fayetitchenal commited on
Commit
510bd2e
·
verified ·
1 Parent(s): 76a9fdb

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_110324",
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:ca6160d2dc2b3d8969930cb0f9ac75d8387c648f46b1f735843c81ad16029821
3
+ size 338562516
checkpoint-18000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28c8d57c60d1476d1bc019d38db892cc647869a0a9b4f192e4c635c1a03828c2
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": 9.137022018432617,
14
+ "learning_rate": 2.9481132075471702e-05,
15
+ "loss": 2.8435,
16
+ "step": 1000
17
+ },
18
+ {
19
+ "epoch": 1.179245283018868,
20
+ "grad_norm": 6.052332401275635,
21
+ "learning_rate": 5.8962264150943403e-05,
22
+ "loss": 1.791,
23
+ "step": 2000
24
+ },
25
+ {
26
+ "epoch": 1.3266509433962264,
27
+ "eval_accuracy_background": 0.6658729859800179,
28
+ "eval_accuracy_double_plant": 0.44796971861142204,
29
+ "eval_accuracy_drydown": 0.838452272819733,
30
+ "eval_accuracy_endrow": 0.6564099936974693,
31
+ "eval_accuracy_nutrient_deficiency": 0.5255760511997533,
32
+ "eval_accuracy_planter_skip": 0.7062276764551357,
33
+ "eval_accuracy_water": 0.9873387227342924,
34
+ "eval_accuracy_waterway": 0.5770296385642565,
35
+ "eval_accuracy_weed_cluster": 0.4428120632960455,
36
+ "eval_iou_background": 0.5904190410028399,
37
+ "eval_iou_double_plant": 0.21713659970128166,
38
+ "eval_iou_drydown": 0.5191145408680865,
39
+ "eval_iou_endrow": 0.146705809299966,
40
+ "eval_iou_nutrient_deficiency": 0.3433700256479391,
41
+ "eval_iou_planter_skip": 0.19867844175237825,
42
+ "eval_iou_water": 0.3151796347651091,
43
+ "eval_iou_waterway": 0.17710299517028666,
44
+ "eval_iou_weed_cluster": 0.2871479144864637,
45
+ "eval_loss": 1.6307891607284546,
46
+ "eval_mean_accuracy": 0.6497432359286807,
47
+ "eval_mean_iou": 0.31053944474381673,
48
+ "eval_overall_accuracy": 0.6703095203666921,
49
+ "eval_runtime": 550.4413,
50
+ "eval_samples_per_second": 15.489,
51
+ "eval_steps_per_second": 0.485,
52
+ "step": 2250
53
+ },
54
+ {
55
+ "epoch": 1.7688679245283019,
56
+ "grad_norm": 10.023482322692871,
57
+ "learning_rate": 8.84433962264151e-05,
58
+ "loss": 1.5821,
59
+ "step": 3000
60
+ },
61
+ {
62
+ "epoch": 2.358490566037736,
63
+ "grad_norm": 3.805853843688965,
64
+ "learning_rate": 9.990216178505835e-05,
65
+ "loss": 1.4645,
66
+ "step": 4000
67
+ },
68
+ {
69
+ "epoch": 2.6533018867924527,
70
+ "eval_accuracy_background": 0.6928149295345922,
71
+ "eval_accuracy_double_plant": 0.520600810240665,
72
+ "eval_accuracy_drydown": 0.9041661302317083,
73
+ "eval_accuracy_endrow": 0.41216652623141536,
74
+ "eval_accuracy_nutrient_deficiency": 0.5572999343661789,
75
+ "eval_accuracy_planter_skip": 0.5885655971148109,
76
+ "eval_accuracy_water": 0.9012349964392468,
77
+ "eval_accuracy_waterway": 0.5122856732827319,
78
+ "eval_accuracy_weed_cluster": 0.3529035479823603,
79
+ "eval_iou_background": 0.6154372685119965,
80
+ "eval_iou_double_plant": 0.286691279808622,
81
+ "eval_iou_drydown": 0.5005889595647872,
82
+ "eval_iou_endrow": 0.23071917367118422,
83
+ "eval_iou_nutrient_deficiency": 0.34903960157986247,
84
+ "eval_iou_planter_skip": 0.3404743586046269,
85
+ "eval_iou_water": 0.4866735755203849,
86
+ "eval_iou_waterway": 0.17740336931417308,
87
+ "eval_iou_weed_cluster": 0.25445442999377565,
88
+ "eval_loss": 1.6076184511184692,
89
+ "eval_mean_accuracy": 0.6046709050470789,
90
+ "eval_mean_iou": 0.36016466850771256,
91
+ "eval_overall_accuracy": 0.6938899108737494,
92
+ "eval_runtime": 529.8506,
93
+ "eval_samples_per_second": 16.091,
94
+ "eval_steps_per_second": 0.504,
95
+ "step": 4500
96
+ },
97
+ {
98
+ "epoch": 2.94811320754717,
99
+ "grad_norm": 3.9986305236816406,
100
+ "learning_rate": 9.931699497285711e-05,
101
+ "loss": 1.3957,
102
+ "step": 5000
103
+ },
104
+ {
105
+ "epoch": 3.5377358490566038,
106
+ "grad_norm": 2.7372334003448486,
107
+ "learning_rate": 9.821001342355031e-05,
108
+ "loss": 1.3124,
109
+ "step": 6000
110
+ },
111
+ {
112
+ "epoch": 3.9799528301886795,
113
+ "eval_accuracy_background": 0.6307230231497055,
114
+ "eval_accuracy_double_plant": 0.48486363452173786,
115
+ "eval_accuracy_drydown": 0.8612367898895216,
116
+ "eval_accuracy_endrow": 0.5075110635399532,
117
+ "eval_accuracy_nutrient_deficiency": 0.6861430352037324,
118
+ "eval_accuracy_planter_skip": 0.6031976315027698,
119
+ "eval_accuracy_water": 0.9802194342104156,
120
+ "eval_accuracy_waterway": 0.48897799511586637,
121
+ "eval_accuracy_weed_cluster": 0.556708376519504,
122
+ "eval_iou_background": 0.5741567963414517,
123
+ "eval_iou_double_plant": 0.2860919486620036,
124
+ "eval_iou_drydown": 0.5029587941632667,
125
+ "eval_iou_endrow": 0.22521629078173816,
126
+ "eval_iou_nutrient_deficiency": 0.3596590524564408,
127
+ "eval_iou_planter_skip": 0.3023475517768152,
128
+ "eval_iou_water": 0.41466030404934484,
129
+ "eval_iou_waterway": 0.11905712637247004,
130
+ "eval_iou_weed_cluster": 0.3161136298740651,
131
+ "eval_loss": 1.6003777980804443,
132
+ "eval_mean_accuracy": 0.6443978870725785,
133
+ "eval_mean_iou": 0.34447349938639954,
134
+ "eval_overall_accuracy": 0.6672287911844751,
135
+ "eval_runtime": 529.5311,
136
+ "eval_samples_per_second": 16.101,
137
+ "eval_steps_per_second": 0.504,
138
+ "step": 6750
139
+ },
140
+ {
141
+ "epoch": 4.127358490566038,
142
+ "grad_norm": 3.5192806720733643,
143
+ "learning_rate": 9.65929299206612e-05,
144
+ "loss": 1.2631,
145
+ "step": 7000
146
+ },
147
+ {
148
+ "epoch": 4.716981132075472,
149
+ "grad_norm": 3.9972991943359375,
150
+ "learning_rate": 9.448285454973738e-05,
151
+ "loss": 1.2003,
152
+ "step": 8000
153
+ },
154
+ {
155
+ "epoch": 5.306603773584905,
156
+ "grad_norm": 5.617863178253174,
157
+ "learning_rate": 9.190211365944541e-05,
158
+ "loss": 1.1545,
159
+ "step": 9000
160
+ },
161
+ {
162
+ "epoch": 5.306603773584905,
163
+ "eval_accuracy_background": 0.7152218890852868,
164
+ "eval_accuracy_double_plant": 0.559446789259611,
165
+ "eval_accuracy_drydown": 0.7926407431922761,
166
+ "eval_accuracy_endrow": 0.5090316219335013,
167
+ "eval_accuracy_nutrient_deficiency": 0.5738896211353212,
168
+ "eval_accuracy_planter_skip": 0.6593127897989867,
169
+ "eval_accuracy_water": 0.9642661896511828,
170
+ "eval_accuracy_waterway": 0.46492080717677636,
171
+ "eval_accuracy_weed_cluster": 0.5152471841227493,
172
+ "eval_iou_background": 0.6309696025011753,
173
+ "eval_iou_double_plant": 0.33576586005371956,
174
+ "eval_iou_drydown": 0.5023456898026126,
175
+ "eval_iou_endrow": 0.24985551240605414,
176
+ "eval_iou_nutrient_deficiency": 0.36416354391893857,
177
+ "eval_iou_planter_skip": 0.3932620146979082,
178
+ "eval_iou_water": 0.46396310962016846,
179
+ "eval_iou_waterway": 0.16987197692528003,
180
+ "eval_iou_weed_cluster": 0.29655626229371107,
181
+ "eval_loss": 1.5461719036102295,
182
+ "eval_mean_accuracy": 0.6393308483728546,
183
+ "eval_mean_iou": 0.3785281746910631,
184
+ "eval_overall_accuracy": 0.7041663407297746,
185
+ "eval_runtime": 529.5113,
186
+ "eval_samples_per_second": 16.102,
187
+ "eval_steps_per_second": 0.504,
188
+ "step": 9000
189
+ },
190
+ {
191
+ "epoch": 5.89622641509434,
192
+ "grad_norm": 1.9349929094314575,
193
+ "learning_rate": 8.887801363027234e-05,
194
+ "loss": 1.1301,
195
+ "step": 10000
196
+ },
197
+ {
198
+ "epoch": 6.485849056603773,
199
+ "grad_norm": 3.736844539642334,
200
+ "learning_rate": 8.544255195035733e-05,
201
+ "loss": 1.0996,
202
+ "step": 11000
203
+ },
204
+ {
205
+ "epoch": 6.633254716981132,
206
+ "eval_accuracy_background": 0.7321345719725818,
207
+ "eval_accuracy_double_plant": 0.5061561579791534,
208
+ "eval_accuracy_drydown": 0.6959542485519916,
209
+ "eval_accuracy_endrow": 0.4876229883472026,
210
+ "eval_accuracy_nutrient_deficiency": 0.6450410877845559,
211
+ "eval_accuracy_planter_skip": 0.7014340631125459,
212
+ "eval_accuracy_water": 0.9722180984130999,
213
+ "eval_accuracy_waterway": 0.4680069536108854,
214
+ "eval_accuracy_weed_cluster": 0.5999105125905921,
215
+ "eval_iou_background": 0.6400560054890143,
216
+ "eval_iou_double_plant": 0.3502816119859484,
217
+ "eval_iou_drydown": 0.48456196377624977,
218
+ "eval_iou_endrow": 0.2639788246911193,
219
+ "eval_iou_nutrient_deficiency": 0.396585559445601,
220
+ "eval_iou_planter_skip": 0.3409834868970738,
221
+ "eval_iou_water": 0.44451378617604653,
222
+ "eval_iou_waterway": 0.16081881352962019,
223
+ "eval_iou_weed_cluster": 0.33812444541681147,
224
+ "eval_loss": 1.492088794708252,
225
+ "eval_mean_accuracy": 0.645386520262512,
226
+ "eval_mean_iou": 0.37998938860083165,
227
+ "eval_overall_accuracy": 0.7115690581547556,
228
+ "eval_runtime": 529.9729,
229
+ "eval_samples_per_second": 16.088,
230
+ "eval_steps_per_second": 0.504,
231
+ "step": 11250
232
+ },
233
+ {
234
+ "epoch": 7.0754716981132075,
235
+ "grad_norm": 1.84828782081604,
236
+ "learning_rate": 8.163207865551112e-05,
237
+ "loss": 1.0727,
238
+ "step": 12000
239
+ },
240
+ {
241
+ "epoch": 7.665094339622642,
242
+ "grad_norm": 2.411391258239746,
243
+ "learning_rate": 7.748691171566864e-05,
244
+ "loss": 1.0468,
245
+ "step": 13000
246
+ },
247
+ {
248
+ "epoch": 7.959905660377358,
249
+ "eval_accuracy_background": 0.7526398125858199,
250
+ "eval_accuracy_double_plant": 0.5617747160876485,
251
+ "eval_accuracy_drydown": 0.7416084130046285,
252
+ "eval_accuracy_endrow": 0.5035272340399668,
253
+ "eval_accuracy_nutrient_deficiency": 0.531805617343839,
254
+ "eval_accuracy_planter_skip": 0.6247280891527807,
255
+ "eval_accuracy_water": 0.9841431099962412,
256
+ "eval_accuracy_waterway": 0.4783465745730863,
257
+ "eval_accuracy_weed_cluster": 0.47373664220884076,
258
+ "eval_iou_background": 0.6514530993735008,
259
+ "eval_iou_double_plant": 0.3221799261469335,
260
+ "eval_iou_drydown": 0.496553324050437,
261
+ "eval_iou_endrow": 0.25636256499323895,
262
+ "eval_iou_nutrient_deficiency": 0.36014685659389084,
263
+ "eval_iou_planter_skip": 0.3605061710097777,
264
+ "eval_iou_water": 0.4365396537110217,
265
+ "eval_iou_waterway": 0.14037628353126624,
266
+ "eval_iou_weed_cluster": 0.3214568250037232,
267
+ "eval_loss": 1.5932934284210205,
268
+ "eval_mean_accuracy": 0.6280344676658723,
269
+ "eval_mean_iou": 0.37173052271264334,
270
+ "eval_overall_accuracy": 0.7170902700988321,
271
+ "eval_runtime": 529.3378,
272
+ "eval_samples_per_second": 16.107,
273
+ "eval_steps_per_second": 0.504,
274
+ "step": 13500
275
+ },
276
+ {
277
+ "epoch": 8.254716981132075,
278
+ "grad_norm": 1.4569978713989258,
279
+ "learning_rate": 7.305091043730557e-05,
280
+ "loss": 1.0271,
281
+ "step": 14000
282
+ },
283
+ {
284
+ "epoch": 8.84433962264151,
285
+ "grad_norm": 2.9907829761505127,
286
+ "learning_rate": 6.837101139557501e-05,
287
+ "loss": 1.0036,
288
+ "step": 15000
289
+ },
290
+ {
291
+ "epoch": 9.286556603773585,
292
+ "eval_accuracy_background": 0.7598414375076388,
293
+ "eval_accuracy_double_plant": 0.5177256602627419,
294
+ "eval_accuracy_drydown": 0.7154793750660278,
295
+ "eval_accuracy_endrow": 0.5012670182902391,
296
+ "eval_accuracy_nutrient_deficiency": 0.5368322761783371,
297
+ "eval_accuracy_planter_skip": 0.5468935952263279,
298
+ "eval_accuracy_water": 0.937019395570021,
299
+ "eval_accuracy_waterway": 0.49325926542068943,
300
+ "eval_accuracy_weed_cluster": 0.4951997709703656,
301
+ "eval_iou_background": 0.653841296382127,
302
+ "eval_iou_double_plant": 0.3556917179689535,
303
+ "eval_iou_drydown": 0.48554419750993316,
304
+ "eval_iou_endrow": 0.2419919783626517,
305
+ "eval_iou_nutrient_deficiency": 0.3544690920377324,
306
+ "eval_iou_planter_skip": 0.3770845151472525,
307
+ "eval_iou_water": 0.4813163648581422,
308
+ "eval_iou_waterway": 0.1553945460962498,
309
+ "eval_iou_weed_cluster": 0.32913578509118946,
310
+ "eval_loss": 1.6111669540405273,
311
+ "eval_mean_accuracy": 0.6115019771658209,
312
+ "eval_mean_iou": 0.3816077214949146,
313
+ "eval_overall_accuracy": 0.7184015448031484,
314
+ "eval_runtime": 530.1576,
315
+ "eval_samples_per_second": 16.082,
316
+ "eval_steps_per_second": 0.504,
317
+ "step": 15750
318
+ },
319
+ {
320
+ "epoch": 9.433962264150944,
321
+ "grad_norm": 1.7846933603286743,
322
+ "learning_rate": 6.34967318063877e-05,
323
+ "loss": 0.9725,
324
+ "step": 16000
325
+ },
326
+ {
327
+ "epoch": 10.023584905660377,
328
+ "grad_norm": 1.7969969511032104,
329
+ "learning_rate": 5.847964559317128e-05,
330
+ "loss": 0.9677,
331
+ "step": 17000
332
+ },
333
+ {
334
+ "epoch": 10.61320754716981,
335
+ "grad_norm": 3.443549156188965,
336
+ "learning_rate": 5.3372837691956955e-05,
337
+ "loss": 0.9443,
338
+ "step": 18000
339
+ },
340
+ {
341
+ "epoch": 10.61320754716981,
342
+ "eval_accuracy_background": 0.7705968530221318,
343
+ "eval_accuracy_double_plant": 0.5181034194377331,
344
+ "eval_accuracy_drydown": 0.6903163741892455,
345
+ "eval_accuracy_endrow": 0.4835231803753014,
346
+ "eval_accuracy_nutrient_deficiency": 0.5916561216478401,
347
+ "eval_accuracy_planter_skip": 0.5335884546146096,
348
+ "eval_accuracy_water": 0.9620806979435361,
349
+ "eval_accuracy_waterway": 0.43880698473219176,
350
+ "eval_accuracy_weed_cluster": 0.3965401585703216,
351
+ "eval_iou_background": 0.6599883230602321,
352
+ "eval_iou_double_plant": 0.34125555031911986,
353
+ "eval_iou_drydown": 0.48090155814983737,
354
+ "eval_iou_endrow": 0.27739686571309646,
355
+ "eval_iou_nutrient_deficiency": 0.37205107367408047,
356
+ "eval_iou_planter_skip": 0.34566214329213907,
357
+ "eval_iou_water": 0.4751891278946817,
358
+ "eval_iou_waterway": 0.18319416480334516,
359
+ "eval_iou_weed_cluster": 0.28830863200785295,
360
+ "eval_loss": 1.7351711988449097,
361
+ "eval_mean_accuracy": 0.5983569160592123,
362
+ "eval_mean_iou": 0.38043860432382054,
363
+ "eval_overall_accuracy": 0.7224762810473321,
364
+ "eval_runtime": 547.3444,
365
+ "eval_samples_per_second": 15.577,
366
+ "eval_steps_per_second": 0.488,
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 sha256:2201adcd7aa770e7087bc147820e84282a45a5e69835d16fedd13a67bc092189
3
+ size 4539