Upload folder using huggingface_hub
Browse files- checkpoint-1120/config.json +32 -0
- checkpoint-1120/model.safetensors +3 -0
- checkpoint-1120/optimizer.pt +3 -0
- checkpoint-1120/preprocessor_config.json +36 -0
- checkpoint-1120/rng_state.pth +3 -0
- checkpoint-1120/scheduler.pt +3 -0
- checkpoint-1120/trainer_state.json +677 -0
- checkpoint-1120/training_args.bin +3 -0
- config.json +32 -0
- model.safetensors +3 -0
- preprocessor_config.json +36 -0
- training_args.bin +3 -0
checkpoint-1120/config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "dima806/deepfake_vs_real_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": "Real",
|
| 13 |
+
"1": "Fake"
|
| 14 |
+
},
|
| 15 |
+
"image_size": 224,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 3072,
|
| 18 |
+
"label2id": {
|
| 19 |
+
"Fake": 1,
|
| 20 |
+
"Real": 0
|
| 21 |
+
},
|
| 22 |
+
"layer_norm_eps": 1e-12,
|
| 23 |
+
"model_type": "vit",
|
| 24 |
+
"num_attention_heads": 12,
|
| 25 |
+
"num_channels": 3,
|
| 26 |
+
"num_hidden_layers": 12,
|
| 27 |
+
"patch_size": 16,
|
| 28 |
+
"problem_type": "single_label_classification",
|
| 29 |
+
"qkv_bias": true,
|
| 30 |
+
"torch_dtype": "float32",
|
| 31 |
+
"transformers_version": "4.41.1"
|
| 32 |
+
}
|
checkpoint-1120/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ae1f3f0da1454b92763e6fbcc02d1365df7df4843ae939c6ce93fe8f97acaa61
|
| 3 |
+
size 343223968
|
checkpoint-1120/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d07d1871b4297e1de258974a0feff30c0eb0a44e1df1e6f479b62f4d50d4f97
|
| 3 |
+
size 686568453
|
checkpoint-1120/preprocessor_config.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_valid_processor_keys": [
|
| 3 |
+
"images",
|
| 4 |
+
"do_resize",
|
| 5 |
+
"size",
|
| 6 |
+
"resample",
|
| 7 |
+
"do_rescale",
|
| 8 |
+
"rescale_factor",
|
| 9 |
+
"do_normalize",
|
| 10 |
+
"image_mean",
|
| 11 |
+
"image_std",
|
| 12 |
+
"return_tensors",
|
| 13 |
+
"data_format",
|
| 14 |
+
"input_data_format"
|
| 15 |
+
],
|
| 16 |
+
"do_normalize": true,
|
| 17 |
+
"do_rescale": true,
|
| 18 |
+
"do_resize": true,
|
| 19 |
+
"image_mean": [
|
| 20 |
+
0.5,
|
| 21 |
+
0.5,
|
| 22 |
+
0.5
|
| 23 |
+
],
|
| 24 |
+
"image_processor_type": "ViTImageProcessor",
|
| 25 |
+
"image_std": [
|
| 26 |
+
0.5,
|
| 27 |
+
0.5,
|
| 28 |
+
0.5
|
| 29 |
+
],
|
| 30 |
+
"resample": 2,
|
| 31 |
+
"rescale_factor": 0.00392156862745098,
|
| 32 |
+
"size": {
|
| 33 |
+
"height": 224,
|
| 34 |
+
"width": 224
|
| 35 |
+
}
|
| 36 |
+
}
|
checkpoint-1120/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95b5909d7f693f99d9eb5ac737096023ba0bd963aa544dc6af82da304839f8f0
|
| 3 |
+
size 14575
|
checkpoint-1120/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:70272f50d5452d4e79b70ac0a09cd865a625b808d0d4c278db7317880c9de22c
|
| 3 |
+
size 627
|
checkpoint-1120/trainer_state.json
ADDED
|
@@ -0,0 +1,677 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.35881221294403076,
|
| 3 |
+
"best_model_checkpoint": "deepfake_vs_real_image_detection/checkpoint-1120",
|
| 4 |
+
"epoch": 70.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1120,
|
| 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": 0.6946107784431138,
|
| 14 |
+
"eval_loss": 1.4309433698654175,
|
| 15 |
+
"eval_runtime": 4.3566,
|
| 16 |
+
"eval_samples_per_second": 76.666,
|
| 17 |
+
"eval_steps_per_second": 9.641,
|
| 18 |
+
"step": 16
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 2.0,
|
| 22 |
+
"eval_accuracy": 0.7035928143712575,
|
| 23 |
+
"eval_loss": 1.3571979999542236,
|
| 24 |
+
"eval_runtime": 4.4448,
|
| 25 |
+
"eval_samples_per_second": 75.144,
|
| 26 |
+
"eval_steps_per_second": 9.449,
|
| 27 |
+
"step": 32
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 3.0,
|
| 31 |
+
"eval_accuracy": 0.7095808383233533,
|
| 32 |
+
"eval_loss": 1.2381948232650757,
|
| 33 |
+
"eval_runtime": 4.5106,
|
| 34 |
+
"eval_samples_per_second": 74.048,
|
| 35 |
+
"eval_steps_per_second": 9.311,
|
| 36 |
+
"step": 48
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 4.0,
|
| 40 |
+
"eval_accuracy": 0.7485029940119761,
|
| 41 |
+
"eval_loss": 1.1571621894836426,
|
| 42 |
+
"eval_runtime": 4.3237,
|
| 43 |
+
"eval_samples_per_second": 77.249,
|
| 44 |
+
"eval_steps_per_second": 9.714,
|
| 45 |
+
"step": 64
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 5.0,
|
| 49 |
+
"eval_accuracy": 0.7544910179640718,
|
| 50 |
+
"eval_loss": 1.1075578927993774,
|
| 51 |
+
"eval_runtime": 4.3443,
|
| 52 |
+
"eval_samples_per_second": 76.883,
|
| 53 |
+
"eval_steps_per_second": 9.668,
|
| 54 |
+
"step": 80
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 6.0,
|
| 58 |
+
"eval_accuracy": 0.7544910179640718,
|
| 59 |
+
"eval_loss": 1.0685527324676514,
|
| 60 |
+
"eval_runtime": 4.4445,
|
| 61 |
+
"eval_samples_per_second": 75.149,
|
| 62 |
+
"eval_steps_per_second": 9.45,
|
| 63 |
+
"step": 96
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 7.0,
|
| 67 |
+
"eval_accuracy": 0.7574850299401198,
|
| 68 |
+
"eval_loss": 1.0224419832229614,
|
| 69 |
+
"eval_runtime": 4.3333,
|
| 70 |
+
"eval_samples_per_second": 77.078,
|
| 71 |
+
"eval_steps_per_second": 9.692,
|
| 72 |
+
"step": 112
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 8.0,
|
| 76 |
+
"eval_accuracy": 0.7604790419161677,
|
| 77 |
+
"eval_loss": 0.9735947847366333,
|
| 78 |
+
"eval_runtime": 4.4514,
|
| 79 |
+
"eval_samples_per_second": 75.033,
|
| 80 |
+
"eval_steps_per_second": 9.435,
|
| 81 |
+
"step": 128
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 9.0,
|
| 85 |
+
"eval_accuracy": 0.7634730538922155,
|
| 86 |
+
"eval_loss": 0.935190737247467,
|
| 87 |
+
"eval_runtime": 4.483,
|
| 88 |
+
"eval_samples_per_second": 74.504,
|
| 89 |
+
"eval_steps_per_second": 9.369,
|
| 90 |
+
"step": 144
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 10.0,
|
| 94 |
+
"eval_accuracy": 0.7574850299401198,
|
| 95 |
+
"eval_loss": 0.8989432454109192,
|
| 96 |
+
"eval_runtime": 4.3947,
|
| 97 |
+
"eval_samples_per_second": 76.0,
|
| 98 |
+
"eval_steps_per_second": 9.557,
|
| 99 |
+
"step": 160
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 11.0,
|
| 103 |
+
"eval_accuracy": 0.7514970059880239,
|
| 104 |
+
"eval_loss": 0.8565046787261963,
|
| 105 |
+
"eval_runtime": 4.4679,
|
| 106 |
+
"eval_samples_per_second": 74.755,
|
| 107 |
+
"eval_steps_per_second": 9.4,
|
| 108 |
+
"step": 176
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 12.0,
|
| 112 |
+
"eval_accuracy": 0.7634730538922155,
|
| 113 |
+
"eval_loss": 0.8191085457801819,
|
| 114 |
+
"eval_runtime": 4.4986,
|
| 115 |
+
"eval_samples_per_second": 74.245,
|
| 116 |
+
"eval_steps_per_second": 9.336,
|
| 117 |
+
"step": 192
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 13.0,
|
| 121 |
+
"eval_accuracy": 0.7634730538922155,
|
| 122 |
+
"eval_loss": 0.783191442489624,
|
| 123 |
+
"eval_runtime": 4.5843,
|
| 124 |
+
"eval_samples_per_second": 72.857,
|
| 125 |
+
"eval_steps_per_second": 9.162,
|
| 126 |
+
"step": 208
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 14.0,
|
| 130 |
+
"eval_accuracy": 0.7724550898203593,
|
| 131 |
+
"eval_loss": 0.7542709708213806,
|
| 132 |
+
"eval_runtime": 4.417,
|
| 133 |
+
"eval_samples_per_second": 75.617,
|
| 134 |
+
"eval_steps_per_second": 9.509,
|
| 135 |
+
"step": 224
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 15.0,
|
| 139 |
+
"eval_accuracy": 0.7694610778443114,
|
| 140 |
+
"eval_loss": 0.7199692130088806,
|
| 141 |
+
"eval_runtime": 4.401,
|
| 142 |
+
"eval_samples_per_second": 75.891,
|
| 143 |
+
"eval_steps_per_second": 9.543,
|
| 144 |
+
"step": 240
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 16.0,
|
| 148 |
+
"eval_accuracy": 0.7754491017964071,
|
| 149 |
+
"eval_loss": 0.6858012676239014,
|
| 150 |
+
"eval_runtime": 4.4254,
|
| 151 |
+
"eval_samples_per_second": 75.474,
|
| 152 |
+
"eval_steps_per_second": 9.491,
|
| 153 |
+
"step": 256
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 17.0,
|
| 157 |
+
"eval_accuracy": 0.7784431137724551,
|
| 158 |
+
"eval_loss": 0.6566184759140015,
|
| 159 |
+
"eval_runtime": 4.4216,
|
| 160 |
+
"eval_samples_per_second": 75.539,
|
| 161 |
+
"eval_steps_per_second": 9.499,
|
| 162 |
+
"step": 272
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 18.0,
|
| 166 |
+
"eval_accuracy": 0.7844311377245509,
|
| 167 |
+
"eval_loss": 0.628990650177002,
|
| 168 |
+
"eval_runtime": 4.4307,
|
| 169 |
+
"eval_samples_per_second": 75.383,
|
| 170 |
+
"eval_steps_per_second": 9.479,
|
| 171 |
+
"step": 288
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 19.0,
|
| 175 |
+
"eval_accuracy": 0.7844311377245509,
|
| 176 |
+
"eval_loss": 0.6014779210090637,
|
| 177 |
+
"eval_runtime": 4.3612,
|
| 178 |
+
"eval_samples_per_second": 76.584,
|
| 179 |
+
"eval_steps_per_second": 9.63,
|
| 180 |
+
"step": 304
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 20.0,
|
| 184 |
+
"eval_accuracy": 0.7904191616766467,
|
| 185 |
+
"eval_loss": 0.5828869938850403,
|
| 186 |
+
"eval_runtime": 4.3656,
|
| 187 |
+
"eval_samples_per_second": 76.508,
|
| 188 |
+
"eval_steps_per_second": 9.621,
|
| 189 |
+
"step": 320
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 21.0,
|
| 193 |
+
"eval_accuracy": 0.7904191616766467,
|
| 194 |
+
"eval_loss": 0.560990571975708,
|
| 195 |
+
"eval_runtime": 4.4037,
|
| 196 |
+
"eval_samples_per_second": 75.846,
|
| 197 |
+
"eval_steps_per_second": 9.538,
|
| 198 |
+
"step": 336
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 22.0,
|
| 202 |
+
"eval_accuracy": 0.7904191616766467,
|
| 203 |
+
"eval_loss": 0.5444124937057495,
|
| 204 |
+
"eval_runtime": 4.4206,
|
| 205 |
+
"eval_samples_per_second": 75.556,
|
| 206 |
+
"eval_steps_per_second": 9.501,
|
| 207 |
+
"step": 352
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 23.0,
|
| 211 |
+
"eval_accuracy": 0.7964071856287425,
|
| 212 |
+
"eval_loss": 0.5321463346481323,
|
| 213 |
+
"eval_runtime": 4.402,
|
| 214 |
+
"eval_samples_per_second": 75.875,
|
| 215 |
+
"eval_steps_per_second": 9.541,
|
| 216 |
+
"step": 368
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 24.0,
|
| 220 |
+
"eval_accuracy": 0.8023952095808383,
|
| 221 |
+
"eval_loss": 0.521594226360321,
|
| 222 |
+
"eval_runtime": 4.4337,
|
| 223 |
+
"eval_samples_per_second": 75.331,
|
| 224 |
+
"eval_steps_per_second": 9.473,
|
| 225 |
+
"step": 384
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 25.0,
|
| 229 |
+
"eval_accuracy": 0.8053892215568862,
|
| 230 |
+
"eval_loss": 0.5053771138191223,
|
| 231 |
+
"eval_runtime": 4.4563,
|
| 232 |
+
"eval_samples_per_second": 74.95,
|
| 233 |
+
"eval_steps_per_second": 9.425,
|
| 234 |
+
"step": 400
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 26.0,
|
| 238 |
+
"eval_accuracy": 0.8143712574850299,
|
| 239 |
+
"eval_loss": 0.4930258095264435,
|
| 240 |
+
"eval_runtime": 4.3955,
|
| 241 |
+
"eval_samples_per_second": 75.986,
|
| 242 |
+
"eval_steps_per_second": 9.555,
|
| 243 |
+
"step": 416
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 27.0,
|
| 247 |
+
"eval_accuracy": 0.811377245508982,
|
| 248 |
+
"eval_loss": 0.4849102795124054,
|
| 249 |
+
"eval_runtime": 4.4566,
|
| 250 |
+
"eval_samples_per_second": 74.945,
|
| 251 |
+
"eval_steps_per_second": 9.424,
|
| 252 |
+
"step": 432
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 28.0,
|
| 256 |
+
"eval_accuracy": 0.811377245508982,
|
| 257 |
+
"eval_loss": 0.47692808508872986,
|
| 258 |
+
"eval_runtime": 4.4557,
|
| 259 |
+
"eval_samples_per_second": 74.959,
|
| 260 |
+
"eval_steps_per_second": 9.426,
|
| 261 |
+
"step": 448
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 29.0,
|
| 265 |
+
"eval_accuracy": 0.8293413173652695,
|
| 266 |
+
"eval_loss": 0.46685636043548584,
|
| 267 |
+
"eval_runtime": 4.3405,
|
| 268 |
+
"eval_samples_per_second": 76.949,
|
| 269 |
+
"eval_steps_per_second": 9.676,
|
| 270 |
+
"step": 464
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 30.0,
|
| 274 |
+
"eval_accuracy": 0.8323353293413174,
|
| 275 |
+
"eval_loss": 0.45782506465911865,
|
| 276 |
+
"eval_runtime": 4.3468,
|
| 277 |
+
"eval_samples_per_second": 76.838,
|
| 278 |
+
"eval_steps_per_second": 9.662,
|
| 279 |
+
"step": 480
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 31.0,
|
| 283 |
+
"eval_accuracy": 0.8323353293413174,
|
| 284 |
+
"eval_loss": 0.44891008734703064,
|
| 285 |
+
"eval_runtime": 4.5111,
|
| 286 |
+
"eval_samples_per_second": 74.039,
|
| 287 |
+
"eval_steps_per_second": 9.31,
|
| 288 |
+
"step": 496
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 31.25,
|
| 292 |
+
"grad_norm": 3.9690632820129395,
|
| 293 |
+
"learning_rate": 8.098055790363483e-07,
|
| 294 |
+
"loss": 0.6987,
|
| 295 |
+
"step": 500
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 32.0,
|
| 299 |
+
"eval_accuracy": 0.8323353293413174,
|
| 300 |
+
"eval_loss": 0.4414936900138855,
|
| 301 |
+
"eval_runtime": 4.3863,
|
| 302 |
+
"eval_samples_per_second": 76.146,
|
| 303 |
+
"eval_steps_per_second": 9.575,
|
| 304 |
+
"step": 512
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 33.0,
|
| 308 |
+
"eval_accuracy": 0.8323353293413174,
|
| 309 |
+
"eval_loss": 0.435556560754776,
|
| 310 |
+
"eval_runtime": 4.3922,
|
| 311 |
+
"eval_samples_per_second": 76.043,
|
| 312 |
+
"eval_steps_per_second": 9.562,
|
| 313 |
+
"step": 528
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 34.0,
|
| 317 |
+
"eval_accuracy": 0.8323353293413174,
|
| 318 |
+
"eval_loss": 0.4296257793903351,
|
| 319 |
+
"eval_runtime": 4.3498,
|
| 320 |
+
"eval_samples_per_second": 76.785,
|
| 321 |
+
"eval_steps_per_second": 9.656,
|
| 322 |
+
"step": 544
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 35.0,
|
| 326 |
+
"eval_accuracy": 0.8323353293413174,
|
| 327 |
+
"eval_loss": 0.4240988492965698,
|
| 328 |
+
"eval_runtime": 4.3863,
|
| 329 |
+
"eval_samples_per_second": 76.146,
|
| 330 |
+
"eval_steps_per_second": 9.575,
|
| 331 |
+
"step": 560
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 36.0,
|
| 335 |
+
"eval_accuracy": 0.8353293413173652,
|
| 336 |
+
"eval_loss": 0.41868290305137634,
|
| 337 |
+
"eval_runtime": 4.4309,
|
| 338 |
+
"eval_samples_per_second": 75.38,
|
| 339 |
+
"eval_steps_per_second": 9.479,
|
| 340 |
+
"step": 576
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 37.0,
|
| 344 |
+
"eval_accuracy": 0.8383233532934131,
|
| 345 |
+
"eval_loss": 0.413825124502182,
|
| 346 |
+
"eval_runtime": 4.3459,
|
| 347 |
+
"eval_samples_per_second": 76.853,
|
| 348 |
+
"eval_steps_per_second": 9.664,
|
| 349 |
+
"step": 592
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 38.0,
|
| 353 |
+
"eval_accuracy": 0.8413173652694611,
|
| 354 |
+
"eval_loss": 0.4092709720134735,
|
| 355 |
+
"eval_runtime": 4.353,
|
| 356 |
+
"eval_samples_per_second": 76.728,
|
| 357 |
+
"eval_steps_per_second": 9.648,
|
| 358 |
+
"step": 608
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 39.0,
|
| 362 |
+
"eval_accuracy": 0.8413173652694611,
|
| 363 |
+
"eval_loss": 0.40586671233177185,
|
| 364 |
+
"eval_runtime": 4.3701,
|
| 365 |
+
"eval_samples_per_second": 76.429,
|
| 366 |
+
"eval_steps_per_second": 9.611,
|
| 367 |
+
"step": 624
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 40.0,
|
| 371 |
+
"eval_accuracy": 0.8502994011976048,
|
| 372 |
+
"eval_loss": 0.401659220457077,
|
| 373 |
+
"eval_runtime": 4.4026,
|
| 374 |
+
"eval_samples_per_second": 75.864,
|
| 375 |
+
"eval_steps_per_second": 9.54,
|
| 376 |
+
"step": 640
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 41.0,
|
| 380 |
+
"eval_accuracy": 0.8502994011976048,
|
| 381 |
+
"eval_loss": 0.3979702293872833,
|
| 382 |
+
"eval_runtime": 4.5672,
|
| 383 |
+
"eval_samples_per_second": 73.13,
|
| 384 |
+
"eval_steps_per_second": 9.196,
|
| 385 |
+
"step": 656
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 42.0,
|
| 389 |
+
"eval_accuracy": 0.8502994011976048,
|
| 390 |
+
"eval_loss": 0.3946226239204407,
|
| 391 |
+
"eval_runtime": 4.3743,
|
| 392 |
+
"eval_samples_per_second": 76.354,
|
| 393 |
+
"eval_steps_per_second": 9.601,
|
| 394 |
+
"step": 672
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 43.0,
|
| 398 |
+
"eval_accuracy": 0.8473053892215568,
|
| 399 |
+
"eval_loss": 0.39317673444747925,
|
| 400 |
+
"eval_runtime": 4.3247,
|
| 401 |
+
"eval_samples_per_second": 77.23,
|
| 402 |
+
"eval_steps_per_second": 9.712,
|
| 403 |
+
"step": 688
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 44.0,
|
| 407 |
+
"eval_accuracy": 0.8473053892215568,
|
| 408 |
+
"eval_loss": 0.389418363571167,
|
| 409 |
+
"eval_runtime": 4.4405,
|
| 410 |
+
"eval_samples_per_second": 75.216,
|
| 411 |
+
"eval_steps_per_second": 9.458,
|
| 412 |
+
"step": 704
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 45.0,
|
| 416 |
+
"eval_accuracy": 0.8473053892215568,
|
| 417 |
+
"eval_loss": 0.3862323462963104,
|
| 418 |
+
"eval_runtime": 4.3815,
|
| 419 |
+
"eval_samples_per_second": 76.23,
|
| 420 |
+
"eval_steps_per_second": 9.586,
|
| 421 |
+
"step": 720
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 46.0,
|
| 425 |
+
"eval_accuracy": 0.8502994011976048,
|
| 426 |
+
"eval_loss": 0.384316086769104,
|
| 427 |
+
"eval_runtime": 4.4286,
|
| 428 |
+
"eval_samples_per_second": 75.42,
|
| 429 |
+
"eval_steps_per_second": 9.484,
|
| 430 |
+
"step": 736
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 47.0,
|
| 434 |
+
"eval_accuracy": 0.844311377245509,
|
| 435 |
+
"eval_loss": 0.383512943983078,
|
| 436 |
+
"eval_runtime": 4.4251,
|
| 437 |
+
"eval_samples_per_second": 75.479,
|
| 438 |
+
"eval_steps_per_second": 9.491,
|
| 439 |
+
"step": 752
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 48.0,
|
| 443 |
+
"eval_accuracy": 0.8622754491017964,
|
| 444 |
+
"eval_loss": 0.3795510232448578,
|
| 445 |
+
"eval_runtime": 4.4626,
|
| 446 |
+
"eval_samples_per_second": 74.844,
|
| 447 |
+
"eval_steps_per_second": 9.412,
|
| 448 |
+
"step": 768
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 49.0,
|
| 452 |
+
"eval_accuracy": 0.8592814371257484,
|
| 453 |
+
"eval_loss": 0.3759188950061798,
|
| 454 |
+
"eval_runtime": 4.3421,
|
| 455 |
+
"eval_samples_per_second": 76.921,
|
| 456 |
+
"eval_steps_per_second": 9.673,
|
| 457 |
+
"step": 784
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 50.0,
|
| 461 |
+
"eval_accuracy": 0.8652694610778443,
|
| 462 |
+
"eval_loss": 0.37549516558647156,
|
| 463 |
+
"eval_runtime": 4.4076,
|
| 464 |
+
"eval_samples_per_second": 75.778,
|
| 465 |
+
"eval_steps_per_second": 9.529,
|
| 466 |
+
"step": 800
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 51.0,
|
| 470 |
+
"eval_accuracy": 0.8622754491017964,
|
| 471 |
+
"eval_loss": 0.3745565414428711,
|
| 472 |
+
"eval_runtime": 4.466,
|
| 473 |
+
"eval_samples_per_second": 74.787,
|
| 474 |
+
"eval_steps_per_second": 9.404,
|
| 475 |
+
"step": 816
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 52.0,
|
| 479 |
+
"eval_accuracy": 0.8652694610778443,
|
| 480 |
+
"eval_loss": 0.37246376276016235,
|
| 481 |
+
"eval_runtime": 4.3519,
|
| 482 |
+
"eval_samples_per_second": 76.748,
|
| 483 |
+
"eval_steps_per_second": 9.651,
|
| 484 |
+
"step": 832
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 53.0,
|
| 488 |
+
"eval_accuracy": 0.8652694610778443,
|
| 489 |
+
"eval_loss": 0.36974743008613586,
|
| 490 |
+
"eval_runtime": 4.4116,
|
| 491 |
+
"eval_samples_per_second": 75.709,
|
| 492 |
+
"eval_steps_per_second": 9.52,
|
| 493 |
+
"step": 848
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 54.0,
|
| 497 |
+
"eval_accuracy": 0.874251497005988,
|
| 498 |
+
"eval_loss": 0.3688933253288269,
|
| 499 |
+
"eval_runtime": 4.373,
|
| 500 |
+
"eval_samples_per_second": 76.377,
|
| 501 |
+
"eval_steps_per_second": 9.604,
|
| 502 |
+
"step": 864
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 55.0,
|
| 506 |
+
"eval_accuracy": 0.874251497005988,
|
| 507 |
+
"eval_loss": 0.3665659427642822,
|
| 508 |
+
"eval_runtime": 4.3847,
|
| 509 |
+
"eval_samples_per_second": 76.173,
|
| 510 |
+
"eval_steps_per_second": 9.579,
|
| 511 |
+
"step": 880
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 56.0,
|
| 515 |
+
"eval_accuracy": 0.874251497005988,
|
| 516 |
+
"eval_loss": 0.3666609227657318,
|
| 517 |
+
"eval_runtime": 4.4937,
|
| 518 |
+
"eval_samples_per_second": 74.326,
|
| 519 |
+
"eval_steps_per_second": 9.346,
|
| 520 |
+
"step": 896
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 57.0,
|
| 524 |
+
"eval_accuracy": 0.8802395209580839,
|
| 525 |
+
"eval_loss": 0.366781622171402,
|
| 526 |
+
"eval_runtime": 4.4056,
|
| 527 |
+
"eval_samples_per_second": 75.812,
|
| 528 |
+
"eval_steps_per_second": 9.533,
|
| 529 |
+
"step": 912
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 58.0,
|
| 533 |
+
"eval_accuracy": 0.8802395209580839,
|
| 534 |
+
"eval_loss": 0.3661949932575226,
|
| 535 |
+
"eval_runtime": 4.4335,
|
| 536 |
+
"eval_samples_per_second": 75.336,
|
| 537 |
+
"eval_steps_per_second": 9.473,
|
| 538 |
+
"step": 928
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 59.0,
|
| 542 |
+
"eval_accuracy": 0.8802395209580839,
|
| 543 |
+
"eval_loss": 0.36339056491851807,
|
| 544 |
+
"eval_runtime": 4.4117,
|
| 545 |
+
"eval_samples_per_second": 75.708,
|
| 546 |
+
"eval_steps_per_second": 9.52,
|
| 547 |
+
"step": 944
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 60.0,
|
| 551 |
+
"eval_accuracy": 0.8772455089820359,
|
| 552 |
+
"eval_loss": 0.36299723386764526,
|
| 553 |
+
"eval_runtime": 4.4436,
|
| 554 |
+
"eval_samples_per_second": 75.165,
|
| 555 |
+
"eval_steps_per_second": 9.452,
|
| 556 |
+
"step": 960
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 61.0,
|
| 560 |
+
"eval_accuracy": 0.8832335329341318,
|
| 561 |
+
"eval_loss": 0.3644971549510956,
|
| 562 |
+
"eval_runtime": 4.3789,
|
| 563 |
+
"eval_samples_per_second": 76.274,
|
| 564 |
+
"eval_steps_per_second": 9.591,
|
| 565 |
+
"step": 976
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 62.0,
|
| 569 |
+
"eval_accuracy": 0.8832335329341318,
|
| 570 |
+
"eval_loss": 0.3652717173099518,
|
| 571 |
+
"eval_runtime": 4.3124,
|
| 572 |
+
"eval_samples_per_second": 77.452,
|
| 573 |
+
"eval_steps_per_second": 9.739,
|
| 574 |
+
"step": 992
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 62.5,
|
| 578 |
+
"grad_norm": 1.597828984260559,
|
| 579 |
+
"learning_rate": 5.984784446322908e-07,
|
| 580 |
+
"loss": 0.1919,
|
| 581 |
+
"step": 1000
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 63.0,
|
| 585 |
+
"eval_accuracy": 0.8832335329341318,
|
| 586 |
+
"eval_loss": 0.3627488911151886,
|
| 587 |
+
"eval_runtime": 4.4759,
|
| 588 |
+
"eval_samples_per_second": 74.622,
|
| 589 |
+
"eval_steps_per_second": 9.384,
|
| 590 |
+
"step": 1008
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 64.0,
|
| 594 |
+
"eval_accuracy": 0.8772455089820359,
|
| 595 |
+
"eval_loss": 0.3642951548099518,
|
| 596 |
+
"eval_runtime": 4.4875,
|
| 597 |
+
"eval_samples_per_second": 74.428,
|
| 598 |
+
"eval_steps_per_second": 9.359,
|
| 599 |
+
"step": 1024
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 65.0,
|
| 603 |
+
"eval_accuracy": 0.874251497005988,
|
| 604 |
+
"eval_loss": 0.3665095567703247,
|
| 605 |
+
"eval_runtime": 4.39,
|
| 606 |
+
"eval_samples_per_second": 76.083,
|
| 607 |
+
"eval_steps_per_second": 9.567,
|
| 608 |
+
"step": 1040
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 66.0,
|
| 612 |
+
"eval_accuracy": 0.8772455089820359,
|
| 613 |
+
"eval_loss": 0.3633537292480469,
|
| 614 |
+
"eval_runtime": 4.4453,
|
| 615 |
+
"eval_samples_per_second": 75.135,
|
| 616 |
+
"eval_steps_per_second": 9.448,
|
| 617 |
+
"step": 1056
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 67.0,
|
| 621 |
+
"eval_accuracy": 0.8772455089820359,
|
| 622 |
+
"eval_loss": 0.36282193660736084,
|
| 623 |
+
"eval_runtime": 4.4608,
|
| 624 |
+
"eval_samples_per_second": 74.875,
|
| 625 |
+
"eval_steps_per_second": 9.415,
|
| 626 |
+
"step": 1072
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 68.0,
|
| 630 |
+
"eval_accuracy": 0.874251497005988,
|
| 631 |
+
"eval_loss": 0.36409279704093933,
|
| 632 |
+
"eval_runtime": 4.3878,
|
| 633 |
+
"eval_samples_per_second": 76.119,
|
| 634 |
+
"eval_steps_per_second": 9.572,
|
| 635 |
+
"step": 1088
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 69.0,
|
| 639 |
+
"eval_accuracy": 0.8772455089820359,
|
| 640 |
+
"eval_loss": 0.3623464107513428,
|
| 641 |
+
"eval_runtime": 4.4706,
|
| 642 |
+
"eval_samples_per_second": 74.711,
|
| 643 |
+
"eval_steps_per_second": 9.395,
|
| 644 |
+
"step": 1104
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 70.0,
|
| 648 |
+
"eval_accuracy": 0.8772455089820359,
|
| 649 |
+
"eval_loss": 0.35881221294403076,
|
| 650 |
+
"eval_runtime": 4.3641,
|
| 651 |
+
"eval_samples_per_second": 76.533,
|
| 652 |
+
"eval_steps_per_second": 9.624,
|
| 653 |
+
"step": 1120
|
| 654 |
+
}
|
| 655 |
+
],
|
| 656 |
+
"logging_steps": 500,
|
| 657 |
+
"max_steps": 2416,
|
| 658 |
+
"num_input_tokens_seen": 0,
|
| 659 |
+
"num_train_epochs": 151,
|
| 660 |
+
"save_steps": 500,
|
| 661 |
+
"stateful_callbacks": {
|
| 662 |
+
"TrainerControl": {
|
| 663 |
+
"args": {
|
| 664 |
+
"should_epoch_stop": false,
|
| 665 |
+
"should_evaluate": false,
|
| 666 |
+
"should_log": false,
|
| 667 |
+
"should_save": true,
|
| 668 |
+
"should_training_stop": false
|
| 669 |
+
},
|
| 670 |
+
"attributes": {}
|
| 671 |
+
}
|
| 672 |
+
},
|
| 673 |
+
"total_flos": 2.71221963651072e+18,
|
| 674 |
+
"train_batch_size": 32,
|
| 675 |
+
"trial_name": null,
|
| 676 |
+
"trial_params": null
|
| 677 |
+
}
|
checkpoint-1120/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f484c0c119950cdacad5f1cbd2aac30d22ab021a11bc6242cc3e2cb5c108c64d
|
| 3 |
+
size 4603
|
config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "dima806/deepfake_vs_real_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": "Real",
|
| 13 |
+
"1": "Fake"
|
| 14 |
+
},
|
| 15 |
+
"image_size": 224,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 3072,
|
| 18 |
+
"label2id": {
|
| 19 |
+
"Fake": 1,
|
| 20 |
+
"Real": 0
|
| 21 |
+
},
|
| 22 |
+
"layer_norm_eps": 1e-12,
|
| 23 |
+
"model_type": "vit",
|
| 24 |
+
"num_attention_heads": 12,
|
| 25 |
+
"num_channels": 3,
|
| 26 |
+
"num_hidden_layers": 12,
|
| 27 |
+
"patch_size": 16,
|
| 28 |
+
"problem_type": "single_label_classification",
|
| 29 |
+
"qkv_bias": true,
|
| 30 |
+
"torch_dtype": "float32",
|
| 31 |
+
"transformers_version": "4.41.1"
|
| 32 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ae1f3f0da1454b92763e6fbcc02d1365df7df4843ae939c6ce93fe8f97acaa61
|
| 3 |
+
size 343223968
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_valid_processor_keys": [
|
| 3 |
+
"images",
|
| 4 |
+
"do_resize",
|
| 5 |
+
"size",
|
| 6 |
+
"resample",
|
| 7 |
+
"do_rescale",
|
| 8 |
+
"rescale_factor",
|
| 9 |
+
"do_normalize",
|
| 10 |
+
"image_mean",
|
| 11 |
+
"image_std",
|
| 12 |
+
"return_tensors",
|
| 13 |
+
"data_format",
|
| 14 |
+
"input_data_format"
|
| 15 |
+
],
|
| 16 |
+
"do_normalize": true,
|
| 17 |
+
"do_rescale": true,
|
| 18 |
+
"do_resize": true,
|
| 19 |
+
"image_mean": [
|
| 20 |
+
0.5,
|
| 21 |
+
0.5,
|
| 22 |
+
0.5
|
| 23 |
+
],
|
| 24 |
+
"image_processor_type": "ViTImageProcessor",
|
| 25 |
+
"image_std": [
|
| 26 |
+
0.5,
|
| 27 |
+
0.5,
|
| 28 |
+
0.5
|
| 29 |
+
],
|
| 30 |
+
"resample": 2,
|
| 31 |
+
"rescale_factor": 0.00392156862745098,
|
| 32 |
+
"size": {
|
| 33 |
+
"height": 224,
|
| 34 |
+
"width": 224
|
| 35 |
+
}
|
| 36 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f484c0c119950cdacad5f1cbd2aac30d22ab021a11bc6242cc3e2cb5c108c64d
|
| 3 |
+
size 4603
|