "grad_norm": 0.5948991179466248,
708
- "learning_rate": 0.00016654136341378157,
709
- "loss": 0.7704,
710
  "step": 495
711
  },
712
  {
713
- "epoch": 0.6915629322268326,
714
- "grad_norm": 0.6586953401565552,
715
- "learning_rate": 0.00016563525941267845,
716
- "loss": 0.7781,
717
  "step": 500
718
  },
719
  {
720
- "epoch": 0.6984785615491009,
721
- "grad_norm": 0.6398725509643555,
722
- "learning_rate": 0.00016471958750136176,
723
- "loss": 0.7707,
724
  "step": 505
725
  },
726
  {
727
- "epoch": 0.7053941908713693,
728
- "grad_norm": 0.6144907474517822,
729
- "learning_rate": 0.00016379448116093156,
730
- "loss": 0.7714,
731
  "step": 510
732
  },
733
  {
734
- "epoch": 0.7123098201936376,
735
- "grad_norm": 0.6330462694168091,
736
- "learning_rate": 0.00016286007524778185,
737
- "loss": 0.7653,
738
  "step": 515
739
  },
740
  {
741
- "epoch": 0.719225449515906,
742
- "grad_norm": 0.5762836933135986,
743
- "learning_rate": 0.00016191650597394198,
744
- "loss": 0.7715,
745
  "step": 520
746
  },
747
  {
748
- "epoch": 0.7261410788381742,
749
- "grad_norm": 0.578916609287262,
750
- "learning_rate": 0.00016096391088722047,
751
- "loss": 0.785,
752
  "step": 525
753
  },
754
  {
755
- "epoch": 0.7330567081604425,
756
- "grad_norm": 0.5828131437301636,
757
- "learning_rate": 0.0001600024288511541,
758
- "loss": 0.7424,
759
  "step": 530
760
  },
761
  {
762
- "epoch": 0.739972337482711,
763
- "grad_norm": 0.5606003999710083,
764
- "learning_rate": 0.00015903220002476515,
765
- "loss": 0.7782,
766
  "step": 535
767
  },
768
  {
769
- "epoch": 0.7468879668049793,
770
- "grad_norm": 0.5754717588424683,
771
- "learning_rate": 0.0001580533658421302,
772
- "loss": 0.7865,
773
  "step": 540
774
  },
775
  {
776
- "epoch": 0.7538035961272476,
777
- "grad_norm": 0.680016279220581,
778
- "learning_rate": 0.0001570660689917623,
779
- "loss": 0.7637,
780
  "step": 545
781
  },
782
  {
783
- "epoch": 0.7607192254495159,
784
- "grad_norm": 0.6306326985359192,
785
- "learning_rate": 0.00015607045339581096,
786
- "loss": 0.7528,
787
  "step": 550
788
  },
789
  {
790
- "epoch": 0.7676348547717843,
791
- "grad_norm": 0.5506445169448853,
792
- "learning_rate": 0.00015506666418908203,
793
- "loss": 0.767,
794
  "step": 555
795
  },
796
  {
797
- "epoch": 0.7745504840940526,
798
- "grad_norm": 0.6151379346847534,
799
- "learning_rate": 0.00015405484769788073,
800
- "loss": 0.7511,
801
  "step": 560
802
  },
803
  {
804
- "epoch": 0.7814661134163209,
805
- "grad_norm": 0.6520763635635376,
806
- "learning_rate": 0.00015303515141868116,
807
- "loss": 0.7529,
808
  "step": 565
809
  },
810
  {
811
- "epoch": 0.7883817427385892,
812
- "grad_norm": 0.5401438474655151,
813
- "learning_rate": 0.00015200772399662514,
814
- "loss": 0.7754,
815
  "step": 570
816
  },
817
  {
818
- "epoch": 0.7952973720608575,
819
- "grad_norm": 0.6025466322898865,
820
- "learning_rate": 0.00015097271520385366,
821
- "loss": 0.7577,
822
  "step": 575
823
  },
824
  {
825
- "epoch": 0.8022130013831259,
826
- "grad_norm": 0.6601846218109131,
827
- "learning_rate": 0.00014993027591767396,
828
- "loss": 0.7406,
829
  "step": 580
830
  },
831
  {
832
- "epoch": 0.8091286307053942,
833
- "grad_norm": 0.6068835854530334,
834
- "learning_rate": 0.0001488805580985655,
835
- "loss": 0.764,
836
  "step": 585
837
  },
838
  {
839
- "epoch": 0.8160442600276625,
840
- "grad_norm": 0.586649477481842,
841
- "learning_rate": 0.00014782371476802824,
842
- "loss": 0.7547,
843
  "step": 590
844
  },
845
  {
846
- "epoch": 0.8229598893499308,
847
- "grad_norm": 0.4942811131477356,
848
- "learning_rate": 0.00014675989998627598,
849
- "loss": 0.7539,
850
  "step": 595
851
  },
852
  {
853
- "epoch": 0.8298755186721992,
854
- "grad_norm": 0.5191232562065125,
855
- "learning_rate": 0.00014568926882977832,
856
- "loss": 0.75,
857
  "step": 600
858
  },
859
  {
860
- "epoch": 0.8367911479944675,
861
- "grad_norm": 0.5124319195747375,
862
- "learning_rate": 0.00014461197736865481,
863
- "loss": 0.7207,
864
  "step": 605
865
  },
866
  {
867
- "epoch": 0.8437067773167358,
868
- "grad_norm": 0.5561709403991699,
869
- "learning_rate": 0.00014352818264392364,
870
- "loss": 0.7504,
871
  "step": 610
872
  },
873
  {
874
- "epoch": 0.8506224066390041,
875
- "grad_norm": 0.5596902370452881,
876
- "learning_rate": 0.00014243804264460957,
877
- "loss": 0.7634,
878
  "step": 615
879
  },
880
  {
881
- "epoch": 0.8575380359612724,
882
- "grad_norm": 0.5537763237953186,
883
- "learning_rate": 0.00014134171628471276,
884
- "loss": 0.7293,
885
  "step": 620
886
  },
887
  {
888
- "epoch": 0.8644536652835408,
889
- "grad_norm": 0.6073982119560242,
890
- "learning_rate": 0.00014023936338004373,
891
- "loss": 0.7358,
892
  "step": 625
893
  },
894
  {
895
- "epoch": 0.8713692946058091,
896
- "grad_norm": 0.5265364050865173,
897
- "learning_rate": 0.00013913114462492601,
898
- "loss": 0.7415,
899
  "step": 630
900
  },
901
  {
902
- "epoch": 0.8782849239280774,
903
- "grad_norm": 0.5465463399887085,
904
- "learning_rate": 0.00013801722156877143,
905
- "loss": 0.7351,
906
  "step": 635
907
  },
908
  {
909
- "epoch": 0.8852005532503457,
910
- "grad_norm": 0.5381340384483337,
911
- "learning_rate": 0.00013689775659253006,
912
- "loss": 0.7403,
913
  "step": 640
914
  },
915
  {
916
- "epoch": 0.8921161825726142,
917
- "grad_norm": 0.5682520270347595,
918
- "learning_rate": 0.00013577291288501952,
919
- "loss": 0.7299,
920
  "step": 645
921
  },
922
  {
923
- "epoch": 0.8990318118948825,
924
- "grad_norm": 0.538758397102356,
925
- "learning_rate": 0.00013464285441913636,
926
- "loss": 0.7501,
927
  "step": 650
928
  },
929
  {
930
- "epoch": 0.9059474412171508,
931
- "grad_norm": 0.6069577932357788,
932
- "learning_rate": 0.00013350774592795292,
933
- "loss": 0.7373,
934
  "step": 655
935
  },
936
  {
937
- "epoch": 0.9128630705394191,
938
- "grad_norm": 0.5356409549713135,
939
- "learning_rate": 0.0001323677528807036,
940
- "loss": 0.7343,
941
  "step": 660
942
  },
943
  {
944
- "epoch": 0.9197786998616874,
945
- "grad_norm": 0.5176509618759155,
946
- "learning_rate": 0.00013122304145866381,
947
- "loss": 0.7298,
948
  "step": 665
949
  },
950
  {
951
- "epoch": 0.9266943291839558,
952
- "grad_norm": 0.6070849895477295,
953
- "learning_rate": 0.00013007377853092503,
954
- "loss": 0.7352,
955
  "step": 670
956
  },
957
  {
958
- "epoch": 0.9336099585062241,
959
- "grad_norm": 0.5752330422401428,
960
- "learning_rate": 0.00012892013163006962,
961
- "loss": 0.7323,
962
  "step": 675
963
  },
964
  {
965
- "epoch": 0.9405255878284924,
966
- "grad_norm": 0.550092339515686,
967
- "learning_rate": 0.00012776226892774903,
968
- "loss": 0.7437,
969
  "step": 680
970
  },
971
  {
972
- "epoch": 0.9474412171507607,
973
- "grad_norm": 0.5363165736198425,
974
- "learning_rate": 0.00012660035921016854,
975
- "loss": 0.7199,
976
  "step": 685
977
  },
978
  {
979
- "epoch": 0.9543568464730291,
980
- "grad_norm": 0.5486807823181152,
981
- "learning_rate": 0.00012543457185348298,
982
- "loss": 0.7159,
983
  "step": 690
984
  },
985
  {
986
- "epoch": 0.9612724757952974,
987
- "grad_norm": 0.6245793104171753,
988
- "learning_rate": 0.00012426507679910576,
989
- "loss": 0.7295,
990
  "step": 695
991
  },
992
  {
993
- "epoch": 0.9681881051175657,
994
- "grad_norm": 0.5306342244148254,
995
- "learning_rate": 0.00012309204452893606,
996
- "loss": 0.7239,
997
  "step": 700
998
  },
999
  {
1000
- "epoch": 0.975103734439834,
1001
- "grad_norm": 0.5488132238388062,
1002
- "learning_rate": 0.00012191564604050683,
1003
- "loss": 0.7027,
1004
  "step": 705
1005
  },
1006
  {
1007
- "epoch": 0.9820193637621023,
1008
- "grad_norm": 0.5506658554077148,
1009
- "learning_rate": 0.00012073605282205802,
1010
- "loss": 0.7397,
1011
  "step": 710
1012
  },
1013
  {
1014
- "epoch": 0.9889349930843707,
1015
- "grad_norm": 0.5438380241394043,
1016
- "learning_rate": 0.00011955343682753794,
1017
- "loss": 0.7241,
1018
  "step": 715
1019
  },
1020
  {
1021
- "epoch": 0.995850622406639,
1022
- "grad_norm": 0.5309740900993347,
1023
- "learning_rate": 0.0001183679704515368,
1024
- "loss": 0.7209,
1025
  "step": 720
1026
  },
1027
  {
1028
- "epoch": 1.0027662517289073,
1029
- "grad_norm": 0.6508172750473022,
1030
- "learning_rate": 0.00011717982650415624,
1031
- "loss": 0.6672,
 
 
 
 
 
 
 
 
1032
  "step": 725
1033
  },
1034
  {
1035
- "epoch": 1.0096818810511756,
1036
- "grad_norm": 0.6566136479377747,
1037
- "learning_rate": 0.00011598917818581791,
1038
- "loss": 0.5846,
1039
  "step": 730
1040
  },
1041
  {
1042
- "epoch": 1.016597510373444,
1043
- "grad_norm": 0.5107012987136841,
1044
- "learning_rate": 0.00011479619906201557,
1045
- "loss": 0.5685,
1046
  "step": 735
1047
  },
1048
  {
1049
- "epoch": 1.0235131396957122,
1050
- "grad_norm": 0.48931655287742615,
1051
- "learning_rate": 0.00011360106303801364,
1052
- "loss": 0.5846,
1053
  "step": 740
1054
  },
1055
  {
1056
- "epoch": 1.0304287690179805,
1057
- "grad_norm": 0.5219966769218445,
1058
- "learning_rate": 0.00011240394433349637,
1059
- "loss": 0.5588,
1060
  "step": 745
1061
  },
1062
  {
1063
- "epoch": 1.037344398340249,
1064
- "grad_norm": 0.5277289152145386,
1065
- "learning_rate": 0.00011120501745717112,
1066
- "loss": 0.5727,
1067
  "step": 750
1068
  },
1069
  {
1070
- "epoch": 1.0442600276625174,
1071
- "grad_norm": 0.5424395799636841,
1072
- "learning_rate": 0.00011000445718132966,
1073
- "loss": 0.566,
1074
  "step": 755
1075
  },
1076
  {
1077
- "epoch": 1.0511756569847857,
1078
- "grad_norm": 0.5607298612594604,
1079
- "learning_rate": 0.00010880243851637078,
1080
- "loss": 0.57,
1081
  "step": 760
1082
  },
1083
  {
1084
- "epoch": 1.058091286307054,
1085
- "grad_norm": 0.5789066553115845,
1086
- "learning_rate": 0.00010759913668528841,
1087
- "loss": 0.5659,
1088
  "step": 765
1089
  },
1090
  {
1091
- "epoch": 1.0650069156293223,
1092
- "grad_norm": 0.5418556928634644,
1093
- "learning_rate": 0.00010639472709812861,
1094
- "loss": 0.573,
1095
  "step": 770
1096
  },
1097
  {
1098
- "epoch": 1.0719225449515906,
1099
- "grad_norm": 0.49530017375946045,
1100
- "learning_rate": 0.0001051893853264195,
1101
- "loss": 0.5695,
1102
  "step": 775
1103
  },
1104
  {
1105
- "epoch": 1.0788381742738589,
1106
- "grad_norm": 0.5037497878074646,
1107
- "learning_rate": 0.00010398328707757738,
1108
- "loss": 0.5651,
1109
  "step": 780
1110
  },
1111
  {
1112
- "epoch": 1.0857538035961272,
1113
- "grad_norm": 0.5208268761634827,
1114
- "learning_rate": 0.00010277660816929313,
1115
- "loss": 0.5883,
1116
  "step": 785
1117
  },
1118
  {
1119
- "epoch": 1.0926694329183957,
1120
- "grad_norm": 0.5543485283851624,
1121
- "learning_rate": 0.00010156952450390269,
1122
- "loss": 0.5537,
1123
  "step": 790
1124
  },
1125
  {
1126
- "epoch": 1.099585062240664,
1127
- "grad_norm": 0.6337606310844421,
1128
- "learning_rate": 0.00010036221204274512,
1129
- "loss": 0.5649,
1130
  "step": 795
1131
  },
1132
  {
1133
- "epoch": 1.1065006915629323,
1134
- "grad_norm": 0.4707985818386078,
1135
- "learning_rate": 9.915484678051175e-05,
1136
- "loss": 0.5471,
1137
  "step": 800
1138
  },
1139
  {
1140
- "epoch": 1.1134163208852006,
1141
- "grad_norm": 0.5109753608703613,
1142
- "learning_rate": 9.794760471959116e-05,
1143
- "loss": 0.5663,
1144
  "step": 805
1145
  },
1146
  {
1147
- "epoch": 1.120331950207469,
1148
- "grad_norm": 0.5100296139717102,
1149
- "learning_rate": 9.674066184441221e-05,
1150
- "loss": 0.5713,
1151
  "step": 810
1152
  },
1153
  {
1154
- "epoch": 1.1272475795297372,
1155
- "grad_norm": 0.5414464473724365,
1156
- "learning_rate": 9.553419409579035e-05,
1157
- "loss": 0.5611,
1158
  "step": 815
1159
  },
1160
  {
1161
- "epoch": 1.1341632088520055,
1162
- "grad_norm": 0.5717695355415344,
1163
- "learning_rate": 9.432837734527995e-05,
1164
- "loss": 0.5817,
1165
  "step": 820
1166
  },
1167
  {
1168
- "epoch": 1.1410788381742738,
1169
- "grad_norm": 0.5524587035179138,
1170
- "learning_rate": 9.312338736953683e-05,
1171
- "loss": 0.5757,
1172
  "step": 825
1173
  },
1174
  {
1175
- "epoch": 1.147994467496542,
1176
- "grad_norm": 0.5250436663627625,
1177
- "learning_rate": 9.191939982469458e-05,
1178
- "loss": 0.569,
1179
  "step": 830
1180
  },
1181
  {
1182
- "epoch": 1.1549100968188104,
1183
- "grad_norm": 0.4710783064365387,
1184
- "learning_rate": 9.071659022075849e-05,
1185
- "loss": 0.565,
1186
  "step": 835
1187
  },
1188
  {
1189
- "epoch": 1.161825726141079,
1190
- "grad_norm": 0.4436459541320801,
1191
- "learning_rate": 8.951513389602076e-05,
1192
- "loss": 0.5666,
1193
  "step": 840
1194
  },
1195
  {
1196
- "epoch": 1.1687413554633472,
1197
- "grad_norm": 0.5180997252464294,
1198
- "learning_rate": 8.831520599150083e-05,
1199
- "loss": 0.5595,
1200
  "step": 845
1201
  },
1202
  {
1203
- "epoch": 1.1756569847856155,
1204
- "grad_norm": 0.49515190720558167,
1205
- "learning_rate": 8.71169814254142e-05,
1206
- "loss": 0.584,
1207
  "step": 850
1208
  },
1209
  {
1210
- "epoch": 1.1825726141078838,
1211
- "grad_norm": 0.5082345008850098,
1212
- "learning_rate": 8.592063486767406e-05,
1213
- "loss": 0.5687,
1214
  "step": 855
1215
  },
1216
  {
1217
- "epoch": 1.1894882434301521,
1218
- "grad_norm": 0.5054699182510376,
1219
- "learning_rate": 8.472634071442896e-05,
1220
- "loss": 0.5796,
1221
  "step": 860
1222
  },
1223
  {
1224
- "epoch": 1.1964038727524204,
1225
- "grad_norm": 0.4742473065853119,
1226
- "learning_rate": 8.353427306264032e-05,
1227
- "loss": 0.5575,
1228
  "step": 865
1229
  },
1230
  {
1231
- "epoch": 1.2033195020746887,
1232
- "grad_norm": 0.46908038854599,
1233
- "learning_rate": 8.23446056847037e-05,
1234
- "loss": 0.5654,
1235
  "step": 870
1236
  },
1237
  {
1238
- "epoch": 1.210235131396957,
1239
- "grad_norm": 0.4599679708480835,
1240
- "learning_rate": 8.115751200311725e-05,
1241
- "loss": 0.5452,
1242
  "step": 875
1243
  },
1244
  {
1245
- "epoch": 1.2171507607192256,
1246
- "grad_norm": 0.48915913701057434,
1247
- "learning_rate": 7.99731650652013e-05,
1248
- "loss": 0.5667,
1249
  "step": 880
1250
  },
1251
  {
1252
- "epoch": 1.2240663900414939,
1253
- "grad_norm": 0.5099295377731323,
1254
- "learning_rate": 7.879173751787259e-05,
1255
- "loss": 0.5673,
1256
  "step": 885
1257
  },
1258
  {
1259
- "epoch": 1.2309820193637622,
1260
- "grad_norm": 0.49911564588546753,
1261
- "learning_rate": 7.761340158247674e-05,
1262
- "loss": 0.5695,
1263
  "step": 890
1264
  },
1265
  {
1266
- "epoch": 1.2378976486860305,
1267
- "grad_norm": 0.4832319915294647,
1268
- "learning_rate": 7.64383290296829e-05,
1269
- "loss": 0.5477,
1270
  "step": 895
1271
  },
1272
  {
1273
- "epoch": 1.2448132780082988,
1274
- "grad_norm": 0.4508809447288513,
1275
- "learning_rate": 7.526669115444414e-05,
1276
- "loss": 0.5738,
1277
  "step": 900
1278
  },
1279
  {
1280
- "epoch": 1.251728907330567,
1281
- "grad_norm": 0.5219544768333435,
1282
- "learning_rate": 7.409865875102704e-05,
1283
- "loss": 0.5581,
1284
  "step": 905
1285
  },
1286
  {
1287
- "epoch": 1.2586445366528354,
1288
- "grad_norm": 0.4636399745941162,
1289
- "learning_rate": 7.293440208811435e-05,
1290
- "loss": 0.5593,
1291
  "step": 910
1292
  },
1293
  {
1294
- "epoch": 1.2655601659751037,
1295
- "grad_norm": 0.4862774908542633,
1296
- "learning_rate": 7.177409088398425e-05,
1297
- "loss": 0.5481,
1298
  "step": 915
1299
  },
1300
  {
1301
- "epoch": 1.272475795297372,
1302
- "grad_norm": 0.5330405235290527,
1303
- "learning_rate": 7.06178942817699e-05,
1304
- "loss": 0.5595,
1305
  "step": 920
1306
  },
1307
  {
1308
- "epoch": 1.2793914246196403,
1309
- "grad_norm": 0.49150362610816956,
1310
- "learning_rate": 6.946598082480268e-05,
1311
- "loss": 0.5429,
1312
  "step": 925
1313
  },
1314
  {
1315
- "epoch": 1.2863070539419086,
1316
- "grad_norm": 0.5385686755180359,
1317
- "learning_rate": 6.831851843204308e-05,
1318
- "loss": 0.5692,
1319
  "step": 930
1320
  },
1321
  {
1322
- "epoch": 1.293222683264177,
1323
- "grad_norm": 0.487090528011322,
1324
- "learning_rate": 6.71756743736024e-05,
1325
- "loss": 0.5484,
1326
  "step": 935
1327
  },
1328
  {
1329
- "epoch": 1.3001383125864454,
1330
- "grad_norm": 0.5285480618476868,
1331
- "learning_rate": 6.603761524635914e-05,
1332
- "loss": 0.5563,
1333
  "step": 940
1334
  },
1335
  {
1336
- "epoch": 1.3070539419087137,
1337
- "grad_norm": 0.48829564452171326,
1338
- "learning_rate": 6.490450694967358e-05,
1339
- "loss": 0.5606,
1340
  "step": 945
1341
  },
1342
  {
1343
- "epoch": 1.313969571230982,
1344
- "grad_norm": 0.4818781912326813,
1345
- "learning_rate": 6.377651466120391e-05,
1346
- "loss": 0.5538,
1347
  "step": 950
1348
  },
1349
  {
1350
- "epoch": 1.3208852005532503,
1351
- "grad_norm": 0.47652438282966614,
1352
- "learning_rate": 6.265380281282762e-05,
1353
- "loss": 0.5584,
1354
  "step": 955
1355
  },
1356
  {
1357
- "epoch": 1.3278008298755186,
1358
- "grad_norm": 0.47311243414878845,
1359
- "learning_rate": 6.15365350666718e-05,
1360
- "loss": 0.5409,
1361
  "step": 960
1362
  },
1363
  {
1364
- "epoch": 1.334716459197787,
1365
- "grad_norm": 0.46478375792503357,
1366
- "learning_rate": 6.042487429125516e-05,
1367
- "loss": 0.5554,
1368
  "step": 965
1369
  },
1370
  {
1371
- "epoch": 1.3416320885200554,
1372
- "grad_norm": 0.4783475399017334,
1373
- "learning_rate": 5.931898253774628e-05,
1374
- "loss": 0.5465,
1375
  "step": 970
1376
  },
1377
  {
1378
- "epoch": 1.3485477178423237,
1379
- "grad_norm": 0.48976966738700867,
1380
- "learning_rate": 5.821902101634069e-05,
1381
- "loss": 0.5563,
1382
  "step": 975
1383
  },
1384
  {
1385
- "epoch": 1.355463347164592,
1386
- "grad_norm": 0.4838036298751831,
1387
- "learning_rate": 5.7125150072760635e-05,
1388
- "loss": 0.5628,
1389
  "step": 980
1390
  },
1391
  {
1392
- "epoch": 1.3623789764868603,
1393
- "grad_norm": 0.46755191683769226,
1394
- "learning_rate": 5.603752916488085e-05,
1395
- "loss": 0.5472,
1396
  "step": 985
1397
  },
1398
  {
1399
- "epoch": 1.3692946058091287,
1400
- "grad_norm": 0.50596022605896,
1401
- "learning_rate": 5.4956316839483734e-05,
1402
- "loss": 0.5632,
1403
  "step": 990
1404
  },
1405
  {
1406
- "epoch": 1.376210235131397,
1407
- "grad_norm": 0.49392423033714294,
1408
- "learning_rate": 5.388167070914738e-05,
1409
- "loss": 0.5363,
1410
  "step": 995
1411
  },
1412
  {
1413
- "epoch": 1.3831258644536653,
1414
- "grad_norm": 0.4692226052284241,
1415
- "learning_rate": 5.281374742926987e-05,
1416
- "loss": 0.536,
1417
  "step": 1000
1418
  },
1419
  {
1420
- "epoch": 1.3900414937759336,
1421
- "grad_norm": 0.5283923745155334,
1422
- "learning_rate": 5.175270267523278e-05,
1423
- "loss": 0.5553,
1424
  "step": 1005
1425
  },
1426
  {
1427
- "epoch": 1.3969571230982019,
1428
- "grad_norm": 0.47653043270111084,
1429
- "learning_rate": 5.069869111970793e-05,
1430
- "loss": 0.5492,
1431
  "step": 1010
1432
  },
1433
  {
1434
- "epoch": 1.4038727524204702,
1435
- "grad_norm": 0.45839253067970276,
1436
- "learning_rate": 4.965186641011013e-05,
1437
- "loss": 0.5297,
1438
  "step": 1015
1439
  },
1440
  {
1441
- "epoch": 1.4107883817427385,
1442
- "grad_norm": 0.4624306261539459,
1443
- "learning_rate": 4.861238114619929e-05,
1444
- "loss": 0.5384,
1445
  "step": 1020
1446
  },
1447
  {
1448
- "epoch": 1.417704011065007,
1449
- "grad_norm": 0.48706522583961487,
1450
- "learning_rate": 4.75803868578355e-05,
1451
- "loss": 0.5369,
1452
  "step": 1025
1453
  },
1454
  {
1455
- "epoch": 1.4246196403872753,
1456
- "grad_norm": 0.4311310350894928,
1457
- "learning_rate": 4.655603398288979e-05,
1458
- "loss": 0.5276,
1459
  "step": 1030
1460
  },
1461
  {
1462
- "epoch": 1.4315352697095436,
1463
- "grad_norm": 0.47203701734542847,
1464
- "learning_rate": 4.5539471845314304e-05,
1465
- "loss": 0.5347,
1466
  "step": 1035
1467
  },
1468
  {
1469
- "epoch": 1.438450899031812,
1470
- "grad_norm": 0.46674981713294983,
1471
- "learning_rate": 4.453084863337471e-05,
1472
- "loss": 0.5299,
1473
  "step": 1040
1474
  },
1475
  {
1476
- "epoch": 1.4453665283540802,
1477
- "grad_norm": 0.42725497484207153,
1478
- "learning_rate": 4.353031137804821e-05,
1479
- "loss": 0.5369,
1480
  "step": 1045
1481
  },
1482
  {
1483
- "epoch": 1.4522821576763485,
1484
- "grad_norm": 0.4412887990474701,
1485
- "learning_rate": 4.253800593159029e-05,
1486
- "loss": 0.5418,
1487
  "step": 1050
1488
  },
1489
  {
1490
- "epoch": 1.4591977869986168,
1491
- "grad_norm": 0.45071831345558167,
1492
- "learning_rate": 4.155407694627322e-05,
1493
- "loss": 0.5551,
1494
  "step": 1055
1495
  },
1496
  {
1497
- "epoch": 1.4661134163208853,
1498
- "grad_norm": 0.44687119126319885,
1499
- "learning_rate": 4.057866785329959e-05,
1500
- "loss": 0.5424,
1501
  "step": 1060
1502
  },
1503
  {
1504
- "epoch": 1.4730290456431536,
1505
- "grad_norm": 0.45058682560920715,
1506
- "learning_rate": 3.96119208418937e-05,
1507
- "loss": 0.5524,
1508
  "step": 1065
1509
  },
1510
  {
1511
- "epoch": 1.479944674965422,
1512
- "grad_norm": 0.5045320391654968,
1513
- "learning_rate": 3.8653976838574104e-05,
1514
- "loss": 0.5403,
1515
  "step": 1070
1516
  },
1517
  {
1518
- "epoch": 1.4868603042876902,
1519
- "grad_norm": 0.4632035791873932,
1520
- "learning_rate": 3.770497548661021e-05,
1521
- "loss": 0.5346,
1522
  "step": 1075
1523
  },
1524
  {
1525
- "epoch": 1.4937759336099585,
1526
- "grad_norm": 0.4648893177509308,
1527
- "learning_rate": 3.676505512566597e-05,
1528
- "loss": 0.5423,
1529
  "step": 1080
1530
  },
1531
  {
1532
- "epoch": 1.5006915629322268,
1533
- "grad_norm": 0.4647049605846405,
1534
- "learning_rate": 3.5834352771633475e-05,
1535
- "loss": 0.5319,
1536
  "step": 1085
1537
  },
1538
  {
1539
- "epoch": 1.5076071922544951,
1540
- "grad_norm": 0.45878130197525024,
1541
- "learning_rate": 3.491300409665963e-05,
1542
- "loss": 0.533,
1543
  "step": 1090
1544
  },
1545
  {
1546
- "epoch": 1.5145228215767634,
1547
- "grad_norm": 0.4348316192626953,
1548
- "learning_rate": 3.4001143409368773e-05,
1549
- "loss": 0.5111,
1550
  "step": 1095
1551
  },
1552
  {
1553
- "epoch": 1.5214384508990317,
1554
- "grad_norm": 0.47867557406425476,
1555
- "learning_rate": 3.309890363528386e-05,
1556
- "loss": 0.527,
1557
  "step": 1100
1558
  },
1559
  {
1560
- "epoch": 1.5283540802213,
1561
- "grad_norm": 0.47667166590690613,
1562
- "learning_rate": 3.220641629744947e-05,
1563
- "loss": 0.5184,
1564
  "step": 1105
1565
  },
1566
  {
1567
- "epoch": 1.5352697095435683,
1568
- "grad_norm": 0.4480649530887604,
1569
- "learning_rate": 3.132381149725916e-05,
1570
- "loss": 0.5218,
1571
  "step": 1110
1572
  },
1573
  {
1574
- "epoch": 1.5421853388658366,
1575
- "grad_norm": 0.4330606460571289,
1576
- "learning_rate": 3.0451217895489992e-05,
1577
- "loss": 0.5317,
1578
  "step": 1115
1579
  },
1580
  {
1581
- "epoch": 1.5491009681881052,
1582
- "grad_norm": 0.45356714725494385,
1583
- "learning_rate": 2.9588762693547355e-05,
1584
- "loss": 0.5307,
1585
  "step": 1120
1586
  },
1587
  {
1588
- "epoch": 1.5560165975103735,
1589
- "grad_norm": 0.46869832277297974,
1590
- "learning_rate": 2.8736571614922046e-05,
1591
- "loss": 0.5231,
1592
  "step": 1125
1593
  },
1594
  {
1595
- "epoch": 1.5629322268326418,
1596
- "grad_norm": 0.5013293623924255,
1597
- "learning_rate": 2.7894768886863233e-05,
1598
- "loss": 0.5272,
1599
  "step": 1130
1600
  },
1601
  {
1602
- "epoch": 1.56984785615491,
1603
- "grad_norm": 0.44263824820518494,
1604
- "learning_rate": 2.7063477222269306e-05,
1605
- "loss": 0.53,
1606
  "step": 1135
1607
  },
1608
  {
1609
- "epoch": 1.5767634854771784,
1610
- "grad_norm": 0.45062586665153503,
1611
- "learning_rate": 2.6242817801799557e-05,
1612
- "loss": 0.5116,
1613
  "step": 1140
1614
  },
1615
  {
1616
- "epoch": 1.583679114799447,
1617
- "grad_norm": 0.4699437618255615,
1618
- "learning_rate": 2.5432910256209187e-05,
1619
- "loss": 0.5314,
1620
  "step": 1145
1621
  },
1622
  {
1623
- "epoch": 1.5905947441217152,
1624
- "grad_norm": 0.45320945978164673,
1625
- "learning_rate": 2.4633872648910252e-05,
1626
- "loss": 0.5244,
1627
  "step": 1150
1628
  },
1629
  {
1630
- "epoch": 1.5975103734439835,
1631
- "grad_norm": 0.45921820402145386,
1632
- "learning_rate": 2.3845821458761063e-05,
1633
- "loss": 0.5265,
1634
  "step": 1155
1635
  },
1636
  {
1637
- "epoch": 1.6044260027662518,
1638
- "grad_norm": 0.4836861789226532,
1639
- "learning_rate": 2.3068871563086757e-05,
1640
- "loss": 0.5397,
1641
  "step": 1160
1642
  },
1643
  {
1644
- "epoch": 1.61134163208852,
1645
- "grad_norm": 0.4829198122024536,
1646
- "learning_rate": 2.230313622093295e-05,
1647
- "loss": 0.5247,
1648
  "step": 1165
1649
  },
1650
  {
1651
- "epoch": 1.6182572614107884,
1652
- "grad_norm": 0.42497992515563965,
1653
- "learning_rate": 2.154872705655566e-05,
1654
- "loss": 0.4998,
1655
  "step": 1170
1656
  },
1657
  {
1658
- "epoch": 1.6251728907330567,
1659
- "grad_norm": 0.4427035450935364,
1660
- "learning_rate": 2.0805754043149394e-05,
1661
- "loss": 0.52,
1662
  "step": 1175
1663
  },
1664
  {
1665
- "epoch": 1.632088520055325,
1666
- "grad_norm": 0.4457587003707886,
1667
- "learning_rate": 2.0074325486815883e-05,
1668
- "loss": 0.4931,
1669
  "step": 1180
1670
  },
1671
  {
1672
- "epoch": 1.6390041493775933,
1673
- "grad_norm": 0.46790486574172974,
1674
- "learning_rate": 1.9354548010775896e-05,
1675
- "loss": 0.529,
1676
  "step": 1185
1677
  },
1678
  {
1679
- "epoch": 1.6459197786998616,
1680
- "grad_norm": 0.4801090955734253,
1681
- "learning_rate": 1.864652653982636e-05,
1682
- "loss": 0.5188,
1683
  "step": 1190
1684
  },
1685
  {
1686
- "epoch": 1.65283540802213,
1687
- "grad_norm": 0.4337711036205292,
1688
- "learning_rate": 1.7950364285044996e-05,
1689
- "loss": 0.5105,
1690
  "step": 1195
1691
  },
1692
  {
1693
- "epoch": 1.6597510373443982,
1694
- "grad_norm": 0.45500195026397705,
1695
- "learning_rate": 1.7266162728744993e-05,
1696
- "loss": 0.5049,
1697
  "step": 1200
1698
  },
1699
  {
1700
- "epoch": 1.6666666666666665,
1701
- "grad_norm": 0.4910541772842407,
1702
- "learning_rate": 1.6594021609681344e-05,
1703
- "loss": 0.5079,
1704
  "step": 1205
1705
  },
1706
  {
1707
- "epoch": 1.673582295988935,
1708
- "grad_norm": 0.4468669295310974,
1709
- "learning_rate": 1.5934038908511616e-05,
1710
- "loss": 0.5135,
1711
  "step": 1210
1712
  },
1713
  {
1714
- "epoch": 1.6804979253112033,
1715
- "grad_norm": 0.4427061975002289,
1716
- "learning_rate": 1.5286310833512963e-05,
1717
- "loss": 0.5019,
1718
  "step": 1215
1719
  },
1720
  {
1721
- "epoch": 1.6874135546334716,
1722
- "grad_norm": 0.44807228446006775,
1723
- "learning_rate": 1.4650931806557389e-05,
1724
- "loss": 0.5121,
1725
  "step": 1220
1726
  },
1727
  {
1728
- "epoch": 1.69432918395574,
1729
- "grad_norm": 0.45542779564857483,
1730
- "learning_rate": 1.402799444934757e-05,
1731
- "loss": 0.5183,
1732
  "step": 1225
1733
  },
1734
  {
1735
- "epoch": 1.7012448132780082,
1736
- "grad_norm": 0.44287553429603577,
1737
- "learning_rate": 1.3417589569914978e-05,
1738
- "loss": 0.4962,
1739
  "step": 1230
1740
  },
1741
  {
1742
- "epoch": 1.7081604426002768,
1743
- "grad_norm": 0.43979722261428833,
1744
- "learning_rate": 1.2819806149382441e-05,
1745
- "loss": 0.5065,
1746
  "step": 1235
1747
  },
1748
  {
1749
- "epoch": 1.715076071922545,
1750
- "grad_norm": 0.4374731481075287,
1751
- "learning_rate": 1.2234731328993055e-05,
1752
- "loss": 0.5144,
1753
  "step": 1240
1754
  },
1755
  {
1756
- "epoch": 1.7219917012448134,
1757
- "grad_norm": 0.4652646481990814,
1758
- "learning_rate": 1.1662450397407188e-05,
1759
- "loss": 0.5017,
1760
  "step": 1245
1761
  },
1762
  {
1763
- "epoch": 1.7289073305670817,
1764
- "grad_norm": 0.4413568377494812,
1765
- "learning_rate": 1.1103046778269687e-05,
1766
- "loss": 0.5109,
1767
  "step": 1250
1768
  },
1769
  {
1770
- "epoch": 1.73582295988935,
1771
- "grad_norm": 0.4495951235294342,
1772
- "learning_rate": 1.0556602018048866e-05,
1773
- "loss": 0.502,
1774
  "step": 1255
1775
  },
1776
  {
1777
- "epoch": 1.7427385892116183,
1778
- "grad_norm": 0.4346306025981903,
1779
- "learning_rate": 1.0023195774149119e-05,
1780
- "loss": 0.4963,
1781
  "step": 1260
1782
  },
1783
  {
1784
- "epoch": 1.7496542185338866,
1785
- "grad_norm": 0.4612555503845215,
1786
- "learning_rate": 9.502905803299e-06,
1787
- "loss": 0.5006,
1788
  "step": 1265
1789
  },
1790
  {
1791
- "epoch": 1.7565698478561549,
1792
- "grad_norm": 0.46448782086372375,
1793
- "learning_rate": 8.995807950216262e-06,
1794
- "loss": 0.5195,
1795
  "step": 1270
1796
  },
1797
  {
1798
- "epoch": 1.7634854771784232,
1799
- "grad_norm": 0.44382619857788086,
1800
- "learning_rate": 8.501976136551749e-06,
1801
- "loss": 0.5061,
1802
  "step": 1275
1803
  },
1804
  {
1805
- "epoch": 1.7704011065006915,
1806
- "grad_norm": 0.45043641328811646,
1807
- "learning_rate": 8.021482350113474e-06,
1808
- "loss": 0.5254,
1809
  "step": 1280
1810
  },
1811
  {
1812
- "epoch": 1.7773167358229598,
1813
- "grad_norm": 0.44958069920539856,
1814
- "learning_rate": 7.554396634372707e-06,
1815
- "loss": 0.5311,
1816
  "step": 1285
1817
  },
1818
  {
1819
- "epoch": 1.784232365145228,
1820
- "grad_norm": 0.433776319026947,
1821
- "learning_rate": 7.100787078253446e-06,
1822
- "loss": 0.486,
1823
  "step": 1290
1824
  },
1825
  {
1826
- "epoch": 1.7911479944674964,
1827
- "grad_norm": 0.41777148842811584,
1828
- "learning_rate": 6.660719806206839e-06,
1829
- "loss": 0.5108,
1830
  "step": 1295
1831
  },
1832
  {
1833
- "epoch": 1.798063623789765,
1834
- "grad_norm": 0.4487934112548828,
1835
- "learning_rate": 6.234258968571971e-06,
1836
- "loss": 0.5006,
1837
  "step": 1300
1838
  },
1839
  {
1840
- "epoch": 1.8049792531120332,
1841
- "grad_norm": 0.4544650614261627,
1842
- "learning_rate": 5.821466732224412e-06,
1843
- "loss": 0.5037,
1844
  "step": 1305
1845
  },
1846
  {
1847
- "epoch": 1.8118948824343015,
1848
- "grad_norm": 0.41929152607917786,
1849
- "learning_rate": 5.422403271513854e-06,
1850
- "loss": 0.4978,
1851
  "step": 1310
1852
  },
1853
  {
1854
- "epoch": 1.8188105117565698,
1855
- "grad_norm": 0.4613235294818878,
1856
- "learning_rate": 5.0371267594923834e-06,
1857
- "loss": 0.5209,
1858
  "step": 1315
1859
  },
1860
  {
1861
- "epoch": 1.8257261410788381,
1862
- "grad_norm": 0.44448205828666687,
1863
- "learning_rate": 4.66569335943422e-06,
1864
- "loss": 0.5014,
1865
  "step": 1320
1866
  },
1867
  {
1868
- "epoch": 1.8326417704011067,
1869
- "grad_norm": 0.43924444913864136,
1870
- "learning_rate": 4.3081572166486675e-06,
1871
- "loss": 0.5247,
1872
  "step": 1325
1873
  },
1874
  {
1875
- "epoch": 1.839557399723375,
1876
- "grad_norm": 0.44769856333732605,
1877
- "learning_rate": 3.964570450587113e-06,
1878
- "loss": 0.5005,
1879
  "step": 1330
1880
  },
1881
  {
1882
- "epoch": 1.8464730290456433,
1883
- "grad_norm": 0.4331699311733246,
1884
- "learning_rate": 3.6349831472453743e-06,
1885
- "loss": 0.5041,
1886
  "step": 1335
1887
  },
1888
  {
1889
- "epoch": 1.8533886583679116,
1890
- "grad_norm": 0.43336230516433716,
1891
- "learning_rate": 3.3194433518624614e-06,
1892
- "loss": 0.5162,
1893
  "step": 1340
1894
  },
1895
  {
1896
- "epoch": 1.8603042876901799,
1897
- "grad_norm": 0.4407755434513092,
1898
- "learning_rate": 3.017997061916833e-06,
1899
- "loss": 0.501,
1900
  "step": 1345
1901
  },
1902
  {
1903
- "epoch": 1.8672199170124482,
1904
- "grad_norm": 0.43583956360816956,
1905
- "learning_rate": 2.7306882204211626e-06,
1906
- "loss": 0.5009,
1907
  "step": 1350
1908
  },
1909
  {
1910
- "epoch": 1.8741355463347165,
1911
- "grad_norm": 0.42576801776885986,
1912
- "learning_rate": 2.4575587095166054e-06,
1913
- "loss": 0.5128,
1914
  "step": 1355
1915
  },
1916
  {
1917
- "epoch": 1.8810511756569848,
1918
- "grad_norm": 0.42050647735595703,
1919
- "learning_rate": 2.198648344367449e-06,
1920
- "loss": 0.5069,
1921
  "step": 1360
1922
  },
1923
  {
1924
- "epoch": 1.887966804979253,
1925
- "grad_norm": 0.4313651919364929,
1926
- "learning_rate": 1.953994867357134e-06,
1927
- "loss": 0.5055,
1928
  "step": 1365
1929
  },
1930
  {
1931
- "epoch": 1.8948824343015214,
1932
- "grad_norm": 0.43365243077278137,
1933
- "learning_rate": 1.7236339425863446e-06,
1934
- "loss": 0.5145,
1935
  "step": 1370
1936
  },
1937
  {
1938
- "epoch": 1.9017980636237897,
1939
- "grad_norm": 0.4598560035228729,
1940
- "learning_rate": 1.507599150674177e-06,
1941
- "loss": 0.5191,
1942
  "step": 1375
1943
  },
1944
  {
1945
- "epoch": 1.908713692946058,
1946
- "grad_norm": 0.4304969310760498,
1947
- "learning_rate": 1.3059219838629234e-06,
1948
- "loss": 0.501,
1949
  "step": 1380
1950
  },
1951
  {
1952
- "epoch": 1.9156293222683263,
1953
- "grad_norm": 0.44647660851478577,
1954
- "learning_rate": 1.11863184142732e-06,
1955
- "loss": 0.5063,
1956
  "step": 1385
1957
  },
1958
  {
1959
- "epoch": 1.9225449515905948,
1960
- "grad_norm": 0.40728816390037537,
1961
- "learning_rate": 9.457560253889219e-07,
1962
- "loss": 0.5009,
1963
  "step": 1390
1964
  },
1965
  {
1966
- "epoch": 1.929460580912863,
1967
- "grad_norm": 0.4395773410797119,
1968
- "learning_rate": 7.873197365361407e-07,
1969
- "loss": 0.4993,
1970
  "step": 1395
1971
  },
1972
  {
1973
- "epoch": 1.9363762102351314,
1974
- "grad_norm": 0.44724011421203613,
1975
- "learning_rate": 6.433460707506722e-07,
1976
- "loss": 0.5065,
1977
  "step": 1400
1978
  },
1979
  {
1980
- "epoch": 1.9432918395573997,
1981
- "grad_norm": 0.41976112127304077,
1982
- "learning_rate": 5.138560156407124e-07,
1983
- "loss": 0.5058,
1984
  "step": 1405
1985
  },
1986
  {
1987
- "epoch": 1.950207468879668,
1988
- "grad_norm": 0.43243739008903503,
1989
- "learning_rate": 3.988684474814819e-07,
1990
- "loss": 0.4919,
1991
  "step": 1410
1992
  },
1993
  {
1994
- "epoch": 1.9571230982019365,
1995
- "grad_norm": 0.4788164496421814,
1996
- "learning_rate": 2.984001284635496e-07,
1997
- "loss": 0.5128,
1998
  "step": 1415
1999
  },
2000
  {
2001
- "epoch": 1.9640387275242048,
2002
- "grad_norm": 0.42695632576942444,
2003
- "learning_rate": 2.1246570424940936e-07,
2004
- "loss": 0.5116,
2005
  "step": 1420
2006
  },
2007
  {
2008
- "epoch": 1.9709543568464731,
2009
- "grad_norm": 0.40510523319244385,
2010
- "learning_rate": 1.4107770183845458e-07,
2011
- "loss": 0.5058,
2012
  "step": 1425
2013
  },
2014
  {
2015
- "epoch": 1.9778699861687414,
2016
- "grad_norm": 0.4062838554382324,
2017
- "learning_rate": 8.424652774089436e-08,
2018
- "loss": 0.5128,
2019
  "step": 1430
2020
  },
2021
  {
2022
- "epoch": 1.9847856154910097,
2023
- "grad_norm": 0.4485253393650055,
2024
- "learning_rate": 4.198046646075593e-08,
2025
- "loss": 0.5062,
2026
  "step": 1435
2027
  },
2028
  {
2029
- "epoch": 1.991701244813278,
2030
- "grad_norm": 0.43608593940734863,
2031
- "learning_rate": 1.4285679288228437e-08,
2032
- "loss": 0.4919,
2033
  "step": 1440
2034
  },
2035
  {
2036
- "epoch": 1.9986168741355463,
2037
- "grad_norm": 0.4265099763870239,
2038
- "learning_rate": 1.166203401481436e-09,
2039
- "loss": 0.5001,
2040
- "step": 1445
 
2041
  },
2042
  {
2043
- "epoch": 2.0,
2044
- "step": 1446,
2045
- "total_flos": 6.011530730704732e+18,
2046
- "train_loss": 0.6795949197244182,
2047
- "train_runtime": 13742.7497,
2048
- "train_samples_per_second": 6.734,
2049
- "train_steps_per_second": 0.105
2050
  }
2051
  ],
2052
  "logging_steps": 5,
2053
- "max_steps": 1446,
2054
  "num_input_tokens_seen": 0,
2055
  "num_train_epochs": 2,
2056
  "save_steps": 500,
@@ -2066,7 +2075,7 @@
2066
  "attributes": {}
2067
  }
2068
  },
2069
- "total_flos": 6.011530730704732e+18,
2070
  "train_batch_size": 4,
2071
  "trial_name": null,
2072
  "trial_params": null
 
2
  "best_global_step": null,
3
  "best_metric": null,
4
  "best_model_checkpoint": null,
5
+ "epoch": 1.9982692973347178,
6
  "eval_steps": 500,
7
+ "global_step": 1444,
8
  "is_hyper_param_search": false,
9
  "is_local_process_zero": true,
10
  "is_world_process_zero": true,
11
  "log_history": [
12
  {
13
+ "epoch": 0.0013845621322256837,
14
+ "grad_norm": 92.96826171875,
15
  "learning_rate": 0.0,
16
+ "loss": 3.4865,
17
  "step": 1
18
  },
19
  {
20
+ "epoch": 0.006922810661128418,
21
+ "grad_norm": 35.61795425415039,
22
  "learning_rate": 5.517241379310345e-06,
23
+ "loss": 3.2186,
24
  "step": 5
25
  },
26
  {
27
+ "epoch": 0.013845621322256836,
28
+ "grad_norm": 11.851616859436035,
29
  "learning_rate": 1.2413793103448277e-05,
30
+ "loss": 2.7414,
31
  "step": 10
32
  },
33
  {
34
+ "epoch": 0.020768431983385256,
35
+ "grad_norm": 7.518247604370117,
36
  "learning_rate": 1.9310344827586207e-05,
37
+ "loss": 2.4447,
38
  "step": 15
39
  },
40
  {
41
+ "epoch": 0.027691242644513673,
42
+ "grad_norm": 7.707947731018066,
43
  "learning_rate": 2.620689655172414e-05,
44
+ "loss": 2.2078,
45
  "step": 20
46
  },
47
  {
48
+ "epoch": 0.034614053305642094,
49
+ "grad_norm": 7.561448574066162,
50
  "learning_rate": 3.310344827586207e-05,
51
+ "loss": 1.904,
52
  "step": 25
53
  },
54
  {
55
+ "epoch": 0.04153686396677051,
56
+ "grad_norm": 7.238735675811768,
57
  "learning_rate": 4e-05,
58
+ "loss": 1.4911,
59
  "step": 30
60
  },
61
  {
62
+ "epoch": 0.04845967462789893,
63
+ "grad_norm": 3.2107925415039062,
64
  "learning_rate": 4.689655172413793e-05,
65
+ "loss": 1.0777,
66
  "step": 35
67
  },
68
  {
69
+ "epoch": 0.055382485289027346,
70
+ "grad_norm": 4.923348426818848,
71
  "learning_rate": 5.379310344827586e-05,
72
+ "loss": 1.0017,
73
  "step": 40
74
  },
75
  {
76
+ "epoch": 0.06230529595015576,
77
+ "grad_norm": 22.305971145629883,
78
  "learning_rate": 6.068965517241379e-05,
79
+ "loss": 1.0557,
80
  "step": 45
81
  },
82
  {
83
+ "epoch": 0.06922810661128419,
84
+ "grad_norm": 38.01321029663086,
85
  "learning_rate": 6.758620689655173e-05,
86
+ "loss": 0.9835,
87
  "step": 50
88
  },
89
  {
90
+ "epoch": 0.0761509172724126,
91
+ "grad_norm": 34.23405838012695,
92
  "learning_rate": 7.448275862068966e-05,
93
+ "loss": 1.4889,
94
  "step": 55
95
  },
96
  {
97
+ "epoch": 0.08307372793354102,
98
+ "grad_norm": 11.111536979675293,
99
  "learning_rate": 8.137931034482759e-05,
100
+ "loss": 1.239,
101
  "step": 60
102
  },
103
  {
104
+ "epoch": 0.08999653859466944,
105
+ "grad_norm": 4.050104141235352,
106
  "learning_rate": 8.827586206896552e-05,
107
+ "loss": 0.9949,
108
  "step": 65
109
  },
110
  {
111
+ "epoch": 0.09691934925579786,
112
+ "grad_norm": 14.96822738647461,
113
  "learning_rate": 9.517241379310345e-05,
114
+ "loss": 1.0695,
115
  "step": 70
116
  },
117
  {
118
+ "epoch": 0.10384215991692627,
119
+ "grad_norm": 92.9261474609375,
120
  "learning_rate": 0.0001020689655172414,
121
+ "loss": 0.952,
122
  "step": 75
123
  },
124
  {
125
+ "epoch": 0.11076497057805469,
126
+ "grad_norm": 20.283737182617188,
127
  "learning_rate": 0.00010896551724137931,
128
+ "loss": 0.9217,
129
  "step": 80
130
  },
131
  {
132
+ "epoch": 0.11768778123918311,
133
+ "grad_norm": 5.715663909912109,
134
  "learning_rate": 0.00011586206896551725,
135
+ "loss": 0.8496,
136
  "step": 85
137
  },
138
  {
139
+ "epoch": 0.12461059190031153,
140
+ "grad_norm": 2.5005342960357666,
141
  "learning_rate": 0.00012275862068965518,
142
+ "loss": 0.8388,
143
  "step": 90
144
  },
145
  {
146
+ "epoch": 0.13153340256143994,
147
+ "grad_norm": 2.399947166442871,
148
  "learning_rate": 0.0001296551724137931,
149
+ "loss": 0.7857,
150
  "step": 95
151
  },
152
  {
153
+ "epoch": 0.13845621322256838,
154
+ "grad_norm": 2.1263132095336914,
155
  "learning_rate": 0.00013655172413793104,
156
+ "loss": 0.7885,
157
  "step": 100
158
  },
159
  {
160
+ "epoch": 0.14537902388369678,
161
+ "grad_norm": 2.0712645053863525,
162
  "learning_rate": 0.00014344827586206896,
163
+ "loss": 0.7858,
164
  "step": 105
165
  },
166
  {
167
+ "epoch": 0.1523018345448252,
168
+ "grad_norm": 1.9345420598983765,
169
  "learning_rate": 0.0001503448275862069,
170
+ "loss": 0.7976,
171
  "step": 110
172
  },
173
  {
174
+ "epoch": 0.1592246452059536,
175
+ "grad_norm": 1.4023919105529785,
176
  "learning_rate": 0.00015724137931034485,
177
+ "loss": 0.7927,
178
  "step": 115
179
  },
180
  {
181
+ "epoch": 0.16614745586708204,
182
+ "grad_norm": 1.31692636013031,
183
  "learning_rate": 0.00016413793103448276,
184
+ "loss": 0.7989,
185
  "step": 120
186
  },
187
  {
188
+ "epoch": 0.17307026652821045,
189
+ "grad_norm": 1.4096266031265259,
190
  "learning_rate": 0.0001710344827586207,
191
+ "loss": 0.8108,
192
  "step": 125
193
  },
194
  {
195
+ "epoch": 0.17999307718933888,
196
+ "grad_norm": 1.2111753225326538,
197
  "learning_rate": 0.00017793103448275862,
198
+ "loss": 0.789,
199
  "step": 130
200
  },
201
  {
202
+ "epoch": 0.18691588785046728,
203
+ "grad_norm": 1.141992449760437,
204
  "learning_rate": 0.00018482758620689654,
205
+ "loss": 0.8205,
206
  "step": 135
207
  },
208
  {
209
+ "epoch": 0.19383869851159571,
210
+ "grad_norm": 1.1753684282302856,
211
  "learning_rate": 0.0001917241379310345,
212
+ "loss": 0.8463,
213
  "step": 140
214
  },
215
  {
216
+ "epoch": 0.20076150917272412,
217
+ "grad_norm": 1.3483130931854248,
218
  "learning_rate": 0.00019862068965517243,
219
+ "loss": 0.8368,
220
  "step": 145
221
  },
222
  {
223
+ "epoch": 0.20768431983385255,
224
+ "grad_norm": 1.7060608863830566,
225
+ "learning_rate": 0.0001999953208384105,
226
+ "loss": 0.8143,
227
  "step": 150
228
  },
229
  {
230
+ "epoch": 0.21460713049498095,
231
+ "grad_norm": 1.2675988674163818,
232
+ "learning_rate": 0.0001999763124949378,
233
+ "loss": 0.8178,
234
  "step": 155
235
  },
236
  {
237
+ "epoch": 0.22152994115610938,
238
+ "grad_norm": 1.2426109313964844,
239
+ "learning_rate": 0.0001999426852993044,
240
+ "loss": 0.8021,
241
  "step": 160
242
  },
243
  {
244
+ "epoch": 0.2284527518172378,
245
+ "grad_norm": 1.486345887184143,
246
+ "learning_rate": 0.00019989444416858502,
247
+ "loss": 0.8061,
248
  "step": 165
249
  },
250
  {
251
+ "epoch": 0.23537556247836622,
252
+ "grad_norm": 1.2734565734863281,
253
+ "learning_rate": 0.0001998315961567502,
254
+ "loss": 0.8516,
255
  "step": 170
256
  },
257
  {
258
+ "epoch": 0.24229837313949462,
259
+ "grad_norm": 1.2226619720458984,
260
+ "learning_rate": 0.00019975415045363467,
261
+ "loss": 0.8117,
262
  "step": 175
263
  },
264
  {
265
+ "epoch": 0.24922118380062305,
266
+ "grad_norm": 1.387458324432373,
267
+ "learning_rate": 0.00019966211838359378,
268
+ "loss": 0.8166,
269
  "step": 180
270
  },
271
  {
272
+ "epoch": 0.25614399446175146,
273
+ "grad_norm": 0.9343878030776978,
274
+ "learning_rate": 0.00019955551340384743,
275
+ "loss": 0.8021,
276
  "step": 185
277
  },
278
  {
279
+ "epoch": 0.2630668051228799,
280
+ "grad_norm": 1.0633678436279297,
281
+ "learning_rate": 0.00019943435110251248,
282
+ "loss": 0.7869,
283
  "step": 190
284
  },
285
  {
286
+ "epoch": 0.2699896157840083,
287
+ "grad_norm": 1.1777832508087158,
288
+ "learning_rate": 0.0001992986491963232,
289
+ "loss": 0.8022,
290
  "step": 195
291
  },
292
  {
293
+ "epoch": 0.27691242644513675,
294
+ "grad_norm": 0.8816802501678467,
295
+ "learning_rate": 0.00019914842752804103,
296
+ "loss": 0.7921,
297
  "step": 200
298
  },
299
  {
300
+ "epoch": 0.2838352371062651,
301
+ "grad_norm": 0.9247435331344604,
302
+ "learning_rate": 0.00019898370806355263,
303
+ "loss": 0.778,
304
  "step": 205
305
  },
306
  {
307
+ "epoch": 0.29075804776739356,
308
+ "grad_norm": 1.0109232664108276,
309
+ "learning_rate": 0.00019880451488865836,
310
+ "loss": 0.8086,
311
  "step": 210
312
  },
313
  {
314
+ "epoch": 0.297680858428522,
315
+ "grad_norm": 1.0167481899261475,
316
+ "learning_rate": 0.00019861087420555018,
317
+ "loss": 0.7866,
318
  "step": 215
319
  },
320
  {
321
+ "epoch": 0.3046036690896504,
322
+ "grad_norm": 1.0388667583465576,
323
+ "learning_rate": 0.0001984028143289803,
324
+ "loss": 0.7767,
325
  "step": 220
326
  },
327
  {
328
+ "epoch": 0.3115264797507788,
329
+ "grad_norm": 0.9146580696105957,
330
+ "learning_rate": 0.00019818036568212106,
331
+ "loss": 0.7524,
332
  "step": 225
333
  },
334
  {
335
+ "epoch": 0.3184492904119072,
336
+ "grad_norm": 0.9698460102081299,
337
+ "learning_rate": 0.00019794356079211604,
338
+ "loss": 0.796,
339
  "step": 230
340
  },
341
  {
342
+ "epoch": 0.32537210107303566,
343
+ "grad_norm": 1.009892463684082,
344
+ "learning_rate": 0.00019769243428532422,
345
+ "loss": 0.7565,
346
  "step": 235
347
  },
348
  {
349
+ "epoch": 0.3322949117341641,
350
+ "grad_norm": 0.7838777303695679,
351
+ "learning_rate": 0.00019742702288225652,
352
+ "loss": 0.7672,
353
  "step": 240
354
  },
355
  {
356
+ "epoch": 0.33921772239529246,
357
+ "grad_norm": 0.986925482749939,
358
+ "learning_rate": 0.00019714736539220648,
359
+ "loss": 0.7657,
360
  "step": 245
361
  },
362
  {
363
+ "epoch": 0.3461405330564209,
364
+ "grad_norm": 0.841692328453064,
365
+ "learning_rate": 0.00019685350270757555,
366
+ "loss": 0.7629,
367
  "step": 250
368
  },
369
  {
370
+ "epoch": 0.3530633437175493,
371
+ "grad_norm": 0.8966616988182068,
372
+ "learning_rate": 0.0001965454777978936,
373
+ "loss": 0.7873,
374
  "step": 255
375
  },
376
  {
377
+ "epoch": 0.35998615437867776,
378
+ "grad_norm": 0.7105309963226318,
379
+ "learning_rate": 0.00019622333570353567,
380
+ "loss": 0.765,
381
  "step": 260
382
  },
383
  {
384
+ "epoch": 0.36690896503980613,
385
+ "grad_norm": 0.8212746381759644,
386
+ "learning_rate": 0.00019588712352913625,
387
+ "loss": 0.7588,
388
  "step": 265
389
  },
390
  {
391
+ "epoch": 0.37383177570093457,
392
+ "grad_norm": 0.797985315322876,
393
+ "learning_rate": 0.00019553689043670127,
394
+ "loss": 0.7452,
395
  "step": 270
396
  },
397
  {
398
+ "epoch": 0.380754586362063,
399
+ "grad_norm": 1.5698399543762207,
400
+ "learning_rate": 0.00019517268763841962,
401
+ "loss": 0.7591,
402
  "step": 275
403
  },
404
  {
405
+ "epoch": 0.38767739702319143,
406
+ "grad_norm": 0.9166404008865356,
407
+ "learning_rate": 0.00019479456838917473,
408
+ "loss": 0.797,
409
  "step": 280
410
  },
411
  {
412
+ "epoch": 0.39460020768431986,
413
+ "grad_norm": 1.1828029155731201,
414
+ "learning_rate": 0.0001944025879787574,
415
+ "loss": 0.7941,
416
  "step": 285
417
  },
418
  {
419
+ "epoch": 0.40152301834544823,
420
+ "grad_norm": 0.7243306040763855,
421
+ "learning_rate": 0.0001939968037237812,
422
+ "loss": 0.7537,
423
  "step": 290
424
  },
425
  {
426
+ "epoch": 0.40844582900657667,
427
+ "grad_norm": 0.7559319734573364,
428
+ "learning_rate": 0.00019357727495930146,
429
+ "loss": 0.7473,
430
  "step": 295
431
  },
432
  {
433
+ "epoch": 0.4153686396677051,
434
+ "grad_norm": 0.7109008431434631,
435
+ "learning_rate": 0.00019314406303013904,
436
+ "loss": 0.7415,
437
  "step": 300
438
  },
439
  {
440
+ "epoch": 0.42229145032883353,
441
+ "grad_norm": 0.7651932835578918,
442
+ "learning_rate": 0.00019269723128191048,
443
+ "loss": 0.7427,
444
  "step": 305
445
  },
446
  {
447
+ "epoch": 0.4292142609899619,
448
+ "grad_norm": 0.7459658980369568,
449
+ "learning_rate": 0.00019223684505176517,
450
+ "loss": 0.7347,
451
  "step": 310
452
  },
453
  {
454
+ "epoch": 0.43613707165109034,
455
+ "grad_norm": 0.6931290626525879,
456
+ "learning_rate": 0.00019176297165883165,
457
+ "loss": 0.7363,
458
  "step": 315
459
  },
460
  {
461
+ "epoch": 0.44305988231221877,
462
+ "grad_norm": 0.668066143989563,
463
+ "learning_rate": 0.00019127568039437406,
464
+ "loss": 0.7378,
465
  "step": 320
466
  },
467
  {
468
+ "epoch": 0.4499826929733472,
469
+ "grad_norm": 0.6254391670227051,
470
+ "learning_rate": 0.00019077504251166006,
471
+ "loss": 0.736,
472
  "step": 325
473
  },
474
  {
475
+ "epoch": 0.4569055036344756,
476
+ "grad_norm": 0.7221892476081848,
477
+ "learning_rate": 0.00019026113121554204,
478
+ "loss": 0.7142,
479
  "step": 330
480
  },
481
  {
482
+ "epoch": 0.463828314295604,
483
+ "grad_norm": 0.6113557815551758,
484
+ "learning_rate": 0.00018973402165175268,
485
+ "loss": 0.7381,
486
  "step": 335
487
  },
488
  {
489
+ "epoch": 0.47075112495673244,
490
+ "grad_norm": 0.6476663947105408,
491
+ "learning_rate": 0.0001891937908959172,
492
+ "loss": 0.7383,
493
  "step": 340
494
  },
495
  {
496
+ "epoch": 0.47767393561786087,
497
+ "grad_norm": 0.6006854176521301,
498
+ "learning_rate": 0.00018864051794228282,
499
+ "loss": 0.7244,
500
  "step": 345
501
  },
502
  {
503
+ "epoch": 0.48459674627898924,
504
+ "grad_norm": 0.8741462230682373,
505
+ "learning_rate": 0.00018807428369216822,
506
+ "loss": 0.7092,
507
  "step": 350
508
  },
509
  {
510
+ "epoch": 0.4915195569401177,
511
+ "grad_norm": 0.6464200615882874,
512
+ "learning_rate": 0.00018749517094213375,
513
+ "loss": 0.721,
514
  "step": 355
515
  },
516
  {
517
+ "epoch": 0.4984423676012461,
518
+ "grad_norm": 0.7166573405265808,
519
+ "learning_rate": 0.00018690326437187472,
520
+ "loss": 0.7219,
521
  "step": 360
522
  },
523
  {
524
+ "epoch": 0.5053651782623745,
525
+ "grad_norm": 0.697205662727356,
526
+ "learning_rate": 0.00018629865053183911,
527
+ "loss": 0.7114,
528
  "step": 365
529
  },
530
  {
531
+ "epoch": 0.5122879889235029,
532
+ "grad_norm": 0.7029978036880493,
533
+ "learning_rate": 0.0001856814178305722,
534
+ "loss": 0.7235,
535
  "step": 370
536
  },
537
  {
538
+ "epoch": 0.5192107995846313,
539
+ "grad_norm": 0.6499993801116943,
540
+ "learning_rate": 0.00018505165652178893,
541
+ "loss": 0.7288,
542
  "step": 375
543
  },
544
  {
545
+ "epoch": 0.5261336102457598,
546
+ "grad_norm": 0.6139202117919922,
547
+ "learning_rate": 0.00018440945869117675,
548
+ "loss": 0.7051,
549
  "step": 380
550
  },
551
  {
552
+ "epoch": 0.5330564209068882,
553
+ "grad_norm": 0.6284235715866089,
554
+ "learning_rate": 0.00018375491824293067,
555
+ "loss": 0.7032,
556
  "step": 385
557
  },
558
  {
559
+ "epoch": 0.5399792315680166,
560
+ "grad_norm": 0.6663042902946472,
561
+ "learning_rate": 0.00018308813088602227,
562
+ "loss": 0.7245,
563
  "step": 390
564
  },
565
  {
566
+ "epoch": 0.5469020422291451,
567
+ "grad_norm": 0.5999610424041748,
568
+ "learning_rate": 0.00018240919412020466,
569
+ "loss": 0.6995,
570
  "step": 395
571
  },
572
  {
573
+ "epoch": 0.5538248528902735,
574
+ "grad_norm": 0.5745204091072083,
575
+ "learning_rate": 0.00018171820722175604,
576
+ "loss": 0.6996,
577
  "step": 400
578
  },
579
  {
580
+ "epoch": 0.5607476635514018,
581
+ "grad_norm": 0.6465938091278076,
582
+ "learning_rate": 0.00018101527122896293,
583
+ "loss": 0.7029,
584
  "step": 405
585
  },
586
  {
587
+ "epoch": 0.5676704742125303,
588
+ "grad_norm": 0.5327824354171753,
589
+ "learning_rate": 0.0001803004889273463,
590
+ "loss": 0.6936,
591
  "step": 410
592
  },
593
  {
594
+ "epoch": 0.5745932848736587,
595
+ "grad_norm": 0.6284375786781311,
596
+ "learning_rate": 0.0001795739648346317,
597
+ "loss": 0.7005,
598
  "step": 415
599
  },
600
  {
601
+ "epoch": 0.5815160955347871,
602
+ "grad_norm": 0.6005917191505432,
603
+ "learning_rate": 0.00017883580518546647,
604
+ "loss": 0.672,
605
  "step": 420
606
  },
607
  {
608
+ "epoch": 0.5884389061959155,
609
+ "grad_norm": 0.547201931476593,
610
+ "learning_rate": 0.00017808611791588584,
611
+ "loss": 0.7139,
612
  "step": 425
613
  },
614
  {
615
+ "epoch": 0.595361716857044,
616
+ "grad_norm": 0.593535840511322,
617
+ "learning_rate": 0.00017732501264752996,
618
+ "loss": 0.6997,
619
  "step": 430
620
  },
621
  {
622
+ "epoch": 0.6022845275181724,
623
+ "grad_norm": 0.5183349847793579,
624
+ "learning_rate": 0.00017655260067161503,
625
+ "loss": 0.7047,
626
  "step": 435
627
  },
628
  {
629
+ "epoch": 0.6092073381793008,
630
+ "grad_norm": 0.5909267067909241,
631
+ "learning_rate": 0.00017576899493265954,
632
+ "loss": 0.7078,
633
  "step": 440
634
  },
635
  {
636
+ "epoch": 0.6161301488404292,
637
+ "grad_norm": 0.5590004920959473,
638
+ "learning_rate": 0.00017497431001196943,
639
+ "loss": 0.685,
640
  "step": 445
641
  },
642
  {
643
+ "epoch": 0.6230529595015576,
644
+ "grad_norm": 0.602851927280426,
645
+ "learning_rate": 0.00017416866211088366,
646
+ "loss": 0.7004,
647
  "step": 450
648
  },
649
  {
650
+ "epoch": 0.629975770162686,
651
+ "grad_norm": 0.517782986164093,
652
+ "learning_rate": 0.00017335216903378267,
653
+ "loss": 0.6971,
654
  "step": 455
655
  },
656
  {
657
+ "epoch": 0.6368985808238145,
658
+ "grad_norm": 0.5401471853256226,
659
+ "learning_rate": 0.00017252495017086296,
660
+ "loss": 0.6905,
661
  "step": 460
662
  },
663
  {
664
+ "epoch": 0.6438213914849429,
665
+ "grad_norm": 0.5384804606437683,
666
+ "learning_rate": 0.0001716871264806791,
667
+ "loss": 0.6711,
668
  "step": 465
669
  },
670
  {
671
+ "epoch": 0.6507442021460713,
672
+ "grad_norm": 0.590003490447998,
673
+ "learning_rate": 0.0001708388204724572,
674
+ "loss": 0.692,
675
  "step": 470
676
  },
677
  {
678
+ "epoch": 0.6576670128071997,
679
+ "grad_norm": 0.5600295662879944,
680
+ "learning_rate": 0.00016998015618818087,
681
+ "loss": 0.6847,
682
  "step": 475
683
  },
684
  {
685
+ "epoch": 0.6645898234683282,
686
+ "grad_norm": 0.5578026175498962,
687
+ "learning_rate": 0.00016911125918445363,
688
+ "loss": 0.6914,
689
  "step": 480
690
  },
691
  {
692
+ "epoch": 0.6715126341294566,
693
+ "grad_norm": 0.5562442541122437,
694
+ "learning_rate": 0.00016823225651413953,
695
+ "loss": 0.7038,
696
  "step": 485
697
  },
698
  {
699
+ "epoch": 0.6784354447905849,
700
+ "grad_norm": 0.5268980860710144,
701
+ "learning_rate": 0.00016734327670778508,
702
+ "loss": 0.6762,
703
  "step": 490
704
  },
705
  {
706
+ "epoch": 0.6853582554517134,
707
+ "grad_norm": 0.5298250913619995,
708
+ "learning_rate": 0.00016644444975482523,
709
+ "loss": 0.692,
710
  "step": 495
711
  },
712
  {
713
+ "epoch": 0.6922810661128418,
714
+ "grad_norm": 0.5908562541007996,
715
+ "learning_rate": 0.0001655359070845757,
716
+ "loss": 0.6912,
717
  "step": 500
718
  },
719
  {
720
+ "epoch": 0.6992038767739702,
721
+ "grad_norm": 0.5218640565872192,
722
+ "learning_rate": 0.0001646177815470151,
723
+ "loss": 0.6734,
724
  "step": 505
725
  },
726
  {
727
+ "epoch": 0.7061266874350987,
728
+ "grad_norm": 0.5327192544937134,
729
+ "learning_rate": 0.00016369020739335924,
730
+ "loss": 0.6678,
731
  "step": 510
732
  },
733
  {
734
+ "epoch": 0.7130494980962271,
735
+ "grad_norm": 0.5872364640235901,
736
+ "learning_rate": 0.00016275332025643028,
737
+ "loss": 0.677,
738
  "step": 515
739
  },
740
  {
741
+ "epoch": 0.7199723087573555,
742
+ "grad_norm": 0.5231550335884094,
743
+ "learning_rate": 0.00016180725713082434,
744
+ "loss": 0.6769,
745
  "step": 520
746
  },
747
  {
748
+ "epoch": 0.726895119418484,
749
+ "grad_norm": 0.5114765763282776,
750
+ "learning_rate": 0.0001608521563528795,
751
+ "loss": 0.6834,
752
  "step": 525
753
  },
754
  {
755
+ "epoch": 0.7338179300796123,
756
+ "grad_norm": 0.5015800595283508,
757
+ "learning_rate": 0.00015988815758044792,
758
+ "loss": 0.6591,
759
  "step": 530
760
  },
761
  {
762
+ "epoch": 0.7407407407407407,
763
+ "grad_norm": 0.5894415378570557,
764
+ "learning_rate": 0.00015891540177247467,
765
+ "loss": 0.6857,
766
  "step": 535
767
  },
768
  {
769
+ "epoch": 0.7476635514018691,
770
+ "grad_norm": 0.536401629447937,
771
+ "learning_rate": 0.00015793403116838613,
772
+ "loss": 0.6627,
773
  "step": 540
774
  },
775
  {
776
+ "epoch": 0.7545863620629976,
777
+ "grad_norm": 0.6664223670959473,
778
+ "learning_rate": 0.00015694418926729146,
779
+ "loss": 0.6821,
780
  "step": 545
781
  },
782
  {
783
+ "epoch": 0.761509172724126,
784
+ "grad_norm": 0.49270617961883545,
785
+ "learning_rate": 0.00015594602080699956,
786
+ "loss": 0.6641,
787
  "step": 550
788
  },
789
  {
790
+ "epoch": 0.7684319833852544,
791
+ "grad_norm": 0.5254787802696228,
792
+ "learning_rate": 0.00015493967174285512,
793
+ "loss": 0.6775,
794
  "step": 555
795
  },
796
  {
797
+ "epoch": 0.7753547940463829,
798
+ "grad_norm": 0.5014247298240662,
799
+ "learning_rate": 0.00015392528922639662,
800
+ "loss": 0.6689,
801
  "step": 560
802
  },
803
  {
804
+ "epoch": 0.7822776047075113,
805
+ "grad_norm": 0.5261066555976868,
806
+ "learning_rate": 0.00015290302158383923,
807
+ "loss": 0.6497,
808
  "step": 565
809
  },
810
  {
811
+ "epoch": 0.7892004153686397,
812
+ "grad_norm": 0.5276380777359009,
813
+ "learning_rate": 0.0001518730182943862,
814
+ "loss": 0.6578,
815
  "step": 570
816
  },
817
  {
818
+ "epoch": 0.796123226029768,
819
+ "grad_norm": 0.6668577790260315,
820
+ "learning_rate": 0.0001508354299683715,
821
+ "loss": 0.6569,
822
  "step": 575
823
  },
824
  {
825
+ "epoch": 0.8030460366908965,
826
+ "grad_norm": 0.7964425683021545,
827
+ "learning_rate": 0.0001497904083252371,
828
+ "loss": 0.6571,
829
  "step": 580
830
  },
831
  {
832
+ "epoch": 0.8099688473520249,
833
+ "grad_norm": 0.5345013737678528,
834
+ "learning_rate": 0.00014873810617134806,
835
+ "loss": 0.6758,
836
  "step": 585
837
  },
838
  {
839
+ "epoch": 0.8168916580131533,
840
+ "grad_norm": 0.5370649695396423,
841
+ "learning_rate": 0.00014767867737764876,
842
+ "loss": 0.6769,
843
  "step": 590
844
  },
845
  {
846
+ "epoch": 0.8238144686742818,
847
+ "grad_norm": 0.4770396053791046,
848
+ "learning_rate": 0.0001466122768571633,
849
+ "loss": 0.6445,
850
  "step": 595
851
  },
852
  {
853
+ "epoch": 0.8307372793354102,
854
+ "grad_norm": 0.5163952708244324,
855
+ "learning_rate": 0.0001455390605423437,
856
+ "loss": 0.6647,
857
  "step": 600
858
  },
859
  {
860
+ "epoch": 0.8376600899965386,
861
+ "grad_norm": 0.5729551315307617,
862
+ "learning_rate": 0.00014445918536226904,
863
+ "loss": 0.6372,
864
  "step": 605
865
  },
866
  {
867
+ "epoch": 0.8445829006576671,
868
+ "grad_norm": 0.5001676082611084,
869
+ "learning_rate": 0.00014337280921969858,
870
+ "loss": 0.679,
871
  "step": 610
872
  },
873
  {
874
+ "epoch": 0.8515057113187954,
875
+ "grad_norm": 0.5549310445785522,
876
+ "learning_rate": 0.0001422800909679829,
877
+ "loss": 0.6553,
878
  "step": 615
879
  },
880
  {
881
+ "epoch": 0.8584285219799238,
882
+ "grad_norm": 0.5330554246902466,
883
+ "learning_rate": 0.00014118119038783588,
884
+ "loss": 0.6484,
885
  "step": 620
886
  },
887
  {
888
+ "epoch": 0.8653513326410522,
889
+ "grad_norm": 0.570789098739624,
890
+ "learning_rate": 0.00014007626816397087,
891
+ "loss": 0.6596,
892
  "step": 625
893
  },
894
  {
895
+ "epoch": 0.8722741433021807,
896
+ "grad_norm": 0.547938883304596,
897
+ "learning_rate": 0.00013896548586160503,
898
+ "loss": 0.6833,
899
  "step": 630
900
  },
901
  {
902
+ "epoch": 0.8791969539633091,
903
+ "grad_norm": 0.4971727132797241,
904
+ "learning_rate": 0.00013784900590283473,
905
+ "loss": 0.6548,
906
  "step": 635
907
  },
908
  {
909
+ "epoch": 0.8861197646244375,
910
+ "grad_norm": 0.48353663086891174,
911
+ "learning_rate": 0.00013672699154288569,
912
+ "loss": 0.6422,
913
  "step": 640
914
  },
915
  {
916
+ "epoch": 0.893042575285566,
917
+ "grad_norm": 0.560279905796051,
918
+ "learning_rate": 0.00013559960684624112,
919
+ "loss": 0.6549,
920
  "step": 645
921
  },
922
  {
923
+ "epoch": 0.8999653859466944,
924
+ "grad_norm": 0.5509752035140991,
925
+ "learning_rate": 0.0001344670166626519,
926
+ "loss": 0.6635,
927
  "step": 650
928
  },
929
  {
930
+ "epoch": 0.9068881966078228,
931
+ "grad_norm": 0.510706901550293,
932
+ "learning_rate": 0.0001333293866030317,
933
+ "loss": 0.6626,
934
  "step": 655
935
  },
936
  {
937
+ "epoch": 0.9138110072689511,
938
+ "grad_norm": 0.5331857800483704,
939
+ "learning_rate": 0.0001321868830152407,
940
+ "loss": 0.6495,
941
  "step": 660
942
  },
943
  {
944
+ "epoch": 0.9207338179300796,
945
+ "grad_norm": 0.5860617756843567,
946
+ "learning_rate": 0.00013103967295976179,
947
+ "loss": 0.6433,
948
  "step": 665
949
  },
950
  {
951
+ "epoch": 0.927656628591208,
952
+ "grad_norm": 0.45142075419425964,
953
+ "learning_rate": 0.0001298879241852724,
954
+ "loss": 0.6499,
955
  "step": 670
956
  },
957
  {
958
+ "epoch": 0.9345794392523364,
959
+ "grad_norm": 0.5519718527793884,
960
+ "learning_rate": 0.0001287318051041159,
961
+ "loss": 0.6366,
962
  "step": 675
963
  },
964
  {
965
+ "epoch": 0.9415022499134649,
966
+ "grad_norm": 0.5027467012405396,
967
+ "learning_rate": 0.00012757148476767553,
968
+ "loss": 0.6445,
969
  "step": 680
970
  },
971
  {
972
+ "epoch": 0.9484250605745933,
973
+ "grad_norm": 0.485603004693985,
974
+ "learning_rate": 0.0001264071328416555,
975
+ "loss": 0.6352,
976
  "step": 685
977
  },
978
  {
979
+ "epoch": 0.9553478712357217,
980
+ "grad_norm": 0.488096684217453,
981
+ "learning_rate": 0.00012523891958127168,
982
+ "loss": 0.6673,
983
  "step": 690
984
  },
985
  {
986
+ "epoch": 0.9622706818968502,
987
+ "grad_norm": 0.49362048506736755,
988
+ "learning_rate": 0.0001240670158063565,
989
+ "loss": 0.6496,
990
  "step": 695
991
  },
992
  {
993
+ "epoch": 0.9691934925579785,
994
+ "grad_norm": 0.49509820342063904,
995
+ "learning_rate": 0.00012289159287638105,
996
+ "loss": 0.6389,
997
  "step": 700
998
  },
999
  {
1000
+ "epoch": 0.9761163032191069,
1001
+ "grad_norm": 0.502849280834198,
1002
+ "learning_rate": 0.00012171282266539845,
1003
+ "loss": 0.6558,
1004
  "step": 705
1005
  },
1006
  {
1007
+ "epoch": 0.9830391138802353,
1008
+ "grad_norm": 0.4513024389743805,
1009
+ "learning_rate": 0.00012053087753691172,
1010
+ "loss": 0.6454,
1011
  "step": 710
1012
  },
1013
  {
1014
+ "epoch": 0.9899619245413638,
1015
+ "grad_norm": 0.47943952679634094,
1016
+ "learning_rate": 0.00011934593031867038,
1017
+ "loss": 0.6506,
1018
  "step": 715
1019
  },
1020
  {
1021
+ "epoch": 0.9968847352024922,
1022
+ "grad_norm": 0.48262372612953186,
1023
+ "learning_rate": 0.00011815815427739911,
1024
+ "loss": 0.638,
1025
  "step": 720
1026
  },
1027
  {
1028
+ "epoch": 1.0,
1029
+ "eval_loss": 0.6328259110450745,
1030
+ "eval_runtime": 78.8724,
1031
+ "eval_samples_per_second": 26.232,
1032
+ "eval_steps_per_second": 1.648,
1033
+ "step": 723
1034
+ },
1035
+ {
1036
+ "epoch": 1.0027691242644514,
1037
+ "grad_norm": 0.48229867219924927,
1038
+ "learning_rate": 0.00011696772309346182,
1039
+ "loss": 0.5821,
1040
  "step": 725
1041
  },
1042
  {
1043
+ "epoch": 1.0096919349255797,
1044
+ "grad_norm": 0.5330725908279419,
1045
+ "learning_rate": 0.0001157748108354659,
1046
+ "loss": 0.5198,
1047
  "step": 730
1048
  },
1049
  {
1050
+ "epoch": 1.0166147455867083,
1051
+ "grad_norm": 0.4326479434967041,
1052
+ "learning_rate": 0.00011457959193480925,
1053
+ "loss": 0.5137,
1054
  "step": 735
1055
  },
1056
  {
1057
+ "epoch": 1.0235375562478366,
1058
+ "grad_norm": 0.4757113456726074,
1059
+ "learning_rate": 0.00011338224116017423,
1060
+ "loss": 0.5175,
1061
  "step": 740
1062
  },
1063
  {
1064
+ "epoch": 1.0304603669089651,
1065
+ "grad_norm": 0.46214616298675537,
1066
+ "learning_rate": 0.00011218293359197267,
1067
+ "loss": 0.5259,
1068
  "step": 745
1069
  },
1070
  {
1071
+ "epoch": 1.0373831775700935,
1072
+ "grad_norm": 0.5037508010864258,
1073
+ "learning_rate": 0.00011098184459674484,
1074
+ "loss": 0.5087,
1075
  "step": 750
1076
  },
1077
  {
1078
+ "epoch": 1.0443059882312218,
1079
+ "grad_norm": 0.46930187940597534,
1080
+ "learning_rate": 0.00010977914980151691,
1081
+ "loss": 0.5119,
1082
  "step": 755
1083
  },
1084
  {
1085
+ "epoch": 1.0512287988923503,
1086
+ "grad_norm": 0.4699100852012634,
1087
+ "learning_rate": 0.00010857502506812029,
1088
+ "loss": 0.4873,
1089
  "step": 760
1090
  },
1091
  {
1092
+ "epoch": 1.0581516095534786,
1093
+ "grad_norm": 0.43981245160102844,
1094
+ "learning_rate": 0.00010736964646747629,
1095
+ "loss": 0.4876,
1096
  "step": 765
1097
  },
1098
  {
1099
+ "epoch": 1.0650744202146072,
1100
+ "grad_norm": 0.489309698343277,
1101
+ "learning_rate": 0.00010616319025385089,
1102
+ "loss": 0.5049,
1103
  "step": 770
1104
  },
1105
  {
1106
+ "epoch": 1.0719972308757355,
1107
+ "grad_norm": 0.44854122400283813,
1108
+ "learning_rate": 0.00010495583283908177,
1109
+ "loss": 0.5125,
1110
  "step": 775
1111
  },
1112
  {
1113
+ "epoch": 1.078920041536864,
1114
+ "grad_norm": 0.3934454023838043,
1115
+ "learning_rate": 0.00010374775076678332,
1116
+ "loss": 0.5008,
1117
  "step": 780
1118
  },
1119
  {
1120
+ "epoch": 1.0858428521979924,
1121
+ "grad_norm": 0.4133076071739197,
1122
+ "learning_rate": 0.00010253912068653146,
1123
+ "loss": 0.5119,
1124
  "step": 785
1125
  },
1126
  {
1127
+ "epoch": 1.092765662859121,
1128
+ "grad_norm": 0.4223884642124176,
1129
+ "learning_rate": 0.00010133011932803378,
1130
+ "loss": 0.5128,
1131
  "step": 790
1132
  },
1133
  {
1134
+ "epoch": 1.0996884735202492,
1135
+ "grad_norm": 0.44975772500038147,
1136
+ "learning_rate": 0.00010012092347528705,
1137
+ "loss": 0.5166,
1138
  "step": 795
1139
  },
1140
  {
1141
+ "epoch": 1.1066112841813776,
1142
+ "grad_norm": 0.5160712599754333,
1143
+ "learning_rate": 9.891170994072793e-05,
1144
+ "loss": 0.5023,
1145
  "step": 800
1146
  },
1147
  {
1148
+ "epoch": 1.113534094842506,
1149
+ "grad_norm": 0.4700876772403717,
1150
+ "learning_rate": 9.770265553937831e-05,
1151
+ "loss": 0.5031,
1152
  "step": 805
1153
  },
1154
  {
1155
+ "epoch": 1.1204569055036344,
1156
+ "grad_norm": 0.44749587774276733,
1157
+ "learning_rate": 9.649393706299127e-05,
1158
+ "loss": 0.5031,
1159
  "step": 810
1160
  },
1161
  {
1162
+ "epoch": 1.127379716164763,
1163
+ "grad_norm": 0.44761285185813904,
1164
+ "learning_rate": 9.52857312541998e-05,
1165
+ "loss": 0.4932,
1166
  "step": 815
1167
  },
1168
  {
1169
+ "epoch": 1.1343025268258913,
1170
+ "grad_norm": 0.41243189573287964,
1171
+ "learning_rate": 9.407821478067304e-05,
1172
+ "loss": 0.4908,
1173
  "step": 820
1174
  },
1175
  {
1176
+ "epoch": 1.1412253374870198,
1177
+ "grad_norm": 0.4481096565723419,
1178
+ "learning_rate": 9.287156420928342e-05,
1179
+ "loss": 0.4906,
1180
  "step": 825
1181
  },
1182
  {
1183
+ "epoch": 1.1481481481481481,
1184
+ "grad_norm": 0.4322609007358551,
1185
+ "learning_rate": 9.166595598028832e-05,
1186
+ "loss": 0.5143,
1187
  "step": 830
1188
  },
1189
  {
1190
+ "epoch": 1.1550709588092767,
1191
+ "grad_norm": 0.46154677867889404,
1192
+ "learning_rate": 9.046156638153056e-05,
1193
+ "loss": 0.493,
1194
  "step": 835
1195
  },
1196
  {
1197
+ "epoch": 1.161993769470405,
1198
+ "grad_norm": 0.4196186065673828,
1199
+ "learning_rate": 8.925857152266103e-05,
1200
+ "loss": 0.4856,
1201
  "step": 840
1202
  },
1203
  {
1204
+ "epoch": 1.1689165801315333,
1205
+ "grad_norm": 0.47523409128189087,
1206
+ "learning_rate": 8.805714730938728e-05,
1207
+ "loss": 0.4978,
1208
  "step": 845
1209
  },
1210
  {
1211
+ "epoch": 1.1758393907926619,
1212
+ "grad_norm": 0.38712936639785767,
1213
+ "learning_rate": 8.685746941775219e-05,
1214
+ "loss": 0.4925,
1215
  "step": 850
1216
  },
1217
  {
1218
+ "epoch": 1.1827622014537902,
1219
+ "grad_norm": 0.4324866533279419,
1220
+ "learning_rate": 8.565971326844584e-05,
1221
+ "loss": 0.503,
1222
  "step": 855
1223
  },
1224
  {
1225
+ "epoch": 1.1896850121149187,
1226
+ "grad_norm": 0.41721028089523315,
1227
+ "learning_rate": 8.44640540011553e-05,
1228
+ "loss": 0.5095,
1229
  "step": 860
1230
  },
1231
  {
1232
+ "epoch": 1.196607822776047,
1233
+ "grad_norm": 0.40640369057655334,
1234
+ "learning_rate": 8.327066644895469e-05,
1235
+ "loss": 0.4946,
1236
  "step": 865
1237
  },
1238
  {
1239
+ "epoch": 1.2035306334371756,
1240
+ "grad_norm": 0.4490497410297394,
1241
+ "learning_rate": 8.207972511274095e-05,
1242
+ "loss": 0.5101,
1243
  "step": 870
1244
  },
1245
  {
1246
+ "epoch": 1.210453444098304,
1247
+ "grad_norm": 0.4271405339241028,
1248
+ "learning_rate": 8.089140413571747e-05,
1249
+ "loss": 0.5019,
1250
  "step": 875
1251
  },
1252
  {
1253
+ "epoch": 1.2173762547594325,
1254
+ "grad_norm": 0.4564451277256012,
1255
+ "learning_rate": 7.970587727793051e-05,
1256
+ "loss": 0.4879,
1257
  "step": 880
1258
  },
1259
  {
1260
+ "epoch": 1.2242990654205608,
1261
+ "grad_norm": 0.4469379186630249,
1262
+ "learning_rate": 7.852331789086135e-05,
1263
+ "loss": 0.4893,
1264
  "step": 885
1265
  },
1266
  {
1267
+ "epoch": 1.231221876081689,
1268
+ "grad_norm": 0.4078274667263031,
1269
+ "learning_rate": 7.73438988920784e-05,
1270
+ "loss": 0.4967,
1271
  "step": 890
1272
  },
1273
  {
1274
+ "epoch": 1.2381446867428176,
1275
+ "grad_norm": 0.44297975301742554,
1276
+ "learning_rate": 7.616779273995252e-05,
1277
+ "loss": 0.5057,
1278
  "step": 895
1279
  },
1280
  {
1281
+ "epoch": 1.245067497403946,
1282
+ "grad_norm": 0.41071122884750366,
1283
+ "learning_rate": 7.499517140843968e-05,
1284
+ "loss": 0.5024,
1285
  "step": 900
1286
  },
1287
  {
1288
+ "epoch": 1.2519903080650745,
1289
+ "grad_norm": 0.4083300232887268,
1290
+ "learning_rate": 7.382620636193438e-05,
1291
+ "loss": 0.4814,
1292
  "step": 905
1293
  },
1294
  {
1295
+ "epoch": 1.2589131187262028,
1296
+ "grad_norm": 0.4223071038722992,
1297
+ "learning_rate": 7.266106853019759e-05,
1298
+ "loss": 0.49,
1299
  "step": 910
1300
  },
1301
  {
1302
+ "epoch": 1.2658359293873311,
1303
+ "grad_norm": 0.42086276412010193,
1304
+ "learning_rate": 7.149992828336274e-05,
1305
+ "loss": 0.4989,
1306
  "step": 915
1307
  },
1308
  {
1309
+ "epoch": 1.2727587400484597,
1310
+ "grad_norm": 0.41196486353874207,
1311
+ "learning_rate": 7.034295540702397e-05,
1312
+ "loss": 0.4854,
1313
  "step": 920
1314
  },
1315
  {
1316
+ "epoch": 1.2796815507095882,
1317
+ "grad_norm": 0.4233442544937134,
1318
+ "learning_rate": 6.919031907740922e-05,
1319
+ "loss": 0.4959,
1320
  "step": 925
1321
  },
1322
  {
1323
+ "epoch": 1.2866043613707165,
1324
+ "grad_norm": 0.4324461817741394,
1325
+ "learning_rate": 6.804218783664288e-05,
1326
+ "loss": 0.5002,
1327
  "step": 930
1328
  },
1329
  {
1330
+ "epoch": 1.2935271720318449,
1331
+ "grad_norm": 0.43228667974472046,
1332
+ "learning_rate": 6.689872956810102e-05,
1333
+ "loss": 0.4947,
1334
  "step": 935
1335
  },
1336
  {
1337
+ "epoch": 1.3004499826929734,
1338
+ "grad_norm": 0.4018300175666809,
1339
+ "learning_rate": 6.5760111471863e-05,
1340
+ "loss": 0.4916,
1341
  "step": 940
1342
  },
1343
  {
1344
+ "epoch": 1.3073727933541017,
1345
+ "grad_norm": 0.42353230714797974,
1346
+ "learning_rate": 6.46265000402628e-05,
1347
+ "loss": 0.498,
1348
  "step": 945
1349
  },
1350
  {
1351
+ "epoch": 1.3142956040152303,
1352
+ "grad_norm": 0.4208320677280426,
1353
+ "learning_rate": 6.349806103354417e-05,
1354
+ "loss": 0.4892,
1355
  "step": 950
1356
  },
1357
  {
1358
+ "epoch": 1.3212184146763586,
1359
+ "grad_norm": 0.4484878182411194,
1360
+ "learning_rate": 6.23749594556226e-05,
1361
+ "loss": 0.4834,
1362
  "step": 955
1363
  },
1364
  {
1365
+ "epoch": 1.328141225337487,
1366
+ "grad_norm": 0.4080387055873871,
1367
+ "learning_rate": 6.125735952995775e-05,
1368
+ "loss": 0.4768,
1369
  "step": 960
1370
  },
1371
  {
1372
+ "epoch": 1.3350640359986155,
1373
+ "grad_norm": 0.4215630888938904,
1374
+ "learning_rate": 6.0145424675540394e-05,
1375
+ "loss": 0.4864,
1376
  "step": 965
1377
  },
1378
  {
1379
+ "epoch": 1.341986846659744,
1380
+ "grad_norm": 0.40287894010543823,
1381
+ "learning_rate": 5.9039317482996735e-05,
1382
+ "loss": 0.492,
1383
  "step": 970
1384
  },
1385
  {
1386
+ "epoch": 1.3489096573208723,
1387
+ "grad_norm": 0.41661080718040466,
1388
+ "learning_rate": 5.793919969081374e-05,
1389
+ "loss": 0.479,
1390
  "step": 975
1391
  },
1392
  {
1393
+ "epoch": 1.3558324679820006,
1394
+ "grad_norm": 0.3967858552932739,
1395
+ "learning_rate": 5.68452321616894e-05,
1396
+ "loss": 0.471,
1397
  "step": 980
1398
  },
1399
  {
1400
+ "epoch": 1.3627552786431292,
1401
+ "grad_norm": 0.39763662219047546,
1402
+ "learning_rate": 5.5757574859010886e-05,
1403
+ "loss": 0.4841,
1404
  "step": 985
1405
  },
1406
  {
1407
+ "epoch": 1.3696780893042575,
1408
+ "grad_norm": 0.40723222494125366,
1409
+ "learning_rate": 5.467638682346403e-05,
1410
+ "loss": 0.4677,
1411
  "step": 990
1412
  },
1413
  {
1414
+ "epoch": 1.376600899965386,
1415
+ "grad_norm": 0.4053124189376831,
1416
+ "learning_rate": 5.3601826149777966e-05,
1417
+ "loss": 0.4863,
1418
  "step": 995
1419
  },
1420
  {
1421
+ "epoch": 1.3835237106265144,
1422
+ "grad_norm": 0.4242284893989563,
1423
+ "learning_rate": 5.25340499636082e-05,
1424
+ "loss": 0.4744,
1425
  "step": 1000
1426
  },
1427
  {
1428
+ "epoch": 1.3904465212876427,
1429
+ "grad_norm": 0.41244375705718994,
1430
+ "learning_rate": 5.147321439856091e-05,
1431
+ "loss": 0.484,
1432
  "step": 1005
1433
  },
1434
  {
1435
+ "epoch": 1.3973693319487712,
1436
+ "grad_norm": 0.3905507028102875,
1437
+ "learning_rate": 5.041947457336274e-05,
1438
+ "loss": 0.4967,
1439
  "step": 1010
1440
  },
1441
  {
1442
+ "epoch": 1.4042921426098995,
1443
+ "grad_norm": 0.40260571241378784,
1444
+ "learning_rate": 4.937298456917889e-05,
1445
+ "loss": 0.4867,
1446
  "step": 1015
1447
  },
1448
  {
1449
+ "epoch": 1.411214953271028,
1450
+ "grad_norm": 0.440318763256073,
1451
+ "learning_rate": 4.8333897407082896e-05,
1452
+ "loss": 0.4799,
1453
  "step": 1020
1454
  },
1455
  {
1456
+ "epoch": 1.4181377639321564,
1457
+ "grad_norm": 0.37970152497291565,
1458
+ "learning_rate": 4.7302365025681206e-05,
1459
+ "loss": 0.4829,
1460
  "step": 1025
1461
  },
1462
  {
1463
+ "epoch": 1.425060574593285,
1464
+ "grad_norm": 0.41383683681488037,
1465
+ "learning_rate": 4.627853825889642e-05,
1466
+ "loss": 0.4838,
1467
  "step": 1030
1468
  },
1469
  {
1470
+ "epoch": 1.4319833852544133,
1471
+ "grad_norm": 0.4296334385871887,
1472
+ "learning_rate": 4.526256681391192e-05,
1473
+ "loss": 0.4723,
1474
  "step": 1035
1475
  },
1476
  {
1477
+ "epoch": 1.4389061959155418,
1478
+ "grad_norm": 0.4446117579936981,
1479
+ "learning_rate": 4.4254599249281016e-05,
1480
+ "loss": 0.4577,
1481
  "step": 1040
1482
  },
1483
  {
1484
+ "epoch": 1.4458290065766701,
1485
+ "grad_norm": 0.41226211190223694,
1486
+ "learning_rate": 4.3254782953204375e-05,
1487
+ "loss": 0.4737,
1488
  "step": 1045
1489
  },
1490
  {
1491
+ "epoch": 1.4527518172377984,
1492
+ "grad_norm": 0.41757309436798096,
1493
+ "learning_rate": 4.2263264121978505e-05,
1494
+ "loss": 0.4884,
1495
  "step": 1050
1496
  },
1497
  {
1498
+ "epoch": 1.459674627898927,
1499
+ "grad_norm": 0.38596871495246887,
1500
+ "learning_rate": 4.12801877386183e-05,
1501
+ "loss": 0.4634,
1502
  "step": 1055
1503
  },
1504
  {
1505
+ "epoch": 1.4665974385600553,
1506
+ "grad_norm": 0.4125303328037262,
1507
+ "learning_rate": 4.030569755165736e-05,
1508
+ "loss": 0.4775,
1509
  "step": 1060
1510
  },
1511
  {
1512
+ "epoch": 1.4735202492211839,
1513
+ "grad_norm": 0.3986310362815857,
1514
+ "learning_rate": 3.9339936054128466e-05,
1515
+ "loss": 0.4686,
1516
  "step": 1065
1517
  },
1518
  {
1519
+ "epoch": 1.4804430598823122,
1520
+ "grad_norm": 0.39551883935928345,
1521
+ "learning_rate": 3.8383044462727826e-05,
1522
+ "loss": 0.4571,
1523
  "step": 1070
1524
  },
1525
  {
1526
+ "epoch": 1.4873658705434407,
1527
+ "grad_norm": 0.4369433522224426,
1528
+ "learning_rate": 3.7435162697166107e-05,
1529
+ "loss": 0.4761,
1530
  "step": 1075
1531
  },
1532
  {
1533
+ "epoch": 1.494288681204569,
1534
+ "grad_norm": 0.40035200119018555,
1535
+ "learning_rate": 3.649642935970859e-05,
1536
+ "loss": 0.4767,
1537
  "step": 1080
1538
  },
1539
  {
1540
+ "epoch": 1.5012114918656976,
1541
+ "grad_norm": 0.3791543245315552,
1542
+ "learning_rate": 3.556698171490871e-05,
1543
+ "loss": 0.4664,
1544
  "step": 1085
1545
  },
1546
  {
1547
+ "epoch": 1.508134302526826,
1548
+ "grad_norm": 0.3993076980113983,
1549
+ "learning_rate": 3.464695566953644e-05,
1550
+ "loss": 0.4803,
1551
  "step": 1090
1552
  },
1553
  {
1554
+ "epoch": 1.5150571131879542,
1555
+ "grad_norm": 0.42704322934150696,
1556
+ "learning_rate": 3.373648575270576e-05,
1557
+ "loss": 0.4776,
1558
  "step": 1095
1559
  },
1560
  {
1561
+ "epoch": 1.5219799238490828,
1562
+ "grad_norm": 0.4107125997543335,
1563
+ "learning_rate": 3.283570509620344e-05,
1564
+ "loss": 0.4777,
1565
  "step": 1100
1566
  },
1567
  {
1568
+ "epoch": 1.5289027345102113,
1569
+ "grad_norm": 0.3953048884868622,
1570
+ "learning_rate": 3.1944745415021915e-05,
1571
+ "loss": 0.468,
1572
  "step": 1105
1573
  },
1574
  {
1575
+ "epoch": 1.5358255451713396,
1576
+ "grad_norm": 0.3960942327976227,
1577
+ "learning_rate": 3.106373698809958e-05,
1578
+ "loss": 0.4631,
1579
  "step": 1110
1580
  },
1581
  {
1582
+ "epoch": 1.542748355832468,
1583
+ "grad_norm": 0.3955947458744049,
1584
+ "learning_rate": 3.0192808639271065e-05,
1585
+ "loss": 0.4707,
1586
  "step": 1115
1587
  },
1588
  {
1589
+ "epoch": 1.5496711664935963,
1590
+ "grad_norm": 0.3740635812282562,
1591
+ "learning_rate": 2.933208771843018e-05,
1592
+ "loss": 0.4793,
1593
  "step": 1120
1594
  },
1595
  {
1596
+ "epoch": 1.5565939771547248,
1597
+ "grad_norm": 0.3782494068145752,
1598
+ "learning_rate": 2.8481700082908268e-05,
1599
+ "loss": 0.4623,
1600
  "step": 1125
1601
  },
1602
  {
1603
+ "epoch": 1.5635167878158533,
1604
+ "grad_norm": 0.3642530143260956,
1605
+ "learning_rate": 2.7641770079071127e-05,
1606
+ "loss": 0.4565,
1607
  "step": 1130
1608
  },
1609
  {
1610
+ "epoch": 1.5704395984769817,
1611
+ "grad_norm": 0.3924564719200134,
1612
+ "learning_rate": 2.681242052413656e-05,
1613
+ "loss": 0.4599,
1614
  "step": 1135
1615
  },
1616
  {
1617
+ "epoch": 1.57736240913811,
1618
+ "grad_norm": 0.38727226853370667,
1619
+ "learning_rate": 2.59937726882158e-05,
1620
+ "loss": 0.466,
1621
  "step": 1140
1622
  },
1623
  {
1624
+ "epoch": 1.5842852197992385,
1625
+ "grad_norm": 0.40863728523254395,
1626
+ "learning_rate": 2.5185946276580918e-05,
1627
+ "loss": 0.4674,
1628
  "step": 1145
1629
  },
1630
  {
1631
+ "epoch": 1.591208030460367,
1632
+ "grad_norm": 0.4037117660045624,
1633
+ "learning_rate": 2.4389059412161087e-05,
1634
+ "loss": 0.4654,
1635
  "step": 1150
1636
  },
1637
  {
1638
+ "epoch": 1.5981308411214954,
1639
+ "grad_norm": 0.3873690962791443,
1640
+ "learning_rate": 2.3603228618270356e-05,
1641
+ "loss": 0.4678,
1642
  "step": 1155
1643
  },
1644
  {
1645
+ "epoch": 1.6050536517826237,
1646
+ "grad_norm": 0.41405242681503296,
1647
+ "learning_rate": 2.2828568801569283e-05,
1648
+ "loss": 0.4472,
1649
  "step": 1160
1650
  },
1651
  {
1652
+ "epoch": 1.611976462443752,
1653
+ "grad_norm": 0.39564672112464905,
1654
+ "learning_rate": 2.2065193235262782e-05,
1655
+ "loss": 0.462,
1656
  "step": 1165
1657
  },
1658
  {
1659
+ "epoch": 1.6188992731048806,
1660
+ "grad_norm": 0.4193384349346161,
1661
+ "learning_rate": 2.1313213542537058e-05,
1662
+ "loss": 0.4574,
1663
  "step": 1170
1664
  },
1665
  {
1666
+ "epoch": 1.6258220837660091,
1667
+ "grad_norm": 0.42125841975212097,
1668
+ "learning_rate": 2.0572739680237717e-05,
1669
+ "loss": 0.4693,
1670
  "step": 1175
1671
  },
1672
  {
1673
+ "epoch": 1.6327448944271374,
1674
+ "grad_norm": 0.38997945189476013,
1675
+ "learning_rate": 1.984387992279153e-05,
1676
+ "loss": 0.4674,
1677
  "step": 1180
1678
  },
1679
  {
1680
+ "epoch": 1.6396677050882658,
1681
+ "grad_norm": 0.4101271331310272,
1682
+ "learning_rate": 1.912674084637409e-05,
1683
+ "loss": 0.453,
1684
  "step": 1185
1685
  },
1686
  {
1687
+ "epoch": 1.6465905157493943,
1688
+ "grad_norm": 0.40208595991134644,
1689
+ "learning_rate": 1.8421427313326046e-05,
1690
+ "loss": 0.461,
1691
  "step": 1190
1692
  },
1693
  {
1694
+ "epoch": 1.6535133264105226,
1695
+ "grad_norm": 0.4130416810512543,
1696
+ "learning_rate": 1.7728042456819704e-05,
1697
+ "loss": 0.4585,
1698
  "step": 1195
1699
  },
1700
  {
1701
+ "epoch": 1.6604361370716512,
1702
+ "grad_norm": 0.39520734548568726,
1703
+ "learning_rate": 1.7046687665778717e-05,
1704
+ "loss": 0.4548,
1705
  "step": 1200
1706
  },
1707
  {
1708
+ "epoch": 1.6673589477327795,
1709
+ "grad_norm": 0.3736555576324463,
1710
+ "learning_rate": 1.6377462570052438e-05,
1711
+ "loss": 0.454,
1712
  "step": 1205
1713
  },
1714
  {
1715
+ "epoch": 1.6742817583939078,
1716
+ "grad_norm": 0.3883906304836273,
1717
+ "learning_rate": 1.5720465025847987e-05,
1718
+ "loss": 0.459,
1719
  "step": 1210
1720
  },
1721
  {
1722
+ "epoch": 1.6812045690550363,
1723
+ "grad_norm": 0.3996312618255615,
1724
+ "learning_rate": 1.5075791101421166e-05,
1725
+ "loss": 0.4559,
1726
  "step": 1215
1727
  },
1728
  {
1729
+ "epoch": 1.6881273797161649,
1730
+ "grad_norm": 0.3835405707359314,
1731
+ "learning_rate": 1.4443535063029279e-05,
1732
+ "loss": 0.4531,
1733
  "step": 1220
1734
  },
1735
  {
1736
+ "epoch": 1.6950501903772932,
1737
+ "grad_norm": 0.40183380246162415,
1738
+ "learning_rate": 1.3823789361147044e-05,
1739
+ "loss": 0.4564,
1740
  "step": 1225
1741
  },
1742
  {
1743
+ "epoch": 1.7019730010384215,
1744
+ "grad_norm": 0.36577045917510986,
1745
+ "learning_rate": 1.3216644616948282e-05,
1746
+ "loss": 0.4562,
1747
  "step": 1230
1748
  },
1749
  {
1750
+ "epoch": 1.70889581169955,
1751
+ "grad_norm": 0.38610365986824036,
1752
+ "learning_rate": 1.2622189609054979e-05,
1753
+ "loss": 0.4517,
1754
  "step": 1235
1755
  },
1756
  {
1757
+ "epoch": 1.7158186223606784,
1758
+ "grad_norm": 0.3951961100101471,
1759
+ "learning_rate": 1.2040511260555831e-05,
1760
+ "loss": 0.4441,
1761
  "step": 1240
1762
  },
1763
  {
1764
+ "epoch": 1.722741433021807,
1765
+ "grad_norm": 0.4014803171157837,
1766
+ "learning_rate": 1.1471694626296049e-05,
1767
+ "loss": 0.4511,
1768
  "step": 1245
1769
  },
1770
  {
1771
+ "epoch": 1.7296642436829353,
1772
+ "grad_norm": 0.4027356505393982,
1773
+ "learning_rate": 1.0915822880440308e-05,
1774
+ "loss": 0.4432,
1775
  "step": 1250
1776
  },
1777
  {
1778
+ "epoch": 1.7365870543440636,
1779
+ "grad_norm": 0.39167362451553345,
1780
+ "learning_rate": 1.037297730431085e-05,
1781
+ "loss": 0.4327,
1782
  "step": 1255
1783
  },
1784
  {
1785
+ "epoch": 1.7435098650051921,
1786
+ "grad_norm": 0.3852311968803406,
1787
+ "learning_rate": 9.843237274502282e-06,
1788
+ "loss": 0.4613,
1789
  "step": 1260
1790
  },
1791
  {
1792
+ "epoch": 1.7504326756663207,
1793
+ "grad_norm": 0.41405415534973145,
1794
+ "learning_rate": 9.326680251274777e-06,
1795
+ "loss": 0.4442,
1796
  "step": 1265
1797
  },
1798
  {
1799
+ "epoch": 1.757355486327449,
1800
+ "grad_norm": 0.405292809009552,
1801
+ "learning_rate": 8.823381767227667e-06,
1802
+ "loss": 0.4536,
1803
  "step": 1270
1804
  },
1805
  {
1806
+ "epoch": 1.7642782969885773,
1807
+ "grad_norm": 0.3843272924423218,
1808
+ "learning_rate": 8.333415416254831e-06,
1809
+ "loss": 0.454,
1810
  "step": 1275
1811
  },
1812
  {
1813
+ "epoch": 1.7712011076497056,
1814
+ "grad_norm": 0.3779032826423645,
1815
+ "learning_rate": 7.856852842783547e-06,
1816
+ "loss": 0.457,
1817
  "step": 1280
1818
  },
1819
  {
1820
+ "epoch": 1.7781239183108342,
1821
+ "grad_norm": 0.38312792778015137,
1822
+ "learning_rate": 7.393763731298387e-06,
1823
+ "loss": 0.4701,
1824
  "step": 1285
1825
  },
1826
  {
1827
+ "epoch": 1.7850467289719627,
1828
+ "grad_norm": 0.3812410831451416,
1829
+ "learning_rate": 6.944215796151765e-06,
1830
+ "loss": 0.4484,
1831
  "step": 1290
1832
  },
1833
  {
1834
+ "epoch": 1.791969539633091,
1835
+ "grad_norm": 0.3971448242664337,
1836
+ "learning_rate": 6.5082747716625255e-06,
1837
+ "loss": 0.4425,
1838
  "step": 1295
1839
  },
1840
  {
1841
+ "epoch": 1.7988923502942193,
1842
+ "grad_norm": 0.3934316039085388,
1843
+ "learning_rate": 6.086004402504098e-06,
1844
+ "loss": 0.4563,
1845
  "step": 1300
1846
  },
1847
  {
1848
+ "epoch": 1.8058151609553479,
1849
+ "grad_norm": 0.38482704758644104,
1850
+ "learning_rate": 5.6774664343834495e-06,
1851
+ "loss": 0.4531,
1852
  "step": 1305
1853
  },
1854
  {
1855
+ "epoch": 1.8127379716164764,
1856
+ "grad_norm": 0.38579967617988586,
1857
+ "learning_rate": 5.2827206050125585e-06,
1858
+ "loss": 0.4631,
1859
  "step": 1310
1860
  },
1861
  {
1862
+ "epoch": 1.8196607822776047,
1863
+ "grad_norm": 0.39094579219818115,
1864
+ "learning_rate": 4.9018246353732755e-06,
1865
+ "loss": 0.462,
1866
  "step": 1315
1867
  },
1868
  {
1869
+ "epoch": 1.826583592938733,
1870
+ "grad_norm": 0.3818303048610687,
1871
+ "learning_rate": 4.534834221277262e-06,
1872
+ "loss": 0.4489,
1873
  "step": 1320
1874
  },
1875
  {
1876
+ "epoch": 1.8335064035998614,
1877
+ "grad_norm": 0.3742011487483978,
1878
+ "learning_rate": 4.181803025221898e-06,
1879
+ "loss": 0.4582,
1880
  "step": 1325
1881
  },
1882
  {
1883
+ "epoch": 1.84042921426099,
1884
+ "grad_norm": 0.38944151997566223,
1885
+ "learning_rate": 3.842782668543599e-06,
1886
+ "loss": 0.4486,
1887
  "step": 1330
1888
  },
1889
  {
1890
+ "epoch": 1.8473520249221185,
1891
+ "grad_norm": 0.34260445833206177,
1892
+ "learning_rate": 3.5178227238696347e-06,
1893
+ "loss": 0.4592,
1894
  "step": 1335
1895
  },
1896
  {
1897
+ "epoch": 1.8542748355832468,
1898
+ "grad_norm": 0.3642573058605194,
1899
+ "learning_rate": 3.2069707078694057e-06,
1900
+ "loss": 0.4391,
1901
  "step": 1340
1902
  },
1903
  {
1904
+ "epoch": 1.8611976462443751,
1905
+ "grad_norm": 0.3616952896118164,
1906
+ "learning_rate": 2.9102720743064527e-06,
1907
+ "loss": 0.4433,
1908
  "step": 1345
1909
  },
1910
  {
1911
+ "epoch": 1.8681204569055037,
1912
+ "grad_norm": 0.37953394651412964,
1913
+ "learning_rate": 2.6277702073919997e-06,
1914
+ "loss": 0.4464,
1915
  "step": 1350
1916
  },
1917
  {
1918
+ "epoch": 1.8750432675666322,
1919
+ "grad_norm": 0.37515413761138916,
1920
+ "learning_rate": 2.3595064154412374e-06,
1921
+ "loss": 0.4483,
1922
  "step": 1355
1923
  },
1924
  {
1925
+ "epoch": 1.8819660782277605,
1926
+ "grad_norm": 0.37095800042152405,
1927
+ "learning_rate": 2.1055199248330546e-06,
1928
+ "loss": 0.4397,
1929
  "step": 1360
1930
  },
1931
  {
1932
+ "epoch": 1.8888888888888888,
1933
+ "grad_norm": 0.39310017228126526,
1934
+ "learning_rate": 1.8658478742742625e-06,
1935
+ "loss": 0.4483,
1936
  "step": 1365
1937
  },
1938
  {
1939
+ "epoch": 1.8958116995500172,
1940
+ "grad_norm": 0.37695300579071045,
1941
+ "learning_rate": 1.6405253093690343e-06,
1942
+ "loss": 0.4538,
1943
  "step": 1370
1944
  },
1945
  {
1946
+ "epoch": 1.9027345102111457,
1947
+ "grad_norm": 0.40717628598213196,
1948
+ "learning_rate": 1.429585177494419e-06,
1949
+ "loss": 0.4518,
1950
  "step": 1375
1951
  },
1952
  {
1953
+ "epoch": 1.9096573208722742,
1954
+ "grad_norm": 0.4086827337741852,
1955
+ "learning_rate": 1.2330583229827053e-06,
1956
+ "loss": 0.4606,
1957
  "step": 1380
1958
  },
1959
  {
1960
+ "epoch": 1.9165801315334026,
1961
+ "grad_norm": 0.3887495994567871,
1962
+ "learning_rate": 1.0509734826112394e-06,
1963
+ "loss": 0.4428,
1964
  "step": 1385
1965
  },
1966
  {
1967
+ "epoch": 1.9235029421945309,
1968
+ "grad_norm": 0.4020722508430481,
1969
+ "learning_rate": 8.833572814004321e-07,
1970
+ "loss": 0.4346,
1971
  "step": 1390
1972
  },
1973
  {
1974
+ "epoch": 1.9304257528556594,
1975
+ "grad_norm": 0.37936893105506897,
1976
+ "learning_rate": 7.302342287205943e-07,
1977
+ "loss": 0.4386,
1978
  "step": 1395
1979
  },
1980
  {
1981
+ "epoch": 1.937348563516788,
1982
+ "grad_norm": 0.35618487000465393,
1983
+ "learning_rate": 5.916267147080934e-07,
1984
+ "loss": 0.4556,
1985
  "step": 1400
1986
  },
1987
  {
1988
+ "epoch": 1.9442713741779163,
1989
+ "grad_norm": 0.40421056747436523,
1990
+ "learning_rate": 4.6755500699138253e-07,
1991
+ "loss": 0.4469,
1992
  "step": 1405
1993
  },
1994
  {
1995
+ "epoch": 1.9511941848390446,
1996
+ "grad_norm": 0.379030704498291,
1997
+ "learning_rate": 3.5803724772740523e-07,
1998
+ "loss": 0.4385,
1999
  "step": 1410
2000
  },
2001
  {
2002
+ "epoch": 1.958116995500173,
2003
+ "grad_norm": 0.3669376075267792,
2004
+ "learning_rate": 2.630894509488058e-07,
2005
+ "loss": 0.4586,
2006
  "step": 1415
2007
  },
2008
  {
2009
+ "epoch": 1.9650398061613015,
2010
+ "grad_norm": 0.37957462668418884,
2011
+ "learning_rate": 1.8272550022230272e-07,
2012
+ "loss": 0.4511,
2013
  "step": 1420
2014
  },
2015
  {
2016
+ "epoch": 1.97196261682243,
2017
+ "grad_norm": 0.3957827389240265,
2018
+ "learning_rate": 1.1695714661856816e-07,
2019
+ "loss": 0.4533,
2020
  "step": 1425
2021
  },
2022
  {
2023
+ "epoch": 1.9788854274835583,
2024
+ "grad_norm": 0.3891640901565552,
2025
+ "learning_rate": 6.579400699397998e-08,
2026
+ "loss": 0.4464,
2027
  "step": 1430
2028
  },
2029
  {
2030
+ "epoch": 1.9858082381446867,
2031
+ "grad_norm": 0.3792797029018402,
2032
+ "learning_rate": 2.9243562584402305e-08,
2033
+ "loss": 0.4454,
2034
  "step": 1435
2035
  },
2036
  {
2037
+ "epoch": 1.9927310488058152,
2038
+ "grad_norm": 0.3780197501182556,
2039
+ "learning_rate": 7.311157911249478e-09,
2040
+ "loss": 0.4515,
2041
  "step": 1440
2042
  },
2043
  {
2044
+ "epoch": 1.9982692973347178,
2045
+ "eval_loss": 0.5933948159217834,
2046
+ "eval_runtime": 78.897,
2047
+ "eval_samples_per_second": 26.224,
2048
+ "eval_steps_per_second": 1.648,
2049
+ "step": 1444
2050
  },
2051
  {
2052
+ "epoch": 1.9982692973347178,
2053
+ "step": 1444,
2054
+ "total_flos": 7.113001462715646e+18,
2055
+ "train_loss": 0.6430129308119374,
2056
+ "train_runtime": 14278.4866,
2057
+ "train_samples_per_second": 6.474,
2058
+ "train_steps_per_second": 0.101
2059
  }
2060
  ],
2061
  "logging_steps": 5,
2062
+ "max_steps": 1444,
2063
  "num_input_tokens_seen": 0,
2064
  "num_train_epochs": 2,
2065
  "save_steps": 500,
 
2075
  "attributes": {}
2076
  }
2077
  },
2078
+ "total_flos": 7.113001462715646e+18,
2079
  "train_batch_size": 4,
2080
  "trial_name": null,
2081
  "trial_params": null
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:866fc066538def660b5fabd18598af79ac48a91e9239f0323747193025ad16fb
3
- size 7096
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52e38047350b995275b0facde1c76bf4478da59cb980ea74aed2d82c48c6ffc6
3
+ size 7032