fayetitchenal commited on
Commit
5cdcbb7
·
verified ·
1 Parent(s): 6613477

Training in progress, step 18000, checkpoint

Browse files
checkpoint-18000/config.json ADDED
@@ -0,0 +1,92 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "fayetitchenal/segformer_finetuned_full_102824",
3
+ "architectures": [
4
+ "SegformerForSemanticSegmentation"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.0,
7
+ "classifier_dropout_prob": 0.1,
8
+ "decoder_hidden_size": 768,
9
+ "depths": [
10
+ 3,
11
+ 6,
12
+ 40,
13
+ 3
14
+ ],
15
+ "downsampling_rates": [
16
+ 1,
17
+ 4,
18
+ 8,
19
+ 16
20
+ ],
21
+ "drop_path_rate": 0.1,
22
+ "hidden_act": "gelu",
23
+ "hidden_dropout_prob": 0.0,
24
+ "hidden_sizes": [
25
+ 64,
26
+ 128,
27
+ 320,
28
+ 512
29
+ ],
30
+ "id2label": {
31
+ "0": "background",
32
+ "1": "double_plant",
33
+ "2": "drydown",
34
+ "3": "endrow",
35
+ "4": "nutrient_deficiency",
36
+ "5": "planter_skip",
37
+ "6": "water",
38
+ "7": "waterway",
39
+ "8": "weed_cluster"
40
+ },
41
+ "image_size": 224,
42
+ "initializer_range": 0.02,
43
+ "label2id": {
44
+ "background": 0,
45
+ "double_plant": 1,
46
+ "drydown": 2,
47
+ "endrow": 3,
48
+ "nutrient_deficiency": 4,
49
+ "planter_skip": 5,
50
+ "water": 6,
51
+ "waterway": 7,
52
+ "weed_cluster": 8
53
+ },
54
+ "layer_norm_eps": 1e-06,
55
+ "mlp_ratios": [
56
+ 4,
57
+ 4,
58
+ 4,
59
+ 4
60
+ ],
61
+ "model_type": "segformer",
62
+ "num_attention_heads": [
63
+ 1,
64
+ 2,
65
+ 5,
66
+ 8
67
+ ],
68
+ "num_channels": 4,
69
+ "num_encoder_blocks": 4,
70
+ "patch_sizes": [
71
+ 7,
72
+ 3,
73
+ 3,
74
+ 3
75
+ ],
76
+ "reshape_last_stage": true,
77
+ "semantic_loss_ignore_index": 255,
78
+ "sr_ratios": [
79
+ 8,
80
+ 4,
81
+ 2,
82
+ 1
83
+ ],
84
+ "strides": [
85
+ 4,
86
+ 2,
87
+ 2,
88
+ 2
89
+ ],
90
+ "torch_dtype": "float32",
91
+ "transformers_version": "4.40.2"
92
+ }
checkpoint-18000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cb52396102a86bf9fe80909a847c3c57983f265ae694ba1ade2c31f7d74b6d76
3
+ size 338562516
checkpoint-18000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:875bfc1a61cba2b4aa52e7a208f607ef30b2d0f004f3a204434202e6405a415a
3
+ size 677518143
checkpoint-18000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70ee661630baaadd8b30bc1c04db32add9b1e4de06e4ce840f9155e76b87ee23
3
+ size 14575
checkpoint-18000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3beb4d4c94e5faaf5905d583deafb0bf937633cc661655a0e95874f918c080e9
3
+ size 563
checkpoint-18000/trainer_state.json ADDED
@@ -0,0 +1,379 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 10.61320754716981,
5
+ "eval_steps": 2250,
6
+ "global_step": 18000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.589622641509434,
13
+ "grad_norm": 9.09252643585205,
14
+ "learning_rate": 2.9481132075471702e-05,
15
+ "loss": 2.8132,
16
+ "step": 1000
17
+ },
18
+ {
19
+ "epoch": 1.179245283018868,
20
+ "grad_norm": 7.212547302246094,
21
+ "learning_rate": 5.8962264150943403e-05,
22
+ "loss": 1.7737,
23
+ "step": 2000
24
+ },
25
+ {
26
+ "epoch": 1.3266509433962264,
27
+ "eval_accuracy_background": 0.6323930092153496,
28
+ "eval_accuracy_double_plant": 0.3849305031210213,
29
+ "eval_accuracy_drydown": 0.8411866405808689,
30
+ "eval_accuracy_endrow": 0.5317305315249954,
31
+ "eval_accuracy_nutrient_deficiency": 0.7313988224684205,
32
+ "eval_accuracy_planter_skip": 0.568298826627077,
33
+ "eval_accuracy_water": 0.9894614449233383,
34
+ "eval_accuracy_waterway": 0.680404792024105,
35
+ "eval_accuracy_weed_cluster": 0.45985419985443066,
36
+ "eval_iou_background": 0.5728108810832201,
37
+ "eval_iou_double_plant": 0.2244358493191135,
38
+ "eval_iou_drydown": 0.5181682211979306,
39
+ "eval_iou_endrow": 0.173903146695287,
40
+ "eval_iou_nutrient_deficiency": 0.3634766707254442,
41
+ "eval_iou_planter_skip": 0.2799366984899054,
42
+ "eval_iou_water": 0.3046660180760937,
43
+ "eval_iou_waterway": 0.27785446752468385,
44
+ "eval_iou_weed_cluster": 0.3040126938826071,
45
+ "eval_loss": 1.6098850965499878,
46
+ "eval_mean_accuracy": 0.6466287522599563,
47
+ "eval_mean_iou": 0.3354738496660317,
48
+ "eval_overall_accuracy": 0.664939675891281,
49
+ "eval_runtime": 548.0884,
50
+ "eval_samples_per_second": 15.556,
51
+ "eval_steps_per_second": 0.487,
52
+ "step": 2250
53
+ },
54
+ {
55
+ "epoch": 1.7688679245283019,
56
+ "grad_norm": 6.368645191192627,
57
+ "learning_rate": 8.84433962264151e-05,
58
+ "loss": 1.5815,
59
+ "step": 3000
60
+ },
61
+ {
62
+ "epoch": 2.358490566037736,
63
+ "grad_norm": 8.585444450378418,
64
+ "learning_rate": 9.990216178505835e-05,
65
+ "loss": 1.472,
66
+ "step": 4000
67
+ },
68
+ {
69
+ "epoch": 2.6533018867924527,
70
+ "eval_accuracy_background": 0.6784285276238243,
71
+ "eval_accuracy_double_plant": 0.45434500662461724,
72
+ "eval_accuracy_drydown": 0.89512745592938,
73
+ "eval_accuracy_endrow": 0.5571884263160238,
74
+ "eval_accuracy_nutrient_deficiency": 0.6123044131155115,
75
+ "eval_accuracy_planter_skip": 0.5606583933264498,
76
+ "eval_accuracy_water": 0.8798561490206811,
77
+ "eval_accuracy_waterway": 0.4483896291979977,
78
+ "eval_accuracy_weed_cluster": 0.33068794060446854,
79
+ "eval_iou_background": 0.6044297199665501,
80
+ "eval_iou_double_plant": 0.28959011610726704,
81
+ "eval_iou_drydown": 0.5010830899198156,
82
+ "eval_iou_endrow": 0.19991054244053805,
83
+ "eval_iou_nutrient_deficiency": 0.36180734635376055,
84
+ "eval_iou_planter_skip": 0.28574898470084736,
85
+ "eval_iou_water": 0.45906365727311366,
86
+ "eval_iou_waterway": 0.1704026096816556,
87
+ "eval_iou_weed_cluster": 0.24703832661933417,
88
+ "eval_loss": 1.5796327590942383,
89
+ "eval_mean_accuracy": 0.601887326862106,
90
+ "eval_mean_iou": 0.34656382145143133,
91
+ "eval_overall_accuracy": 0.6864635619347936,
92
+ "eval_runtime": 528.9743,
93
+ "eval_samples_per_second": 16.118,
94
+ "eval_steps_per_second": 0.505,
95
+ "step": 4500
96
+ },
97
+ {
98
+ "epoch": 2.94811320754717,
99
+ "grad_norm": 6.215964317321777,
100
+ "learning_rate": 9.931699497285711e-05,
101
+ "loss": 1.3994,
102
+ "step": 5000
103
+ },
104
+ {
105
+ "epoch": 3.5377358490566038,
106
+ "grad_norm": 2.051684856414795,
107
+ "learning_rate": 9.821001342355031e-05,
108
+ "loss": 1.3177,
109
+ "step": 6000
110
+ },
111
+ {
112
+ "epoch": 3.9799528301886795,
113
+ "eval_accuracy_background": 0.6462387800902261,
114
+ "eval_accuracy_double_plant": 0.5789862724446915,
115
+ "eval_accuracy_drydown": 0.8528714242099724,
116
+ "eval_accuracy_endrow": 0.4707707581928247,
117
+ "eval_accuracy_nutrient_deficiency": 0.6952529134275348,
118
+ "eval_accuracy_planter_skip": 0.6016348257166644,
119
+ "eval_accuracy_water": 0.980488748185343,
120
+ "eval_accuracy_waterway": 0.520362876821302,
121
+ "eval_accuracy_weed_cluster": 0.48490926389178757,
122
+ "eval_iou_background": 0.5865287998573112,
123
+ "eval_iou_double_plant": 0.23644210203746396,
124
+ "eval_iou_drydown": 0.5074338458069501,
125
+ "eval_iou_endrow": 0.22486725808431682,
126
+ "eval_iou_nutrient_deficiency": 0.38304612346670297,
127
+ "eval_iou_planter_skip": 0.2354710865951827,
128
+ "eval_iou_water": 0.4030757215089252,
129
+ "eval_iou_waterway": 0.11303841029835258,
130
+ "eval_iou_weed_cluster": 0.303994759745169,
131
+ "eval_loss": 1.576890230178833,
132
+ "eval_mean_accuracy": 0.6479462069978162,
133
+ "eval_mean_iou": 0.33265534526670826,
134
+ "eval_overall_accuracy": 0.6749021834627369,
135
+ "eval_runtime": 526.8336,
136
+ "eval_samples_per_second": 16.183,
137
+ "eval_steps_per_second": 0.507,
138
+ "step": 6750
139
+ },
140
+ {
141
+ "epoch": 4.127358490566038,
142
+ "grad_norm": 8.216894149780273,
143
+ "learning_rate": 9.65929299206612e-05,
144
+ "loss": 1.2741,
145
+ "step": 7000
146
+ },
147
+ {
148
+ "epoch": 4.716981132075472,
149
+ "grad_norm": 3.159677743911743,
150
+ "learning_rate": 9.448285454973738e-05,
151
+ "loss": 1.2126,
152
+ "step": 8000
153
+ },
154
+ {
155
+ "epoch": 5.306603773584905,
156
+ "grad_norm": 1.5673192739486694,
157
+ "learning_rate": 9.190211365944541e-05,
158
+ "loss": 1.171,
159
+ "step": 9000
160
+ },
161
+ {
162
+ "epoch": 5.306603773584905,
163
+ "eval_accuracy_background": 0.7269166066590328,
164
+ "eval_accuracy_double_plant": 0.5245596485272079,
165
+ "eval_accuracy_drydown": 0.8277967088801947,
166
+ "eval_accuracy_endrow": 0.48676704241624213,
167
+ "eval_accuracy_nutrient_deficiency": 0.5977026838803114,
168
+ "eval_accuracy_planter_skip": 0.660893993754415,
169
+ "eval_accuracy_water": 0.9632686366062749,
170
+ "eval_accuracy_waterway": 0.36961777988306216,
171
+ "eval_accuracy_weed_cluster": 0.4562712731630422,
172
+ "eval_iou_background": 0.6430143634959152,
173
+ "eval_iou_double_plant": 0.28282184588015774,
174
+ "eval_iou_drydown": 0.5184629485658176,
175
+ "eval_iou_endrow": 0.28061928855749285,
176
+ "eval_iou_nutrient_deficiency": 0.3791407898501335,
177
+ "eval_iou_planter_skip": 0.3393091165595382,
178
+ "eval_iou_water": 0.44762851379996027,
179
+ "eval_iou_waterway": 0.15789185948205015,
180
+ "eval_iou_weed_cluster": 0.31112156295187404,
181
+ "eval_loss": 1.4945807456970215,
182
+ "eval_mean_accuracy": 0.6237549304188649,
183
+ "eval_mean_iou": 0.3733344765714377,
184
+ "eval_overall_accuracy": 0.7158101182141803,
185
+ "eval_runtime": 526.9805,
186
+ "eval_samples_per_second": 16.179,
187
+ "eval_steps_per_second": 0.507,
188
+ "step": 9000
189
+ },
190
+ {
191
+ "epoch": 5.89622641509434,
192
+ "grad_norm": 2.707547903060913,
193
+ "learning_rate": 8.887801363027234e-05,
194
+ "loss": 1.1505,
195
+ "step": 10000
196
+ },
197
+ {
198
+ "epoch": 6.485849056603773,
199
+ "grad_norm": 2.3729190826416016,
200
+ "learning_rate": 8.544255195035733e-05,
201
+ "loss": 1.1169,
202
+ "step": 11000
203
+ },
204
+ {
205
+ "epoch": 6.633254716981132,
206
+ "eval_accuracy_background": 0.7347674332960046,
207
+ "eval_accuracy_double_plant": 0.5163814236541484,
208
+ "eval_accuracy_drydown": 0.6856827555491075,
209
+ "eval_accuracy_endrow": 0.39937713978730166,
210
+ "eval_accuracy_nutrient_deficiency": 0.6915369481048043,
211
+ "eval_accuracy_planter_skip": 0.6270521933807838,
212
+ "eval_accuracy_water": 0.969516685620151,
213
+ "eval_accuracy_waterway": 0.38301412838614074,
214
+ "eval_accuracy_weed_cluster": 0.4826223016858449,
215
+ "eval_iou_background": 0.639678261208408,
216
+ "eval_iou_double_plant": 0.3481368436310948,
217
+ "eval_iou_drydown": 0.46861831796483266,
218
+ "eval_iou_endrow": 0.2434744213529937,
219
+ "eval_iou_nutrient_deficiency": 0.3926333742769746,
220
+ "eval_iou_planter_skip": 0.34354493532185115,
221
+ "eval_iou_water": 0.45169225338047775,
222
+ "eval_iou_waterway": 0.1686818262862102,
223
+ "eval_iou_weed_cluster": 0.3206403315046765,
224
+ "eval_loss": 1.541970133781433,
225
+ "eval_mean_accuracy": 0.609994556607143,
226
+ "eval_mean_iou": 0.37523339610305767,
227
+ "eval_overall_accuracy": 0.7094913536171235,
228
+ "eval_runtime": 527.9323,
229
+ "eval_samples_per_second": 16.15,
230
+ "eval_steps_per_second": 0.506,
231
+ "step": 11250
232
+ },
233
+ {
234
+ "epoch": 7.0754716981132075,
235
+ "grad_norm": 2.201885223388672,
236
+ "learning_rate": 8.163207865551112e-05,
237
+ "loss": 1.0955,
238
+ "step": 12000
239
+ },
240
+ {
241
+ "epoch": 7.665094339622642,
242
+ "grad_norm": 1.980070948600769,
243
+ "learning_rate": 7.748691171566864e-05,
244
+ "loss": 1.0659,
245
+ "step": 13000
246
+ },
247
+ {
248
+ "epoch": 7.959905660377358,
249
+ "eval_accuracy_background": 0.7691083980404577,
250
+ "eval_accuracy_double_plant": 0.6024440209211166,
251
+ "eval_accuracy_drydown": 0.6814714434518159,
252
+ "eval_accuracy_endrow": 0.5464485717344214,
253
+ "eval_accuracy_nutrient_deficiency": 0.509203127689722,
254
+ "eval_accuracy_planter_skip": 0.6650384781390429,
255
+ "eval_accuracy_water": 0.9454965193897085,
256
+ "eval_accuracy_waterway": 0.3698600004151499,
257
+ "eval_accuracy_weed_cluster": 0.5154108026712857,
258
+ "eval_iou_background": 0.6578031962911103,
259
+ "eval_iou_double_plant": 0.31294833900910296,
260
+ "eval_iou_drydown": 0.47689665403570886,
261
+ "eval_iou_endrow": 0.2618940193906661,
262
+ "eval_iou_nutrient_deficiency": 0.3534970492371128,
263
+ "eval_iou_planter_skip": 0.31625822947683646,
264
+ "eval_iou_water": 0.4817152186884349,
265
+ "eval_iou_waterway": 0.15124054678647988,
266
+ "eval_iou_weed_cluster": 0.3316746531778273,
267
+ "eval_loss": 1.533487319946289,
268
+ "eval_mean_accuracy": 0.6227201513836357,
269
+ "eval_mean_iou": 0.37154754512147553,
270
+ "eval_overall_accuracy": 0.7196279695998185,
271
+ "eval_runtime": 527.6877,
272
+ "eval_samples_per_second": 16.157,
273
+ "eval_steps_per_second": 0.506,
274
+ "step": 13500
275
+ },
276
+ {
277
+ "epoch": 8.254716981132075,
278
+ "grad_norm": 2.4944000244140625,
279
+ "learning_rate": 7.305091043730557e-05,
280
+ "loss": 1.0462,
281
+ "step": 14000
282
+ },
283
+ {
284
+ "epoch": 8.84433962264151,
285
+ "grad_norm": 2.1022963523864746,
286
+ "learning_rate": 6.837101139557501e-05,
287
+ "loss": 1.0253,
288
+ "step": 15000
289
+ },
290
+ {
291
+ "epoch": 9.286556603773585,
292
+ "eval_accuracy_background": 0.7728507820265633,
293
+ "eval_accuracy_double_plant": 0.5148991261591159,
294
+ "eval_accuracy_drydown": 0.705680052147375,
295
+ "eval_accuracy_endrow": 0.5623466153698641,
296
+ "eval_accuracy_nutrient_deficiency": 0.5199237782061494,
297
+ "eval_accuracy_planter_skip": 0.5901816685685527,
298
+ "eval_accuracy_water": 0.900331474456696,
299
+ "eval_accuracy_waterway": 0.4231801763849317,
300
+ "eval_accuracy_weed_cluster": 0.4348939770827169,
301
+ "eval_iou_background": 0.6602399849319134,
302
+ "eval_iou_double_plant": 0.3486919692513433,
303
+ "eval_iou_drydown": 0.4745067424738569,
304
+ "eval_iou_endrow": 0.26713541595159784,
305
+ "eval_iou_nutrient_deficiency": 0.35050766552393176,
306
+ "eval_iou_planter_skip": 0.35530947653735145,
307
+ "eval_iou_water": 0.4922300137725779,
308
+ "eval_iou_waterway": 0.16175899595022267,
309
+ "eval_iou_weed_cluster": 0.31610597752344755,
310
+ "eval_loss": 1.595482587814331,
311
+ "eval_mean_accuracy": 0.6026986278224407,
312
+ "eval_mean_iou": 0.38072069354624916,
313
+ "eval_overall_accuracy": 0.7215420453829048,
314
+ "eval_runtime": 528.6777,
315
+ "eval_samples_per_second": 16.127,
316
+ "eval_steps_per_second": 0.505,
317
+ "step": 15750
318
+ },
319
+ {
320
+ "epoch": 9.433962264150944,
321
+ "grad_norm": 1.7042616605758667,
322
+ "learning_rate": 6.34967318063877e-05,
323
+ "loss": 0.9886,
324
+ "step": 16000
325
+ },
326
+ {
327
+ "epoch": 10.023584905660377,
328
+ "grad_norm": 2.8781015872955322,
329
+ "learning_rate": 5.847964559317128e-05,
330
+ "loss": 0.9908,
331
+ "step": 17000
332
+ },
333
+ {
334
+ "epoch": 10.61320754716981,
335
+ "grad_norm": 3.1057872772216797,
336
+ "learning_rate": 5.3372837691956955e-05,
337
+ "loss": 0.9665,
338
+ "step": 18000
339
+ },
340
+ {
341
+ "epoch": 10.61320754716981,
342
+ "eval_accuracy_background": 0.779746789817052,
343
+ "eval_accuracy_double_plant": 0.42980582389486105,
344
+ "eval_accuracy_drydown": 0.7098823662359411,
345
+ "eval_accuracy_endrow": 0.5570065199747799,
346
+ "eval_accuracy_nutrient_deficiency": 0.4918399226013093,
347
+ "eval_accuracy_planter_skip": 0.5889699117229129,
348
+ "eval_accuracy_water": 0.9671706616607829,
349
+ "eval_accuracy_waterway": 0.2686522876474846,
350
+ "eval_accuracy_weed_cluster": 0.4629757010799936,
351
+ "eval_iou_background": 0.6649244448853501,
352
+ "eval_iou_double_plant": 0.3192541496243385,
353
+ "eval_iou_drydown": 0.47846096564055196,
354
+ "eval_iou_endrow": 0.2738811943652846,
355
+ "eval_iou_nutrient_deficiency": 0.34461702917680226,
356
+ "eval_iou_planter_skip": 0.3595814747520095,
357
+ "eval_iou_water": 0.46891808902847,
358
+ "eval_iou_waterway": 0.1499735440891693,
359
+ "eval_iou_weed_cluster": 0.32812183598439904,
360
+ "eval_loss": 1.6341685056686401,
361
+ "eval_mean_accuracy": 0.5840055538483464,
362
+ "eval_mean_iou": 0.3764147475051528,
363
+ "eval_overall_accuracy": 0.7252505242891358,
364
+ "eval_runtime": 551.0227,
365
+ "eval_samples_per_second": 15.473,
366
+ "eval_steps_per_second": 0.485,
367
+ "step": 18000
368
+ }
369
+ ],
370
+ "logging_steps": 1000,
371
+ "max_steps": 33920,
372
+ "num_input_tokens_seen": 0,
373
+ "num_train_epochs": 20,
374
+ "save_steps": 2250,
375
+ "total_flos": 3.0653390846567645e+20,
376
+ "train_batch_size": 32,
377
+ "trial_name": null,
378
+ "trial_params": null
379
+ }
checkpoint-18000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fc3fa49520d3f99fe06147ccb1c775910e89ea989beb4ae81c5b0c6fc998032
3
+ size 4539