Upload folder using huggingface_hub
Browse files- checkpoint-4220/config.json +228 -0
- checkpoint-4220/model.safetensors +3 -0
- checkpoint-4220/optimizer.pt +3 -0
- checkpoint-4220/preprocessor_config.json +22 -0
- checkpoint-4220/rng_state.pth +3 -0
- checkpoint-4220/scheduler.pt +3 -0
- checkpoint-4220/trainer_state.json +122 -0
- checkpoint-4220/training_args.bin +3 -0
- model.safetensors +1 -1
- training_args.bin +1 -1
checkpoint-4220/config.json
ADDED
|
@@ -0,0 +1,228 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "dima806/fruit_100_types_image_detection",
|
| 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": "abiu",
|
| 13 |
+
"1": "acai",
|
| 14 |
+
"2": "acerola",
|
| 15 |
+
"3": "ackee",
|
| 16 |
+
"4": "ambarella",
|
| 17 |
+
"5": "apple",
|
| 18 |
+
"6": "apricot",
|
| 19 |
+
"7": "avocado",
|
| 20 |
+
"8": "banana",
|
| 21 |
+
"9": "barbadine",
|
| 22 |
+
"10": "barberry",
|
| 23 |
+
"11": "betel_nut",
|
| 24 |
+
"12": "bitter_gourd",
|
| 25 |
+
"13": "black_berry",
|
| 26 |
+
"14": "black_mullberry",
|
| 27 |
+
"15": "brazil_nut",
|
| 28 |
+
"16": "camu_camu",
|
| 29 |
+
"17": "cashew",
|
| 30 |
+
"18": "cempedak",
|
| 31 |
+
"19": "chenet",
|
| 32 |
+
"20": "cherimoya",
|
| 33 |
+
"21": "chico",
|
| 34 |
+
"22": "chokeberry",
|
| 35 |
+
"23": "cluster_fig",
|
| 36 |
+
"24": "coconut",
|
| 37 |
+
"25": "corn_kernel",
|
| 38 |
+
"26": "cranberry",
|
| 39 |
+
"27": "cupuacu",
|
| 40 |
+
"28": "custard_apple",
|
| 41 |
+
"29": "damson",
|
| 42 |
+
"30": "dewberry",
|
| 43 |
+
"31": "dragonfruit",
|
| 44 |
+
"32": "durian",
|
| 45 |
+
"33": "eggplant",
|
| 46 |
+
"34": "elderberry",
|
| 47 |
+
"35": "emblic",
|
| 48 |
+
"36": "feijoa",
|
| 49 |
+
"37": "fig",
|
| 50 |
+
"38": "finger_lime",
|
| 51 |
+
"39": "gooseberry",
|
| 52 |
+
"40": "goumi",
|
| 53 |
+
"41": "grape",
|
| 54 |
+
"42": "grapefruit",
|
| 55 |
+
"43": "greengage",
|
| 56 |
+
"44": "grenadilla",
|
| 57 |
+
"45": "guava",
|
| 58 |
+
"46": "hard_kiwi",
|
| 59 |
+
"47": "hawthorn",
|
| 60 |
+
"48": "hog_plum",
|
| 61 |
+
"49": "horned_melon",
|
| 62 |
+
"50": "indian_strawberry",
|
| 63 |
+
"51": "jaboticaba",
|
| 64 |
+
"52": "jackfruit",
|
| 65 |
+
"53": "jalapeno",
|
| 66 |
+
"54": "jamaica_cherry",
|
| 67 |
+
"55": "jambul",
|
| 68 |
+
"56": "jocote",
|
| 69 |
+
"57": "jujube",
|
| 70 |
+
"58": "kaffir_lime",
|
| 71 |
+
"59": "kumquat",
|
| 72 |
+
"60": "lablab",
|
| 73 |
+
"61": "langsat",
|
| 74 |
+
"62": "longan",
|
| 75 |
+
"63": "mabolo",
|
| 76 |
+
"64": "malay_apple",
|
| 77 |
+
"65": "mandarine",
|
| 78 |
+
"66": "mango",
|
| 79 |
+
"67": "mangosteen",
|
| 80 |
+
"68": "medlar",
|
| 81 |
+
"69": "mock_strawberry",
|
| 82 |
+
"70": "morinda",
|
| 83 |
+
"71": "mountain_soursop",
|
| 84 |
+
"72": "oil_palm",
|
| 85 |
+
"73": "olive",
|
| 86 |
+
"74": "otaheite_apple",
|
| 87 |
+
"75": "papaya",
|
| 88 |
+
"76": "passion_fruit",
|
| 89 |
+
"77": "pawpaw",
|
| 90 |
+
"78": "pea",
|
| 91 |
+
"79": "pineapple",
|
| 92 |
+
"80": "plumcot",
|
| 93 |
+
"81": "pomegranate",
|
| 94 |
+
"82": "prikly_pear",
|
| 95 |
+
"83": "quince",
|
| 96 |
+
"84": "rambutan",
|
| 97 |
+
"85": "raspberry",
|
| 98 |
+
"86": "redcurrant",
|
| 99 |
+
"87": "rose_hip",
|
| 100 |
+
"88": "rose_leaf_bramble",
|
| 101 |
+
"89": "salak",
|
| 102 |
+
"90": "santol",
|
| 103 |
+
"91": "sapodilla",
|
| 104 |
+
"92": "sea_buckthorn",
|
| 105 |
+
"93": "strawberry_guava",
|
| 106 |
+
"94": "sugar_apple",
|
| 107 |
+
"95": "taxus_baccata",
|
| 108 |
+
"96": "ugli_fruit",
|
| 109 |
+
"97": "white_currant",
|
| 110 |
+
"98": "yali_pear",
|
| 111 |
+
"99": "yellow_plum"
|
| 112 |
+
},
|
| 113 |
+
"image_size": 224,
|
| 114 |
+
"initializer_range": 0.02,
|
| 115 |
+
"intermediate_size": 3072,
|
| 116 |
+
"label2id": {
|
| 117 |
+
"abiu": 0,
|
| 118 |
+
"acai": 1,
|
| 119 |
+
"acerola": 2,
|
| 120 |
+
"ackee": 3,
|
| 121 |
+
"ambarella": 4,
|
| 122 |
+
"apple": 5,
|
| 123 |
+
"apricot": 6,
|
| 124 |
+
"avocado": 7,
|
| 125 |
+
"banana": 8,
|
| 126 |
+
"barbadine": 9,
|
| 127 |
+
"barberry": 10,
|
| 128 |
+
"betel_nut": 11,
|
| 129 |
+
"bitter_gourd": 12,
|
| 130 |
+
"black_berry": 13,
|
| 131 |
+
"black_mullberry": 14,
|
| 132 |
+
"brazil_nut": 15,
|
| 133 |
+
"camu_camu": 16,
|
| 134 |
+
"cashew": 17,
|
| 135 |
+
"cempedak": 18,
|
| 136 |
+
"chenet": 19,
|
| 137 |
+
"cherimoya": 20,
|
| 138 |
+
"chico": 21,
|
| 139 |
+
"chokeberry": 22,
|
| 140 |
+
"cluster_fig": 23,
|
| 141 |
+
"coconut": 24,
|
| 142 |
+
"corn_kernel": 25,
|
| 143 |
+
"cranberry": 26,
|
| 144 |
+
"cupuacu": 27,
|
| 145 |
+
"custard_apple": 28,
|
| 146 |
+
"damson": 29,
|
| 147 |
+
"dewberry": 30,
|
| 148 |
+
"dragonfruit": 31,
|
| 149 |
+
"durian": 32,
|
| 150 |
+
"eggplant": 33,
|
| 151 |
+
"elderberry": 34,
|
| 152 |
+
"emblic": 35,
|
| 153 |
+
"feijoa": 36,
|
| 154 |
+
"fig": 37,
|
| 155 |
+
"finger_lime": 38,
|
| 156 |
+
"gooseberry": 39,
|
| 157 |
+
"goumi": 40,
|
| 158 |
+
"grape": 41,
|
| 159 |
+
"grapefruit": 42,
|
| 160 |
+
"greengage": 43,
|
| 161 |
+
"grenadilla": 44,
|
| 162 |
+
"guava": 45,
|
| 163 |
+
"hard_kiwi": 46,
|
| 164 |
+
"hawthorn": 47,
|
| 165 |
+
"hog_plum": 48,
|
| 166 |
+
"horned_melon": 49,
|
| 167 |
+
"indian_strawberry": 50,
|
| 168 |
+
"jaboticaba": 51,
|
| 169 |
+
"jackfruit": 52,
|
| 170 |
+
"jalapeno": 53,
|
| 171 |
+
"jamaica_cherry": 54,
|
| 172 |
+
"jambul": 55,
|
| 173 |
+
"jocote": 56,
|
| 174 |
+
"jujube": 57,
|
| 175 |
+
"kaffir_lime": 58,
|
| 176 |
+
"kumquat": 59,
|
| 177 |
+
"lablab": 60,
|
| 178 |
+
"langsat": 61,
|
| 179 |
+
"longan": 62,
|
| 180 |
+
"mabolo": 63,
|
| 181 |
+
"malay_apple": 64,
|
| 182 |
+
"mandarine": 65,
|
| 183 |
+
"mango": 66,
|
| 184 |
+
"mangosteen": 67,
|
| 185 |
+
"medlar": 68,
|
| 186 |
+
"mock_strawberry": 69,
|
| 187 |
+
"morinda": 70,
|
| 188 |
+
"mountain_soursop": 71,
|
| 189 |
+
"oil_palm": 72,
|
| 190 |
+
"olive": 73,
|
| 191 |
+
"otaheite_apple": 74,
|
| 192 |
+
"papaya": 75,
|
| 193 |
+
"passion_fruit": 76,
|
| 194 |
+
"pawpaw": 77,
|
| 195 |
+
"pea": 78,
|
| 196 |
+
"pineapple": 79,
|
| 197 |
+
"plumcot": 80,
|
| 198 |
+
"pomegranate": 81,
|
| 199 |
+
"prikly_pear": 82,
|
| 200 |
+
"quince": 83,
|
| 201 |
+
"rambutan": 84,
|
| 202 |
+
"raspberry": 85,
|
| 203 |
+
"redcurrant": 86,
|
| 204 |
+
"rose_hip": 87,
|
| 205 |
+
"rose_leaf_bramble": 88,
|
| 206 |
+
"salak": 89,
|
| 207 |
+
"santol": 90,
|
| 208 |
+
"sapodilla": 91,
|
| 209 |
+
"sea_buckthorn": 92,
|
| 210 |
+
"strawberry_guava": 93,
|
| 211 |
+
"sugar_apple": 94,
|
| 212 |
+
"taxus_baccata": 95,
|
| 213 |
+
"ugli_fruit": 96,
|
| 214 |
+
"white_currant": 97,
|
| 215 |
+
"yali_pear": 98,
|
| 216 |
+
"yellow_plum": 99
|
| 217 |
+
},
|
| 218 |
+
"layer_norm_eps": 1e-12,
|
| 219 |
+
"model_type": "vit",
|
| 220 |
+
"num_attention_heads": 12,
|
| 221 |
+
"num_channels": 3,
|
| 222 |
+
"num_hidden_layers": 12,
|
| 223 |
+
"patch_size": 16,
|
| 224 |
+
"problem_type": "single_label_classification",
|
| 225 |
+
"qkv_bias": true,
|
| 226 |
+
"torch_dtype": "float32",
|
| 227 |
+
"transformers_version": "4.38.2"
|
| 228 |
+
}
|
checkpoint-4220/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4ade855defe46783b7c9ca8a606fb72248a37b021769cbc025fe2953d9474cdc
|
| 3 |
+
size 343525432
|
checkpoint-4220/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:23537072bf8cfa7b91c51314b2da7bf5d9954e8a7e8c6b3e6b0bf6ccfc2cfd41
|
| 3 |
+
size 687171333
|
checkpoint-4220/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-4220/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:da8c74bbcbdc41c3fb5277ac43bd4728c9d71bc5143c0676c04da75c211c1477
|
| 3 |
+
size 14575
|
checkpoint-4220/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:714839a7eaee9eac22146fec65132eecd45ea328a4c5692db2bfb56d0f6673e2
|
| 3 |
+
size 627
|
checkpoint-4220/trainer_state.json
ADDED
|
@@ -0,0 +1,122 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.6452902555465698,
|
| 3 |
+
"best_model_checkpoint": "fruit_100_types_image_detection/checkpoint-4220",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4220,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.59,
|
| 13 |
+
"grad_norm": 2.6183602809906006,
|
| 14 |
+
"learning_rate": 1.7841726618705034e-06,
|
| 15 |
+
"loss": 1.8505,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_accuracy": 0.8403333333333334,
|
| 21 |
+
"eval_loss": 1.8119271993637085,
|
| 22 |
+
"eval_runtime": 167.1143,
|
| 23 |
+
"eval_samples_per_second": 107.711,
|
| 24 |
+
"eval_steps_per_second": 6.732,
|
| 25 |
+
"step": 844
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 1.18,
|
| 29 |
+
"grad_norm": 3.343961000442505,
|
| 30 |
+
"learning_rate": 1.5443645083932852e-06,
|
| 31 |
+
"loss": 1.7915,
|
| 32 |
+
"step": 1000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 1.78,
|
| 36 |
+
"grad_norm": 3.8417131900787354,
|
| 37 |
+
"learning_rate": 1.3045563549160672e-06,
|
| 38 |
+
"loss": 1.7335,
|
| 39 |
+
"step": 1500
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 2.0,
|
| 43 |
+
"eval_accuracy": 0.8433333333333334,
|
| 44 |
+
"eval_loss": 1.7391046285629272,
|
| 45 |
+
"eval_runtime": 169.3646,
|
| 46 |
+
"eval_samples_per_second": 106.28,
|
| 47 |
+
"eval_steps_per_second": 6.642,
|
| 48 |
+
"step": 1688
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 2.37,
|
| 52 |
+
"grad_norm": 2.7001466751098633,
|
| 53 |
+
"learning_rate": 1.0647482014388488e-06,
|
| 54 |
+
"loss": 1.7029,
|
| 55 |
+
"step": 2000
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 2.96,
|
| 59 |
+
"grad_norm": 3.4065492153167725,
|
| 60 |
+
"learning_rate": 8.249400479616306e-07,
|
| 61 |
+
"loss": 1.6636,
|
| 62 |
+
"step": 2500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 3.0,
|
| 66 |
+
"eval_accuracy": 0.8451666666666666,
|
| 67 |
+
"eval_loss": 1.6863653659820557,
|
| 68 |
+
"eval_runtime": 183.0164,
|
| 69 |
+
"eval_samples_per_second": 98.352,
|
| 70 |
+
"eval_steps_per_second": 6.147,
|
| 71 |
+
"step": 2532
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 3.55,
|
| 75 |
+
"grad_norm": 2.9417073726654053,
|
| 76 |
+
"learning_rate": 5.851318944844125e-07,
|
| 77 |
+
"loss": 1.6379,
|
| 78 |
+
"step": 3000
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 4.0,
|
| 82 |
+
"eval_accuracy": 0.8458333333333333,
|
| 83 |
+
"eval_loss": 1.655003309249878,
|
| 84 |
+
"eval_runtime": 182.964,
|
| 85 |
+
"eval_samples_per_second": 98.38,
|
| 86 |
+
"eval_steps_per_second": 6.149,
|
| 87 |
+
"step": 3376
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 4.15,
|
| 91 |
+
"grad_norm": 3.41286563873291,
|
| 92 |
+
"learning_rate": 3.4532374100719426e-07,
|
| 93 |
+
"loss": 1.617,
|
| 94 |
+
"step": 3500
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 4.74,
|
| 98 |
+
"grad_norm": 3.1814427375793457,
|
| 99 |
+
"learning_rate": 1.0551558752997601e-07,
|
| 100 |
+
"loss": 1.6055,
|
| 101 |
+
"step": 4000
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 5.0,
|
| 105 |
+
"eval_accuracy": 0.8466666666666667,
|
| 106 |
+
"eval_loss": 1.6452902555465698,
|
| 107 |
+
"eval_runtime": 165.9895,
|
| 108 |
+
"eval_samples_per_second": 108.441,
|
| 109 |
+
"eval_steps_per_second": 6.778,
|
| 110 |
+
"step": 4220
|
| 111 |
+
}
|
| 112 |
+
],
|
| 113 |
+
"logging_steps": 500,
|
| 114 |
+
"max_steps": 4220,
|
| 115 |
+
"num_input_tokens_seen": 0,
|
| 116 |
+
"num_train_epochs": 5,
|
| 117 |
+
"save_steps": 500,
|
| 118 |
+
"total_flos": 1.047060731179008e+19,
|
| 119 |
+
"train_batch_size": 32,
|
| 120 |
+
"trial_name": null,
|
| 121 |
+
"trial_params": null
|
| 122 |
+
}
|
checkpoint-4220/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:491f27a549b7d56714436a3db1387752a569ea6a924c8f46da6769e33fa4ade0
|
| 3 |
+
size 4411
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 343525432
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4ade855defe46783b7c9ca8a606fb72248a37b021769cbc025fe2953d9474cdc
|
| 3 |
size 343525432
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4411
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:491f27a549b7d56714436a3db1387752a569ea6a924c8f46da6769e33fa4ade0
|
| 3 |
size 4411
|