Upload 8 files
Browse files- config.json +201 -0
- optimizer.pt +3 -0
- preprocessor_config.json +28 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +118 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,201 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "lamnt2008/bicycle_base",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"BeitForImageClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"auxiliary_channels": 256,
|
| 8 |
+
"auxiliary_concat_input": false,
|
| 9 |
+
"auxiliary_loss_weight": 0.4,
|
| 10 |
+
"auxiliary_num_convs": 1,
|
| 11 |
+
"drop_path_rate": 0.1,
|
| 12 |
+
"finetuning_task": "image-classification",
|
| 13 |
+
"hidden_act": "gelu",
|
| 14 |
+
"hidden_dropout_prob": 0.0,
|
| 15 |
+
"hidden_size": 768,
|
| 16 |
+
"id2label": {
|
| 17 |
+
"0": "Ducati-Diavel",
|
| 18 |
+
"1": "Ducati-Monster",
|
| 19 |
+
"10": "HarleyDavidson-Street(BOB,750)",
|
| 20 |
+
"11": "HarleyDavidson-Ultra",
|
| 21 |
+
"12": "Honda-AfricaTwin",
|
| 22 |
+
"13": "Honda-AirBlade",
|
| 23 |
+
"14": "Honda-Blade",
|
| 24 |
+
"15": "Honda-CB(CBF,Fury)",
|
| 25 |
+
"16": "Honda-CBR",
|
| 26 |
+
"17": "Honda-CD(CL)",
|
| 27 |
+
"18": "Honda-CRF",
|
| 28 |
+
"19": "Honda-Cub(Cseries)",
|
| 29 |
+
"2": "Ducati-Multistrada",
|
| 30 |
+
"20": "Honda-Dream(Super_Neos_Yuga)",
|
| 31 |
+
"21": "Honda-Forza(PCX)",
|
| 32 |
+
"22": "Honda-Future",
|
| 33 |
+
"23": "Honda-Giorno(SuperCub)",
|
| 34 |
+
"24": "Honda-GoldWing",
|
| 35 |
+
"25": "Honda-Lead(Elite)",
|
| 36 |
+
"26": "Honda-Magna",
|
| 37 |
+
"27": "Honda-Monkey(MSX)",
|
| 38 |
+
"28": "Honda-Rebel(Phantom,Shadow)",
|
| 39 |
+
"29": "Honda-SH",
|
| 40 |
+
"3": "Ducati-Panigale",
|
| 41 |
+
"30": "Honda-SHMode",
|
| 42 |
+
"31": "Honda-Scoopy",
|
| 43 |
+
"32": "Honda-Spacy(Dio)",
|
| 44 |
+
"33": "Honda-Vario",
|
| 45 |
+
"34": "Honda-Vision(Beat)",
|
| 46 |
+
"35": "Honda-Wave",
|
| 47 |
+
"36": "Honda-WaveRSX(Supra)",
|
| 48 |
+
"37": "Honda-Winner",
|
| 49 |
+
"38": "Piaggio-Beverly",
|
| 50 |
+
"39": "Piaggio-Fly(NRG)",
|
| 51 |
+
"4": "Ducati-Scrambler",
|
| 52 |
+
"40": "Piaggio-Liberty",
|
| 53 |
+
"41": "Piaggio-MP3",
|
| 54 |
+
"42": "Piaggio-Medley",
|
| 55 |
+
"43": "Suzuki-Address",
|
| 56 |
+
"44": "Suzuki-Axelo(Sport,Smash,Shogun)",
|
| 57 |
+
"45": "Suzuki-Burgman",
|
| 58 |
+
"46": "Suzuki-DR(EN,GN)",
|
| 59 |
+
"47": "Suzuki-GD(SV,RV,TU)",
|
| 60 |
+
"48": "Suzuki-GSX",
|
| 61 |
+
"49": "Suzuki-Gixxer",
|
| 62 |
+
"5": "HarleyDavidson-FatBob",
|
| 63 |
+
"50": "Suzuki-Hayate",
|
| 64 |
+
"51": "Suzuki-Impulse",
|
| 65 |
+
"52": "Suzuki-Viva",
|
| 66 |
+
"53": "Sym-Attila",
|
| 67 |
+
"54": "Sym-Elegant",
|
| 68 |
+
"55": "Vespa-946",
|
| 69 |
+
"56": "Vespa-GT(Piaggio-Vespa,GTS)",
|
| 70 |
+
"57": "Vespa-Primavera(LX,LXV,PX)",
|
| 71 |
+
"58": "Vespa-SeiGiorni(GTV)",
|
| 72 |
+
"59": "Vespa-Sprint(S)",
|
| 73 |
+
"6": "HarleyDavidson-FortyEight(Iron)",
|
| 74 |
+
"60": "Yamaha-Acruzo",
|
| 75 |
+
"61": "Yamaha-Exciter",
|
| 76 |
+
"62": "Yamaha-Freego",
|
| 77 |
+
"63": "Yamaha-Grande",
|
| 78 |
+
"64": "Yamaha-Janus(Sporty,X-ride,Gravita)",
|
| 79 |
+
"65": "Yamaha-Jupiter(FI,MX)",
|
| 80 |
+
"66": "Yamaha-MTSeries(TFX)",
|
| 81 |
+
"67": "Yamaha-NVX(Aerox,Cygnus)",
|
| 82 |
+
"68": "Yamaha-Nouvo(Luvias)",
|
| 83 |
+
"69": "Yamaha-Sirius",
|
| 84 |
+
"7": "HarleyDavidson-RoadKing",
|
| 85 |
+
"70": "Yamaha-SiriusFI(Crypton)",
|
| 86 |
+
"71": "Yamaha-Tracer(FJR)",
|
| 87 |
+
"72": "Yamaha-Virago(VSeries,XSR)",
|
| 88 |
+
"73": "Yamaha-YZF(YZR,FZ,TFX)",
|
| 89 |
+
"8": "HarleyDavidson-Softail",
|
| 90 |
+
"9": "HarleyDavidson-Sportster"
|
| 91 |
+
},
|
| 92 |
+
"image_size": 224,
|
| 93 |
+
"initializer_range": 0.02,
|
| 94 |
+
"intermediate_size": 3072,
|
| 95 |
+
"label2id": {
|
| 96 |
+
"Ducati-Diavel": "0",
|
| 97 |
+
"Ducati-Monster": "1",
|
| 98 |
+
"Ducati-Multistrada": "2",
|
| 99 |
+
"Ducati-Panigale": "3",
|
| 100 |
+
"Ducati-Scrambler": "4",
|
| 101 |
+
"HarleyDavidson-FatBob": "5",
|
| 102 |
+
"HarleyDavidson-FortyEight(Iron)": "6",
|
| 103 |
+
"HarleyDavidson-RoadKing": "7",
|
| 104 |
+
"HarleyDavidson-Softail": "8",
|
| 105 |
+
"HarleyDavidson-Sportster": "9",
|
| 106 |
+
"HarleyDavidson-Street(BOB,750)": "10",
|
| 107 |
+
"HarleyDavidson-Ultra": "11",
|
| 108 |
+
"Honda-AfricaTwin": "12",
|
| 109 |
+
"Honda-AirBlade": "13",
|
| 110 |
+
"Honda-Blade": "14",
|
| 111 |
+
"Honda-CB(CBF,Fury)": "15",
|
| 112 |
+
"Honda-CBR": "16",
|
| 113 |
+
"Honda-CD(CL)": "17",
|
| 114 |
+
"Honda-CRF": "18",
|
| 115 |
+
"Honda-Cub(Cseries)": "19",
|
| 116 |
+
"Honda-Dream(Super_Neos_Yuga)": "20",
|
| 117 |
+
"Honda-Forza(PCX)": "21",
|
| 118 |
+
"Honda-Future": "22",
|
| 119 |
+
"Honda-Giorno(SuperCub)": "23",
|
| 120 |
+
"Honda-GoldWing": "24",
|
| 121 |
+
"Honda-Lead(Elite)": "25",
|
| 122 |
+
"Honda-Magna": "26",
|
| 123 |
+
"Honda-Monkey(MSX)": "27",
|
| 124 |
+
"Honda-Rebel(Phantom,Shadow)": "28",
|
| 125 |
+
"Honda-SH": "29",
|
| 126 |
+
"Honda-SHMode": "30",
|
| 127 |
+
"Honda-Scoopy": "31",
|
| 128 |
+
"Honda-Spacy(Dio)": "32",
|
| 129 |
+
"Honda-Vario": "33",
|
| 130 |
+
"Honda-Vision(Beat)": "34",
|
| 131 |
+
"Honda-Wave": "35",
|
| 132 |
+
"Honda-WaveRSX(Supra)": "36",
|
| 133 |
+
"Honda-Winner": "37",
|
| 134 |
+
"Piaggio-Beverly": "38",
|
| 135 |
+
"Piaggio-Fly(NRG)": "39",
|
| 136 |
+
"Piaggio-Liberty": "40",
|
| 137 |
+
"Piaggio-MP3": "41",
|
| 138 |
+
"Piaggio-Medley": "42",
|
| 139 |
+
"Suzuki-Address": "43",
|
| 140 |
+
"Suzuki-Axelo(Sport,Smash,Shogun)": "44",
|
| 141 |
+
"Suzuki-Burgman": "45",
|
| 142 |
+
"Suzuki-DR(EN,GN)": "46",
|
| 143 |
+
"Suzuki-GD(SV,RV,TU)": "47",
|
| 144 |
+
"Suzuki-GSX": "48",
|
| 145 |
+
"Suzuki-Gixxer": "49",
|
| 146 |
+
"Suzuki-Hayate": "50",
|
| 147 |
+
"Suzuki-Impulse": "51",
|
| 148 |
+
"Suzuki-Viva": "52",
|
| 149 |
+
"Sym-Attila": "53",
|
| 150 |
+
"Sym-Elegant": "54",
|
| 151 |
+
"Vespa-946": "55",
|
| 152 |
+
"Vespa-GT(Piaggio-Vespa,GTS)": "56",
|
| 153 |
+
"Vespa-Primavera(LX,LXV,PX)": "57",
|
| 154 |
+
"Vespa-SeiGiorni(GTV)": "58",
|
| 155 |
+
"Vespa-Sprint(S)": "59",
|
| 156 |
+
"Yamaha-Acruzo": "60",
|
| 157 |
+
"Yamaha-Exciter": "61",
|
| 158 |
+
"Yamaha-Freego": "62",
|
| 159 |
+
"Yamaha-Grande": "63",
|
| 160 |
+
"Yamaha-Janus(Sporty,X-ride,Gravita)": "64",
|
| 161 |
+
"Yamaha-Jupiter(FI,MX)": "65",
|
| 162 |
+
"Yamaha-MTSeries(TFX)": "66",
|
| 163 |
+
"Yamaha-NVX(Aerox,Cygnus)": "67",
|
| 164 |
+
"Yamaha-Nouvo(Luvias)": "68",
|
| 165 |
+
"Yamaha-Sirius": "69",
|
| 166 |
+
"Yamaha-SiriusFI(Crypton)": "70",
|
| 167 |
+
"Yamaha-Tracer(FJR)": "71",
|
| 168 |
+
"Yamaha-Virago(VSeries,XSR)": "72",
|
| 169 |
+
"Yamaha-YZF(YZR,FZ,TFX)": "73"
|
| 170 |
+
},
|
| 171 |
+
"layer_norm_eps": 1e-12,
|
| 172 |
+
"layer_scale_init_value": 0.1,
|
| 173 |
+
"model_type": "beit",
|
| 174 |
+
"num_attention_heads": 12,
|
| 175 |
+
"num_channels": 3,
|
| 176 |
+
"num_hidden_layers": 12,
|
| 177 |
+
"out_indices": [
|
| 178 |
+
3,
|
| 179 |
+
5,
|
| 180 |
+
7,
|
| 181 |
+
11
|
| 182 |
+
],
|
| 183 |
+
"patch_size": 16,
|
| 184 |
+
"pool_scales": [
|
| 185 |
+
1,
|
| 186 |
+
2,
|
| 187 |
+
3,
|
| 188 |
+
6
|
| 189 |
+
],
|
| 190 |
+
"problem_type": "single_label_classification",
|
| 191 |
+
"semantic_loss_ignore_index": 255,
|
| 192 |
+
"torch_dtype": "float32",
|
| 193 |
+
"transformers_version": "4.28.0.dev0",
|
| 194 |
+
"use_absolute_position_embeddings": false,
|
| 195 |
+
"use_auxiliary_head": true,
|
| 196 |
+
"use_mask_token": false,
|
| 197 |
+
"use_mean_pooling": true,
|
| 198 |
+
"use_relative_position_bias": true,
|
| 199 |
+
"use_shared_relative_position_bias": false,
|
| 200 |
+
"vocab_size": 8192
|
| 201 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c169775c366010e4c5eaffadb34df1ccf4d9628fd6397bae1c7164e585adaf43
|
| 3 |
+
size 686681861
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"crop_size": {
|
| 3 |
+
"height": 224,
|
| 4 |
+
"width": 224
|
| 5 |
+
},
|
| 6 |
+
"do_center_crop": false,
|
| 7 |
+
"do_normalize": true,
|
| 8 |
+
"do_reduce_labels": false,
|
| 9 |
+
"do_rescale": true,
|
| 10 |
+
"do_resize": true,
|
| 11 |
+
"image_mean": [
|
| 12 |
+
0.5,
|
| 13 |
+
0.5,
|
| 14 |
+
0.5
|
| 15 |
+
],
|
| 16 |
+
"image_processor_type": "BeitImageProcessor",
|
| 17 |
+
"image_std": [
|
| 18 |
+
0.5,
|
| 19 |
+
0.5,
|
| 20 |
+
0.5
|
| 21 |
+
],
|
| 22 |
+
"resample": 2,
|
| 23 |
+
"rescale_factor": 0.00392156862745098,
|
| 24 |
+
"size": {
|
| 25 |
+
"height": 224,
|
| 26 |
+
"width": 224
|
| 27 |
+
}
|
| 28 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:93ab5ed813a4061484bf42a72dcb86ac4bd68e8e621bd367d2d73de57acf6916
|
| 3 |
+
size 347081849
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c800d296d274356dc7e1009bd7b9538d79376fb6a4db1bb2d6965f7da8688b3b
|
| 3 |
+
size 14639
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b91afddbb850d225c9270f7a06401619e44b4668e24d6085fb2561753e178a88
|
| 3 |
+
size 627
|
trainer_state.json
ADDED
|
@@ -0,0 +1,118 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.763363091874788,
|
| 5 |
+
"global_step": 51000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.34,
|
| 12 |
+
"learning_rate": 9.915244660413606e-06,
|
| 13 |
+
"loss": 3.3726,
|
| 14 |
+
"step": 3000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.68,
|
| 18 |
+
"learning_rate": 9.830489320827214e-06,
|
| 19 |
+
"loss": 2.1283,
|
| 20 |
+
"step": 6000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.02,
|
| 24 |
+
"learning_rate": 9.745733981240819e-06,
|
| 25 |
+
"loss": 1.8301,
|
| 26 |
+
"step": 9000
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 1.36,
|
| 30 |
+
"learning_rate": 9.660978641654425e-06,
|
| 31 |
+
"loss": 1.6611,
|
| 32 |
+
"step": 12000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 1.7,
|
| 36 |
+
"learning_rate": 9.57622330206803e-06,
|
| 37 |
+
"loss": 1.6012,
|
| 38 |
+
"step": 15000
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 2.03,
|
| 42 |
+
"learning_rate": 9.491467962481637e-06,
|
| 43 |
+
"loss": 1.5255,
|
| 44 |
+
"step": 18000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 2.37,
|
| 48 |
+
"learning_rate": 9.406712622895243e-06,
|
| 49 |
+
"loss": 1.4598,
|
| 50 |
+
"step": 21000
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 2.71,
|
| 54 |
+
"learning_rate": 9.321957283308849e-06,
|
| 55 |
+
"loss": 1.4156,
|
| 56 |
+
"step": 24000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 3.05,
|
| 60 |
+
"learning_rate": 9.237201943722456e-06,
|
| 61 |
+
"loss": 1.3908,
|
| 62 |
+
"step": 27000
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 3.39,
|
| 66 |
+
"learning_rate": 9.152446604136061e-06,
|
| 67 |
+
"loss": 1.3406,
|
| 68 |
+
"step": 30000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 3.73,
|
| 72 |
+
"learning_rate": 9.067691264549667e-06,
|
| 73 |
+
"loss": 1.3277,
|
| 74 |
+
"step": 33000
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 4.07,
|
| 78 |
+
"learning_rate": 8.982935924963272e-06,
|
| 79 |
+
"loss": 1.3095,
|
| 80 |
+
"step": 36000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 4.41,
|
| 84 |
+
"learning_rate": 8.89818058537688e-06,
|
| 85 |
+
"loss": 1.253,
|
| 86 |
+
"step": 39000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 4.75,
|
| 90 |
+
"learning_rate": 8.813425245790485e-06,
|
| 91 |
+
"loss": 1.2553,
|
| 92 |
+
"step": 42000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 5.09,
|
| 96 |
+
"learning_rate": 8.72866990620409e-06,
|
| 97 |
+
"loss": 1.2177,
|
| 98 |
+
"step": 45000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 5.42,
|
| 102 |
+
"learning_rate": 8.643914566617696e-06,
|
| 103 |
+
"loss": 1.198,
|
| 104 |
+
"step": 48000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 5.76,
|
| 108 |
+
"learning_rate": 8.559159227031304e-06,
|
| 109 |
+
"loss": 1.2163,
|
| 110 |
+
"step": 51000
|
| 111 |
+
}
|
| 112 |
+
],
|
| 113 |
+
"max_steps": 353960,
|
| 114 |
+
"num_train_epochs": 40,
|
| 115 |
+
"total_flos": 3.162362109935616e+19,
|
| 116 |
+
"trial_name": null,
|
| 117 |
+
"trial_params": null
|
| 118 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:02ae9bc04ec11a5436f1c2b5d27a074129d73b2e6339507f9d2f4304a8062bf8
|
| 3 |
+
size 3643
|