Upload folder using huggingface_hub
Browse files- README.md +31 -0
- checkpoint-4587/config.json +33 -0
- checkpoint-4587/model.safetensors +3 -0
- checkpoint-4587/optimizer.pt +3 -0
- checkpoint-4587/rng_state.pth +3 -0
- checkpoint-4587/scheduler.pt +3 -0
- checkpoint-4587/trainer_state.json +1362 -0
- checkpoint-4587/training_args.bin +3 -0
- config.json +33 -0
- model.safetensors +3 -0
- preprocessor_config.json +36 -0
- runs/Jun14_18-14-50_r-luminoussg-autotrain-s6ik9skt-775cb-hpjwc/events.out.tfevents.1718388891.r-luminoussg-autotrain-s6ik9skt-775cb-hpjwc.147.0 +2 -2
- runs/Jun14_18-14-50_r-luminoussg-autotrain-s6ik9skt-775cb-hpjwc/events.out.tfevents.1718389367.r-luminoussg-autotrain-s6ik9skt-775cb-hpjwc.147.1 +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: 0.08583971858024597
|
| 22 |
+
|
| 23 |
+
f1: 0.9841642228739003
|
| 24 |
+
|
| 25 |
+
precision: 0.9940758293838863
|
| 26 |
+
|
| 27 |
+
recall: 0.9744483159117305
|
| 28 |
+
|
| 29 |
+
auc: 0.9978146585977138
|
| 30 |
+
|
| 31 |
+
accuracy: 0.976824034334764
|
checkpoint-4587/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": "NORMAL",
|
| 14 |
+
"1": "PNEUMONIA"
|
| 15 |
+
},
|
| 16 |
+
"image_size": 224,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 3072,
|
| 19 |
+
"label2id": {
|
| 20 |
+
"NORMAL": 0,
|
| 21 |
+
"PNEUMONIA": 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": "4.41.2"
|
| 33 |
+
}
|
checkpoint-4587/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d0d15fb80234d91522f046bab233bf009c7f9fec36b2a1bd1b48d1dbef6aad00
|
| 3 |
+
size 343223968
|
checkpoint-4587/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:08024d8d18ea2471bdd83cff8217eef60625529ef1b3a3e198d7246e517e603b
|
| 3 |
+
size 686568890
|
checkpoint-4587/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d091fb1e783f68a2ade35d8ed8e89d2def1ec0c6bf54e824bc50f82371741f40
|
| 3 |
+
size 14244
|
checkpoint-4587/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:02599d5d666b7daf9afcf6444cd68a2c70bc5202b1cef3e096d191f638b23127
|
| 3 |
+
size 1064
|
checkpoint-4587/trainer_state.json
ADDED
|
@@ -0,0 +1,1362 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.08583971858024597,
|
| 3 |
+
"best_model_checkpoint": "xraynewww/checkpoint-4587",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4587,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.016350555918901243,
|
| 13 |
+
"grad_norm": 8.776586532592773,
|
| 14 |
+
"learning_rate": 2.6143790849673204e-06,
|
| 15 |
+
"loss": 0.769,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.032701111837802485,
|
| 20 |
+
"grad_norm": 6.286694526672363,
|
| 21 |
+
"learning_rate": 5.33769063180828e-06,
|
| 22 |
+
"loss": 0.4887,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.04905166775670373,
|
| 27 |
+
"grad_norm": 3.120753288269043,
|
| 28 |
+
"learning_rate": 8.061002178649239e-06,
|
| 29 |
+
"loss": 0.5396,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.06540222367560497,
|
| 34 |
+
"grad_norm": 4.589162826538086,
|
| 35 |
+
"learning_rate": 1.0784313725490197e-05,
|
| 36 |
+
"loss": 0.4676,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.08175277959450622,
|
| 41 |
+
"grad_norm": 5.849193096160889,
|
| 42 |
+
"learning_rate": 1.3507625272331156e-05,
|
| 43 |
+
"loss": 0.3214,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.09810333551340746,
|
| 48 |
+
"grad_norm": 16.743791580200195,
|
| 49 |
+
"learning_rate": 1.6230936819172112e-05,
|
| 50 |
+
"loss": 0.3655,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.1144538914323087,
|
| 55 |
+
"grad_norm": 4.951746463775635,
|
| 56 |
+
"learning_rate": 1.895424836601307e-05,
|
| 57 |
+
"loss": 0.282,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.13080444735120994,
|
| 62 |
+
"grad_norm": 2.425187349319458,
|
| 63 |
+
"learning_rate": 2.1677559912854033e-05,
|
| 64 |
+
"loss": 0.2814,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.1471550032701112,
|
| 69 |
+
"grad_norm": 1.2229609489440918,
|
| 70 |
+
"learning_rate": 2.4400871459694992e-05,
|
| 71 |
+
"loss": 0.1735,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.16350555918901244,
|
| 76 |
+
"grad_norm": 6.4670515060424805,
|
| 77 |
+
"learning_rate": 2.7124183006535947e-05,
|
| 78 |
+
"loss": 0.3867,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.17985611510791366,
|
| 83 |
+
"grad_norm": 8.571281433105469,
|
| 84 |
+
"learning_rate": 2.984749455337691e-05,
|
| 85 |
+
"loss": 0.2893,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.1962066710268149,
|
| 90 |
+
"grad_norm": 10.381059646606445,
|
| 91 |
+
"learning_rate": 3.257080610021787e-05,
|
| 92 |
+
"loss": 0.2292,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.21255722694571616,
|
| 97 |
+
"grad_norm": 1.9075127840042114,
|
| 98 |
+
"learning_rate": 3.529411764705883e-05,
|
| 99 |
+
"loss": 0.3212,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2289077828646174,
|
| 104 |
+
"grad_norm": 0.7908616065979004,
|
| 105 |
+
"learning_rate": 3.8017429193899786e-05,
|
| 106 |
+
"loss": 0.4011,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.24525833878351863,
|
| 111 |
+
"grad_norm": 11.509167671203613,
|
| 112 |
+
"learning_rate": 4.074074074074074e-05,
|
| 113 |
+
"loss": 0.3098,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2616088947024199,
|
| 118 |
+
"grad_norm": 18.186838150024414,
|
| 119 |
+
"learning_rate": 4.3464052287581704e-05,
|
| 120 |
+
"loss": 0.1825,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.2779594506213211,
|
| 125 |
+
"grad_norm": 4.45954704284668,
|
| 126 |
+
"learning_rate": 4.6187363834422656e-05,
|
| 127 |
+
"loss": 0.4669,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.2943100065402224,
|
| 132 |
+
"grad_norm": 5.498913764953613,
|
| 133 |
+
"learning_rate": 4.891067538126362e-05,
|
| 134 |
+
"loss": 0.2676,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3106605624591236,
|
| 139 |
+
"grad_norm": 7.379275798797607,
|
| 140 |
+
"learning_rate": 4.981831395348838e-05,
|
| 141 |
+
"loss": 0.2986,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.3270111183780249,
|
| 146 |
+
"grad_norm": 4.307358741760254,
|
| 147 |
+
"learning_rate": 4.9515503875968994e-05,
|
| 148 |
+
"loss": 0.2586,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.3433616742969261,
|
| 153 |
+
"grad_norm": 10.809741973876953,
|
| 154 |
+
"learning_rate": 4.9212693798449616e-05,
|
| 155 |
+
"loss": 0.2444,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.3597122302158273,
|
| 160 |
+
"grad_norm": 4.582003593444824,
|
| 161 |
+
"learning_rate": 4.890988372093023e-05,
|
| 162 |
+
"loss": 0.4484,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.3760627861347286,
|
| 167 |
+
"grad_norm": 2.763364315032959,
|
| 168 |
+
"learning_rate": 4.860707364341086e-05,
|
| 169 |
+
"loss": 0.3377,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.3924133420536298,
|
| 174 |
+
"grad_norm": 11.982911109924316,
|
| 175 |
+
"learning_rate": 4.830426356589148e-05,
|
| 176 |
+
"loss": 0.2219,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.40876389797253104,
|
| 181 |
+
"grad_norm": 14.225927352905273,
|
| 182 |
+
"learning_rate": 4.8001453488372095e-05,
|
| 183 |
+
"loss": 0.2669,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.4251144538914323,
|
| 188 |
+
"grad_norm": 16.698148727416992,
|
| 189 |
+
"learning_rate": 4.7698643410852716e-05,
|
| 190 |
+
"loss": 0.3172,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.44146500981033354,
|
| 195 |
+
"grad_norm": 11.465031623840332,
|
| 196 |
+
"learning_rate": 4.739583333333333e-05,
|
| 197 |
+
"loss": 0.3953,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.4578155657292348,
|
| 202 |
+
"grad_norm": 2.0661489963531494,
|
| 203 |
+
"learning_rate": 4.709302325581396e-05,
|
| 204 |
+
"loss": 0.3295,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.47416612164813604,
|
| 209 |
+
"grad_norm": 4.683841228485107,
|
| 210 |
+
"learning_rate": 4.679021317829458e-05,
|
| 211 |
+
"loss": 0.2518,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.49051667756703726,
|
| 216 |
+
"grad_norm": 0.2995409369468689,
|
| 217 |
+
"learning_rate": 4.6487403100775196e-05,
|
| 218 |
+
"loss": 0.2238,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.5068672334859385,
|
| 223 |
+
"grad_norm": 8.76974105834961,
|
| 224 |
+
"learning_rate": 4.618459302325582e-05,
|
| 225 |
+
"loss": 0.4002,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5232177894048398,
|
| 230 |
+
"grad_norm": 16.188955307006836,
|
| 231 |
+
"learning_rate": 4.588178294573643e-05,
|
| 232 |
+
"loss": 0.2211,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.539568345323741,
|
| 237 |
+
"grad_norm": 1.0120298862457275,
|
| 238 |
+
"learning_rate": 4.557897286821706e-05,
|
| 239 |
+
"loss": 0.3034,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5559189012426422,
|
| 244 |
+
"grad_norm": 3.9240946769714355,
|
| 245 |
+
"learning_rate": 4.527616279069768e-05,
|
| 246 |
+
"loss": 0.3112,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5722694571615435,
|
| 251 |
+
"grad_norm": 0.4812540113925934,
|
| 252 |
+
"learning_rate": 4.4973352713178296e-05,
|
| 253 |
+
"loss": 0.1609,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5886200130804448,
|
| 258 |
+
"grad_norm": 3.5718393325805664,
|
| 259 |
+
"learning_rate": 4.467054263565892e-05,
|
| 260 |
+
"loss": 0.3033,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.604970568999346,
|
| 265 |
+
"grad_norm": 16.025508880615234,
|
| 266 |
+
"learning_rate": 4.436773255813953e-05,
|
| 267 |
+
"loss": 0.3427,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.6213211249182472,
|
| 272 |
+
"grad_norm": 6.246650218963623,
|
| 273 |
+
"learning_rate": 4.406492248062016e-05,
|
| 274 |
+
"loss": 0.1998,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6376716808371484,
|
| 279 |
+
"grad_norm": 0.6095386743545532,
|
| 280 |
+
"learning_rate": 4.3762112403100776e-05,
|
| 281 |
+
"loss": 0.2766,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.6540222367560498,
|
| 286 |
+
"grad_norm": 2.4750990867614746,
|
| 287 |
+
"learning_rate": 4.34593023255814e-05,
|
| 288 |
+
"loss": 0.1986,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.670372792674951,
|
| 293 |
+
"grad_norm": 4.894515514373779,
|
| 294 |
+
"learning_rate": 4.315649224806202e-05,
|
| 295 |
+
"loss": 0.3955,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.6867233485938522,
|
| 300 |
+
"grad_norm": 0.23163600265979767,
|
| 301 |
+
"learning_rate": 4.2853682170542634e-05,
|
| 302 |
+
"loss": 0.205,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.7030739045127534,
|
| 307 |
+
"grad_norm": 2.4058098793029785,
|
| 308 |
+
"learning_rate": 4.255087209302326e-05,
|
| 309 |
+
"loss": 0.3028,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.7194244604316546,
|
| 314 |
+
"grad_norm": 8.029851913452148,
|
| 315 |
+
"learning_rate": 4.2248062015503877e-05,
|
| 316 |
+
"loss": 0.1737,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.7357750163505559,
|
| 321 |
+
"grad_norm": 20.8834228515625,
|
| 322 |
+
"learning_rate": 4.19452519379845e-05,
|
| 323 |
+
"loss": 0.2956,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.7521255722694572,
|
| 328 |
+
"grad_norm": 3.3043746948242188,
|
| 329 |
+
"learning_rate": 4.164244186046512e-05,
|
| 330 |
+
"loss": 0.2918,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.7684761281883584,
|
| 335 |
+
"grad_norm": 3.6136388778686523,
|
| 336 |
+
"learning_rate": 4.1339631782945734e-05,
|
| 337 |
+
"loss": 0.2564,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.7848266841072596,
|
| 342 |
+
"grad_norm": 4.633103370666504,
|
| 343 |
+
"learning_rate": 4.103682170542636e-05,
|
| 344 |
+
"loss": 0.158,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.8011772400261609,
|
| 349 |
+
"grad_norm": 6.264617443084717,
|
| 350 |
+
"learning_rate": 4.073401162790698e-05,
|
| 351 |
+
"loss": 0.3824,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.8175277959450621,
|
| 356 |
+
"grad_norm": 4.16187047958374,
|
| 357 |
+
"learning_rate": 4.04312015503876e-05,
|
| 358 |
+
"loss": 0.1884,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.8338783518639634,
|
| 363 |
+
"grad_norm": 7.170223236083984,
|
| 364 |
+
"learning_rate": 4.012839147286822e-05,
|
| 365 |
+
"loss": 0.2477,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.8502289077828646,
|
| 370 |
+
"grad_norm": 9.301641464233398,
|
| 371 |
+
"learning_rate": 3.9825581395348835e-05,
|
| 372 |
+
"loss": 0.2557,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.8665794637017659,
|
| 377 |
+
"grad_norm": 0.13785265386104584,
|
| 378 |
+
"learning_rate": 3.9522771317829463e-05,
|
| 379 |
+
"loss": 0.2382,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.8829300196206671,
|
| 384 |
+
"grad_norm": 8.949559211730957,
|
| 385 |
+
"learning_rate": 3.921996124031008e-05,
|
| 386 |
+
"loss": 0.3123,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.8992805755395683,
|
| 391 |
+
"grad_norm": 5.080151557922363,
|
| 392 |
+
"learning_rate": 3.89171511627907e-05,
|
| 393 |
+
"loss": 0.1705,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.9156311314584696,
|
| 398 |
+
"grad_norm": 4.757380485534668,
|
| 399 |
+
"learning_rate": 3.861434108527132e-05,
|
| 400 |
+
"loss": 0.121,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.9319816873773709,
|
| 405 |
+
"grad_norm": 8.166597366333008,
|
| 406 |
+
"learning_rate": 3.8311531007751936e-05,
|
| 407 |
+
"loss": 0.2547,
|
| 408 |
+
"step": 1425
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.9483322432962721,
|
| 412 |
+
"grad_norm": 0.18423116207122803,
|
| 413 |
+
"learning_rate": 3.8008720930232564e-05,
|
| 414 |
+
"loss": 0.1343,
|
| 415 |
+
"step": 1450
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.9646827992151733,
|
| 419 |
+
"grad_norm": 9.6463623046875,
|
| 420 |
+
"learning_rate": 3.770591085271318e-05,
|
| 421 |
+
"loss": 0.2026,
|
| 422 |
+
"step": 1475
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.9810333551340745,
|
| 426 |
+
"grad_norm": 6.031147003173828,
|
| 427 |
+
"learning_rate": 3.74031007751938e-05,
|
| 428 |
+
"loss": 0.2007,
|
| 429 |
+
"step": 1500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.9973839110529757,
|
| 433 |
+
"grad_norm": 15.847230911254883,
|
| 434 |
+
"learning_rate": 3.710029069767442e-05,
|
| 435 |
+
"loss": 0.287,
|
| 436 |
+
"step": 1525
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 1.0,
|
| 440 |
+
"eval_accuracy": 0.9502145922746781,
|
| 441 |
+
"eval_auc": 0.9945156335961857,
|
| 442 |
+
"eval_f1": 0.965352449223417,
|
| 443 |
+
"eval_loss": 0.15939411520957947,
|
| 444 |
+
"eval_precision": 0.993849938499385,
|
| 445 |
+
"eval_recall": 0.9384436701509872,
|
| 446 |
+
"eval_runtime": 5.7507,
|
| 447 |
+
"eval_samples_per_second": 202.584,
|
| 448 |
+
"eval_steps_per_second": 12.694,
|
| 449 |
+
"step": 1529
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.013734466971877,
|
| 453 |
+
"grad_norm": 5.32667875289917,
|
| 454 |
+
"learning_rate": 3.679748062015504e-05,
|
| 455 |
+
"loss": 0.2212,
|
| 456 |
+
"step": 1550
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0300850228907783,
|
| 460 |
+
"grad_norm": 4.328026294708252,
|
| 461 |
+
"learning_rate": 3.6494670542635665e-05,
|
| 462 |
+
"loss": 0.1949,
|
| 463 |
+
"step": 1575
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.0464355788096795,
|
| 467 |
+
"grad_norm": 8.589631080627441,
|
| 468 |
+
"learning_rate": 3.619186046511628e-05,
|
| 469 |
+
"loss": 0.1544,
|
| 470 |
+
"step": 1600
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.0627861347285807,
|
| 474 |
+
"grad_norm": 1.7674341201782227,
|
| 475 |
+
"learning_rate": 3.58890503875969e-05,
|
| 476 |
+
"loss": 0.103,
|
| 477 |
+
"step": 1625
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.079136690647482,
|
| 481 |
+
"grad_norm": 6.062548637390137,
|
| 482 |
+
"learning_rate": 3.558624031007752e-05,
|
| 483 |
+
"loss": 0.2805,
|
| 484 |
+
"step": 1650
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 1.0954872465663832,
|
| 488 |
+
"grad_norm": 8.54261302947998,
|
| 489 |
+
"learning_rate": 3.528343023255814e-05,
|
| 490 |
+
"loss": 0.1475,
|
| 491 |
+
"step": 1675
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 1.1118378024852844,
|
| 495 |
+
"grad_norm": 6.659710884094238,
|
| 496 |
+
"learning_rate": 3.4980620155038766e-05,
|
| 497 |
+
"loss": 0.2388,
|
| 498 |
+
"step": 1700
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 1.1281883584041856,
|
| 502 |
+
"grad_norm": 4.354561805725098,
|
| 503 |
+
"learning_rate": 3.467781007751938e-05,
|
| 504 |
+
"loss": 0.1795,
|
| 505 |
+
"step": 1725
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 1.144538914323087,
|
| 509 |
+
"grad_norm": 0.20215673744678497,
|
| 510 |
+
"learning_rate": 3.4375e-05,
|
| 511 |
+
"loss": 0.1543,
|
| 512 |
+
"step": 1750
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 1.1608894702419883,
|
| 516 |
+
"grad_norm": 0.9856411218643188,
|
| 517 |
+
"learning_rate": 3.4072189922480624e-05,
|
| 518 |
+
"loss": 0.2781,
|
| 519 |
+
"step": 1775
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 1.1772400261608895,
|
| 523 |
+
"grad_norm": 0.7960185408592224,
|
| 524 |
+
"learning_rate": 3.376937984496124e-05,
|
| 525 |
+
"loss": 0.2001,
|
| 526 |
+
"step": 1800
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 1.1935905820797907,
|
| 530 |
+
"grad_norm": 4.385952949523926,
|
| 531 |
+
"learning_rate": 3.346656976744187e-05,
|
| 532 |
+
"loss": 0.262,
|
| 533 |
+
"step": 1825
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 1.209941137998692,
|
| 537 |
+
"grad_norm": 5.5558576583862305,
|
| 538 |
+
"learning_rate": 3.316375968992248e-05,
|
| 539 |
+
"loss": 0.1862,
|
| 540 |
+
"step": 1850
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 1.2262916939175932,
|
| 544 |
+
"grad_norm": 8.943402290344238,
|
| 545 |
+
"learning_rate": 3.28609496124031e-05,
|
| 546 |
+
"loss": 0.1657,
|
| 547 |
+
"step": 1875
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 1.2426422498364944,
|
| 551 |
+
"grad_norm": 0.3145776689052582,
|
| 552 |
+
"learning_rate": 3.2558139534883724e-05,
|
| 553 |
+
"loss": 0.213,
|
| 554 |
+
"step": 1900
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 1.2589928057553956,
|
| 558 |
+
"grad_norm": 0.6332132816314697,
|
| 559 |
+
"learning_rate": 3.225532945736434e-05,
|
| 560 |
+
"loss": 0.1174,
|
| 561 |
+
"step": 1925
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 1.2753433616742968,
|
| 565 |
+
"grad_norm": 3.167943239212036,
|
| 566 |
+
"learning_rate": 3.195251937984496e-05,
|
| 567 |
+
"loss": 0.1622,
|
| 568 |
+
"step": 1950
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 1.2916939175931983,
|
| 572 |
+
"grad_norm": 0.24524331092834473,
|
| 573 |
+
"learning_rate": 3.164970930232558e-05,
|
| 574 |
+
"loss": 0.2144,
|
| 575 |
+
"step": 1975
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 1.3080444735120995,
|
| 579 |
+
"grad_norm": 5.5770263671875,
|
| 580 |
+
"learning_rate": 3.1346899224806204e-05,
|
| 581 |
+
"loss": 0.1812,
|
| 582 |
+
"step": 2000
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.3243950294310007,
|
| 586 |
+
"grad_norm": 0.6289435029029846,
|
| 587 |
+
"learning_rate": 3.1044089147286825e-05,
|
| 588 |
+
"loss": 0.1762,
|
| 589 |
+
"step": 2025
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.340745585349902,
|
| 593 |
+
"grad_norm": 6.443571090698242,
|
| 594 |
+
"learning_rate": 3.074127906976744e-05,
|
| 595 |
+
"loss": 0.1759,
|
| 596 |
+
"step": 2050
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 1.3570961412688032,
|
| 600 |
+
"grad_norm": 6.263022422790527,
|
| 601 |
+
"learning_rate": 3.0450581395348838e-05,
|
| 602 |
+
"loss": 0.411,
|
| 603 |
+
"step": 2075
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 1.3734466971877044,
|
| 607 |
+
"grad_norm": 4.602290630340576,
|
| 608 |
+
"learning_rate": 3.014777131782946e-05,
|
| 609 |
+
"loss": 0.1595,
|
| 610 |
+
"step": 2100
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 1.3897972531066056,
|
| 614 |
+
"grad_norm": 13.215031623840332,
|
| 615 |
+
"learning_rate": 2.9844961240310077e-05,
|
| 616 |
+
"loss": 0.2196,
|
| 617 |
+
"step": 2125
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 1.4061478090255068,
|
| 621 |
+
"grad_norm": 0.35713547468185425,
|
| 622 |
+
"learning_rate": 2.95421511627907e-05,
|
| 623 |
+
"loss": 0.1646,
|
| 624 |
+
"step": 2150
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 1.422498364944408,
|
| 628 |
+
"grad_norm": 10.571084022521973,
|
| 629 |
+
"learning_rate": 2.923934108527132e-05,
|
| 630 |
+
"loss": 0.2415,
|
| 631 |
+
"step": 2175
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 1.4388489208633093,
|
| 635 |
+
"grad_norm": 0.12735210359096527,
|
| 636 |
+
"learning_rate": 2.893653100775194e-05,
|
| 637 |
+
"loss": 0.1596,
|
| 638 |
+
"step": 2200
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 1.4551994767822105,
|
| 642 |
+
"grad_norm": 0.8274198174476624,
|
| 643 |
+
"learning_rate": 2.863372093023256e-05,
|
| 644 |
+
"loss": 0.2063,
|
| 645 |
+
"step": 2225
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 1.4715500327011117,
|
| 649 |
+
"grad_norm": 0.18884733319282532,
|
| 650 |
+
"learning_rate": 2.8330910852713178e-05,
|
| 651 |
+
"loss": 0.2319,
|
| 652 |
+
"step": 2250
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"epoch": 1.487900588620013,
|
| 656 |
+
"grad_norm": 7.244335651397705,
|
| 657 |
+
"learning_rate": 2.80281007751938e-05,
|
| 658 |
+
"loss": 0.171,
|
| 659 |
+
"step": 2275
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 1.5042511445389142,
|
| 663 |
+
"grad_norm": 1.030834436416626,
|
| 664 |
+
"learning_rate": 2.772529069767442e-05,
|
| 665 |
+
"loss": 0.1566,
|
| 666 |
+
"step": 2300
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 1.5206017004578156,
|
| 670 |
+
"grad_norm": 0.2723897099494934,
|
| 671 |
+
"learning_rate": 2.742248062015504e-05,
|
| 672 |
+
"loss": 0.2864,
|
| 673 |
+
"step": 2325
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 1.5369522563767168,
|
| 677 |
+
"grad_norm": 0.3210669457912445,
|
| 678 |
+
"learning_rate": 2.711967054263566e-05,
|
| 679 |
+
"loss": 0.1941,
|
| 680 |
+
"step": 2350
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 1.553302812295618,
|
| 684 |
+
"grad_norm": 5.555069446563721,
|
| 685 |
+
"learning_rate": 2.681686046511628e-05,
|
| 686 |
+
"loss": 0.2085,
|
| 687 |
+
"step": 2375
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 1.5696533682145193,
|
| 691 |
+
"grad_norm": 8.625024795532227,
|
| 692 |
+
"learning_rate": 2.65140503875969e-05,
|
| 693 |
+
"loss": 0.1789,
|
| 694 |
+
"step": 2400
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.5860039241334205,
|
| 698 |
+
"grad_norm": 7.455594539642334,
|
| 699 |
+
"learning_rate": 2.6211240310077522e-05,
|
| 700 |
+
"loss": 0.2819,
|
| 701 |
+
"step": 2425
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.6023544800523217,
|
| 705 |
+
"grad_norm": 2.1804699897766113,
|
| 706 |
+
"learning_rate": 2.590843023255814e-05,
|
| 707 |
+
"loss": 0.2866,
|
| 708 |
+
"step": 2450
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.6187050359712232,
|
| 712 |
+
"grad_norm": 10.467452049255371,
|
| 713 |
+
"learning_rate": 2.560562015503876e-05,
|
| 714 |
+
"loss": 0.2576,
|
| 715 |
+
"step": 2475
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6350555918901244,
|
| 719 |
+
"grad_norm": 4.638566493988037,
|
| 720 |
+
"learning_rate": 2.530281007751938e-05,
|
| 721 |
+
"loss": 0.275,
|
| 722 |
+
"step": 2500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6514061478090256,
|
| 726 |
+
"grad_norm": 6.57110595703125,
|
| 727 |
+
"learning_rate": 2.5e-05,
|
| 728 |
+
"loss": 0.1353,
|
| 729 |
+
"step": 2525
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.6677567037279268,
|
| 733 |
+
"grad_norm": 0.1900484710931778,
|
| 734 |
+
"learning_rate": 2.469718992248062e-05,
|
| 735 |
+
"loss": 0.2028,
|
| 736 |
+
"step": 2550
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.684107259646828,
|
| 740 |
+
"grad_norm": 0.9844114184379578,
|
| 741 |
+
"learning_rate": 2.439437984496124e-05,
|
| 742 |
+
"loss": 0.1779,
|
| 743 |
+
"step": 2575
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.7004578155657293,
|
| 747 |
+
"grad_norm": 3.4427802562713623,
|
| 748 |
+
"learning_rate": 2.4091569767441862e-05,
|
| 749 |
+
"loss": 0.2193,
|
| 750 |
+
"step": 2600
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.7168083714846305,
|
| 754 |
+
"grad_norm": 0.2224160134792328,
|
| 755 |
+
"learning_rate": 2.3788759689922484e-05,
|
| 756 |
+
"loss": 0.2097,
|
| 757 |
+
"step": 2625
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 1.7331589274035317,
|
| 761 |
+
"grad_norm": 5.424374580383301,
|
| 762 |
+
"learning_rate": 2.3485949612403102e-05,
|
| 763 |
+
"loss": 0.1106,
|
| 764 |
+
"step": 2650
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 1.749509483322433,
|
| 768 |
+
"grad_norm": 0.08824160695075989,
|
| 769 |
+
"learning_rate": 2.318313953488372e-05,
|
| 770 |
+
"loss": 0.1152,
|
| 771 |
+
"step": 2675
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 1.7658600392413342,
|
| 775 |
+
"grad_norm": 13.905054092407227,
|
| 776 |
+
"learning_rate": 2.288032945736434e-05,
|
| 777 |
+
"loss": 0.1858,
|
| 778 |
+
"step": 2700
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.7822105951602354,
|
| 782 |
+
"grad_norm": 1.2964977025985718,
|
| 783 |
+
"learning_rate": 2.2577519379844963e-05,
|
| 784 |
+
"loss": 0.1714,
|
| 785 |
+
"step": 2725
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 1.7985611510791366,
|
| 789 |
+
"grad_norm": 40.13843536376953,
|
| 790 |
+
"learning_rate": 2.2274709302325585e-05,
|
| 791 |
+
"loss": 0.0823,
|
| 792 |
+
"step": 2750
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 1.8149117069980378,
|
| 796 |
+
"grad_norm": 6.79445219039917,
|
| 797 |
+
"learning_rate": 2.1971899224806203e-05,
|
| 798 |
+
"loss": 0.0833,
|
| 799 |
+
"step": 2775
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 1.831262262916939,
|
| 803 |
+
"grad_norm": 5.254680156707764,
|
| 804 |
+
"learning_rate": 2.166908914728682e-05,
|
| 805 |
+
"loss": 0.126,
|
| 806 |
+
"step": 2800
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 1.8476128188358403,
|
| 810 |
+
"grad_norm": 0.21236732602119446,
|
| 811 |
+
"learning_rate": 2.1366279069767442e-05,
|
| 812 |
+
"loss": 0.1087,
|
| 813 |
+
"step": 2825
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 1.8639633747547415,
|
| 817 |
+
"grad_norm": 1.6276271343231201,
|
| 818 |
+
"learning_rate": 2.1063468992248064e-05,
|
| 819 |
+
"loss": 0.0624,
|
| 820 |
+
"step": 2850
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 1.880313930673643,
|
| 824 |
+
"grad_norm": 0.12902870774269104,
|
| 825 |
+
"learning_rate": 2.0760658914728685e-05,
|
| 826 |
+
"loss": 0.2461,
|
| 827 |
+
"step": 2875
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 1.8966644865925442,
|
| 831 |
+
"grad_norm": 0.7700151801109314,
|
| 832 |
+
"learning_rate": 2.0457848837209304e-05,
|
| 833 |
+
"loss": 0.1109,
|
| 834 |
+
"step": 2900
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 1.9130150425114454,
|
| 838 |
+
"grad_norm": 9.981553077697754,
|
| 839 |
+
"learning_rate": 2.0155038759689922e-05,
|
| 840 |
+
"loss": 0.1031,
|
| 841 |
+
"step": 2925
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 1.9293655984303466,
|
| 845 |
+
"grad_norm": 7.399847507476807,
|
| 846 |
+
"learning_rate": 1.9852228682170543e-05,
|
| 847 |
+
"loss": 0.2301,
|
| 848 |
+
"step": 2950
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 1.9457161543492478,
|
| 852 |
+
"grad_norm": 10.187629699707031,
|
| 853 |
+
"learning_rate": 1.954941860465116e-05,
|
| 854 |
+
"loss": 0.2124,
|
| 855 |
+
"step": 2975
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 1.9620667102681493,
|
| 859 |
+
"grad_norm": 6.082677841186523,
|
| 860 |
+
"learning_rate": 1.9246608527131786e-05,
|
| 861 |
+
"loss": 0.248,
|
| 862 |
+
"step": 3000
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 1.9784172661870505,
|
| 866 |
+
"grad_norm": 11.070582389831543,
|
| 867 |
+
"learning_rate": 1.8943798449612404e-05,
|
| 868 |
+
"loss": 0.11,
|
| 869 |
+
"step": 3025
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 1.9947678221059517,
|
| 873 |
+
"grad_norm": 1.658309817314148,
|
| 874 |
+
"learning_rate": 1.8640988372093023e-05,
|
| 875 |
+
"loss": 0.2387,
|
| 876 |
+
"step": 3050
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 2.0,
|
| 880 |
+
"eval_accuracy": 0.9699570815450643,
|
| 881 |
+
"eval_auc": 0.9967430007946696,
|
| 882 |
+
"eval_f1": 0.9799196787148594,
|
| 883 |
+
"eval_loss": 0.08820199966430664,
|
| 884 |
+
"eval_precision": 0.9682539682539683,
|
| 885 |
+
"eval_recall": 0.991869918699187,
|
| 886 |
+
"eval_runtime": 5.7323,
|
| 887 |
+
"eval_samples_per_second": 203.234,
|
| 888 |
+
"eval_steps_per_second": 12.735,
|
| 889 |
+
"step": 3058
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 2.011118378024853,
|
| 893 |
+
"grad_norm": 0.42580923438072205,
|
| 894 |
+
"learning_rate": 1.8338178294573644e-05,
|
| 895 |
+
"loss": 0.1586,
|
| 896 |
+
"step": 3075
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 2.027468933943754,
|
| 900 |
+
"grad_norm": 12.402567863464355,
|
| 901 |
+
"learning_rate": 1.8035368217054262e-05,
|
| 902 |
+
"loss": 0.1163,
|
| 903 |
+
"step": 3100
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 2.0438194898626554,
|
| 907 |
+
"grad_norm": 9.509015083312988,
|
| 908 |
+
"learning_rate": 1.7732558139534887e-05,
|
| 909 |
+
"loss": 0.1893,
|
| 910 |
+
"step": 3125
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 2.0601700457815566,
|
| 914 |
+
"grad_norm": 0.024427318945527077,
|
| 915 |
+
"learning_rate": 1.7429748062015505e-05,
|
| 916 |
+
"loss": 0.1121,
|
| 917 |
+
"step": 3150
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 2.076520601700458,
|
| 921 |
+
"grad_norm": 3.6162619590759277,
|
| 922 |
+
"learning_rate": 1.7126937984496123e-05,
|
| 923 |
+
"loss": 0.1592,
|
| 924 |
+
"step": 3175
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 2.092871157619359,
|
| 928 |
+
"grad_norm": 0.3668271601200104,
|
| 929 |
+
"learning_rate": 1.6824127906976745e-05,
|
| 930 |
+
"loss": 0.1724,
|
| 931 |
+
"step": 3200
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 2.1092217135382603,
|
| 935 |
+
"grad_norm": 0.10604169964790344,
|
| 936 |
+
"learning_rate": 1.6521317829457363e-05,
|
| 937 |
+
"loss": 0.1133,
|
| 938 |
+
"step": 3225
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 2.1255722694571615,
|
| 942 |
+
"grad_norm": 5.482110977172852,
|
| 943 |
+
"learning_rate": 1.6218507751937988e-05,
|
| 944 |
+
"loss": 0.2885,
|
| 945 |
+
"step": 3250
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 2.1419228253760627,
|
| 949 |
+
"grad_norm": 2.177529811859131,
|
| 950 |
+
"learning_rate": 1.5915697674418606e-05,
|
| 951 |
+
"loss": 0.1656,
|
| 952 |
+
"step": 3275
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 2.158273381294964,
|
| 956 |
+
"grad_norm": 0.7062684893608093,
|
| 957 |
+
"learning_rate": 1.5612887596899224e-05,
|
| 958 |
+
"loss": 0.1784,
|
| 959 |
+
"step": 3300
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 2.174623937213865,
|
| 963 |
+
"grad_norm": 0.6271941065788269,
|
| 964 |
+
"learning_rate": 1.5310077519379846e-05,
|
| 965 |
+
"loss": 0.0416,
|
| 966 |
+
"step": 3325
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 2.1909744931327664,
|
| 970 |
+
"grad_norm": 0.13655735552310944,
|
| 971 |
+
"learning_rate": 1.5007267441860465e-05,
|
| 972 |
+
"loss": 0.1002,
|
| 973 |
+
"step": 3350
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 2.2073250490516676,
|
| 977 |
+
"grad_norm": 8.668305397033691,
|
| 978 |
+
"learning_rate": 1.4704457364341087e-05,
|
| 979 |
+
"loss": 0.1661,
|
| 980 |
+
"step": 3375
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 2.223675604970569,
|
| 984 |
+
"grad_norm": 5.533019065856934,
|
| 985 |
+
"learning_rate": 1.4401647286821707e-05,
|
| 986 |
+
"loss": 0.3131,
|
| 987 |
+
"step": 3400
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 2.24002616088947,
|
| 991 |
+
"grad_norm": 9.50761890411377,
|
| 992 |
+
"learning_rate": 1.4098837209302327e-05,
|
| 993 |
+
"loss": 0.155,
|
| 994 |
+
"step": 3425
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 2.2563767168083713,
|
| 998 |
+
"grad_norm": 14.682671546936035,
|
| 999 |
+
"learning_rate": 1.3796027131782946e-05,
|
| 1000 |
+
"loss": 0.1707,
|
| 1001 |
+
"step": 3450
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 2.2727272727272725,
|
| 1005 |
+
"grad_norm": 14.890127182006836,
|
| 1006 |
+
"learning_rate": 1.3493217054263566e-05,
|
| 1007 |
+
"loss": 0.1336,
|
| 1008 |
+
"step": 3475
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 2.289077828646174,
|
| 1012 |
+
"grad_norm": 0.08246014267206192,
|
| 1013 |
+
"learning_rate": 1.3190406976744188e-05,
|
| 1014 |
+
"loss": 0.1234,
|
| 1015 |
+
"step": 3500
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 2.3054283845650754,
|
| 1019 |
+
"grad_norm": 0.00616547791287303,
|
| 1020 |
+
"learning_rate": 1.2887596899224808e-05,
|
| 1021 |
+
"loss": 0.1608,
|
| 1022 |
+
"step": 3525
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 2.3217789404839766,
|
| 1026 |
+
"grad_norm": 6.7408576011657715,
|
| 1027 |
+
"learning_rate": 1.2584786821705427e-05,
|
| 1028 |
+
"loss": 0.1328,
|
| 1029 |
+
"step": 3550
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 2.338129496402878,
|
| 1033 |
+
"grad_norm": 13.762513160705566,
|
| 1034 |
+
"learning_rate": 1.2281976744186047e-05,
|
| 1035 |
+
"loss": 0.0673,
|
| 1036 |
+
"step": 3575
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 2.354480052321779,
|
| 1040 |
+
"grad_norm": 0.3772618770599365,
|
| 1041 |
+
"learning_rate": 1.1979166666666667e-05,
|
| 1042 |
+
"loss": 0.1469,
|
| 1043 |
+
"step": 3600
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 2.3708306082406803,
|
| 1047 |
+
"grad_norm": 0.21671177446842194,
|
| 1048 |
+
"learning_rate": 1.1676356589147287e-05,
|
| 1049 |
+
"loss": 0.1849,
|
| 1050 |
+
"step": 3625
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 2.3871811641595815,
|
| 1054 |
+
"grad_norm": 4.807242393493652,
|
| 1055 |
+
"learning_rate": 1.1373546511627907e-05,
|
| 1056 |
+
"loss": 0.202,
|
| 1057 |
+
"step": 3650
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 2.4035317200784827,
|
| 1061 |
+
"grad_norm": 2.1541972160339355,
|
| 1062 |
+
"learning_rate": 1.1070736434108528e-05,
|
| 1063 |
+
"loss": 0.2248,
|
| 1064 |
+
"step": 3675
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 2.419882275997384,
|
| 1068 |
+
"grad_norm": 0.06315416097640991,
|
| 1069 |
+
"learning_rate": 1.0767926356589148e-05,
|
| 1070 |
+
"loss": 0.1071,
|
| 1071 |
+
"step": 3700
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 2.436232831916285,
|
| 1075 |
+
"grad_norm": 0.3411709666252136,
|
| 1076 |
+
"learning_rate": 1.0465116279069768e-05,
|
| 1077 |
+
"loss": 0.073,
|
| 1078 |
+
"step": 3725
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 2.4525833878351864,
|
| 1082 |
+
"grad_norm": 0.4479942321777344,
|
| 1083 |
+
"learning_rate": 1.0162306201550388e-05,
|
| 1084 |
+
"loss": 0.1289,
|
| 1085 |
+
"step": 3750
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 2.4689339437540876,
|
| 1089 |
+
"grad_norm": 3.079694986343384,
|
| 1090 |
+
"learning_rate": 9.859496124031007e-06,
|
| 1091 |
+
"loss": 0.1315,
|
| 1092 |
+
"step": 3775
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 2.485284499672989,
|
| 1096 |
+
"grad_norm": 0.02924235165119171,
|
| 1097 |
+
"learning_rate": 9.556686046511629e-06,
|
| 1098 |
+
"loss": 0.152,
|
| 1099 |
+
"step": 3800
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 2.50163505559189,
|
| 1103 |
+
"grad_norm": 0.04543924704194069,
|
| 1104 |
+
"learning_rate": 9.253875968992249e-06,
|
| 1105 |
+
"loss": 0.115,
|
| 1106 |
+
"step": 3825
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 2.5179856115107913,
|
| 1110 |
+
"grad_norm": 13.066108703613281,
|
| 1111 |
+
"learning_rate": 8.951065891472869e-06,
|
| 1112 |
+
"loss": 0.154,
|
| 1113 |
+
"step": 3850
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 2.5343361674296925,
|
| 1117 |
+
"grad_norm": 0.13062244653701782,
|
| 1118 |
+
"learning_rate": 8.648255813953488e-06,
|
| 1119 |
+
"loss": 0.0953,
|
| 1120 |
+
"step": 3875
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 2.5506867233485937,
|
| 1124 |
+
"grad_norm": 14.618424415588379,
|
| 1125 |
+
"learning_rate": 8.345445736434108e-06,
|
| 1126 |
+
"loss": 0.1167,
|
| 1127 |
+
"step": 3900
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 2.5670372792674954,
|
| 1131 |
+
"grad_norm": 4.791504859924316,
|
| 1132 |
+
"learning_rate": 8.04263565891473e-06,
|
| 1133 |
+
"loss": 0.2192,
|
| 1134 |
+
"step": 3925
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 2.5833878351863966,
|
| 1138 |
+
"grad_norm": 0.2636285126209259,
|
| 1139 |
+
"learning_rate": 7.73982558139535e-06,
|
| 1140 |
+
"loss": 0.1261,
|
| 1141 |
+
"step": 3950
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 2.599738391105298,
|
| 1145 |
+
"grad_norm": 5.333641529083252,
|
| 1146 |
+
"learning_rate": 7.4370155038759686e-06,
|
| 1147 |
+
"loss": 0.1271,
|
| 1148 |
+
"step": 3975
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 2.616088947024199,
|
| 1152 |
+
"grad_norm": 5.496675491333008,
|
| 1153 |
+
"learning_rate": 7.13420542635659e-06,
|
| 1154 |
+
"loss": 0.1443,
|
| 1155 |
+
"step": 4000
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 2.6324395029431003,
|
| 1159 |
+
"grad_norm": 0.13467253744602203,
|
| 1160 |
+
"learning_rate": 6.831395348837209e-06,
|
| 1161 |
+
"loss": 0.061,
|
| 1162 |
+
"step": 4025
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 2.6487900588620015,
|
| 1166 |
+
"grad_norm": 0.5136555433273315,
|
| 1167 |
+
"learning_rate": 6.5285852713178306e-06,
|
| 1168 |
+
"loss": 0.255,
|
| 1169 |
+
"step": 4050
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 2.6651406147809027,
|
| 1173 |
+
"grad_norm": 0.19155253469944,
|
| 1174 |
+
"learning_rate": 6.2257751937984495e-06,
|
| 1175 |
+
"loss": 0.1159,
|
| 1176 |
+
"step": 4075
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 2.681491170699804,
|
| 1180 |
+
"grad_norm": 0.2983386814594269,
|
| 1181 |
+
"learning_rate": 5.92296511627907e-06,
|
| 1182 |
+
"loss": 0.1332,
|
| 1183 |
+
"step": 4100
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 2.697841726618705,
|
| 1187 |
+
"grad_norm": 1.9400678873062134,
|
| 1188 |
+
"learning_rate": 5.620155038759691e-06,
|
| 1189 |
+
"loss": 0.1556,
|
| 1190 |
+
"step": 4125
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 2.7141922825376064,
|
| 1194 |
+
"grad_norm": 5.343267440795898,
|
| 1195 |
+
"learning_rate": 5.31734496124031e-06,
|
| 1196 |
+
"loss": 0.1416,
|
| 1197 |
+
"step": 4150
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 2.7305428384565076,
|
| 1201 |
+
"grad_norm": 6.421015739440918,
|
| 1202 |
+
"learning_rate": 5.0145348837209305e-06,
|
| 1203 |
+
"loss": 0.1812,
|
| 1204 |
+
"step": 4175
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"epoch": 2.746893394375409,
|
| 1208 |
+
"grad_norm": 0.15666760504245758,
|
| 1209 |
+
"learning_rate": 4.71172480620155e-06,
|
| 1210 |
+
"loss": 0.1108,
|
| 1211 |
+
"step": 4200
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 2.76324395029431,
|
| 1215 |
+
"grad_norm": 0.2742402255535126,
|
| 1216 |
+
"learning_rate": 4.408914728682171e-06,
|
| 1217 |
+
"loss": 0.1291,
|
| 1218 |
+
"step": 4225
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 2.7795945062132112,
|
| 1222 |
+
"grad_norm": 0.05212033912539482,
|
| 1223 |
+
"learning_rate": 4.106104651162791e-06,
|
| 1224 |
+
"loss": 0.0938,
|
| 1225 |
+
"step": 4250
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 2.7959450621321125,
|
| 1229 |
+
"grad_norm": 0.1697097271680832,
|
| 1230 |
+
"learning_rate": 3.8032945736434107e-06,
|
| 1231 |
+
"loss": 0.0349,
|
| 1232 |
+
"step": 4275
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 2.8122956180510137,
|
| 1236 |
+
"grad_norm": 1.2600692510604858,
|
| 1237 |
+
"learning_rate": 3.5004844961240313e-06,
|
| 1238 |
+
"loss": 0.0619,
|
| 1239 |
+
"step": 4300
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 2.828646173969915,
|
| 1243 |
+
"grad_norm": 0.0772242546081543,
|
| 1244 |
+
"learning_rate": 3.1976744186046516e-06,
|
| 1245 |
+
"loss": 0.107,
|
| 1246 |
+
"step": 4325
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 2.844996729888816,
|
| 1250 |
+
"grad_norm": 0.17465172708034515,
|
| 1251 |
+
"learning_rate": 2.8948643410852714e-06,
|
| 1252 |
+
"loss": 0.0803,
|
| 1253 |
+
"step": 4350
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 2.8613472858077174,
|
| 1257 |
+
"grad_norm": 11.832673072814941,
|
| 1258 |
+
"learning_rate": 2.5920542635658916e-06,
|
| 1259 |
+
"loss": 0.1322,
|
| 1260 |
+
"step": 4375
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 2.8776978417266186,
|
| 1264 |
+
"grad_norm": 10.022361755371094,
|
| 1265 |
+
"learning_rate": 2.2892441860465114e-06,
|
| 1266 |
+
"loss": 0.1041,
|
| 1267 |
+
"step": 4400
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 2.89404839764552,
|
| 1271 |
+
"grad_norm": 5.453954219818115,
|
| 1272 |
+
"learning_rate": 1.986434108527132e-06,
|
| 1273 |
+
"loss": 0.1611,
|
| 1274 |
+
"step": 4425
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 2.910398953564421,
|
| 1278 |
+
"grad_norm": 0.10524914413690567,
|
| 1279 |
+
"learning_rate": 1.683624031007752e-06,
|
| 1280 |
+
"loss": 0.1443,
|
| 1281 |
+
"step": 4450
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 2.9267495094833222,
|
| 1285 |
+
"grad_norm": 8.652328491210938,
|
| 1286 |
+
"learning_rate": 1.3808139534883722e-06,
|
| 1287 |
+
"loss": 0.1896,
|
| 1288 |
+
"step": 4475
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 2.9431000654022235,
|
| 1292 |
+
"grad_norm": 0.2993433177471161,
|
| 1293 |
+
"learning_rate": 1.0780038759689922e-06,
|
| 1294 |
+
"loss": 0.0849,
|
| 1295 |
+
"step": 4500
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 2.9594506213211247,
|
| 1299 |
+
"grad_norm": 0.6885517239570618,
|
| 1300 |
+
"learning_rate": 7.751937984496125e-07,
|
| 1301 |
+
"loss": 0.1334,
|
| 1302 |
+
"step": 4525
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 2.975801177240026,
|
| 1306 |
+
"grad_norm": 7.389411926269531,
|
| 1307 |
+
"learning_rate": 4.7238372093023254e-07,
|
| 1308 |
+
"loss": 0.202,
|
| 1309 |
+
"step": 4550
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 2.992151733158927,
|
| 1313 |
+
"grad_norm": 2.6400258541107178,
|
| 1314 |
+
"learning_rate": 1.6957364341085273e-07,
|
| 1315 |
+
"loss": 0.1566,
|
| 1316 |
+
"step": 4575
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 3.0,
|
| 1320 |
+
"eval_accuracy": 0.976824034334764,
|
| 1321 |
+
"eval_auc": 0.9978146585977138,
|
| 1322 |
+
"eval_f1": 0.9841642228739003,
|
| 1323 |
+
"eval_loss": 0.08583971858024597,
|
| 1324 |
+
"eval_precision": 0.9940758293838863,
|
| 1325 |
+
"eval_recall": 0.9744483159117305,
|
| 1326 |
+
"eval_runtime": 5.747,
|
| 1327 |
+
"eval_samples_per_second": 202.716,
|
| 1328 |
+
"eval_steps_per_second": 12.702,
|
| 1329 |
+
"step": 4587
|
| 1330 |
+
}
|
| 1331 |
+
],
|
| 1332 |
+
"logging_steps": 25,
|
| 1333 |
+
"max_steps": 4587,
|
| 1334 |
+
"num_input_tokens_seen": 0,
|
| 1335 |
+
"num_train_epochs": 3,
|
| 1336 |
+
"save_steps": 500,
|
| 1337 |
+
"stateful_callbacks": {
|
| 1338 |
+
"EarlyStoppingCallback": {
|
| 1339 |
+
"args": {
|
| 1340 |
+
"early_stopping_patience": 5,
|
| 1341 |
+
"early_stopping_threshold": 0.01
|
| 1342 |
+
},
|
| 1343 |
+
"attributes": {
|
| 1344 |
+
"early_stopping_patience_counter": 0
|
| 1345 |
+
}
|
| 1346 |
+
},
|
| 1347 |
+
"TrainerControl": {
|
| 1348 |
+
"args": {
|
| 1349 |
+
"should_epoch_stop": false,
|
| 1350 |
+
"should_evaluate": false,
|
| 1351 |
+
"should_log": false,
|
| 1352 |
+
"should_save": true,
|
| 1353 |
+
"should_training_stop": true
|
| 1354 |
+
},
|
| 1355 |
+
"attributes": {}
|
| 1356 |
+
}
|
| 1357 |
+
},
|
| 1358 |
+
"total_flos": 2.8429486229905367e+18,
|
| 1359 |
+
"train_batch_size": 8,
|
| 1360 |
+
"trial_name": null,
|
| 1361 |
+
"trial_params": null
|
| 1362 |
+
}
|
checkpoint-4587/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b4cf403581691c765ff41ff652af940801d26c6fc86df0eed77a36f82b198d04
|
| 3 |
+
size 5112
|
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": "NORMAL",
|
| 14 |
+
"1": "PNEUMONIA"
|
| 15 |
+
},
|
| 16 |
+
"image_size": 224,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 3072,
|
| 19 |
+
"label2id": {
|
| 20 |
+
"NORMAL": 0,
|
| 21 |
+
"PNEUMONIA": 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": "4.41.2"
|
| 33 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d0d15fb80234d91522f046bab233bf009c7f9fec36b2a1bd1b48d1dbef6aad00
|
| 3 |
+
size 343223968
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_valid_processor_keys": [
|
| 3 |
+
"images",
|
| 4 |
+
"do_resize",
|
| 5 |
+
"size",
|
| 6 |
+
"resample",
|
| 7 |
+
"do_rescale",
|
| 8 |
+
"rescale_factor",
|
| 9 |
+
"do_normalize",
|
| 10 |
+
"image_mean",
|
| 11 |
+
"image_std",
|
| 12 |
+
"return_tensors",
|
| 13 |
+
"data_format",
|
| 14 |
+
"input_data_format"
|
| 15 |
+
],
|
| 16 |
+
"do_normalize": true,
|
| 17 |
+
"do_rescale": true,
|
| 18 |
+
"do_resize": true,
|
| 19 |
+
"image_mean": [
|
| 20 |
+
0.5,
|
| 21 |
+
0.5,
|
| 22 |
+
0.5
|
| 23 |
+
],
|
| 24 |
+
"image_processor_type": "ViTImageProcessor",
|
| 25 |
+
"image_std": [
|
| 26 |
+
0.5,
|
| 27 |
+
0.5,
|
| 28 |
+
0.5
|
| 29 |
+
],
|
| 30 |
+
"resample": 2,
|
| 31 |
+
"rescale_factor": 0.00392156862745098,
|
| 32 |
+
"size": {
|
| 33 |
+
"height": 224,
|
| 34 |
+
"width": 224
|
| 35 |
+
}
|
| 36 |
+
}
|
runs/Jun14_18-14-50_r-luminoussg-autotrain-s6ik9skt-775cb-hpjwc/events.out.tfevents.1718388891.r-luminoussg-autotrain-s6ik9skt-775cb-hpjwc.147.0
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 sha256:53ad7376dd4d8d520e2e522b84bc2a016fc6416b5ec0fe7b9f68645fc527b1ab
|
| 3 |
+
size 45346
|
runs/Jun14_18-14-50_r-luminoussg-autotrain-s6ik9skt-775cb-hpjwc/events.out.tfevents.1718389367.r-luminoussg-autotrain-s6ik9skt-775cb-hpjwc.147.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:26c7f5f1512cb23abecfd68209712116f055490d7067ebc6232949c102f49655
|
| 3 |
+
size 607
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b4cf403581691c765ff41ff652af940801d26c6fc86df0eed77a36f82b198d04
|
| 3 |
+
size 5112
|
training_params.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "xraynewww/autotrain-data",
|
| 3 |
+
"model": "google/vit-base-patch16-224",
|
| 4 |
+
"username": "luminoussg",
|
| 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": "xraynewww",
|
| 19 |
+
"auto_find_batch_size": false,
|
| 20 |
+
"mixed_precision": "fp16",
|
| 21 |
+
"save_total_limit": 1,
|
| 22 |
+
"push_to_hub": true,
|
| 23 |
+
"evaluation_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 |
+
}
|