Upload folder using huggingface_hub
Browse files- checkpoint-28700/config.json +168 -0
- checkpoint-28700/model.safetensors +3 -0
- checkpoint-28700/optimizer.pt +3 -0
- checkpoint-28700/preprocessor_config.json +22 -0
- checkpoint-28700/rng_state.pth +3 -0
- checkpoint-28700/scheduler.pt +3 -0
- checkpoint-28700/trainer_state.json +1432 -0
- checkpoint-28700/training_args.bin +3 -0
- config.json +134 -136
- model.safetensors +2 -2
checkpoint-28700/config.json
ADDED
|
@@ -0,0 +1,168 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-base-patch16-224-in21k",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"ViTForImageClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"encoder_stride": 16,
|
| 8 |
+
"hidden_act": "gelu",
|
| 9 |
+
"hidden_dropout_prob": 0.0,
|
| 10 |
+
"hidden_size": 768,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "Afghan",
|
| 13 |
+
"1": "African Wild Dog",
|
| 14 |
+
"2": "Airedale",
|
| 15 |
+
"3": "American Hairless",
|
| 16 |
+
"4": "American Spaniel",
|
| 17 |
+
"5": "Basenji",
|
| 18 |
+
"6": "Basset",
|
| 19 |
+
"7": "Beagle",
|
| 20 |
+
"8": "Bearded Collie",
|
| 21 |
+
"9": "Bermaise",
|
| 22 |
+
"10": "Bichon Frise",
|
| 23 |
+
"11": "Blenheim",
|
| 24 |
+
"12": "Bloodhound",
|
| 25 |
+
"13": "Bluetick",
|
| 26 |
+
"14": "Border Collie",
|
| 27 |
+
"15": "Borzoi",
|
| 28 |
+
"16": "Boston Terrier",
|
| 29 |
+
"17": "Boxer",
|
| 30 |
+
"18": "Bull Mastiff",
|
| 31 |
+
"19": "Bull Terrier",
|
| 32 |
+
"20": "Bulldog",
|
| 33 |
+
"21": "Cairn",
|
| 34 |
+
"22": "Chihuahua",
|
| 35 |
+
"23": "Chinese Crested",
|
| 36 |
+
"24": "Chow",
|
| 37 |
+
"25": "Clumber",
|
| 38 |
+
"26": "Cockapoo",
|
| 39 |
+
"27": "Cocker",
|
| 40 |
+
"28": "Collie",
|
| 41 |
+
"29": "Corgi",
|
| 42 |
+
"30": "Coyote",
|
| 43 |
+
"31": "Dalmation",
|
| 44 |
+
"32": "Dhole",
|
| 45 |
+
"33": "Dingo",
|
| 46 |
+
"34": "Doberman",
|
| 47 |
+
"35": "Elk Hound",
|
| 48 |
+
"36": "French Bulldog",
|
| 49 |
+
"37": "German Sheperd",
|
| 50 |
+
"38": "Golden Retriever",
|
| 51 |
+
"39": "Great Dane",
|
| 52 |
+
"40": "Great Perenees",
|
| 53 |
+
"41": "Greyhound",
|
| 54 |
+
"42": "Groenendael",
|
| 55 |
+
"43": "Irish Spaniel",
|
| 56 |
+
"44": "Irish Wolfhound",
|
| 57 |
+
"45": "Japanese Spaniel",
|
| 58 |
+
"46": "Komondor",
|
| 59 |
+
"47": "Labradoodle",
|
| 60 |
+
"48": "Labrador",
|
| 61 |
+
"49": "Lhasa",
|
| 62 |
+
"50": "Malinois",
|
| 63 |
+
"51": "Maltese",
|
| 64 |
+
"52": "Mex Hairless",
|
| 65 |
+
"53": "Newfoundland",
|
| 66 |
+
"54": "Pekinese",
|
| 67 |
+
"55": "Pit Bull",
|
| 68 |
+
"56": "Pomeranian",
|
| 69 |
+
"57": "Poodle",
|
| 70 |
+
"58": "Pug",
|
| 71 |
+
"59": "Rhodesian",
|
| 72 |
+
"60": "Rottweiler",
|
| 73 |
+
"61": "Saint Bernard",
|
| 74 |
+
"62": "Schnauzer",
|
| 75 |
+
"63": "Scotch Terrier",
|
| 76 |
+
"64": "Shar_Pei",
|
| 77 |
+
"65": "Shiba Inu",
|
| 78 |
+
"66": "Shih-Tzu",
|
| 79 |
+
"67": "Siberian Husky",
|
| 80 |
+
"68": "Vizsla",
|
| 81 |
+
"69": "Yorkie"
|
| 82 |
+
},
|
| 83 |
+
"image_size": 224,
|
| 84 |
+
"initializer_range": 0.02,
|
| 85 |
+
"intermediate_size": 3072,
|
| 86 |
+
"label2id": {
|
| 87 |
+
"Afghan": 0,
|
| 88 |
+
"African Wild Dog": 1,
|
| 89 |
+
"Airedale": 2,
|
| 90 |
+
"American Hairless": 3,
|
| 91 |
+
"American Spaniel": 4,
|
| 92 |
+
"Basenji": 5,
|
| 93 |
+
"Basset": 6,
|
| 94 |
+
"Beagle": 7,
|
| 95 |
+
"Bearded Collie": 8,
|
| 96 |
+
"Bermaise": 9,
|
| 97 |
+
"Bichon Frise": 10,
|
| 98 |
+
"Blenheim": 11,
|
| 99 |
+
"Bloodhound": 12,
|
| 100 |
+
"Bluetick": 13,
|
| 101 |
+
"Border Collie": 14,
|
| 102 |
+
"Borzoi": 15,
|
| 103 |
+
"Boston Terrier": 16,
|
| 104 |
+
"Boxer": 17,
|
| 105 |
+
"Bull Mastiff": 18,
|
| 106 |
+
"Bull Terrier": 19,
|
| 107 |
+
"Bulldog": 20,
|
| 108 |
+
"Cairn": 21,
|
| 109 |
+
"Chihuahua": 22,
|
| 110 |
+
"Chinese Crested": 23,
|
| 111 |
+
"Chow": 24,
|
| 112 |
+
"Clumber": 25,
|
| 113 |
+
"Cockapoo": 26,
|
| 114 |
+
"Cocker": 27,
|
| 115 |
+
"Collie": 28,
|
| 116 |
+
"Corgi": 29,
|
| 117 |
+
"Coyote": 30,
|
| 118 |
+
"Dalmation": 31,
|
| 119 |
+
"Dhole": 32,
|
| 120 |
+
"Dingo": 33,
|
| 121 |
+
"Doberman": 34,
|
| 122 |
+
"Elk Hound": 35,
|
| 123 |
+
"French Bulldog": 36,
|
| 124 |
+
"German Sheperd": 37,
|
| 125 |
+
"Golden Retriever": 38,
|
| 126 |
+
"Great Dane": 39,
|
| 127 |
+
"Great Perenees": 40,
|
| 128 |
+
"Greyhound": 41,
|
| 129 |
+
"Groenendael": 42,
|
| 130 |
+
"Irish Spaniel": 43,
|
| 131 |
+
"Irish Wolfhound": 44,
|
| 132 |
+
"Japanese Spaniel": 45,
|
| 133 |
+
"Komondor": 46,
|
| 134 |
+
"Labradoodle": 47,
|
| 135 |
+
"Labrador": 48,
|
| 136 |
+
"Lhasa": 49,
|
| 137 |
+
"Malinois": 50,
|
| 138 |
+
"Maltese": 51,
|
| 139 |
+
"Mex Hairless": 52,
|
| 140 |
+
"Newfoundland": 53,
|
| 141 |
+
"Pekinese": 54,
|
| 142 |
+
"Pit Bull": 55,
|
| 143 |
+
"Pomeranian": 56,
|
| 144 |
+
"Poodle": 57,
|
| 145 |
+
"Pug": 58,
|
| 146 |
+
"Rhodesian": 59,
|
| 147 |
+
"Rottweiler": 60,
|
| 148 |
+
"Saint Bernard": 61,
|
| 149 |
+
"Schnauzer": 62,
|
| 150 |
+
"Scotch Terrier": 63,
|
| 151 |
+
"Shar_Pei": 64,
|
| 152 |
+
"Shiba Inu": 65,
|
| 153 |
+
"Shih-Tzu": 66,
|
| 154 |
+
"Siberian Husky": 67,
|
| 155 |
+
"Vizsla": 68,
|
| 156 |
+
"Yorkie": 69
|
| 157 |
+
},
|
| 158 |
+
"layer_norm_eps": 1e-12,
|
| 159 |
+
"model_type": "vit",
|
| 160 |
+
"num_attention_heads": 12,
|
| 161 |
+
"num_channels": 3,
|
| 162 |
+
"num_hidden_layers": 12,
|
| 163 |
+
"patch_size": 16,
|
| 164 |
+
"problem_type": "single_label_classification",
|
| 165 |
+
"qkv_bias": true,
|
| 166 |
+
"torch_dtype": "float32",
|
| 167 |
+
"transformers_version": "4.45.2"
|
| 168 |
+
}
|
checkpoint-28700/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fa6f5b1f9c44c3c29d9c4c7c230929790148180797e2a7c74658d6a0569b7889
|
| 3 |
+
size 343433152
|
checkpoint-28700/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9441119e17fbce8c02c123f0a6bc137190b493c4a02ea94d0eb91c98ce0d197f
|
| 3 |
+
size 686986757
|
checkpoint-28700/preprocessor_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_rescale": true,
|
| 4 |
+
"do_resize": true,
|
| 5 |
+
"image_mean": [
|
| 6 |
+
0.5,
|
| 7 |
+
0.5,
|
| 8 |
+
0.5
|
| 9 |
+
],
|
| 10 |
+
"image_processor_type": "ViTImageProcessor",
|
| 11 |
+
"image_std": [
|
| 12 |
+
0.5,
|
| 13 |
+
0.5,
|
| 14 |
+
0.5
|
| 15 |
+
],
|
| 16 |
+
"resample": 2,
|
| 17 |
+
"rescale_factor": 0.00392156862745098,
|
| 18 |
+
"size": {
|
| 19 |
+
"height": 224,
|
| 20 |
+
"width": 224
|
| 21 |
+
}
|
| 22 |
+
}
|
checkpoint-28700/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d4e46c392585b57f3d919f401bc84426e9917e75fa9450d0b2300bd39adcd700
|
| 3 |
+
size 14575
|
checkpoint-28700/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b5f5db57e2e71196f015b9687e3a2120516fab7ddf9b02d267209859f3b74fef
|
| 3 |
+
size 627
|
checkpoint-28700/trainer_state.json
ADDED
|
@@ -0,0 +1,1432 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 2.204108953475952,
|
| 3 |
+
"best_model_checkpoint": "dogs_70_breeds_image_detection/checkpoint-28700",
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 28700,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_accuracy": 0.02490170380078637,
|
| 14 |
+
"eval_loss": 4.22898006439209,
|
| 15 |
+
"eval_model_preparation_time": 0.0056,
|
| 16 |
+
"eval_runtime": 53.0352,
|
| 17 |
+
"eval_samples_per_second": 115.093,
|
| 18 |
+
"eval_steps_per_second": 14.387,
|
| 19 |
+
"step": 287
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 1.7421602787456445,
|
| 23 |
+
"grad_norm": 1.2073975801467896,
|
| 24 |
+
"learning_rate": 9.842931937172775e-07,
|
| 25 |
+
"loss": 4.2284,
|
| 26 |
+
"step": 500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.0,
|
| 30 |
+
"eval_accuracy": 0.048820445609436436,
|
| 31 |
+
"eval_loss": 4.192336559295654,
|
| 32 |
+
"eval_model_preparation_time": 0.0056,
|
| 33 |
+
"eval_runtime": 51.9294,
|
| 34 |
+
"eval_samples_per_second": 117.544,
|
| 35 |
+
"eval_steps_per_second": 14.693,
|
| 36 |
+
"step": 574
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 3.0,
|
| 40 |
+
"eval_accuracy": 0.09518348623853211,
|
| 41 |
+
"eval_loss": 4.150681972503662,
|
| 42 |
+
"eval_model_preparation_time": 0.0056,
|
| 43 |
+
"eval_runtime": 51.3038,
|
| 44 |
+
"eval_samples_per_second": 118.977,
|
| 45 |
+
"eval_steps_per_second": 14.872,
|
| 46 |
+
"step": 861
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 3.484320557491289,
|
| 50 |
+
"grad_norm": 1.3359559774398804,
|
| 51 |
+
"learning_rate": 9.668411867364747e-07,
|
| 52 |
+
"loss": 4.1524,
|
| 53 |
+
"step": 1000
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 4.0,
|
| 57 |
+
"eval_accuracy": 0.18299475753604194,
|
| 58 |
+
"eval_loss": 4.1059417724609375,
|
| 59 |
+
"eval_model_preparation_time": 0.0056,
|
| 60 |
+
"eval_runtime": 51.1368,
|
| 61 |
+
"eval_samples_per_second": 119.366,
|
| 62 |
+
"eval_steps_per_second": 14.921,
|
| 63 |
+
"step": 1148
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 5.0,
|
| 67 |
+
"eval_accuracy": 0.29701834862385323,
|
| 68 |
+
"eval_loss": 4.060353755950928,
|
| 69 |
+
"eval_model_preparation_time": 0.0056,
|
| 70 |
+
"eval_runtime": 52.6533,
|
| 71 |
+
"eval_samples_per_second": 115.928,
|
| 72 |
+
"eval_steps_per_second": 14.491,
|
| 73 |
+
"step": 1435
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 5.2264808362369335,
|
| 77 |
+
"grad_norm": 1.33736252784729,
|
| 78 |
+
"learning_rate": 9.493891797556719e-07,
|
| 79 |
+
"loss": 4.07,
|
| 80 |
+
"step": 1500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 6.0,
|
| 84 |
+
"eval_accuracy": 0.42480340760157276,
|
| 85 |
+
"eval_loss": 4.015019416809082,
|
| 86 |
+
"eval_model_preparation_time": 0.0056,
|
| 87 |
+
"eval_runtime": 52.6896,
|
| 88 |
+
"eval_samples_per_second": 115.848,
|
| 89 |
+
"eval_steps_per_second": 14.481,
|
| 90 |
+
"step": 1722
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 6.968641114982578,
|
| 94 |
+
"grad_norm": 1.47151780128479,
|
| 95 |
+
"learning_rate": 9.31937172774869e-07,
|
| 96 |
+
"loss": 3.9838,
|
| 97 |
+
"step": 2000
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 7.0,
|
| 101 |
+
"eval_accuracy": 0.5329292267365662,
|
| 102 |
+
"eval_loss": 3.9675402641296387,
|
| 103 |
+
"eval_model_preparation_time": 0.0056,
|
| 104 |
+
"eval_runtime": 53.5702,
|
| 105 |
+
"eval_samples_per_second": 113.944,
|
| 106 |
+
"eval_steps_per_second": 14.243,
|
| 107 |
+
"step": 2009
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 8.0,
|
| 111 |
+
"eval_accuracy": 0.6192660550458715,
|
| 112 |
+
"eval_loss": 3.920508623123169,
|
| 113 |
+
"eval_model_preparation_time": 0.0056,
|
| 114 |
+
"eval_runtime": 54.4646,
|
| 115 |
+
"eval_samples_per_second": 112.073,
|
| 116 |
+
"eval_steps_per_second": 14.009,
|
| 117 |
+
"step": 2296
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 8.710801393728223,
|
| 121 |
+
"grad_norm": 1.5192607641220093,
|
| 122 |
+
"learning_rate": 9.144851657940664e-07,
|
| 123 |
+
"loss": 3.8977,
|
| 124 |
+
"step": 2500
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 9.0,
|
| 128 |
+
"eval_accuracy": 0.6908584534731324,
|
| 129 |
+
"eval_loss": 3.8727903366088867,
|
| 130 |
+
"eval_model_preparation_time": 0.0056,
|
| 131 |
+
"eval_runtime": 53.3384,
|
| 132 |
+
"eval_samples_per_second": 114.439,
|
| 133 |
+
"eval_steps_per_second": 14.305,
|
| 134 |
+
"step": 2583
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 10.0,
|
| 138 |
+
"eval_accuracy": 0.7378768020969856,
|
| 139 |
+
"eval_loss": 3.8257367610931396,
|
| 140 |
+
"eval_model_preparation_time": 0.0056,
|
| 141 |
+
"eval_runtime": 52.9624,
|
| 142 |
+
"eval_samples_per_second": 115.252,
|
| 143 |
+
"eval_steps_per_second": 14.406,
|
| 144 |
+
"step": 2870
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 10.452961672473867,
|
| 148 |
+
"grad_norm": 1.5286623239517212,
|
| 149 |
+
"learning_rate": 8.970331588132635e-07,
|
| 150 |
+
"loss": 3.8106,
|
| 151 |
+
"step": 3000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 11.0,
|
| 155 |
+
"eval_accuracy": 0.7714613368283093,
|
| 156 |
+
"eval_loss": 3.7787516117095947,
|
| 157 |
+
"eval_model_preparation_time": 0.0056,
|
| 158 |
+
"eval_runtime": 52.9786,
|
| 159 |
+
"eval_samples_per_second": 115.216,
|
| 160 |
+
"eval_steps_per_second": 14.402,
|
| 161 |
+
"step": 3157
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 12.0,
|
| 165 |
+
"eval_accuracy": 0.7980013106159896,
|
| 166 |
+
"eval_loss": 3.7332425117492676,
|
| 167 |
+
"eval_model_preparation_time": 0.0056,
|
| 168 |
+
"eval_runtime": 53.2337,
|
| 169 |
+
"eval_samples_per_second": 114.664,
|
| 170 |
+
"eval_steps_per_second": 14.333,
|
| 171 |
+
"step": 3444
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 12.195121951219512,
|
| 175 |
+
"grad_norm": 2.0538368225097656,
|
| 176 |
+
"learning_rate": 8.795811518324607e-07,
|
| 177 |
+
"loss": 3.7266,
|
| 178 |
+
"step": 3500
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 13.0,
|
| 182 |
+
"eval_accuracy": 0.8112712975098296,
|
| 183 |
+
"eval_loss": 3.68833065032959,
|
| 184 |
+
"eval_model_preparation_time": 0.0056,
|
| 185 |
+
"eval_runtime": 53.3885,
|
| 186 |
+
"eval_samples_per_second": 114.332,
|
| 187 |
+
"eval_steps_per_second": 14.291,
|
| 188 |
+
"step": 3731
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 13.937282229965156,
|
| 192 |
+
"grad_norm": 1.6309235095977783,
|
| 193 |
+
"learning_rate": 8.621291448516579e-07,
|
| 194 |
+
"loss": 3.6456,
|
| 195 |
+
"step": 4000
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 14.0,
|
| 199 |
+
"eval_accuracy": 0.8211009174311926,
|
| 200 |
+
"eval_loss": 3.642503023147583,
|
| 201 |
+
"eval_model_preparation_time": 0.0056,
|
| 202 |
+
"eval_runtime": 53.7562,
|
| 203 |
+
"eval_samples_per_second": 113.55,
|
| 204 |
+
"eval_steps_per_second": 14.194,
|
| 205 |
+
"step": 4018
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 15.0,
|
| 209 |
+
"eval_accuracy": 0.8299475753604194,
|
| 210 |
+
"eval_loss": 3.5993118286132812,
|
| 211 |
+
"eval_model_preparation_time": 0.0056,
|
| 212 |
+
"eval_runtime": 54.1732,
|
| 213 |
+
"eval_samples_per_second": 112.676,
|
| 214 |
+
"eval_steps_per_second": 14.084,
|
| 215 |
+
"step": 4305
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 15.679442508710801,
|
| 219 |
+
"grad_norm": 1.6514416933059692,
|
| 220 |
+
"learning_rate": 8.446771378708551e-07,
|
| 221 |
+
"loss": 3.5667,
|
| 222 |
+
"step": 4500
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 16.0,
|
| 226 |
+
"eval_accuracy": 0.8381389252948886,
|
| 227 |
+
"eval_loss": 3.5562214851379395,
|
| 228 |
+
"eval_model_preparation_time": 0.0056,
|
| 229 |
+
"eval_runtime": 53.3732,
|
| 230 |
+
"eval_samples_per_second": 114.365,
|
| 231 |
+
"eval_steps_per_second": 14.296,
|
| 232 |
+
"step": 4592
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 17.0,
|
| 236 |
+
"eval_accuracy": 0.8423984272608126,
|
| 237 |
+
"eval_loss": 3.514224052429199,
|
| 238 |
+
"eval_model_preparation_time": 0.0056,
|
| 239 |
+
"eval_runtime": 53.3166,
|
| 240 |
+
"eval_samples_per_second": 114.486,
|
| 241 |
+
"eval_steps_per_second": 14.311,
|
| 242 |
+
"step": 4879
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 17.421602787456447,
|
| 246 |
+
"grad_norm": 1.6095918416976929,
|
| 247 |
+
"learning_rate": 8.272251308900523e-07,
|
| 248 |
+
"loss": 3.4909,
|
| 249 |
+
"step": 5000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 18.0,
|
| 253 |
+
"eval_accuracy": 0.851572739187418,
|
| 254 |
+
"eval_loss": 3.4752180576324463,
|
| 255 |
+
"eval_model_preparation_time": 0.0056,
|
| 256 |
+
"eval_runtime": 52.6463,
|
| 257 |
+
"eval_samples_per_second": 115.944,
|
| 258 |
+
"eval_steps_per_second": 14.493,
|
| 259 |
+
"step": 5166
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 19.0,
|
| 263 |
+
"eval_accuracy": 0.85370249017038,
|
| 264 |
+
"eval_loss": 3.435464859008789,
|
| 265 |
+
"eval_model_preparation_time": 0.0056,
|
| 266 |
+
"eval_runtime": 52.9099,
|
| 267 |
+
"eval_samples_per_second": 115.366,
|
| 268 |
+
"eval_steps_per_second": 14.421,
|
| 269 |
+
"step": 5453
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 19.16376306620209,
|
| 273 |
+
"grad_norm": 1.7376962900161743,
|
| 274 |
+
"learning_rate": 8.097731239092496e-07,
|
| 275 |
+
"loss": 3.419,
|
| 276 |
+
"step": 5500
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 20.0,
|
| 280 |
+
"eval_accuracy": 0.8582896461336829,
|
| 281 |
+
"eval_loss": 3.3971521854400635,
|
| 282 |
+
"eval_model_preparation_time": 0.0056,
|
| 283 |
+
"eval_runtime": 52.7361,
|
| 284 |
+
"eval_samples_per_second": 115.746,
|
| 285 |
+
"eval_steps_per_second": 14.468,
|
| 286 |
+
"step": 5740
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 20.905923344947734,
|
| 290 |
+
"grad_norm": 1.8516736030578613,
|
| 291 |
+
"learning_rate": 7.923211169284468e-07,
|
| 292 |
+
"loss": 3.349,
|
| 293 |
+
"step": 6000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 21.0,
|
| 297 |
+
"eval_accuracy": 0.862221494102228,
|
| 298 |
+
"eval_loss": 3.35935640335083,
|
| 299 |
+
"eval_model_preparation_time": 0.0056,
|
| 300 |
+
"eval_runtime": 53.0085,
|
| 301 |
+
"eval_samples_per_second": 115.151,
|
| 302 |
+
"eval_steps_per_second": 14.394,
|
| 303 |
+
"step": 6027
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 22.0,
|
| 307 |
+
"eval_accuracy": 0.8640235910878112,
|
| 308 |
+
"eval_loss": 3.323974370956421,
|
| 309 |
+
"eval_model_preparation_time": 0.0056,
|
| 310 |
+
"eval_runtime": 53.6052,
|
| 311 |
+
"eval_samples_per_second": 113.87,
|
| 312 |
+
"eval_steps_per_second": 14.234,
|
| 313 |
+
"step": 6314
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 22.64808362369338,
|
| 317 |
+
"grad_norm": 1.8237162828445435,
|
| 318 |
+
"learning_rate": 7.74869109947644e-07,
|
| 319 |
+
"loss": 3.2859,
|
| 320 |
+
"step": 6500
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 23.0,
|
| 324 |
+
"eval_accuracy": 0.8663171690694627,
|
| 325 |
+
"eval_loss": 3.288198232650757,
|
| 326 |
+
"eval_model_preparation_time": 0.0056,
|
| 327 |
+
"eval_runtime": 52.8638,
|
| 328 |
+
"eval_samples_per_second": 115.467,
|
| 329 |
+
"eval_steps_per_second": 14.433,
|
| 330 |
+
"step": 6601
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 24.0,
|
| 334 |
+
"eval_accuracy": 0.8699213630406291,
|
| 335 |
+
"eval_loss": 3.2544491291046143,
|
| 336 |
+
"eval_model_preparation_time": 0.0056,
|
| 337 |
+
"eval_runtime": 52.3248,
|
| 338 |
+
"eval_samples_per_second": 116.656,
|
| 339 |
+
"eval_steps_per_second": 14.582,
|
| 340 |
+
"step": 6888
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 24.390243902439025,
|
| 344 |
+
"grad_norm": 1.9635404348373413,
|
| 345 |
+
"learning_rate": 7.574171029668411e-07,
|
| 346 |
+
"loss": 3.2226,
|
| 347 |
+
"step": 7000
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 25.0,
|
| 351 |
+
"eval_accuracy": 0.8697575360419397,
|
| 352 |
+
"eval_loss": 3.2205216884613037,
|
| 353 |
+
"eval_model_preparation_time": 0.0056,
|
| 354 |
+
"eval_runtime": 52.8748,
|
| 355 |
+
"eval_samples_per_second": 115.443,
|
| 356 |
+
"eval_steps_per_second": 14.43,
|
| 357 |
+
"step": 7175
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 26.0,
|
| 361 |
+
"eval_accuracy": 0.8733617300131061,
|
| 362 |
+
"eval_loss": 3.1889891624450684,
|
| 363 |
+
"eval_model_preparation_time": 0.0056,
|
| 364 |
+
"eval_runtime": 54.213,
|
| 365 |
+
"eval_samples_per_second": 112.593,
|
| 366 |
+
"eval_steps_per_second": 14.074,
|
| 367 |
+
"step": 7462
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 26.13240418118467,
|
| 371 |
+
"grad_norm": 1.7023582458496094,
|
| 372 |
+
"learning_rate": 7.399650959860384e-07,
|
| 373 |
+
"loss": 3.1622,
|
| 374 |
+
"step": 7500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 27.0,
|
| 378 |
+
"eval_accuracy": 0.8756553079947575,
|
| 379 |
+
"eval_loss": 3.157259941101074,
|
| 380 |
+
"eval_model_preparation_time": 0.0056,
|
| 381 |
+
"eval_runtime": 52.9974,
|
| 382 |
+
"eval_samples_per_second": 115.176,
|
| 383 |
+
"eval_steps_per_second": 14.397,
|
| 384 |
+
"step": 7749
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 27.874564459930312,
|
| 388 |
+
"grad_norm": 1.7533528804779053,
|
| 389 |
+
"learning_rate": 7.225130890052355e-07,
|
| 390 |
+
"loss": 3.1082,
|
| 391 |
+
"step": 8000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 28.0,
|
| 395 |
+
"eval_accuracy": 0.8787680209698558,
|
| 396 |
+
"eval_loss": 3.126617908477783,
|
| 397 |
+
"eval_model_preparation_time": 0.0056,
|
| 398 |
+
"eval_runtime": 52.8562,
|
| 399 |
+
"eval_samples_per_second": 115.483,
|
| 400 |
+
"eval_steps_per_second": 14.435,
|
| 401 |
+
"step": 8036
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 29.0,
|
| 405 |
+
"eval_accuracy": 0.8807339449541285,
|
| 406 |
+
"eval_loss": 3.0970399379730225,
|
| 407 |
+
"eval_model_preparation_time": 0.0056,
|
| 408 |
+
"eval_runtime": 52.9688,
|
| 409 |
+
"eval_samples_per_second": 115.238,
|
| 410 |
+
"eval_steps_per_second": 14.405,
|
| 411 |
+
"step": 8323
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 29.61672473867596,
|
| 415 |
+
"grad_norm": 1.8451896905899048,
|
| 416 |
+
"learning_rate": 7.050610820244328e-07,
|
| 417 |
+
"loss": 3.0527,
|
| 418 |
+
"step": 8500
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 30.0,
|
| 422 |
+
"eval_accuracy": 0.882699868938401,
|
| 423 |
+
"eval_loss": 3.068537950515747,
|
| 424 |
+
"eval_model_preparation_time": 0.0056,
|
| 425 |
+
"eval_runtime": 52.723,
|
| 426 |
+
"eval_samples_per_second": 115.775,
|
| 427 |
+
"eval_steps_per_second": 14.472,
|
| 428 |
+
"step": 8610
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 31.0,
|
| 432 |
+
"eval_accuracy": 0.8822083879423329,
|
| 433 |
+
"eval_loss": 3.0398900508880615,
|
| 434 |
+
"eval_model_preparation_time": 0.0056,
|
| 435 |
+
"eval_runtime": 52.8402,
|
| 436 |
+
"eval_samples_per_second": 115.518,
|
| 437 |
+
"eval_steps_per_second": 14.44,
|
| 438 |
+
"step": 8897
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 31.358885017421603,
|
| 442 |
+
"grad_norm": 1.9000401496887207,
|
| 443 |
+
"learning_rate": 6.8760907504363e-07,
|
| 444 |
+
"loss": 3.0013,
|
| 445 |
+
"step": 9000
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 32.0,
|
| 449 |
+
"eval_accuracy": 0.8843381389252949,
|
| 450 |
+
"eval_loss": 3.0114569664001465,
|
| 451 |
+
"eval_model_preparation_time": 0.0056,
|
| 452 |
+
"eval_runtime": 52.8176,
|
| 453 |
+
"eval_samples_per_second": 115.568,
|
| 454 |
+
"eval_steps_per_second": 14.446,
|
| 455 |
+
"step": 9184
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 33.0,
|
| 459 |
+
"eval_accuracy": 0.8863040629095675,
|
| 460 |
+
"eval_loss": 2.9846866130828857,
|
| 461 |
+
"eval_model_preparation_time": 0.0056,
|
| 462 |
+
"eval_runtime": 52.9145,
|
| 463 |
+
"eval_samples_per_second": 115.356,
|
| 464 |
+
"eval_steps_per_second": 14.419,
|
| 465 |
+
"step": 9471
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 33.10104529616725,
|
| 469 |
+
"grad_norm": 1.8424733877182007,
|
| 470 |
+
"learning_rate": 6.701570680628273e-07,
|
| 471 |
+
"loss": 2.9544,
|
| 472 |
+
"step": 9500
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 34.0,
|
| 476 |
+
"eval_accuracy": 0.8881061598951507,
|
| 477 |
+
"eval_loss": 2.9583957195281982,
|
| 478 |
+
"eval_model_preparation_time": 0.0056,
|
| 479 |
+
"eval_runtime": 53.5577,
|
| 480 |
+
"eval_samples_per_second": 113.97,
|
| 481 |
+
"eval_steps_per_second": 14.246,
|
| 482 |
+
"step": 9758
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 34.84320557491289,
|
| 486 |
+
"grad_norm": 1.8269978761672974,
|
| 487 |
+
"learning_rate": 6.527050610820244e-07,
|
| 488 |
+
"loss": 2.9042,
|
| 489 |
+
"step": 10000
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 35.0,
|
| 493 |
+
"eval_accuracy": 0.8890891218872871,
|
| 494 |
+
"eval_loss": 2.93341326713562,
|
| 495 |
+
"eval_model_preparation_time": 0.0056,
|
| 496 |
+
"eval_runtime": 53.4697,
|
| 497 |
+
"eval_samples_per_second": 114.158,
|
| 498 |
+
"eval_steps_per_second": 14.27,
|
| 499 |
+
"step": 10045
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 36.0,
|
| 503 |
+
"eval_accuracy": 0.890399737876802,
|
| 504 |
+
"eval_loss": 2.908048152923584,
|
| 505 |
+
"eval_model_preparation_time": 0.0056,
|
| 506 |
+
"eval_runtime": 53.4437,
|
| 507 |
+
"eval_samples_per_second": 114.214,
|
| 508 |
+
"eval_steps_per_second": 14.277,
|
| 509 |
+
"step": 10332
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 36.58536585365854,
|
| 513 |
+
"grad_norm": 2.0479116439819336,
|
| 514 |
+
"learning_rate": 6.352530541012216e-07,
|
| 515 |
+
"loss": 2.8575,
|
| 516 |
+
"step": 10500
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 37.0,
|
| 520 |
+
"eval_accuracy": 0.8907273918741808,
|
| 521 |
+
"eval_loss": 2.8831241130828857,
|
| 522 |
+
"eval_model_preparation_time": 0.0056,
|
| 523 |
+
"eval_runtime": 52.6355,
|
| 524 |
+
"eval_samples_per_second": 115.967,
|
| 525 |
+
"eval_steps_per_second": 14.496,
|
| 526 |
+
"step": 10619
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 38.0,
|
| 530 |
+
"eval_accuracy": 0.8918741808650066,
|
| 531 |
+
"eval_loss": 2.860414981842041,
|
| 532 |
+
"eval_model_preparation_time": 0.0056,
|
| 533 |
+
"eval_runtime": 54.5384,
|
| 534 |
+
"eval_samples_per_second": 111.921,
|
| 535 |
+
"eval_steps_per_second": 13.99,
|
| 536 |
+
"step": 10906
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 38.32752613240418,
|
| 540 |
+
"grad_norm": 2.2141685485839844,
|
| 541 |
+
"learning_rate": 6.178010471204188e-07,
|
| 542 |
+
"loss": 2.8171,
|
| 543 |
+
"step": 11000
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 39.0,
|
| 547 |
+
"eval_accuracy": 0.8926933158584535,
|
| 548 |
+
"eval_loss": 2.8362042903900146,
|
| 549 |
+
"eval_model_preparation_time": 0.0056,
|
| 550 |
+
"eval_runtime": 52.4046,
|
| 551 |
+
"eval_samples_per_second": 116.478,
|
| 552 |
+
"eval_steps_per_second": 14.56,
|
| 553 |
+
"step": 11193
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 40.0,
|
| 557 |
+
"eval_accuracy": 0.8940039318479686,
|
| 558 |
+
"eval_loss": 2.813732147216797,
|
| 559 |
+
"eval_model_preparation_time": 0.0056,
|
| 560 |
+
"eval_runtime": 53.0566,
|
| 561 |
+
"eval_samples_per_second": 115.047,
|
| 562 |
+
"eval_steps_per_second": 14.381,
|
| 563 |
+
"step": 11480
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 40.069686411149824,
|
| 567 |
+
"grad_norm": 1.753510594367981,
|
| 568 |
+
"learning_rate": 6.00349040139616e-07,
|
| 569 |
+
"loss": 2.7729,
|
| 570 |
+
"step": 11500
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 41.0,
|
| 574 |
+
"eval_accuracy": 0.8948230668414154,
|
| 575 |
+
"eval_loss": 2.7919933795928955,
|
| 576 |
+
"eval_model_preparation_time": 0.0056,
|
| 577 |
+
"eval_runtime": 53.4649,
|
| 578 |
+
"eval_samples_per_second": 114.168,
|
| 579 |
+
"eval_steps_per_second": 14.271,
|
| 580 |
+
"step": 11767
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 41.81184668989547,
|
| 584 |
+
"grad_norm": 2.048759937286377,
|
| 585 |
+
"learning_rate": 5.828970331588133e-07,
|
| 586 |
+
"loss": 2.7358,
|
| 587 |
+
"step": 12000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 42.0,
|
| 591 |
+
"eval_accuracy": 0.8956422018348624,
|
| 592 |
+
"eval_loss": 2.7705893516540527,
|
| 593 |
+
"eval_model_preparation_time": 0.0056,
|
| 594 |
+
"eval_runtime": 52.8824,
|
| 595 |
+
"eval_samples_per_second": 115.426,
|
| 596 |
+
"eval_steps_per_second": 14.428,
|
| 597 |
+
"step": 12054
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 43.0,
|
| 601 |
+
"eval_accuracy": 0.89629750982962,
|
| 602 |
+
"eval_loss": 2.7493233680725098,
|
| 603 |
+
"eval_model_preparation_time": 0.0056,
|
| 604 |
+
"eval_runtime": 53.4446,
|
| 605 |
+
"eval_samples_per_second": 114.212,
|
| 606 |
+
"eval_steps_per_second": 14.276,
|
| 607 |
+
"step": 12341
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 43.55400696864111,
|
| 611 |
+
"grad_norm": 1.7660369873046875,
|
| 612 |
+
"learning_rate": 5.654450261780105e-07,
|
| 613 |
+
"loss": 2.6966,
|
| 614 |
+
"step": 12500
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 44.0,
|
| 618 |
+
"eval_accuracy": 0.8985910878112713,
|
| 619 |
+
"eval_loss": 2.7291805744171143,
|
| 620 |
+
"eval_model_preparation_time": 0.0056,
|
| 621 |
+
"eval_runtime": 52.7207,
|
| 622 |
+
"eval_samples_per_second": 115.78,
|
| 623 |
+
"eval_steps_per_second": 14.472,
|
| 624 |
+
"step": 12628
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 45.0,
|
| 628 |
+
"eval_accuracy": 0.8989187418086501,
|
| 629 |
+
"eval_loss": 2.708031415939331,
|
| 630 |
+
"eval_model_preparation_time": 0.0056,
|
| 631 |
+
"eval_runtime": 52.6994,
|
| 632 |
+
"eval_samples_per_second": 115.827,
|
| 633 |
+
"eval_steps_per_second": 14.478,
|
| 634 |
+
"step": 12915
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 45.29616724738676,
|
| 638 |
+
"grad_norm": 1.8633924722671509,
|
| 639 |
+
"learning_rate": 5.479930191972077e-07,
|
| 640 |
+
"loss": 2.6579,
|
| 641 |
+
"step": 13000
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 46.0,
|
| 645 |
+
"eval_accuracy": 0.8989187418086501,
|
| 646 |
+
"eval_loss": 2.6885175704956055,
|
| 647 |
+
"eval_model_preparation_time": 0.0056,
|
| 648 |
+
"eval_runtime": 52.82,
|
| 649 |
+
"eval_samples_per_second": 115.562,
|
| 650 |
+
"eval_steps_per_second": 14.445,
|
| 651 |
+
"step": 13202
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 47.0,
|
| 655 |
+
"eval_accuracy": 0.9008846657929227,
|
| 656 |
+
"eval_loss": 2.669055461883545,
|
| 657 |
+
"eval_model_preparation_time": 0.0056,
|
| 658 |
+
"eval_runtime": 52.9106,
|
| 659 |
+
"eval_samples_per_second": 115.364,
|
| 660 |
+
"eval_steps_per_second": 14.421,
|
| 661 |
+
"step": 13489
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 47.038327526132406,
|
| 665 |
+
"grad_norm": 2.1170654296875,
|
| 666 |
+
"learning_rate": 5.305410122164049e-07,
|
| 667 |
+
"loss": 2.6229,
|
| 668 |
+
"step": 13500
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 48.0,
|
| 672 |
+
"eval_accuracy": 0.901048492791612,
|
| 673 |
+
"eval_loss": 2.6509358882904053,
|
| 674 |
+
"eval_model_preparation_time": 0.0056,
|
| 675 |
+
"eval_runtime": 53.208,
|
| 676 |
+
"eval_samples_per_second": 114.72,
|
| 677 |
+
"eval_steps_per_second": 14.34,
|
| 678 |
+
"step": 13776
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 48.78048780487805,
|
| 682 |
+
"grad_norm": 2.170200824737549,
|
| 683 |
+
"learning_rate": 5.13089005235602e-07,
|
| 684 |
+
"loss": 2.5897,
|
| 685 |
+
"step": 14000
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 49.0,
|
| 689 |
+
"eval_accuracy": 0.9023591087811271,
|
| 690 |
+
"eval_loss": 2.6330766677856445,
|
| 691 |
+
"eval_model_preparation_time": 0.0056,
|
| 692 |
+
"eval_runtime": 53.5861,
|
| 693 |
+
"eval_samples_per_second": 113.91,
|
| 694 |
+
"eval_steps_per_second": 14.239,
|
| 695 |
+
"step": 14063
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 50.0,
|
| 699 |
+
"eval_accuracy": 0.9026867627785059,
|
| 700 |
+
"eval_loss": 2.615280866622925,
|
| 701 |
+
"eval_model_preparation_time": 0.0056,
|
| 702 |
+
"eval_runtime": 53.1713,
|
| 703 |
+
"eval_samples_per_second": 114.799,
|
| 704 |
+
"eval_steps_per_second": 14.35,
|
| 705 |
+
"step": 14350
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 50.52264808362369,
|
| 709 |
+
"grad_norm": 2.208359956741333,
|
| 710 |
+
"learning_rate": 4.956369982547993e-07,
|
| 711 |
+
"loss": 2.5583,
|
| 712 |
+
"step": 14500
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 51.0,
|
| 716 |
+
"eval_accuracy": 0.9025229357798165,
|
| 717 |
+
"eval_loss": 2.5973732471466064,
|
| 718 |
+
"eval_model_preparation_time": 0.0056,
|
| 719 |
+
"eval_runtime": 52.7169,
|
| 720 |
+
"eval_samples_per_second": 115.788,
|
| 721 |
+
"eval_steps_per_second": 14.474,
|
| 722 |
+
"step": 14637
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 52.0,
|
| 726 |
+
"eval_accuracy": 0.9043250327653998,
|
| 727 |
+
"eval_loss": 2.581191062927246,
|
| 728 |
+
"eval_model_preparation_time": 0.0056,
|
| 729 |
+
"eval_runtime": 54.1621,
|
| 730 |
+
"eval_samples_per_second": 112.699,
|
| 731 |
+
"eval_steps_per_second": 14.087,
|
| 732 |
+
"step": 14924
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 52.26480836236934,
|
| 736 |
+
"grad_norm": 2.159825325012207,
|
| 737 |
+
"learning_rate": 4.781849912739964e-07,
|
| 738 |
+
"loss": 2.5257,
|
| 739 |
+
"step": 15000
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 53.0,
|
| 743 |
+
"eval_accuracy": 0.9035058977719528,
|
| 744 |
+
"eval_loss": 2.564519166946411,
|
| 745 |
+
"eval_model_preparation_time": 0.0056,
|
| 746 |
+
"eval_runtime": 54.0888,
|
| 747 |
+
"eval_samples_per_second": 112.851,
|
| 748 |
+
"eval_steps_per_second": 14.106,
|
| 749 |
+
"step": 15211
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 54.0,
|
| 753 |
+
"eval_accuracy": 0.9041612057667103,
|
| 754 |
+
"eval_loss": 2.5490972995758057,
|
| 755 |
+
"eval_model_preparation_time": 0.0056,
|
| 756 |
+
"eval_runtime": 54.8624,
|
| 757 |
+
"eval_samples_per_second": 111.26,
|
| 758 |
+
"eval_steps_per_second": 13.908,
|
| 759 |
+
"step": 15498
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 54.00696864111498,
|
| 763 |
+
"grad_norm": 1.8598278760910034,
|
| 764 |
+
"learning_rate": 4.607329842931937e-07,
|
| 765 |
+
"loss": 2.4986,
|
| 766 |
+
"step": 15500
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 55.0,
|
| 770 |
+
"eval_accuracy": 0.9059633027522935,
|
| 771 |
+
"eval_loss": 2.533395528793335,
|
| 772 |
+
"eval_model_preparation_time": 0.0056,
|
| 773 |
+
"eval_runtime": 52.7918,
|
| 774 |
+
"eval_samples_per_second": 115.624,
|
| 775 |
+
"eval_steps_per_second": 14.453,
|
| 776 |
+
"step": 15785
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 55.749128919860624,
|
| 780 |
+
"grad_norm": 1.9237112998962402,
|
| 781 |
+
"learning_rate": 4.4328097731239087e-07,
|
| 782 |
+
"loss": 2.4703,
|
| 783 |
+
"step": 16000
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 56.0,
|
| 787 |
+
"eval_accuracy": 0.9056356487549148,
|
| 788 |
+
"eval_loss": 2.5186452865600586,
|
| 789 |
+
"eval_model_preparation_time": 0.0056,
|
| 790 |
+
"eval_runtime": 52.6994,
|
| 791 |
+
"eval_samples_per_second": 115.827,
|
| 792 |
+
"eval_steps_per_second": 14.478,
|
| 793 |
+
"step": 16072
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 57.0,
|
| 797 |
+
"eval_accuracy": 0.906127129750983,
|
| 798 |
+
"eval_loss": 2.504196882247925,
|
| 799 |
+
"eval_model_preparation_time": 0.0056,
|
| 800 |
+
"eval_runtime": 53.0824,
|
| 801 |
+
"eval_samples_per_second": 114.991,
|
| 802 |
+
"eval_steps_per_second": 14.374,
|
| 803 |
+
"step": 16359
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 57.491289198606275,
|
| 807 |
+
"grad_norm": 2.3951334953308105,
|
| 808 |
+
"learning_rate": 4.258289703315881e-07,
|
| 809 |
+
"loss": 2.444,
|
| 810 |
+
"step": 16500
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 58.0,
|
| 814 |
+
"eval_accuracy": 0.9092398427260813,
|
| 815 |
+
"eval_loss": 2.48964262008667,
|
| 816 |
+
"eval_model_preparation_time": 0.0056,
|
| 817 |
+
"eval_runtime": 52.6693,
|
| 818 |
+
"eval_samples_per_second": 115.893,
|
| 819 |
+
"eval_steps_per_second": 14.487,
|
| 820 |
+
"step": 16646
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 59.0,
|
| 824 |
+
"eval_accuracy": 0.9092398427260813,
|
| 825 |
+
"eval_loss": 2.4755215644836426,
|
| 826 |
+
"eval_model_preparation_time": 0.0056,
|
| 827 |
+
"eval_runtime": 53.0222,
|
| 828 |
+
"eval_samples_per_second": 115.121,
|
| 829 |
+
"eval_steps_per_second": 14.39,
|
| 830 |
+
"step": 16933
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 59.23344947735192,
|
| 834 |
+
"grad_norm": 2.315953493118286,
|
| 835 |
+
"learning_rate": 4.083769633507853e-07,
|
| 836 |
+
"loss": 2.4157,
|
| 837 |
+
"step": 17000
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 60.0,
|
| 841 |
+
"eval_accuracy": 0.908748361730013,
|
| 842 |
+
"eval_loss": 2.4624128341674805,
|
| 843 |
+
"eval_model_preparation_time": 0.0056,
|
| 844 |
+
"eval_runtime": 53.0917,
|
| 845 |
+
"eval_samples_per_second": 114.971,
|
| 846 |
+
"eval_steps_per_second": 14.371,
|
| 847 |
+
"step": 17220
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 60.97560975609756,
|
| 851 |
+
"grad_norm": 2.262585401535034,
|
| 852 |
+
"learning_rate": 3.909249563699825e-07,
|
| 853 |
+
"loss": 2.3904,
|
| 854 |
+
"step": 17500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 61.0,
|
| 858 |
+
"eval_accuracy": 0.90956749672346,
|
| 859 |
+
"eval_loss": 2.448362112045288,
|
| 860 |
+
"eval_model_preparation_time": 0.0056,
|
| 861 |
+
"eval_runtime": 53.2314,
|
| 862 |
+
"eval_samples_per_second": 114.669,
|
| 863 |
+
"eval_steps_per_second": 14.334,
|
| 864 |
+
"step": 17507
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 62.0,
|
| 868 |
+
"eval_accuracy": 0.9102228047182176,
|
| 869 |
+
"eval_loss": 2.435769557952881,
|
| 870 |
+
"eval_model_preparation_time": 0.0056,
|
| 871 |
+
"eval_runtime": 55.5115,
|
| 872 |
+
"eval_samples_per_second": 109.959,
|
| 873 |
+
"eval_steps_per_second": 13.745,
|
| 874 |
+
"step": 17794
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 62.717770034843205,
|
| 878 |
+
"grad_norm": 1.9645085334777832,
|
| 879 |
+
"learning_rate": 3.7347294938917974e-07,
|
| 880 |
+
"loss": 2.367,
|
| 881 |
+
"step": 18000
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 63.0,
|
| 885 |
+
"eval_accuracy": 0.9110419397116645,
|
| 886 |
+
"eval_loss": 2.422717809677124,
|
| 887 |
+
"eval_model_preparation_time": 0.0056,
|
| 888 |
+
"eval_runtime": 55.4553,
|
| 889 |
+
"eval_samples_per_second": 110.071,
|
| 890 |
+
"eval_steps_per_second": 13.759,
|
| 891 |
+
"step": 18081
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 64.0,
|
| 895 |
+
"eval_accuracy": 0.9118610747051114,
|
| 896 |
+
"eval_loss": 2.4112889766693115,
|
| 897 |
+
"eval_model_preparation_time": 0.0056,
|
| 898 |
+
"eval_runtime": 54.3035,
|
| 899 |
+
"eval_samples_per_second": 112.405,
|
| 900 |
+
"eval_steps_per_second": 14.051,
|
| 901 |
+
"step": 18368
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 64.45993031358886,
|
| 905 |
+
"grad_norm": 2.300833225250244,
|
| 906 |
+
"learning_rate": 3.5602094240837693e-07,
|
| 907 |
+
"loss": 2.3506,
|
| 908 |
+
"step": 18500
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 65.0,
|
| 912 |
+
"eval_accuracy": 0.9120249017038008,
|
| 913 |
+
"eval_loss": 2.399442434310913,
|
| 914 |
+
"eval_model_preparation_time": 0.0056,
|
| 915 |
+
"eval_runtime": 54.4675,
|
| 916 |
+
"eval_samples_per_second": 112.067,
|
| 917 |
+
"eval_steps_per_second": 14.008,
|
| 918 |
+
"step": 18655
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 66.0,
|
| 922 |
+
"eval_accuracy": 0.9125163826998689,
|
| 923 |
+
"eval_loss": 2.3884286880493164,
|
| 924 |
+
"eval_model_preparation_time": 0.0056,
|
| 925 |
+
"eval_runtime": 54.201,
|
| 926 |
+
"eval_samples_per_second": 112.618,
|
| 927 |
+
"eval_steps_per_second": 14.077,
|
| 928 |
+
"step": 18942
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 66.2020905923345,
|
| 932 |
+
"grad_norm": 2.1156551837921143,
|
| 933 |
+
"learning_rate": 3.3856893542757413e-07,
|
| 934 |
+
"loss": 2.3237,
|
| 935 |
+
"step": 19000
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 67.0,
|
| 939 |
+
"eval_accuracy": 0.9130078636959371,
|
| 940 |
+
"eval_loss": 2.3775436878204346,
|
| 941 |
+
"eval_model_preparation_time": 0.0056,
|
| 942 |
+
"eval_runtime": 54.1909,
|
| 943 |
+
"eval_samples_per_second": 112.639,
|
| 944 |
+
"eval_steps_per_second": 14.08,
|
| 945 |
+
"step": 19229
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 67.94425087108014,
|
| 949 |
+
"grad_norm": 1.9504848718643188,
|
| 950 |
+
"learning_rate": 3.2111692844677137e-07,
|
| 951 |
+
"loss": 2.3053,
|
| 952 |
+
"step": 19500
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 68.0,
|
| 956 |
+
"eval_accuracy": 0.9133355176933159,
|
| 957 |
+
"eval_loss": 2.3665952682495117,
|
| 958 |
+
"eval_model_preparation_time": 0.0056,
|
| 959 |
+
"eval_runtime": 54.7045,
|
| 960 |
+
"eval_samples_per_second": 111.581,
|
| 961 |
+
"eval_steps_per_second": 13.948,
|
| 962 |
+
"step": 19516
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 69.0,
|
| 966 |
+
"eval_accuracy": 0.9131716906946264,
|
| 967 |
+
"eval_loss": 2.356461763381958,
|
| 968 |
+
"eval_model_preparation_time": 0.0056,
|
| 969 |
+
"eval_runtime": 53.8382,
|
| 970 |
+
"eval_samples_per_second": 113.377,
|
| 971 |
+
"eval_steps_per_second": 14.172,
|
| 972 |
+
"step": 19803
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 69.68641114982579,
|
| 976 |
+
"grad_norm": 1.9092929363250732,
|
| 977 |
+
"learning_rate": 3.0366492146596856e-07,
|
| 978 |
+
"loss": 2.285,
|
| 979 |
+
"step": 20000
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 70.0,
|
| 983 |
+
"eval_accuracy": 0.9139908256880734,
|
| 984 |
+
"eval_loss": 2.34682559967041,
|
| 985 |
+
"eval_model_preparation_time": 0.0056,
|
| 986 |
+
"eval_runtime": 53.9142,
|
| 987 |
+
"eval_samples_per_second": 113.217,
|
| 988 |
+
"eval_steps_per_second": 14.152,
|
| 989 |
+
"step": 20090
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 71.0,
|
| 993 |
+
"eval_accuracy": 0.9141546526867628,
|
| 994 |
+
"eval_loss": 2.3373236656188965,
|
| 995 |
+
"eval_model_preparation_time": 0.0056,
|
| 996 |
+
"eval_runtime": 56.1241,
|
| 997 |
+
"eval_samples_per_second": 108.759,
|
| 998 |
+
"eval_steps_per_second": 13.595,
|
| 999 |
+
"step": 20377
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 71.42857142857143,
|
| 1003 |
+
"grad_norm": 2.050865411758423,
|
| 1004 |
+
"learning_rate": 2.8621291448516576e-07,
|
| 1005 |
+
"loss": 2.2706,
|
| 1006 |
+
"step": 20500
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 72.0,
|
| 1010 |
+
"eval_accuracy": 0.9141546526867628,
|
| 1011 |
+
"eval_loss": 2.3278188705444336,
|
| 1012 |
+
"eval_model_preparation_time": 0.0056,
|
| 1013 |
+
"eval_runtime": 52.6008,
|
| 1014 |
+
"eval_samples_per_second": 116.044,
|
| 1015 |
+
"eval_steps_per_second": 14.505,
|
| 1016 |
+
"step": 20664
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 73.0,
|
| 1020 |
+
"eval_accuracy": 0.9148099606815203,
|
| 1021 |
+
"eval_loss": 2.3193538188934326,
|
| 1022 |
+
"eval_model_preparation_time": 0.0056,
|
| 1023 |
+
"eval_runtime": 53.1272,
|
| 1024 |
+
"eval_samples_per_second": 114.894,
|
| 1025 |
+
"eval_steps_per_second": 14.362,
|
| 1026 |
+
"step": 20951
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 73.17073170731707,
|
| 1030 |
+
"grad_norm": 2.2429611682891846,
|
| 1031 |
+
"learning_rate": 2.68760907504363e-07,
|
| 1032 |
+
"loss": 2.2531,
|
| 1033 |
+
"step": 21000
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 74.0,
|
| 1037 |
+
"eval_accuracy": 0.9161205766710354,
|
| 1038 |
+
"eval_loss": 2.3109400272369385,
|
| 1039 |
+
"eval_model_preparation_time": 0.0056,
|
| 1040 |
+
"eval_runtime": 54.1757,
|
| 1041 |
+
"eval_samples_per_second": 112.671,
|
| 1042 |
+
"eval_steps_per_second": 14.084,
|
| 1043 |
+
"step": 21238
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 74.91289198606272,
|
| 1047 |
+
"grad_norm": 2.5070650577545166,
|
| 1048 |
+
"learning_rate": 2.513089005235602e-07,
|
| 1049 |
+
"loss": 2.2372,
|
| 1050 |
+
"step": 21500
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 75.0,
|
| 1054 |
+
"eval_accuracy": 0.9157929226736566,
|
| 1055 |
+
"eval_loss": 2.3027830123901367,
|
| 1056 |
+
"eval_model_preparation_time": 0.0056,
|
| 1057 |
+
"eval_runtime": 53.9429,
|
| 1058 |
+
"eval_samples_per_second": 113.157,
|
| 1059 |
+
"eval_steps_per_second": 14.145,
|
| 1060 |
+
"step": 21525
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 76.0,
|
| 1064 |
+
"eval_accuracy": 0.915956749672346,
|
| 1065 |
+
"eval_loss": 2.2948713302612305,
|
| 1066 |
+
"eval_model_preparation_time": 0.0056,
|
| 1067 |
+
"eval_runtime": 53.7032,
|
| 1068 |
+
"eval_samples_per_second": 113.662,
|
| 1069 |
+
"eval_steps_per_second": 14.208,
|
| 1070 |
+
"step": 21812
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 76.65505226480836,
|
| 1074 |
+
"grad_norm": 1.9022382497787476,
|
| 1075 |
+
"learning_rate": 2.338568935427574e-07,
|
| 1076 |
+
"loss": 2.223,
|
| 1077 |
+
"step": 22000
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 77.0,
|
| 1081 |
+
"eval_accuracy": 0.9151376146788991,
|
| 1082 |
+
"eval_loss": 2.2872207164764404,
|
| 1083 |
+
"eval_model_preparation_time": 0.0056,
|
| 1084 |
+
"eval_runtime": 52.4931,
|
| 1085 |
+
"eval_samples_per_second": 116.282,
|
| 1086 |
+
"eval_steps_per_second": 14.535,
|
| 1087 |
+
"step": 22099
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 78.0,
|
| 1091 |
+
"eval_accuracy": 0.9161205766710354,
|
| 1092 |
+
"eval_loss": 2.2804179191589355,
|
| 1093 |
+
"eval_model_preparation_time": 0.0056,
|
| 1094 |
+
"eval_runtime": 52.9333,
|
| 1095 |
+
"eval_samples_per_second": 115.315,
|
| 1096 |
+
"eval_steps_per_second": 14.414,
|
| 1097 |
+
"step": 22386
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 78.397212543554,
|
| 1101 |
+
"grad_norm": 2.776266574859619,
|
| 1102 |
+
"learning_rate": 2.164048865619546e-07,
|
| 1103 |
+
"loss": 2.2134,
|
| 1104 |
+
"step": 22500
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 79.0,
|
| 1108 |
+
"eval_accuracy": 0.9164482306684142,
|
| 1109 |
+
"eval_loss": 2.2735037803649902,
|
| 1110 |
+
"eval_model_preparation_time": 0.0056,
|
| 1111 |
+
"eval_runtime": 51.885,
|
| 1112 |
+
"eval_samples_per_second": 117.645,
|
| 1113 |
+
"eval_steps_per_second": 14.706,
|
| 1114 |
+
"step": 22673
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 80.0,
|
| 1118 |
+
"eval_accuracy": 0.9166120576671035,
|
| 1119 |
+
"eval_loss": 2.266995668411255,
|
| 1120 |
+
"eval_model_preparation_time": 0.0056,
|
| 1121 |
+
"eval_runtime": 52.3579,
|
| 1122 |
+
"eval_samples_per_second": 116.582,
|
| 1123 |
+
"eval_steps_per_second": 14.573,
|
| 1124 |
+
"step": 22960
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 80.13937282229965,
|
| 1128 |
+
"grad_norm": 2.38444447517395,
|
| 1129 |
+
"learning_rate": 1.989528795811518e-07,
|
| 1130 |
+
"loss": 2.195,
|
| 1131 |
+
"step": 23000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 81.0,
|
| 1135 |
+
"eval_accuracy": 0.9166120576671035,
|
| 1136 |
+
"eval_loss": 2.2608718872070312,
|
| 1137 |
+
"eval_model_preparation_time": 0.0056,
|
| 1138 |
+
"eval_runtime": 52.4181,
|
| 1139 |
+
"eval_samples_per_second": 116.448,
|
| 1140 |
+
"eval_steps_per_second": 14.556,
|
| 1141 |
+
"step": 23247
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 81.88153310104529,
|
| 1145 |
+
"grad_norm": 2.126260757446289,
|
| 1146 |
+
"learning_rate": 1.8150087260034902e-07,
|
| 1147 |
+
"loss": 2.1856,
|
| 1148 |
+
"step": 23500
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 82.0,
|
| 1152 |
+
"eval_accuracy": 0.9166120576671035,
|
| 1153 |
+
"eval_loss": 2.2549705505371094,
|
| 1154 |
+
"eval_model_preparation_time": 0.0056,
|
| 1155 |
+
"eval_runtime": 52.0305,
|
| 1156 |
+
"eval_samples_per_second": 117.316,
|
| 1157 |
+
"eval_steps_per_second": 14.664,
|
| 1158 |
+
"step": 23534
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 83.0,
|
| 1162 |
+
"eval_accuracy": 0.9175950196592398,
|
| 1163 |
+
"eval_loss": 2.2493834495544434,
|
| 1164 |
+
"eval_model_preparation_time": 0.0056,
|
| 1165 |
+
"eval_runtime": 51.6617,
|
| 1166 |
+
"eval_samples_per_second": 118.153,
|
| 1167 |
+
"eval_steps_per_second": 14.769,
|
| 1168 |
+
"step": 23821
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 83.62369337979094,
|
| 1172 |
+
"grad_norm": 2.363184928894043,
|
| 1173 |
+
"learning_rate": 1.6404886561954624e-07,
|
| 1174 |
+
"loss": 2.1759,
|
| 1175 |
+
"step": 24000
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 84.0,
|
| 1179 |
+
"eval_accuracy": 0.9174311926605505,
|
| 1180 |
+
"eval_loss": 2.244133949279785,
|
| 1181 |
+
"eval_model_preparation_time": 0.0056,
|
| 1182 |
+
"eval_runtime": 51.6512,
|
| 1183 |
+
"eval_samples_per_second": 118.177,
|
| 1184 |
+
"eval_steps_per_second": 14.772,
|
| 1185 |
+
"step": 24108
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 85.0,
|
| 1189 |
+
"eval_accuracy": 0.918086500655308,
|
| 1190 |
+
"eval_loss": 2.2390334606170654,
|
| 1191 |
+
"eval_model_preparation_time": 0.0056,
|
| 1192 |
+
"eval_runtime": 50.8842,
|
| 1193 |
+
"eval_samples_per_second": 119.959,
|
| 1194 |
+
"eval_steps_per_second": 14.995,
|
| 1195 |
+
"step": 24395
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 85.36585365853658,
|
| 1199 |
+
"grad_norm": 2.263583183288574,
|
| 1200 |
+
"learning_rate": 1.4659685863874343e-07,
|
| 1201 |
+
"loss": 2.1669,
|
| 1202 |
+
"step": 24500
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 86.0,
|
| 1206 |
+
"eval_accuracy": 0.9184141546526867,
|
| 1207 |
+
"eval_loss": 2.2345142364501953,
|
| 1208 |
+
"eval_model_preparation_time": 0.0056,
|
| 1209 |
+
"eval_runtime": 52.6434,
|
| 1210 |
+
"eval_samples_per_second": 115.95,
|
| 1211 |
+
"eval_steps_per_second": 14.494,
|
| 1212 |
+
"step": 24682
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 87.0,
|
| 1216 |
+
"eval_accuracy": 0.9182503276539974,
|
| 1217 |
+
"eval_loss": 2.230360984802246,
|
| 1218 |
+
"eval_model_preparation_time": 0.0056,
|
| 1219 |
+
"eval_runtime": 53.4573,
|
| 1220 |
+
"eval_samples_per_second": 114.185,
|
| 1221 |
+
"eval_steps_per_second": 14.273,
|
| 1222 |
+
"step": 24969
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 87.10801393728222,
|
| 1226 |
+
"grad_norm": 2.30412220954895,
|
| 1227 |
+
"learning_rate": 1.2914485165794065e-07,
|
| 1228 |
+
"loss": 2.1574,
|
| 1229 |
+
"step": 25000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 88.0,
|
| 1233 |
+
"eval_accuracy": 0.9185779816513762,
|
| 1234 |
+
"eval_loss": 2.226450204849243,
|
| 1235 |
+
"eval_model_preparation_time": 0.0056,
|
| 1236 |
+
"eval_runtime": 53.9596,
|
| 1237 |
+
"eval_samples_per_second": 113.122,
|
| 1238 |
+
"eval_steps_per_second": 14.14,
|
| 1239 |
+
"step": 25256
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 88.85017421602788,
|
| 1243 |
+
"grad_norm": 2.065523147583008,
|
| 1244 |
+
"learning_rate": 1.1169284467713787e-07,
|
| 1245 |
+
"loss": 2.152,
|
| 1246 |
+
"step": 25500
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 89.0,
|
| 1250 |
+
"eval_accuracy": 0.9189056356487549,
|
| 1251 |
+
"eval_loss": 2.2227816581726074,
|
| 1252 |
+
"eval_model_preparation_time": 0.0056,
|
| 1253 |
+
"eval_runtime": 54.0621,
|
| 1254 |
+
"eval_samples_per_second": 112.907,
|
| 1255 |
+
"eval_steps_per_second": 14.113,
|
| 1256 |
+
"step": 25543
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 90.0,
|
| 1260 |
+
"eval_accuracy": 0.9185779816513762,
|
| 1261 |
+
"eval_loss": 2.2196154594421387,
|
| 1262 |
+
"eval_model_preparation_time": 0.0056,
|
| 1263 |
+
"eval_runtime": 54.1756,
|
| 1264 |
+
"eval_samples_per_second": 112.671,
|
| 1265 |
+
"eval_steps_per_second": 14.084,
|
| 1266 |
+
"step": 25830
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 90.59233449477352,
|
| 1270 |
+
"grad_norm": 2.453361749649048,
|
| 1271 |
+
"learning_rate": 9.424083769633507e-08,
|
| 1272 |
+
"loss": 2.1462,
|
| 1273 |
+
"step": 26000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 91.0,
|
| 1277 |
+
"eval_accuracy": 0.9182503276539974,
|
| 1278 |
+
"eval_loss": 2.216709613800049,
|
| 1279 |
+
"eval_model_preparation_time": 0.0056,
|
| 1280 |
+
"eval_runtime": 53.7343,
|
| 1281 |
+
"eval_samples_per_second": 113.596,
|
| 1282 |
+
"eval_steps_per_second": 14.199,
|
| 1283 |
+
"step": 26117
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 92.0,
|
| 1287 |
+
"eval_accuracy": 0.9185779816513762,
|
| 1288 |
+
"eval_loss": 2.214052438735962,
|
| 1289 |
+
"eval_model_preparation_time": 0.0056,
|
| 1290 |
+
"eval_runtime": 53.4012,
|
| 1291 |
+
"eval_samples_per_second": 114.305,
|
| 1292 |
+
"eval_steps_per_second": 14.288,
|
| 1293 |
+
"step": 26404
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 92.33449477351917,
|
| 1297 |
+
"grad_norm": 2.680849313735962,
|
| 1298 |
+
"learning_rate": 7.678883071553228e-08,
|
| 1299 |
+
"loss": 2.1384,
|
| 1300 |
+
"step": 26500
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 93.0,
|
| 1304 |
+
"eval_accuracy": 0.9185779816513762,
|
| 1305 |
+
"eval_loss": 2.2115318775177,
|
| 1306 |
+
"eval_model_preparation_time": 0.0056,
|
| 1307 |
+
"eval_runtime": 54.7905,
|
| 1308 |
+
"eval_samples_per_second": 111.406,
|
| 1309 |
+
"eval_steps_per_second": 13.926,
|
| 1310 |
+
"step": 26691
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 94.0,
|
| 1314 |
+
"eval_accuracy": 0.9184141546526867,
|
| 1315 |
+
"eval_loss": 2.209531784057617,
|
| 1316 |
+
"eval_model_preparation_time": 0.0056,
|
| 1317 |
+
"eval_runtime": 53.296,
|
| 1318 |
+
"eval_samples_per_second": 114.53,
|
| 1319 |
+
"eval_steps_per_second": 14.316,
|
| 1320 |
+
"step": 26978
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 94.07665505226481,
|
| 1324 |
+
"grad_norm": 1.967248558998108,
|
| 1325 |
+
"learning_rate": 5.933682373472949e-08,
|
| 1326 |
+
"loss": 2.1353,
|
| 1327 |
+
"step": 27000
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 95.0,
|
| 1331 |
+
"eval_accuracy": 0.9190694626474443,
|
| 1332 |
+
"eval_loss": 2.207920551300049,
|
| 1333 |
+
"eval_model_preparation_time": 0.0056,
|
| 1334 |
+
"eval_runtime": 53.7765,
|
| 1335 |
+
"eval_samples_per_second": 113.507,
|
| 1336 |
+
"eval_steps_per_second": 14.188,
|
| 1337 |
+
"step": 27265
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 95.81881533101046,
|
| 1341 |
+
"grad_norm": 2.0648317337036133,
|
| 1342 |
+
"learning_rate": 4.1884816753926704e-08,
|
| 1343 |
+
"loss": 2.1324,
|
| 1344 |
+
"step": 27500
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 96.0,
|
| 1348 |
+
"eval_accuracy": 0.9190694626474443,
|
| 1349 |
+
"eval_loss": 2.2065632343292236,
|
| 1350 |
+
"eval_model_preparation_time": 0.0056,
|
| 1351 |
+
"eval_runtime": 54.6903,
|
| 1352 |
+
"eval_samples_per_second": 111.61,
|
| 1353 |
+
"eval_steps_per_second": 13.951,
|
| 1354 |
+
"step": 27552
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 97.0,
|
| 1358 |
+
"eval_accuracy": 0.9190694626474443,
|
| 1359 |
+
"eval_loss": 2.2054483890533447,
|
| 1360 |
+
"eval_model_preparation_time": 0.0056,
|
| 1361 |
+
"eval_runtime": 54.5908,
|
| 1362 |
+
"eval_samples_per_second": 111.814,
|
| 1363 |
+
"eval_steps_per_second": 13.977,
|
| 1364 |
+
"step": 27839
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 97.5609756097561,
|
| 1368 |
+
"grad_norm": 2.1506221294403076,
|
| 1369 |
+
"learning_rate": 2.4432809773123906e-08,
|
| 1370 |
+
"loss": 2.1307,
|
| 1371 |
+
"step": 28000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 98.0,
|
| 1375 |
+
"eval_accuracy": 0.9190694626474443,
|
| 1376 |
+
"eval_loss": 2.2046942710876465,
|
| 1377 |
+
"eval_model_preparation_time": 0.0056,
|
| 1378 |
+
"eval_runtime": 54.8026,
|
| 1379 |
+
"eval_samples_per_second": 111.382,
|
| 1380 |
+
"eval_steps_per_second": 13.923,
|
| 1381 |
+
"step": 28126
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 99.0,
|
| 1385 |
+
"eval_accuracy": 0.9192332896461337,
|
| 1386 |
+
"eval_loss": 2.2042276859283447,
|
| 1387 |
+
"eval_model_preparation_time": 0.0056,
|
| 1388 |
+
"eval_runtime": 54.9177,
|
| 1389 |
+
"eval_samples_per_second": 111.148,
|
| 1390 |
+
"eval_steps_per_second": 13.894,
|
| 1391 |
+
"step": 28413
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 99.30313588850174,
|
| 1395 |
+
"grad_norm": 1.9334636926651,
|
| 1396 |
+
"learning_rate": 6.980802792321117e-09,
|
| 1397 |
+
"loss": 2.1278,
|
| 1398 |
+
"step": 28500
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 100.0,
|
| 1402 |
+
"eval_accuracy": 0.9192332896461337,
|
| 1403 |
+
"eval_loss": 2.204108953475952,
|
| 1404 |
+
"eval_model_preparation_time": 0.0056,
|
| 1405 |
+
"eval_runtime": 54.2227,
|
| 1406 |
+
"eval_samples_per_second": 112.573,
|
| 1407 |
+
"eval_steps_per_second": 14.072,
|
| 1408 |
+
"step": 28700
|
| 1409 |
+
}
|
| 1410 |
+
],
|
| 1411 |
+
"logging_steps": 500,
|
| 1412 |
+
"max_steps": 28700,
|
| 1413 |
+
"num_input_tokens_seen": 0,
|
| 1414 |
+
"num_train_epochs": 100,
|
| 1415 |
+
"save_steps": 500,
|
| 1416 |
+
"stateful_callbacks": {
|
| 1417 |
+
"TrainerControl": {
|
| 1418 |
+
"args": {
|
| 1419 |
+
"should_epoch_stop": false,
|
| 1420 |
+
"should_evaluate": false,
|
| 1421 |
+
"should_log": false,
|
| 1422 |
+
"should_save": true,
|
| 1423 |
+
"should_training_stop": true
|
| 1424 |
+
},
|
| 1425 |
+
"attributes": {}
|
| 1426 |
+
}
|
| 1427 |
+
},
|
| 1428 |
+
"total_flos": 7.099490807006331e+19,
|
| 1429 |
+
"train_batch_size": 32,
|
| 1430 |
+
"trial_name": null,
|
| 1431 |
+
"trial_params": null
|
| 1432 |
+
}
|
checkpoint-28700/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:375fecfe3250f0d7d4d281c005412dc987d92419d67a86bffe4969cb248991cd
|
| 3 |
+
size 4731
|
config.json
CHANGED
|
@@ -12,74 +12,73 @@
|
|
| 12 |
"0": "Afghan",
|
| 13 |
"1": "African Wild Dog",
|
| 14 |
"2": "Airedale",
|
| 15 |
-
"3": "American
|
| 16 |
-
"4": "American
|
| 17 |
-
"5": "
|
| 18 |
-
"6": "
|
| 19 |
-
"7": "
|
| 20 |
-
"8": "
|
| 21 |
-
"9": "
|
| 22 |
-
"10": "
|
| 23 |
-
"11": "
|
| 24 |
-
"12": "
|
| 25 |
-
"13": "
|
| 26 |
-
"14": "
|
| 27 |
-
"15": "
|
| 28 |
-
"16": "
|
| 29 |
-
"17": "
|
| 30 |
-
"18": "
|
| 31 |
-
"19": "Bull
|
| 32 |
-
"20": "
|
| 33 |
-
"21": "
|
| 34 |
-
"22": "
|
| 35 |
-
"23": "
|
| 36 |
-
"24": "
|
| 37 |
-
"25": "
|
| 38 |
-
"26": "
|
| 39 |
-
"27": "
|
| 40 |
-
"28": "
|
| 41 |
-
"29": "
|
| 42 |
-
"30": "
|
| 43 |
-
"31": "
|
| 44 |
-
"32": "
|
| 45 |
-
"33": "
|
| 46 |
-
"34": "
|
| 47 |
-
"35": "
|
| 48 |
-
"36": "
|
| 49 |
-
"37": "
|
| 50 |
-
"38": "
|
| 51 |
-
"39": "
|
| 52 |
-
"40": "Great
|
| 53 |
-
"41": "
|
| 54 |
-
"42": "
|
| 55 |
-
"43": "
|
| 56 |
-
"44": "Irish
|
| 57 |
-
"45": "
|
| 58 |
-
"46": "
|
| 59 |
-
"47": "
|
| 60 |
-
"48": "
|
| 61 |
-
"49": "
|
| 62 |
-
"50": "
|
| 63 |
-
"51": "
|
| 64 |
-
"52": "
|
| 65 |
-
"53": "
|
| 66 |
-
"54": "
|
| 67 |
-
"55": "
|
| 68 |
-
"56": "
|
| 69 |
-
"57": "
|
| 70 |
-
"58": "
|
| 71 |
-
"59": "
|
| 72 |
-
"60": "
|
| 73 |
-
"61": "
|
| 74 |
-
"62": "
|
| 75 |
-
"63": "
|
| 76 |
-
"64": "
|
| 77 |
-
"65": "
|
| 78 |
-
"66": "
|
| 79 |
-
"67": "
|
| 80 |
-
"68": "
|
| 81 |
-
"69": "
|
| 82 |
-
"70": "Yorkie"
|
| 83 |
},
|
| 84 |
"image_size": 224,
|
| 85 |
"initializer_range": 0.02,
|
|
@@ -88,74 +87,73 @@
|
|
| 88 |
"Afghan": 0,
|
| 89 |
"African Wild Dog": 1,
|
| 90 |
"Airedale": 2,
|
| 91 |
-
"American
|
| 92 |
-
"American
|
| 93 |
-
"
|
| 94 |
-
"
|
| 95 |
-
"
|
| 96 |
-
"
|
| 97 |
-
"
|
| 98 |
-
"
|
| 99 |
-
"
|
| 100 |
-
"
|
| 101 |
-
"
|
| 102 |
-
"
|
| 103 |
-
"
|
| 104 |
-
"
|
| 105 |
-
"
|
| 106 |
-
"
|
| 107 |
-
"Bull
|
| 108 |
-
"
|
| 109 |
-
"
|
| 110 |
-
"
|
| 111 |
-
"
|
| 112 |
-
"
|
| 113 |
-
"
|
| 114 |
-
"
|
| 115 |
-
"
|
| 116 |
-
"
|
| 117 |
-
"
|
| 118 |
-
"
|
| 119 |
-
"
|
| 120 |
-
"
|
| 121 |
-
"
|
| 122 |
-
"
|
| 123 |
-
"
|
| 124 |
-
"
|
| 125 |
-
"
|
| 126 |
-
"
|
| 127 |
-
"
|
| 128 |
-
"Great
|
| 129 |
-
"
|
| 130 |
-
"
|
| 131 |
-
"
|
| 132 |
-
"Irish
|
| 133 |
-
"
|
| 134 |
-
"
|
| 135 |
-
"
|
| 136 |
-
"
|
| 137 |
-
"
|
| 138 |
-
"
|
| 139 |
-
"
|
| 140 |
-
"
|
| 141 |
-
"
|
| 142 |
-
"
|
| 143 |
-
"
|
| 144 |
-
"
|
| 145 |
-
"
|
| 146 |
-
"
|
| 147 |
-
"
|
| 148 |
-
"
|
| 149 |
-
"
|
| 150 |
-
"
|
| 151 |
-
"
|
| 152 |
-
"
|
| 153 |
-
"
|
| 154 |
-
"
|
| 155 |
-
"
|
| 156 |
-
"
|
| 157 |
-
"
|
| 158 |
-
"Yorkie": 70
|
| 159 |
},
|
| 160 |
"layer_norm_eps": 1e-12,
|
| 161 |
"model_type": "vit",
|
|
|
|
| 12 |
"0": "Afghan",
|
| 13 |
"1": "African Wild Dog",
|
| 14 |
"2": "Airedale",
|
| 15 |
+
"3": "American Hairless",
|
| 16 |
+
"4": "American Spaniel",
|
| 17 |
+
"5": "Basenji",
|
| 18 |
+
"6": "Basset",
|
| 19 |
+
"7": "Beagle",
|
| 20 |
+
"8": "Bearded Collie",
|
| 21 |
+
"9": "Bermaise",
|
| 22 |
+
"10": "Bichon Frise",
|
| 23 |
+
"11": "Blenheim",
|
| 24 |
+
"12": "Bloodhound",
|
| 25 |
+
"13": "Bluetick",
|
| 26 |
+
"14": "Border Collie",
|
| 27 |
+
"15": "Borzoi",
|
| 28 |
+
"16": "Boston Terrier",
|
| 29 |
+
"17": "Boxer",
|
| 30 |
+
"18": "Bull Mastiff",
|
| 31 |
+
"19": "Bull Terrier",
|
| 32 |
+
"20": "Bulldog",
|
| 33 |
+
"21": "Cairn",
|
| 34 |
+
"22": "Chihuahua",
|
| 35 |
+
"23": "Chinese Crested",
|
| 36 |
+
"24": "Chow",
|
| 37 |
+
"25": "Clumber",
|
| 38 |
+
"26": "Cockapoo",
|
| 39 |
+
"27": "Cocker",
|
| 40 |
+
"28": "Collie",
|
| 41 |
+
"29": "Corgi",
|
| 42 |
+
"30": "Coyote",
|
| 43 |
+
"31": "Dalmation",
|
| 44 |
+
"32": "Dhole",
|
| 45 |
+
"33": "Dingo",
|
| 46 |
+
"34": "Doberman",
|
| 47 |
+
"35": "Elk Hound",
|
| 48 |
+
"36": "French Bulldog",
|
| 49 |
+
"37": "German Sheperd",
|
| 50 |
+
"38": "Golden Retriever",
|
| 51 |
+
"39": "Great Dane",
|
| 52 |
+
"40": "Great Perenees",
|
| 53 |
+
"41": "Greyhound",
|
| 54 |
+
"42": "Groenendael",
|
| 55 |
+
"43": "Irish Spaniel",
|
| 56 |
+
"44": "Irish Wolfhound",
|
| 57 |
+
"45": "Japanese Spaniel",
|
| 58 |
+
"46": "Komondor",
|
| 59 |
+
"47": "Labradoodle",
|
| 60 |
+
"48": "Labrador",
|
| 61 |
+
"49": "Lhasa",
|
| 62 |
+
"50": "Malinois",
|
| 63 |
+
"51": "Maltese",
|
| 64 |
+
"52": "Mex Hairless",
|
| 65 |
+
"53": "Newfoundland",
|
| 66 |
+
"54": "Pekinese",
|
| 67 |
+
"55": "Pit Bull",
|
| 68 |
+
"56": "Pomeranian",
|
| 69 |
+
"57": "Poodle",
|
| 70 |
+
"58": "Pug",
|
| 71 |
+
"59": "Rhodesian",
|
| 72 |
+
"60": "Rottweiler",
|
| 73 |
+
"61": "Saint Bernard",
|
| 74 |
+
"62": "Schnauzer",
|
| 75 |
+
"63": "Scotch Terrier",
|
| 76 |
+
"64": "Shar_Pei",
|
| 77 |
+
"65": "Shiba Inu",
|
| 78 |
+
"66": "Shih-Tzu",
|
| 79 |
+
"67": "Siberian Husky",
|
| 80 |
+
"68": "Vizsla",
|
| 81 |
+
"69": "Yorkie"
|
|
|
|
| 82 |
},
|
| 83 |
"image_size": 224,
|
| 84 |
"initializer_range": 0.02,
|
|
|
|
| 87 |
"Afghan": 0,
|
| 88 |
"African Wild Dog": 1,
|
| 89 |
"Airedale": 2,
|
| 90 |
+
"American Hairless": 3,
|
| 91 |
+
"American Spaniel": 4,
|
| 92 |
+
"Basenji": 5,
|
| 93 |
+
"Basset": 6,
|
| 94 |
+
"Beagle": 7,
|
| 95 |
+
"Bearded Collie": 8,
|
| 96 |
+
"Bermaise": 9,
|
| 97 |
+
"Bichon Frise": 10,
|
| 98 |
+
"Blenheim": 11,
|
| 99 |
+
"Bloodhound": 12,
|
| 100 |
+
"Bluetick": 13,
|
| 101 |
+
"Border Collie": 14,
|
| 102 |
+
"Borzoi": 15,
|
| 103 |
+
"Boston Terrier": 16,
|
| 104 |
+
"Boxer": 17,
|
| 105 |
+
"Bull Mastiff": 18,
|
| 106 |
+
"Bull Terrier": 19,
|
| 107 |
+
"Bulldog": 20,
|
| 108 |
+
"Cairn": 21,
|
| 109 |
+
"Chihuahua": 22,
|
| 110 |
+
"Chinese Crested": 23,
|
| 111 |
+
"Chow": 24,
|
| 112 |
+
"Clumber": 25,
|
| 113 |
+
"Cockapoo": 26,
|
| 114 |
+
"Cocker": 27,
|
| 115 |
+
"Collie": 28,
|
| 116 |
+
"Corgi": 29,
|
| 117 |
+
"Coyote": 30,
|
| 118 |
+
"Dalmation": 31,
|
| 119 |
+
"Dhole": 32,
|
| 120 |
+
"Dingo": 33,
|
| 121 |
+
"Doberman": 34,
|
| 122 |
+
"Elk Hound": 35,
|
| 123 |
+
"French Bulldog": 36,
|
| 124 |
+
"German Sheperd": 37,
|
| 125 |
+
"Golden Retriever": 38,
|
| 126 |
+
"Great Dane": 39,
|
| 127 |
+
"Great Perenees": 40,
|
| 128 |
+
"Greyhound": 41,
|
| 129 |
+
"Groenendael": 42,
|
| 130 |
+
"Irish Spaniel": 43,
|
| 131 |
+
"Irish Wolfhound": 44,
|
| 132 |
+
"Japanese Spaniel": 45,
|
| 133 |
+
"Komondor": 46,
|
| 134 |
+
"Labradoodle": 47,
|
| 135 |
+
"Labrador": 48,
|
| 136 |
+
"Lhasa": 49,
|
| 137 |
+
"Malinois": 50,
|
| 138 |
+
"Maltese": 51,
|
| 139 |
+
"Mex Hairless": 52,
|
| 140 |
+
"Newfoundland": 53,
|
| 141 |
+
"Pekinese": 54,
|
| 142 |
+
"Pit Bull": 55,
|
| 143 |
+
"Pomeranian": 56,
|
| 144 |
+
"Poodle": 57,
|
| 145 |
+
"Pug": 58,
|
| 146 |
+
"Rhodesian": 59,
|
| 147 |
+
"Rottweiler": 60,
|
| 148 |
+
"Saint Bernard": 61,
|
| 149 |
+
"Schnauzer": 62,
|
| 150 |
+
"Scotch Terrier": 63,
|
| 151 |
+
"Shar_Pei": 64,
|
| 152 |
+
"Shiba Inu": 65,
|
| 153 |
+
"Shih-Tzu": 66,
|
| 154 |
+
"Siberian Husky": 67,
|
| 155 |
+
"Vizsla": 68,
|
| 156 |
+
"Yorkie": 69
|
|
|
|
| 157 |
},
|
| 158 |
"layer_norm_eps": 1e-12,
|
| 159 |
"model_type": "vit",
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fa6f5b1f9c44c3c29d9c4c7c230929790148180797e2a7c74658d6a0569b7889
|
| 3 |
+
size 343433152
|