Upload folder using huggingface_hub
Browse files- checkpoint-1400/config.json +32 -0
- checkpoint-1400/model.safetensors +3 -0
- checkpoint-1400/optimizer.pt +3 -0
- checkpoint-1400/preprocessor_config.json +22 -0
- checkpoint-1400/rng_state.pth +3 -0
- checkpoint-1400/scheduler.pt +3 -0
- checkpoint-1400/trainer_state.json +931 -0
- checkpoint-1400/training_args.bin +3 -0
- config.json +32 -0
- model.safetensors +3 -0
- preprocessor_config.json +22 -0
- training_args.bin +3 -0
checkpoint-1400/config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "notsmoking",
|
| 13 |
+
"1": "smoking"
|
| 14 |
+
},
|
| 15 |
+
"image_size": 224,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 3072,
|
| 18 |
+
"label2id": {
|
| 19 |
+
"notsmoking": 0,
|
| 20 |
+
"smoking": 1
|
| 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": "4.35.2"
|
| 32 |
+
}
|
checkpoint-1400/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9fb95ff652cd152efa361ce3b9aa37d471539c6407c4cf663b87eef8dc832fd9
|
| 3 |
+
size 343223968
|
checkpoint-1400/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:99496f0a2ae72128b92c999afddcf5baad2ecab9870535336529dae8e36eb2f4
|
| 3 |
+
size 686568453
|
checkpoint-1400/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-1400/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7b289ce341ba729396ddddf2e288b3e9997176c1ca553f42210e0e4126f6d798
|
| 3 |
+
size 14575
|
checkpoint-1400/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e91c4211581725916ac765c790776fd468ed84fa34fbdbd540ac48bcd3d3f766
|
| 3 |
+
size 627
|
checkpoint-1400/trainer_state.json
ADDED
|
@@ -0,0 +1,931 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.2659357786178589,
|
| 3 |
+
"best_model_checkpoint": "smoker_image_classification/checkpoint-1400",
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1400,
|
| 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": 0.6071428571428571,
|
| 14 |
+
"eval_loss": 0.6707378029823303,
|
| 15 |
+
"eval_runtime": 2.2733,
|
| 16 |
+
"eval_samples_per_second": 98.533,
|
| 17 |
+
"eval_steps_per_second": 3.079,
|
| 18 |
+
"step": 14
|
| 19 |
+
},
|
| 20 |
+
{
|
| 21 |
+
"epoch": 2.0,
|
| 22 |
+
"eval_accuracy": 0.6160714285714286,
|
| 23 |
+
"eval_loss": 0.6684125065803528,
|
| 24 |
+
"eval_runtime": 2.22,
|
| 25 |
+
"eval_samples_per_second": 100.9,
|
| 26 |
+
"eval_steps_per_second": 3.153,
|
| 27 |
+
"step": 28
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 3.0,
|
| 31 |
+
"eval_accuracy": 0.6383928571428571,
|
| 32 |
+
"eval_loss": 0.6645717620849609,
|
| 33 |
+
"eval_runtime": 2.1931,
|
| 34 |
+
"eval_samples_per_second": 102.136,
|
| 35 |
+
"eval_steps_per_second": 3.192,
|
| 36 |
+
"step": 42
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 4.0,
|
| 40 |
+
"eval_accuracy": 0.6785714285714286,
|
| 41 |
+
"eval_loss": 0.659195065498352,
|
| 42 |
+
"eval_runtime": 2.3039,
|
| 43 |
+
"eval_samples_per_second": 97.228,
|
| 44 |
+
"eval_steps_per_second": 3.038,
|
| 45 |
+
"step": 56
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 5.0,
|
| 49 |
+
"eval_accuracy": 0.7142857142857143,
|
| 50 |
+
"eval_loss": 0.6535929441452026,
|
| 51 |
+
"eval_runtime": 2.244,
|
| 52 |
+
"eval_samples_per_second": 99.823,
|
| 53 |
+
"eval_steps_per_second": 3.119,
|
| 54 |
+
"step": 70
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 6.0,
|
| 58 |
+
"eval_accuracy": 0.7455357142857143,
|
| 59 |
+
"eval_loss": 0.6479587554931641,
|
| 60 |
+
"eval_runtime": 2.2785,
|
| 61 |
+
"eval_samples_per_second": 98.312,
|
| 62 |
+
"eval_steps_per_second": 3.072,
|
| 63 |
+
"step": 84
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 7.0,
|
| 67 |
+
"eval_accuracy": 0.7633928571428571,
|
| 68 |
+
"eval_loss": 0.6423625946044922,
|
| 69 |
+
"eval_runtime": 2.2435,
|
| 70 |
+
"eval_samples_per_second": 99.842,
|
| 71 |
+
"eval_steps_per_second": 3.12,
|
| 72 |
+
"step": 98
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 8.0,
|
| 76 |
+
"eval_accuracy": 0.78125,
|
| 77 |
+
"eval_loss": 0.6365941762924194,
|
| 78 |
+
"eval_runtime": 2.2344,
|
| 79 |
+
"eval_samples_per_second": 100.249,
|
| 80 |
+
"eval_steps_per_second": 3.133,
|
| 81 |
+
"step": 112
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 9.0,
|
| 85 |
+
"eval_accuracy": 0.7946428571428571,
|
| 86 |
+
"eval_loss": 0.6308039426803589,
|
| 87 |
+
"eval_runtime": 2.2742,
|
| 88 |
+
"eval_samples_per_second": 98.496,
|
| 89 |
+
"eval_steps_per_second": 3.078,
|
| 90 |
+
"step": 126
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 10.0,
|
| 94 |
+
"eval_accuracy": 0.8125,
|
| 95 |
+
"eval_loss": 0.6248693466186523,
|
| 96 |
+
"eval_runtime": 2.2018,
|
| 97 |
+
"eval_samples_per_second": 101.734,
|
| 98 |
+
"eval_steps_per_second": 3.179,
|
| 99 |
+
"step": 140
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 11.0,
|
| 103 |
+
"eval_accuracy": 0.8214285714285714,
|
| 104 |
+
"eval_loss": 0.6189534068107605,
|
| 105 |
+
"eval_runtime": 2.265,
|
| 106 |
+
"eval_samples_per_second": 98.897,
|
| 107 |
+
"eval_steps_per_second": 3.091,
|
| 108 |
+
"step": 154
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 12.0,
|
| 112 |
+
"eval_accuracy": 0.8348214285714286,
|
| 113 |
+
"eval_loss": 0.6127967834472656,
|
| 114 |
+
"eval_runtime": 2.3514,
|
| 115 |
+
"eval_samples_per_second": 95.261,
|
| 116 |
+
"eval_steps_per_second": 2.977,
|
| 117 |
+
"step": 168
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 13.0,
|
| 121 |
+
"eval_accuracy": 0.8348214285714286,
|
| 122 |
+
"eval_loss": 0.6062906980514526,
|
| 123 |
+
"eval_runtime": 2.227,
|
| 124 |
+
"eval_samples_per_second": 100.583,
|
| 125 |
+
"eval_steps_per_second": 3.143,
|
| 126 |
+
"step": 182
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 14.0,
|
| 130 |
+
"eval_accuracy": 0.84375,
|
| 131 |
+
"eval_loss": 0.5997238755226135,
|
| 132 |
+
"eval_runtime": 2.2875,
|
| 133 |
+
"eval_samples_per_second": 97.924,
|
| 134 |
+
"eval_steps_per_second": 3.06,
|
| 135 |
+
"step": 196
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 15.0,
|
| 139 |
+
"eval_accuracy": 0.8482142857142857,
|
| 140 |
+
"eval_loss": 0.5930039286613464,
|
| 141 |
+
"eval_runtime": 2.2251,
|
| 142 |
+
"eval_samples_per_second": 100.668,
|
| 143 |
+
"eval_steps_per_second": 3.146,
|
| 144 |
+
"step": 210
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 16.0,
|
| 148 |
+
"eval_accuracy": 0.8571428571428571,
|
| 149 |
+
"eval_loss": 0.5864587426185608,
|
| 150 |
+
"eval_runtime": 2.224,
|
| 151 |
+
"eval_samples_per_second": 100.719,
|
| 152 |
+
"eval_steps_per_second": 3.147,
|
| 153 |
+
"step": 224
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 17.0,
|
| 157 |
+
"eval_accuracy": 0.875,
|
| 158 |
+
"eval_loss": 0.5794734358787537,
|
| 159 |
+
"eval_runtime": 2.2613,
|
| 160 |
+
"eval_samples_per_second": 99.057,
|
| 161 |
+
"eval_steps_per_second": 3.096,
|
| 162 |
+
"step": 238
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 18.0,
|
| 166 |
+
"eval_accuracy": 0.8973214285714286,
|
| 167 |
+
"eval_loss": 0.5723555684089661,
|
| 168 |
+
"eval_runtime": 2.2068,
|
| 169 |
+
"eval_samples_per_second": 101.503,
|
| 170 |
+
"eval_steps_per_second": 3.172,
|
| 171 |
+
"step": 252
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 19.0,
|
| 175 |
+
"eval_accuracy": 0.9107142857142857,
|
| 176 |
+
"eval_loss": 0.5650845766067505,
|
| 177 |
+
"eval_runtime": 2.2914,
|
| 178 |
+
"eval_samples_per_second": 97.755,
|
| 179 |
+
"eval_steps_per_second": 3.055,
|
| 180 |
+
"step": 266
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 20.0,
|
| 184 |
+
"eval_accuracy": 0.9107142857142857,
|
| 185 |
+
"eval_loss": 0.5578051209449768,
|
| 186 |
+
"eval_runtime": 2.2659,
|
| 187 |
+
"eval_samples_per_second": 98.857,
|
| 188 |
+
"eval_steps_per_second": 3.089,
|
| 189 |
+
"step": 280
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 21.0,
|
| 193 |
+
"eval_accuracy": 0.9107142857142857,
|
| 194 |
+
"eval_loss": 0.550781786441803,
|
| 195 |
+
"eval_runtime": 2.2624,
|
| 196 |
+
"eval_samples_per_second": 99.009,
|
| 197 |
+
"eval_steps_per_second": 3.094,
|
| 198 |
+
"step": 294
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 22.0,
|
| 202 |
+
"eval_accuracy": 0.9107142857142857,
|
| 203 |
+
"eval_loss": 0.5435484647750854,
|
| 204 |
+
"eval_runtime": 2.2295,
|
| 205 |
+
"eval_samples_per_second": 100.471,
|
| 206 |
+
"eval_steps_per_second": 3.14,
|
| 207 |
+
"step": 308
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 23.0,
|
| 211 |
+
"eval_accuracy": 0.9151785714285714,
|
| 212 |
+
"eval_loss": 0.5361891388893127,
|
| 213 |
+
"eval_runtime": 2.2063,
|
| 214 |
+
"eval_samples_per_second": 101.529,
|
| 215 |
+
"eval_steps_per_second": 3.173,
|
| 216 |
+
"step": 322
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 24.0,
|
| 220 |
+
"eval_accuracy": 0.9151785714285714,
|
| 221 |
+
"eval_loss": 0.5289903283119202,
|
| 222 |
+
"eval_runtime": 2.2591,
|
| 223 |
+
"eval_samples_per_second": 99.153,
|
| 224 |
+
"eval_steps_per_second": 3.099,
|
| 225 |
+
"step": 336
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 25.0,
|
| 229 |
+
"eval_accuracy": 0.9196428571428571,
|
| 230 |
+
"eval_loss": 0.5216715931892395,
|
| 231 |
+
"eval_runtime": 2.2043,
|
| 232 |
+
"eval_samples_per_second": 101.621,
|
| 233 |
+
"eval_steps_per_second": 3.176,
|
| 234 |
+
"step": 350
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 26.0,
|
| 238 |
+
"eval_accuracy": 0.9241071428571429,
|
| 239 |
+
"eval_loss": 0.5144416689872742,
|
| 240 |
+
"eval_runtime": 2.1638,
|
| 241 |
+
"eval_samples_per_second": 103.523,
|
| 242 |
+
"eval_steps_per_second": 3.235,
|
| 243 |
+
"step": 364
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 27.0,
|
| 247 |
+
"eval_accuracy": 0.9241071428571429,
|
| 248 |
+
"eval_loss": 0.5072270631790161,
|
| 249 |
+
"eval_runtime": 2.2434,
|
| 250 |
+
"eval_samples_per_second": 99.851,
|
| 251 |
+
"eval_steps_per_second": 3.12,
|
| 252 |
+
"step": 378
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 28.0,
|
| 256 |
+
"eval_accuracy": 0.9285714285714286,
|
| 257 |
+
"eval_loss": 0.49997568130493164,
|
| 258 |
+
"eval_runtime": 2.2107,
|
| 259 |
+
"eval_samples_per_second": 101.325,
|
| 260 |
+
"eval_steps_per_second": 3.166,
|
| 261 |
+
"step": 392
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 29.0,
|
| 265 |
+
"eval_accuracy": 0.9375,
|
| 266 |
+
"eval_loss": 0.4927251636981964,
|
| 267 |
+
"eval_runtime": 2.3185,
|
| 268 |
+
"eval_samples_per_second": 96.615,
|
| 269 |
+
"eval_steps_per_second": 3.019,
|
| 270 |
+
"step": 406
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 30.0,
|
| 274 |
+
"eval_accuracy": 0.9330357142857143,
|
| 275 |
+
"eval_loss": 0.48569080233573914,
|
| 276 |
+
"eval_runtime": 2.2126,
|
| 277 |
+
"eval_samples_per_second": 101.24,
|
| 278 |
+
"eval_steps_per_second": 3.164,
|
| 279 |
+
"step": 420
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 31.0,
|
| 283 |
+
"eval_accuracy": 0.9330357142857143,
|
| 284 |
+
"eval_loss": 0.47862449288368225,
|
| 285 |
+
"eval_runtime": 2.1706,
|
| 286 |
+
"eval_samples_per_second": 103.196,
|
| 287 |
+
"eval_steps_per_second": 3.225,
|
| 288 |
+
"step": 434
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 32.0,
|
| 292 |
+
"eval_accuracy": 0.9375,
|
| 293 |
+
"eval_loss": 0.47164279222488403,
|
| 294 |
+
"eval_runtime": 2.3354,
|
| 295 |
+
"eval_samples_per_second": 95.915,
|
| 296 |
+
"eval_steps_per_second": 2.997,
|
| 297 |
+
"step": 448
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 33.0,
|
| 301 |
+
"eval_accuracy": 0.9419642857142857,
|
| 302 |
+
"eval_loss": 0.4649314284324646,
|
| 303 |
+
"eval_runtime": 2.1861,
|
| 304 |
+
"eval_samples_per_second": 102.464,
|
| 305 |
+
"eval_steps_per_second": 3.202,
|
| 306 |
+
"step": 462
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 34.0,
|
| 310 |
+
"eval_accuracy": 0.9419642857142857,
|
| 311 |
+
"eval_loss": 0.4582762122154236,
|
| 312 |
+
"eval_runtime": 2.2888,
|
| 313 |
+
"eval_samples_per_second": 97.867,
|
| 314 |
+
"eval_steps_per_second": 3.058,
|
| 315 |
+
"step": 476
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 35.0,
|
| 319 |
+
"eval_accuracy": 0.9419642857142857,
|
| 320 |
+
"eval_loss": 0.45151466131210327,
|
| 321 |
+
"eval_runtime": 2.1977,
|
| 322 |
+
"eval_samples_per_second": 101.923,
|
| 323 |
+
"eval_steps_per_second": 3.185,
|
| 324 |
+
"step": 490
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 35.71,
|
| 328 |
+
"learning_rate": 6.666666666666666e-07,
|
| 329 |
+
"loss": 0.5638,
|
| 330 |
+
"step": 500
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 36.0,
|
| 334 |
+
"eval_accuracy": 0.9419642857142857,
|
| 335 |
+
"eval_loss": 0.44504764676094055,
|
| 336 |
+
"eval_runtime": 2.2179,
|
| 337 |
+
"eval_samples_per_second": 100.997,
|
| 338 |
+
"eval_steps_per_second": 3.156,
|
| 339 |
+
"step": 504
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 37.0,
|
| 343 |
+
"eval_accuracy": 0.9419642857142857,
|
| 344 |
+
"eval_loss": 0.43859127163887024,
|
| 345 |
+
"eval_runtime": 2.2014,
|
| 346 |
+
"eval_samples_per_second": 101.752,
|
| 347 |
+
"eval_steps_per_second": 3.18,
|
| 348 |
+
"step": 518
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 38.0,
|
| 352 |
+
"eval_accuracy": 0.9508928571428571,
|
| 353 |
+
"eval_loss": 0.432064026594162,
|
| 354 |
+
"eval_runtime": 2.2444,
|
| 355 |
+
"eval_samples_per_second": 99.806,
|
| 356 |
+
"eval_steps_per_second": 3.119,
|
| 357 |
+
"step": 532
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 39.0,
|
| 361 |
+
"eval_accuracy": 0.9508928571428571,
|
| 362 |
+
"eval_loss": 0.4258166253566742,
|
| 363 |
+
"eval_runtime": 2.2378,
|
| 364 |
+
"eval_samples_per_second": 100.097,
|
| 365 |
+
"eval_steps_per_second": 3.128,
|
| 366 |
+
"step": 546
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 40.0,
|
| 370 |
+
"eval_accuracy": 0.9508928571428571,
|
| 371 |
+
"eval_loss": 0.4194658696651459,
|
| 372 |
+
"eval_runtime": 2.2746,
|
| 373 |
+
"eval_samples_per_second": 98.48,
|
| 374 |
+
"eval_steps_per_second": 3.078,
|
| 375 |
+
"step": 560
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 41.0,
|
| 379 |
+
"eval_accuracy": 0.9553571428571429,
|
| 380 |
+
"eval_loss": 0.4134449362754822,
|
| 381 |
+
"eval_runtime": 2.1981,
|
| 382 |
+
"eval_samples_per_second": 101.908,
|
| 383 |
+
"eval_steps_per_second": 3.185,
|
| 384 |
+
"step": 574
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 42.0,
|
| 388 |
+
"eval_accuracy": 0.9553571428571429,
|
| 389 |
+
"eval_loss": 0.40731653571128845,
|
| 390 |
+
"eval_runtime": 2.2057,
|
| 391 |
+
"eval_samples_per_second": 101.554,
|
| 392 |
+
"eval_steps_per_second": 3.174,
|
| 393 |
+
"step": 588
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 43.0,
|
| 397 |
+
"eval_accuracy": 0.9553571428571429,
|
| 398 |
+
"eval_loss": 0.4012937545776367,
|
| 399 |
+
"eval_runtime": 2.1965,
|
| 400 |
+
"eval_samples_per_second": 101.979,
|
| 401 |
+
"eval_steps_per_second": 3.187,
|
| 402 |
+
"step": 602
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 44.0,
|
| 406 |
+
"eval_accuracy": 0.9553571428571429,
|
| 407 |
+
"eval_loss": 0.39546194672584534,
|
| 408 |
+
"eval_runtime": 2.2606,
|
| 409 |
+
"eval_samples_per_second": 99.09,
|
| 410 |
+
"eval_steps_per_second": 3.097,
|
| 411 |
+
"step": 616
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 45.0,
|
| 415 |
+
"eval_accuracy": 0.9553571428571429,
|
| 416 |
+
"eval_loss": 0.389827698469162,
|
| 417 |
+
"eval_runtime": 2.1892,
|
| 418 |
+
"eval_samples_per_second": 102.32,
|
| 419 |
+
"eval_steps_per_second": 3.198,
|
| 420 |
+
"step": 630
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 46.0,
|
| 424 |
+
"eval_accuracy": 0.9553571428571429,
|
| 425 |
+
"eval_loss": 0.384507954120636,
|
| 426 |
+
"eval_runtime": 2.2493,
|
| 427 |
+
"eval_samples_per_second": 99.586,
|
| 428 |
+
"eval_steps_per_second": 3.112,
|
| 429 |
+
"step": 644
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 47.0,
|
| 433 |
+
"eval_accuracy": 0.9508928571428571,
|
| 434 |
+
"eval_loss": 0.37909093499183655,
|
| 435 |
+
"eval_runtime": 2.2271,
|
| 436 |
+
"eval_samples_per_second": 100.579,
|
| 437 |
+
"eval_steps_per_second": 3.143,
|
| 438 |
+
"step": 658
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 48.0,
|
| 442 |
+
"eval_accuracy": 0.9508928571428571,
|
| 443 |
+
"eval_loss": 0.3739430606365204,
|
| 444 |
+
"eval_runtime": 2.1894,
|
| 445 |
+
"eval_samples_per_second": 102.31,
|
| 446 |
+
"eval_steps_per_second": 3.197,
|
| 447 |
+
"step": 672
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 49.0,
|
| 451 |
+
"eval_accuracy": 0.9508928571428571,
|
| 452 |
+
"eval_loss": 0.3690197169780731,
|
| 453 |
+
"eval_runtime": 2.2024,
|
| 454 |
+
"eval_samples_per_second": 101.707,
|
| 455 |
+
"eval_steps_per_second": 3.178,
|
| 456 |
+
"step": 686
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 50.0,
|
| 460 |
+
"eval_accuracy": 0.9508928571428571,
|
| 461 |
+
"eval_loss": 0.3642566502094269,
|
| 462 |
+
"eval_runtime": 2.1527,
|
| 463 |
+
"eval_samples_per_second": 104.057,
|
| 464 |
+
"eval_steps_per_second": 3.252,
|
| 465 |
+
"step": 700
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 51.0,
|
| 469 |
+
"eval_accuracy": 0.9553571428571429,
|
| 470 |
+
"eval_loss": 0.35963085293769836,
|
| 471 |
+
"eval_runtime": 2.1982,
|
| 472 |
+
"eval_samples_per_second": 101.904,
|
| 473 |
+
"eval_steps_per_second": 3.184,
|
| 474 |
+
"step": 714
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 52.0,
|
| 478 |
+
"eval_accuracy": 0.9553571428571429,
|
| 479 |
+
"eval_loss": 0.355037122964859,
|
| 480 |
+
"eval_runtime": 2.2747,
|
| 481 |
+
"eval_samples_per_second": 98.474,
|
| 482 |
+
"eval_steps_per_second": 3.077,
|
| 483 |
+
"step": 728
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 53.0,
|
| 487 |
+
"eval_accuracy": 0.9553571428571429,
|
| 488 |
+
"eval_loss": 0.3504486680030823,
|
| 489 |
+
"eval_runtime": 2.2554,
|
| 490 |
+
"eval_samples_per_second": 99.319,
|
| 491 |
+
"eval_steps_per_second": 3.104,
|
| 492 |
+
"step": 742
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 54.0,
|
| 496 |
+
"eval_accuracy": 0.9553571428571429,
|
| 497 |
+
"eval_loss": 0.34610915184020996,
|
| 498 |
+
"eval_runtime": 2.2044,
|
| 499 |
+
"eval_samples_per_second": 101.613,
|
| 500 |
+
"eval_steps_per_second": 3.175,
|
| 501 |
+
"step": 756
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 55.0,
|
| 505 |
+
"eval_accuracy": 0.9553571428571429,
|
| 506 |
+
"eval_loss": 0.3419116139411926,
|
| 507 |
+
"eval_runtime": 2.214,
|
| 508 |
+
"eval_samples_per_second": 101.173,
|
| 509 |
+
"eval_steps_per_second": 3.162,
|
| 510 |
+
"step": 770
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 56.0,
|
| 514 |
+
"eval_accuracy": 0.9553571428571429,
|
| 515 |
+
"eval_loss": 0.3379717469215393,
|
| 516 |
+
"eval_runtime": 2.2916,
|
| 517 |
+
"eval_samples_per_second": 97.746,
|
| 518 |
+
"eval_steps_per_second": 3.055,
|
| 519 |
+
"step": 784
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 57.0,
|
| 523 |
+
"eval_accuracy": 0.9553571428571429,
|
| 524 |
+
"eval_loss": 0.3341245651245117,
|
| 525 |
+
"eval_runtime": 2.2633,
|
| 526 |
+
"eval_samples_per_second": 98.971,
|
| 527 |
+
"eval_steps_per_second": 3.093,
|
| 528 |
+
"step": 798
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 58.0,
|
| 532 |
+
"eval_accuracy": 0.9553571428571429,
|
| 533 |
+
"eval_loss": 0.3303578197956085,
|
| 534 |
+
"eval_runtime": 2.2319,
|
| 535 |
+
"eval_samples_per_second": 100.363,
|
| 536 |
+
"eval_steps_per_second": 3.136,
|
| 537 |
+
"step": 812
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 59.0,
|
| 541 |
+
"eval_accuracy": 0.9553571428571429,
|
| 542 |
+
"eval_loss": 0.326800674200058,
|
| 543 |
+
"eval_runtime": 2.2327,
|
| 544 |
+
"eval_samples_per_second": 100.328,
|
| 545 |
+
"eval_steps_per_second": 3.135,
|
| 546 |
+
"step": 826
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 60.0,
|
| 550 |
+
"eval_accuracy": 0.9553571428571429,
|
| 551 |
+
"eval_loss": 0.3234170377254486,
|
| 552 |
+
"eval_runtime": 2.1782,
|
| 553 |
+
"eval_samples_per_second": 102.837,
|
| 554 |
+
"eval_steps_per_second": 3.214,
|
| 555 |
+
"step": 840
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 61.0,
|
| 559 |
+
"eval_accuracy": 0.9553571428571429,
|
| 560 |
+
"eval_loss": 0.3202018737792969,
|
| 561 |
+
"eval_runtime": 2.1926,
|
| 562 |
+
"eval_samples_per_second": 102.16,
|
| 563 |
+
"eval_steps_per_second": 3.192,
|
| 564 |
+
"step": 854
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 62.0,
|
| 568 |
+
"eval_accuracy": 0.9553571428571429,
|
| 569 |
+
"eval_loss": 0.3171311914920807,
|
| 570 |
+
"eval_runtime": 2.2213,
|
| 571 |
+
"eval_samples_per_second": 100.843,
|
| 572 |
+
"eval_steps_per_second": 3.151,
|
| 573 |
+
"step": 868
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 63.0,
|
| 577 |
+
"eval_accuracy": 0.9553571428571429,
|
| 578 |
+
"eval_loss": 0.3142254054546356,
|
| 579 |
+
"eval_runtime": 2.2136,
|
| 580 |
+
"eval_samples_per_second": 101.194,
|
| 581 |
+
"eval_steps_per_second": 3.162,
|
| 582 |
+
"step": 882
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 64.0,
|
| 586 |
+
"eval_accuracy": 0.9553571428571429,
|
| 587 |
+
"eval_loss": 0.31125956773757935,
|
| 588 |
+
"eval_runtime": 2.3194,
|
| 589 |
+
"eval_samples_per_second": 96.576,
|
| 590 |
+
"eval_steps_per_second": 3.018,
|
| 591 |
+
"step": 896
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 65.0,
|
| 595 |
+
"eval_accuracy": 0.9553571428571429,
|
| 596 |
+
"eval_loss": 0.30836549401283264,
|
| 597 |
+
"eval_runtime": 2.291,
|
| 598 |
+
"eval_samples_per_second": 97.772,
|
| 599 |
+
"eval_steps_per_second": 3.055,
|
| 600 |
+
"step": 910
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 66.0,
|
| 604 |
+
"eval_accuracy": 0.9553571428571429,
|
| 605 |
+
"eval_loss": 0.3055935800075531,
|
| 606 |
+
"eval_runtime": 2.2305,
|
| 607 |
+
"eval_samples_per_second": 100.425,
|
| 608 |
+
"eval_steps_per_second": 3.138,
|
| 609 |
+
"step": 924
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 67.0,
|
| 613 |
+
"eval_accuracy": 0.9553571428571429,
|
| 614 |
+
"eval_loss": 0.30298447608947754,
|
| 615 |
+
"eval_runtime": 2.2233,
|
| 616 |
+
"eval_samples_per_second": 100.751,
|
| 617 |
+
"eval_steps_per_second": 3.148,
|
| 618 |
+
"step": 938
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 68.0,
|
| 622 |
+
"eval_accuracy": 0.9553571428571429,
|
| 623 |
+
"eval_loss": 0.3004678189754486,
|
| 624 |
+
"eval_runtime": 2.2309,
|
| 625 |
+
"eval_samples_per_second": 100.406,
|
| 626 |
+
"eval_steps_per_second": 3.138,
|
| 627 |
+
"step": 952
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 69.0,
|
| 631 |
+
"eval_accuracy": 0.9553571428571429,
|
| 632 |
+
"eval_loss": 0.29803305864334106,
|
| 633 |
+
"eval_runtime": 2.2138,
|
| 634 |
+
"eval_samples_per_second": 101.185,
|
| 635 |
+
"eval_steps_per_second": 3.162,
|
| 636 |
+
"step": 966
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 70.0,
|
| 640 |
+
"eval_accuracy": 0.9553571428571429,
|
| 641 |
+
"eval_loss": 0.29582586884498596,
|
| 642 |
+
"eval_runtime": 2.192,
|
| 643 |
+
"eval_samples_per_second": 102.189,
|
| 644 |
+
"eval_steps_per_second": 3.193,
|
| 645 |
+
"step": 980
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 71.0,
|
| 649 |
+
"eval_accuracy": 0.9553571428571429,
|
| 650 |
+
"eval_loss": 0.2937617897987366,
|
| 651 |
+
"eval_runtime": 2.1747,
|
| 652 |
+
"eval_samples_per_second": 103.004,
|
| 653 |
+
"eval_steps_per_second": 3.219,
|
| 654 |
+
"step": 994
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 71.43,
|
| 658 |
+
"learning_rate": 2.962962962962963e-07,
|
| 659 |
+
"loss": 0.3154,
|
| 660 |
+
"step": 1000
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 72.0,
|
| 664 |
+
"eval_accuracy": 0.9553571428571429,
|
| 665 |
+
"eval_loss": 0.29180124402046204,
|
| 666 |
+
"eval_runtime": 2.2133,
|
| 667 |
+
"eval_samples_per_second": 101.205,
|
| 668 |
+
"eval_steps_per_second": 3.163,
|
| 669 |
+
"step": 1008
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 73.0,
|
| 673 |
+
"eval_accuracy": 0.9553571428571429,
|
| 674 |
+
"eval_loss": 0.28983354568481445,
|
| 675 |
+
"eval_runtime": 2.2226,
|
| 676 |
+
"eval_samples_per_second": 100.781,
|
| 677 |
+
"eval_steps_per_second": 3.149,
|
| 678 |
+
"step": 1022
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 74.0,
|
| 682 |
+
"eval_accuracy": 0.9553571428571429,
|
| 683 |
+
"eval_loss": 0.2879674732685089,
|
| 684 |
+
"eval_runtime": 2.1942,
|
| 685 |
+
"eval_samples_per_second": 102.086,
|
| 686 |
+
"eval_steps_per_second": 3.19,
|
| 687 |
+
"step": 1036
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 75.0,
|
| 691 |
+
"eval_accuracy": 0.9553571428571429,
|
| 692 |
+
"eval_loss": 0.2861447036266327,
|
| 693 |
+
"eval_runtime": 2.2049,
|
| 694 |
+
"eval_samples_per_second": 101.59,
|
| 695 |
+
"eval_steps_per_second": 3.175,
|
| 696 |
+
"step": 1050
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 76.0,
|
| 700 |
+
"eval_accuracy": 0.9553571428571429,
|
| 701 |
+
"eval_loss": 0.2843897044658661,
|
| 702 |
+
"eval_runtime": 2.1835,
|
| 703 |
+
"eval_samples_per_second": 102.586,
|
| 704 |
+
"eval_steps_per_second": 3.206,
|
| 705 |
+
"step": 1064
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 77.0,
|
| 709 |
+
"eval_accuracy": 0.9553571428571429,
|
| 710 |
+
"eval_loss": 0.282759428024292,
|
| 711 |
+
"eval_runtime": 2.2185,
|
| 712 |
+
"eval_samples_per_second": 100.967,
|
| 713 |
+
"eval_steps_per_second": 3.155,
|
| 714 |
+
"step": 1078
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 78.0,
|
| 718 |
+
"eval_accuracy": 0.9553571428571429,
|
| 719 |
+
"eval_loss": 0.28130826354026794,
|
| 720 |
+
"eval_runtime": 2.2043,
|
| 721 |
+
"eval_samples_per_second": 101.618,
|
| 722 |
+
"eval_steps_per_second": 3.176,
|
| 723 |
+
"step": 1092
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 79.0,
|
| 727 |
+
"eval_accuracy": 0.9553571428571429,
|
| 728 |
+
"eval_loss": 0.2799042761325836,
|
| 729 |
+
"eval_runtime": 2.237,
|
| 730 |
+
"eval_samples_per_second": 100.135,
|
| 731 |
+
"eval_steps_per_second": 3.129,
|
| 732 |
+
"step": 1106
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 80.0,
|
| 736 |
+
"eval_accuracy": 0.9553571428571429,
|
| 737 |
+
"eval_loss": 0.2784968912601471,
|
| 738 |
+
"eval_runtime": 2.1855,
|
| 739 |
+
"eval_samples_per_second": 102.494,
|
| 740 |
+
"eval_steps_per_second": 3.203,
|
| 741 |
+
"step": 1120
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 81.0,
|
| 745 |
+
"eval_accuracy": 0.9553571428571429,
|
| 746 |
+
"eval_loss": 0.27718737721443176,
|
| 747 |
+
"eval_runtime": 2.1979,
|
| 748 |
+
"eval_samples_per_second": 101.917,
|
| 749 |
+
"eval_steps_per_second": 3.185,
|
| 750 |
+
"step": 1134
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 82.0,
|
| 754 |
+
"eval_accuracy": 0.9553571428571429,
|
| 755 |
+
"eval_loss": 0.27603691816329956,
|
| 756 |
+
"eval_runtime": 2.2236,
|
| 757 |
+
"eval_samples_per_second": 100.737,
|
| 758 |
+
"eval_steps_per_second": 3.148,
|
| 759 |
+
"step": 1148
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 83.0,
|
| 763 |
+
"eval_accuracy": 0.9553571428571429,
|
| 764 |
+
"eval_loss": 0.2748844027519226,
|
| 765 |
+
"eval_runtime": 2.2411,
|
| 766 |
+
"eval_samples_per_second": 99.95,
|
| 767 |
+
"eval_steps_per_second": 3.123,
|
| 768 |
+
"step": 1162
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 84.0,
|
| 772 |
+
"eval_accuracy": 0.9553571428571429,
|
| 773 |
+
"eval_loss": 0.27382367849349976,
|
| 774 |
+
"eval_runtime": 2.2069,
|
| 775 |
+
"eval_samples_per_second": 101.498,
|
| 776 |
+
"eval_steps_per_second": 3.172,
|
| 777 |
+
"step": 1176
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 85.0,
|
| 781 |
+
"eval_accuracy": 0.9553571428571429,
|
| 782 |
+
"eval_loss": 0.27282026410102844,
|
| 783 |
+
"eval_runtime": 2.2306,
|
| 784 |
+
"eval_samples_per_second": 100.422,
|
| 785 |
+
"eval_steps_per_second": 3.138,
|
| 786 |
+
"step": 1190
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 86.0,
|
| 790 |
+
"eval_accuracy": 0.9553571428571429,
|
| 791 |
+
"eval_loss": 0.2719047963619232,
|
| 792 |
+
"eval_runtime": 2.335,
|
| 793 |
+
"eval_samples_per_second": 95.931,
|
| 794 |
+
"eval_steps_per_second": 2.998,
|
| 795 |
+
"step": 1204
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 87.0,
|
| 799 |
+
"eval_accuracy": 0.9553571428571429,
|
| 800 |
+
"eval_loss": 0.2710733115673065,
|
| 801 |
+
"eval_runtime": 2.2352,
|
| 802 |
+
"eval_samples_per_second": 100.216,
|
| 803 |
+
"eval_steps_per_second": 3.132,
|
| 804 |
+
"step": 1218
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 88.0,
|
| 808 |
+
"eval_accuracy": 0.9553571428571429,
|
| 809 |
+
"eval_loss": 0.27031686902046204,
|
| 810 |
+
"eval_runtime": 2.2134,
|
| 811 |
+
"eval_samples_per_second": 101.201,
|
| 812 |
+
"eval_steps_per_second": 3.163,
|
| 813 |
+
"step": 1232
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 89.0,
|
| 817 |
+
"eval_accuracy": 0.9553571428571429,
|
| 818 |
+
"eval_loss": 0.2695934474468231,
|
| 819 |
+
"eval_runtime": 2.2019,
|
| 820 |
+
"eval_samples_per_second": 101.73,
|
| 821 |
+
"eval_steps_per_second": 3.179,
|
| 822 |
+
"step": 1246
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 90.0,
|
| 826 |
+
"eval_accuracy": 0.9553571428571429,
|
| 827 |
+
"eval_loss": 0.2689376771450043,
|
| 828 |
+
"eval_runtime": 2.1925,
|
| 829 |
+
"eval_samples_per_second": 102.166,
|
| 830 |
+
"eval_steps_per_second": 3.193,
|
| 831 |
+
"step": 1260
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 91.0,
|
| 835 |
+
"eval_accuracy": 0.9598214285714286,
|
| 836 |
+
"eval_loss": 0.2683904469013214,
|
| 837 |
+
"eval_runtime": 2.2557,
|
| 838 |
+
"eval_samples_per_second": 99.303,
|
| 839 |
+
"eval_steps_per_second": 3.103,
|
| 840 |
+
"step": 1274
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 92.0,
|
| 844 |
+
"eval_accuracy": 0.9598214285714286,
|
| 845 |
+
"eval_loss": 0.267875611782074,
|
| 846 |
+
"eval_runtime": 2.2017,
|
| 847 |
+
"eval_samples_per_second": 101.738,
|
| 848 |
+
"eval_steps_per_second": 3.179,
|
| 849 |
+
"step": 1288
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 93.0,
|
| 853 |
+
"eval_accuracy": 0.9598214285714286,
|
| 854 |
+
"eval_loss": 0.2674213945865631,
|
| 855 |
+
"eval_runtime": 2.2295,
|
| 856 |
+
"eval_samples_per_second": 100.473,
|
| 857 |
+
"eval_steps_per_second": 3.14,
|
| 858 |
+
"step": 1302
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 94.0,
|
| 862 |
+
"eval_accuracy": 0.9598214285714286,
|
| 863 |
+
"eval_loss": 0.26701620221138,
|
| 864 |
+
"eval_runtime": 2.2017,
|
| 865 |
+
"eval_samples_per_second": 101.74,
|
| 866 |
+
"eval_steps_per_second": 3.179,
|
| 867 |
+
"step": 1316
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 95.0,
|
| 871 |
+
"eval_accuracy": 0.9598214285714286,
|
| 872 |
+
"eval_loss": 0.2666829526424408,
|
| 873 |
+
"eval_runtime": 2.2533,
|
| 874 |
+
"eval_samples_per_second": 99.409,
|
| 875 |
+
"eval_steps_per_second": 3.107,
|
| 876 |
+
"step": 1330
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 96.0,
|
| 880 |
+
"eval_accuracy": 0.9598214285714286,
|
| 881 |
+
"eval_loss": 0.266412615776062,
|
| 882 |
+
"eval_runtime": 2.2717,
|
| 883 |
+
"eval_samples_per_second": 98.603,
|
| 884 |
+
"eval_steps_per_second": 3.081,
|
| 885 |
+
"step": 1344
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 97.0,
|
| 889 |
+
"eval_accuracy": 0.9598214285714286,
|
| 890 |
+
"eval_loss": 0.2662082016468048,
|
| 891 |
+
"eval_runtime": 2.2818,
|
| 892 |
+
"eval_samples_per_second": 98.168,
|
| 893 |
+
"eval_steps_per_second": 3.068,
|
| 894 |
+
"step": 1358
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 98.0,
|
| 898 |
+
"eval_accuracy": 0.9598214285714286,
|
| 899 |
+
"eval_loss": 0.2660551369190216,
|
| 900 |
+
"eval_runtime": 2.3429,
|
| 901 |
+
"eval_samples_per_second": 95.607,
|
| 902 |
+
"eval_steps_per_second": 2.988,
|
| 903 |
+
"step": 1372
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 99.0,
|
| 907 |
+
"eval_accuracy": 0.9598214285714286,
|
| 908 |
+
"eval_loss": 0.2659624516963959,
|
| 909 |
+
"eval_runtime": 2.1858,
|
| 910 |
+
"eval_samples_per_second": 102.479,
|
| 911 |
+
"eval_steps_per_second": 3.202,
|
| 912 |
+
"step": 1386
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 100.0,
|
| 916 |
+
"eval_accuracy": 0.9598214285714286,
|
| 917 |
+
"eval_loss": 0.2659357786178589,
|
| 918 |
+
"eval_runtime": 2.212,
|
| 919 |
+
"eval_samples_per_second": 101.268,
|
| 920 |
+
"eval_steps_per_second": 3.165,
|
| 921 |
+
"step": 1400
|
| 922 |
+
}
|
| 923 |
+
],
|
| 924 |
+
"logging_steps": 500,
|
| 925 |
+
"max_steps": 1400,
|
| 926 |
+
"num_train_epochs": 100,
|
| 927 |
+
"save_steps": 500,
|
| 928 |
+
"total_flos": 6.943282269467443e+18,
|
| 929 |
+
"trial_name": null,
|
| 930 |
+
"trial_params": null
|
| 931 |
+
}
|
checkpoint-1400/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28dcd3f688e2a05864d4128ea4a295370c2c16728b4b9157b715afdb36bc2cad
|
| 3 |
+
size 4091
|
config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "notsmoking",
|
| 13 |
+
"1": "smoking"
|
| 14 |
+
},
|
| 15 |
+
"image_size": 224,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 3072,
|
| 18 |
+
"label2id": {
|
| 19 |
+
"notsmoking": 0,
|
| 20 |
+
"smoking": 1
|
| 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": "4.35.2"
|
| 32 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9fb95ff652cd152efa361ce3b9aa37d471539c6407c4cf663b87eef8dc832fd9
|
| 3 |
+
size 343223968
|
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 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28dcd3f688e2a05864d4128ea4a295370c2c16728b4b9157b715afdb36bc2cad
|
| 3 |
+
size 4091
|