Upload 8 files
Browse files- config.json +39 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- preprocessor_config.json +26 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +253 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "MCG-NJU/videomae-base",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"VideoMAEForVideoClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.2,
|
| 7 |
+
"decoder_hidden_size": 384,
|
| 8 |
+
"decoder_intermediate_size": 1536,
|
| 9 |
+
"decoder_num_attention_heads": 6,
|
| 10 |
+
"decoder_num_hidden_layers": 4,
|
| 11 |
+
"hidden_act": "gelu",
|
| 12 |
+
"hidden_dropout_prob": 0.2,
|
| 13 |
+
"hidden_size": 768,
|
| 14 |
+
"id2label": {
|
| 15 |
+
"0": "real",
|
| 16 |
+
"1": "fake"
|
| 17 |
+
},
|
| 18 |
+
"image_size": 224,
|
| 19 |
+
"initializer_range": 0.02,
|
| 20 |
+
"intermediate_size": 3072,
|
| 21 |
+
"label2id": {
|
| 22 |
+
"fake": 1,
|
| 23 |
+
"real": 0
|
| 24 |
+
},
|
| 25 |
+
"layer_norm_eps": 1e-12,
|
| 26 |
+
"model_type": "videomae",
|
| 27 |
+
"norm_pix_loss": true,
|
| 28 |
+
"num_attention_heads": 12,
|
| 29 |
+
"num_channels": 3,
|
| 30 |
+
"num_frames": 16,
|
| 31 |
+
"num_hidden_layers": 12,
|
| 32 |
+
"patch_size": 16,
|
| 33 |
+
"problem_type": "single_label_classification",
|
| 34 |
+
"qkv_bias": true,
|
| 35 |
+
"torch_dtype": "float32",
|
| 36 |
+
"transformers_version": "4.45.2",
|
| 37 |
+
"tubelet_size": 2,
|
| 38 |
+
"use_mean_pooling": false
|
| 39 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7676545d09a8e011ccf410d74d3a57d7a3cc0aee949c255d306d698ca2aa82d
|
| 3 |
+
size 344937368
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1d7a50889a0732a3b0689fc4c01a04fc35e04cc0d2200885db7b26873b2386a0
|
| 3 |
+
size 689986059
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"crop_size": {
|
| 3 |
+
"height": 224,
|
| 4 |
+
"width": 224
|
| 5 |
+
},
|
| 6 |
+
"do_center_crop": true,
|
| 7 |
+
"do_normalize": true,
|
| 8 |
+
"do_rescale": true,
|
| 9 |
+
"do_resize": true,
|
| 10 |
+
"image_mean": [
|
| 11 |
+
0.485,
|
| 12 |
+
0.456,
|
| 13 |
+
0.406
|
| 14 |
+
],
|
| 15 |
+
"image_processor_type": "VideoMAEImageProcessor",
|
| 16 |
+
"image_std": [
|
| 17 |
+
0.229,
|
| 18 |
+
0.224,
|
| 19 |
+
0.225
|
| 20 |
+
],
|
| 21 |
+
"resample": 2,
|
| 22 |
+
"rescale_factor": 0.00392156862745098,
|
| 23 |
+
"size": {
|
| 24 |
+
"shortest_edge": 224
|
| 25 |
+
}
|
| 26 |
+
}
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8acc7552f0381c1c69ec889c7cf60ffa7491bac598edceb8a09350334e5cd265
|
| 3 |
+
size 14645
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7624e1c1854cde810c826a1d33b122dcdab39ef35afff70319167f7aacff63fd
|
| 3 |
+
size 1465
|
trainer_state.json
ADDED
|
@@ -0,0 +1,253 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.4863013698630137,
|
| 3 |
+
"best_model_checkpoint": "deepfake-video-model-balanced/checkpoint-100",
|
| 4 |
+
"epoch": 1.7241379310344827,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 250,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.06896551724137931,
|
| 13 |
+
"grad_norm": 11.416295051574707,
|
| 14 |
+
"learning_rate": 6.896551724137932e-06,
|
| 15 |
+
"loss": 0.7193,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.13793103448275862,
|
| 20 |
+
"grad_norm": 8.326750755310059,
|
| 21 |
+
"learning_rate": 1.3793103448275863e-05,
|
| 22 |
+
"loss": 0.6911,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.20689655172413793,
|
| 27 |
+
"grad_norm": 7.880892753601074,
|
| 28 |
+
"learning_rate": 1.9999275591576767e-05,
|
| 29 |
+
"loss": 0.7207,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.27586206896551724,
|
| 34 |
+
"grad_norm": 7.71298885345459,
|
| 35 |
+
"learning_rate": 1.991247350280358e-05,
|
| 36 |
+
"loss": 0.7034,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.3448275862068966,
|
| 41 |
+
"grad_norm": 14.676814079284668,
|
| 42 |
+
"learning_rate": 1.9682229406025635e-05,
|
| 43 |
+
"loss": 0.697,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.3448275862068966,
|
| 48 |
+
"eval_accuracy": 0.4726027397260274,
|
| 49 |
+
"eval_loss": 0.7181810140609741,
|
| 50 |
+
"eval_runtime": 96.4229,
|
| 51 |
+
"eval_samples_per_second": 1.514,
|
| 52 |
+
"eval_steps_per_second": 0.384,
|
| 53 |
+
"step": 50
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.41379310344827586,
|
| 57 |
+
"grad_norm": 19.170063018798828,
|
| 58 |
+
"learning_rate": 1.931187513114025e-05,
|
| 59 |
+
"loss": 0.7374,
|
| 60 |
+
"step": 60
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.4827586206896552,
|
| 64 |
+
"grad_norm": 12.284372329711914,
|
| 65 |
+
"learning_rate": 1.880677002269928e-05,
|
| 66 |
+
"loss": 0.7261,
|
| 67 |
+
"step": 70
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.5517241379310345,
|
| 71 |
+
"grad_norm": 10.028695106506348,
|
| 72 |
+
"learning_rate": 1.817422338558892e-05,
|
| 73 |
+
"loss": 0.7085,
|
| 74 |
+
"step": 80
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.6206896551724138,
|
| 78 |
+
"grad_norm": 8.536717414855957,
|
| 79 |
+
"learning_rate": 1.74233887131081e-05,
|
| 80 |
+
"loss": 0.7193,
|
| 81 |
+
"step": 90
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.6896551724137931,
|
| 85 |
+
"grad_norm": 11.210798263549805,
|
| 86 |
+
"learning_rate": 1.6565131228056136e-05,
|
| 87 |
+
"loss": 0.7254,
|
| 88 |
+
"step": 100
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.6896551724137931,
|
| 92 |
+
"eval_accuracy": 0.4863013698630137,
|
| 93 |
+
"eval_loss": 0.7144510746002197,
|
| 94 |
+
"eval_runtime": 95.1394,
|
| 95 |
+
"eval_samples_per_second": 1.535,
|
| 96 |
+
"eval_steps_per_second": 0.389,
|
| 97 |
+
"step": 100
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.7586206896551724,
|
| 101 |
+
"grad_norm": 15.714051246643066,
|
| 102 |
+
"learning_rate": 1.5611870653623826e-05,
|
| 103 |
+
"loss": 0.6968,
|
| 104 |
+
"step": 110
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.8275862068965517,
|
| 108 |
+
"grad_norm": 17.525775909423828,
|
| 109 |
+
"learning_rate": 1.4577401489328335e-05,
|
| 110 |
+
"loss": 0.7214,
|
| 111 |
+
"step": 120
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.896551724137931,
|
| 115 |
+
"grad_norm": 8.84438705444336,
|
| 116 |
+
"learning_rate": 1.3476693392753477e-05,
|
| 117 |
+
"loss": 0.7145,
|
| 118 |
+
"step": 130
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.9655172413793104,
|
| 122 |
+
"grad_norm": 17.513769149780273,
|
| 123 |
+
"learning_rate": 1.2325674555743106e-05,
|
| 124 |
+
"loss": 0.7781,
|
| 125 |
+
"step": 140
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.0344827586206897,
|
| 129 |
+
"grad_norm": 7.130011081695557,
|
| 130 |
+
"learning_rate": 1.1141001209780249e-05,
|
| 131 |
+
"loss": 0.6942,
|
| 132 |
+
"step": 150
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 1.0344827586206897,
|
| 136 |
+
"eval_accuracy": 0.4794520547945205,
|
| 137 |
+
"eval_loss": 0.7006052136421204,
|
| 138 |
+
"eval_runtime": 94.9661,
|
| 139 |
+
"eval_samples_per_second": 1.537,
|
| 140 |
+
"eval_steps_per_second": 0.39,
|
| 141 |
+
"step": 150
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 1.103448275862069,
|
| 145 |
+
"grad_norm": 11.245115280151367,
|
| 146 |
+
"learning_rate": 9.939816596007147e-06,
|
| 147 |
+
"loss": 0.7334,
|
| 148 |
+
"step": 160
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 1.1724137931034484,
|
| 152 |
+
"grad_norm": 17.212377548217773,
|
| 153 |
+
"learning_rate": 8.739502887797108e-06,
|
| 154 |
+
"loss": 0.648,
|
| 155 |
+
"step": 170
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.2413793103448276,
|
| 159 |
+
"grad_norm": 7.2269606590271,
|
| 160 |
+
"learning_rate": 7.557429655771691e-06,
|
| 161 |
+
"loss": 0.7214,
|
| 162 |
+
"step": 180
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 1.3103448275862069,
|
| 166 |
+
"grad_norm": 16.559890747070312,
|
| 167 |
+
"learning_rate": 6.410702515190544e-06,
|
| 168 |
+
"loss": 0.7197,
|
| 169 |
+
"step": 190
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 1.3793103448275863,
|
| 173 |
+
"grad_norm": 7.4253387451171875,
|
| 174 |
+
"learning_rate": 5.3159155930021e-06,
|
| 175 |
+
"loss": 0.6836,
|
| 176 |
+
"step": 200
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.3793103448275863,
|
| 180 |
+
"eval_accuracy": 0.4178082191780822,
|
| 181 |
+
"eval_loss": 0.7102890014648438,
|
| 182 |
+
"eval_runtime": 97.3566,
|
| 183 |
+
"eval_samples_per_second": 1.5,
|
| 184 |
+
"eval_steps_per_second": 0.38,
|
| 185 |
+
"step": 200
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.4482758620689655,
|
| 189 |
+
"grad_norm": 7.56271505355835,
|
| 190 |
+
"learning_rate": 4.288911396569599e-06,
|
| 191 |
+
"loss": 0.6892,
|
| 192 |
+
"step": 210
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 1.5172413793103448,
|
| 196 |
+
"grad_norm": 7.661104679107666,
|
| 197 |
+
"learning_rate": 3.3445515589782574e-06,
|
| 198 |
+
"loss": 0.6895,
|
| 199 |
+
"step": 220
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 1.5862068965517242,
|
| 203 |
+
"grad_norm": 7.989816188812256,
|
| 204 |
+
"learning_rate": 2.496501778435977e-06,
|
| 205 |
+
"loss": 0.6854,
|
| 206 |
+
"step": 230
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.6551724137931034,
|
| 210 |
+
"grad_norm": 7.145473957061768,
|
| 211 |
+
"learning_rate": 1.7570340638792349e-06,
|
| 212 |
+
"loss": 0.7028,
|
| 213 |
+
"step": 240
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 1.7241379310344827,
|
| 217 |
+
"grad_norm": 8.94724178314209,
|
| 218 |
+
"learning_rate": 1.136849148460125e-06,
|
| 219 |
+
"loss": 0.6861,
|
| 220 |
+
"step": 250
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.7241379310344827,
|
| 224 |
+
"eval_accuracy": 0.4520547945205479,
|
| 225 |
+
"eval_loss": 0.7077414989471436,
|
| 226 |
+
"eval_runtime": 80.9468,
|
| 227 |
+
"eval_samples_per_second": 1.804,
|
| 228 |
+
"eval_steps_per_second": 0.457,
|
| 229 |
+
"step": 250
|
| 230 |
+
}
|
| 231 |
+
],
|
| 232 |
+
"logging_steps": 10,
|
| 233 |
+
"max_steps": 290,
|
| 234 |
+
"num_input_tokens_seen": 0,
|
| 235 |
+
"num_train_epochs": 2,
|
| 236 |
+
"save_steps": 50,
|
| 237 |
+
"stateful_callbacks": {
|
| 238 |
+
"TrainerControl": {
|
| 239 |
+
"args": {
|
| 240 |
+
"should_epoch_stop": false,
|
| 241 |
+
"should_evaluate": false,
|
| 242 |
+
"should_log": false,
|
| 243 |
+
"should_save": true,
|
| 244 |
+
"should_training_stop": false
|
| 245 |
+
},
|
| 246 |
+
"attributes": {}
|
| 247 |
+
}
|
| 248 |
+
},
|
| 249 |
+
"total_flos": 1.246064589471744e+18,
|
| 250 |
+
"train_batch_size": 4,
|
| 251 |
+
"trial_name": null,
|
| 252 |
+
"trial_params": null
|
| 253 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b2a1cf5116bd008dc6a67e79db7280fd55e533974b93e1fb4d1cdd3f4eee1a5
|
| 3 |
+
size 5649
|