kevinoli commited on
Commit
c523e50
1 Parent(s): 596b6fd

Training in progress, step 9000, checkpoint

Browse files
checkpoint-9000/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6c7006edb64df6aeea55de1d9383a1d5ba7df07e0289eae0b67e50d27f10834f
3
  size 1711848436
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:01d9ce051c5a8f3422c4d52c3025c914787a9690cb81fc657427c8c74784cb64
3
  size 1711848436
checkpoint-9000/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:77ab61faab3cb01fd64b5f30980bee98d3009c9a4ce9100e935bd2c31f93268f
3
  size 3424043887
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:255c9301a46eb4a25029510a5819799534027da52f07785e08e2543a38170593
3
  size 3424043887
checkpoint-9000/rng_state.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:75e7a66179ff03d27874bf0d0676cd79fe7137c908da603458cd4b183abc4a42
3
  size 14503
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26c99df967694e66808fb8d4e4814367449c627fb6d4991b78dafe1ced9ade3e
3
  size 14503
checkpoint-9000/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:be738b6e943cb0505f19818033ed2ab90e8e660c03ff20089783742289e05acc
3
  size 623
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:17921b0cd7ff814e07f1683e2287860a992d82b3e0a99e8e8f19c90e3ae1cfe4
3
  size 623
checkpoint-9000/trainer_state.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
- "best_metric": 0.9536784291267395,
3
  "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e4l58-l/checkpoint-9000",
4
- "epoch": 1.658069270449521,
5
  "eval_steps": 500,
6
  "global_step": 9000,
7
  "is_hyper_param_search": false,
@@ -9,278 +9,278 @@
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
- "epoch": 0.09211495946941783,
13
- "grad_norm": 20.04043960571289,
14
- "learning_rate": 4.8848563006632274e-08,
15
- "loss": 0.3758,
16
  "step": 500
17
  },
18
  {
19
- "epoch": 0.09211495946941783,
20
- "eval_loss": 1.4184927940368652,
21
- "eval_runtime": 74.3562,
22
- "eval_samples_per_second": 16.233,
23
- "eval_steps_per_second": 2.031,
24
  "step": 500
25
  },
26
  {
27
- "epoch": 0.18422991893883567,
28
- "grad_norm": 74.25025939941406,
29
- "learning_rate": 4.7697126013264556e-08,
30
- "loss": 0.4103,
31
  "step": 1000
32
  },
33
  {
34
- "epoch": 0.18422991893883567,
35
- "eval_loss": 1.3500770330429077,
36
- "eval_runtime": 75.7013,
37
- "eval_samples_per_second": 15.944,
38
- "eval_steps_per_second": 1.995,
39
  "step": 1000
40
  },
41
  {
42
- "epoch": 0.2763448784082535,
43
- "grad_norm": 0.5102410912513733,
44
- "learning_rate": 4.6545689019896826e-08,
45
- "loss": 0.433,
46
  "step": 1500
47
  },
48
  {
49
- "epoch": 0.2763448784082535,
50
- "eval_loss": 1.2885024547576904,
51
- "eval_runtime": 75.197,
52
- "eval_samples_per_second": 16.051,
53
- "eval_steps_per_second": 2.008,
54
  "step": 1500
55
  },
56
  {
57
- "epoch": 0.36845983787767134,
58
- "grad_norm": 0.1667678952217102,
59
- "learning_rate": 4.539425202652911e-08,
60
- "loss": 0.3424,
61
  "step": 2000
62
  },
63
  {
64
- "epoch": 0.36845983787767134,
65
- "eval_loss": 1.239119052886963,
66
- "eval_runtime": 75.387,
67
- "eval_samples_per_second": 16.011,
68
- "eval_steps_per_second": 2.003,
69
  "step": 2000
70
  },
71
  {
72
- "epoch": 0.46057479734708917,
73
- "grad_norm": 343.8175048828125,
74
- "learning_rate": 4.4242815033161385e-08,
75
- "loss": 0.3645,
76
  "step": 2500
77
  },
78
  {
79
- "epoch": 0.46057479734708917,
80
- "eval_loss": 1.1902339458465576,
81
- "eval_runtime": 75.9899,
82
- "eval_samples_per_second": 15.884,
83
- "eval_steps_per_second": 1.987,
84
  "step": 2500
85
  },
86
  {
87
- "epoch": 0.552689756816507,
88
- "grad_norm": 305.5626525878906,
89
- "learning_rate": 4.309137803979366e-08,
90
- "loss": 0.3172,
91
  "step": 3000
92
  },
93
  {
94
- "epoch": 0.552689756816507,
95
- "eval_loss": 1.1506118774414062,
96
- "eval_runtime": 74.8655,
97
- "eval_samples_per_second": 16.122,
98
- "eval_steps_per_second": 2.017,
99
  "step": 3000
100
  },
101
  {
102
- "epoch": 0.6448047162859248,
103
- "grad_norm": 65.59048461914062,
104
- "learning_rate": 4.193994104642594e-08,
105
- "loss": 0.2751,
106
  "step": 3500
107
  },
108
  {
109
- "epoch": 0.6448047162859248,
110
- "eval_loss": 1.1169357299804688,
111
- "eval_runtime": 76.2936,
112
- "eval_samples_per_second": 15.82,
113
- "eval_steps_per_second": 1.979,
114
  "step": 3500
115
  },
116
  {
117
- "epoch": 0.7369196757553427,
118
- "grad_norm": 18.106201171875,
119
- "learning_rate": 4.078850405305821e-08,
120
- "loss": 0.2919,
121
  "step": 4000
122
  },
123
  {
124
- "epoch": 0.7369196757553427,
125
- "eval_loss": 1.0920671224594116,
126
- "eval_runtime": 74.9101,
127
- "eval_samples_per_second": 16.113,
128
- "eval_steps_per_second": 2.016,
129
  "step": 4000
130
  },
131
  {
132
- "epoch": 0.8290346352247605,
133
- "grad_norm": 506.40570068359375,
134
- "learning_rate": 3.9637067059690496e-08,
135
- "loss": 0.2583,
136
  "step": 4500
137
  },
138
  {
139
- "epoch": 0.8290346352247605,
140
- "eval_loss": 1.0721209049224854,
141
- "eval_runtime": 75.8053,
142
- "eval_samples_per_second": 15.922,
143
- "eval_steps_per_second": 1.992,
144
  "step": 4500
145
  },
146
  {
147
- "epoch": 0.9211495946941783,
148
- "grad_norm": 1.1508910655975342,
149
- "learning_rate": 3.848563006632277e-08,
150
- "loss": 0.2679,
151
  "step": 5000
152
  },
153
  {
154
- "epoch": 0.9211495946941783,
155
- "eval_loss": 1.0519349575042725,
156
- "eval_runtime": 75.226,
157
- "eval_samples_per_second": 16.045,
158
- "eval_steps_per_second": 2.007,
159
  "step": 5000
160
  },
161
  {
162
- "epoch": 1.013264554163596,
163
- "grad_norm": 0.1393290013074875,
164
- "learning_rate": 3.733419307295505e-08,
165
- "loss": 0.2472,
166
  "step": 5500
167
  },
168
  {
169
- "epoch": 1.013264554163596,
170
- "eval_loss": 1.0355629920959473,
171
- "eval_runtime": 74.6535,
172
- "eval_samples_per_second": 16.168,
173
- "eval_steps_per_second": 2.023,
174
  "step": 5500
175
  },
176
  {
177
- "epoch": 1.105379513633014,
178
- "grad_norm": 200.34664916992188,
179
- "learning_rate": 3.6182756079587324e-08,
180
- "loss": 0.26,
181
  "step": 6000
182
  },
183
  {
184
- "epoch": 1.105379513633014,
185
- "eval_loss": 1.0176944732666016,
186
- "eval_runtime": 74.4374,
187
- "eval_samples_per_second": 16.215,
188
- "eval_steps_per_second": 2.029,
189
  "step": 6000
190
  },
191
  {
192
- "epoch": 1.1974944731024317,
193
- "grad_norm": 431.1804504394531,
194
- "learning_rate": 3.50313190862196e-08,
195
- "loss": 0.2153,
196
  "step": 6500
197
  },
198
  {
199
- "epoch": 1.1974944731024317,
200
- "eval_loss": 1.0044586658477783,
201
- "eval_runtime": 74.3247,
202
- "eval_samples_per_second": 16.24,
203
- "eval_steps_per_second": 2.032,
204
  "step": 6500
205
  },
206
  {
207
- "epoch": 1.2896094325718497,
208
- "grad_norm": 0.6204941272735596,
209
- "learning_rate": 3.387988209285188e-08,
210
- "loss": 0.1791,
211
  "step": 7000
212
  },
213
  {
214
- "epoch": 1.2896094325718497,
215
- "eval_loss": 0.9926707148551941,
216
- "eval_runtime": 75.7317,
217
- "eval_samples_per_second": 15.938,
218
- "eval_steps_per_second": 1.994,
219
  "step": 7000
220
  },
221
  {
222
- "epoch": 1.3817243920412676,
223
- "grad_norm": 52.27273178100586,
224
- "learning_rate": 3.272844509948415e-08,
225
- "loss": 0.2082,
226
  "step": 7500
227
  },
228
  {
229
- "epoch": 1.3817243920412676,
230
- "eval_loss": 0.9803809523582458,
231
- "eval_runtime": 74.9936,
232
- "eval_samples_per_second": 16.095,
233
- "eval_steps_per_second": 2.014,
234
  "step": 7500
235
  },
236
  {
237
- "epoch": 1.4738393515106853,
238
- "grad_norm": 0.0976635068655014,
239
- "learning_rate": 3.1577008106116435e-08,
240
- "loss": 0.196,
241
  "step": 8000
242
  },
243
  {
244
- "epoch": 1.4738393515106853,
245
- "eval_loss": 0.9711655378341675,
246
- "eval_runtime": 74.3537,
247
- "eval_samples_per_second": 16.233,
248
- "eval_steps_per_second": 2.031,
249
  "step": 8000
250
  },
251
  {
252
- "epoch": 1.565954310980103,
253
- "grad_norm": 46.64162826538086,
254
- "learning_rate": 3.042557111274871e-08,
255
- "loss": 0.1946,
256
  "step": 8500
257
  },
258
  {
259
- "epoch": 1.565954310980103,
260
- "eval_loss": 0.9621181488037109,
261
- "eval_runtime": 75.7205,
262
- "eval_samples_per_second": 15.94,
263
- "eval_steps_per_second": 1.994,
264
  "step": 8500
265
  },
266
  {
267
- "epoch": 1.658069270449521,
268
- "grad_norm": 0.05690698325634003,
269
- "learning_rate": 2.927413411938099e-08,
270
- "loss": 0.2422,
271
  "step": 9000
272
  },
273
  {
274
- "epoch": 1.658069270449521,
275
- "eval_loss": 0.9536784291267395,
276
- "eval_runtime": 75.1745,
277
- "eval_samples_per_second": 16.056,
278
- "eval_steps_per_second": 2.009,
279
  "step": 9000
280
  }
281
  ],
282
  "logging_steps": 500,
283
- "max_steps": 21712,
284
  "num_input_tokens_seen": 0,
285
  "num_train_epochs": 4,
286
  "save_steps": 500,
 
1
  {
2
+ "best_metric": 0.992453396320343,
3
  "best_model_checkpoint": "./output/clip-finetuned-csu-p14-336-e4l58-l/checkpoint-9000",
4
+ "epoch": 0.9598976109215017,
5
  "eval_steps": 500,
6
  "global_step": 9000,
7
  "is_hyper_param_search": false,
 
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
+ "epoch": 0.05332764505119454,
13
+ "grad_norm": 217.55360412597656,
14
+ "learning_rate": 4.933340443686007e-08,
15
+ "loss": 0.4667,
16
  "step": 500
17
  },
18
  {
19
+ "epoch": 0.05332764505119454,
20
+ "eval_loss": 1.4426143169403076,
21
+ "eval_runtime": 61.9901,
22
+ "eval_samples_per_second": 15.922,
23
+ "eval_steps_per_second": 2.0,
24
  "step": 500
25
  },
26
  {
27
+ "epoch": 0.10665529010238908,
28
+ "grad_norm": 73.73340606689453,
29
+ "learning_rate": 4.8666808873720136e-08,
30
+ "loss": 0.4532,
31
  "step": 1000
32
  },
33
  {
34
+ "epoch": 0.10665529010238908,
35
+ "eval_loss": 1.3815597295761108,
36
+ "eval_runtime": 62.4624,
37
+ "eval_samples_per_second": 15.801,
38
+ "eval_steps_per_second": 1.985,
39
  "step": 1000
40
  },
41
  {
42
+ "epoch": 0.1599829351535836,
43
+ "grad_norm": 471.4011535644531,
44
+ "learning_rate": 4.80002133105802e-08,
45
+ "loss": 0.3749,
46
  "step": 1500
47
  },
48
  {
49
+ "epoch": 0.1599829351535836,
50
+ "eval_loss": 1.3310539722442627,
51
+ "eval_runtime": 63.3918,
52
+ "eval_samples_per_second": 15.57,
53
+ "eval_steps_per_second": 1.956,
54
  "step": 1500
55
  },
56
  {
57
+ "epoch": 0.21331058020477817,
58
+ "grad_norm": 11.962693214416504,
59
+ "learning_rate": 4.733361774744027e-08,
60
+ "loss": 0.336,
61
  "step": 2000
62
  },
63
  {
64
+ "epoch": 0.21331058020477817,
65
+ "eval_loss": 1.2890639305114746,
66
+ "eval_runtime": 63.177,
67
+ "eval_samples_per_second": 15.623,
68
+ "eval_steps_per_second": 1.963,
69
  "step": 2000
70
  },
71
  {
72
+ "epoch": 0.2666382252559727,
73
+ "grad_norm": 375.1609191894531,
74
+ "learning_rate": 4.666702218430034e-08,
75
+ "loss": 0.3585,
76
  "step": 2500
77
  },
78
  {
79
+ "epoch": 0.2666382252559727,
80
+ "eval_loss": 1.2536433935165405,
81
+ "eval_runtime": 63.1512,
82
+ "eval_samples_per_second": 15.629,
83
+ "eval_steps_per_second": 1.964,
84
  "step": 2500
85
  },
86
  {
87
+ "epoch": 0.3199658703071672,
88
+ "grad_norm": 0.0022936267778277397,
89
+ "learning_rate": 4.600042662116041e-08,
90
+ "loss": 0.303,
91
  "step": 3000
92
  },
93
  {
94
+ "epoch": 0.3199658703071672,
95
+ "eval_loss": 1.2202869653701782,
96
+ "eval_runtime": 63.5857,
97
+ "eval_samples_per_second": 15.522,
98
+ "eval_steps_per_second": 1.95,
99
  "step": 3000
100
  },
101
  {
102
+ "epoch": 0.37329351535836175,
103
+ "grad_norm": 15.805087089538574,
104
+ "learning_rate": 4.5333831058020476e-08,
105
+ "loss": 0.3242,
106
  "step": 3500
107
  },
108
  {
109
+ "epoch": 0.37329351535836175,
110
+ "eval_loss": 1.195627212524414,
111
+ "eval_runtime": 63.4024,
112
+ "eval_samples_per_second": 15.567,
113
+ "eval_steps_per_second": 1.956,
114
  "step": 3500
115
  },
116
  {
117
+ "epoch": 0.42662116040955633,
118
+ "grad_norm": 5.377908229827881,
119
+ "learning_rate": 4.4667235494880546e-08,
120
+ "loss": 0.2427,
121
  "step": 4000
122
  },
123
  {
124
+ "epoch": 0.42662116040955633,
125
+ "eval_loss": 1.169384241104126,
126
+ "eval_runtime": 63.5082,
127
+ "eval_samples_per_second": 15.541,
128
+ "eval_steps_per_second": 1.953,
129
  "step": 4000
130
  },
131
  {
132
+ "epoch": 0.47994880546075086,
133
+ "grad_norm": 150.5334930419922,
134
+ "learning_rate": 4.4000639931740615e-08,
135
+ "loss": 0.2993,
136
  "step": 4500
137
  },
138
  {
139
+ "epoch": 0.47994880546075086,
140
+ "eval_loss": 1.145558476448059,
141
+ "eval_runtime": 62.3888,
142
+ "eval_samples_per_second": 15.82,
143
+ "eval_steps_per_second": 1.988,
144
  "step": 4500
145
  },
146
  {
147
+ "epoch": 0.5332764505119454,
148
+ "grad_norm": 0.12246542423963547,
149
+ "learning_rate": 4.333404436860068e-08,
150
+ "loss": 0.3183,
151
  "step": 5000
152
  },
153
  {
154
+ "epoch": 0.5332764505119454,
155
+ "eval_loss": 1.1201218366622925,
156
+ "eval_runtime": 63.7154,
157
+ "eval_samples_per_second": 15.491,
158
+ "eval_steps_per_second": 1.946,
159
  "step": 5000
160
  },
161
  {
162
+ "epoch": 0.58660409556314,
163
+ "grad_norm": 0.3306196630001068,
164
+ "learning_rate": 4.266744880546075e-08,
165
+ "loss": 0.307,
166
  "step": 5500
167
  },
168
  {
169
+ "epoch": 0.58660409556314,
170
+ "eval_loss": 1.0982334613800049,
171
+ "eval_runtime": 62.4882,
172
+ "eval_samples_per_second": 15.795,
173
+ "eval_steps_per_second": 1.984,
174
  "step": 5500
175
  },
176
  {
177
+ "epoch": 0.6399317406143344,
178
+ "grad_norm": 149.38189697265625,
179
+ "learning_rate": 4.2000853242320816e-08,
180
+ "loss": 0.2638,
181
  "step": 6000
182
  },
183
  {
184
+ "epoch": 0.6399317406143344,
185
+ "eval_loss": 1.0779541730880737,
186
+ "eval_runtime": 63.4101,
187
+ "eval_samples_per_second": 15.565,
188
+ "eval_steps_per_second": 1.956,
189
  "step": 6000
190
  },
191
  {
192
+ "epoch": 0.693259385665529,
193
+ "grad_norm": 353.0036315917969,
194
+ "learning_rate": 4.1334257679180886e-08,
195
+ "loss": 0.2226,
196
  "step": 6500
197
  },
198
  {
199
+ "epoch": 0.693259385665529,
200
+ "eval_loss": 1.0613036155700684,
201
+ "eval_runtime": 62.8292,
202
+ "eval_samples_per_second": 15.709,
203
+ "eval_steps_per_second": 1.974,
204
  "step": 6500
205
  },
206
  {
207
+ "epoch": 0.7465870307167235,
208
+ "grad_norm": 0.13188917934894562,
209
+ "learning_rate": 4.0667662116040955e-08,
210
+ "loss": 0.2453,
211
  "step": 7000
212
  },
213
  {
214
+ "epoch": 0.7465870307167235,
215
+ "eval_loss": 1.0443507432937622,
216
+ "eval_runtime": 62.7457,
217
+ "eval_samples_per_second": 15.73,
218
+ "eval_steps_per_second": 1.976,
219
  "step": 7000
220
  },
221
  {
222
+ "epoch": 0.7999146757679181,
223
+ "grad_norm": 59.391639709472656,
224
+ "learning_rate": 4.000106655290102e-08,
225
+ "loss": 0.272,
226
  "step": 7500
227
  },
228
  {
229
+ "epoch": 0.7999146757679181,
230
+ "eval_loss": 1.0300519466400146,
231
+ "eval_runtime": 62.3719,
232
+ "eval_samples_per_second": 15.824,
233
+ "eval_steps_per_second": 1.988,
234
  "step": 7500
235
  },
236
  {
237
+ "epoch": 0.8532423208191127,
238
+ "grad_norm": 495.4797668457031,
239
+ "learning_rate": 3.9334470989761093e-08,
240
+ "loss": 0.283,
241
  "step": 8000
242
  },
243
  {
244
+ "epoch": 0.8532423208191127,
245
+ "eval_loss": 1.0166898965835571,
246
+ "eval_runtime": 62.0338,
247
+ "eval_samples_per_second": 15.911,
248
+ "eval_steps_per_second": 1.999,
249
  "step": 8000
250
  },
251
  {
252
+ "epoch": 0.9065699658703071,
253
+ "grad_norm": 159.0310821533203,
254
+ "learning_rate": 3.866787542662116e-08,
255
+ "loss": 0.2331,
256
  "step": 8500
257
  },
258
  {
259
+ "epoch": 0.9065699658703071,
260
+ "eval_loss": 1.003544807434082,
261
+ "eval_runtime": 62.9704,
262
+ "eval_samples_per_second": 15.674,
263
+ "eval_steps_per_second": 1.969,
264
  "step": 8500
265
  },
266
  {
267
+ "epoch": 0.9598976109215017,
268
+ "grad_norm": 45.40327835083008,
269
+ "learning_rate": 3.8001279863481225e-08,
270
+ "loss": 0.2362,
271
  "step": 9000
272
  },
273
  {
274
+ "epoch": 0.9598976109215017,
275
+ "eval_loss": 0.992453396320343,
276
+ "eval_runtime": 63.1405,
277
+ "eval_samples_per_second": 15.632,
278
+ "eval_steps_per_second": 1.964,
279
  "step": 9000
280
  }
281
  ],
282
  "logging_steps": 500,
283
+ "max_steps": 37504,
284
  "num_input_tokens_seen": 0,
285
  "num_train_epochs": 4,
286
  "save_steps": 500,
checkpoint-9000/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9a29c2374e59ce3f57b34db9cb20d2d1bdc870c3dad128e9a4d462636c37e8b0
3
  size 4847
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87e35dca972e13467c750106c19301b098fbd2918fda1087487e227bac3aca37
3
  size 4847