Upload folder using huggingface_hub
Browse files- README.md +31 -0
- checkpoint-7029/config.json +33 -0
- checkpoint-7029/model.safetensors +3 -0
- checkpoint-7029/optimizer.pt +3 -0
- checkpoint-7029/rng_state.pth +3 -0
- checkpoint-7029/scheduler.pt +3 -0
- checkpoint-7029/trainer_state.json +2048 -0
- checkpoint-7029/training_args.bin +3 -0
- config.json +33 -0
- model.safetensors +3 -0
- preprocessor_config.json +22 -0
- runs/Sep24_16-03-26_r-pandalori-autotrain-cats-and-dogs-8cfe05bk-8ef12-7ju5g/events.out.tfevents.1727193808.r-pandalori-autotrain-cats-and-dogs-8cfe05bk-8ef12-7ju5g.207.0 +2 -2
- runs/Sep24_16-03-26_r-pandalori-autotrain-cats-and-dogs-8cfe05bk-8ef12-7ju5g/events.out.tfevents.1727249929.r-pandalori-autotrain-cats-and-dogs-8cfe05bk-8ef12-7ju5g.207.1 +3 -0
- training_args.bin +3 -0
- training_params.json +29 -0
README.md
ADDED
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
---
|
3 |
+
tags:
|
4 |
+
- autotrain
|
5 |
+
- image-classification
|
6 |
+
base_model: google/vit-base-patch16-224
|
7 |
+
widget:
|
8 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/tiger.jpg
|
9 |
+
example_title: Tiger
|
10 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/teapot.jpg
|
11 |
+
example_title: Teapot
|
12 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/palace.jpg
|
13 |
+
example_title: Palace
|
14 |
+
---
|
15 |
+
|
16 |
+
# Model Trained Using AutoTrain
|
17 |
+
|
18 |
+
- Problem type: Image Classification
|
19 |
+
|
20 |
+
## Validation Metrics
|
21 |
+
loss: 0.0158307533711195
|
22 |
+
|
23 |
+
f1: 0.9961538461538462
|
24 |
+
|
25 |
+
precision: 0.9940298507462687
|
26 |
+
|
27 |
+
recall: 0.9982869379014989
|
28 |
+
|
29 |
+
auc: 0.9994886327395326
|
30 |
+
|
31 |
+
accuracy: 0.9961579509071505
|
checkpoint-7029/config.json
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "google/vit-base-patch16-224",
|
3 |
+
"_num_labels": 2,
|
4 |
+
"architectures": [
|
5 |
+
"ViTForImageClassification"
|
6 |
+
],
|
7 |
+
"attention_probs_dropout_prob": 0.0,
|
8 |
+
"encoder_stride": 16,
|
9 |
+
"hidden_act": "gelu",
|
10 |
+
"hidden_dropout_prob": 0.0,
|
11 |
+
"hidden_size": 768,
|
12 |
+
"id2label": {
|
13 |
+
"0": "Cat",
|
14 |
+
"1": "Dog"
|
15 |
+
},
|
16 |
+
"image_size": 224,
|
17 |
+
"initializer_range": 0.02,
|
18 |
+
"intermediate_size": 3072,
|
19 |
+
"label2id": {
|
20 |
+
"Cat": 0,
|
21 |
+
"Dog": 1
|
22 |
+
},
|
23 |
+
"layer_norm_eps": 1e-12,
|
24 |
+
"model_type": "vit",
|
25 |
+
"num_attention_heads": 12,
|
26 |
+
"num_channels": 3,
|
27 |
+
"num_hidden_layers": 12,
|
28 |
+
"patch_size": 16,
|
29 |
+
"problem_type": "single_label_classification",
|
30 |
+
"qkv_bias": true,
|
31 |
+
"torch_dtype": "float32",
|
32 |
+
"transformers_version": "4.44.2"
|
33 |
+
}
|
checkpoint-7029/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dee3eed218313e8ff29b584802246af13c768a49b4c314f646c4fa450b081f8d
|
3 |
+
size 343223968
|
checkpoint-7029/optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6530ffd08ba009a70fae44eaf51e791bb695d95ef4886227a233727009bad46a
|
3 |
+
size 686563258
|
checkpoint-7029/rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:191d38bc28e3cef21d465729870c6f5f9e5cd0a4ebcd4b05940fe058bf3fcecd
|
3 |
+
size 13990
|
checkpoint-7029/scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bf772930a9908c481bf75c081cf19a2311659fefe0f2d5acffeb2329985b9a67
|
3 |
+
size 1064
|
checkpoint-7029/trainer_state.json
ADDED
@@ -0,0 +1,2048 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.0158307533711195,
|
3 |
+
"best_model_checkpoint": "autotrain-ltphq-1aue6/checkpoint-7029",
|
4 |
+
"epoch": 3.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 7029,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.010670081092616303,
|
13 |
+
"grad_norm": 10.576315879821777,
|
14 |
+
"learning_rate": 1.7780938833570414e-06,
|
15 |
+
"loss": 0.5983,
|
16 |
+
"step": 25
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.021340162185232606,
|
20 |
+
"grad_norm": 6.441564559936523,
|
21 |
+
"learning_rate": 3.556187766714083e-06,
|
22 |
+
"loss": 0.5878,
|
23 |
+
"step": 50
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.03201024327784891,
|
27 |
+
"grad_norm": 5.742478370666504,
|
28 |
+
"learning_rate": 5.334281650071124e-06,
|
29 |
+
"loss": 0.4551,
|
30 |
+
"step": 75
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.04268032437046521,
|
34 |
+
"grad_norm": 3.9501702785491943,
|
35 |
+
"learning_rate": 7.112375533428166e-06,
|
36 |
+
"loss": 0.3351,
|
37 |
+
"step": 100
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.05335040546308152,
|
41 |
+
"grad_norm": 3.156757354736328,
|
42 |
+
"learning_rate": 8.890469416785207e-06,
|
43 |
+
"loss": 0.2165,
|
44 |
+
"step": 125
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.06402048655569782,
|
48 |
+
"grad_norm": 1.531072735786438,
|
49 |
+
"learning_rate": 1.0668563300142247e-05,
|
50 |
+
"loss": 0.1089,
|
51 |
+
"step": 150
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.07469056764831412,
|
55 |
+
"grad_norm": 0.6878227591514587,
|
56 |
+
"learning_rate": 1.244665718349929e-05,
|
57 |
+
"loss": 0.1164,
|
58 |
+
"step": 175
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.08536064874093043,
|
62 |
+
"grad_norm": 0.3712044358253479,
|
63 |
+
"learning_rate": 1.4224751066856332e-05,
|
64 |
+
"loss": 0.0563,
|
65 |
+
"step": 200
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.09603072983354674,
|
69 |
+
"grad_norm": 5.825244903564453,
|
70 |
+
"learning_rate": 1.6002844950213374e-05,
|
71 |
+
"loss": 0.0577,
|
72 |
+
"step": 225
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.10670081092616304,
|
76 |
+
"grad_norm": 7.699280261993408,
|
77 |
+
"learning_rate": 1.7780938833570414e-05,
|
78 |
+
"loss": 0.0995,
|
79 |
+
"step": 250
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.11737089201877934,
|
83 |
+
"grad_norm": 1.6869832277297974,
|
84 |
+
"learning_rate": 1.9559032716927454e-05,
|
85 |
+
"loss": 0.0223,
|
86 |
+
"step": 275
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.12804097311139565,
|
90 |
+
"grad_norm": 0.5479092597961426,
|
91 |
+
"learning_rate": 2.1337126600284495e-05,
|
92 |
+
"loss": 0.0396,
|
93 |
+
"step": 300
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.13871105420401195,
|
97 |
+
"grad_norm": 17.6248836517334,
|
98 |
+
"learning_rate": 2.3115220483641535e-05,
|
99 |
+
"loss": 0.0629,
|
100 |
+
"step": 325
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.14938113529662825,
|
104 |
+
"grad_norm": 1.2864736318588257,
|
105 |
+
"learning_rate": 2.489331436699858e-05,
|
106 |
+
"loss": 0.0815,
|
107 |
+
"step": 350
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.16005121638924455,
|
111 |
+
"grad_norm": 14.023919105529785,
|
112 |
+
"learning_rate": 2.6671408250355616e-05,
|
113 |
+
"loss": 0.1471,
|
114 |
+
"step": 375
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.17072129748186085,
|
118 |
+
"grad_norm": 0.7266956567764282,
|
119 |
+
"learning_rate": 2.8449502133712663e-05,
|
120 |
+
"loss": 0.1421,
|
121 |
+
"step": 400
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.18139137857447715,
|
125 |
+
"grad_norm": 1.591202735900879,
|
126 |
+
"learning_rate": 3.0227596017069704e-05,
|
127 |
+
"loss": 0.0837,
|
128 |
+
"step": 425
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.19206145966709348,
|
132 |
+
"grad_norm": 0.03316282480955124,
|
133 |
+
"learning_rate": 3.200568990042675e-05,
|
134 |
+
"loss": 0.0419,
|
135 |
+
"step": 450
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.20273154075970978,
|
139 |
+
"grad_norm": 0.0206840131431818,
|
140 |
+
"learning_rate": 3.3783783783783784e-05,
|
141 |
+
"loss": 0.0621,
|
142 |
+
"step": 475
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.21340162185232608,
|
146 |
+
"grad_norm": 0.017792997881770134,
|
147 |
+
"learning_rate": 3.556187766714083e-05,
|
148 |
+
"loss": 0.1231,
|
149 |
+
"step": 500
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.22407170294494239,
|
153 |
+
"grad_norm": 5.664891242980957,
|
154 |
+
"learning_rate": 3.7339971550497865e-05,
|
155 |
+
"loss": 0.1255,
|
156 |
+
"step": 525
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.2347417840375587,
|
160 |
+
"grad_norm": 0.08337491750717163,
|
161 |
+
"learning_rate": 3.911806543385491e-05,
|
162 |
+
"loss": 0.1177,
|
163 |
+
"step": 550
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.245411865130175,
|
167 |
+
"grad_norm": 11.999979019165039,
|
168 |
+
"learning_rate": 4.089615931721195e-05,
|
169 |
+
"loss": 0.0796,
|
170 |
+
"step": 575
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.2560819462227913,
|
174 |
+
"grad_norm": 0.007278278470039368,
|
175 |
+
"learning_rate": 4.267425320056899e-05,
|
176 |
+
"loss": 0.0677,
|
177 |
+
"step": 600
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.2667520273154076,
|
181 |
+
"grad_norm": 16.901777267456055,
|
182 |
+
"learning_rate": 4.4452347083926033e-05,
|
183 |
+
"loss": 0.2031,
|
184 |
+
"step": 625
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.2774221084080239,
|
188 |
+
"grad_norm": 8.973575592041016,
|
189 |
+
"learning_rate": 4.623044096728307e-05,
|
190 |
+
"loss": 0.0724,
|
191 |
+
"step": 650
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.2880921895006402,
|
195 |
+
"grad_norm": 0.5288171768188477,
|
196 |
+
"learning_rate": 4.8008534850640114e-05,
|
197 |
+
"loss": 0.0616,
|
198 |
+
"step": 675
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.2987622705932565,
|
202 |
+
"grad_norm": 0.03448805212974548,
|
203 |
+
"learning_rate": 4.978662873399716e-05,
|
204 |
+
"loss": 0.0951,
|
205 |
+
"step": 700
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.3094323516858728,
|
209 |
+
"grad_norm": 37.160011291503906,
|
210 |
+
"learning_rate": 4.982611444830857e-05,
|
211 |
+
"loss": 0.2065,
|
212 |
+
"step": 725
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.3201024327784891,
|
216 |
+
"grad_norm": 0.19236384332180023,
|
217 |
+
"learning_rate": 4.9628517230477395e-05,
|
218 |
+
"loss": 0.2069,
|
219 |
+
"step": 750
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.3307725138711054,
|
223 |
+
"grad_norm": 0.1479603499174118,
|
224 |
+
"learning_rate": 4.9430920012646225e-05,
|
225 |
+
"loss": 0.1349,
|
226 |
+
"step": 775
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.3414425949637217,
|
230 |
+
"grad_norm": 0.3465137779712677,
|
231 |
+
"learning_rate": 4.923332279481505e-05,
|
232 |
+
"loss": 0.13,
|
233 |
+
"step": 800
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.352112676056338,
|
237 |
+
"grad_norm": 0.05050384998321533,
|
238 |
+
"learning_rate": 4.903572557698388e-05,
|
239 |
+
"loss": 0.137,
|
240 |
+
"step": 825
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.3627827571489543,
|
244 |
+
"grad_norm": 0.16721266508102417,
|
245 |
+
"learning_rate": 4.883812835915271e-05,
|
246 |
+
"loss": 0.1692,
|
247 |
+
"step": 850
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.37345283824157066,
|
251 |
+
"grad_norm": 0.015766268596053123,
|
252 |
+
"learning_rate": 4.864053114132153e-05,
|
253 |
+
"loss": 0.1324,
|
254 |
+
"step": 875
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.38412291933418696,
|
258 |
+
"grad_norm": 0.04455697163939476,
|
259 |
+
"learning_rate": 4.8442933923490356e-05,
|
260 |
+
"loss": 0.1173,
|
261 |
+
"step": 900
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.39479300042680326,
|
265 |
+
"grad_norm": 0.045407216995954514,
|
266 |
+
"learning_rate": 4.8245336705659186e-05,
|
267 |
+
"loss": 0.1282,
|
268 |
+
"step": 925
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.40546308151941957,
|
272 |
+
"grad_norm": 0.014451881870627403,
|
273 |
+
"learning_rate": 4.804773948782802e-05,
|
274 |
+
"loss": 0.156,
|
275 |
+
"step": 950
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.41613316261203587,
|
279 |
+
"grad_norm": 10.56628131866455,
|
280 |
+
"learning_rate": 4.785014226999684e-05,
|
281 |
+
"loss": 0.1354,
|
282 |
+
"step": 975
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.42680324370465217,
|
286 |
+
"grad_norm": 9.949240684509277,
|
287 |
+
"learning_rate": 4.765254505216567e-05,
|
288 |
+
"loss": 0.1674,
|
289 |
+
"step": 1000
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.43747332479726847,
|
293 |
+
"grad_norm": 0.8956235647201538,
|
294 |
+
"learning_rate": 4.7454947834334494e-05,
|
295 |
+
"loss": 0.0663,
|
296 |
+
"step": 1025
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.44814340588988477,
|
300 |
+
"grad_norm": 11.43295669555664,
|
301 |
+
"learning_rate": 4.725735061650332e-05,
|
302 |
+
"loss": 0.106,
|
303 |
+
"step": 1050
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.4588134869825011,
|
307 |
+
"grad_norm": 0.10002151876688004,
|
308 |
+
"learning_rate": 4.7059753398672155e-05,
|
309 |
+
"loss": 0.0652,
|
310 |
+
"step": 1075
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.4694835680751174,
|
314 |
+
"grad_norm": 0.018603714182972908,
|
315 |
+
"learning_rate": 4.686215618084098e-05,
|
316 |
+
"loss": 0.0544,
|
317 |
+
"step": 1100
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.4801536491677337,
|
321 |
+
"grad_norm": 0.002447799313813448,
|
322 |
+
"learning_rate": 4.66645589630098e-05,
|
323 |
+
"loss": 0.0874,
|
324 |
+
"step": 1125
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.49082373026035,
|
328 |
+
"grad_norm": 0.029736997559666634,
|
329 |
+
"learning_rate": 4.6466961745178625e-05,
|
330 |
+
"loss": 0.1893,
|
331 |
+
"step": 1150
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.5014938113529663,
|
335 |
+
"grad_norm": 10.314343452453613,
|
336 |
+
"learning_rate": 4.6269364527347456e-05,
|
337 |
+
"loss": 0.1898,
|
338 |
+
"step": 1175
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.5121638924455826,
|
342 |
+
"grad_norm": 0.6686179041862488,
|
343 |
+
"learning_rate": 4.6071767309516286e-05,
|
344 |
+
"loss": 0.0865,
|
345 |
+
"step": 1200
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.5228339735381989,
|
349 |
+
"grad_norm": 1.8905715942382812,
|
350 |
+
"learning_rate": 4.587417009168511e-05,
|
351 |
+
"loss": 0.1412,
|
352 |
+
"step": 1225
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 0.5335040546308152,
|
356 |
+
"grad_norm": 2.919725179672241,
|
357 |
+
"learning_rate": 4.567657287385394e-05,
|
358 |
+
"loss": 0.1305,
|
359 |
+
"step": 1250
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 0.5441741357234315,
|
363 |
+
"grad_norm": 0.24073974788188934,
|
364 |
+
"learning_rate": 4.547897565602276e-05,
|
365 |
+
"loss": 0.1203,
|
366 |
+
"step": 1275
|
367 |
+
},
|
368 |
+
{
|
369 |
+
"epoch": 0.5548442168160478,
|
370 |
+
"grad_norm": 0.010738029144704342,
|
371 |
+
"learning_rate": 4.5281378438191594e-05,
|
372 |
+
"loss": 0.0627,
|
373 |
+
"step": 1300
|
374 |
+
},
|
375 |
+
{
|
376 |
+
"epoch": 0.5655142979086641,
|
377 |
+
"grad_norm": 0.008832808583974838,
|
378 |
+
"learning_rate": 4.5083781220360424e-05,
|
379 |
+
"loss": 0.1611,
|
380 |
+
"step": 1325
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 0.5761843790012804,
|
384 |
+
"grad_norm": 3.6089112758636475,
|
385 |
+
"learning_rate": 4.488618400252925e-05,
|
386 |
+
"loss": 0.146,
|
387 |
+
"step": 1350
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 0.5868544600938967,
|
391 |
+
"grad_norm": 0.08832165598869324,
|
392 |
+
"learning_rate": 4.468858678469807e-05,
|
393 |
+
"loss": 0.1005,
|
394 |
+
"step": 1375
|
395 |
+
},
|
396 |
+
{
|
397 |
+
"epoch": 0.597524541186513,
|
398 |
+
"grad_norm": 0.03827153891324997,
|
399 |
+
"learning_rate": 4.44909895668669e-05,
|
400 |
+
"loss": 0.1597,
|
401 |
+
"step": 1400
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 0.6081946222791293,
|
405 |
+
"grad_norm": 0.02119293250143528,
|
406 |
+
"learning_rate": 4.429339234903573e-05,
|
407 |
+
"loss": 0.0933,
|
408 |
+
"step": 1425
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 0.6188647033717456,
|
412 |
+
"grad_norm": 0.004805543925613165,
|
413 |
+
"learning_rate": 4.4095795131204555e-05,
|
414 |
+
"loss": 0.0904,
|
415 |
+
"step": 1450
|
416 |
+
},
|
417 |
+
{
|
418 |
+
"epoch": 0.629534784464362,
|
419 |
+
"grad_norm": 0.560352623462677,
|
420 |
+
"learning_rate": 4.3898197913373385e-05,
|
421 |
+
"loss": 0.1021,
|
422 |
+
"step": 1475
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 0.6402048655569782,
|
426 |
+
"grad_norm": 0.13053584098815918,
|
427 |
+
"learning_rate": 4.370060069554221e-05,
|
428 |
+
"loss": 0.1822,
|
429 |
+
"step": 1500
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 0.6508749466495946,
|
433 |
+
"grad_norm": 0.020129365846514702,
|
434 |
+
"learning_rate": 4.350300347771103e-05,
|
435 |
+
"loss": 0.1116,
|
436 |
+
"step": 1525
|
437 |
+
},
|
438 |
+
{
|
439 |
+
"epoch": 0.6615450277422108,
|
440 |
+
"grad_norm": 0.08490480482578278,
|
441 |
+
"learning_rate": 4.330540625987986e-05,
|
442 |
+
"loss": 0.1514,
|
443 |
+
"step": 1550
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 0.6722151088348272,
|
447 |
+
"grad_norm": 17.20114517211914,
|
448 |
+
"learning_rate": 4.310780904204869e-05,
|
449 |
+
"loss": 0.1683,
|
450 |
+
"step": 1575
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 0.6828851899274434,
|
454 |
+
"grad_norm": 20.063180923461914,
|
455 |
+
"learning_rate": 4.2910211824217516e-05,
|
456 |
+
"loss": 0.0969,
|
457 |
+
"step": 1600
|
458 |
+
},
|
459 |
+
{
|
460 |
+
"epoch": 0.6935552710200598,
|
461 |
+
"grad_norm": 10.146587371826172,
|
462 |
+
"learning_rate": 4.271261460638634e-05,
|
463 |
+
"loss": 0.1382,
|
464 |
+
"step": 1625
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 0.704225352112676,
|
468 |
+
"grad_norm": 0.04244920238852501,
|
469 |
+
"learning_rate": 4.251501738855517e-05,
|
470 |
+
"loss": 0.102,
|
471 |
+
"step": 1650
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 0.7148954332052924,
|
475 |
+
"grad_norm": 3.3513338565826416,
|
476 |
+
"learning_rate": 4.2317420170724e-05,
|
477 |
+
"loss": 0.0844,
|
478 |
+
"step": 1675
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 0.7255655142979086,
|
482 |
+
"grad_norm": 9.942954063415527,
|
483 |
+
"learning_rate": 4.2119822952892824e-05,
|
484 |
+
"loss": 0.1169,
|
485 |
+
"step": 1700
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 0.736235595390525,
|
489 |
+
"grad_norm": 0.02176540717482567,
|
490 |
+
"learning_rate": 4.1922225735061654e-05,
|
491 |
+
"loss": 0.0546,
|
492 |
+
"step": 1725
|
493 |
+
},
|
494 |
+
{
|
495 |
+
"epoch": 0.7469056764831413,
|
496 |
+
"grad_norm": 0.02619314193725586,
|
497 |
+
"learning_rate": 4.172462851723048e-05,
|
498 |
+
"loss": 0.1489,
|
499 |
+
"step": 1750
|
500 |
+
},
|
501 |
+
{
|
502 |
+
"epoch": 0.7575757575757576,
|
503 |
+
"grad_norm": 4.979767799377441,
|
504 |
+
"learning_rate": 4.152703129939931e-05,
|
505 |
+
"loss": 0.1087,
|
506 |
+
"step": 1775
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.7682458386683739,
|
510 |
+
"grad_norm": 0.22245188057422638,
|
511 |
+
"learning_rate": 4.132943408156814e-05,
|
512 |
+
"loss": 0.0505,
|
513 |
+
"step": 1800
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 0.7789159197609902,
|
517 |
+
"grad_norm": 3.7733798027038574,
|
518 |
+
"learning_rate": 4.113183686373696e-05,
|
519 |
+
"loss": 0.1281,
|
520 |
+
"step": 1825
|
521 |
+
},
|
522 |
+
{
|
523 |
+
"epoch": 0.7895860008536065,
|
524 |
+
"grad_norm": 0.004993762820959091,
|
525 |
+
"learning_rate": 4.0934239645905786e-05,
|
526 |
+
"loss": 0.0967,
|
527 |
+
"step": 1850
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 0.8002560819462228,
|
531 |
+
"grad_norm": 6.705662250518799,
|
532 |
+
"learning_rate": 4.0736642428074616e-05,
|
533 |
+
"loss": 0.088,
|
534 |
+
"step": 1875
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 0.8109261630388391,
|
538 |
+
"grad_norm": 0.09842755645513535,
|
539 |
+
"learning_rate": 4.053904521024344e-05,
|
540 |
+
"loss": 0.09,
|
541 |
+
"step": 1900
|
542 |
+
},
|
543 |
+
{
|
544 |
+
"epoch": 0.8215962441314554,
|
545 |
+
"grad_norm": 0.006683106068521738,
|
546 |
+
"learning_rate": 4.034144799241227e-05,
|
547 |
+
"loss": 0.0329,
|
548 |
+
"step": 1925
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.8322663252240717,
|
552 |
+
"grad_norm": 0.8531517386436462,
|
553 |
+
"learning_rate": 4.014385077458109e-05,
|
554 |
+
"loss": 0.1405,
|
555 |
+
"step": 1950
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 0.842936406316688,
|
559 |
+
"grad_norm": 0.011011715978384018,
|
560 |
+
"learning_rate": 3.9946253556749924e-05,
|
561 |
+
"loss": 0.1381,
|
562 |
+
"step": 1975
|
563 |
+
},
|
564 |
+
{
|
565 |
+
"epoch": 0.8536064874093043,
|
566 |
+
"grad_norm": 0.20137256383895874,
|
567 |
+
"learning_rate": 3.974865633891875e-05,
|
568 |
+
"loss": 0.1315,
|
569 |
+
"step": 2000
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 0.8642765685019206,
|
573 |
+
"grad_norm": 4.466578006744385,
|
574 |
+
"learning_rate": 3.955105912108758e-05,
|
575 |
+
"loss": 0.0826,
|
576 |
+
"step": 2025
|
577 |
+
},
|
578 |
+
{
|
579 |
+
"epoch": 0.8749466495945369,
|
580 |
+
"grad_norm": 0.007189568132162094,
|
581 |
+
"learning_rate": 3.935346190325641e-05,
|
582 |
+
"loss": 0.0696,
|
583 |
+
"step": 2050
|
584 |
+
},
|
585 |
+
{
|
586 |
+
"epoch": 0.8856167306871532,
|
587 |
+
"grad_norm": 0.13521578907966614,
|
588 |
+
"learning_rate": 3.915586468542523e-05,
|
589 |
+
"loss": 0.0647,
|
590 |
+
"step": 2075
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.8962868117797695,
|
594 |
+
"grad_norm": 0.09942985326051712,
|
595 |
+
"learning_rate": 3.8958267467594055e-05,
|
596 |
+
"loss": 0.0337,
|
597 |
+
"step": 2100
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 0.9069568928723858,
|
601 |
+
"grad_norm": 0.003063550451770425,
|
602 |
+
"learning_rate": 3.8760670249762885e-05,
|
603 |
+
"loss": 0.1349,
|
604 |
+
"step": 2125
|
605 |
+
},
|
606 |
+
{
|
607 |
+
"epoch": 0.9176269739650021,
|
608 |
+
"grad_norm": 0.03615418076515198,
|
609 |
+
"learning_rate": 3.8563073031931715e-05,
|
610 |
+
"loss": 0.1884,
|
611 |
+
"step": 2150
|
612 |
+
},
|
613 |
+
{
|
614 |
+
"epoch": 0.9282970550576184,
|
615 |
+
"grad_norm": 2.4063897132873535,
|
616 |
+
"learning_rate": 3.836547581410054e-05,
|
617 |
+
"loss": 0.0942,
|
618 |
+
"step": 2175
|
619 |
+
},
|
620 |
+
{
|
621 |
+
"epoch": 0.9389671361502347,
|
622 |
+
"grad_norm": 0.10824164003133774,
|
623 |
+
"learning_rate": 3.816787859626937e-05,
|
624 |
+
"loss": 0.1643,
|
625 |
+
"step": 2200
|
626 |
+
},
|
627 |
+
{
|
628 |
+
"epoch": 0.949637217242851,
|
629 |
+
"grad_norm": 0.23877990245819092,
|
630 |
+
"learning_rate": 3.797028137843819e-05,
|
631 |
+
"loss": 0.2041,
|
632 |
+
"step": 2225
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 0.9603072983354674,
|
636 |
+
"grad_norm": 0.12965470552444458,
|
637 |
+
"learning_rate": 3.7772684160607016e-05,
|
638 |
+
"loss": 0.0686,
|
639 |
+
"step": 2250
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 0.9709773794280837,
|
643 |
+
"grad_norm": 9.515990257263184,
|
644 |
+
"learning_rate": 3.757508694277585e-05,
|
645 |
+
"loss": 0.1288,
|
646 |
+
"step": 2275
|
647 |
+
},
|
648 |
+
{
|
649 |
+
"epoch": 0.9816474605207,
|
650 |
+
"grad_norm": 0.010902081616222858,
|
651 |
+
"learning_rate": 3.737748972494468e-05,
|
652 |
+
"loss": 0.1204,
|
653 |
+
"step": 2300
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 0.9923175416133163,
|
657 |
+
"grad_norm": 0.12074346095323563,
|
658 |
+
"learning_rate": 3.71798925071135e-05,
|
659 |
+
"loss": 0.0735,
|
660 |
+
"step": 2325
|
661 |
+
},
|
662 |
+
{
|
663 |
+
"epoch": 1.0,
|
664 |
+
"eval_accuracy": 0.9942369263607257,
|
665 |
+
"eval_auc": 0.9994051665223928,
|
666 |
+
"eval_f1": 0.9942369263607257,
|
667 |
+
"eval_loss": 0.02018207497894764,
|
668 |
+
"eval_precision": 0.9910638297872341,
|
669 |
+
"eval_recall": 0.9974304068522484,
|
670 |
+
"eval_runtime": 1286.73,
|
671 |
+
"eval_samples_per_second": 3.641,
|
672 |
+
"eval_steps_per_second": 0.228,
|
673 |
+
"step": 2343
|
674 |
+
},
|
675 |
+
{
|
676 |
+
"epoch": 1.0029876227059327,
|
677 |
+
"grad_norm": 0.13857701420783997,
|
678 |
+
"learning_rate": 3.6982295289282324e-05,
|
679 |
+
"loss": 0.1404,
|
680 |
+
"step": 2350
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 1.0136577037985488,
|
684 |
+
"grad_norm": 16.29566764831543,
|
685 |
+
"learning_rate": 3.6784698071451154e-05,
|
686 |
+
"loss": 0.1023,
|
687 |
+
"step": 2375
|
688 |
+
},
|
689 |
+
{
|
690 |
+
"epoch": 1.0243277848911652,
|
691 |
+
"grad_norm": 0.007060299627482891,
|
692 |
+
"learning_rate": 3.6587100853619984e-05,
|
693 |
+
"loss": 0.0314,
|
694 |
+
"step": 2400
|
695 |
+
},
|
696 |
+
{
|
697 |
+
"epoch": 1.0349978659837815,
|
698 |
+
"grad_norm": 0.1959122270345688,
|
699 |
+
"learning_rate": 3.638950363578881e-05,
|
700 |
+
"loss": 0.0698,
|
701 |
+
"step": 2425
|
702 |
+
},
|
703 |
+
{
|
704 |
+
"epoch": 1.0456679470763979,
|
705 |
+
"grad_norm": 2.1727123260498047,
|
706 |
+
"learning_rate": 3.619190641795764e-05,
|
707 |
+
"loss": 0.1026,
|
708 |
+
"step": 2450
|
709 |
+
},
|
710 |
+
{
|
711 |
+
"epoch": 1.056338028169014,
|
712 |
+
"grad_norm": 0.14948531985282898,
|
713 |
+
"learning_rate": 3.599430920012646e-05,
|
714 |
+
"loss": 0.1336,
|
715 |
+
"step": 2475
|
716 |
+
},
|
717 |
+
{
|
718 |
+
"epoch": 1.0670081092616304,
|
719 |
+
"grad_norm": 0.0020017814822494984,
|
720 |
+
"learning_rate": 3.579671198229529e-05,
|
721 |
+
"loss": 0.0718,
|
722 |
+
"step": 2500
|
723 |
+
},
|
724 |
+
{
|
725 |
+
"epoch": 1.0776781903542467,
|
726 |
+
"grad_norm": 0.040247637778520584,
|
727 |
+
"learning_rate": 3.559911476446412e-05,
|
728 |
+
"loss": 0.1003,
|
729 |
+
"step": 2525
|
730 |
+
},
|
731 |
+
{
|
732 |
+
"epoch": 1.088348271446863,
|
733 |
+
"grad_norm": 0.0026400748174637556,
|
734 |
+
"learning_rate": 3.5401517546632946e-05,
|
735 |
+
"loss": 0.0103,
|
736 |
+
"step": 2550
|
737 |
+
},
|
738 |
+
{
|
739 |
+
"epoch": 1.0990183525394792,
|
740 |
+
"grad_norm": 13.71839714050293,
|
741 |
+
"learning_rate": 3.520392032880177e-05,
|
742 |
+
"loss": 0.1549,
|
743 |
+
"step": 2575
|
744 |
+
},
|
745 |
+
{
|
746 |
+
"epoch": 1.1096884336320956,
|
747 |
+
"grad_norm": 0.016973601654171944,
|
748 |
+
"learning_rate": 3.50063231109706e-05,
|
749 |
+
"loss": 0.0833,
|
750 |
+
"step": 2600
|
751 |
+
},
|
752 |
+
{
|
753 |
+
"epoch": 1.120358514724712,
|
754 |
+
"grad_norm": 0.023715652525424957,
|
755 |
+
"learning_rate": 3.480872589313943e-05,
|
756 |
+
"loss": 0.1606,
|
757 |
+
"step": 2625
|
758 |
+
},
|
759 |
+
{
|
760 |
+
"epoch": 1.1310285958173283,
|
761 |
+
"grad_norm": 0.12024948000907898,
|
762 |
+
"learning_rate": 3.4611128675308254e-05,
|
763 |
+
"loss": 0.0659,
|
764 |
+
"step": 2650
|
765 |
+
},
|
766 |
+
{
|
767 |
+
"epoch": 1.1416986769099444,
|
768 |
+
"grad_norm": 0.013869931921362877,
|
769 |
+
"learning_rate": 3.4413531457477084e-05,
|
770 |
+
"loss": 0.1145,
|
771 |
+
"step": 2675
|
772 |
+
},
|
773 |
+
{
|
774 |
+
"epoch": 1.1523687580025608,
|
775 |
+
"grad_norm": 0.003865574486553669,
|
776 |
+
"learning_rate": 3.421593423964591e-05,
|
777 |
+
"loss": 0.0742,
|
778 |
+
"step": 2700
|
779 |
+
},
|
780 |
+
{
|
781 |
+
"epoch": 1.1630388390951771,
|
782 |
+
"grad_norm": 6.1942524909973145,
|
783 |
+
"learning_rate": 3.401833702181473e-05,
|
784 |
+
"loss": 0.1685,
|
785 |
+
"step": 2725
|
786 |
+
},
|
787 |
+
{
|
788 |
+
"epoch": 1.1737089201877935,
|
789 |
+
"grad_norm": 13.037029266357422,
|
790 |
+
"learning_rate": 3.382073980398356e-05,
|
791 |
+
"loss": 0.0882,
|
792 |
+
"step": 2750
|
793 |
+
},
|
794 |
+
{
|
795 |
+
"epoch": 1.1843790012804098,
|
796 |
+
"grad_norm": 0.0354326069355011,
|
797 |
+
"learning_rate": 3.362314258615239e-05,
|
798 |
+
"loss": 0.1236,
|
799 |
+
"step": 2775
|
800 |
+
},
|
801 |
+
{
|
802 |
+
"epoch": 1.195049082373026,
|
803 |
+
"grad_norm": 0.18212293088436127,
|
804 |
+
"learning_rate": 3.3425545368321215e-05,
|
805 |
+
"loss": 0.0612,
|
806 |
+
"step": 2800
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"epoch": 1.2057191634656423,
|
810 |
+
"grad_norm": 0.019807470962405205,
|
811 |
+
"learning_rate": 3.322794815049004e-05,
|
812 |
+
"loss": 0.0237,
|
813 |
+
"step": 2825
|
814 |
+
},
|
815 |
+
{
|
816 |
+
"epoch": 1.2163892445582587,
|
817 |
+
"grad_norm": 0.48490288853645325,
|
818 |
+
"learning_rate": 3.303035093265887e-05,
|
819 |
+
"loss": 0.1864,
|
820 |
+
"step": 2850
|
821 |
+
},
|
822 |
+
{
|
823 |
+
"epoch": 1.2270593256508748,
|
824 |
+
"grad_norm": 15.739009857177734,
|
825 |
+
"learning_rate": 3.28327537148277e-05,
|
826 |
+
"loss": 0.0783,
|
827 |
+
"step": 2875
|
828 |
+
},
|
829 |
+
{
|
830 |
+
"epoch": 1.2377294067434912,
|
831 |
+
"grad_norm": 0.00520035345107317,
|
832 |
+
"learning_rate": 3.263515649699652e-05,
|
833 |
+
"loss": 0.056,
|
834 |
+
"step": 2900
|
835 |
+
},
|
836 |
+
{
|
837 |
+
"epoch": 1.2483994878361075,
|
838 |
+
"grad_norm": 0.007584866136312485,
|
839 |
+
"learning_rate": 3.243755927916535e-05,
|
840 |
+
"loss": 0.1054,
|
841 |
+
"step": 2925
|
842 |
+
},
|
843 |
+
{
|
844 |
+
"epoch": 1.259069568928724,
|
845 |
+
"grad_norm": 0.009660612791776657,
|
846 |
+
"learning_rate": 3.2239962061334176e-05,
|
847 |
+
"loss": 0.1001,
|
848 |
+
"step": 2950
|
849 |
+
},
|
850 |
+
{
|
851 |
+
"epoch": 1.2697396500213403,
|
852 |
+
"grad_norm": 5.9183735847473145,
|
853 |
+
"learning_rate": 3.2042364843503e-05,
|
854 |
+
"loss": 0.0959,
|
855 |
+
"step": 2975
|
856 |
+
},
|
857 |
+
{
|
858 |
+
"epoch": 1.2804097311139564,
|
859 |
+
"grad_norm": 0.017659351229667664,
|
860 |
+
"learning_rate": 3.184476762567184e-05,
|
861 |
+
"loss": 0.0771,
|
862 |
+
"step": 3000
|
863 |
+
},
|
864 |
+
{
|
865 |
+
"epoch": 1.2910798122065728,
|
866 |
+
"grad_norm": 0.05887264013290405,
|
867 |
+
"learning_rate": 3.164717040784066e-05,
|
868 |
+
"loss": 0.0527,
|
869 |
+
"step": 3025
|
870 |
+
},
|
871 |
+
{
|
872 |
+
"epoch": 1.301749893299189,
|
873 |
+
"grad_norm": 47.84782028198242,
|
874 |
+
"learning_rate": 3.1449573190009484e-05,
|
875 |
+
"loss": 0.0658,
|
876 |
+
"step": 3050
|
877 |
+
},
|
878 |
+
{
|
879 |
+
"epoch": 1.3124199743918052,
|
880 |
+
"grad_norm": 0.046731848269701004,
|
881 |
+
"learning_rate": 3.1251975972178314e-05,
|
882 |
+
"loss": 0.0446,
|
883 |
+
"step": 3075
|
884 |
+
},
|
885 |
+
{
|
886 |
+
"epoch": 1.3230900554844216,
|
887 |
+
"grad_norm": 23.92389678955078,
|
888 |
+
"learning_rate": 3.105437875434714e-05,
|
889 |
+
"loss": 0.1645,
|
890 |
+
"step": 3100
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 1.333760136577038,
|
894 |
+
"grad_norm": 0.23928241431713104,
|
895 |
+
"learning_rate": 3.085678153651597e-05,
|
896 |
+
"loss": 0.0808,
|
897 |
+
"step": 3125
|
898 |
+
},
|
899 |
+
{
|
900 |
+
"epoch": 1.3444302176696543,
|
901 |
+
"grad_norm": 0.004650407936424017,
|
902 |
+
"learning_rate": 3.06591843186848e-05,
|
903 |
+
"loss": 0.0788,
|
904 |
+
"step": 3150
|
905 |
+
},
|
906 |
+
{
|
907 |
+
"epoch": 1.3551002987622707,
|
908 |
+
"grad_norm": 2.407900333404541,
|
909 |
+
"learning_rate": 3.0461587100853622e-05,
|
910 |
+
"loss": 0.0663,
|
911 |
+
"step": 3175
|
912 |
+
},
|
913 |
+
{
|
914 |
+
"epoch": 1.365770379854887,
|
915 |
+
"grad_norm": 0.024215010926127434,
|
916 |
+
"learning_rate": 3.026398988302245e-05,
|
917 |
+
"loss": 0.0836,
|
918 |
+
"step": 3200
|
919 |
+
},
|
920 |
+
{
|
921 |
+
"epoch": 1.3764404609475032,
|
922 |
+
"grad_norm": 8.639642715454102,
|
923 |
+
"learning_rate": 3.0066392665191273e-05,
|
924 |
+
"loss": 0.0826,
|
925 |
+
"step": 3225
|
926 |
+
},
|
927 |
+
{
|
928 |
+
"epoch": 1.3871105420401195,
|
929 |
+
"grad_norm": 0.4214279353618622,
|
930 |
+
"learning_rate": 2.9868795447360103e-05,
|
931 |
+
"loss": 0.0407,
|
932 |
+
"step": 3250
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 1.3977806231327359,
|
936 |
+
"grad_norm": 3.026334285736084,
|
937 |
+
"learning_rate": 2.967119822952893e-05,
|
938 |
+
"loss": 0.0902,
|
939 |
+
"step": 3275
|
940 |
+
},
|
941 |
+
{
|
942 |
+
"epoch": 1.408450704225352,
|
943 |
+
"grad_norm": 5.5475568771362305,
|
944 |
+
"learning_rate": 2.9473601011697753e-05,
|
945 |
+
"loss": 0.0213,
|
946 |
+
"step": 3300
|
947 |
+
},
|
948 |
+
{
|
949 |
+
"epoch": 1.4191207853179684,
|
950 |
+
"grad_norm": 0.4343937337398529,
|
951 |
+
"learning_rate": 2.9276003793866587e-05,
|
952 |
+
"loss": 0.0485,
|
953 |
+
"step": 3325
|
954 |
+
},
|
955 |
+
{
|
956 |
+
"epoch": 1.4297908664105847,
|
957 |
+
"grad_norm": 4.497000217437744,
|
958 |
+
"learning_rate": 2.907840657603541e-05,
|
959 |
+
"loss": 0.0254,
|
960 |
+
"step": 3350
|
961 |
+
},
|
962 |
+
{
|
963 |
+
"epoch": 1.440460947503201,
|
964 |
+
"grad_norm": 9.44819164276123,
|
965 |
+
"learning_rate": 2.8880809358204237e-05,
|
966 |
+
"loss": 0.1441,
|
967 |
+
"step": 3375
|
968 |
+
},
|
969 |
+
{
|
970 |
+
"epoch": 1.4511310285958174,
|
971 |
+
"grad_norm": 0.5778021812438965,
|
972 |
+
"learning_rate": 2.8683212140373068e-05,
|
973 |
+
"loss": 0.0601,
|
974 |
+
"step": 3400
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 1.4618011096884336,
|
978 |
+
"grad_norm": 0.0012221585493534803,
|
979 |
+
"learning_rate": 2.848561492254189e-05,
|
980 |
+
"loss": 0.0134,
|
981 |
+
"step": 3425
|
982 |
+
},
|
983 |
+
{
|
984 |
+
"epoch": 1.47247119078105,
|
985 |
+
"grad_norm": 0.003332477994263172,
|
986 |
+
"learning_rate": 2.8288017704710718e-05,
|
987 |
+
"loss": 0.0872,
|
988 |
+
"step": 3450
|
989 |
+
},
|
990 |
+
{
|
991 |
+
"epoch": 1.4831412718736663,
|
992 |
+
"grad_norm": 19.613975524902344,
|
993 |
+
"learning_rate": 2.809042048687955e-05,
|
994 |
+
"loss": 0.0975,
|
995 |
+
"step": 3475
|
996 |
+
},
|
997 |
+
{
|
998 |
+
"epoch": 1.4938113529662824,
|
999 |
+
"grad_norm": 0.0008805744582787156,
|
1000 |
+
"learning_rate": 2.7892823269048375e-05,
|
1001 |
+
"loss": 0.0348,
|
1002 |
+
"step": 3500
|
1003 |
+
},
|
1004 |
+
{
|
1005 |
+
"epoch": 1.5044814340588988,
|
1006 |
+
"grad_norm": 0.745171070098877,
|
1007 |
+
"learning_rate": 2.76952260512172e-05,
|
1008 |
+
"loss": 0.0709,
|
1009 |
+
"step": 3525
|
1010 |
+
},
|
1011 |
+
{
|
1012 |
+
"epoch": 1.5151515151515151,
|
1013 |
+
"grad_norm": 0.0018973586848005652,
|
1014 |
+
"learning_rate": 2.749762883338603e-05,
|
1015 |
+
"loss": 0.1068,
|
1016 |
+
"step": 3550
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 1.5258215962441315,
|
1020 |
+
"grad_norm": 7.711709022521973,
|
1021 |
+
"learning_rate": 2.7300031615554856e-05,
|
1022 |
+
"loss": 0.1518,
|
1023 |
+
"step": 3575
|
1024 |
+
},
|
1025 |
+
{
|
1026 |
+
"epoch": 1.5364916773367479,
|
1027 |
+
"grad_norm": 0.08238033205270767,
|
1028 |
+
"learning_rate": 2.710243439772368e-05,
|
1029 |
+
"loss": 0.0806,
|
1030 |
+
"step": 3600
|
1031 |
+
},
|
1032 |
+
{
|
1033 |
+
"epoch": 1.5471617584293642,
|
1034 |
+
"grad_norm": 0.00853784941136837,
|
1035 |
+
"learning_rate": 2.6904837179892507e-05,
|
1036 |
+
"loss": 0.0889,
|
1037 |
+
"step": 3625
|
1038 |
+
},
|
1039 |
+
{
|
1040 |
+
"epoch": 1.5578318395219803,
|
1041 |
+
"grad_norm": 0.06024482846260071,
|
1042 |
+
"learning_rate": 2.6707239962061337e-05,
|
1043 |
+
"loss": 0.0724,
|
1044 |
+
"step": 3650
|
1045 |
+
},
|
1046 |
+
{
|
1047 |
+
"epoch": 1.5685019206145967,
|
1048 |
+
"grad_norm": 0.06635627895593643,
|
1049 |
+
"learning_rate": 2.650964274423016e-05,
|
1050 |
+
"loss": 0.0477,
|
1051 |
+
"step": 3675
|
1052 |
+
},
|
1053 |
+
{
|
1054 |
+
"epoch": 1.5791720017072128,
|
1055 |
+
"grad_norm": 0.057497624307870865,
|
1056 |
+
"learning_rate": 2.6312045526398987e-05,
|
1057 |
+
"loss": 0.0362,
|
1058 |
+
"step": 3700
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 1.5898420827998292,
|
1062 |
+
"grad_norm": 5.420880317687988,
|
1063 |
+
"learning_rate": 2.6114448308567818e-05,
|
1064 |
+
"loss": 0.0996,
|
1065 |
+
"step": 3725
|
1066 |
+
},
|
1067 |
+
{
|
1068 |
+
"epoch": 1.6005121638924455,
|
1069 |
+
"grad_norm": 0.03688732162117958,
|
1070 |
+
"learning_rate": 2.5916851090736644e-05,
|
1071 |
+
"loss": 0.1065,
|
1072 |
+
"step": 3750
|
1073 |
+
},
|
1074 |
+
{
|
1075 |
+
"epoch": 1.611182244985062,
|
1076 |
+
"grad_norm": 0.01889336109161377,
|
1077 |
+
"learning_rate": 2.5719253872905468e-05,
|
1078 |
+
"loss": 0.0675,
|
1079 |
+
"step": 3775
|
1080 |
+
},
|
1081 |
+
{
|
1082 |
+
"epoch": 1.6218523260776783,
|
1083 |
+
"grad_norm": 0.041272666305303574,
|
1084 |
+
"learning_rate": 2.5521656655074298e-05,
|
1085 |
+
"loss": 0.0783,
|
1086 |
+
"step": 3800
|
1087 |
+
},
|
1088 |
+
{
|
1089 |
+
"epoch": 1.6325224071702946,
|
1090 |
+
"grad_norm": 0.0283421128988266,
|
1091 |
+
"learning_rate": 2.5324059437243125e-05,
|
1092 |
+
"loss": 0.0535,
|
1093 |
+
"step": 3825
|
1094 |
+
},
|
1095 |
+
{
|
1096 |
+
"epoch": 1.6431924882629108,
|
1097 |
+
"grad_norm": 0.03971688076853752,
|
1098 |
+
"learning_rate": 2.512646221941195e-05,
|
1099 |
+
"loss": 0.0916,
|
1100 |
+
"step": 3850
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"epoch": 1.6538625693555271,
|
1104 |
+
"grad_norm": 0.7453581690788269,
|
1105 |
+
"learning_rate": 2.492886500158078e-05,
|
1106 |
+
"loss": 0.0275,
|
1107 |
+
"step": 3875
|
1108 |
+
},
|
1109 |
+
{
|
1110 |
+
"epoch": 1.6645326504481432,
|
1111 |
+
"grad_norm": 0.0038267953786998987,
|
1112 |
+
"learning_rate": 2.4731267783749606e-05,
|
1113 |
+
"loss": 0.0636,
|
1114 |
+
"step": 3900
|
1115 |
+
},
|
1116 |
+
{
|
1117 |
+
"epoch": 1.6752027315407596,
|
1118 |
+
"grad_norm": 17.677406311035156,
|
1119 |
+
"learning_rate": 2.4533670565918433e-05,
|
1120 |
+
"loss": 0.0811,
|
1121 |
+
"step": 3925
|
1122 |
+
},
|
1123 |
+
{
|
1124 |
+
"epoch": 1.685872812633376,
|
1125 |
+
"grad_norm": 0.056949540972709656,
|
1126 |
+
"learning_rate": 2.433607334808726e-05,
|
1127 |
+
"loss": 0.1143,
|
1128 |
+
"step": 3950
|
1129 |
+
},
|
1130 |
+
{
|
1131 |
+
"epoch": 1.6965428937259923,
|
1132 |
+
"grad_norm": 0.06619152426719666,
|
1133 |
+
"learning_rate": 2.4138476130256087e-05,
|
1134 |
+
"loss": 0.0825,
|
1135 |
+
"step": 3975
|
1136 |
+
},
|
1137 |
+
{
|
1138 |
+
"epoch": 1.7072129748186087,
|
1139 |
+
"grad_norm": 29.96984100341797,
|
1140 |
+
"learning_rate": 2.3940878912424914e-05,
|
1141 |
+
"loss": 0.1266,
|
1142 |
+
"step": 4000
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 1.717883055911225,
|
1146 |
+
"grad_norm": 7.958868026733398,
|
1147 |
+
"learning_rate": 2.374328169459374e-05,
|
1148 |
+
"loss": 0.0979,
|
1149 |
+
"step": 4025
|
1150 |
+
},
|
1151 |
+
{
|
1152 |
+
"epoch": 1.7285531370038414,
|
1153 |
+
"grad_norm": 0.0707533210515976,
|
1154 |
+
"learning_rate": 2.354568447676257e-05,
|
1155 |
+
"loss": 0.1238,
|
1156 |
+
"step": 4050
|
1157 |
+
},
|
1158 |
+
{
|
1159 |
+
"epoch": 1.7392232180964575,
|
1160 |
+
"grad_norm": 5.3426408767700195,
|
1161 |
+
"learning_rate": 2.3348087258931394e-05,
|
1162 |
+
"loss": 0.0941,
|
1163 |
+
"step": 4075
|
1164 |
+
},
|
1165 |
+
{
|
1166 |
+
"epoch": 1.7498932991890739,
|
1167 |
+
"grad_norm": 0.8626427054405212,
|
1168 |
+
"learning_rate": 2.315049004110022e-05,
|
1169 |
+
"loss": 0.037,
|
1170 |
+
"step": 4100
|
1171 |
+
},
|
1172 |
+
{
|
1173 |
+
"epoch": 1.76056338028169,
|
1174 |
+
"grad_norm": 0.007791485637426376,
|
1175 |
+
"learning_rate": 2.2952892823269048e-05,
|
1176 |
+
"loss": 0.039,
|
1177 |
+
"step": 4125
|
1178 |
+
},
|
1179 |
+
{
|
1180 |
+
"epoch": 1.7712334613743064,
|
1181 |
+
"grad_norm": 0.0019313797820359468,
|
1182 |
+
"learning_rate": 2.2755295605437875e-05,
|
1183 |
+
"loss": 0.0719,
|
1184 |
+
"step": 4150
|
1185 |
+
},
|
1186 |
+
{
|
1187 |
+
"epoch": 1.7819035424669227,
|
1188 |
+
"grad_norm": 0.058322928845882416,
|
1189 |
+
"learning_rate": 2.2557698387606705e-05,
|
1190 |
+
"loss": 0.0242,
|
1191 |
+
"step": 4175
|
1192 |
+
},
|
1193 |
+
{
|
1194 |
+
"epoch": 1.792573623559539,
|
1195 |
+
"grad_norm": 6.136843204498291,
|
1196 |
+
"learning_rate": 2.236010116977553e-05,
|
1197 |
+
"loss": 0.1471,
|
1198 |
+
"step": 4200
|
1199 |
+
},
|
1200 |
+
{
|
1201 |
+
"epoch": 1.8032437046521554,
|
1202 |
+
"grad_norm": 16.353532791137695,
|
1203 |
+
"learning_rate": 2.216250395194436e-05,
|
1204 |
+
"loss": 0.034,
|
1205 |
+
"step": 4225
|
1206 |
+
},
|
1207 |
+
{
|
1208 |
+
"epoch": 1.8139137857447718,
|
1209 |
+
"grad_norm": 18.50251007080078,
|
1210 |
+
"learning_rate": 2.1964906734113186e-05,
|
1211 |
+
"loss": 0.0398,
|
1212 |
+
"step": 4250
|
1213 |
+
},
|
1214 |
+
{
|
1215 |
+
"epoch": 1.824583866837388,
|
1216 |
+
"grad_norm": 0.3170296251773834,
|
1217 |
+
"learning_rate": 2.176730951628201e-05,
|
1218 |
+
"loss": 0.0841,
|
1219 |
+
"step": 4275
|
1220 |
+
},
|
1221 |
+
{
|
1222 |
+
"epoch": 1.8352539479300043,
|
1223 |
+
"grad_norm": 0.0018119533779099584,
|
1224 |
+
"learning_rate": 2.156971229845084e-05,
|
1225 |
+
"loss": 0.0634,
|
1226 |
+
"step": 4300
|
1227 |
+
},
|
1228 |
+
{
|
1229 |
+
"epoch": 1.8459240290226204,
|
1230 |
+
"grad_norm": 0.035546980798244476,
|
1231 |
+
"learning_rate": 2.1372115080619663e-05,
|
1232 |
+
"loss": 0.1411,
|
1233 |
+
"step": 4325
|
1234 |
+
},
|
1235 |
+
{
|
1236 |
+
"epoch": 1.8565941101152368,
|
1237 |
+
"grad_norm": 0.5771517753601074,
|
1238 |
+
"learning_rate": 2.1174517862788494e-05,
|
1239 |
+
"loss": 0.0917,
|
1240 |
+
"step": 4350
|
1241 |
+
},
|
1242 |
+
{
|
1243 |
+
"epoch": 1.8672641912078531,
|
1244 |
+
"grad_norm": 0.013703244738280773,
|
1245 |
+
"learning_rate": 2.097692064495732e-05,
|
1246 |
+
"loss": 0.0859,
|
1247 |
+
"step": 4375
|
1248 |
+
},
|
1249 |
+
{
|
1250 |
+
"epoch": 1.8779342723004695,
|
1251 |
+
"grad_norm": 20.912160873413086,
|
1252 |
+
"learning_rate": 2.0779323427126148e-05,
|
1253 |
+
"loss": 0.0813,
|
1254 |
+
"step": 4400
|
1255 |
+
},
|
1256 |
+
{
|
1257 |
+
"epoch": 1.8886043533930859,
|
1258 |
+
"grad_norm": 0.2195858508348465,
|
1259 |
+
"learning_rate": 2.0581726209294974e-05,
|
1260 |
+
"loss": 0.1836,
|
1261 |
+
"step": 4425
|
1262 |
+
},
|
1263 |
+
{
|
1264 |
+
"epoch": 1.8992744344857022,
|
1265 |
+
"grad_norm": 8.048083305358887,
|
1266 |
+
"learning_rate": 2.03841289914638e-05,
|
1267 |
+
"loss": 0.0874,
|
1268 |
+
"step": 4450
|
1269 |
+
},
|
1270 |
+
{
|
1271 |
+
"epoch": 1.9099445155783183,
|
1272 |
+
"grad_norm": 0.010916252620518208,
|
1273 |
+
"learning_rate": 2.0186531773632628e-05,
|
1274 |
+
"loss": 0.0122,
|
1275 |
+
"step": 4475
|
1276 |
+
},
|
1277 |
+
{
|
1278 |
+
"epoch": 1.9206145966709347,
|
1279 |
+
"grad_norm": 29.587345123291016,
|
1280 |
+
"learning_rate": 1.9988934555801455e-05,
|
1281 |
+
"loss": 0.0956,
|
1282 |
+
"step": 4500
|
1283 |
+
},
|
1284 |
+
{
|
1285 |
+
"epoch": 1.9312846777635508,
|
1286 |
+
"grad_norm": 24.480087280273438,
|
1287 |
+
"learning_rate": 1.9791337337970282e-05,
|
1288 |
+
"loss": 0.0754,
|
1289 |
+
"step": 4525
|
1290 |
+
},
|
1291 |
+
{
|
1292 |
+
"epoch": 1.9419547588561672,
|
1293 |
+
"grad_norm": 15.161417961120605,
|
1294 |
+
"learning_rate": 1.959374012013911e-05,
|
1295 |
+
"loss": 0.0414,
|
1296 |
+
"step": 4550
|
1297 |
+
},
|
1298 |
+
{
|
1299 |
+
"epoch": 1.9526248399487836,
|
1300 |
+
"grad_norm": 0.2126699984073639,
|
1301 |
+
"learning_rate": 1.9396142902307936e-05,
|
1302 |
+
"loss": 0.0727,
|
1303 |
+
"step": 4575
|
1304 |
+
},
|
1305 |
+
{
|
1306 |
+
"epoch": 1.9632949210414,
|
1307 |
+
"grad_norm": 9.748570442199707,
|
1308 |
+
"learning_rate": 1.9198545684476763e-05,
|
1309 |
+
"loss": 0.0911,
|
1310 |
+
"step": 4600
|
1311 |
+
},
|
1312 |
+
{
|
1313 |
+
"epoch": 1.9739650021340163,
|
1314 |
+
"grad_norm": 0.005681836046278477,
|
1315 |
+
"learning_rate": 1.900094846664559e-05,
|
1316 |
+
"loss": 0.0652,
|
1317 |
+
"step": 4625
|
1318 |
+
},
|
1319 |
+
{
|
1320 |
+
"epoch": 1.9846350832266326,
|
1321 |
+
"grad_norm": 0.010802343487739563,
|
1322 |
+
"learning_rate": 1.880335124881442e-05,
|
1323 |
+
"loss": 0.0306,
|
1324 |
+
"step": 4650
|
1325 |
+
},
|
1326 |
+
{
|
1327 |
+
"epoch": 1.995305164319249,
|
1328 |
+
"grad_norm": 0.0069721778854727745,
|
1329 |
+
"learning_rate": 1.8605754030983244e-05,
|
1330 |
+
"loss": 0.0705,
|
1331 |
+
"step": 4675
|
1332 |
+
},
|
1333 |
+
{
|
1334 |
+
"epoch": 2.0,
|
1335 |
+
"eval_accuracy": 0.9955176093916756,
|
1336 |
+
"eval_auc": 0.9994811608729328,
|
1337 |
+
"eval_f1": 0.9955137791070284,
|
1338 |
+
"eval_loss": 0.01761646941304207,
|
1339 |
+
"eval_precision": 0.9931798806479113,
|
1340 |
+
"eval_recall": 0.9978586723768736,
|
1341 |
+
"eval_runtime": 1381.3155,
|
1342 |
+
"eval_samples_per_second": 3.392,
|
1343 |
+
"eval_steps_per_second": 0.212,
|
1344 |
+
"step": 4686
|
1345 |
+
},
|
1346 |
+
{
|
1347 |
+
"epoch": 2.0059752454118653,
|
1348 |
+
"grad_norm": 0.010495180264115334,
|
1349 |
+
"learning_rate": 1.840815681315207e-05,
|
1350 |
+
"loss": 0.0779,
|
1351 |
+
"step": 4700
|
1352 |
+
},
|
1353 |
+
{
|
1354 |
+
"epoch": 2.0166453265044813,
|
1355 |
+
"grad_norm": 1.1236854791641235,
|
1356 |
+
"learning_rate": 1.8210559595320897e-05,
|
1357 |
+
"loss": 0.07,
|
1358 |
+
"step": 4725
|
1359 |
+
},
|
1360 |
+
{
|
1361 |
+
"epoch": 2.0273154075970976,
|
1362 |
+
"grad_norm": 0.09067221730947495,
|
1363 |
+
"learning_rate": 1.8012962377489724e-05,
|
1364 |
+
"loss": 0.0346,
|
1365 |
+
"step": 4750
|
1366 |
+
},
|
1367 |
+
{
|
1368 |
+
"epoch": 2.037985488689714,
|
1369 |
+
"grad_norm": 0.005663533229380846,
|
1370 |
+
"learning_rate": 1.7815365159658555e-05,
|
1371 |
+
"loss": 0.0374,
|
1372 |
+
"step": 4775
|
1373 |
+
},
|
1374 |
+
{
|
1375 |
+
"epoch": 2.0486555697823303,
|
1376 |
+
"grad_norm": 0.004197725094854832,
|
1377 |
+
"learning_rate": 1.7617767941827378e-05,
|
1378 |
+
"loss": 0.0368,
|
1379 |
+
"step": 4800
|
1380 |
+
},
|
1381 |
+
{
|
1382 |
+
"epoch": 2.0593256508749467,
|
1383 |
+
"grad_norm": 44.16444778442383,
|
1384 |
+
"learning_rate": 1.742017072399621e-05,
|
1385 |
+
"loss": 0.0744,
|
1386 |
+
"step": 4825
|
1387 |
+
},
|
1388 |
+
{
|
1389 |
+
"epoch": 2.069995731967563,
|
1390 |
+
"grad_norm": 0.021498629823327065,
|
1391 |
+
"learning_rate": 1.7222573506165035e-05,
|
1392 |
+
"loss": 0.1031,
|
1393 |
+
"step": 4850
|
1394 |
+
},
|
1395 |
+
{
|
1396 |
+
"epoch": 2.0806658130601794,
|
1397 |
+
"grad_norm": 0.057749390602111816,
|
1398 |
+
"learning_rate": 1.702497628833386e-05,
|
1399 |
+
"loss": 0.0597,
|
1400 |
+
"step": 4875
|
1401 |
+
},
|
1402 |
+
{
|
1403 |
+
"epoch": 2.0913358941527957,
|
1404 |
+
"grad_norm": 0.15275585651397705,
|
1405 |
+
"learning_rate": 1.682737907050269e-05,
|
1406 |
+
"loss": 0.0305,
|
1407 |
+
"step": 4900
|
1408 |
+
},
|
1409 |
+
{
|
1410 |
+
"epoch": 2.1020059752454117,
|
1411 |
+
"grad_norm": 0.13239429891109467,
|
1412 |
+
"learning_rate": 1.6629781852671516e-05,
|
1413 |
+
"loss": 0.0338,
|
1414 |
+
"step": 4925
|
1415 |
+
},
|
1416 |
+
{
|
1417 |
+
"epoch": 2.112676056338028,
|
1418 |
+
"grad_norm": 9.17479419708252,
|
1419 |
+
"learning_rate": 1.6432184634840343e-05,
|
1420 |
+
"loss": 0.0517,
|
1421 |
+
"step": 4950
|
1422 |
+
},
|
1423 |
+
{
|
1424 |
+
"epoch": 2.1233461374306444,
|
1425 |
+
"grad_norm": 0.002612267853692174,
|
1426 |
+
"learning_rate": 1.623458741700917e-05,
|
1427 |
+
"loss": 0.0472,
|
1428 |
+
"step": 4975
|
1429 |
+
},
|
1430 |
+
{
|
1431 |
+
"epoch": 2.1340162185232607,
|
1432 |
+
"grad_norm": 6.946547031402588,
|
1433 |
+
"learning_rate": 1.6036990199177997e-05,
|
1434 |
+
"loss": 0.0854,
|
1435 |
+
"step": 5000
|
1436 |
+
},
|
1437 |
+
{
|
1438 |
+
"epoch": 2.144686299615877,
|
1439 |
+
"grad_norm": 0.0008439666125923395,
|
1440 |
+
"learning_rate": 1.5839392981346824e-05,
|
1441 |
+
"loss": 0.0394,
|
1442 |
+
"step": 5025
|
1443 |
+
},
|
1444 |
+
{
|
1445 |
+
"epoch": 2.1553563807084934,
|
1446 |
+
"grad_norm": 0.03294059634208679,
|
1447 |
+
"learning_rate": 1.564179576351565e-05,
|
1448 |
+
"loss": 0.0745,
|
1449 |
+
"step": 5050
|
1450 |
+
},
|
1451 |
+
{
|
1452 |
+
"epoch": 2.16602646180111,
|
1453 |
+
"grad_norm": 5.746143817901611,
|
1454 |
+
"learning_rate": 1.5444198545684478e-05,
|
1455 |
+
"loss": 0.0595,
|
1456 |
+
"step": 5075
|
1457 |
+
},
|
1458 |
+
{
|
1459 |
+
"epoch": 2.176696542893726,
|
1460 |
+
"grad_norm": 0.020633000880479813,
|
1461 |
+
"learning_rate": 1.5246601327853304e-05,
|
1462 |
+
"loss": 0.0027,
|
1463 |
+
"step": 5100
|
1464 |
+
},
|
1465 |
+
{
|
1466 |
+
"epoch": 2.1873666239863425,
|
1467 |
+
"grad_norm": 0.08342266827821732,
|
1468 |
+
"learning_rate": 1.5049004110022133e-05,
|
1469 |
+
"loss": 0.0944,
|
1470 |
+
"step": 5125
|
1471 |
+
},
|
1472 |
+
{
|
1473 |
+
"epoch": 2.1980367050789584,
|
1474 |
+
"grad_norm": 6.828884124755859,
|
1475 |
+
"learning_rate": 1.4851406892190958e-05,
|
1476 |
+
"loss": 0.0776,
|
1477 |
+
"step": 5150
|
1478 |
+
},
|
1479 |
+
{
|
1480 |
+
"epoch": 2.208706786171575,
|
1481 |
+
"grad_norm": 2.5025949478149414,
|
1482 |
+
"learning_rate": 1.4653809674359787e-05,
|
1483 |
+
"loss": 0.0812,
|
1484 |
+
"step": 5175
|
1485 |
+
},
|
1486 |
+
{
|
1487 |
+
"epoch": 2.219376867264191,
|
1488 |
+
"grad_norm": 0.010205933824181557,
|
1489 |
+
"learning_rate": 1.4456212456528612e-05,
|
1490 |
+
"loss": 0.0416,
|
1491 |
+
"step": 5200
|
1492 |
+
},
|
1493 |
+
{
|
1494 |
+
"epoch": 2.2300469483568075,
|
1495 |
+
"grad_norm": 0.014995824545621872,
|
1496 |
+
"learning_rate": 1.4258615238697439e-05,
|
1497 |
+
"loss": 0.0476,
|
1498 |
+
"step": 5225
|
1499 |
+
},
|
1500 |
+
{
|
1501 |
+
"epoch": 2.240717029449424,
|
1502 |
+
"grad_norm": 0.011430823244154453,
|
1503 |
+
"learning_rate": 1.4061018020866268e-05,
|
1504 |
+
"loss": 0.0688,
|
1505 |
+
"step": 5250
|
1506 |
+
},
|
1507 |
+
{
|
1508 |
+
"epoch": 2.25138711054204,
|
1509 |
+
"grad_norm": 0.024417445063591003,
|
1510 |
+
"learning_rate": 1.3863420803035093e-05,
|
1511 |
+
"loss": 0.0445,
|
1512 |
+
"step": 5275
|
1513 |
+
},
|
1514 |
+
{
|
1515 |
+
"epoch": 2.2620571916346566,
|
1516 |
+
"grad_norm": 0.04339329153299332,
|
1517 |
+
"learning_rate": 1.3665823585203921e-05,
|
1518 |
+
"loss": 0.0443,
|
1519 |
+
"step": 5300
|
1520 |
+
},
|
1521 |
+
{
|
1522 |
+
"epoch": 2.2727272727272725,
|
1523 |
+
"grad_norm": 0.006649048998951912,
|
1524 |
+
"learning_rate": 1.346822636737275e-05,
|
1525 |
+
"loss": 0.076,
|
1526 |
+
"step": 5325
|
1527 |
+
},
|
1528 |
+
{
|
1529 |
+
"epoch": 2.283397353819889,
|
1530 |
+
"grad_norm": 11.57873821258545,
|
1531 |
+
"learning_rate": 1.3270629149541575e-05,
|
1532 |
+
"loss": 0.0266,
|
1533 |
+
"step": 5350
|
1534 |
+
},
|
1535 |
+
{
|
1536 |
+
"epoch": 2.294067434912505,
|
1537 |
+
"grad_norm": 0.12145959585905075,
|
1538 |
+
"learning_rate": 1.3073031931710402e-05,
|
1539 |
+
"loss": 0.0953,
|
1540 |
+
"step": 5375
|
1541 |
+
},
|
1542 |
+
{
|
1543 |
+
"epoch": 2.3047375160051216,
|
1544 |
+
"grad_norm": 0.03219663351774216,
|
1545 |
+
"learning_rate": 1.2875434713879227e-05,
|
1546 |
+
"loss": 0.0797,
|
1547 |
+
"step": 5400
|
1548 |
+
},
|
1549 |
+
{
|
1550 |
+
"epoch": 2.315407597097738,
|
1551 |
+
"grad_norm": 0.011474654078483582,
|
1552 |
+
"learning_rate": 1.2677837496048056e-05,
|
1553 |
+
"loss": 0.0274,
|
1554 |
+
"step": 5425
|
1555 |
+
},
|
1556 |
+
{
|
1557 |
+
"epoch": 2.3260776781903543,
|
1558 |
+
"grad_norm": 0.001972693484276533,
|
1559 |
+
"learning_rate": 1.2480240278216883e-05,
|
1560 |
+
"loss": 0.0563,
|
1561 |
+
"step": 5450
|
1562 |
+
},
|
1563 |
+
{
|
1564 |
+
"epoch": 2.3367477592829706,
|
1565 |
+
"grad_norm": 0.9134419560432434,
|
1566 |
+
"learning_rate": 1.2282643060385712e-05,
|
1567 |
+
"loss": 0.0292,
|
1568 |
+
"step": 5475
|
1569 |
+
},
|
1570 |
+
{
|
1571 |
+
"epoch": 2.347417840375587,
|
1572 |
+
"grad_norm": 0.002546141389757395,
|
1573 |
+
"learning_rate": 1.2085045842554538e-05,
|
1574 |
+
"loss": 0.0428,
|
1575 |
+
"step": 5500
|
1576 |
+
},
|
1577 |
+
{
|
1578 |
+
"epoch": 2.3580879214682033,
|
1579 |
+
"grad_norm": 0.0010626994771882892,
|
1580 |
+
"learning_rate": 1.1887448624723364e-05,
|
1581 |
+
"loss": 0.076,
|
1582 |
+
"step": 5525
|
1583 |
+
},
|
1584 |
+
{
|
1585 |
+
"epoch": 2.3687580025608197,
|
1586 |
+
"grad_norm": 0.0040688286535441875,
|
1587 |
+
"learning_rate": 1.168985140689219e-05,
|
1588 |
+
"loss": 0.0782,
|
1589 |
+
"step": 5550
|
1590 |
+
},
|
1591 |
+
{
|
1592 |
+
"epoch": 2.3794280836534356,
|
1593 |
+
"grad_norm": 0.02786325477063656,
|
1594 |
+
"learning_rate": 1.149225418906102e-05,
|
1595 |
+
"loss": 0.0123,
|
1596 |
+
"step": 5575
|
1597 |
+
},
|
1598 |
+
{
|
1599 |
+
"epoch": 2.390098164746052,
|
1600 |
+
"grad_norm": 0.4415593147277832,
|
1601 |
+
"learning_rate": 1.1294656971229846e-05,
|
1602 |
+
"loss": 0.0589,
|
1603 |
+
"step": 5600
|
1604 |
+
},
|
1605 |
+
{
|
1606 |
+
"epoch": 2.4007682458386683,
|
1607 |
+
"grad_norm": 0.0095005938783288,
|
1608 |
+
"learning_rate": 1.1097059753398673e-05,
|
1609 |
+
"loss": 0.1002,
|
1610 |
+
"step": 5625
|
1611 |
+
},
|
1612 |
+
{
|
1613 |
+
"epoch": 2.4114383269312847,
|
1614 |
+
"grad_norm": 0.11727124452590942,
|
1615 |
+
"learning_rate": 1.08994625355675e-05,
|
1616 |
+
"loss": 0.0324,
|
1617 |
+
"step": 5650
|
1618 |
+
},
|
1619 |
+
{
|
1620 |
+
"epoch": 2.422108408023901,
|
1621 |
+
"grad_norm": 3.638735294342041,
|
1622 |
+
"learning_rate": 1.0701865317736327e-05,
|
1623 |
+
"loss": 0.1456,
|
1624 |
+
"step": 5675
|
1625 |
+
},
|
1626 |
+
{
|
1627 |
+
"epoch": 2.4327784891165174,
|
1628 |
+
"grad_norm": 0.025659436360001564,
|
1629 |
+
"learning_rate": 1.0504268099905154e-05,
|
1630 |
+
"loss": 0.0323,
|
1631 |
+
"step": 5700
|
1632 |
+
},
|
1633 |
+
{
|
1634 |
+
"epoch": 2.4434485702091338,
|
1635 |
+
"grad_norm": 0.000763273739721626,
|
1636 |
+
"learning_rate": 1.030667088207398e-05,
|
1637 |
+
"loss": 0.019,
|
1638 |
+
"step": 5725
|
1639 |
+
},
|
1640 |
+
{
|
1641 |
+
"epoch": 2.4541186513017497,
|
1642 |
+
"grad_norm": 13.73189926147461,
|
1643 |
+
"learning_rate": 1.0109073664242808e-05,
|
1644 |
+
"loss": 0.0804,
|
1645 |
+
"step": 5750
|
1646 |
+
},
|
1647 |
+
{
|
1648 |
+
"epoch": 2.464788732394366,
|
1649 |
+
"grad_norm": 13.580681800842285,
|
1650 |
+
"learning_rate": 9.911476446411636e-06,
|
1651 |
+
"loss": 0.0628,
|
1652 |
+
"step": 5775
|
1653 |
+
},
|
1654 |
+
{
|
1655 |
+
"epoch": 2.4754588134869824,
|
1656 |
+
"grad_norm": 0.0029777430463582277,
|
1657 |
+
"learning_rate": 9.713879228580461e-06,
|
1658 |
+
"loss": 0.0807,
|
1659 |
+
"step": 5800
|
1660 |
+
},
|
1661 |
+
{
|
1662 |
+
"epoch": 2.4861288945795987,
|
1663 |
+
"grad_norm": 0.003513498930260539,
|
1664 |
+
"learning_rate": 9.516282010749288e-06,
|
1665 |
+
"loss": 0.0826,
|
1666 |
+
"step": 5825
|
1667 |
+
},
|
1668 |
+
{
|
1669 |
+
"epoch": 2.496798975672215,
|
1670 |
+
"grad_norm": 0.009089035913348198,
|
1671 |
+
"learning_rate": 9.318684792918115e-06,
|
1672 |
+
"loss": 0.0406,
|
1673 |
+
"step": 5850
|
1674 |
+
},
|
1675 |
+
{
|
1676 |
+
"epoch": 2.5074690567648314,
|
1677 |
+
"grad_norm": 0.0068649169988930225,
|
1678 |
+
"learning_rate": 9.121087575086944e-06,
|
1679 |
+
"loss": 0.1065,
|
1680 |
+
"step": 5875
|
1681 |
+
},
|
1682 |
+
{
|
1683 |
+
"epoch": 2.518139137857448,
|
1684 |
+
"grad_norm": 20.251689910888672,
|
1685 |
+
"learning_rate": 8.92349035725577e-06,
|
1686 |
+
"loss": 0.1227,
|
1687 |
+
"step": 5900
|
1688 |
+
},
|
1689 |
+
{
|
1690 |
+
"epoch": 2.528809218950064,
|
1691 |
+
"grad_norm": 0.1939096450805664,
|
1692 |
+
"learning_rate": 8.725893139424598e-06,
|
1693 |
+
"loss": 0.0343,
|
1694 |
+
"step": 5925
|
1695 |
+
},
|
1696 |
+
{
|
1697 |
+
"epoch": 2.5394793000426805,
|
1698 |
+
"grad_norm": 0.6377553939819336,
|
1699 |
+
"learning_rate": 8.528295921593425e-06,
|
1700 |
+
"loss": 0.0247,
|
1701 |
+
"step": 5950
|
1702 |
+
},
|
1703 |
+
{
|
1704 |
+
"epoch": 2.550149381135297,
|
1705 |
+
"grad_norm": 1.8269633054733276,
|
1706 |
+
"learning_rate": 8.330698703762251e-06,
|
1707 |
+
"loss": 0.057,
|
1708 |
+
"step": 5975
|
1709 |
+
},
|
1710 |
+
{
|
1711 |
+
"epoch": 2.560819462227913,
|
1712 |
+
"grad_norm": 0.004431420471519232,
|
1713 |
+
"learning_rate": 8.133101485931078e-06,
|
1714 |
+
"loss": 0.0184,
|
1715 |
+
"step": 6000
|
1716 |
+
},
|
1717 |
+
{
|
1718 |
+
"epoch": 2.571489543320529,
|
1719 |
+
"grad_norm": 0.0886378139257431,
|
1720 |
+
"learning_rate": 7.935504268099905e-06,
|
1721 |
+
"loss": 0.1559,
|
1722 |
+
"step": 6025
|
1723 |
+
},
|
1724 |
+
{
|
1725 |
+
"epoch": 2.5821596244131455,
|
1726 |
+
"grad_norm": 0.011996462009847164,
|
1727 |
+
"learning_rate": 7.737907050268732e-06,
|
1728 |
+
"loss": 0.1207,
|
1729 |
+
"step": 6050
|
1730 |
+
},
|
1731 |
+
{
|
1732 |
+
"epoch": 2.592829705505762,
|
1733 |
+
"grad_norm": 0.025645237416028976,
|
1734 |
+
"learning_rate": 7.54030983243756e-06,
|
1735 |
+
"loss": 0.0327,
|
1736 |
+
"step": 6075
|
1737 |
+
},
|
1738 |
+
{
|
1739 |
+
"epoch": 2.603499786598378,
|
1740 |
+
"grad_norm": 0.05047876015305519,
|
1741 |
+
"learning_rate": 7.342712614606387e-06,
|
1742 |
+
"loss": 0.07,
|
1743 |
+
"step": 6100
|
1744 |
+
},
|
1745 |
+
{
|
1746 |
+
"epoch": 2.6141698676909946,
|
1747 |
+
"grad_norm": 0.24467694759368896,
|
1748 |
+
"learning_rate": 7.145115396775214e-06,
|
1749 |
+
"loss": 0.0104,
|
1750 |
+
"step": 6125
|
1751 |
+
},
|
1752 |
+
{
|
1753 |
+
"epoch": 2.6248399487836105,
|
1754 |
+
"grad_norm": 0.07860807329416275,
|
1755 |
+
"learning_rate": 6.94751817894404e-06,
|
1756 |
+
"loss": 0.073,
|
1757 |
+
"step": 6150
|
1758 |
+
},
|
1759 |
+
{
|
1760 |
+
"epoch": 2.635510029876227,
|
1761 |
+
"grad_norm": 0.03573083132505417,
|
1762 |
+
"learning_rate": 6.7499209611128685e-06,
|
1763 |
+
"loss": 0.0616,
|
1764 |
+
"step": 6175
|
1765 |
+
},
|
1766 |
+
{
|
1767 |
+
"epoch": 2.646180110968843,
|
1768 |
+
"grad_norm": 1.0572718381881714,
|
1769 |
+
"learning_rate": 6.552323743281695e-06,
|
1770 |
+
"loss": 0.0872,
|
1771 |
+
"step": 6200
|
1772 |
+
},
|
1773 |
+
{
|
1774 |
+
"epoch": 2.6568501920614596,
|
1775 |
+
"grad_norm": 0.009367382153868675,
|
1776 |
+
"learning_rate": 6.3547265254505215e-06,
|
1777 |
+
"loss": 0.0388,
|
1778 |
+
"step": 6225
|
1779 |
+
},
|
1780 |
+
{
|
1781 |
+
"epoch": 2.667520273154076,
|
1782 |
+
"grad_norm": 0.02948431856930256,
|
1783 |
+
"learning_rate": 6.157129307619349e-06,
|
1784 |
+
"loss": 0.0831,
|
1785 |
+
"step": 6250
|
1786 |
+
},
|
1787 |
+
{
|
1788 |
+
"epoch": 2.6781903542466923,
|
1789 |
+
"grad_norm": 0.0797591432929039,
|
1790 |
+
"learning_rate": 5.959532089788176e-06,
|
1791 |
+
"loss": 0.049,
|
1792 |
+
"step": 6275
|
1793 |
+
},
|
1794 |
+
{
|
1795 |
+
"epoch": 2.6888604353393086,
|
1796 |
+
"grad_norm": 0.034841641783714294,
|
1797 |
+
"learning_rate": 5.761934871957003e-06,
|
1798 |
+
"loss": 0.0055,
|
1799 |
+
"step": 6300
|
1800 |
+
},
|
1801 |
+
{
|
1802 |
+
"epoch": 2.699530516431925,
|
1803 |
+
"grad_norm": 0.07419008761644363,
|
1804 |
+
"learning_rate": 5.56433765412583e-06,
|
1805 |
+
"loss": 0.1128,
|
1806 |
+
"step": 6325
|
1807 |
+
},
|
1808 |
+
{
|
1809 |
+
"epoch": 2.7102005975245413,
|
1810 |
+
"grad_norm": 0.0030936244875192642,
|
1811 |
+
"learning_rate": 5.366740436294658e-06,
|
1812 |
+
"loss": 0.0564,
|
1813 |
+
"step": 6350
|
1814 |
+
},
|
1815 |
+
{
|
1816 |
+
"epoch": 2.7208706786171577,
|
1817 |
+
"grad_norm": 0.003594167297706008,
|
1818 |
+
"learning_rate": 5.169143218463484e-06,
|
1819 |
+
"loss": 0.0928,
|
1820 |
+
"step": 6375
|
1821 |
+
},
|
1822 |
+
{
|
1823 |
+
"epoch": 2.731540759709774,
|
1824 |
+
"grad_norm": 0.03685923293232918,
|
1825 |
+
"learning_rate": 4.9715460006323115e-06,
|
1826 |
+
"loss": 0.0502,
|
1827 |
+
"step": 6400
|
1828 |
+
},
|
1829 |
+
{
|
1830 |
+
"epoch": 2.74221084080239,
|
1831 |
+
"grad_norm": 0.019324608147144318,
|
1832 |
+
"learning_rate": 4.7739487828011385e-06,
|
1833 |
+
"loss": 0.0204,
|
1834 |
+
"step": 6425
|
1835 |
+
},
|
1836 |
+
{
|
1837 |
+
"epoch": 2.7528809218950063,
|
1838 |
+
"grad_norm": 22.45241355895996,
|
1839 |
+
"learning_rate": 4.576351564969965e-06,
|
1840 |
+
"loss": 0.0867,
|
1841 |
+
"step": 6450
|
1842 |
+
},
|
1843 |
+
{
|
1844 |
+
"epoch": 2.7635510029876227,
|
1845 |
+
"grad_norm": 0.14366178214550018,
|
1846 |
+
"learning_rate": 4.378754347138792e-06,
|
1847 |
+
"loss": 0.0366,
|
1848 |
+
"step": 6475
|
1849 |
+
},
|
1850 |
+
{
|
1851 |
+
"epoch": 2.774221084080239,
|
1852 |
+
"grad_norm": 1.7778751850128174,
|
1853 |
+
"learning_rate": 4.181157129307619e-06,
|
1854 |
+
"loss": 0.097,
|
1855 |
+
"step": 6500
|
1856 |
+
},
|
1857 |
+
{
|
1858 |
+
"epoch": 2.7848911651728554,
|
1859 |
+
"grad_norm": 4.860283851623535,
|
1860 |
+
"learning_rate": 3.983559911476446e-06,
|
1861 |
+
"loss": 0.0884,
|
1862 |
+
"step": 6525
|
1863 |
+
},
|
1864 |
+
{
|
1865 |
+
"epoch": 2.7955612462654718,
|
1866 |
+
"grad_norm": 0.00165728444699198,
|
1867 |
+
"learning_rate": 3.785962693645274e-06,
|
1868 |
+
"loss": 0.0195,
|
1869 |
+
"step": 6550
|
1870 |
+
},
|
1871 |
+
{
|
1872 |
+
"epoch": 2.8062313273580877,
|
1873 |
+
"grad_norm": 0.03730342909693718,
|
1874 |
+
"learning_rate": 3.5883654758141003e-06,
|
1875 |
+
"loss": 0.0412,
|
1876 |
+
"step": 6575
|
1877 |
+
},
|
1878 |
+
{
|
1879 |
+
"epoch": 2.816901408450704,
|
1880 |
+
"grad_norm": 0.0034714387729763985,
|
1881 |
+
"learning_rate": 3.3907682579829277e-06,
|
1882 |
+
"loss": 0.0714,
|
1883 |
+
"step": 6600
|
1884 |
+
},
|
1885 |
+
{
|
1886 |
+
"epoch": 2.8275714895433204,
|
1887 |
+
"grad_norm": 0.06972959637641907,
|
1888 |
+
"learning_rate": 3.1931710401517546e-06,
|
1889 |
+
"loss": 0.0082,
|
1890 |
+
"step": 6625
|
1891 |
+
},
|
1892 |
+
{
|
1893 |
+
"epoch": 2.8382415706359367,
|
1894 |
+
"grad_norm": 0.022793615236878395,
|
1895 |
+
"learning_rate": 2.995573822320582e-06,
|
1896 |
+
"loss": 0.0705,
|
1897 |
+
"step": 6650
|
1898 |
+
},
|
1899 |
+
{
|
1900 |
+
"epoch": 2.848911651728553,
|
1901 |
+
"grad_norm": 0.05906020104885101,
|
1902 |
+
"learning_rate": 2.797976604489409e-06,
|
1903 |
+
"loss": 0.0435,
|
1904 |
+
"step": 6675
|
1905 |
+
},
|
1906 |
+
{
|
1907 |
+
"epoch": 2.8595817328211695,
|
1908 |
+
"grad_norm": 0.012936658225953579,
|
1909 |
+
"learning_rate": 2.600379386658236e-06,
|
1910 |
+
"loss": 0.0852,
|
1911 |
+
"step": 6700
|
1912 |
+
},
|
1913 |
+
{
|
1914 |
+
"epoch": 2.870251813913786,
|
1915 |
+
"grad_norm": 0.004626471549272537,
|
1916 |
+
"learning_rate": 2.402782168827063e-06,
|
1917 |
+
"loss": 0.03,
|
1918 |
+
"step": 6725
|
1919 |
+
},
|
1920 |
+
{
|
1921 |
+
"epoch": 2.880921895006402,
|
1922 |
+
"grad_norm": 0.08372853696346283,
|
1923 |
+
"learning_rate": 2.20518495099589e-06,
|
1924 |
+
"loss": 0.0152,
|
1925 |
+
"step": 6750
|
1926 |
+
},
|
1927 |
+
{
|
1928 |
+
"epoch": 2.8915919760990185,
|
1929 |
+
"grad_norm": 1.3282454013824463,
|
1930 |
+
"learning_rate": 2.0075877331647173e-06,
|
1931 |
+
"loss": 0.0816,
|
1932 |
+
"step": 6775
|
1933 |
+
},
|
1934 |
+
{
|
1935 |
+
"epoch": 2.902262057191635,
|
1936 |
+
"grad_norm": 3.8434557914733887,
|
1937 |
+
"learning_rate": 1.8099905153335442e-06,
|
1938 |
+
"loss": 0.05,
|
1939 |
+
"step": 6800
|
1940 |
+
},
|
1941 |
+
{
|
1942 |
+
"epoch": 2.9129321382842512,
|
1943 |
+
"grad_norm": 0.008098805323243141,
|
1944 |
+
"learning_rate": 1.6123932975023712e-06,
|
1945 |
+
"loss": 0.0347,
|
1946 |
+
"step": 6825
|
1947 |
+
},
|
1948 |
+
{
|
1949 |
+
"epoch": 2.923602219376867,
|
1950 |
+
"grad_norm": 0.0022435523569583893,
|
1951 |
+
"learning_rate": 1.4147960796711983e-06,
|
1952 |
+
"loss": 0.0214,
|
1953 |
+
"step": 6850
|
1954 |
+
},
|
1955 |
+
{
|
1956 |
+
"epoch": 2.9342723004694835,
|
1957 |
+
"grad_norm": 7.002129554748535,
|
1958 |
+
"learning_rate": 1.2171988618400254e-06,
|
1959 |
+
"loss": 0.0121,
|
1960 |
+
"step": 6875
|
1961 |
+
},
|
1962 |
+
{
|
1963 |
+
"epoch": 2.9449423815621,
|
1964 |
+
"grad_norm": 0.11791983246803284,
|
1965 |
+
"learning_rate": 1.0196016440088523e-06,
|
1966 |
+
"loss": 0.1015,
|
1967 |
+
"step": 6900
|
1968 |
+
},
|
1969 |
+
{
|
1970 |
+
"epoch": 2.955612462654716,
|
1971 |
+
"grad_norm": 0.0031008380465209484,
|
1972 |
+
"learning_rate": 8.220044261776794e-07,
|
1973 |
+
"loss": 0.0016,
|
1974 |
+
"step": 6925
|
1975 |
+
},
|
1976 |
+
{
|
1977 |
+
"epoch": 2.9662825437473326,
|
1978 |
+
"grad_norm": 0.00397999444976449,
|
1979 |
+
"learning_rate": 6.244072083465065e-07,
|
1980 |
+
"loss": 0.0238,
|
1981 |
+
"step": 6950
|
1982 |
+
},
|
1983 |
+
{
|
1984 |
+
"epoch": 2.976952624839949,
|
1985 |
+
"grad_norm": 0.14921312034130096,
|
1986 |
+
"learning_rate": 4.2680999051533353e-07,
|
1987 |
+
"loss": 0.0147,
|
1988 |
+
"step": 6975
|
1989 |
+
},
|
1990 |
+
{
|
1991 |
+
"epoch": 2.987622705932565,
|
1992 |
+
"grad_norm": 0.005675365682691336,
|
1993 |
+
"learning_rate": 2.2921277268416063e-07,
|
1994 |
+
"loss": 0.0175,
|
1995 |
+
"step": 7000
|
1996 |
+
},
|
1997 |
+
{
|
1998 |
+
"epoch": 2.998292787025181,
|
1999 |
+
"grad_norm": 0.009994860738515854,
|
2000 |
+
"learning_rate": 3.1615554852987675e-08,
|
2001 |
+
"loss": 0.0694,
|
2002 |
+
"step": 7025
|
2003 |
+
},
|
2004 |
+
{
|
2005 |
+
"epoch": 3.0,
|
2006 |
+
"eval_accuracy": 0.9961579509071505,
|
2007 |
+
"eval_auc": 0.9994886327395326,
|
2008 |
+
"eval_f1": 0.9961538461538462,
|
2009 |
+
"eval_loss": 0.0158307533711195,
|
2010 |
+
"eval_precision": 0.9940298507462687,
|
2011 |
+
"eval_recall": 0.9982869379014989,
|
2012 |
+
"eval_runtime": 1329.756,
|
2013 |
+
"eval_samples_per_second": 3.523,
|
2014 |
+
"eval_steps_per_second": 0.22,
|
2015 |
+
"step": 7029
|
2016 |
+
}
|
2017 |
+
],
|
2018 |
+
"logging_steps": 25,
|
2019 |
+
"max_steps": 7029,
|
2020 |
+
"num_input_tokens_seen": 0,
|
2021 |
+
"num_train_epochs": 3,
|
2022 |
+
"save_steps": 500,
|
2023 |
+
"stateful_callbacks": {
|
2024 |
+
"EarlyStoppingCallback": {
|
2025 |
+
"args": {
|
2026 |
+
"early_stopping_patience": 5,
|
2027 |
+
"early_stopping_threshold": 0.01
|
2028 |
+
},
|
2029 |
+
"attributes": {
|
2030 |
+
"early_stopping_patience_counter": 0
|
2031 |
+
}
|
2032 |
+
},
|
2033 |
+
"TrainerControl": {
|
2034 |
+
"args": {
|
2035 |
+
"should_epoch_stop": false,
|
2036 |
+
"should_evaluate": false,
|
2037 |
+
"should_log": false,
|
2038 |
+
"should_save": true,
|
2039 |
+
"should_training_stop": true
|
2040 |
+
},
|
2041 |
+
"attributes": {}
|
2042 |
+
}
|
2043 |
+
},
|
2044 |
+
"total_flos": 4.355902228225831e+18,
|
2045 |
+
"train_batch_size": 8,
|
2046 |
+
"trial_name": null,
|
2047 |
+
"trial_params": null
|
2048 |
+
}
|
checkpoint-7029/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a662cb1b8424c7697385438632e076ac211cd4a2f771733a58f0b62740b3dfd4
|
3 |
+
size 5240
|
config.json
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "google/vit-base-patch16-224",
|
3 |
+
"_num_labels": 2,
|
4 |
+
"architectures": [
|
5 |
+
"ViTForImageClassification"
|
6 |
+
],
|
7 |
+
"attention_probs_dropout_prob": 0.0,
|
8 |
+
"encoder_stride": 16,
|
9 |
+
"hidden_act": "gelu",
|
10 |
+
"hidden_dropout_prob": 0.0,
|
11 |
+
"hidden_size": 768,
|
12 |
+
"id2label": {
|
13 |
+
"0": "Cat",
|
14 |
+
"1": "Dog"
|
15 |
+
},
|
16 |
+
"image_size": 224,
|
17 |
+
"initializer_range": 0.02,
|
18 |
+
"intermediate_size": 3072,
|
19 |
+
"label2id": {
|
20 |
+
"Cat": 0,
|
21 |
+
"Dog": 1
|
22 |
+
},
|
23 |
+
"layer_norm_eps": 1e-12,
|
24 |
+
"model_type": "vit",
|
25 |
+
"num_attention_heads": 12,
|
26 |
+
"num_channels": 3,
|
27 |
+
"num_hidden_layers": 12,
|
28 |
+
"patch_size": 16,
|
29 |
+
"problem_type": "single_label_classification",
|
30 |
+
"qkv_bias": true,
|
31 |
+
"torch_dtype": "float32",
|
32 |
+
"transformers_version": "4.44.2"
|
33 |
+
}
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dee3eed218313e8ff29b584802246af13c768a49b4c314f646c4fa450b081f8d
|
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 |
+
}
|
runs/Sep24_16-03-26_r-pandalori-autotrain-cats-and-dogs-8cfe05bk-8ef12-7ju5g/events.out.tfevents.1727193808.r-pandalori-autotrain-cats-and-dogs-8cfe05bk-8ef12-7ju5g.207.0
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2a210962bd98f4f7f9b809fbeef5897cb3aecdbbe48c10640bbc470cab29968f
|
3 |
+
size 66148
|
runs/Sep24_16-03-26_r-pandalori-autotrain-cats-and-dogs-8cfe05bk-8ef12-7ju5g/events.out.tfevents.1727249929.r-pandalori-autotrain-cats-and-dogs-8cfe05bk-8ef12-7ju5g.207.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:46d02389d22e10b07145a8199dc76a8d90bd471571df0068b3c72494345043ef
|
3 |
+
size 607
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a662cb1b8424c7697385438632e076ac211cd4a2f771733a58f0b62740b3dfd4
|
3 |
+
size 5240
|
training_params.json
ADDED
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"data_path": "autotrain-ltphq-1aue6/autotrain-data",
|
3 |
+
"model": "google/vit-base-patch16-224",
|
4 |
+
"username": "pandalori",
|
5 |
+
"lr": 5e-05,
|
6 |
+
"epochs": 3,
|
7 |
+
"batch_size": 8,
|
8 |
+
"warmup_ratio": 0.1,
|
9 |
+
"gradient_accumulation": 1,
|
10 |
+
"optimizer": "adamw_torch",
|
11 |
+
"scheduler": "linear",
|
12 |
+
"weight_decay": 0.0,
|
13 |
+
"max_grad_norm": 1.0,
|
14 |
+
"seed": 42,
|
15 |
+
"train_split": "train",
|
16 |
+
"valid_split": "validation",
|
17 |
+
"logging_steps": -1,
|
18 |
+
"project_name": "autotrain-ltphq-1aue6",
|
19 |
+
"auto_find_batch_size": false,
|
20 |
+
"mixed_precision": "fp16",
|
21 |
+
"save_total_limit": 1,
|
22 |
+
"push_to_hub": true,
|
23 |
+
"eval_strategy": "epoch",
|
24 |
+
"image_column": "autotrain_image",
|
25 |
+
"target_column": "autotrain_label",
|
26 |
+
"log": "tensorboard",
|
27 |
+
"early_stopping_patience": 5,
|
28 |
+
"early_stopping_threshold": 0.01
|
29 |
+
}
|