Anonymous1Researcher commited on
Commit
af890a2
·
verified ·
1 Parent(s): 796d04e

Upload image_classification.json

Browse files
Files changed (1) hide show
  1. image_classification.json +683 -0
image_classification.json ADDED
@@ -0,0 +1,683 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "task_name": "image_classification",
3
+ "models": [
4
+ {
5
+ "model_name": "convnextpico",
6
+ "model_script_path": "/data/chl0045/project/benchmark_v2/models/convnextpico.py",
7
+ "runs": [
8
+ {
9
+ "dataset": "cifar10",
10
+ "model_id": "convnextpi_000",
11
+ "hardware": {
12
+ "gpu": "NVIDIA H100 NVL",
13
+ "training_time_hrs": 0.045935
14
+ },
15
+ "stage": "classification",
16
+ "base_model_name": "convnext_pico",
17
+ "base_model_path": null,
18
+ "config_path": "train/image_classification/convnextpico/config/cifar10.json",
19
+ "augmentation_path": null,
20
+ "hyperparams": {
21
+ "lr": 0.0005,
22
+ "optimizer": "adamw",
23
+ "weight_decay": 0.0001,
24
+ "batch_size": 256,
25
+ "epochs": 4,
26
+ "seed": 42,
27
+ "scheduler": "cosine_annealing",
28
+ "image_size": 224,
29
+ "input_channels": 3
30
+ },
31
+ "training_info": {
32
+ "best_epoch": 4,
33
+ "best_val_acc": 96.99,
34
+ "best_val_loss": 0.104373,
35
+ "total_params": 8537802,
36
+ "total_params_in_m": "8.537802M",
37
+ "trainable_params": 8537802,
38
+ "trainable_params_in_m": "8.537802M",
39
+ "embedding_params": null,
40
+ "embedding_params_in_m": null,
41
+ "non_embedding_params": null,
42
+ "non_embedding_params_in_m": null,
43
+ "epoch_logs": []
44
+ },
45
+ "ckpt_path": "zoo/cifar10/convnextpico/convnextpi_000.pth"
46
+ }
47
+ ]
48
+ },
49
+ {
50
+ "model_name": "mobilenetlarge",
51
+ "model_script_path": "/data/chl0045/project/benchmark_v2/models/mobilenetlarge.py",
52
+ "runs": [
53
+ {
54
+ "dataset": "cifar10",
55
+ "model_id": "mobilenetl_000",
56
+ "hardware": {
57
+ "gpu": "NVIDIA H100 NVL",
58
+ "training_time_hrs": 0.034388
59
+ },
60
+ "stage": "classification",
61
+ "base_model_name": "mobilenetv3_large_100",
62
+ "base_model_path": null,
63
+ "config_path": "train/image_classification/mobilenetlarge/config/cifar10.json",
64
+ "augmentation_path": null,
65
+ "hyperparams": {
66
+ "lr": 0.0005,
67
+ "optimizer": "adamw",
68
+ "weight_decay": 0.0001,
69
+ "batch_size": 256,
70
+ "epochs": 5,
71
+ "seed": 42,
72
+ "scheduler": "cosine_annealing",
73
+ "image_size": 224,
74
+ "input_channels": 3
75
+ },
76
+ "training_info": {
77
+ "best_epoch": 4,
78
+ "best_val_acc": 95.53,
79
+ "best_val_loss": 0.181964,
80
+ "total_params": 4214842,
81
+ "total_params_in_m": "4.214842M",
82
+ "trainable_params": 4214842,
83
+ "trainable_params_in_m": "4.214842M",
84
+ "embedding_params": null,
85
+ "embedding_params_in_m": null,
86
+ "non_embedding_params": null,
87
+ "non_embedding_params_in_m": null,
88
+ "epoch_logs": []
89
+ },
90
+ "ckpt_path": "zoo/cifar10/mobilenetlarge/mobilenetl_000.pth"
91
+ },
92
+ {
93
+ "dataset": "mnist",
94
+ "model_id": "mobilenetl_001",
95
+ "hardware": {
96
+ "gpu": "NVIDIA H100 NVL",
97
+ "training_time_hrs": 0.040502
98
+ },
99
+ "stage": "classification",
100
+ "base_model_name": "mobilenetv3_large_100",
101
+ "base_model_path": null,
102
+ "config_path": "train/image_classification/mobilenetlarge/config/mnist.json",
103
+ "augmentation_path": null,
104
+ "hyperparams": {
105
+ "lr": 0.0005,
106
+ "optimizer": "adamw",
107
+ "weight_decay": 0.0001,
108
+ "batch_size": 256,
109
+ "epochs": 5,
110
+ "seed": 42,
111
+ "scheduler": "cosine_annealing",
112
+ "image_size": 224,
113
+ "input_channels": 3
114
+ },
115
+ "training_info": {
116
+ "best_epoch": 4,
117
+ "best_val_acc": 99.48,
118
+ "best_val_loss": 0.017188,
119
+ "total_params": 4214842,
120
+ "total_params_in_m": "4.214842M",
121
+ "trainable_params": 4214842,
122
+ "trainable_params_in_m": "4.214842M",
123
+ "embedding_params": null,
124
+ "embedding_params_in_m": null,
125
+ "non_embedding_params": null,
126
+ "non_embedding_params_in_m": null,
127
+ "epoch_logs": []
128
+ },
129
+ "ckpt_path": "zoo/mnist/mobilenetlarge/mobilenetl_001.pth"
130
+ }
131
+ ]
132
+ },
133
+ {
134
+ "model_name": "image_cnn",
135
+ "model_script_path": "/data/chl0045/project/benchmark_v2/models/image_cnn.py",
136
+ "runs": [
137
+ {
138
+ "dataset": "mnist",
139
+ "model_id": "imagecnn_000",
140
+ "hardware": {
141
+ "gpu": "NVIDIA H100 NVL",
142
+ "training_time_hrs": 0.01513
143
+ },
144
+ "stage": "classification",
145
+ "base_model_name": null,
146
+ "base_model_path": null,
147
+ "config_path": "train/image_classification/image_cnn/config/mnist.json",
148
+ "augmentation_path": null,
149
+ "hyperparams": {
150
+ "lr": 0.001,
151
+ "optimizer": "adamw",
152
+ "weight_decay": 0.0001,
153
+ "batch_size": 256,
154
+ "epochs": 20,
155
+ "seed": 42,
156
+ "scheduler": "cosine_annealing",
157
+ "image_size": 32,
158
+ "input_channels": 3
159
+ },
160
+ "training_info": {
161
+ "best_epoch": 16,
162
+ "best_val_acc": 99.33,
163
+ "best_val_loss": 0.025129,
164
+ "total_params": 153962,
165
+ "total_params_in_m": "0.153962M",
166
+ "trainable_params": 153962,
167
+ "trainable_params_in_m": "0.153962M",
168
+ "embedding_params": null,
169
+ "embedding_params_in_m": null,
170
+ "non_embedding_params": null,
171
+ "non_embedding_params_in_m": null,
172
+ "epoch_logs": []
173
+ },
174
+ "ckpt_path": "zoo/mnist/image_cnn/imagecnn_000.pth"
175
+ },
176
+ {
177
+ "dataset": "cifar10",
178
+ "model_id": "imagecnn_001",
179
+ "hardware": {
180
+ "gpu": "NVIDIA H100 NVL",
181
+ "training_time_hrs": 0.009136
182
+ },
183
+ "stage": "classification",
184
+ "base_model_name": null,
185
+ "base_model_path": null,
186
+ "config_path": "train/image_classification/image_cnn/config/cifar10.json",
187
+ "augmentation_path": null,
188
+ "hyperparams": {
189
+ "lr": 0.001,
190
+ "optimizer": "adamw",
191
+ "weight_decay": 0.0001,
192
+ "batch_size": 256,
193
+ "epochs": 20,
194
+ "seed": 42,
195
+ "scheduler": "cosine_annealing",
196
+ "image_size": 32,
197
+ "input_channels": 3
198
+ },
199
+ "training_info": {
200
+ "best_epoch": 16,
201
+ "best_val_acc": 69.14,
202
+ "best_val_loss": 0.917102,
203
+ "total_params": 153962,
204
+ "total_params_in_m": "0.153962M",
205
+ "trainable_params": 153962,
206
+ "trainable_params_in_m": "0.153962M",
207
+ "embedding_params": null,
208
+ "embedding_params_in_m": null,
209
+ "non_embedding_params": null,
210
+ "non_embedding_params_in_m": null,
211
+ "epoch_logs": []
212
+ },
213
+ "ckpt_path": "zoo/cifar10/image_cnn/imagecnn_001.pth"
214
+ },
215
+ {
216
+ "dataset": "cifar10",
217
+ "model_id": "imagecnn_002",
218
+ "hardware": {
219
+ "gpu": "NVIDIA H100 NVL",
220
+ "training_time_hrs": 0.011604
221
+ },
222
+ "stage": "classification",
223
+ "base_model_name": null,
224
+ "base_model_path": null,
225
+ "config_path": "train/image_classification/image_cnn/config/cifar10.json",
226
+ "augmentation_path": null,
227
+ "hyperparams": {
228
+ "lr": 0.001,
229
+ "optimizer": "adamw",
230
+ "weight_decay": 0.0001,
231
+ "batch_size": 256,
232
+ "epochs": 20,
233
+ "seed": 42,
234
+ "scheduler": "cosine_annealing",
235
+ "image_size": 32,
236
+ "input_channels": 3
237
+ },
238
+ "training_info": {
239
+ "best_epoch": 18,
240
+ "best_val_acc": 69.47,
241
+ "best_val_loss": 0.912958,
242
+ "total_params": 153962,
243
+ "total_params_in_m": "0.153962M",
244
+ "trainable_params": 153962,
245
+ "trainable_params_in_m": "0.153962M",
246
+ "embedding_params": null,
247
+ "embedding_params_in_m": null,
248
+ "non_embedding_params": null,
249
+ "non_embedding_params_in_m": null,
250
+ "epoch_logs": []
251
+ },
252
+ "ckpt_path": "zoo/cifar10/image_cnn/imagecnn_002.pth"
253
+ },
254
+ {
255
+ "dataset": "svhn",
256
+ "model_id": "imagecnn_003",
257
+ "hardware": {
258
+ "gpu": "NVIDIA H100 NVL",
259
+ "training_time_hrs": 0.01523
260
+ },
261
+ "stage": "classification",
262
+ "base_model_name": null,
263
+ "base_model_path": null,
264
+ "config_path": "train/image_classification/image_cnn/config/svhn.json",
265
+ "augmentation_path": null,
266
+ "hyperparams": {
267
+ "lr": 0.001,
268
+ "optimizer": "adamw",
269
+ "weight_decay": 0.0001,
270
+ "batch_size": 256,
271
+ "epochs": 20,
272
+ "seed": 42,
273
+ "scheduler": "cosine_annealing",
274
+ "image_size": 32,
275
+ "input_channels": 3
276
+ },
277
+ "training_info": {
278
+ "best_epoch": 14,
279
+ "best_val_acc": 89.27858,
280
+ "best_val_loss": 0.407848,
281
+ "total_params": 153962,
282
+ "total_params_in_m": "0.153962M",
283
+ "trainable_params": 153962,
284
+ "trainable_params_in_m": "0.153962M",
285
+ "embedding_params": null,
286
+ "embedding_params_in_m": null,
287
+ "non_embedding_params": null,
288
+ "non_embedding_params_in_m": null,
289
+ "epoch_logs": []
290
+ },
291
+ "ckpt_path": "zoo/svhn/image_cnn/imagecnn_003.pth"
292
+ },
293
+ {
294
+ "dataset": "stl10",
295
+ "model_id": "imagecnn_004",
296
+ "hardware": {
297
+ "gpu": "NVIDIA H100 NVL",
298
+ "training_time_hrs": 0.006417
299
+ },
300
+ "stage": "classification",
301
+ "base_model_name": null,
302
+ "base_model_path": null,
303
+ "config_path": "train/image_classification/image_cnn/config/stl10.json",
304
+ "augmentation_path": null,
305
+ "hyperparams": {
306
+ "lr": 0.001,
307
+ "optimizer": "adamw",
308
+ "weight_decay": 0.0001,
309
+ "batch_size": 256,
310
+ "epochs": 20,
311
+ "seed": 42,
312
+ "scheduler": "cosine_annealing",
313
+ "image_size": 32,
314
+ "input_channels": 3
315
+ },
316
+ "training_info": {
317
+ "best_epoch": 18,
318
+ "best_val_acc": 51.675,
319
+ "best_val_loss": 1.341974,
320
+ "total_params": 153962,
321
+ "total_params_in_m": "0.153962M",
322
+ "trainable_params": 153962,
323
+ "trainable_params_in_m": "0.153962M",
324
+ "embedding_params": null,
325
+ "embedding_params_in_m": null,
326
+ "non_embedding_params": null,
327
+ "non_embedding_params_in_m": null,
328
+ "epoch_logs": []
329
+ },
330
+ "ckpt_path": "zoo/stl10/image_cnn/imagecnn_004.pth"
331
+ }
332
+ ]
333
+ },
334
+ {
335
+ "model_name": "vitmini_full",
336
+ "model_script_path": "/data/chl0045/project/benchmark_v2/models/vitmini.py",
337
+ "runs": [
338
+ {
339
+ "dataset": "mnist",
340
+ "model_id": "vitminiful_000",
341
+ "hardware": {
342
+ "gpu": "NVIDIA H100 NVL",
343
+ "training_time_hrs": 0.076045
344
+ },
345
+ "stage": "classification",
346
+ "base_model_name": "vit_tiny_patch16_224",
347
+ "base_model_path": "/data/chl0045/project/benchmark_v2/zoo/cifar100/vitmini/vitmini_002.pth",
348
+ "config_path": "train/image_classification/vitmini/config/mnist.json",
349
+ "augmentation_path": null,
350
+ "hyperparams": {
351
+ "lr": 0.001,
352
+ "optimizer": "adamw",
353
+ "weight_decay": 0.0001,
354
+ "batch_size": 256,
355
+ "epochs": 10,
356
+ "seed": 42,
357
+ "scheduler": "cosine_annealing",
358
+ "image_size": 224,
359
+ "input_channels": 3,
360
+ "finetune_mode": "full"
361
+ },
362
+ "training_info": {
363
+ "best_epoch": 9,
364
+ "best_val_acc": 98.58,
365
+ "best_val_loss": 0.050301,
366
+ "total_params": 918410,
367
+ "total_params_in_m": "0.918410M",
368
+ "trainable_params": 918410,
369
+ "trainable_params_in_m": "0.918410M",
370
+ "embedding_params": null,
371
+ "embedding_params_in_m": null,
372
+ "non_embedding_params": null,
373
+ "non_embedding_params_in_m": null,
374
+ "epoch_logs": []
375
+ },
376
+ "ckpt_path": "zoo/mnist/vitmini_full/vitminiful_000.pth"
377
+ },
378
+ {
379
+ "dataset": "cifar10",
380
+ "model_id": "vitminiful_001",
381
+ "hardware": {
382
+ "gpu": "NVIDIA H100 NVL",
383
+ "training_time_hrs": 0.055205
384
+ },
385
+ "stage": "classification",
386
+ "base_model_name": "vit_tiny_patch16_224",
387
+ "base_model_path": "/data/chl0045/project/benchmark_v2/zoo/cifar100/vitmini/vitmini_002.pth",
388
+ "config_path": "train/image_classification/vitmini/config/cifar10.json",
389
+ "augmentation_path": null,
390
+ "hyperparams": {
391
+ "lr": 0.001,
392
+ "optimizer": "adamw",
393
+ "weight_decay": 0.0001,
394
+ "batch_size": 256,
395
+ "epochs": 10,
396
+ "seed": 42,
397
+ "scheduler": "cosine_annealing",
398
+ "image_size": 224,
399
+ "input_channels": 3,
400
+ "finetune_mode": "full"
401
+ },
402
+ "training_info": {
403
+ "best_epoch": 10,
404
+ "best_val_acc": 72.07,
405
+ "best_val_loss": 0.931169,
406
+ "total_params": 918410,
407
+ "total_params_in_m": "0.918410M",
408
+ "trainable_params": 918410,
409
+ "trainable_params_in_m": "0.918410M",
410
+ "embedding_params": null,
411
+ "embedding_params_in_m": null,
412
+ "non_embedding_params": null,
413
+ "non_embedding_params_in_m": null,
414
+ "epoch_logs": []
415
+ },
416
+ "ckpt_path": "zoo/cifar10/vitmini_full/vitminiful_001.pth"
417
+ }
418
+ ]
419
+ },
420
+ {
421
+ "model_name": "vitmini_lora",
422
+ "model_script_path": "/data/chl0045/project/benchmark_v2/models/vitmini.py",
423
+ "runs": [
424
+ {
425
+ "dataset": "mnist",
426
+ "model_id": "vitminilor_000",
427
+ "hardware": {
428
+ "gpu": "NVIDIA H100 NVL",
429
+ "training_time_hrs": 0.076512
430
+ },
431
+ "stage": "classification",
432
+ "base_model_name": "vit_tiny_patch16_224",
433
+ "base_model_path": "/data/chl0045/project/benchmark_v2/zoo/cifar100/vitmini/vitmini_002.pth",
434
+ "config_path": "train/image_classification/vitmini/config/mnist.json",
435
+ "augmentation_path": null,
436
+ "hyperparams": {
437
+ "lr": 0.001,
438
+ "optimizer": "adamw",
439
+ "weight_decay": 0.0001,
440
+ "batch_size": 256,
441
+ "epochs": 10,
442
+ "seed": 42,
443
+ "scheduler": "cosine_annealing",
444
+ "image_size": 224,
445
+ "input_channels": 3,
446
+ "finetune_mode": "lora",
447
+ "lora_r": 8,
448
+ "lora_alpha": 16,
449
+ "lora_dropout": 0.0
450
+ },
451
+ "training_info": {
452
+ "best_epoch": 10,
453
+ "best_val_acc": 95.17,
454
+ "best_val_loss": 0.154653,
455
+ "total_params": 942986,
456
+ "total_params_in_m": "0.942986M",
457
+ "trainable_params": 25866,
458
+ "trainable_params_in_m": "0.025866M",
459
+ "embedding_params": null,
460
+ "embedding_params_in_m": null,
461
+ "non_embedding_params": null,
462
+ "non_embedding_params_in_m": null,
463
+ "epoch_logs": []
464
+ },
465
+ "ckpt_path": "zoo/mnist/vitmini_lora/vitminilor_000.pth"
466
+ },
467
+ {
468
+ "dataset": "cifar10",
469
+ "model_id": "vitminilor_001",
470
+ "hardware": {
471
+ "gpu": "NVIDIA H100 NVL",
472
+ "training_time_hrs": 0.055041
473
+ },
474
+ "stage": "classification",
475
+ "base_model_name": "vit_tiny_patch16_224",
476
+ "base_model_path": "/data/chl0045/project/benchmark_v2/zoo/cifar100/vitmini/vitmini_002.pth",
477
+ "config_path": "train/image_classification/vitmini/config/cifar10.json",
478
+ "augmentation_path": null,
479
+ "hyperparams": {
480
+ "lr": 0.001,
481
+ "optimizer": "adamw",
482
+ "weight_decay": 0.0001,
483
+ "batch_size": 256,
484
+ "epochs": 10,
485
+ "seed": 42,
486
+ "scheduler": "cosine_annealing",
487
+ "image_size": 224,
488
+ "input_channels": 3,
489
+ "finetune_mode": "lora",
490
+ "lora_r": 8,
491
+ "lora_alpha": 16,
492
+ "lora_dropout": 0.0
493
+ },
494
+ "training_info": {
495
+ "best_epoch": 10,
496
+ "best_val_acc": 65.39,
497
+ "best_val_loss": 0.974446,
498
+ "total_params": 942986,
499
+ "total_params_in_m": "0.942986M",
500
+ "trainable_params": 25866,
501
+ "trainable_params_in_m": "0.025866M",
502
+ "embedding_params": null,
503
+ "embedding_params_in_m": null,
504
+ "non_embedding_params": null,
505
+ "non_embedding_params_in_m": null,
506
+ "epoch_logs": []
507
+ },
508
+ "ckpt_path": "zoo/cifar10/vitmini_lora/vitminilor_001.pth"
509
+ },
510
+ {
511
+ "dataset": "mnist",
512
+ "model_id": "vitminilor_002",
513
+ "hardware": {
514
+ "gpu": "NVIDIA H100 NVL",
515
+ "training_time_hrs": 0.151413
516
+ },
517
+ "stage": "classification",
518
+ "base_model_name": "vit_tiny_patch16_224",
519
+ "base_model_path": "/data/chl0045/project/benchmark_v2/zoo/cifar100/vitmini/vitmini_002.pth",
520
+ "config_path": "train/image_classification/vitmini/config/mnist.json",
521
+ "augmentation_path": null,
522
+ "hyperparams": {
523
+ "lr": 0.001,
524
+ "optimizer": "adamw",
525
+ "weight_decay": 0.0001,
526
+ "batch_size": 256,
527
+ "epochs": 20,
528
+ "seed": 42,
529
+ "scheduler": "cosine_annealing",
530
+ "image_size": 224,
531
+ "input_channels": 3,
532
+ "finetune_mode": "lora",
533
+ "lora_r": 8,
534
+ "lora_alpha": 16,
535
+ "lora_dropout": 0.0
536
+ },
537
+ "training_info": {
538
+ "best_epoch": 20,
539
+ "best_val_acc": 96.73,
540
+ "best_val_loss": 0.1081,
541
+ "total_params": 942986,
542
+ "total_params_in_m": "0.942986M",
543
+ "trainable_params": 25866,
544
+ "trainable_params_in_m": "0.025866M",
545
+ "embedding_params": null,
546
+ "embedding_params_in_m": null,
547
+ "non_embedding_params": null,
548
+ "non_embedding_params_in_m": null,
549
+ "epoch_logs": []
550
+ },
551
+ "ckpt_path": "zoo/mnist/vitmini_lora/vitminilor_002.pth"
552
+ },
553
+ {
554
+ "dataset": "cifar10",
555
+ "model_id": "vitminilor_003",
556
+ "hardware": {
557
+ "gpu": "NVIDIA H100 NVL",
558
+ "training_time_hrs": 0.110626
559
+ },
560
+ "stage": "classification",
561
+ "base_model_name": "vit_tiny_patch16_224",
562
+ "base_model_path": "/data/chl0045/project/benchmark_v2/zoo/cifar100/vitmini/vitmini_002.pth",
563
+ "config_path": "train/image_classification/vitmini/config/cifar10.json",
564
+ "augmentation_path": null,
565
+ "hyperparams": {
566
+ "lr": 0.001,
567
+ "optimizer": "adamw",
568
+ "weight_decay": 0.0001,
569
+ "batch_size": 256,
570
+ "epochs": 20,
571
+ "seed": 42,
572
+ "scheduler": "cosine_annealing",
573
+ "image_size": 224,
574
+ "input_channels": 3,
575
+ "finetune_mode": "lora",
576
+ "lora_r": 8,
577
+ "lora_alpha": 16,
578
+ "lora_dropout": 0.0
579
+ },
580
+ "training_info": {
581
+ "best_epoch": 17,
582
+ "best_val_acc": 66.83,
583
+ "best_val_loss": 0.942915,
584
+ "total_params": 942986,
585
+ "total_params_in_m": "0.942986M",
586
+ "trainable_params": 25866,
587
+ "trainable_params_in_m": "0.025866M",
588
+ "embedding_params": null,
589
+ "embedding_params_in_m": null,
590
+ "non_embedding_params": null,
591
+ "non_embedding_params_in_m": null,
592
+ "epoch_logs": []
593
+ },
594
+ "ckpt_path": "zoo/cifar10/vitmini_lora/vitminilor_003.pth"
595
+ }
596
+ ]
597
+ },
598
+ {
599
+ "model_name": "resnetmini",
600
+ "model_script_path": "/data/chl0045/project/benchmark_v2/models/resnetmini.py",
601
+ "runs": [
602
+ {
603
+ "dataset": "mnist",
604
+ "model_id": "resnetmini_000",
605
+ "hardware": {
606
+ "gpu": "NVIDIA H100 NVL",
607
+ "training_time_hrs": 0.017248
608
+ },
609
+ "stage": "classification",
610
+ "base_model_name": null,
611
+ "base_model_path": null,
612
+ "config_path": "train/image_classification/resnetmini/config/mnist.json",
613
+ "augmentation_path": null,
614
+ "hyperparams": {
615
+ "lr": 0.001,
616
+ "optimizer": "adamw",
617
+ "weight_decay": 0.0001,
618
+ "batch_size": 256,
619
+ "epochs": 20,
620
+ "seed": 42,
621
+ "scheduler": "cosine_annealing",
622
+ "image_size": 32,
623
+ "input_channels": 3
624
+ },
625
+ "training_info": {
626
+ "best_epoch": 18,
627
+ "best_val_acc": 99.53,
628
+ "best_val_loss": 0.016165,
629
+ "total_params": 927398,
630
+ "total_params_in_m": "0.927398M",
631
+ "trainable_params": 927398,
632
+ "trainable_params_in_m": "0.927398M",
633
+ "embedding_params": null,
634
+ "embedding_params_in_m": null,
635
+ "non_embedding_params": null,
636
+ "non_embedding_params_in_m": null,
637
+ "epoch_logs": []
638
+ },
639
+ "ckpt_path": "zoo/mnist/resnetmini/resnetmini_000.pth"
640
+ },
641
+ {
642
+ "dataset": "cifar10",
643
+ "model_id": "resnetmini_001",
644
+ "hardware": {
645
+ "gpu": "NVIDIA H100 NVL",
646
+ "training_time_hrs": 0.010546
647
+ },
648
+ "stage": "classification",
649
+ "base_model_name": null,
650
+ "base_model_path": null,
651
+ "config_path": "train/image_classification/resnetmini/config/cifar10.json",
652
+ "augmentation_path": null,
653
+ "hyperparams": {
654
+ "lr": 0.001,
655
+ "optimizer": "adamw",
656
+ "weight_decay": 0.0001,
657
+ "batch_size": 256,
658
+ "epochs": 20,
659
+ "seed": 42,
660
+ "scheduler": "cosine_annealing",
661
+ "image_size": 32,
662
+ "input_channels": 3
663
+ },
664
+ "training_info": {
665
+ "best_epoch": 18,
666
+ "best_val_acc": 79.32,
667
+ "best_val_loss": 0.924181,
668
+ "total_params": 927398,
669
+ "total_params_in_m": "0.927398M",
670
+ "trainable_params": 927398,
671
+ "trainable_params_in_m": "0.927398M",
672
+ "embedding_params": null,
673
+ "embedding_params_in_m": null,
674
+ "non_embedding_params": null,
675
+ "non_embedding_params_in_m": null,
676
+ "epoch_logs": []
677
+ },
678
+ "ckpt_path": "zoo/cifar10/resnetmini/resnetmini_001.pth"
679
+ }
680
+ ]
681
+ }
682
+ ]
683
+ }