Upload folder using huggingface_hub
Browse files- checkpoint-4900/config.json +38 -0
- checkpoint-4900/model.safetensors +3 -0
- checkpoint-4900/optimizer.pt +3 -0
- checkpoint-4900/preprocessor_config.json +22 -0
- checkpoint-4900/rng_state.pth +3 -0
- checkpoint-4900/scheduler.pt +3 -0
- checkpoint-4900/trainer_state.json +1096 -0
- checkpoint-4900/training_args.bin +3 -0
- config.json +38 -0
- model.safetensors +3 -0
- preprocessor_config.json +22 -0
- training_args.bin +3 -0
checkpoint-4900/config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "curly",
|
| 13 |
+
"1": "dreadlocks",
|
| 14 |
+
"2": "kinky",
|
| 15 |
+
"3": "straight",
|
| 16 |
+
"4": "wavy"
|
| 17 |
+
},
|
| 18 |
+
"image_size": 224,
|
| 19 |
+
"initializer_range": 0.02,
|
| 20 |
+
"intermediate_size": 3072,
|
| 21 |
+
"label2id": {
|
| 22 |
+
"curly": 0,
|
| 23 |
+
"dreadlocks": 1,
|
| 24 |
+
"kinky": 2,
|
| 25 |
+
"straight": 3,
|
| 26 |
+
"wavy": 4
|
| 27 |
+
},
|
| 28 |
+
"layer_norm_eps": 1e-12,
|
| 29 |
+
"model_type": "vit",
|
| 30 |
+
"num_attention_heads": 12,
|
| 31 |
+
"num_channels": 3,
|
| 32 |
+
"num_hidden_layers": 12,
|
| 33 |
+
"patch_size": 16,
|
| 34 |
+
"problem_type": "single_label_classification",
|
| 35 |
+
"qkv_bias": true,
|
| 36 |
+
"torch_dtype": "float32",
|
| 37 |
+
"transformers_version": "4.45.2"
|
| 38 |
+
}
|
checkpoint-4900/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:880839388dee970fc3c9a41a9ea13295c6093fea1653d77780af5a05743c9cf2
|
| 3 |
+
size 343233204
|
checkpoint-4900/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a02b69d9cb3517d77d6efff670a7edd44fb0565ba5a9d703e54737429f44a9ff
|
| 3 |
+
size 686587322
|
checkpoint-4900/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-4900/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fa6d96f7ee624f7c57ede23ba0321ef010817bd064e65eb4db242f525babee44
|
| 3 |
+
size 14244
|
checkpoint-4900/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc8ff882b25de7a093a1ca37813f32aa219aeac50ec93044513164d388e33b66
|
| 3 |
+
size 1064
|
checkpoint-4900/trainer_state.json
ADDED
|
@@ -0,0 +1,1096 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.5298713445663452,
|
| 3 |
+
"best_model_checkpoint": "hair_type_image_detection/checkpoint-4900",
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4900,
|
| 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": 0.17704280155642024,
|
| 14 |
+
"eval_loss": 1.6219326257705688,
|
| 15 |
+
"eval_model_preparation_time": 0.0053,
|
| 16 |
+
"eval_runtime": 16.3274,
|
| 17 |
+
"eval_samples_per_second": 62.962,
|
| 18 |
+
"eval_steps_per_second": 7.901,
|
| 19 |
+
"step": 49
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 2.0,
|
| 23 |
+
"eval_accuracy": 0.22568093385214008,
|
| 24 |
+
"eval_loss": 1.6067438125610352,
|
| 25 |
+
"eval_model_preparation_time": 0.0053,
|
| 26 |
+
"eval_runtime": 16.2468,
|
| 27 |
+
"eval_samples_per_second": 63.274,
|
| 28 |
+
"eval_steps_per_second": 7.94,
|
| 29 |
+
"step": 98
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 3.0,
|
| 33 |
+
"eval_accuracy": 0.27529182879377434,
|
| 34 |
+
"eval_loss": 1.5912580490112305,
|
| 35 |
+
"eval_model_preparation_time": 0.0053,
|
| 36 |
+
"eval_runtime": 16.4806,
|
| 37 |
+
"eval_samples_per_second": 62.377,
|
| 38 |
+
"eval_steps_per_second": 7.827,
|
| 39 |
+
"step": 147
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 4.0,
|
| 43 |
+
"eval_accuracy": 0.33073929961089493,
|
| 44 |
+
"eval_loss": 1.5748045444488525,
|
| 45 |
+
"eval_model_preparation_time": 0.0053,
|
| 46 |
+
"eval_runtime": 16.0231,
|
| 47 |
+
"eval_samples_per_second": 64.157,
|
| 48 |
+
"eval_steps_per_second": 8.051,
|
| 49 |
+
"step": 196
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 5.0,
|
| 53 |
+
"eval_accuracy": 0.39299610894941633,
|
| 54 |
+
"eval_loss": 1.5561603307724,
|
| 55 |
+
"eval_model_preparation_time": 0.0053,
|
| 56 |
+
"eval_runtime": 16.3268,
|
| 57 |
+
"eval_samples_per_second": 62.964,
|
| 58 |
+
"eval_steps_per_second": 7.901,
|
| 59 |
+
"step": 245
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 6.0,
|
| 63 |
+
"eval_accuracy": 0.46400778210116733,
|
| 64 |
+
"eval_loss": 1.5354892015457153,
|
| 65 |
+
"eval_model_preparation_time": 0.0053,
|
| 66 |
+
"eval_runtime": 16.3682,
|
| 67 |
+
"eval_samples_per_second": 62.805,
|
| 68 |
+
"eval_steps_per_second": 7.881,
|
| 69 |
+
"step": 294
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 7.0,
|
| 73 |
+
"eval_accuracy": 0.5223735408560312,
|
| 74 |
+
"eval_loss": 1.5132097005844116,
|
| 75 |
+
"eval_model_preparation_time": 0.0053,
|
| 76 |
+
"eval_runtime": 16.1784,
|
| 77 |
+
"eval_samples_per_second": 63.542,
|
| 78 |
+
"eval_steps_per_second": 7.974,
|
| 79 |
+
"step": 343
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 8.0,
|
| 83 |
+
"eval_accuracy": 0.5700389105058365,
|
| 84 |
+
"eval_loss": 1.4904340505599976,
|
| 85 |
+
"eval_model_preparation_time": 0.0053,
|
| 86 |
+
"eval_runtime": 15.9362,
|
| 87 |
+
"eval_samples_per_second": 64.507,
|
| 88 |
+
"eval_steps_per_second": 8.095,
|
| 89 |
+
"step": 392
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 9.0,
|
| 93 |
+
"eval_accuracy": 0.6138132295719845,
|
| 94 |
+
"eval_loss": 1.4679334163665771,
|
| 95 |
+
"eval_model_preparation_time": 0.0053,
|
| 96 |
+
"eval_runtime": 16.4082,
|
| 97 |
+
"eval_samples_per_second": 62.652,
|
| 98 |
+
"eval_steps_per_second": 7.862,
|
| 99 |
+
"step": 441
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 10.0,
|
| 103 |
+
"eval_accuracy": 0.6517509727626459,
|
| 104 |
+
"eval_loss": 1.4443553686141968,
|
| 105 |
+
"eval_model_preparation_time": 0.0053,
|
| 106 |
+
"eval_runtime": 16.0563,
|
| 107 |
+
"eval_samples_per_second": 64.025,
|
| 108 |
+
"eval_steps_per_second": 8.034,
|
| 109 |
+
"step": 490
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 10.204081632653061,
|
| 113 |
+
"grad_norm": 1.3639816045761108,
|
| 114 |
+
"learning_rate": 9.072164948453608e-07,
|
| 115 |
+
"loss": 1.5205,
|
| 116 |
+
"step": 500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 11.0,
|
| 120 |
+
"eval_accuracy": 0.683852140077821,
|
| 121 |
+
"eval_loss": 1.4206485748291016,
|
| 122 |
+
"eval_model_preparation_time": 0.0053,
|
| 123 |
+
"eval_runtime": 16.0088,
|
| 124 |
+
"eval_samples_per_second": 64.214,
|
| 125 |
+
"eval_steps_per_second": 8.058,
|
| 126 |
+
"step": 539
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 12.0,
|
| 130 |
+
"eval_accuracy": 0.7178988326848249,
|
| 131 |
+
"eval_loss": 1.3944063186645508,
|
| 132 |
+
"eval_model_preparation_time": 0.0053,
|
| 133 |
+
"eval_runtime": 15.9211,
|
| 134 |
+
"eval_samples_per_second": 64.568,
|
| 135 |
+
"eval_steps_per_second": 8.102,
|
| 136 |
+
"step": 588
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 13.0,
|
| 140 |
+
"eval_accuracy": 0.7441634241245136,
|
| 141 |
+
"eval_loss": 1.3676106929779053,
|
| 142 |
+
"eval_model_preparation_time": 0.0053,
|
| 143 |
+
"eval_runtime": 16.1365,
|
| 144 |
+
"eval_samples_per_second": 63.706,
|
| 145 |
+
"eval_steps_per_second": 7.994,
|
| 146 |
+
"step": 637
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 14.0,
|
| 150 |
+
"eval_accuracy": 0.7684824902723736,
|
| 151 |
+
"eval_loss": 1.3409146070480347,
|
| 152 |
+
"eval_model_preparation_time": 0.0053,
|
| 153 |
+
"eval_runtime": 16.0749,
|
| 154 |
+
"eval_samples_per_second": 63.95,
|
| 155 |
+
"eval_steps_per_second": 8.025,
|
| 156 |
+
"step": 686
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 15.0,
|
| 160 |
+
"eval_accuracy": 0.7830739299610895,
|
| 161 |
+
"eval_loss": 1.312709927558899,
|
| 162 |
+
"eval_model_preparation_time": 0.0053,
|
| 163 |
+
"eval_runtime": 16.1852,
|
| 164 |
+
"eval_samples_per_second": 63.515,
|
| 165 |
+
"eval_steps_per_second": 7.97,
|
| 166 |
+
"step": 735
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 16.0,
|
| 170 |
+
"eval_accuracy": 0.7976653696498055,
|
| 171 |
+
"eval_loss": 1.2852919101715088,
|
| 172 |
+
"eval_model_preparation_time": 0.0053,
|
| 173 |
+
"eval_runtime": 16.5033,
|
| 174 |
+
"eval_samples_per_second": 62.291,
|
| 175 |
+
"eval_steps_per_second": 7.817,
|
| 176 |
+
"step": 784
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 17.0,
|
| 180 |
+
"eval_accuracy": 0.8151750972762646,
|
| 181 |
+
"eval_loss": 1.2579333782196045,
|
| 182 |
+
"eval_model_preparation_time": 0.0053,
|
| 183 |
+
"eval_runtime": 16.716,
|
| 184 |
+
"eval_samples_per_second": 61.498,
|
| 185 |
+
"eval_steps_per_second": 7.717,
|
| 186 |
+
"step": 833
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 18.0,
|
| 190 |
+
"eval_accuracy": 0.8258754863813229,
|
| 191 |
+
"eval_loss": 1.2285292148590088,
|
| 192 |
+
"eval_model_preparation_time": 0.0053,
|
| 193 |
+
"eval_runtime": 16.3546,
|
| 194 |
+
"eval_samples_per_second": 62.857,
|
| 195 |
+
"eval_steps_per_second": 7.888,
|
| 196 |
+
"step": 882
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 19.0,
|
| 200 |
+
"eval_accuracy": 0.8317120622568094,
|
| 201 |
+
"eval_loss": 1.2019933462142944,
|
| 202 |
+
"eval_model_preparation_time": 0.0053,
|
| 203 |
+
"eval_runtime": 16.435,
|
| 204 |
+
"eval_samples_per_second": 62.549,
|
| 205 |
+
"eval_steps_per_second": 7.849,
|
| 206 |
+
"step": 931
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 20.0,
|
| 210 |
+
"eval_accuracy": 0.8404669260700389,
|
| 211 |
+
"eval_loss": 1.1748104095458984,
|
| 212 |
+
"eval_model_preparation_time": 0.0053,
|
| 213 |
+
"eval_runtime": 16.4968,
|
| 214 |
+
"eval_samples_per_second": 62.315,
|
| 215 |
+
"eval_steps_per_second": 7.82,
|
| 216 |
+
"step": 980
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 20.408163265306122,
|
| 220 |
+
"grad_norm": 1.599041223526001,
|
| 221 |
+
"learning_rate": 8.041237113402062e-07,
|
| 222 |
+
"loss": 1.2347,
|
| 223 |
+
"step": 1000
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 21.0,
|
| 227 |
+
"eval_accuracy": 0.8492217898832685,
|
| 228 |
+
"eval_loss": 1.1474989652633667,
|
| 229 |
+
"eval_model_preparation_time": 0.0053,
|
| 230 |
+
"eval_runtime": 16.0812,
|
| 231 |
+
"eval_samples_per_second": 63.925,
|
| 232 |
+
"eval_steps_per_second": 8.022,
|
| 233 |
+
"step": 1029
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 22.0,
|
| 237 |
+
"eval_accuracy": 0.8521400778210116,
|
| 238 |
+
"eval_loss": 1.121047019958496,
|
| 239 |
+
"eval_model_preparation_time": 0.0053,
|
| 240 |
+
"eval_runtime": 16.0658,
|
| 241 |
+
"eval_samples_per_second": 63.987,
|
| 242 |
+
"eval_steps_per_second": 8.03,
|
| 243 |
+
"step": 1078
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 23.0,
|
| 247 |
+
"eval_accuracy": 0.8628404669260701,
|
| 248 |
+
"eval_loss": 1.0952664613723755,
|
| 249 |
+
"eval_model_preparation_time": 0.0053,
|
| 250 |
+
"eval_runtime": 16.283,
|
| 251 |
+
"eval_samples_per_second": 63.133,
|
| 252 |
+
"eval_steps_per_second": 7.922,
|
| 253 |
+
"step": 1127
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 24.0,
|
| 257 |
+
"eval_accuracy": 0.8638132295719845,
|
| 258 |
+
"eval_loss": 1.070285439491272,
|
| 259 |
+
"eval_model_preparation_time": 0.0053,
|
| 260 |
+
"eval_runtime": 15.9544,
|
| 261 |
+
"eval_samples_per_second": 64.434,
|
| 262 |
+
"eval_steps_per_second": 8.086,
|
| 263 |
+
"step": 1176
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 25.0,
|
| 267 |
+
"eval_accuracy": 0.8638132295719845,
|
| 268 |
+
"eval_loss": 1.046401023864746,
|
| 269 |
+
"eval_model_preparation_time": 0.0053,
|
| 270 |
+
"eval_runtime": 16.1019,
|
| 271 |
+
"eval_samples_per_second": 63.843,
|
| 272 |
+
"eval_steps_per_second": 8.011,
|
| 273 |
+
"step": 1225
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 26.0,
|
| 277 |
+
"eval_accuracy": 0.8686770428015564,
|
| 278 |
+
"eval_loss": 1.0234460830688477,
|
| 279 |
+
"eval_model_preparation_time": 0.0053,
|
| 280 |
+
"eval_runtime": 16.0897,
|
| 281 |
+
"eval_samples_per_second": 63.892,
|
| 282 |
+
"eval_steps_per_second": 8.018,
|
| 283 |
+
"step": 1274
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 27.0,
|
| 287 |
+
"eval_accuracy": 0.8696498054474708,
|
| 288 |
+
"eval_loss": 1.0003376007080078,
|
| 289 |
+
"eval_model_preparation_time": 0.0053,
|
| 290 |
+
"eval_runtime": 16.2835,
|
| 291 |
+
"eval_samples_per_second": 63.131,
|
| 292 |
+
"eval_steps_per_second": 7.922,
|
| 293 |
+
"step": 1323
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 28.0,
|
| 297 |
+
"eval_accuracy": 0.8696498054474708,
|
| 298 |
+
"eval_loss": 0.9786379337310791,
|
| 299 |
+
"eval_model_preparation_time": 0.0053,
|
| 300 |
+
"eval_runtime": 16.2428,
|
| 301 |
+
"eval_samples_per_second": 63.29,
|
| 302 |
+
"eval_steps_per_second": 7.942,
|
| 303 |
+
"step": 1372
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 29.0,
|
| 307 |
+
"eval_accuracy": 0.8715953307392996,
|
| 308 |
+
"eval_loss": 0.9573078751564026,
|
| 309 |
+
"eval_model_preparation_time": 0.0053,
|
| 310 |
+
"eval_runtime": 15.9987,
|
| 311 |
+
"eval_samples_per_second": 64.255,
|
| 312 |
+
"eval_steps_per_second": 8.063,
|
| 313 |
+
"step": 1421
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 30.0,
|
| 317 |
+
"eval_accuracy": 0.872568093385214,
|
| 318 |
+
"eval_loss": 0.9362879395484924,
|
| 319 |
+
"eval_model_preparation_time": 0.0053,
|
| 320 |
+
"eval_runtime": 16.1691,
|
| 321 |
+
"eval_samples_per_second": 63.578,
|
| 322 |
+
"eval_steps_per_second": 7.978,
|
| 323 |
+
"step": 1470
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 30.612244897959183,
|
| 327 |
+
"grad_norm": 1.5271226167678833,
|
| 328 |
+
"learning_rate": 7.010309278350515e-07,
|
| 329 |
+
"loss": 0.9467,
|
| 330 |
+
"step": 1500
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 31.0,
|
| 334 |
+
"eval_accuracy": 0.8754863813229572,
|
| 335 |
+
"eval_loss": 0.9171885848045349,
|
| 336 |
+
"eval_model_preparation_time": 0.0053,
|
| 337 |
+
"eval_runtime": 16.3696,
|
| 338 |
+
"eval_samples_per_second": 62.799,
|
| 339 |
+
"eval_steps_per_second": 7.88,
|
| 340 |
+
"step": 1519
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 32.0,
|
| 344 |
+
"eval_accuracy": 0.8803501945525292,
|
| 345 |
+
"eval_loss": 0.8980955481529236,
|
| 346 |
+
"eval_model_preparation_time": 0.0053,
|
| 347 |
+
"eval_runtime": 16.4392,
|
| 348 |
+
"eval_samples_per_second": 62.533,
|
| 349 |
+
"eval_steps_per_second": 7.847,
|
| 350 |
+
"step": 1568
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 33.0,
|
| 354 |
+
"eval_accuracy": 0.8861867704280155,
|
| 355 |
+
"eval_loss": 0.8809771537780762,
|
| 356 |
+
"eval_model_preparation_time": 0.0053,
|
| 357 |
+
"eval_runtime": 16.4449,
|
| 358 |
+
"eval_samples_per_second": 62.512,
|
| 359 |
+
"eval_steps_per_second": 7.844,
|
| 360 |
+
"step": 1617
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 34.0,
|
| 364 |
+
"eval_accuracy": 0.8861867704280155,
|
| 365 |
+
"eval_loss": 0.8635137677192688,
|
| 366 |
+
"eval_model_preparation_time": 0.0053,
|
| 367 |
+
"eval_runtime": 16.4485,
|
| 368 |
+
"eval_samples_per_second": 62.498,
|
| 369 |
+
"eval_steps_per_second": 7.843,
|
| 370 |
+
"step": 1666
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 35.0,
|
| 374 |
+
"eval_accuracy": 0.8881322957198443,
|
| 375 |
+
"eval_loss": 0.8468510508537292,
|
| 376 |
+
"eval_model_preparation_time": 0.0053,
|
| 377 |
+
"eval_runtime": 16.2536,
|
| 378 |
+
"eval_samples_per_second": 63.247,
|
| 379 |
+
"eval_steps_per_second": 7.937,
|
| 380 |
+
"step": 1715
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 36.0,
|
| 384 |
+
"eval_accuracy": 0.8881322957198443,
|
| 385 |
+
"eval_loss": 0.8328165411949158,
|
| 386 |
+
"eval_model_preparation_time": 0.0053,
|
| 387 |
+
"eval_runtime": 16.503,
|
| 388 |
+
"eval_samples_per_second": 62.292,
|
| 389 |
+
"eval_steps_per_second": 7.817,
|
| 390 |
+
"step": 1764
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 37.0,
|
| 394 |
+
"eval_accuracy": 0.8881322957198443,
|
| 395 |
+
"eval_loss": 0.8181366920471191,
|
| 396 |
+
"eval_model_preparation_time": 0.0053,
|
| 397 |
+
"eval_runtime": 16.8306,
|
| 398 |
+
"eval_samples_per_second": 61.079,
|
| 399 |
+
"eval_steps_per_second": 7.665,
|
| 400 |
+
"step": 1813
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 38.0,
|
| 404 |
+
"eval_accuracy": 0.8891050583657587,
|
| 405 |
+
"eval_loss": 0.8040880560874939,
|
| 406 |
+
"eval_model_preparation_time": 0.0053,
|
| 407 |
+
"eval_runtime": 16.772,
|
| 408 |
+
"eval_samples_per_second": 61.293,
|
| 409 |
+
"eval_steps_per_second": 7.691,
|
| 410 |
+
"step": 1862
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 39.0,
|
| 414 |
+
"eval_accuracy": 0.8891050583657587,
|
| 415 |
+
"eval_loss": 0.790228009223938,
|
| 416 |
+
"eval_model_preparation_time": 0.0053,
|
| 417 |
+
"eval_runtime": 16.3652,
|
| 418 |
+
"eval_samples_per_second": 62.816,
|
| 419 |
+
"eval_steps_per_second": 7.883,
|
| 420 |
+
"step": 1911
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 40.0,
|
| 424 |
+
"eval_accuracy": 0.8910505836575876,
|
| 425 |
+
"eval_loss": 0.7775859832763672,
|
| 426 |
+
"eval_model_preparation_time": 0.0053,
|
| 427 |
+
"eval_runtime": 16.2631,
|
| 428 |
+
"eval_samples_per_second": 63.211,
|
| 429 |
+
"eval_steps_per_second": 7.932,
|
| 430 |
+
"step": 1960
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 40.816326530612244,
|
| 434 |
+
"grad_norm": 1.7460684776306152,
|
| 435 |
+
"learning_rate": 5.979381443298969e-07,
|
| 436 |
+
"loss": 0.7398,
|
| 437 |
+
"step": 2000
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 41.0,
|
| 441 |
+
"eval_accuracy": 0.892023346303502,
|
| 442 |
+
"eval_loss": 0.766076922416687,
|
| 443 |
+
"eval_model_preparation_time": 0.0053,
|
| 444 |
+
"eval_runtime": 16.3183,
|
| 445 |
+
"eval_samples_per_second": 62.997,
|
| 446 |
+
"eval_steps_per_second": 7.905,
|
| 447 |
+
"step": 2009
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 42.0,
|
| 451 |
+
"eval_accuracy": 0.8949416342412452,
|
| 452 |
+
"eval_loss": 0.7547051906585693,
|
| 453 |
+
"eval_model_preparation_time": 0.0053,
|
| 454 |
+
"eval_runtime": 16.274,
|
| 455 |
+
"eval_samples_per_second": 63.168,
|
| 456 |
+
"eval_steps_per_second": 7.927,
|
| 457 |
+
"step": 2058
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 43.0,
|
| 461 |
+
"eval_accuracy": 0.8998054474708171,
|
| 462 |
+
"eval_loss": 0.7428992986679077,
|
| 463 |
+
"eval_model_preparation_time": 0.0053,
|
| 464 |
+
"eval_runtime": 16.2531,
|
| 465 |
+
"eval_samples_per_second": 63.25,
|
| 466 |
+
"eval_steps_per_second": 7.937,
|
| 467 |
+
"step": 2107
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 44.0,
|
| 471 |
+
"eval_accuracy": 0.9027237354085603,
|
| 472 |
+
"eval_loss": 0.7334992289543152,
|
| 473 |
+
"eval_model_preparation_time": 0.0053,
|
| 474 |
+
"eval_runtime": 16.1775,
|
| 475 |
+
"eval_samples_per_second": 63.545,
|
| 476 |
+
"eval_steps_per_second": 7.974,
|
| 477 |
+
"step": 2156
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 45.0,
|
| 481 |
+
"eval_accuracy": 0.9027237354085603,
|
| 482 |
+
"eval_loss": 0.7230122089385986,
|
| 483 |
+
"eval_model_preparation_time": 0.0053,
|
| 484 |
+
"eval_runtime": 16.4977,
|
| 485 |
+
"eval_samples_per_second": 62.312,
|
| 486 |
+
"eval_steps_per_second": 7.819,
|
| 487 |
+
"step": 2205
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 46.0,
|
| 491 |
+
"eval_accuracy": 0.9056420233463035,
|
| 492 |
+
"eval_loss": 0.7135947942733765,
|
| 493 |
+
"eval_model_preparation_time": 0.0053,
|
| 494 |
+
"eval_runtime": 16.7331,
|
| 495 |
+
"eval_samples_per_second": 61.435,
|
| 496 |
+
"eval_steps_per_second": 7.709,
|
| 497 |
+
"step": 2254
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 47.0,
|
| 501 |
+
"eval_accuracy": 0.9085603112840467,
|
| 502 |
+
"eval_loss": 0.7030473351478577,
|
| 503 |
+
"eval_model_preparation_time": 0.0053,
|
| 504 |
+
"eval_runtime": 16.265,
|
| 505 |
+
"eval_samples_per_second": 63.203,
|
| 506 |
+
"eval_steps_per_second": 7.931,
|
| 507 |
+
"step": 2303
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 48.0,
|
| 511 |
+
"eval_accuracy": 0.9066147859922179,
|
| 512 |
+
"eval_loss": 0.694833517074585,
|
| 513 |
+
"eval_model_preparation_time": 0.0053,
|
| 514 |
+
"eval_runtime": 16.6017,
|
| 515 |
+
"eval_samples_per_second": 61.921,
|
| 516 |
+
"eval_steps_per_second": 7.77,
|
| 517 |
+
"step": 2352
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 49.0,
|
| 521 |
+
"eval_accuracy": 0.9075875486381323,
|
| 522 |
+
"eval_loss": 0.6868679523468018,
|
| 523 |
+
"eval_model_preparation_time": 0.0053,
|
| 524 |
+
"eval_runtime": 16.3769,
|
| 525 |
+
"eval_samples_per_second": 62.771,
|
| 526 |
+
"eval_steps_per_second": 7.877,
|
| 527 |
+
"step": 2401
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 50.0,
|
| 531 |
+
"eval_accuracy": 0.9105058365758755,
|
| 532 |
+
"eval_loss": 0.6781201958656311,
|
| 533 |
+
"eval_model_preparation_time": 0.0053,
|
| 534 |
+
"eval_runtime": 16.5365,
|
| 535 |
+
"eval_samples_per_second": 62.166,
|
| 536 |
+
"eval_steps_per_second": 7.801,
|
| 537 |
+
"step": 2450
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 51.0,
|
| 541 |
+
"eval_accuracy": 0.9114785992217899,
|
| 542 |
+
"eval_loss": 0.6707582473754883,
|
| 543 |
+
"eval_model_preparation_time": 0.0053,
|
| 544 |
+
"eval_runtime": 16.7029,
|
| 545 |
+
"eval_samples_per_second": 61.546,
|
| 546 |
+
"eval_steps_per_second": 7.723,
|
| 547 |
+
"step": 2499
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 51.02040816326531,
|
| 551 |
+
"grad_norm": 1.5139410495758057,
|
| 552 |
+
"learning_rate": 4.948453608247422e-07,
|
| 553 |
+
"loss": 0.6135,
|
| 554 |
+
"step": 2500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 52.0,
|
| 558 |
+
"eval_accuracy": 0.914396887159533,
|
| 559 |
+
"eval_loss": 0.6635868549346924,
|
| 560 |
+
"eval_model_preparation_time": 0.0053,
|
| 561 |
+
"eval_runtime": 16.7917,
|
| 562 |
+
"eval_samples_per_second": 61.221,
|
| 563 |
+
"eval_steps_per_second": 7.682,
|
| 564 |
+
"step": 2548
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 53.0,
|
| 568 |
+
"eval_accuracy": 0.9134241245136187,
|
| 569 |
+
"eval_loss": 0.6572186946868896,
|
| 570 |
+
"eval_model_preparation_time": 0.0053,
|
| 571 |
+
"eval_runtime": 16.5427,
|
| 572 |
+
"eval_samples_per_second": 62.142,
|
| 573 |
+
"eval_steps_per_second": 7.798,
|
| 574 |
+
"step": 2597
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 54.0,
|
| 578 |
+
"eval_accuracy": 0.9153696498054474,
|
| 579 |
+
"eval_loss": 0.6508779525756836,
|
| 580 |
+
"eval_model_preparation_time": 0.0053,
|
| 581 |
+
"eval_runtime": 16.3209,
|
| 582 |
+
"eval_samples_per_second": 62.987,
|
| 583 |
+
"eval_steps_per_second": 7.904,
|
| 584 |
+
"step": 2646
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 55.0,
|
| 588 |
+
"eval_accuracy": 0.9173151750972762,
|
| 589 |
+
"eval_loss": 0.6439871191978455,
|
| 590 |
+
"eval_model_preparation_time": 0.0053,
|
| 591 |
+
"eval_runtime": 16.5127,
|
| 592 |
+
"eval_samples_per_second": 62.255,
|
| 593 |
+
"eval_steps_per_second": 7.812,
|
| 594 |
+
"step": 2695
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 56.0,
|
| 598 |
+
"eval_accuracy": 0.9153696498054474,
|
| 599 |
+
"eval_loss": 0.6389443874359131,
|
| 600 |
+
"eval_model_preparation_time": 0.0053,
|
| 601 |
+
"eval_runtime": 16.1819,
|
| 602 |
+
"eval_samples_per_second": 63.528,
|
| 603 |
+
"eval_steps_per_second": 7.972,
|
| 604 |
+
"step": 2744
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 57.0,
|
| 608 |
+
"eval_accuracy": 0.919260700389105,
|
| 609 |
+
"eval_loss": 0.6327943801879883,
|
| 610 |
+
"eval_model_preparation_time": 0.0053,
|
| 611 |
+
"eval_runtime": 16.6521,
|
| 612 |
+
"eval_samples_per_second": 61.734,
|
| 613 |
+
"eval_steps_per_second": 7.747,
|
| 614 |
+
"step": 2793
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 58.0,
|
| 618 |
+
"eval_accuracy": 0.9212062256809338,
|
| 619 |
+
"eval_loss": 0.6252166032791138,
|
| 620 |
+
"eval_model_preparation_time": 0.0053,
|
| 621 |
+
"eval_runtime": 16.374,
|
| 622 |
+
"eval_samples_per_second": 62.782,
|
| 623 |
+
"eval_steps_per_second": 7.878,
|
| 624 |
+
"step": 2842
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 59.0,
|
| 628 |
+
"eval_accuracy": 0.9202334630350194,
|
| 629 |
+
"eval_loss": 0.6206849217414856,
|
| 630 |
+
"eval_model_preparation_time": 0.0053,
|
| 631 |
+
"eval_runtime": 16.2753,
|
| 632 |
+
"eval_samples_per_second": 63.163,
|
| 633 |
+
"eval_steps_per_second": 7.926,
|
| 634 |
+
"step": 2891
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 60.0,
|
| 638 |
+
"eval_accuracy": 0.919260700389105,
|
| 639 |
+
"eval_loss": 0.6157888174057007,
|
| 640 |
+
"eval_model_preparation_time": 0.0053,
|
| 641 |
+
"eval_runtime": 16.4343,
|
| 642 |
+
"eval_samples_per_second": 62.552,
|
| 643 |
+
"eval_steps_per_second": 7.849,
|
| 644 |
+
"step": 2940
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 61.0,
|
| 648 |
+
"eval_accuracy": 0.9221789883268483,
|
| 649 |
+
"eval_loss": 0.6105452179908752,
|
| 650 |
+
"eval_model_preparation_time": 0.0053,
|
| 651 |
+
"eval_runtime": 16.3327,
|
| 652 |
+
"eval_samples_per_second": 62.941,
|
| 653 |
+
"eval_steps_per_second": 7.898,
|
| 654 |
+
"step": 2989
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 61.224489795918366,
|
| 658 |
+
"grad_norm": 2.1964964866638184,
|
| 659 |
+
"learning_rate": 3.917525773195876e-07,
|
| 660 |
+
"loss": 0.5351,
|
| 661 |
+
"step": 3000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 62.0,
|
| 665 |
+
"eval_accuracy": 0.9221789883268483,
|
| 666 |
+
"eval_loss": 0.6059720516204834,
|
| 667 |
+
"eval_model_preparation_time": 0.0053,
|
| 668 |
+
"eval_runtime": 15.9025,
|
| 669 |
+
"eval_samples_per_second": 64.644,
|
| 670 |
+
"eval_steps_per_second": 8.112,
|
| 671 |
+
"step": 3038
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 63.0,
|
| 675 |
+
"eval_accuracy": 0.9221789883268483,
|
| 676 |
+
"eval_loss": 0.6012532114982605,
|
| 677 |
+
"eval_model_preparation_time": 0.0053,
|
| 678 |
+
"eval_runtime": 16.1752,
|
| 679 |
+
"eval_samples_per_second": 63.554,
|
| 680 |
+
"eval_steps_per_second": 7.975,
|
| 681 |
+
"step": 3087
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 64.0,
|
| 685 |
+
"eval_accuracy": 0.9241245136186771,
|
| 686 |
+
"eval_loss": 0.5974491238594055,
|
| 687 |
+
"eval_model_preparation_time": 0.0053,
|
| 688 |
+
"eval_runtime": 16.006,
|
| 689 |
+
"eval_samples_per_second": 64.226,
|
| 690 |
+
"eval_steps_per_second": 8.059,
|
| 691 |
+
"step": 3136
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 65.0,
|
| 695 |
+
"eval_accuracy": 0.9250972762645915,
|
| 696 |
+
"eval_loss": 0.5935858488082886,
|
| 697 |
+
"eval_model_preparation_time": 0.0053,
|
| 698 |
+
"eval_runtime": 15.9581,
|
| 699 |
+
"eval_samples_per_second": 64.419,
|
| 700 |
+
"eval_steps_per_second": 8.084,
|
| 701 |
+
"step": 3185
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 66.0,
|
| 705 |
+
"eval_accuracy": 0.9260700389105059,
|
| 706 |
+
"eval_loss": 0.588865339756012,
|
| 707 |
+
"eval_model_preparation_time": 0.0053,
|
| 708 |
+
"eval_runtime": 16.1822,
|
| 709 |
+
"eval_samples_per_second": 63.526,
|
| 710 |
+
"eval_steps_per_second": 7.972,
|
| 711 |
+
"step": 3234
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 67.0,
|
| 715 |
+
"eval_accuracy": 0.9260700389105059,
|
| 716 |
+
"eval_loss": 0.5851794481277466,
|
| 717 |
+
"eval_model_preparation_time": 0.0053,
|
| 718 |
+
"eval_runtime": 16.116,
|
| 719 |
+
"eval_samples_per_second": 63.787,
|
| 720 |
+
"eval_steps_per_second": 8.004,
|
| 721 |
+
"step": 3283
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 68.0,
|
| 725 |
+
"eval_accuracy": 0.9221789883268483,
|
| 726 |
+
"eval_loss": 0.5818396210670471,
|
| 727 |
+
"eval_model_preparation_time": 0.0053,
|
| 728 |
+
"eval_runtime": 16.5601,
|
| 729 |
+
"eval_samples_per_second": 62.077,
|
| 730 |
+
"eval_steps_per_second": 7.79,
|
| 731 |
+
"step": 3332
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 69.0,
|
| 735 |
+
"eval_accuracy": 0.9231517509727627,
|
| 736 |
+
"eval_loss": 0.578502357006073,
|
| 737 |
+
"eval_model_preparation_time": 0.0053,
|
| 738 |
+
"eval_runtime": 16.364,
|
| 739 |
+
"eval_samples_per_second": 62.821,
|
| 740 |
+
"eval_steps_per_second": 7.883,
|
| 741 |
+
"step": 3381
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 70.0,
|
| 745 |
+
"eval_accuracy": 0.9241245136186771,
|
| 746 |
+
"eval_loss": 0.5750812888145447,
|
| 747 |
+
"eval_model_preparation_time": 0.0053,
|
| 748 |
+
"eval_runtime": 16.3061,
|
| 749 |
+
"eval_samples_per_second": 63.044,
|
| 750 |
+
"eval_steps_per_second": 7.911,
|
| 751 |
+
"step": 3430
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 71.0,
|
| 755 |
+
"eval_accuracy": 0.9241245136186771,
|
| 756 |
+
"eval_loss": 0.5718052387237549,
|
| 757 |
+
"eval_model_preparation_time": 0.0053,
|
| 758 |
+
"eval_runtime": 16.3828,
|
| 759 |
+
"eval_samples_per_second": 62.749,
|
| 760 |
+
"eval_steps_per_second": 7.874,
|
| 761 |
+
"step": 3479
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 71.42857142857143,
|
| 765 |
+
"grad_norm": 1.5724107027053833,
|
| 766 |
+
"learning_rate": 2.8865979381443296e-07,
|
| 767 |
+
"loss": 0.483,
|
| 768 |
+
"step": 3500
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 72.0,
|
| 772 |
+
"eval_accuracy": 0.9250972762645915,
|
| 773 |
+
"eval_loss": 0.5689519643783569,
|
| 774 |
+
"eval_model_preparation_time": 0.0053,
|
| 775 |
+
"eval_runtime": 16.3228,
|
| 776 |
+
"eval_samples_per_second": 62.98,
|
| 777 |
+
"eval_steps_per_second": 7.903,
|
| 778 |
+
"step": 3528
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 73.0,
|
| 782 |
+
"eval_accuracy": 0.9241245136186771,
|
| 783 |
+
"eval_loss": 0.565991997718811,
|
| 784 |
+
"eval_model_preparation_time": 0.0053,
|
| 785 |
+
"eval_runtime": 16.0251,
|
| 786 |
+
"eval_samples_per_second": 64.149,
|
| 787 |
+
"eval_steps_per_second": 8.05,
|
| 788 |
+
"step": 3577
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 74.0,
|
| 792 |
+
"eval_accuracy": 0.9231517509727627,
|
| 793 |
+
"eval_loss": 0.5631707906723022,
|
| 794 |
+
"eval_model_preparation_time": 0.0053,
|
| 795 |
+
"eval_runtime": 16.0794,
|
| 796 |
+
"eval_samples_per_second": 63.933,
|
| 797 |
+
"eval_steps_per_second": 8.023,
|
| 798 |
+
"step": 3626
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 75.0,
|
| 802 |
+
"eval_accuracy": 0.9250972762645915,
|
| 803 |
+
"eval_loss": 0.5604016780853271,
|
| 804 |
+
"eval_model_preparation_time": 0.0053,
|
| 805 |
+
"eval_runtime": 16.1767,
|
| 806 |
+
"eval_samples_per_second": 63.548,
|
| 807 |
+
"eval_steps_per_second": 7.974,
|
| 808 |
+
"step": 3675
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 76.0,
|
| 812 |
+
"eval_accuracy": 0.9241245136186771,
|
| 813 |
+
"eval_loss": 0.5577030777931213,
|
| 814 |
+
"eval_model_preparation_time": 0.0053,
|
| 815 |
+
"eval_runtime": 16.2849,
|
| 816 |
+
"eval_samples_per_second": 63.126,
|
| 817 |
+
"eval_steps_per_second": 7.921,
|
| 818 |
+
"step": 3724
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 77.0,
|
| 822 |
+
"eval_accuracy": 0.9250972762645915,
|
| 823 |
+
"eval_loss": 0.5554907917976379,
|
| 824 |
+
"eval_model_preparation_time": 0.0053,
|
| 825 |
+
"eval_runtime": 16.1225,
|
| 826 |
+
"eval_samples_per_second": 63.762,
|
| 827 |
+
"eval_steps_per_second": 8.001,
|
| 828 |
+
"step": 3773
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 78.0,
|
| 832 |
+
"eval_accuracy": 0.9260700389105059,
|
| 833 |
+
"eval_loss": 0.5532063841819763,
|
| 834 |
+
"eval_model_preparation_time": 0.0053,
|
| 835 |
+
"eval_runtime": 16.1489,
|
| 836 |
+
"eval_samples_per_second": 63.658,
|
| 837 |
+
"eval_steps_per_second": 7.988,
|
| 838 |
+
"step": 3822
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 79.0,
|
| 842 |
+
"eval_accuracy": 0.9241245136186771,
|
| 843 |
+
"eval_loss": 0.5517746806144714,
|
| 844 |
+
"eval_model_preparation_time": 0.0053,
|
| 845 |
+
"eval_runtime": 16.1776,
|
| 846 |
+
"eval_samples_per_second": 63.545,
|
| 847 |
+
"eval_steps_per_second": 7.974,
|
| 848 |
+
"step": 3871
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 80.0,
|
| 852 |
+
"eval_accuracy": 0.9260700389105059,
|
| 853 |
+
"eval_loss": 0.5491300821304321,
|
| 854 |
+
"eval_model_preparation_time": 0.0053,
|
| 855 |
+
"eval_runtime": 16.4646,
|
| 856 |
+
"eval_samples_per_second": 62.437,
|
| 857 |
+
"eval_steps_per_second": 7.835,
|
| 858 |
+
"step": 3920
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 81.0,
|
| 862 |
+
"eval_accuracy": 0.9260700389105059,
|
| 863 |
+
"eval_loss": 0.5468857884407043,
|
| 864 |
+
"eval_model_preparation_time": 0.0053,
|
| 865 |
+
"eval_runtime": 16.5897,
|
| 866 |
+
"eval_samples_per_second": 61.966,
|
| 867 |
+
"eval_steps_per_second": 7.776,
|
| 868 |
+
"step": 3969
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 81.63265306122449,
|
| 872 |
+
"grad_norm": 1.5352323055267334,
|
| 873 |
+
"learning_rate": 1.8556701030927835e-07,
|
| 874 |
+
"loss": 0.4496,
|
| 875 |
+
"step": 4000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 82.0,
|
| 879 |
+
"eval_accuracy": 0.9260700389105059,
|
| 880 |
+
"eval_loss": 0.5453611016273499,
|
| 881 |
+
"eval_model_preparation_time": 0.0053,
|
| 882 |
+
"eval_runtime": 16.5158,
|
| 883 |
+
"eval_samples_per_second": 62.243,
|
| 884 |
+
"eval_steps_per_second": 7.811,
|
| 885 |
+
"step": 4018
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 83.0,
|
| 889 |
+
"eval_accuracy": 0.9289883268482491,
|
| 890 |
+
"eval_loss": 0.5431452989578247,
|
| 891 |
+
"eval_model_preparation_time": 0.0053,
|
| 892 |
+
"eval_runtime": 16.2358,
|
| 893 |
+
"eval_samples_per_second": 63.317,
|
| 894 |
+
"eval_steps_per_second": 7.945,
|
| 895 |
+
"step": 4067
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 84.0,
|
| 899 |
+
"eval_accuracy": 0.9270428015564203,
|
| 900 |
+
"eval_loss": 0.5419163107872009,
|
| 901 |
+
"eval_model_preparation_time": 0.0053,
|
| 902 |
+
"eval_runtime": 16.3758,
|
| 903 |
+
"eval_samples_per_second": 62.776,
|
| 904 |
+
"eval_steps_per_second": 7.877,
|
| 905 |
+
"step": 4116
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 85.0,
|
| 909 |
+
"eval_accuracy": 0.9280155642023347,
|
| 910 |
+
"eval_loss": 0.5407156348228455,
|
| 911 |
+
"eval_model_preparation_time": 0.0053,
|
| 912 |
+
"eval_runtime": 16.7946,
|
| 913 |
+
"eval_samples_per_second": 61.21,
|
| 914 |
+
"eval_steps_per_second": 7.681,
|
| 915 |
+
"step": 4165
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 86.0,
|
| 919 |
+
"eval_accuracy": 0.9280155642023347,
|
| 920 |
+
"eval_loss": 0.5391236543655396,
|
| 921 |
+
"eval_model_preparation_time": 0.0053,
|
| 922 |
+
"eval_runtime": 16.961,
|
| 923 |
+
"eval_samples_per_second": 60.609,
|
| 924 |
+
"eval_steps_per_second": 7.606,
|
| 925 |
+
"step": 4214
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 87.0,
|
| 929 |
+
"eval_accuracy": 0.9280155642023347,
|
| 930 |
+
"eval_loss": 0.5377916693687439,
|
| 931 |
+
"eval_model_preparation_time": 0.0053,
|
| 932 |
+
"eval_runtime": 16.9464,
|
| 933 |
+
"eval_samples_per_second": 60.662,
|
| 934 |
+
"eval_steps_per_second": 7.612,
|
| 935 |
+
"step": 4263
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 88.0,
|
| 939 |
+
"eval_accuracy": 0.9280155642023347,
|
| 940 |
+
"eval_loss": 0.5364987850189209,
|
| 941 |
+
"eval_model_preparation_time": 0.0053,
|
| 942 |
+
"eval_runtime": 16.757,
|
| 943 |
+
"eval_samples_per_second": 61.347,
|
| 944 |
+
"eval_steps_per_second": 7.698,
|
| 945 |
+
"step": 4312
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 89.0,
|
| 949 |
+
"eval_accuracy": 0.9280155642023347,
|
| 950 |
+
"eval_loss": 0.5354925394058228,
|
| 951 |
+
"eval_model_preparation_time": 0.0053,
|
| 952 |
+
"eval_runtime": 16.8288,
|
| 953 |
+
"eval_samples_per_second": 61.086,
|
| 954 |
+
"eval_steps_per_second": 7.665,
|
| 955 |
+
"step": 4361
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 90.0,
|
| 959 |
+
"eval_accuracy": 0.9270428015564203,
|
| 960 |
+
"eval_loss": 0.534569501876831,
|
| 961 |
+
"eval_model_preparation_time": 0.0053,
|
| 962 |
+
"eval_runtime": 16.5639,
|
| 963 |
+
"eval_samples_per_second": 62.063,
|
| 964 |
+
"eval_steps_per_second": 7.788,
|
| 965 |
+
"step": 4410
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 91.0,
|
| 969 |
+
"eval_accuracy": 0.9280155642023347,
|
| 970 |
+
"eval_loss": 0.5335658192634583,
|
| 971 |
+
"eval_model_preparation_time": 0.0053,
|
| 972 |
+
"eval_runtime": 16.5932,
|
| 973 |
+
"eval_samples_per_second": 61.953,
|
| 974 |
+
"eval_steps_per_second": 7.774,
|
| 975 |
+
"step": 4459
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 91.83673469387755,
|
| 979 |
+
"grad_norm": 1.4989055395126343,
|
| 980 |
+
"learning_rate": 8.24742268041237e-08,
|
| 981 |
+
"loss": 0.4299,
|
| 982 |
+
"step": 4500
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 92.0,
|
| 986 |
+
"eval_accuracy": 0.9280155642023347,
|
| 987 |
+
"eval_loss": 0.5327633023262024,
|
| 988 |
+
"eval_model_preparation_time": 0.0053,
|
| 989 |
+
"eval_runtime": 16.6218,
|
| 990 |
+
"eval_samples_per_second": 61.846,
|
| 991 |
+
"eval_steps_per_second": 7.761,
|
| 992 |
+
"step": 4508
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 93.0,
|
| 996 |
+
"eval_accuracy": 0.9280155642023347,
|
| 997 |
+
"eval_loss": 0.5321099758148193,
|
| 998 |
+
"eval_model_preparation_time": 0.0053,
|
| 999 |
+
"eval_runtime": 16.4686,
|
| 1000 |
+
"eval_samples_per_second": 62.422,
|
| 1001 |
+
"eval_steps_per_second": 7.833,
|
| 1002 |
+
"step": 4557
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 94.0,
|
| 1006 |
+
"eval_accuracy": 0.9280155642023347,
|
| 1007 |
+
"eval_loss": 0.5314825177192688,
|
| 1008 |
+
"eval_model_preparation_time": 0.0053,
|
| 1009 |
+
"eval_runtime": 16.4592,
|
| 1010 |
+
"eval_samples_per_second": 62.458,
|
| 1011 |
+
"eval_steps_per_second": 7.838,
|
| 1012 |
+
"step": 4606
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 95.0,
|
| 1016 |
+
"eval_accuracy": 0.9270428015564203,
|
| 1017 |
+
"eval_loss": 0.5310559868812561,
|
| 1018 |
+
"eval_model_preparation_time": 0.0053,
|
| 1019 |
+
"eval_runtime": 16.6737,
|
| 1020 |
+
"eval_samples_per_second": 61.654,
|
| 1021 |
+
"eval_steps_per_second": 7.737,
|
| 1022 |
+
"step": 4655
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 96.0,
|
| 1026 |
+
"eval_accuracy": 0.9270428015564203,
|
| 1027 |
+
"eval_loss": 0.5306681990623474,
|
| 1028 |
+
"eval_model_preparation_time": 0.0053,
|
| 1029 |
+
"eval_runtime": 16.4895,
|
| 1030 |
+
"eval_samples_per_second": 62.343,
|
| 1031 |
+
"eval_steps_per_second": 7.823,
|
| 1032 |
+
"step": 4704
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 97.0,
|
| 1036 |
+
"eval_accuracy": 0.9280155642023347,
|
| 1037 |
+
"eval_loss": 0.5302944183349609,
|
| 1038 |
+
"eval_model_preparation_time": 0.0053,
|
| 1039 |
+
"eval_runtime": 16.5288,
|
| 1040 |
+
"eval_samples_per_second": 62.194,
|
| 1041 |
+
"eval_steps_per_second": 7.805,
|
| 1042 |
+
"step": 4753
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 98.0,
|
| 1046 |
+
"eval_accuracy": 0.9270428015564203,
|
| 1047 |
+
"eval_loss": 0.5300623774528503,
|
| 1048 |
+
"eval_model_preparation_time": 0.0053,
|
| 1049 |
+
"eval_runtime": 16.4477,
|
| 1050 |
+
"eval_samples_per_second": 62.501,
|
| 1051 |
+
"eval_steps_per_second": 7.843,
|
| 1052 |
+
"step": 4802
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 99.0,
|
| 1056 |
+
"eval_accuracy": 0.9280155642023347,
|
| 1057 |
+
"eval_loss": 0.5299115180969238,
|
| 1058 |
+
"eval_model_preparation_time": 0.0053,
|
| 1059 |
+
"eval_runtime": 16.5643,
|
| 1060 |
+
"eval_samples_per_second": 62.061,
|
| 1061 |
+
"eval_steps_per_second": 7.788,
|
| 1062 |
+
"step": 4851
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 100.0,
|
| 1066 |
+
"eval_accuracy": 0.9280155642023347,
|
| 1067 |
+
"eval_loss": 0.5298713445663452,
|
| 1068 |
+
"eval_model_preparation_time": 0.0053,
|
| 1069 |
+
"eval_runtime": 16.9421,
|
| 1070 |
+
"eval_samples_per_second": 60.677,
|
| 1071 |
+
"eval_steps_per_second": 7.614,
|
| 1072 |
+
"step": 4900
|
| 1073 |
+
}
|
| 1074 |
+
],
|
| 1075 |
+
"logging_steps": 500,
|
| 1076 |
+
"max_steps": 4900,
|
| 1077 |
+
"num_input_tokens_seen": 0,
|
| 1078 |
+
"num_train_epochs": 100,
|
| 1079 |
+
"save_steps": 500,
|
| 1080 |
+
"stateful_callbacks": {
|
| 1081 |
+
"TrainerControl": {
|
| 1082 |
+
"args": {
|
| 1083 |
+
"should_epoch_stop": false,
|
| 1084 |
+
"should_evaluate": false,
|
| 1085 |
+
"should_log": false,
|
| 1086 |
+
"should_save": true,
|
| 1087 |
+
"should_training_stop": true
|
| 1088 |
+
},
|
| 1089 |
+
"attributes": {}
|
| 1090 |
+
}
|
| 1091 |
+
},
|
| 1092 |
+
"total_flos": 1.1949586091012506e+19,
|
| 1093 |
+
"train_batch_size": 32,
|
| 1094 |
+
"trial_name": null,
|
| 1095 |
+
"trial_params": null
|
| 1096 |
+
}
|
checkpoint-4900/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:858525d77a2aca146b3415e12c5a888e5f5b65064bb27a757b2697e4eabc4024
|
| 3 |
+
size 5176
|
config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "curly",
|
| 13 |
+
"1": "dreadlocks",
|
| 14 |
+
"2": "kinky",
|
| 15 |
+
"3": "straight",
|
| 16 |
+
"4": "wavy"
|
| 17 |
+
},
|
| 18 |
+
"image_size": 224,
|
| 19 |
+
"initializer_range": 0.02,
|
| 20 |
+
"intermediate_size": 3072,
|
| 21 |
+
"label2id": {
|
| 22 |
+
"curly": 0,
|
| 23 |
+
"dreadlocks": 1,
|
| 24 |
+
"kinky": 2,
|
| 25 |
+
"straight": 3,
|
| 26 |
+
"wavy": 4
|
| 27 |
+
},
|
| 28 |
+
"layer_norm_eps": 1e-12,
|
| 29 |
+
"model_type": "vit",
|
| 30 |
+
"num_attention_heads": 12,
|
| 31 |
+
"num_channels": 3,
|
| 32 |
+
"num_hidden_layers": 12,
|
| 33 |
+
"patch_size": 16,
|
| 34 |
+
"problem_type": "single_label_classification",
|
| 35 |
+
"qkv_bias": true,
|
| 36 |
+
"torch_dtype": "float32",
|
| 37 |
+
"transformers_version": "4.45.2"
|
| 38 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:880839388dee970fc3c9a41a9ea13295c6093fea1653d77780af5a05743c9cf2
|
| 3 |
+
size 343233204
|
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 sha256:858525d77a2aca146b3415e12c5a888e5f5b65064bb27a757b2697e4eabc4024
|
| 3 |
+
size 5176
|