khanhvinh9 commited on
Commit
08b240a
·
verified ·
1 Parent(s): ef43d66

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +35 -0
  2. cifar100/middle/learnable-naive/0.0_00seed20/config.json +524 -0
  3. cifar100/middle/learnable-naive/0.0_00seed20/flax_model.msgpack +3 -0
  4. cifar100/middle/learnable-naive/0.125_00seed20/config.json +525 -0
  5. cifar100/middle/learnable-naive/0.125_00seed20/finetune/best_31280/config.json +525 -0
  6. cifar100/middle/learnable-naive/0.125_00seed20/finetune/best_31280/flax_model.msgpack +3 -0
  7. cifar100/middle/learnable-naive/0.125_00seed20/finetune/config.json +525 -0
  8. cifar100/middle/learnable-naive/0.125_00seed20/finetune/last_78200/config.json +525 -0
  9. cifar100/middle/learnable-naive/0.125_00seed20/finetune/last_78200/flax_model.msgpack +3 -0
  10. cifar100/middle/learnable-naive/0.125_00seed20/flax_model.msgpack +3 -0
  11. cifar100/middle/learnable-naive/0.25_00seed20/config.json +525 -0
  12. cifar100/middle/learnable-naive/0.25_00seed20/finetune/best_25806/config.json +525 -0
  13. cifar100/middle/learnable-naive/0.25_00seed20/finetune/best_25806/flax_model.msgpack +3 -0
  14. cifar100/middle/learnable-naive/0.25_00seed20/finetune/config.json +525 -0
  15. cifar100/middle/learnable-naive/0.25_00seed20/finetune/last_78200/config.json +525 -0
  16. cifar100/middle/learnable-naive/0.25_00seed20/finetune/last_78200/flax_model.msgpack +3 -0
  17. cifar100/middle/learnable-naive/0.25_00seed20/flax_model.msgpack +3 -0
  18. cifar100/middle/learnable-naive/0.375_00seed20/config.json +525 -0
  19. cifar100/middle/learnable-naive/0.375_00seed20/finetune/best_57868/config.json +525 -0
  20. cifar100/middle/learnable-naive/0.375_00seed20/finetune/best_57868/flax_model.msgpack +3 -0
  21. cifar100/middle/learnable-naive/0.375_00seed20/finetune/config.json +525 -0
  22. cifar100/middle/learnable-naive/0.375_00seed20/finetune/last_78200/config.json +525 -0
  23. cifar100/middle/learnable-naive/0.375_00seed20/finetune/last_78200/flax_model.msgpack +3 -0
  24. cifar100/middle/learnable-naive/0.375_00seed20/flax_model.msgpack +3 -0
  25. cifar100/middle/learnable-naive/0.5_00seed20/config.json +524 -0
  26. cifar100/middle/learnable-naive/0.5_00seed20/finetune/best_40664/config.json +524 -0
  27. cifar100/middle/learnable-naive/0.5_00seed20/finetune/best_40664/flax_model.msgpack +3 -0
  28. cifar100/middle/learnable-naive/0.5_00seed20/finetune/config.json +524 -0
  29. cifar100/middle/learnable-naive/0.5_00seed20/finetune/last_78200/config.json +524 -0
  30. cifar100/middle/learnable-naive/0.5_00seed20/finetune/last_78200/flax_model.msgpack +3 -0
  31. cifar100/middle/learnable-naive/0.5_00seed20/flax_model.msgpack +3 -0
  32. cifar100/middle/learnable-naive/0.625_00seed20/config.json +525 -0
  33. cifar100/middle/learnable-naive/0.625_00seed20/finetune/best_19550/config.json +525 -0
  34. cifar100/middle/learnable-naive/0.625_00seed20/finetune/best_19550/flax_model.msgpack +3 -0
  35. cifar100/middle/learnable-naive/0.625_00seed20/finetune/config.json +525 -0
  36. cifar100/middle/learnable-naive/0.625_00seed20/finetune/last_78200/config.json +525 -0
  37. cifar100/middle/learnable-naive/0.625_00seed20/finetune/last_78200/flax_model.msgpack +3 -0
  38. cifar100/middle/learnable-naive/0.625_00seed20/flax_model.msgpack +3 -0
  39. cifar100/middle/learnable-naive/0.75_00seed20/config.json +525 -0
  40. cifar100/middle/learnable-naive/0.75_00seed20/finetune/best_34408/config.json +525 -0
  41. cifar100/middle/learnable-naive/0.75_00seed20/finetune/best_34408/flax_model.msgpack +3 -0
  42. cifar100/middle/learnable-naive/0.75_00seed20/finetune/config.json +525 -0
  43. cifar100/middle/learnable-naive/0.75_00seed20/finetune/last_78200/config.json +525 -0
  44. cifar100/middle/learnable-naive/0.75_00seed20/finetune/last_78200/flax_model.msgpack +3 -0
  45. cifar100/middle/learnable-naive/0.75_00seed20/flax_model.msgpack +3 -0
  46. cifar100/middle/learnable-naive/0.875_00seed20/config.json +525 -0
  47. cifar100/middle/learnable-naive/0.875_00seed20/finetune/best_35190/config.json +525 -0
  48. cifar100/middle/learnable-naive/0.875_00seed20/finetune/best_35190/flax_model.msgpack +3 -0
  49. cifar100/middle/learnable-naive/0.875_00seed20/finetune/config.json +525 -0
  50. cifar100/middle/learnable-naive/0.875_00seed20/finetune/last_78200/config.json +525 -0
.gitattributes CHANGED
@@ -33,3 +33,38 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-192/best_675450/d/ab46af029fcc971be7cdb4434f4c399a filter=lfs diff=lfs merge=lfs -text
37
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-192/best_675450/ocdbt.process_0/d/011ebb354036267a7989690219fc04a0 filter=lfs diff=lfs merge=lfs -text
38
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-192/best_675450/ocdbt.process_0/d/bc1f5220a7d00d05bd6e08589bc5279f filter=lfs diff=lfs merge=lfs -text
39
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-192/best_675450/ocdbt.process_0/d/c43783a4ebaa2d07d94e16c0e1c9dfe4 filter=lfs diff=lfs merge=lfs -text
40
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-192/last_900600/d/96a7c6afbe5c6ddb29421af72f42e17b filter=lfs diff=lfs merge=lfs -text
41
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-192/last_900600/ocdbt.process_0/d/982e45f8e9df68ad5a497d969e184bf4 filter=lfs diff=lfs merge=lfs -text
42
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-192/last_900600/ocdbt.process_0/d/af155e1b7577cd01c592121f7b2c3149 filter=lfs diff=lfs merge=lfs -text
43
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-192/last_900600/ocdbt.process_0/d/e0e94acb071c72b5bfc0be095d646349 filter=lfs diff=lfs merge=lfs -text
44
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-768/best_414276/ocdbt.process_0/d/8a9fae9a5dbb43cf532edcfff6449712 filter=lfs diff=lfs merge=lfs -text
45
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-768/best_414276/ocdbt.process_0/d/c8b1faad25fdc5fb962ae6a23e04ccc2 filter=lfs diff=lfs merge=lfs -text
46
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-768/best_414276/ocdbt.process_0/d/f514e4a5dc80a236c87de21c7799ff1d filter=lfs diff=lfs merge=lfs -text
47
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-768/last_900600/ocdbt.process_0/d/10ac345c5e431392da9375af5ac069ab filter=lfs diff=lfs merge=lfs -text
48
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-768/last_900600/ocdbt.process_0/d/248fbe750d504b2d2df32128d01be317 filter=lfs diff=lfs merge=lfs -text
49
+ generalization60/imagenet0.6-lr0.0005-learnable-epochs300-batch256-shared1-768/last_900600/ocdbt.process_0/d/f288621493b1935749c6b084d68b0889 filter=lfs diff=lfs merge=lfs -text
50
+ imagenet/lr0.0005-learnable-epochs300-batch256-shared1-routed0-topk0/best_870696/ocdbt.process_0/d/26c12dc454dbb03bbca099dbb98e02a7 filter=lfs diff=lfs merge=lfs -text
51
+ imagenet/lr0.0005-learnable-epochs300-batch256-shared1-routed0-topk0/best_870696/ocdbt.process_0/d/c79ec19dea96827c08e0cdf6ef0db427 filter=lfs diff=lfs merge=lfs -text
52
+ imagenet/lr0.0005-learnable-epochs300-batch256-shared1-routed0-topk0/best_870696/ocdbt.process_0/d/db6fbbc979b2db0db6e28f1876384985 filter=lfs diff=lfs merge=lfs -text
53
+ imagenet/lr0.0005-learnable-epochs300-batch256-shared1-routed0-topk0/best_870696/ocdbt.process_0/d/e275749ee5f088372ac9475ba8f1d007 filter=lfs diff=lfs merge=lfs -text
54
+ imagenet/lr0.0005-learnable-epochs300-batch256-shared1-routed0-topk0/last_1516212/ocdbt.process_0/d/391dd188d26232546bb24c01b829b6d4 filter=lfs diff=lfs merge=lfs -text
55
+ imagenet/lr0.0005-learnable-epochs300-batch256-shared1-routed0-topk0/last_1516212/ocdbt.process_0/d/3c56b975d43e9b3fbe67ab318e9bb06c filter=lfs diff=lfs merge=lfs -text
56
+ imagenet/lr0.0005-learnable-epochs300-batch256-shared1-routed0-topk0/last_1516212/ocdbt.process_0/d/9ee9aa6faf71b0c10f4999d29ec2bb97 filter=lfs diff=lfs merge=lfs -text
57
+ imagenet/lr0.0005-learnable-epochs300-batch256-shared1-routed0-topk0/last_1516212/ocdbt.process_0/d/b83ad273855fd9350c0df88d1e72fc43 filter=lfs diff=lfs merge=lfs -text
58
+ imagenet/lr0.0005-rope-epochs300-batch256-shared1-routed0-topk0/best_1266012/ocdbt.process_0/d/0c5cecbeaa6d41e897be68690d4b48d6 filter=lfs diff=lfs merge=lfs -text
59
+ imagenet/lr0.0005-rope-epochs300-batch256-shared1-routed0-topk0/best_1266012/ocdbt.process_0/d/15a43f3bb66536aada52183a5f7565ef filter=lfs diff=lfs merge=lfs -text
60
+ imagenet/lr0.0005-rope-epochs300-batch256-shared1-routed0-topk0/best_1266012/ocdbt.process_0/d/2b93456ba33e7869ca3ebc08ab47b923 filter=lfs diff=lfs merge=lfs -text
61
+ imagenet/lr0.0005-rope-epochs300-batch256-shared1-routed0-topk0/best_1266012/ocdbt.process_0/d/86aa3e6ab23a6024c65b3d2d2da4dc99 filter=lfs diff=lfs merge=lfs -text
62
+ imagenet/lr0.0005-rope-epochs300-batch256-shared1-routed0-topk0/last_1456164/ocdbt.process_0/d/47db9796067721e87392a183fef81a2a filter=lfs diff=lfs merge=lfs -text
63
+ imagenet/lr0.0005-rope-epochs300-batch256-shared1-routed0-topk0/last_1456164/ocdbt.process_0/d/625b90664dcfb74332dbf35766adf5df filter=lfs diff=lfs merge=lfs -text
64
+ imagenet/lr0.0005-rope-epochs300-batch256-shared1-routed0-topk0/last_1456164/ocdbt.process_0/d/806ac1e690d758fcceffbcf9a5dc7561 filter=lfs diff=lfs merge=lfs -text
65
+ imagenet/lr0.0005-sinusoidal-epochs300-batch256-shared1-routed0-topk0/best_1451160/ocdbt.process_0/d/91b0d95ba2e59c5d942784b9ce8e383b filter=lfs diff=lfs merge=lfs -text
66
+ imagenet/lr0.0005-sinusoidal-epochs300-batch256-shared1-routed0-topk0/best_1451160/ocdbt.process_0/d/dd17ab8882bc715e52f6cc107420fbda filter=lfs diff=lfs merge=lfs -text
67
+ imagenet/lr0.0005-sinusoidal-epochs300-batch256-shared1-routed0-topk0/best_1451160/ocdbt.process_0/d/f9c000b1e5a85db813e2bd2862be93e2 filter=lfs diff=lfs merge=lfs -text
68
+ imagenet/lr0.0005-sinusoidal-epochs300-batch256-shared1-routed0-topk0/last_1501200/ocdbt.process_0/d/916125a0dd9e3fb8dafd1b0d090c0eac filter=lfs diff=lfs merge=lfs -text
69
+ imagenet/lr0.0005-sinusoidal-epochs300-batch256-shared1-routed0-topk0/last_1501200/ocdbt.process_0/d/bc5c790df29ef36080c40cd03c190014 filter=lfs diff=lfs merge=lfs -text
70
+ imagenet/lr0.0005-sinusoidal-epochs300-batch256-shared1-routed0-topk0/last_1501200/ocdbt.process_0/d/f12d6a45fa12a1ef9479f76088551fde filter=lfs diff=lfs merge=lfs -text
cifar100/middle/learnable-naive/0.0_00seed20/config.json ADDED
@@ -0,0 +1,524 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torchscript": false,
498
+ "typical_p": 1.0,
499
+ "use_bfloat16": false
500
+ },
501
+ "lmc_layer_indices": [
502
+ 0,
503
+ 1,
504
+ 2,
505
+ 3,
506
+ 4,
507
+ 5
508
+ ],
509
+ "model_type": "vit",
510
+ "num_attention_heads": 8,
511
+ "num_channels": 3,
512
+ "num_hidden_layers": 6,
513
+ "num_routed_experts": 0,
514
+ "num_shared_experts": 1,
515
+ "patch_size": 4,
516
+ "pooler_act": "tanh",
517
+ "pooler_output_size": 768,
518
+ "position_embeddings": "learnable",
519
+ "qkv_bias": true,
520
+ "rotary_value": false,
521
+ "routed_scaling_factor": 1.0,
522
+ "topk": 0,
523
+ "transformers_version": "4.56.1"
524
+ }
cifar100/middle/learnable-naive/0.0_00seed20/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e976a532823148c28727664c7143f981a7f21979e05e7d40674e5f2085ded64
3
+ size 76105920
cifar100/middle/learnable-naive/0.125_00seed20/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.125_00seed20/finetune/best_31280/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.125_00seed20/finetune/best_31280/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9572b432a405080d882cf9b8d2c6d659e014b7efffc3185539b42a6f65113eab
3
+ size 76105920
cifar100/middle/learnable-naive/0.125_00seed20/finetune/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.125_00seed20/finetune/last_78200/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.125_00seed20/finetune/last_78200/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9af270ae6b27bdd23bfc4f7fbc2982f335a6f567ca639849bcad6b7c6a361514
3
+ size 76105920
cifar100/middle/learnable-naive/0.125_00seed20/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85cafdae0a3ee61d2be97012ea829e651e9bf7b7c7aeb81786c96db34fe57f20
3
+ size 76105920
cifar100/middle/learnable-naive/0.25_00seed20/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.25_00seed20/finetune/best_25806/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.25_00seed20/finetune/best_25806/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55ab0b75475af1d9d0681063949478aa4c3e0643338afb2ff7be41756f32b3ad
3
+ size 76105920
cifar100/middle/learnable-naive/0.25_00seed20/finetune/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.25_00seed20/finetune/last_78200/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.25_00seed20/finetune/last_78200/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:785fcfeb38a98f3c50531764724738e0ceba5818fea08e2c3e877e509381aa83
3
+ size 76105920
cifar100/middle/learnable-naive/0.25_00seed20/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:abae1a3b3ab680ff26f1d9fa174af25e41aa108e16ef4b15b1ebac7b0a2a82ee
3
+ size 76105920
cifar100/middle/learnable-naive/0.375_00seed20/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.375_00seed20/finetune/best_57868/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.375_00seed20/finetune/best_57868/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:419d26c3e06a2c14cf7b0c7022e4fe6de7cd191653a08df7951e145a1004c6bf
3
+ size 76105920
cifar100/middle/learnable-naive/0.375_00seed20/finetune/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.375_00seed20/finetune/last_78200/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.375_00seed20/finetune/last_78200/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:00752b4ff13c7b76d65fbabbdbb70d24dc3ccf59347881c61e1ad35e55992a66
3
+ size 76105920
cifar100/middle/learnable-naive/0.375_00seed20/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35894f52094d259020e2be8ace84c6d596421bd2a8492c29f785268daff06532
3
+ size 76105920
cifar100/middle/learnable-naive/0.5_00seed20/config.json ADDED
@@ -0,0 +1,524 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torchscript": false,
498
+ "typical_p": 1.0,
499
+ "use_bfloat16": false
500
+ },
501
+ "lmc_layer_indices": [
502
+ 0,
503
+ 1,
504
+ 2,
505
+ 3,
506
+ 4,
507
+ 5
508
+ ],
509
+ "model_type": "vit",
510
+ "num_attention_heads": 8,
511
+ "num_channels": 3,
512
+ "num_hidden_layers": 6,
513
+ "num_routed_experts": 0,
514
+ "num_shared_experts": 1,
515
+ "patch_size": 4,
516
+ "pooler_act": "tanh",
517
+ "pooler_output_size": 768,
518
+ "position_embeddings": "learnable",
519
+ "qkv_bias": true,
520
+ "rotary_value": false,
521
+ "routed_scaling_factor": 1.0,
522
+ "topk": 0,
523
+ "transformers_version": "4.56.1"
524
+ }
cifar100/middle/learnable-naive/0.5_00seed20/finetune/best_40664/config.json ADDED
@@ -0,0 +1,524 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torchscript": false,
498
+ "typical_p": 1.0,
499
+ "use_bfloat16": false
500
+ },
501
+ "lmc_layer_indices": [
502
+ 0,
503
+ 1,
504
+ 2,
505
+ 3,
506
+ 4,
507
+ 5
508
+ ],
509
+ "model_type": "vit",
510
+ "num_attention_heads": 8,
511
+ "num_channels": 3,
512
+ "num_hidden_layers": 6,
513
+ "num_routed_experts": 0,
514
+ "num_shared_experts": 1,
515
+ "patch_size": 4,
516
+ "pooler_act": "tanh",
517
+ "pooler_output_size": 768,
518
+ "position_embeddings": "learnable",
519
+ "qkv_bias": true,
520
+ "rotary_value": false,
521
+ "routed_scaling_factor": 1.0,
522
+ "topk": 0,
523
+ "transformers_version": "4.56.1"
524
+ }
cifar100/middle/learnable-naive/0.5_00seed20/finetune/best_40664/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f2a62f17caaa936669a8c4f12a7f956249828bdd36ab47d5a1afa14a8c4827b
3
+ size 76105920
cifar100/middle/learnable-naive/0.5_00seed20/finetune/config.json ADDED
@@ -0,0 +1,524 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torchscript": false,
498
+ "typical_p": 1.0,
499
+ "use_bfloat16": false
500
+ },
501
+ "lmc_layer_indices": [
502
+ 0,
503
+ 1,
504
+ 2,
505
+ 3,
506
+ 4,
507
+ 5
508
+ ],
509
+ "model_type": "vit",
510
+ "num_attention_heads": 8,
511
+ "num_channels": 3,
512
+ "num_hidden_layers": 6,
513
+ "num_routed_experts": 0,
514
+ "num_shared_experts": 1,
515
+ "patch_size": 4,
516
+ "pooler_act": "tanh",
517
+ "pooler_output_size": 768,
518
+ "position_embeddings": "learnable",
519
+ "qkv_bias": true,
520
+ "rotary_value": false,
521
+ "routed_scaling_factor": 1.0,
522
+ "topk": 0,
523
+ "transformers_version": "4.56.1"
524
+ }
cifar100/middle/learnable-naive/0.5_00seed20/finetune/last_78200/config.json ADDED
@@ -0,0 +1,524 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torchscript": false,
498
+ "typical_p": 1.0,
499
+ "use_bfloat16": false
500
+ },
501
+ "lmc_layer_indices": [
502
+ 0,
503
+ 1,
504
+ 2,
505
+ 3,
506
+ 4,
507
+ 5
508
+ ],
509
+ "model_type": "vit",
510
+ "num_attention_heads": 8,
511
+ "num_channels": 3,
512
+ "num_hidden_layers": 6,
513
+ "num_routed_experts": 0,
514
+ "num_shared_experts": 1,
515
+ "patch_size": 4,
516
+ "pooler_act": "tanh",
517
+ "pooler_output_size": 768,
518
+ "position_embeddings": "learnable",
519
+ "qkv_bias": true,
520
+ "rotary_value": false,
521
+ "routed_scaling_factor": 1.0,
522
+ "topk": 0,
523
+ "transformers_version": "4.56.1"
524
+ }
cifar100/middle/learnable-naive/0.5_00seed20/finetune/last_78200/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:27f7a4cc5022870972f40ea59f4a00f94433e3d99fc6a6e58454a20aa42e927b
3
+ size 76105920
cifar100/middle/learnable-naive/0.5_00seed20/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:476666e756ae33d350f2d0ff475fccf4c06ffb1a28cbc33fc6d2b8c4780f8715
3
+ size 76105920
cifar100/middle/learnable-naive/0.625_00seed20/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.625_00seed20/finetune/best_19550/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.625_00seed20/finetune/best_19550/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6c64d579d1b9ca4902004b3018e348e35a6ffc50f7bb872f0ad73d88e586b3b7
3
+ size 76105920
cifar100/middle/learnable-naive/0.625_00seed20/finetune/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.625_00seed20/finetune/last_78200/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.625_00seed20/finetune/last_78200/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dfe62d5318e6c7c923f777f6734ff79fcc64db05166476870d4ef97b5b935900
3
+ size 76105920
cifar100/middle/learnable-naive/0.625_00seed20/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a0c857a97983b1d98552e734f067f5830136ba9cc33a520ceb414025c9774d7
3
+ size 76105920
cifar100/middle/learnable-naive/0.75_00seed20/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.75_00seed20/finetune/best_34408/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.75_00seed20/finetune/best_34408/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ed2f5e3c535ba4901eeba4831940830856903f3aca14598dfab1caee3aff2ce
3
+ size 76105920
cifar100/middle/learnable-naive/0.75_00seed20/finetune/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.75_00seed20/finetune/last_78200/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.75_00seed20/finetune/last_78200/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cda7d318ca9d4f453dc924b10283bcfc8ce0fe7a789b86e7ff6c0d51577ef9c9
3
+ size 76105920
cifar100/middle/learnable-naive/0.75_00seed20/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47014ba90fe03c27fd91de2c95da26fff4aa6ca1a51ac525a620a86319fcf4fe
3
+ size 76105920
cifar100/middle/learnable-naive/0.875_00seed20/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.875_00seed20/finetune/best_35190/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.875_00seed20/finetune/best_35190/flax_model.msgpack ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef15e764e53a2c8b1f364e2c6c2d027dbe9bd77531884d76857418e738580b28
3
+ size 76105920
cifar100/middle/learnable-naive/0.875_00seed20/finetune/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }
cifar100/middle/learnable-naive/0.875_00seed20/finetune/last_78200/config.json ADDED
@@ -0,0 +1,525 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "laxViTForImageClassification"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.0,
6
+ "encoder_stride": 16,
7
+ "finetune_mlp": true,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 512,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2",
15
+ "3": "LABEL_3",
16
+ "4": "LABEL_4",
17
+ "5": "LABEL_5",
18
+ "6": "LABEL_6",
19
+ "7": "LABEL_7",
20
+ "8": "LABEL_8",
21
+ "9": "LABEL_9",
22
+ "10": "LABEL_10",
23
+ "11": "LABEL_11",
24
+ "12": "LABEL_12",
25
+ "13": "LABEL_13",
26
+ "14": "LABEL_14",
27
+ "15": "LABEL_15",
28
+ "16": "LABEL_16",
29
+ "17": "LABEL_17",
30
+ "18": "LABEL_18",
31
+ "19": "LABEL_19",
32
+ "20": "LABEL_20",
33
+ "21": "LABEL_21",
34
+ "22": "LABEL_22",
35
+ "23": "LABEL_23",
36
+ "24": "LABEL_24",
37
+ "25": "LABEL_25",
38
+ "26": "LABEL_26",
39
+ "27": "LABEL_27",
40
+ "28": "LABEL_28",
41
+ "29": "LABEL_29",
42
+ "30": "LABEL_30",
43
+ "31": "LABEL_31",
44
+ "32": "LABEL_32",
45
+ "33": "LABEL_33",
46
+ "34": "LABEL_34",
47
+ "35": "LABEL_35",
48
+ "36": "LABEL_36",
49
+ "37": "LABEL_37",
50
+ "38": "LABEL_38",
51
+ "39": "LABEL_39",
52
+ "40": "LABEL_40",
53
+ "41": "LABEL_41",
54
+ "42": "LABEL_42",
55
+ "43": "LABEL_43",
56
+ "44": "LABEL_44",
57
+ "45": "LABEL_45",
58
+ "46": "LABEL_46",
59
+ "47": "LABEL_47",
60
+ "48": "LABEL_48",
61
+ "49": "LABEL_49",
62
+ "50": "LABEL_50",
63
+ "51": "LABEL_51",
64
+ "52": "LABEL_52",
65
+ "53": "LABEL_53",
66
+ "54": "LABEL_54",
67
+ "55": "LABEL_55",
68
+ "56": "LABEL_56",
69
+ "57": "LABEL_57",
70
+ "58": "LABEL_58",
71
+ "59": "LABEL_59",
72
+ "60": "LABEL_60",
73
+ "61": "LABEL_61",
74
+ "62": "LABEL_62",
75
+ "63": "LABEL_63",
76
+ "64": "LABEL_64",
77
+ "65": "LABEL_65",
78
+ "66": "LABEL_66",
79
+ "67": "LABEL_67",
80
+ "68": "LABEL_68",
81
+ "69": "LABEL_69",
82
+ "70": "LABEL_70",
83
+ "71": "LABEL_71",
84
+ "72": "LABEL_72",
85
+ "73": "LABEL_73",
86
+ "74": "LABEL_74",
87
+ "75": "LABEL_75",
88
+ "76": "LABEL_76",
89
+ "77": "LABEL_77",
90
+ "78": "LABEL_78",
91
+ "79": "LABEL_79",
92
+ "80": "LABEL_80",
93
+ "81": "LABEL_81",
94
+ "82": "LABEL_82",
95
+ "83": "LABEL_83",
96
+ "84": "LABEL_84",
97
+ "85": "LABEL_85",
98
+ "86": "LABEL_86",
99
+ "87": "LABEL_87",
100
+ "88": "LABEL_88",
101
+ "89": "LABEL_89",
102
+ "90": "LABEL_90",
103
+ "91": "LABEL_91",
104
+ "92": "LABEL_92",
105
+ "93": "LABEL_93",
106
+ "94": "LABEL_94",
107
+ "95": "LABEL_95",
108
+ "96": "LABEL_96",
109
+ "97": "LABEL_97",
110
+ "98": "LABEL_98",
111
+ "99": "LABEL_99"
112
+ },
113
+ "image_size": 32,
114
+ "initializer_range": 0.02,
115
+ "intermediate_size": 2048,
116
+ "label2id": {
117
+ "LABEL_0": 0,
118
+ "LABEL_1": 1,
119
+ "LABEL_10": 10,
120
+ "LABEL_11": 11,
121
+ "LABEL_12": 12,
122
+ "LABEL_13": 13,
123
+ "LABEL_14": 14,
124
+ "LABEL_15": 15,
125
+ "LABEL_16": 16,
126
+ "LABEL_17": 17,
127
+ "LABEL_18": 18,
128
+ "LABEL_19": 19,
129
+ "LABEL_2": 2,
130
+ "LABEL_20": 20,
131
+ "LABEL_21": 21,
132
+ "LABEL_22": 22,
133
+ "LABEL_23": 23,
134
+ "LABEL_24": 24,
135
+ "LABEL_25": 25,
136
+ "LABEL_26": 26,
137
+ "LABEL_27": 27,
138
+ "LABEL_28": 28,
139
+ "LABEL_29": 29,
140
+ "LABEL_3": 3,
141
+ "LABEL_30": 30,
142
+ "LABEL_31": 31,
143
+ "LABEL_32": 32,
144
+ "LABEL_33": 33,
145
+ "LABEL_34": 34,
146
+ "LABEL_35": 35,
147
+ "LABEL_36": 36,
148
+ "LABEL_37": 37,
149
+ "LABEL_38": 38,
150
+ "LABEL_39": 39,
151
+ "LABEL_4": 4,
152
+ "LABEL_40": 40,
153
+ "LABEL_41": 41,
154
+ "LABEL_42": 42,
155
+ "LABEL_43": 43,
156
+ "LABEL_44": 44,
157
+ "LABEL_45": 45,
158
+ "LABEL_46": 46,
159
+ "LABEL_47": 47,
160
+ "LABEL_48": 48,
161
+ "LABEL_49": 49,
162
+ "LABEL_5": 5,
163
+ "LABEL_50": 50,
164
+ "LABEL_51": 51,
165
+ "LABEL_52": 52,
166
+ "LABEL_53": 53,
167
+ "LABEL_54": 54,
168
+ "LABEL_55": 55,
169
+ "LABEL_56": 56,
170
+ "LABEL_57": 57,
171
+ "LABEL_58": 58,
172
+ "LABEL_59": 59,
173
+ "LABEL_6": 6,
174
+ "LABEL_60": 60,
175
+ "LABEL_61": 61,
176
+ "LABEL_62": 62,
177
+ "LABEL_63": 63,
178
+ "LABEL_64": 64,
179
+ "LABEL_65": 65,
180
+ "LABEL_66": 66,
181
+ "LABEL_67": 67,
182
+ "LABEL_68": 68,
183
+ "LABEL_69": 69,
184
+ "LABEL_7": 7,
185
+ "LABEL_70": 70,
186
+ "LABEL_71": 71,
187
+ "LABEL_72": 72,
188
+ "LABEL_73": 73,
189
+ "LABEL_74": 74,
190
+ "LABEL_75": 75,
191
+ "LABEL_76": 76,
192
+ "LABEL_77": 77,
193
+ "LABEL_78": 78,
194
+ "LABEL_79": 79,
195
+ "LABEL_8": 8,
196
+ "LABEL_80": 80,
197
+ "LABEL_81": 81,
198
+ "LABEL_82": 82,
199
+ "LABEL_83": 83,
200
+ "LABEL_84": 84,
201
+ "LABEL_85": 85,
202
+ "LABEL_86": 86,
203
+ "LABEL_87": 87,
204
+ "LABEL_88": 88,
205
+ "LABEL_89": 89,
206
+ "LABEL_9": 9,
207
+ "LABEL_90": 90,
208
+ "LABEL_91": 91,
209
+ "LABEL_92": 92,
210
+ "LABEL_93": 93,
211
+ "LABEL_94": 94,
212
+ "LABEL_95": 95,
213
+ "LABEL_96": 96,
214
+ "LABEL_97": 97,
215
+ "LABEL_98": 98,
216
+ "LABEL_99": 99
217
+ },
218
+ "layer_norm_eps": 1e-12,
219
+ "lmc_config": {
220
+ "_name_or_path": "",
221
+ "add_cross_attention": false,
222
+ "architectures": null,
223
+ "attention_probs_dropout_prob": 0.0,
224
+ "bad_words_ids": null,
225
+ "begin_suppress_tokens": null,
226
+ "bos_token_id": null,
227
+ "chunk_size_feed_forward": 0,
228
+ "cross_attention_hidden_size": null,
229
+ "decoder_start_token_id": null,
230
+ "diversity_penalty": 0.0,
231
+ "do_sample": false,
232
+ "dtype": null,
233
+ "early_stopping": false,
234
+ "encoder_no_repeat_ngram_size": 0,
235
+ "encoder_stride": 16,
236
+ "eos_token_id": null,
237
+ "exponential_decay_length_penalty": null,
238
+ "finetuning_task": null,
239
+ "forced_bos_token_id": null,
240
+ "forced_eos_token_id": null,
241
+ "hidden_act": "gelu",
242
+ "hidden_dropout_prob": 0.0,
243
+ "hidden_size": 512,
244
+ "id2label": {
245
+ "0": "LABEL_0",
246
+ "1": "LABEL_1",
247
+ "2": "LABEL_2",
248
+ "3": "LABEL_3",
249
+ "4": "LABEL_4",
250
+ "5": "LABEL_5",
251
+ "6": "LABEL_6",
252
+ "7": "LABEL_7",
253
+ "8": "LABEL_8",
254
+ "9": "LABEL_9",
255
+ "10": "LABEL_10",
256
+ "11": "LABEL_11",
257
+ "12": "LABEL_12",
258
+ "13": "LABEL_13",
259
+ "14": "LABEL_14",
260
+ "15": "LABEL_15",
261
+ "16": "LABEL_16",
262
+ "17": "LABEL_17",
263
+ "18": "LABEL_18",
264
+ "19": "LABEL_19",
265
+ "20": "LABEL_20",
266
+ "21": "LABEL_21",
267
+ "22": "LABEL_22",
268
+ "23": "LABEL_23",
269
+ "24": "LABEL_24",
270
+ "25": "LABEL_25",
271
+ "26": "LABEL_26",
272
+ "27": "LABEL_27",
273
+ "28": "LABEL_28",
274
+ "29": "LABEL_29",
275
+ "30": "LABEL_30",
276
+ "31": "LABEL_31",
277
+ "32": "LABEL_32",
278
+ "33": "LABEL_33",
279
+ "34": "LABEL_34",
280
+ "35": "LABEL_35",
281
+ "36": "LABEL_36",
282
+ "37": "LABEL_37",
283
+ "38": "LABEL_38",
284
+ "39": "LABEL_39",
285
+ "40": "LABEL_40",
286
+ "41": "LABEL_41",
287
+ "42": "LABEL_42",
288
+ "43": "LABEL_43",
289
+ "44": "LABEL_44",
290
+ "45": "LABEL_45",
291
+ "46": "LABEL_46",
292
+ "47": "LABEL_47",
293
+ "48": "LABEL_48",
294
+ "49": "LABEL_49",
295
+ "50": "LABEL_50",
296
+ "51": "LABEL_51",
297
+ "52": "LABEL_52",
298
+ "53": "LABEL_53",
299
+ "54": "LABEL_54",
300
+ "55": "LABEL_55",
301
+ "56": "LABEL_56",
302
+ "57": "LABEL_57",
303
+ "58": "LABEL_58",
304
+ "59": "LABEL_59",
305
+ "60": "LABEL_60",
306
+ "61": "LABEL_61",
307
+ "62": "LABEL_62",
308
+ "63": "LABEL_63",
309
+ "64": "LABEL_64",
310
+ "65": "LABEL_65",
311
+ "66": "LABEL_66",
312
+ "67": "LABEL_67",
313
+ "68": "LABEL_68",
314
+ "69": "LABEL_69",
315
+ "70": "LABEL_70",
316
+ "71": "LABEL_71",
317
+ "72": "LABEL_72",
318
+ "73": "LABEL_73",
319
+ "74": "LABEL_74",
320
+ "75": "LABEL_75",
321
+ "76": "LABEL_76",
322
+ "77": "LABEL_77",
323
+ "78": "LABEL_78",
324
+ "79": "LABEL_79",
325
+ "80": "LABEL_80",
326
+ "81": "LABEL_81",
327
+ "82": "LABEL_82",
328
+ "83": "LABEL_83",
329
+ "84": "LABEL_84",
330
+ "85": "LABEL_85",
331
+ "86": "LABEL_86",
332
+ "87": "LABEL_87",
333
+ "88": "LABEL_88",
334
+ "89": "LABEL_89",
335
+ "90": "LABEL_90",
336
+ "91": "LABEL_91",
337
+ "92": "LABEL_92",
338
+ "93": "LABEL_93",
339
+ "94": "LABEL_94",
340
+ "95": "LABEL_95",
341
+ "96": "LABEL_96",
342
+ "97": "LABEL_97",
343
+ "98": "LABEL_98",
344
+ "99": "LABEL_99"
345
+ },
346
+ "image_size": 32,
347
+ "initializer_range": 0.02,
348
+ "intermediate_size": 2048,
349
+ "is_decoder": false,
350
+ "is_encoder_decoder": false,
351
+ "label2id": {
352
+ "LABEL_0": 0,
353
+ "LABEL_1": 1,
354
+ "LABEL_10": 10,
355
+ "LABEL_11": 11,
356
+ "LABEL_12": 12,
357
+ "LABEL_13": 13,
358
+ "LABEL_14": 14,
359
+ "LABEL_15": 15,
360
+ "LABEL_16": 16,
361
+ "LABEL_17": 17,
362
+ "LABEL_18": 18,
363
+ "LABEL_19": 19,
364
+ "LABEL_2": 2,
365
+ "LABEL_20": 20,
366
+ "LABEL_21": 21,
367
+ "LABEL_22": 22,
368
+ "LABEL_23": 23,
369
+ "LABEL_24": 24,
370
+ "LABEL_25": 25,
371
+ "LABEL_26": 26,
372
+ "LABEL_27": 27,
373
+ "LABEL_28": 28,
374
+ "LABEL_29": 29,
375
+ "LABEL_3": 3,
376
+ "LABEL_30": 30,
377
+ "LABEL_31": 31,
378
+ "LABEL_32": 32,
379
+ "LABEL_33": 33,
380
+ "LABEL_34": 34,
381
+ "LABEL_35": 35,
382
+ "LABEL_36": 36,
383
+ "LABEL_37": 37,
384
+ "LABEL_38": 38,
385
+ "LABEL_39": 39,
386
+ "LABEL_4": 4,
387
+ "LABEL_40": 40,
388
+ "LABEL_41": 41,
389
+ "LABEL_42": 42,
390
+ "LABEL_43": 43,
391
+ "LABEL_44": 44,
392
+ "LABEL_45": 45,
393
+ "LABEL_46": 46,
394
+ "LABEL_47": 47,
395
+ "LABEL_48": 48,
396
+ "LABEL_49": 49,
397
+ "LABEL_5": 5,
398
+ "LABEL_50": 50,
399
+ "LABEL_51": 51,
400
+ "LABEL_52": 52,
401
+ "LABEL_53": 53,
402
+ "LABEL_54": 54,
403
+ "LABEL_55": 55,
404
+ "LABEL_56": 56,
405
+ "LABEL_57": 57,
406
+ "LABEL_58": 58,
407
+ "LABEL_59": 59,
408
+ "LABEL_6": 6,
409
+ "LABEL_60": 60,
410
+ "LABEL_61": 61,
411
+ "LABEL_62": 62,
412
+ "LABEL_63": 63,
413
+ "LABEL_64": 64,
414
+ "LABEL_65": 65,
415
+ "LABEL_66": 66,
416
+ "LABEL_67": 67,
417
+ "LABEL_68": 68,
418
+ "LABEL_69": 69,
419
+ "LABEL_7": 7,
420
+ "LABEL_70": 70,
421
+ "LABEL_71": 71,
422
+ "LABEL_72": 72,
423
+ "LABEL_73": 73,
424
+ "LABEL_74": 74,
425
+ "LABEL_75": 75,
426
+ "LABEL_76": 76,
427
+ "LABEL_77": 77,
428
+ "LABEL_78": 78,
429
+ "LABEL_79": 79,
430
+ "LABEL_8": 8,
431
+ "LABEL_80": 80,
432
+ "LABEL_81": 81,
433
+ "LABEL_82": 82,
434
+ "LABEL_83": 83,
435
+ "LABEL_84": 84,
436
+ "LABEL_85": 85,
437
+ "LABEL_86": 86,
438
+ "LABEL_87": 87,
439
+ "LABEL_88": 88,
440
+ "LABEL_89": 89,
441
+ "LABEL_9": 9,
442
+ "LABEL_90": 90,
443
+ "LABEL_91": 91,
444
+ "LABEL_92": 92,
445
+ "LABEL_93": 93,
446
+ "LABEL_94": 94,
447
+ "LABEL_95": 95,
448
+ "LABEL_96": 96,
449
+ "LABEL_97": 97,
450
+ "LABEL_98": 98,
451
+ "LABEL_99": 99
452
+ },
453
+ "layer_norm_eps": 1e-12,
454
+ "length_penalty": 1.0,
455
+ "lmc_layer_indices": [],
456
+ "max_length": 20,
457
+ "min_length": 0,
458
+ "model_type": "vit",
459
+ "no_repeat_ngram_size": 0,
460
+ "num_attention_heads": 8,
461
+ "num_beam_groups": 1,
462
+ "num_beams": 1,
463
+ "num_channels": 3,
464
+ "num_hidden_layers": 6,
465
+ "num_return_sequences": 1,
466
+ "num_routed_experts": 0,
467
+ "num_shared_experts": 1,
468
+ "output_attentions": false,
469
+ "output_hidden_states": false,
470
+ "output_scores": false,
471
+ "pad_token_id": null,
472
+ "patch_size": 4,
473
+ "pooler_act": "tanh",
474
+ "pooler_output_size": 768,
475
+ "position_embeddings": "learnable",
476
+ "prefix": null,
477
+ "problem_type": null,
478
+ "pruned_heads": {},
479
+ "qkv_bias": true,
480
+ "remove_invalid_values": false,
481
+ "repetition_penalty": 1.0,
482
+ "return_dict": true,
483
+ "return_dict_in_generate": false,
484
+ "rotary_value": false,
485
+ "routed_scaling_factor": 1.0,
486
+ "sep_token_id": null,
487
+ "suppress_tokens": null,
488
+ "task_specific_params": null,
489
+ "temperature": 1.0,
490
+ "tf_legacy_loss": false,
491
+ "tie_encoder_decoder": false,
492
+ "tie_word_embeddings": true,
493
+ "tokenizer_class": null,
494
+ "top_k": 50,
495
+ "top_p": 1.0,
496
+ "topk": 0,
497
+ "torch_dtype": null,
498
+ "torchscript": false,
499
+ "typical_p": 1.0,
500
+ "use_bfloat16": false
501
+ },
502
+ "lmc_layer_indices": [
503
+ 0,
504
+ 1,
505
+ 2,
506
+ 3,
507
+ 4,
508
+ 5
509
+ ],
510
+ "model_type": "vit",
511
+ "num_attention_heads": 8,
512
+ "num_channels": 3,
513
+ "num_hidden_layers": 6,
514
+ "num_routed_experts": 0,
515
+ "num_shared_experts": 1,
516
+ "patch_size": 4,
517
+ "pooler_act": "tanh",
518
+ "pooler_output_size": 768,
519
+ "position_embeddings": "learnable",
520
+ "qkv_bias": true,
521
+ "rotary_value": false,
522
+ "routed_scaling_factor": 1.0,
523
+ "topk": 0,
524
+ "transformers_version": "4.55.2"
525
+ }