laszlokiss27 commited on
Commit
074860c
1 Parent(s): 8c792db

doodle-dash

Browse files
README.md ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: other
3
+ base_model: apple/mobilevit-small
4
+ tags:
5
+ - generated_from_trainer
6
+ metrics:
7
+ - accuracy
8
+ model-index:
9
+ - name: results
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ # results
17
+
18
+ This model is a fine-tuned version of [apple/mobilevit-small](https://huggingface.co/apple/mobilevit-small) on an unknown dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 1.3757
21
+ - Accuracy: 0.6597
22
+
23
+ ## Model description
24
+
25
+ More information needed
26
+
27
+ ## Intended uses & limitations
28
+
29
+ More information needed
30
+
31
+ ## Training and evaluation data
32
+
33
+ More information needed
34
+
35
+ ## Training procedure
36
+
37
+ ### Training hyperparameters
38
+
39
+ The following hyperparameters were used during training:
40
+ - learning_rate: 0.0008
41
+ - train_batch_size: 256
42
+ - eval_batch_size: 256
43
+ - seed: 42
44
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
45
+ - lr_scheduler_type: linear
46
+ - num_epochs: 5
47
+ - mixed_precision_training: Native AMP
48
+
49
+ ### Training results
50
+
51
+
52
+
53
+ ### Framework versions
54
+
55
+ - Transformers 4.38.2
56
+ - Pytorch 2.2.1+cu121
57
+ - Datasets 2.18.0
58
+ - Tokenizers 0.15.2
all_results.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "eval_accuracy": 0.6597333333333333,
4
+ "eval_loss": 1.3756999969482422,
5
+ "eval_runtime": 16.2859,
6
+ "eval_samples_per_second": 460.521,
7
+ "eval_steps_per_second": 1.842,
8
+ "total_flos": 8.55731423232e+16,
9
+ "train_loss": 1.5635583357377485,
10
+ "train_runtime": 1391.129,
11
+ "train_samples_per_second": 485.217,
12
+ "train_steps_per_second": 1.898
13
+ }
config.json ADDED
@@ -0,0 +1,741 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "apple/mobilevit-small",
3
+ "architectures": [
4
+ "MobileViTForImageClassification"
5
+ ],
6
+ "aspp_dropout_prob": 0.1,
7
+ "aspp_out_channels": 256,
8
+ "atrous_rates": [
9
+ 6,
10
+ 12,
11
+ 18
12
+ ],
13
+ "attention_probs_dropout_prob": 0.0,
14
+ "classifier_dropout_prob": 0.1,
15
+ "conv_kernel_size": 3,
16
+ "expand_ratio": 4.0,
17
+ "hidden_act": "silu",
18
+ "hidden_dropout_prob": 0.1,
19
+ "hidden_sizes": [
20
+ 144,
21
+ 192,
22
+ 240
23
+ ],
24
+ "id2label": {
25
+ "0": "aircraft carrier",
26
+ "1": "airplane",
27
+ "10": "asparagus",
28
+ "100": "dumbbell",
29
+ "101": "ear",
30
+ "102": "elbow",
31
+ "103": "elephant",
32
+ "104": "envelope",
33
+ "105": "eraser",
34
+ "106": "eye",
35
+ "107": "eyeglasses",
36
+ "108": "face",
37
+ "109": "fan",
38
+ "11": "axe",
39
+ "110": "feather",
40
+ "111": "fence",
41
+ "112": "finger",
42
+ "113": "fire hydrant",
43
+ "114": "fireplace",
44
+ "115": "firetruck",
45
+ "116": "fish",
46
+ "117": "flamingo",
47
+ "118": "flashlight",
48
+ "119": "flip flops",
49
+ "12": "backpack",
50
+ "120": "floor lamp",
51
+ "121": "flower",
52
+ "122": "flying saucer",
53
+ "123": "foot",
54
+ "124": "fork",
55
+ "125": "frog",
56
+ "126": "frying pan",
57
+ "127": "garden hose",
58
+ "128": "garden",
59
+ "129": "giraffe",
60
+ "13": "banana",
61
+ "130": "goatee",
62
+ "131": "golf club",
63
+ "132": "grapes",
64
+ "133": "grass",
65
+ "134": "guitar",
66
+ "135": "hamburger",
67
+ "136": "hammer",
68
+ "137": "hand",
69
+ "138": "harp",
70
+ "139": "hat",
71
+ "14": "bandage",
72
+ "140": "headphones",
73
+ "141": "hedgehog",
74
+ "142": "helicopter",
75
+ "143": "helmet",
76
+ "144": "hexagon",
77
+ "145": "hockey puck",
78
+ "146": "hockey stick",
79
+ "147": "horse",
80
+ "148": "hospital",
81
+ "149": "hot air balloon",
82
+ "15": "barn",
83
+ "150": "hot dog",
84
+ "151": "hot tub",
85
+ "152": "hourglass",
86
+ "153": "house plant",
87
+ "154": "house",
88
+ "155": "hurricane",
89
+ "156": "ice cream",
90
+ "157": "jacket",
91
+ "158": "jail",
92
+ "159": "kangaroo",
93
+ "16": "baseball bat",
94
+ "160": "key",
95
+ "161": "keyboard",
96
+ "162": "knee",
97
+ "163": "knife",
98
+ "164": "ladder",
99
+ "165": "lantern",
100
+ "166": "laptop",
101
+ "167": "leaf",
102
+ "168": "leg",
103
+ "169": "light bulb",
104
+ "17": "baseball",
105
+ "170": "lighter",
106
+ "171": "lighthouse",
107
+ "172": "lightning",
108
+ "173": "line",
109
+ "174": "lion",
110
+ "175": "lipstick",
111
+ "176": "lobster",
112
+ "177": "lollipop",
113
+ "178": "mailbox",
114
+ "179": "map",
115
+ "18": "basket",
116
+ "180": "marker",
117
+ "181": "matches",
118
+ "182": "megaphone",
119
+ "183": "mermaid",
120
+ "184": "microphone",
121
+ "185": "microwave",
122
+ "186": "monkey",
123
+ "187": "moon",
124
+ "188": "mosquito",
125
+ "189": "motorbike",
126
+ "19": "basketball",
127
+ "190": "mountain",
128
+ "191": "mouse",
129
+ "192": "moustache",
130
+ "193": "mouth",
131
+ "194": "mug",
132
+ "195": "mushroom",
133
+ "196": "nail",
134
+ "197": "necklace",
135
+ "198": "nose",
136
+ "199": "ocean",
137
+ "2": "alarm clock",
138
+ "20": "bat",
139
+ "200": "octagon",
140
+ "201": "octopus",
141
+ "202": "onion",
142
+ "203": "oven",
143
+ "204": "owl",
144
+ "205": "paint can",
145
+ "206": "paintbrush",
146
+ "207": "palm tree",
147
+ "208": "panda",
148
+ "209": "pants",
149
+ "21": "bathtub",
150
+ "210": "paper clip",
151
+ "211": "parachute",
152
+ "212": "parrot",
153
+ "213": "passport",
154
+ "214": "peanut",
155
+ "215": "pear",
156
+ "216": "peas",
157
+ "217": "pencil",
158
+ "218": "penguin",
159
+ "219": "piano",
160
+ "22": "beach",
161
+ "220": "pickup truck",
162
+ "221": "picture frame",
163
+ "222": "pig",
164
+ "223": "pillow",
165
+ "224": "pineapple",
166
+ "225": "pizza",
167
+ "226": "pliers",
168
+ "227": "police car",
169
+ "228": "pond",
170
+ "229": "pool",
171
+ "23": "bear",
172
+ "230": "popsicle",
173
+ "231": "postcard",
174
+ "232": "potato",
175
+ "233": "power outlet",
176
+ "234": "purse",
177
+ "235": "rabbit",
178
+ "236": "raccoon",
179
+ "237": "radio",
180
+ "238": "rain",
181
+ "239": "rainbow",
182
+ "24": "beard",
183
+ "240": "rake",
184
+ "241": "remote control",
185
+ "242": "rhinoceros",
186
+ "243": "rifle",
187
+ "244": "river",
188
+ "245": "roller coaster",
189
+ "246": "rollerskates",
190
+ "247": "sailboat",
191
+ "248": "sandwich",
192
+ "249": "saw",
193
+ "25": "bed",
194
+ "250": "saxophone",
195
+ "251": "school bus",
196
+ "252": "scissors",
197
+ "253": "scorpion",
198
+ "254": "screwdriver",
199
+ "255": "sea turtle",
200
+ "256": "see saw",
201
+ "257": "shark",
202
+ "258": "sheep",
203
+ "259": "shoe",
204
+ "26": "bee",
205
+ "260": "shorts",
206
+ "261": "shovel",
207
+ "262": "sink",
208
+ "263": "skateboard",
209
+ "264": "skull",
210
+ "265": "skyscraper",
211
+ "266": "sleeping bag",
212
+ "267": "smiley face",
213
+ "268": "snail",
214
+ "269": "snake",
215
+ "27": "belt",
216
+ "270": "snorkel",
217
+ "271": "snowflake",
218
+ "272": "snowman",
219
+ "273": "soccer ball",
220
+ "274": "sock",
221
+ "275": "speedboat",
222
+ "276": "spider",
223
+ "277": "spoon",
224
+ "278": "spreadsheet",
225
+ "279": "square",
226
+ "28": "bench",
227
+ "280": "squiggle",
228
+ "281": "squirrel",
229
+ "282": "stairs",
230
+ "283": "star",
231
+ "284": "steak",
232
+ "285": "stereo",
233
+ "286": "stethoscope",
234
+ "287": "stitches",
235
+ "288": "stop sign",
236
+ "289": "stove",
237
+ "29": "bicycle",
238
+ "290": "strawberry",
239
+ "291": "streetlight",
240
+ "292": "string bean",
241
+ "293": "submarine",
242
+ "294": "suitcase",
243
+ "295": "sun",
244
+ "296": "swan",
245
+ "297": "sweater",
246
+ "298": "swing set",
247
+ "299": "sword",
248
+ "3": "ambulance",
249
+ "30": "binoculars",
250
+ "300": "syringe",
251
+ "301": "t-shirt",
252
+ "302": "table",
253
+ "303": "teapot",
254
+ "304": "teddy-bear",
255
+ "305": "telephone",
256
+ "306": "television",
257
+ "307": "tennis racquet",
258
+ "308": "tent",
259
+ "309": "The Eiffel Tower",
260
+ "31": "bird",
261
+ "310": "The Great Wall of China",
262
+ "311": "The Mona Lisa",
263
+ "312": "tiger",
264
+ "313": "toaster",
265
+ "314": "toe",
266
+ "315": "toilet",
267
+ "316": "tooth",
268
+ "317": "toothbrush",
269
+ "318": "toothpaste",
270
+ "319": "tornado",
271
+ "32": "birthday cake",
272
+ "320": "tractor",
273
+ "321": "traffic light",
274
+ "322": "train",
275
+ "323": "tree",
276
+ "324": "triangle",
277
+ "325": "trombone",
278
+ "326": "truck",
279
+ "327": "trumpet",
280
+ "328": "umbrella",
281
+ "329": "underwear",
282
+ "33": "blackberry",
283
+ "330": "van",
284
+ "331": "vase",
285
+ "332": "violin",
286
+ "333": "washing machine",
287
+ "334": "watermelon",
288
+ "335": "waterslide",
289
+ "336": "whale",
290
+ "337": "wheel",
291
+ "338": "windmill",
292
+ "339": "wine bottle",
293
+ "34": "blueberry",
294
+ "340": "wine glass",
295
+ "341": "wristwatch",
296
+ "342": "yoga",
297
+ "343": "zebra",
298
+ "344": "zigzag",
299
+ "35": "book",
300
+ "36": "boomerang",
301
+ "37": "bottlecap",
302
+ "38": "bowtie",
303
+ "39": "bracelet",
304
+ "4": "angel",
305
+ "40": "brain",
306
+ "41": "bread",
307
+ "42": "bridge",
308
+ "43": "broccoli",
309
+ "44": "broom",
310
+ "45": "bucket",
311
+ "46": "bulldozer",
312
+ "47": "bus",
313
+ "48": "bush",
314
+ "49": "butterfly",
315
+ "5": "animal migration",
316
+ "50": "cactus",
317
+ "51": "cake",
318
+ "52": "calculator",
319
+ "53": "calendar",
320
+ "54": "camel",
321
+ "55": "camera",
322
+ "56": "camouflage",
323
+ "57": "campfire",
324
+ "58": "candle",
325
+ "59": "cannon",
326
+ "6": "ant",
327
+ "60": "canoe",
328
+ "61": "car",
329
+ "62": "carrot",
330
+ "63": "castle",
331
+ "64": "cat",
332
+ "65": "ceiling fan",
333
+ "66": "cell phone",
334
+ "67": "cello",
335
+ "68": "chair",
336
+ "69": "chandelier",
337
+ "7": "anvil",
338
+ "70": "church",
339
+ "71": "circle",
340
+ "72": "clarinet",
341
+ "73": "clock",
342
+ "74": "cloud",
343
+ "75": "coffee cup",
344
+ "76": "compass",
345
+ "77": "computer",
346
+ "78": "cookie",
347
+ "79": "cooler",
348
+ "8": "apple",
349
+ "80": "couch",
350
+ "81": "cow",
351
+ "82": "crab",
352
+ "83": "crayon",
353
+ "84": "crocodile",
354
+ "85": "crown",
355
+ "86": "cruise ship",
356
+ "87": "cup",
357
+ "88": "diamond",
358
+ "89": "dishwasher",
359
+ "9": "arm",
360
+ "90": "diving board",
361
+ "91": "dog",
362
+ "92": "dolphin",
363
+ "93": "donut",
364
+ "94": "door",
365
+ "95": "dragon",
366
+ "96": "dresser",
367
+ "97": "drill",
368
+ "98": "drums",
369
+ "99": "duck"
370
+ },
371
+ "image_size": 64,
372
+ "initializer_range": 0.02,
373
+ "label2id": {
374
+ "The Eiffel Tower": "309",
375
+ "The Great Wall of China": "310",
376
+ "The Mona Lisa": "311",
377
+ "aircraft carrier": "0",
378
+ "airplane": "1",
379
+ "alarm clock": "2",
380
+ "ambulance": "3",
381
+ "angel": "4",
382
+ "animal migration": "5",
383
+ "ant": "6",
384
+ "anvil": "7",
385
+ "apple": "8",
386
+ "arm": "9",
387
+ "asparagus": "10",
388
+ "axe": "11",
389
+ "backpack": "12",
390
+ "banana": "13",
391
+ "bandage": "14",
392
+ "barn": "15",
393
+ "baseball": "17",
394
+ "baseball bat": "16",
395
+ "basket": "18",
396
+ "basketball": "19",
397
+ "bat": "20",
398
+ "bathtub": "21",
399
+ "beach": "22",
400
+ "bear": "23",
401
+ "beard": "24",
402
+ "bed": "25",
403
+ "bee": "26",
404
+ "belt": "27",
405
+ "bench": "28",
406
+ "bicycle": "29",
407
+ "binoculars": "30",
408
+ "bird": "31",
409
+ "birthday cake": "32",
410
+ "blackberry": "33",
411
+ "blueberry": "34",
412
+ "book": "35",
413
+ "boomerang": "36",
414
+ "bottlecap": "37",
415
+ "bowtie": "38",
416
+ "bracelet": "39",
417
+ "brain": "40",
418
+ "bread": "41",
419
+ "bridge": "42",
420
+ "broccoli": "43",
421
+ "broom": "44",
422
+ "bucket": "45",
423
+ "bulldozer": "46",
424
+ "bus": "47",
425
+ "bush": "48",
426
+ "butterfly": "49",
427
+ "cactus": "50",
428
+ "cake": "51",
429
+ "calculator": "52",
430
+ "calendar": "53",
431
+ "camel": "54",
432
+ "camera": "55",
433
+ "camouflage": "56",
434
+ "campfire": "57",
435
+ "candle": "58",
436
+ "cannon": "59",
437
+ "canoe": "60",
438
+ "car": "61",
439
+ "carrot": "62",
440
+ "castle": "63",
441
+ "cat": "64",
442
+ "ceiling fan": "65",
443
+ "cell phone": "66",
444
+ "cello": "67",
445
+ "chair": "68",
446
+ "chandelier": "69",
447
+ "church": "70",
448
+ "circle": "71",
449
+ "clarinet": "72",
450
+ "clock": "73",
451
+ "cloud": "74",
452
+ "coffee cup": "75",
453
+ "compass": "76",
454
+ "computer": "77",
455
+ "cookie": "78",
456
+ "cooler": "79",
457
+ "couch": "80",
458
+ "cow": "81",
459
+ "crab": "82",
460
+ "crayon": "83",
461
+ "crocodile": "84",
462
+ "crown": "85",
463
+ "cruise ship": "86",
464
+ "cup": "87",
465
+ "diamond": "88",
466
+ "dishwasher": "89",
467
+ "diving board": "90",
468
+ "dog": "91",
469
+ "dolphin": "92",
470
+ "donut": "93",
471
+ "door": "94",
472
+ "dragon": "95",
473
+ "dresser": "96",
474
+ "drill": "97",
475
+ "drums": "98",
476
+ "duck": "99",
477
+ "dumbbell": "100",
478
+ "ear": "101",
479
+ "elbow": "102",
480
+ "elephant": "103",
481
+ "envelope": "104",
482
+ "eraser": "105",
483
+ "eye": "106",
484
+ "eyeglasses": "107",
485
+ "face": "108",
486
+ "fan": "109",
487
+ "feather": "110",
488
+ "fence": "111",
489
+ "finger": "112",
490
+ "fire hydrant": "113",
491
+ "fireplace": "114",
492
+ "firetruck": "115",
493
+ "fish": "116",
494
+ "flamingo": "117",
495
+ "flashlight": "118",
496
+ "flip flops": "119",
497
+ "floor lamp": "120",
498
+ "flower": "121",
499
+ "flying saucer": "122",
500
+ "foot": "123",
501
+ "fork": "124",
502
+ "frog": "125",
503
+ "frying pan": "126",
504
+ "garden": "128",
505
+ "garden hose": "127",
506
+ "giraffe": "129",
507
+ "goatee": "130",
508
+ "golf club": "131",
509
+ "grapes": "132",
510
+ "grass": "133",
511
+ "guitar": "134",
512
+ "hamburger": "135",
513
+ "hammer": "136",
514
+ "hand": "137",
515
+ "harp": "138",
516
+ "hat": "139",
517
+ "headphones": "140",
518
+ "hedgehog": "141",
519
+ "helicopter": "142",
520
+ "helmet": "143",
521
+ "hexagon": "144",
522
+ "hockey puck": "145",
523
+ "hockey stick": "146",
524
+ "horse": "147",
525
+ "hospital": "148",
526
+ "hot air balloon": "149",
527
+ "hot dog": "150",
528
+ "hot tub": "151",
529
+ "hourglass": "152",
530
+ "house": "154",
531
+ "house plant": "153",
532
+ "hurricane": "155",
533
+ "ice cream": "156",
534
+ "jacket": "157",
535
+ "jail": "158",
536
+ "kangaroo": "159",
537
+ "key": "160",
538
+ "keyboard": "161",
539
+ "knee": "162",
540
+ "knife": "163",
541
+ "ladder": "164",
542
+ "lantern": "165",
543
+ "laptop": "166",
544
+ "leaf": "167",
545
+ "leg": "168",
546
+ "light bulb": "169",
547
+ "lighter": "170",
548
+ "lighthouse": "171",
549
+ "lightning": "172",
550
+ "line": "173",
551
+ "lion": "174",
552
+ "lipstick": "175",
553
+ "lobster": "176",
554
+ "lollipop": "177",
555
+ "mailbox": "178",
556
+ "map": "179",
557
+ "marker": "180",
558
+ "matches": "181",
559
+ "megaphone": "182",
560
+ "mermaid": "183",
561
+ "microphone": "184",
562
+ "microwave": "185",
563
+ "monkey": "186",
564
+ "moon": "187",
565
+ "mosquito": "188",
566
+ "motorbike": "189",
567
+ "mountain": "190",
568
+ "mouse": "191",
569
+ "moustache": "192",
570
+ "mouth": "193",
571
+ "mug": "194",
572
+ "mushroom": "195",
573
+ "nail": "196",
574
+ "necklace": "197",
575
+ "nose": "198",
576
+ "ocean": "199",
577
+ "octagon": "200",
578
+ "octopus": "201",
579
+ "onion": "202",
580
+ "oven": "203",
581
+ "owl": "204",
582
+ "paint can": "205",
583
+ "paintbrush": "206",
584
+ "palm tree": "207",
585
+ "panda": "208",
586
+ "pants": "209",
587
+ "paper clip": "210",
588
+ "parachute": "211",
589
+ "parrot": "212",
590
+ "passport": "213",
591
+ "peanut": "214",
592
+ "pear": "215",
593
+ "peas": "216",
594
+ "pencil": "217",
595
+ "penguin": "218",
596
+ "piano": "219",
597
+ "pickup truck": "220",
598
+ "picture frame": "221",
599
+ "pig": "222",
600
+ "pillow": "223",
601
+ "pineapple": "224",
602
+ "pizza": "225",
603
+ "pliers": "226",
604
+ "police car": "227",
605
+ "pond": "228",
606
+ "pool": "229",
607
+ "popsicle": "230",
608
+ "postcard": "231",
609
+ "potato": "232",
610
+ "power outlet": "233",
611
+ "purse": "234",
612
+ "rabbit": "235",
613
+ "raccoon": "236",
614
+ "radio": "237",
615
+ "rain": "238",
616
+ "rainbow": "239",
617
+ "rake": "240",
618
+ "remote control": "241",
619
+ "rhinoceros": "242",
620
+ "rifle": "243",
621
+ "river": "244",
622
+ "roller coaster": "245",
623
+ "rollerskates": "246",
624
+ "sailboat": "247",
625
+ "sandwich": "248",
626
+ "saw": "249",
627
+ "saxophone": "250",
628
+ "school bus": "251",
629
+ "scissors": "252",
630
+ "scorpion": "253",
631
+ "screwdriver": "254",
632
+ "sea turtle": "255",
633
+ "see saw": "256",
634
+ "shark": "257",
635
+ "sheep": "258",
636
+ "shoe": "259",
637
+ "shorts": "260",
638
+ "shovel": "261",
639
+ "sink": "262",
640
+ "skateboard": "263",
641
+ "skull": "264",
642
+ "skyscraper": "265",
643
+ "sleeping bag": "266",
644
+ "smiley face": "267",
645
+ "snail": "268",
646
+ "snake": "269",
647
+ "snorkel": "270",
648
+ "snowflake": "271",
649
+ "snowman": "272",
650
+ "soccer ball": "273",
651
+ "sock": "274",
652
+ "speedboat": "275",
653
+ "spider": "276",
654
+ "spoon": "277",
655
+ "spreadsheet": "278",
656
+ "square": "279",
657
+ "squiggle": "280",
658
+ "squirrel": "281",
659
+ "stairs": "282",
660
+ "star": "283",
661
+ "steak": "284",
662
+ "stereo": "285",
663
+ "stethoscope": "286",
664
+ "stitches": "287",
665
+ "stop sign": "288",
666
+ "stove": "289",
667
+ "strawberry": "290",
668
+ "streetlight": "291",
669
+ "string bean": "292",
670
+ "submarine": "293",
671
+ "suitcase": "294",
672
+ "sun": "295",
673
+ "swan": "296",
674
+ "sweater": "297",
675
+ "swing set": "298",
676
+ "sword": "299",
677
+ "syringe": "300",
678
+ "t-shirt": "301",
679
+ "table": "302",
680
+ "teapot": "303",
681
+ "teddy-bear": "304",
682
+ "telephone": "305",
683
+ "television": "306",
684
+ "tennis racquet": "307",
685
+ "tent": "308",
686
+ "tiger": "312",
687
+ "toaster": "313",
688
+ "toe": "314",
689
+ "toilet": "315",
690
+ "tooth": "316",
691
+ "toothbrush": "317",
692
+ "toothpaste": "318",
693
+ "tornado": "319",
694
+ "tractor": "320",
695
+ "traffic light": "321",
696
+ "train": "322",
697
+ "tree": "323",
698
+ "triangle": "324",
699
+ "trombone": "325",
700
+ "truck": "326",
701
+ "trumpet": "327",
702
+ "umbrella": "328",
703
+ "underwear": "329",
704
+ "van": "330",
705
+ "vase": "331",
706
+ "violin": "332",
707
+ "washing machine": "333",
708
+ "watermelon": "334",
709
+ "waterslide": "335",
710
+ "whale": "336",
711
+ "wheel": "337",
712
+ "windmill": "338",
713
+ "wine bottle": "339",
714
+ "wine glass": "340",
715
+ "wristwatch": "341",
716
+ "yoga": "342",
717
+ "zebra": "343",
718
+ "zigzag": "344"
719
+ },
720
+ "layer_norm_eps": 1e-05,
721
+ "mlp_ratio": 2.0,
722
+ "model_type": "mobilevit",
723
+ "neck_hidden_sizes": [
724
+ 16,
725
+ 32,
726
+ 64,
727
+ 96,
728
+ 128,
729
+ 160,
730
+ 640
731
+ ],
732
+ "num_attention_heads": 4,
733
+ "num_channels": 1,
734
+ "output_stride": 32,
735
+ "patch_size": 2,
736
+ "problem_type": "single_label_classification",
737
+ "qkv_bias": true,
738
+ "semantic_loss_ignore_index": 255,
739
+ "torch_dtype": "float32",
740
+ "transformers_version": "4.38.2"
741
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4594fe1cdfb2d8f9baf4a3c32c01066461f5b9e25044f0208a0a12061d6d2c49
3
+ size 20730036
preprocessor_config.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": {
3
+ "height": 64,
4
+ "width": 64
5
+ },
6
+ "do_center_crop": true,
7
+ "do_convert_rgb": false,
8
+ "do_flip_channel_order": false,
9
+ "do_rescale": true,
10
+ "do_resize": true,
11
+ "image_processor_type": "MobileViTImageProcessor",
12
+ "resample": 2,
13
+ "rescale_factor": 0.00392156862745098,
14
+ "size": {
15
+ "shortest_edge": 64
16
+ }
17
+ }
test_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "eval_accuracy": 0.6597333333333333,
4
+ "eval_loss": 1.3756999969482422,
5
+ "eval_runtime": 16.2859,
6
+ "eval_samples_per_second": 460.521,
7
+ "eval_steps_per_second": 1.842
8
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "total_flos": 8.55731423232e+16,
4
+ "train_loss": 1.5635583357377485,
5
+ "train_runtime": 1391.129,
6
+ "train_samples_per_second": 485.217,
7
+ "train_steps_per_second": 1.898
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 5.0,
5
+ "eval_steps": 5000,
6
+ "global_step": 2640,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 1.89,
13
+ "grad_norm": 1.7177796363830566,
14
+ "learning_rate": 0.0004978787878787879,
15
+ "loss": 2.2138,
16
+ "step": 1000
17
+ },
18
+ {
19
+ "epoch": 3.79,
20
+ "grad_norm": 2.4473989009857178,
21
+ "learning_rate": 0.00019484848484848486,
22
+ "loss": 1.2577,
23
+ "step": 2000
24
+ },
25
+ {
26
+ "epoch": 5.0,
27
+ "step": 2640,
28
+ "total_flos": 8.55731423232e+16,
29
+ "train_loss": 1.5635583357377485,
30
+ "train_runtime": 1391.129,
31
+ "train_samples_per_second": 485.217,
32
+ "train_steps_per_second": 1.898
33
+ }
34
+ ],
35
+ "logging_steps": 1000,
36
+ "max_steps": 2640,
37
+ "num_input_tokens_seen": 0,
38
+ "num_train_epochs": 5,
39
+ "save_steps": 5000,
40
+ "total_flos": 8.55731423232e+16,
41
+ "train_batch_size": 256,
42
+ "trial_name": null,
43
+ "trial_params": null
44
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:57eb28f43f42cafdd51441529e7f5eee73bdb0c9dd9f65304199c2aea12458a4
3
+ size 4856