mikhail-panzo commited on
Commit
987c5da
1 Parent(s): e88feb8

Training in progress, step 1500, checkpoint

Browse files
checkpoint-1500/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3830d89e0fa8ae59498d20c4e968944b0c77e29ea4958b2bdd0800438026af47
3
  size 577789320
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:88b8e054922c9713da9bbd6ecea0a42adf9fbf8501f7c1d812e6674d1da19619
3
  size 577789320
checkpoint-1500/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3f5d4f8cc150cbbc26ba2002c2c2104082487cbb90f6ebd50ee3c186623c6b8b
3
  size 1155772233
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:330e12eaf2821997654f91e38f3ab458f393092724d199c73735596d222d318c
3
  size 1155772233
checkpoint-1500/rng_state.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c13fbf434ac483950ee79aff8c53d4bdc90dc463b90e984c384ab9704b308359
3
  size 14244
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:24e53203e58a87ed5daa21c65691317d73a381c3361e5604b87f75165301ceaf
3
  size 14244
checkpoint-1500/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f32a07717606d664792d4ebfd434223710fe948a637ff5f34234da98aa96ac43
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d814612c6068e7d514ebd9b781f2fe7c957c26fb16d0f984f282d01053c1b46a
3
  size 1064
checkpoint-1500/trainer_state.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "best_metric": 0.43601250648498535,
3
  "best_model_checkpoint": "mikhail-panzo/fil_b32_le5_s4000/checkpoint-1500",
4
- "epoch": 33.333333333333336,
5
  "eval_steps": 500,
6
  "global_step": 1500,
7
  "is_hyper_param_search": false,
@@ -9,244 +9,244 @@
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
- "epoch": 1.1111111111111112,
13
- "grad_norm": 4.398683071136475,
14
- "learning_rate": 2.4500000000000004e-07,
15
- "loss": 0.8108,
16
  "step": 50
17
  },
18
  {
19
- "epoch": 2.2222222222222223,
20
- "grad_norm": 3.726436138153076,
21
- "learning_rate": 4.95e-07,
22
- "loss": 0.792,
23
  "step": 100
24
  },
25
  {
26
- "epoch": 3.3333333333333335,
27
- "grad_norm": 2.0379676818847656,
28
- "learning_rate": 7.450000000000001e-07,
29
- "loss": 0.7413,
30
  "step": 150
31
  },
32
  {
33
- "epoch": 4.444444444444445,
34
- "grad_norm": 2.0460364818573,
35
- "learning_rate": 9.950000000000002e-07,
36
- "loss": 0.7342,
37
  "step": 200
38
  },
39
  {
40
- "epoch": 5.555555555555555,
41
- "grad_norm": 2.803539276123047,
42
- "learning_rate": 1.2450000000000002e-06,
43
- "loss": 0.7147,
44
  "step": 250
45
  },
46
  {
47
- "epoch": 6.666666666666667,
48
- "grad_norm": 1.9818241596221924,
49
- "learning_rate": 1.495e-06,
50
- "loss": 0.7072,
51
  "step": 300
52
  },
53
  {
54
- "epoch": 7.777777777777778,
55
- "grad_norm": 3.081399917602539,
56
- "learning_rate": 1.745e-06,
57
- "loss": 0.6843,
58
  "step": 350
59
  },
60
  {
61
- "epoch": 8.88888888888889,
62
- "grad_norm": 2.4465785026550293,
63
- "learning_rate": 1.9950000000000004e-06,
64
- "loss": 0.657,
65
  "step": 400
66
  },
67
  {
68
- "epoch": 10.0,
69
- "grad_norm": 2.929323434829712,
70
- "learning_rate": 2.245e-06,
71
- "loss": 0.6597,
72
  "step": 450
73
  },
74
  {
75
- "epoch": 11.11111111111111,
76
- "grad_norm": 1.419311285018921,
77
- "learning_rate": 2.4950000000000003e-06,
78
- "loss": 0.6246,
79
  "step": 500
80
  },
81
  {
82
- "epoch": 11.11111111111111,
83
- "eval_loss": 0.5375720262527466,
84
- "eval_runtime": 9.2568,
85
- "eval_samples_per_second": 17.177,
86
- "eval_steps_per_second": 2.161,
87
  "step": 500
88
  },
89
  {
90
- "epoch": 12.222222222222221,
91
- "grad_norm": 2.3438374996185303,
92
- "learning_rate": 2.7450000000000004e-06,
93
- "loss": 0.6018,
94
  "step": 550
95
  },
96
  {
97
- "epoch": 13.333333333333334,
98
- "grad_norm": 1.7441222667694092,
99
- "learning_rate": 2.995e-06,
100
- "loss": 0.5547,
101
  "step": 600
102
  },
103
  {
104
- "epoch": 14.444444444444445,
105
- "grad_norm": 1.5592021942138672,
106
- "learning_rate": 3.2450000000000003e-06,
107
- "loss": 0.5466,
108
  "step": 650
109
  },
110
  {
111
- "epoch": 15.555555555555555,
112
- "grad_norm": 1.305444359779358,
113
- "learning_rate": 3.495e-06,
114
- "loss": 0.535,
115
  "step": 700
116
  },
117
  {
118
- "epoch": 16.666666666666668,
119
- "grad_norm": 2.2067477703094482,
120
- "learning_rate": 3.745e-06,
121
- "loss": 0.5343,
122
  "step": 750
123
  },
124
  {
125
- "epoch": 17.77777777777778,
126
- "grad_norm": 1.667480707168579,
127
- "learning_rate": 3.995000000000001e-06,
128
- "loss": 0.5296,
129
  "step": 800
130
  },
131
  {
132
- "epoch": 18.88888888888889,
133
- "grad_norm": 1.5465797185897827,
134
- "learning_rate": 4.245e-06,
135
- "loss": 0.5198,
136
  "step": 850
137
  },
138
  {
139
- "epoch": 20.0,
140
- "grad_norm": 2.760589361190796,
141
- "learning_rate": 4.495e-06,
142
- "loss": 0.5193,
143
  "step": 900
144
  },
145
  {
146
- "epoch": 21.11111111111111,
147
- "grad_norm": 1.4570050239562988,
148
- "learning_rate": 4.745e-06,
149
- "loss": 0.5124,
150
  "step": 950
151
  },
152
  {
153
- "epoch": 22.22222222222222,
154
- "grad_norm": 2.1914265155792236,
155
- "learning_rate": 4.9950000000000005e-06,
156
- "loss": 0.5034,
157
  "step": 1000
158
  },
159
  {
160
- "epoch": 22.22222222222222,
161
- "eval_loss": 0.45661377906799316,
162
- "eval_runtime": 9.2158,
163
- "eval_samples_per_second": 17.253,
164
- "eval_steps_per_second": 2.17,
165
  "step": 1000
166
  },
167
  {
168
- "epoch": 23.333333333333332,
169
- "grad_norm": 1.941751480102539,
170
- "learning_rate": 5.245e-06,
171
- "loss": 0.5046,
172
  "step": 1050
173
  },
174
  {
175
- "epoch": 24.444444444444443,
176
- "grad_norm": 1.9293475151062012,
177
- "learning_rate": 5.495000000000001e-06,
178
- "loss": 0.4948,
179
  "step": 1100
180
  },
181
  {
182
- "epoch": 25.555555555555557,
183
- "grad_norm": 1.3278849124908447,
184
- "learning_rate": 5.745000000000001e-06,
185
- "loss": 0.4931,
186
  "step": 1150
187
  },
188
  {
189
- "epoch": 26.666666666666668,
190
- "grad_norm": 1.1657841205596924,
191
- "learning_rate": 5.995000000000001e-06,
192
- "loss": 0.496,
193
  "step": 1200
194
  },
195
  {
196
- "epoch": 27.77777777777778,
197
- "grad_norm": 1.8730145692825317,
198
- "learning_rate": 6.245000000000001e-06,
199
- "loss": 0.4939,
200
  "step": 1250
201
  },
202
  {
203
- "epoch": 28.88888888888889,
204
- "grad_norm": 2.1156740188598633,
205
- "learning_rate": 6.4950000000000005e-06,
206
- "loss": 0.4908,
207
  "step": 1300
208
  },
209
  {
210
- "epoch": 30.0,
211
- "grad_norm": 1.799806833267212,
212
- "learning_rate": 6.745000000000001e-06,
213
- "loss": 0.4875,
214
  "step": 1350
215
  },
216
  {
217
- "epoch": 31.11111111111111,
218
- "grad_norm": 1.762253761291504,
219
- "learning_rate": 6.995000000000001e-06,
220
- "loss": 0.4836,
221
  "step": 1400
222
  },
223
  {
224
- "epoch": 32.22222222222222,
225
- "grad_norm": 1.305191993713379,
226
- "learning_rate": 7.245000000000001e-06,
227
- "loss": 0.4811,
228
  "step": 1450
229
  },
230
  {
231
- "epoch": 33.333333333333336,
232
- "grad_norm": 1.5870022773742676,
233
- "learning_rate": 7.495000000000001e-06,
234
- "loss": 0.4807,
235
  "step": 1500
236
  },
237
  {
238
- "epoch": 33.333333333333336,
239
- "eval_loss": 0.43601250648498535,
240
- "eval_runtime": 9.5435,
241
- "eval_samples_per_second": 16.66,
242
- "eval_steps_per_second": 2.096,
243
  "step": 1500
244
  }
245
  ],
246
  "logging_steps": 50,
247
  "max_steps": 4000,
248
  "num_input_tokens_seen": 0,
249
- "num_train_epochs": 89,
250
  "save_steps": 500,
251
  "stateful_callbacks": {
252
  "TrainerControl": {
@@ -260,7 +260,7 @@
260
  "attributes": {}
261
  }
262
  },
263
- "total_flos": 1.03221946153656e+16,
264
  "train_batch_size": 16,
265
  "trial_name": null,
266
  "trial_params": null
 
1
  {
2
+ "best_metric": 0.4245499074459076,
3
  "best_model_checkpoint": "mikhail-panzo/fil_b32_le5_s4000/checkpoint-1500",
4
+ "epoch": 133.33333333333334,
5
  "eval_steps": 500,
6
  "global_step": 1500,
7
  "is_hyper_param_search": false,
 
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
+ "epoch": 4.444444444444445,
13
+ "grad_norm": 4.288165092468262,
14
+ "learning_rate": 2.5000000000000004e-07,
15
+ "loss": 0.7989,
16
  "step": 50
17
  },
18
  {
19
+ "epoch": 8.88888888888889,
20
+ "grad_norm": 2.4493463039398193,
21
+ "learning_rate": 5.000000000000001e-07,
22
+ "loss": 0.7701,
23
  "step": 100
24
  },
25
  {
26
+ "epoch": 13.333333333333334,
27
+ "grad_norm": 1.3798907995224,
28
+ "learning_rate": 7.5e-07,
29
+ "loss": 0.7374,
30
  "step": 150
31
  },
32
  {
33
+ "epoch": 17.77777777777778,
34
+ "grad_norm": 2.6173877716064453,
35
+ "learning_rate": 1.0000000000000002e-06,
36
+ "loss": 0.7101,
37
  "step": 200
38
  },
39
  {
40
+ "epoch": 22.22222222222222,
41
+ "grad_norm": 1.3756701946258545,
42
+ "learning_rate": 1.25e-06,
43
+ "loss": 0.6887,
44
  "step": 250
45
  },
46
  {
47
+ "epoch": 26.666666666666668,
48
+ "grad_norm": 1.0456843376159668,
49
+ "learning_rate": 1.5e-06,
50
+ "loss": 0.672,
51
  "step": 300
52
  },
53
  {
54
+ "epoch": 31.11111111111111,
55
+ "grad_norm": 0.9518398642539978,
56
+ "learning_rate": 1.75e-06,
57
+ "loss": 0.6583,
58
  "step": 350
59
  },
60
  {
61
+ "epoch": 35.55555555555556,
62
+ "grad_norm": 1.305978775024414,
63
+ "learning_rate": 2.0000000000000003e-06,
64
+ "loss": 0.6258,
65
  "step": 400
66
  },
67
  {
68
+ "epoch": 40.0,
69
+ "grad_norm": 0.8857623338699341,
70
+ "learning_rate": 2.25e-06,
71
+ "loss": 0.5835,
72
  "step": 450
73
  },
74
  {
75
+ "epoch": 44.44444444444444,
76
+ "grad_norm": 1.052182912826538,
77
+ "learning_rate": 2.5e-06,
78
+ "loss": 0.5539,
79
  "step": 500
80
  },
81
  {
82
+ "epoch": 44.44444444444444,
83
+ "eval_loss": 0.48496538400650024,
84
+ "eval_runtime": 9.4574,
85
+ "eval_samples_per_second": 16.812,
86
+ "eval_steps_per_second": 2.115,
87
  "step": 500
88
  },
89
  {
90
+ "epoch": 48.888888888888886,
91
+ "grad_norm": 1.249458909034729,
92
+ "learning_rate": 2.7500000000000004e-06,
93
+ "loss": 0.5387,
94
  "step": 550
95
  },
96
  {
97
+ "epoch": 53.333333333333336,
98
+ "grad_norm": 0.801243245601654,
99
+ "learning_rate": 3e-06,
100
+ "loss": 0.5344,
101
  "step": 600
102
  },
103
  {
104
+ "epoch": 57.77777777777778,
105
+ "grad_norm": 1.1922223567962646,
106
+ "learning_rate": 3.2500000000000002e-06,
107
+ "loss": 0.5259,
108
  "step": 650
109
  },
110
  {
111
+ "epoch": 62.22222222222222,
112
+ "grad_norm": 0.9428519606590271,
113
+ "learning_rate": 3.5e-06,
114
+ "loss": 0.5179,
115
  "step": 700
116
  },
117
  {
118
+ "epoch": 66.66666666666667,
119
+ "grad_norm": 0.755260705947876,
120
+ "learning_rate": 3.7500000000000005e-06,
121
+ "loss": 0.5082,
122
  "step": 750
123
  },
124
  {
125
+ "epoch": 71.11111111111111,
126
+ "grad_norm": 1.8587590456008911,
127
+ "learning_rate": 4.000000000000001e-06,
128
+ "loss": 0.5096,
129
  "step": 800
130
  },
131
  {
132
+ "epoch": 75.55555555555556,
133
+ "grad_norm": 0.6854829788208008,
134
+ "learning_rate": 4.25e-06,
135
+ "loss": 0.4952,
136
  "step": 850
137
  },
138
  {
139
+ "epoch": 80.0,
140
+ "grad_norm": 0.9088854789733887,
141
+ "learning_rate": 4.5e-06,
142
+ "loss": 0.4941,
143
  "step": 900
144
  },
145
  {
146
+ "epoch": 84.44444444444444,
147
+ "grad_norm": 0.8932746648788452,
148
+ "learning_rate": 4.75e-06,
149
+ "loss": 0.4936,
150
  "step": 950
151
  },
152
  {
153
+ "epoch": 88.88888888888889,
154
+ "grad_norm": 0.9161014556884766,
155
+ "learning_rate": 5e-06,
156
+ "loss": 0.4867,
157
  "step": 1000
158
  },
159
  {
160
+ "epoch": 88.88888888888889,
161
+ "eval_loss": 0.4438903331756592,
162
+ "eval_runtime": 9.2831,
163
+ "eval_samples_per_second": 17.128,
164
+ "eval_steps_per_second": 2.154,
165
  "step": 1000
166
  },
167
  {
168
+ "epoch": 93.33333333333333,
169
+ "grad_norm": 0.6858564615249634,
170
+ "learning_rate": 5.2500000000000006e-06,
171
+ "loss": 0.486,
172
  "step": 1050
173
  },
174
  {
175
+ "epoch": 97.77777777777777,
176
+ "grad_norm": 1.2040996551513672,
177
+ "learning_rate": 5.500000000000001e-06,
178
+ "loss": 0.4789,
179
  "step": 1100
180
  },
181
  {
182
+ "epoch": 102.22222222222223,
183
+ "grad_norm": 1.0067086219787598,
184
+ "learning_rate": 5.75e-06,
185
+ "loss": 0.477,
186
  "step": 1150
187
  },
188
  {
189
+ "epoch": 106.66666666666667,
190
+ "grad_norm": 1.4230589866638184,
191
+ "learning_rate": 6e-06,
192
+ "loss": 0.473,
193
  "step": 1200
194
  },
195
  {
196
+ "epoch": 111.11111111111111,
197
+ "grad_norm": 0.916904091835022,
198
+ "learning_rate": 6.25e-06,
199
+ "loss": 0.4698,
200
  "step": 1250
201
  },
202
  {
203
+ "epoch": 115.55555555555556,
204
+ "grad_norm": 0.9870907664299011,
205
+ "learning_rate": 6.5000000000000004e-06,
206
+ "loss": 0.4693,
207
  "step": 1300
208
  },
209
  {
210
+ "epoch": 120.0,
211
+ "grad_norm": 1.0495694875717163,
212
+ "learning_rate": 6.750000000000001e-06,
213
+ "loss": 0.4651,
214
  "step": 1350
215
  },
216
  {
217
+ "epoch": 124.44444444444444,
218
+ "grad_norm": 1.6615877151489258,
219
+ "learning_rate": 7e-06,
220
+ "loss": 0.4648,
221
  "step": 1400
222
  },
223
  {
224
+ "epoch": 128.88888888888889,
225
+ "grad_norm": 0.89169842004776,
226
+ "learning_rate": 7.25e-06,
227
+ "loss": 0.4643,
228
  "step": 1450
229
  },
230
  {
231
+ "epoch": 133.33333333333334,
232
+ "grad_norm": 0.8503329157829285,
233
+ "learning_rate": 7.500000000000001e-06,
234
+ "loss": 0.4595,
235
  "step": 1500
236
  },
237
  {
238
+ "epoch": 133.33333333333334,
239
+ "eval_loss": 0.4245499074459076,
240
+ "eval_runtime": 8.4972,
241
+ "eval_samples_per_second": 18.712,
242
+ "eval_steps_per_second": 2.354,
243
  "step": 1500
244
  }
245
  ],
246
  "logging_steps": 50,
247
  "max_steps": 4000,
248
  "num_input_tokens_seen": 0,
249
+ "num_train_epochs": 364,
250
  "save_steps": 500,
251
  "stateful_callbacks": {
252
  "TrainerControl": {
 
260
  "attributes": {}
261
  }
262
  },
263
+ "total_flos": 4.125426981308266e+16,
264
  "train_batch_size": 16,
265
  "trial_name": null,
266
  "trial_params": null
checkpoint-1500/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:05b7ecdded947540b9670cedf7cfe87afc7199b237efb642b915ffbb505d51dc
3
  size 5304
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2de69a241d012253ce4b1526ecd0f9fd0f7dfbedfea74bfff08144d42d9f523a
3
  size 5304