Upload folder using huggingface_hub
Browse files- README.md +31 -0
- checkpoint-6223/config.json +33 -0
- checkpoint-6223/model.safetensors +3 -0
- checkpoint-6223/optimizer.pt +3 -0
- checkpoint-6223/rng_state.pth +3 -0
- checkpoint-6223/scheduler.pt +3 -0
- checkpoint-6223/trainer_state.json +2023 -0
- checkpoint-6223/training_args.bin +3 -0
- config.json +33 -0
- model.safetensors +3 -0
- preprocessor_config.json +22 -0
- runs/Aug18_14-29-31_PcMasterCarrera/events.out.tfevents.1724005772.PcMasterCarrera.5764.0 +3 -0
- runs/Aug18_14-29-31_PcMasterCarrera/events.out.tfevents.1724008308.PcMasterCarrera.5764.1 +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-large-patch32-384
|
| 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: 0.3894089162349701
|
| 22 |
+
|
| 23 |
+
f1: 0.4625
|
| 24 |
+
|
| 25 |
+
precision: 0.6379310344827587
|
| 26 |
+
|
| 27 |
+
recall: 0.3627450980392157
|
| 28 |
+
|
| 29 |
+
auc: 0.8453785257418476
|
| 30 |
+
|
| 31 |
+
accuracy: 0.819327731092437
|
checkpoint-6223/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-large-patch32-384",
|
| 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": 1024,
|
| 12 |
+
"id2label": {
|
| 13 |
+
"0": "no",
|
| 14 |
+
"1": "yes"
|
| 15 |
+
},
|
| 16 |
+
"image_size": 384,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 4096,
|
| 19 |
+
"label2id": {
|
| 20 |
+
"no": 0,
|
| 21 |
+
"yes": 1
|
| 22 |
+
},
|
| 23 |
+
"layer_norm_eps": 1e-12,
|
| 24 |
+
"model_type": "vit",
|
| 25 |
+
"num_attention_heads": 16,
|
| 26 |
+
"num_channels": 3,
|
| 27 |
+
"num_hidden_layers": 24,
|
| 28 |
+
"patch_size": 32,
|
| 29 |
+
"problem_type": "single_label_classification",
|
| 30 |
+
"qkv_bias": true,
|
| 31 |
+
"torch_dtype": "float32",
|
| 32 |
+
"transformers_version": "4.44.0"
|
| 33 |
+
}
|
checkpoint-6223/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8c74ff4bcdac4d551a8a078b510f9615f76250ffa5bbe19830f82a09e570c7a6
|
| 3 |
+
size 1222485480
|
checkpoint-6223/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:20a9ae6ffce200b72c51d9484bf01a9f0b50ca390589ad4ede694332542aac4c
|
| 3 |
+
size 2445205930
|
checkpoint-6223/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:322d2101024269ac2cb4f301b8c0c780df48b85580401b609562e8d789f77724
|
| 3 |
+
size 14180
|
checkpoint-6223/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:37902947342502e99e569f31337b39526dc42e84946dcbff84963b04146018b4
|
| 3 |
+
size 1064
|
checkpoint-6223/trainer_state.json
ADDED
|
@@ -0,0 +1,2023 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.3894089162349701,
|
| 3 |
+
"best_model_checkpoint": "prostate-mri-T2w-v05/checkpoint-6223",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 6223,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.025871766029246346,
|
| 13 |
+
"grad_norm": 62.908626556396484,
|
| 14 |
+
"learning_rate": 2.5871766029246344e-07,
|
| 15 |
+
"loss": 0.9082,
|
| 16 |
+
"step": 23
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.05174353205849269,
|
| 20 |
+
"grad_norm": 17.856958389282227,
|
| 21 |
+
"learning_rate": 5.174353205849269e-07,
|
| 22 |
+
"loss": 0.9712,
|
| 23 |
+
"step": 46
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.07761529808773904,
|
| 27 |
+
"grad_norm": 14.738483428955078,
|
| 28 |
+
"learning_rate": 7.761529808773904e-07,
|
| 29 |
+
"loss": 0.7274,
|
| 30 |
+
"step": 69
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.10348706411698538,
|
| 34 |
+
"grad_norm": 11.495190620422363,
|
| 35 |
+
"learning_rate": 1.0348706411698538e-06,
|
| 36 |
+
"loss": 0.5668,
|
| 37 |
+
"step": 92
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.12935883014623173,
|
| 41 |
+
"grad_norm": 14.365918159484863,
|
| 42 |
+
"learning_rate": 1.2935883014623174e-06,
|
| 43 |
+
"loss": 0.5766,
|
| 44 |
+
"step": 115
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.15523059617547807,
|
| 48 |
+
"grad_norm": 4.656925678253174,
|
| 49 |
+
"learning_rate": 1.5523059617547808e-06,
|
| 50 |
+
"loss": 0.5059,
|
| 51 |
+
"step": 138
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.18110236220472442,
|
| 55 |
+
"grad_norm": 9.39604663848877,
|
| 56 |
+
"learning_rate": 1.8110236220472441e-06,
|
| 57 |
+
"loss": 0.4405,
|
| 58 |
+
"step": 161
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.20697412823397077,
|
| 62 |
+
"grad_norm": 9.823530197143555,
|
| 63 |
+
"learning_rate": 2.0697412823397075e-06,
|
| 64 |
+
"loss": 0.3319,
|
| 65 |
+
"step": 184
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.23284589426321708,
|
| 69 |
+
"grad_norm": 10.687150955200195,
|
| 70 |
+
"learning_rate": 2.328458942632171e-06,
|
| 71 |
+
"loss": 0.4602,
|
| 72 |
+
"step": 207
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.25871766029246346,
|
| 76 |
+
"grad_norm": 1.1277879476547241,
|
| 77 |
+
"learning_rate": 2.5871766029246347e-06,
|
| 78 |
+
"loss": 0.3568,
|
| 79 |
+
"step": 230
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.2845894263217098,
|
| 83 |
+
"grad_norm": 0.47847771644592285,
|
| 84 |
+
"learning_rate": 2.845894263217098e-06,
|
| 85 |
+
"loss": 0.6281,
|
| 86 |
+
"step": 253
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.31046119235095615,
|
| 90 |
+
"grad_norm": 0.5296943187713623,
|
| 91 |
+
"learning_rate": 3.1046119235095615e-06,
|
| 92 |
+
"loss": 0.3346,
|
| 93 |
+
"step": 276
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.3363329583802025,
|
| 97 |
+
"grad_norm": 11.51565170288086,
|
| 98 |
+
"learning_rate": 3.3633295838020245e-06,
|
| 99 |
+
"loss": 0.5892,
|
| 100 |
+
"step": 299
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.36220472440944884,
|
| 104 |
+
"grad_norm": 18.972782135009766,
|
| 105 |
+
"learning_rate": 3.6220472440944883e-06,
|
| 106 |
+
"loss": 0.7544,
|
| 107 |
+
"step": 322
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.3880764904386952,
|
| 111 |
+
"grad_norm": 10.762083053588867,
|
| 112 |
+
"learning_rate": 3.880764904386952e-06,
|
| 113 |
+
"loss": 0.6232,
|
| 114 |
+
"step": 345
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.41394825646794153,
|
| 118 |
+
"grad_norm": 11.391585350036621,
|
| 119 |
+
"learning_rate": 4.139482564679415e-06,
|
| 120 |
+
"loss": 0.4856,
|
| 121 |
+
"step": 368
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.4398200224971879,
|
| 125 |
+
"grad_norm": 1.076779842376709,
|
| 126 |
+
"learning_rate": 4.3982002249718785e-06,
|
| 127 |
+
"loss": 0.3987,
|
| 128 |
+
"step": 391
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.46569178852643417,
|
| 132 |
+
"grad_norm": 11.604365348815918,
|
| 133 |
+
"learning_rate": 4.656917885264342e-06,
|
| 134 |
+
"loss": 0.5967,
|
| 135 |
+
"step": 414
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.4915635545556805,
|
| 139 |
+
"grad_norm": 10.939277648925781,
|
| 140 |
+
"learning_rate": 4.915635545556806e-06,
|
| 141 |
+
"loss": 0.3712,
|
| 142 |
+
"step": 437
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.5174353205849269,
|
| 146 |
+
"grad_norm": 10.872827529907227,
|
| 147 |
+
"learning_rate": 5.1743532058492695e-06,
|
| 148 |
+
"loss": 0.4234,
|
| 149 |
+
"step": 460
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.5433070866141733,
|
| 153 |
+
"grad_norm": 13.358455657958984,
|
| 154 |
+
"learning_rate": 5.433070866141733e-06,
|
| 155 |
+
"loss": 0.6415,
|
| 156 |
+
"step": 483
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.5691788526434196,
|
| 160 |
+
"grad_norm": 0.5069867968559265,
|
| 161 |
+
"learning_rate": 5.691788526434196e-06,
|
| 162 |
+
"loss": 0.3304,
|
| 163 |
+
"step": 506
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.595050618672666,
|
| 167 |
+
"grad_norm": 11.854174613952637,
|
| 168 |
+
"learning_rate": 5.95050618672666e-06,
|
| 169 |
+
"loss": 0.2956,
|
| 170 |
+
"step": 529
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.6209223847019123,
|
| 174 |
+
"grad_norm": 0.6237831711769104,
|
| 175 |
+
"learning_rate": 6.209223847019123e-06,
|
| 176 |
+
"loss": 0.5922,
|
| 177 |
+
"step": 552
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.6467941507311586,
|
| 181 |
+
"grad_norm": 18.399309158325195,
|
| 182 |
+
"learning_rate": 6.467941507311586e-06,
|
| 183 |
+
"loss": 0.5529,
|
| 184 |
+
"step": 575
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.672665916760405,
|
| 188 |
+
"grad_norm": 1.0723528861999512,
|
| 189 |
+
"learning_rate": 6.726659167604049e-06,
|
| 190 |
+
"loss": 0.5464,
|
| 191 |
+
"step": 598
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.6985376827896513,
|
| 195 |
+
"grad_norm": 10.857867240905762,
|
| 196 |
+
"learning_rate": 6.985376827896513e-06,
|
| 197 |
+
"loss": 0.5258,
|
| 198 |
+
"step": 621
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.7244094488188977,
|
| 202 |
+
"grad_norm": 11.417038917541504,
|
| 203 |
+
"learning_rate": 7.244094488188977e-06,
|
| 204 |
+
"loss": 0.4599,
|
| 205 |
+
"step": 644
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.750281214848144,
|
| 209 |
+
"grad_norm": 12.366191864013672,
|
| 210 |
+
"learning_rate": 7.50281214848144e-06,
|
| 211 |
+
"loss": 0.5278,
|
| 212 |
+
"step": 667
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.7761529808773904,
|
| 216 |
+
"grad_norm": 59.656211853027344,
|
| 217 |
+
"learning_rate": 7.761529808773903e-06,
|
| 218 |
+
"loss": 0.4802,
|
| 219 |
+
"step": 690
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.8020247469066367,
|
| 223 |
+
"grad_norm": 1.6525479555130005,
|
| 224 |
+
"learning_rate": 8.020247469066367e-06,
|
| 225 |
+
"loss": 0.3855,
|
| 226 |
+
"step": 713
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.8278965129358831,
|
| 230 |
+
"grad_norm": 11.62271499633789,
|
| 231 |
+
"learning_rate": 8.27896512935883e-06,
|
| 232 |
+
"loss": 0.5377,
|
| 233 |
+
"step": 736
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.8537682789651294,
|
| 237 |
+
"grad_norm": 11.242469787597656,
|
| 238 |
+
"learning_rate": 8.537682789651294e-06,
|
| 239 |
+
"loss": 0.6707,
|
| 240 |
+
"step": 759
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.8796400449943758,
|
| 244 |
+
"grad_norm": 1.673377513885498,
|
| 245 |
+
"learning_rate": 8.796400449943757e-06,
|
| 246 |
+
"loss": 0.434,
|
| 247 |
+
"step": 782
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.905511811023622,
|
| 251 |
+
"grad_norm": 0.3748597800731659,
|
| 252 |
+
"learning_rate": 9.055118110236222e-06,
|
| 253 |
+
"loss": 0.4709,
|
| 254 |
+
"step": 805
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.9313835770528683,
|
| 258 |
+
"grad_norm": 10.111214637756348,
|
| 259 |
+
"learning_rate": 9.313835770528684e-06,
|
| 260 |
+
"loss": 0.5416,
|
| 261 |
+
"step": 828
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.9572553430821147,
|
| 265 |
+
"grad_norm": 11.05431842803955,
|
| 266 |
+
"learning_rate": 9.572553430821147e-06,
|
| 267 |
+
"loss": 0.4895,
|
| 268 |
+
"step": 851
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.983127109111361,
|
| 272 |
+
"grad_norm": 0.5102259516716003,
|
| 273 |
+
"learning_rate": 9.831271091113612e-06,
|
| 274 |
+
"loss": 0.6304,
|
| 275 |
+
"step": 874
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.0,
|
| 279 |
+
"eval_accuracy": 0.7857142857142857,
|
| 280 |
+
"eval_auc": 0.8011167033658384,
|
| 281 |
+
"eval_f1": 0.0,
|
| 282 |
+
"eval_loss": 0.7771126627922058,
|
| 283 |
+
"eval_precision": 0.0,
|
| 284 |
+
"eval_recall": 0.0,
|
| 285 |
+
"eval_runtime": 2.7441,
|
| 286 |
+
"eval_samples_per_second": 173.466,
|
| 287 |
+
"eval_steps_per_second": 21.865,
|
| 288 |
+
"step": 889
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 1.0089988751406074,
|
| 292 |
+
"grad_norm": 13.26599407196045,
|
| 293 |
+
"learning_rate": 1.0089988751406076e-05,
|
| 294 |
+
"loss": 0.3247,
|
| 295 |
+
"step": 897
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.0348706411698538,
|
| 299 |
+
"grad_norm": 11.029207229614258,
|
| 300 |
+
"learning_rate": 1.0348706411698539e-05,
|
| 301 |
+
"loss": 0.6095,
|
| 302 |
+
"step": 920
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.0607424071991,
|
| 306 |
+
"grad_norm": 0.07530930638313293,
|
| 307 |
+
"learning_rate": 1.0607424071991e-05,
|
| 308 |
+
"loss": 0.1472,
|
| 309 |
+
"step": 943
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 1.0866141732283465,
|
| 313 |
+
"grad_norm": 0.6263224482536316,
|
| 314 |
+
"learning_rate": 1.0866141732283466e-05,
|
| 315 |
+
"loss": 0.6615,
|
| 316 |
+
"step": 966
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 1.1124859392575928,
|
| 320 |
+
"grad_norm": 9.975605964660645,
|
| 321 |
+
"learning_rate": 1.1124859392575929e-05,
|
| 322 |
+
"loss": 0.4082,
|
| 323 |
+
"step": 989
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.1383577052868392,
|
| 327 |
+
"grad_norm": 10.850635528564453,
|
| 328 |
+
"learning_rate": 1.1383577052868393e-05,
|
| 329 |
+
"loss": 0.4151,
|
| 330 |
+
"step": 1012
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 1.1642294713160855,
|
| 334 |
+
"grad_norm": 10.857563972473145,
|
| 335 |
+
"learning_rate": 1.1642294713160854e-05,
|
| 336 |
+
"loss": 0.4441,
|
| 337 |
+
"step": 1035
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 1.190101237345332,
|
| 341 |
+
"grad_norm": 4.742630958557129,
|
| 342 |
+
"learning_rate": 1.190101237345332e-05,
|
| 343 |
+
"loss": 0.7853,
|
| 344 |
+
"step": 1058
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 1.2159730033745781,
|
| 348 |
+
"grad_norm": 0.09981119632720947,
|
| 349 |
+
"learning_rate": 1.2159730033745783e-05,
|
| 350 |
+
"loss": 0.2316,
|
| 351 |
+
"step": 1081
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 1.2418447694038246,
|
| 355 |
+
"grad_norm": 0.37830597162246704,
|
| 356 |
+
"learning_rate": 1.2418447694038246e-05,
|
| 357 |
+
"loss": 0.5479,
|
| 358 |
+
"step": 1104
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 1.2677165354330708,
|
| 362 |
+
"grad_norm": 1.0691417455673218,
|
| 363 |
+
"learning_rate": 1.2677165354330708e-05,
|
| 364 |
+
"loss": 0.342,
|
| 365 |
+
"step": 1127
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 1.2935883014623173,
|
| 369 |
+
"grad_norm": 10.283402442932129,
|
| 370 |
+
"learning_rate": 1.2935883014623173e-05,
|
| 371 |
+
"loss": 0.591,
|
| 372 |
+
"step": 1150
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 1.3194600674915635,
|
| 376 |
+
"grad_norm": 2.4355926513671875,
|
| 377 |
+
"learning_rate": 1.3194600674915636e-05,
|
| 378 |
+
"loss": 0.486,
|
| 379 |
+
"step": 1173
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 1.34533183352081,
|
| 383 |
+
"grad_norm": 7.244462966918945,
|
| 384 |
+
"learning_rate": 1.3453318335208098e-05,
|
| 385 |
+
"loss": 0.5206,
|
| 386 |
+
"step": 1196
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 1.3712035995500562,
|
| 390 |
+
"grad_norm": 9.888749122619629,
|
| 391 |
+
"learning_rate": 1.3712035995500563e-05,
|
| 392 |
+
"loss": 0.4504,
|
| 393 |
+
"step": 1219
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 1.3970753655793025,
|
| 397 |
+
"grad_norm": 3.3459055423736572,
|
| 398 |
+
"learning_rate": 1.3970753655793026e-05,
|
| 399 |
+
"loss": 0.4415,
|
| 400 |
+
"step": 1242
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 1.422947131608549,
|
| 404 |
+
"grad_norm": 0.175747811794281,
|
| 405 |
+
"learning_rate": 1.4229471316085491e-05,
|
| 406 |
+
"loss": 0.2943,
|
| 407 |
+
"step": 1265
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.4488188976377954,
|
| 411 |
+
"grad_norm": 4.791985988616943,
|
| 412 |
+
"learning_rate": 1.4488188976377953e-05,
|
| 413 |
+
"loss": 0.566,
|
| 414 |
+
"step": 1288
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 1.4746906636670416,
|
| 418 |
+
"grad_norm": 12.70669174194336,
|
| 419 |
+
"learning_rate": 1.4746906636670415e-05,
|
| 420 |
+
"loss": 0.4687,
|
| 421 |
+
"step": 1311
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 1.5005624296962878,
|
| 425 |
+
"grad_norm": 0.04389326646924019,
|
| 426 |
+
"learning_rate": 1.500562429696288e-05,
|
| 427 |
+
"loss": 0.2629,
|
| 428 |
+
"step": 1334
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 1.5264341957255343,
|
| 432 |
+
"grad_norm": 4.369255542755127,
|
| 433 |
+
"learning_rate": 1.5264341957255343e-05,
|
| 434 |
+
"loss": 0.472,
|
| 435 |
+
"step": 1357
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 1.5523059617547807,
|
| 439 |
+
"grad_norm": 4.454276084899902,
|
| 440 |
+
"learning_rate": 1.5523059617547807e-05,
|
| 441 |
+
"loss": 0.4307,
|
| 442 |
+
"step": 1380
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.578177727784027,
|
| 446 |
+
"grad_norm": 1.1173179149627686,
|
| 447 |
+
"learning_rate": 1.578177727784027e-05,
|
| 448 |
+
"loss": 0.3431,
|
| 449 |
+
"step": 1403
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.6040494938132732,
|
| 453 |
+
"grad_norm": 3.0332882404327393,
|
| 454 |
+
"learning_rate": 1.6040494938132734e-05,
|
| 455 |
+
"loss": 0.2545,
|
| 456 |
+
"step": 1426
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.6299212598425197,
|
| 460 |
+
"grad_norm": 10.521413803100586,
|
| 461 |
+
"learning_rate": 1.6299212598425197e-05,
|
| 462 |
+
"loss": 0.6555,
|
| 463 |
+
"step": 1449
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.6557930258717661,
|
| 467 |
+
"grad_norm": 3.8012301921844482,
|
| 468 |
+
"learning_rate": 1.655793025871766e-05,
|
| 469 |
+
"loss": 0.37,
|
| 470 |
+
"step": 1472
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.6816647919010124,
|
| 474 |
+
"grad_norm": 0.05941835045814514,
|
| 475 |
+
"learning_rate": 1.6816647919010127e-05,
|
| 476 |
+
"loss": 0.3091,
|
| 477 |
+
"step": 1495
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 1.7075365579302586,
|
| 481 |
+
"grad_norm": 0.29451972246170044,
|
| 482 |
+
"learning_rate": 1.7075365579302587e-05,
|
| 483 |
+
"loss": 0.3511,
|
| 484 |
+
"step": 1518
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 1.733408323959505,
|
| 488 |
+
"grad_norm": 3.2587544918060303,
|
| 489 |
+
"learning_rate": 1.733408323959505e-05,
|
| 490 |
+
"loss": 0.6164,
|
| 491 |
+
"step": 1541
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 1.7592800899887515,
|
| 495 |
+
"grad_norm": 3.4113192558288574,
|
| 496 |
+
"learning_rate": 1.7592800899887514e-05,
|
| 497 |
+
"loss": 0.3817,
|
| 498 |
+
"step": 1564
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 1.7851518560179978,
|
| 502 |
+
"grad_norm": 11.975400924682617,
|
| 503 |
+
"learning_rate": 1.7851518560179977e-05,
|
| 504 |
+
"loss": 0.3421,
|
| 505 |
+
"step": 1587
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 1.811023622047244,
|
| 509 |
+
"grad_norm": 1.5508025884628296,
|
| 510 |
+
"learning_rate": 1.8110236220472444e-05,
|
| 511 |
+
"loss": 0.6197,
|
| 512 |
+
"step": 1610
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 1.8368953880764904,
|
| 516 |
+
"grad_norm": 0.24332457780838013,
|
| 517 |
+
"learning_rate": 1.8368953880764904e-05,
|
| 518 |
+
"loss": 0.4227,
|
| 519 |
+
"step": 1633
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 1.862767154105737,
|
| 523 |
+
"grad_norm": 0.10154075175523758,
|
| 524 |
+
"learning_rate": 1.8627671541057367e-05,
|
| 525 |
+
"loss": 0.4287,
|
| 526 |
+
"step": 1656
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 1.8886389201349831,
|
| 530 |
+
"grad_norm": 10.398853302001953,
|
| 531 |
+
"learning_rate": 1.8886389201349834e-05,
|
| 532 |
+
"loss": 0.3014,
|
| 533 |
+
"step": 1679
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 1.9145106861642294,
|
| 537 |
+
"grad_norm": 0.8489835858345032,
|
| 538 |
+
"learning_rate": 1.9145106861642294e-05,
|
| 539 |
+
"loss": 0.3982,
|
| 540 |
+
"step": 1702
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 1.9403824521934758,
|
| 544 |
+
"grad_norm": 0.8622331619262695,
|
| 545 |
+
"learning_rate": 1.940382452193476e-05,
|
| 546 |
+
"loss": 0.5771,
|
| 547 |
+
"step": 1725
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 1.9662542182227223,
|
| 551 |
+
"grad_norm": 2.847560405731201,
|
| 552 |
+
"learning_rate": 1.9662542182227224e-05,
|
| 553 |
+
"loss": 0.4758,
|
| 554 |
+
"step": 1748
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 1.9921259842519685,
|
| 558 |
+
"grad_norm": 0.5839571952819824,
|
| 559 |
+
"learning_rate": 1.9921259842519684e-05,
|
| 560 |
+
"loss": 0.2732,
|
| 561 |
+
"step": 1771
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 2.0,
|
| 565 |
+
"eval_accuracy": 0.7857142857142857,
|
| 566 |
+
"eval_auc": 0.7965817342979973,
|
| 567 |
+
"eval_f1": 0.0,
|
| 568 |
+
"eval_loss": 0.9842509627342224,
|
| 569 |
+
"eval_precision": 0.0,
|
| 570 |
+
"eval_recall": 0.0,
|
| 571 |
+
"eval_runtime": 2.74,
|
| 572 |
+
"eval_samples_per_second": 173.72,
|
| 573 |
+
"eval_steps_per_second": 21.897,
|
| 574 |
+
"step": 1778
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 2.0179977502812148,
|
| 578 |
+
"grad_norm": 0.23481443524360657,
|
| 579 |
+
"learning_rate": 2.017997750281215e-05,
|
| 580 |
+
"loss": 0.2917,
|
| 581 |
+
"step": 1794
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 2.043869516310461,
|
| 585 |
+
"grad_norm": 0.1679057627916336,
|
| 586 |
+
"learning_rate": 2.043869516310461e-05,
|
| 587 |
+
"loss": 0.5162,
|
| 588 |
+
"step": 1817
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 2.0697412823397077,
|
| 592 |
+
"grad_norm": 5.455684185028076,
|
| 593 |
+
"learning_rate": 2.0697412823397078e-05,
|
| 594 |
+
"loss": 0.5789,
|
| 595 |
+
"step": 1840
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 2.095613048368954,
|
| 599 |
+
"grad_norm": 7.542656421661377,
|
| 600 |
+
"learning_rate": 2.095613048368954e-05,
|
| 601 |
+
"loss": 0.5314,
|
| 602 |
+
"step": 1863
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 2.1214848143982,
|
| 606 |
+
"grad_norm": 2.186872720718384,
|
| 607 |
+
"learning_rate": 2.1214848143982e-05,
|
| 608 |
+
"loss": 0.3816,
|
| 609 |
+
"step": 1886
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 2.1473565804274464,
|
| 613 |
+
"grad_norm": 0.19364489614963531,
|
| 614 |
+
"learning_rate": 2.1473565804274468e-05,
|
| 615 |
+
"loss": 0.336,
|
| 616 |
+
"step": 1909
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 2.173228346456693,
|
| 620 |
+
"grad_norm": 0.20303942263126373,
|
| 621 |
+
"learning_rate": 2.173228346456693e-05,
|
| 622 |
+
"loss": 0.421,
|
| 623 |
+
"step": 1932
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 2.1991001124859393,
|
| 627 |
+
"grad_norm": 0.5988159775733948,
|
| 628 |
+
"learning_rate": 2.1991001124859395e-05,
|
| 629 |
+
"loss": 0.4587,
|
| 630 |
+
"step": 1955
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 2.2249718785151855,
|
| 634 |
+
"grad_norm": 7.407308101654053,
|
| 635 |
+
"learning_rate": 2.2249718785151858e-05,
|
| 636 |
+
"loss": 0.6277,
|
| 637 |
+
"step": 1978
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 2.2508436445444318,
|
| 641 |
+
"grad_norm": 2.7972147464752197,
|
| 642 |
+
"learning_rate": 2.2508436445444318e-05,
|
| 643 |
+
"loss": 0.4025,
|
| 644 |
+
"step": 2001
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 2.2767154105736784,
|
| 648 |
+
"grad_norm": 14.71440601348877,
|
| 649 |
+
"learning_rate": 2.2767154105736785e-05,
|
| 650 |
+
"loss": 0.4362,
|
| 651 |
+
"step": 2024
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 2.3025871766029247,
|
| 655 |
+
"grad_norm": 0.6409655213356018,
|
| 656 |
+
"learning_rate": 2.302587176602925e-05,
|
| 657 |
+
"loss": 0.4014,
|
| 658 |
+
"step": 2047
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 2.328458942632171,
|
| 662 |
+
"grad_norm": 0.0657491534948349,
|
| 663 |
+
"learning_rate": 2.328458942632171e-05,
|
| 664 |
+
"loss": 0.3942,
|
| 665 |
+
"step": 2070
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 2.354330708661417,
|
| 669 |
+
"grad_norm": 0.9892528653144836,
|
| 670 |
+
"learning_rate": 2.3543307086614175e-05,
|
| 671 |
+
"loss": 0.5464,
|
| 672 |
+
"step": 2093
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 2.380202474690664,
|
| 676 |
+
"grad_norm": 0.6635844111442566,
|
| 677 |
+
"learning_rate": 2.380202474690664e-05,
|
| 678 |
+
"loss": 0.4767,
|
| 679 |
+
"step": 2116
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 2.40607424071991,
|
| 683 |
+
"grad_norm": 7.151478290557861,
|
| 684 |
+
"learning_rate": 2.4060742407199102e-05,
|
| 685 |
+
"loss": 0.4637,
|
| 686 |
+
"step": 2139
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 2.4319460067491563,
|
| 690 |
+
"grad_norm": 0.7382416129112244,
|
| 691 |
+
"learning_rate": 2.4319460067491565e-05,
|
| 692 |
+
"loss": 0.3082,
|
| 693 |
+
"step": 2162
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 2.4578177727784025,
|
| 697 |
+
"grad_norm": 9.286602973937988,
|
| 698 |
+
"learning_rate": 2.4578177727784025e-05,
|
| 699 |
+
"loss": 0.5282,
|
| 700 |
+
"step": 2185
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 2.483689538807649,
|
| 704 |
+
"grad_norm": 4.055200099945068,
|
| 705 |
+
"learning_rate": 2.4836895388076492e-05,
|
| 706 |
+
"loss": 0.457,
|
| 707 |
+
"step": 2208
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 2.5095613048368954,
|
| 711 |
+
"grad_norm": 0.12764699757099152,
|
| 712 |
+
"learning_rate": 2.5095613048368956e-05,
|
| 713 |
+
"loss": 0.3318,
|
| 714 |
+
"step": 2231
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 2.5354330708661417,
|
| 718 |
+
"grad_norm": 2.9867072105407715,
|
| 719 |
+
"learning_rate": 2.5354330708661416e-05,
|
| 720 |
+
"loss": 0.3542,
|
| 721 |
+
"step": 2254
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 2.5613048368953883,
|
| 725 |
+
"grad_norm": 4.703469753265381,
|
| 726 |
+
"learning_rate": 2.5613048368953886e-05,
|
| 727 |
+
"loss": 0.4748,
|
| 728 |
+
"step": 2277
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 2.5871766029246346,
|
| 732 |
+
"grad_norm": 9.484485626220703,
|
| 733 |
+
"learning_rate": 2.5871766029246346e-05,
|
| 734 |
+
"loss": 0.4135,
|
| 735 |
+
"step": 2300
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.613048368953881,
|
| 739 |
+
"grad_norm": 18.093868255615234,
|
| 740 |
+
"learning_rate": 2.613048368953881e-05,
|
| 741 |
+
"loss": 0.4656,
|
| 742 |
+
"step": 2323
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 2.638920134983127,
|
| 746 |
+
"grad_norm": 0.21779067814350128,
|
| 747 |
+
"learning_rate": 2.6389201349831272e-05,
|
| 748 |
+
"loss": 0.3057,
|
| 749 |
+
"step": 2346
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 2.6647919010123733,
|
| 753 |
+
"grad_norm": 0.7615957260131836,
|
| 754 |
+
"learning_rate": 2.6647919010123736e-05,
|
| 755 |
+
"loss": 0.4652,
|
| 756 |
+
"step": 2369
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 2.69066366704162,
|
| 760 |
+
"grad_norm": 0.35096365213394165,
|
| 761 |
+
"learning_rate": 2.6906636670416196e-05,
|
| 762 |
+
"loss": 0.3751,
|
| 763 |
+
"step": 2392
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 2.716535433070866,
|
| 767 |
+
"grad_norm": 0.5412973761558533,
|
| 768 |
+
"learning_rate": 2.7165354330708666e-05,
|
| 769 |
+
"loss": 0.1823,
|
| 770 |
+
"step": 2415
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 2.7424071991001124,
|
| 774 |
+
"grad_norm": 1.9220446348190308,
|
| 775 |
+
"learning_rate": 2.7424071991001126e-05,
|
| 776 |
+
"loss": 0.6485,
|
| 777 |
+
"step": 2438
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 2.7682789651293587,
|
| 781 |
+
"grad_norm": 11.08309268951416,
|
| 782 |
+
"learning_rate": 2.768278965129359e-05,
|
| 783 |
+
"loss": 0.4777,
|
| 784 |
+
"step": 2461
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"epoch": 2.794150731158605,
|
| 788 |
+
"grad_norm": 0.0344441793859005,
|
| 789 |
+
"learning_rate": 2.7941507311586053e-05,
|
| 790 |
+
"loss": 0.2201,
|
| 791 |
+
"step": 2484
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 2.8200224971878516,
|
| 795 |
+
"grad_norm": 0.9013801217079163,
|
| 796 |
+
"learning_rate": 2.8200224971878513e-05,
|
| 797 |
+
"loss": 0.471,
|
| 798 |
+
"step": 2507
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 2.845894263217098,
|
| 802 |
+
"grad_norm": 1.6073826551437378,
|
| 803 |
+
"learning_rate": 2.8458942632170983e-05,
|
| 804 |
+
"loss": 0.5535,
|
| 805 |
+
"step": 2530
|
| 806 |
+
},
|
| 807 |
+
{
|
| 808 |
+
"epoch": 2.871766029246344,
|
| 809 |
+
"grad_norm": 3.7488861083984375,
|
| 810 |
+
"learning_rate": 2.8717660292463443e-05,
|
| 811 |
+
"loss": 0.525,
|
| 812 |
+
"step": 2553
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 2.8976377952755907,
|
| 816 |
+
"grad_norm": 0.2803422808647156,
|
| 817 |
+
"learning_rate": 2.8976377952755906e-05,
|
| 818 |
+
"loss": 0.4241,
|
| 819 |
+
"step": 2576
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 2.923509561304837,
|
| 823 |
+
"grad_norm": 2.339632749557495,
|
| 824 |
+
"learning_rate": 2.923509561304837e-05,
|
| 825 |
+
"loss": 0.3422,
|
| 826 |
+
"step": 2599
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 2.949381327334083,
|
| 830 |
+
"grad_norm": 0.10544496774673462,
|
| 831 |
+
"learning_rate": 2.949381327334083e-05,
|
| 832 |
+
"loss": 0.2698,
|
| 833 |
+
"step": 2622
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 2.9752530933633294,
|
| 837 |
+
"grad_norm": 5.151175498962402,
|
| 838 |
+
"learning_rate": 2.97525309336333e-05,
|
| 839 |
+
"loss": 0.5485,
|
| 840 |
+
"step": 2645
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 3.0,
|
| 844 |
+
"eval_accuracy": 0.7815126050420168,
|
| 845 |
+
"eval_auc": 0.7921778337003251,
|
| 846 |
+
"eval_f1": 0.0,
|
| 847 |
+
"eval_loss": 0.7073759436607361,
|
| 848 |
+
"eval_precision": 0.0,
|
| 849 |
+
"eval_recall": 0.0,
|
| 850 |
+
"eval_runtime": 2.7321,
|
| 851 |
+
"eval_samples_per_second": 174.227,
|
| 852 |
+
"eval_steps_per_second": 21.961,
|
| 853 |
+
"step": 2667
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 3.001124859392576,
|
| 857 |
+
"grad_norm": 7.475840091705322,
|
| 858 |
+
"learning_rate": 3.001124859392576e-05,
|
| 859 |
+
"loss": 0.2998,
|
| 860 |
+
"step": 2668
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 3.0269966254218224,
|
| 864 |
+
"grad_norm": 0.36270514130592346,
|
| 865 |
+
"learning_rate": 3.0269966254218223e-05,
|
| 866 |
+
"loss": 0.3207,
|
| 867 |
+
"step": 2691
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 3.0528683914510686,
|
| 871 |
+
"grad_norm": 3.655841112136841,
|
| 872 |
+
"learning_rate": 3.052868391451069e-05,
|
| 873 |
+
"loss": 0.376,
|
| 874 |
+
"step": 2714
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 3.078740157480315,
|
| 878 |
+
"grad_norm": 0.28142863512039185,
|
| 879 |
+
"learning_rate": 3.078740157480315e-05,
|
| 880 |
+
"loss": 0.3376,
|
| 881 |
+
"step": 2737
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 3.1046119235095615,
|
| 885 |
+
"grad_norm": 0.1056133359670639,
|
| 886 |
+
"learning_rate": 3.1046119235095613e-05,
|
| 887 |
+
"loss": 0.43,
|
| 888 |
+
"step": 2760
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 3.1304836895388077,
|
| 892 |
+
"grad_norm": 10.104964256286621,
|
| 893 |
+
"learning_rate": 3.130483689538808e-05,
|
| 894 |
+
"loss": 0.4603,
|
| 895 |
+
"step": 2783
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 3.156355455568054,
|
| 899 |
+
"grad_norm": 8.072303771972656,
|
| 900 |
+
"learning_rate": 3.156355455568054e-05,
|
| 901 |
+
"loss": 0.3976,
|
| 902 |
+
"step": 2806
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 3.1822272215973,
|
| 906 |
+
"grad_norm": 10.715200424194336,
|
| 907 |
+
"learning_rate": 3.1822272215973004e-05,
|
| 908 |
+
"loss": 0.4361,
|
| 909 |
+
"step": 2829
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 3.208098987626547,
|
| 913 |
+
"grad_norm": 1.3668488264083862,
|
| 914 |
+
"learning_rate": 3.208098987626547e-05,
|
| 915 |
+
"loss": 0.2668,
|
| 916 |
+
"step": 2852
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 3.233970753655793,
|
| 920 |
+
"grad_norm": 1.127758502960205,
|
| 921 |
+
"learning_rate": 3.233970753655794e-05,
|
| 922 |
+
"loss": 0.3974,
|
| 923 |
+
"step": 2875
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 3.2598425196850394,
|
| 927 |
+
"grad_norm": 0.872011125087738,
|
| 928 |
+
"learning_rate": 3.2598425196850394e-05,
|
| 929 |
+
"loss": 0.3557,
|
| 930 |
+
"step": 2898
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 3.2857142857142856,
|
| 934 |
+
"grad_norm": 15.174445152282715,
|
| 935 |
+
"learning_rate": 3.285714285714286e-05,
|
| 936 |
+
"loss": 0.4542,
|
| 937 |
+
"step": 2921
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 3.3115860517435323,
|
| 941 |
+
"grad_norm": 3.535675048828125,
|
| 942 |
+
"learning_rate": 3.311586051743532e-05,
|
| 943 |
+
"loss": 0.3665,
|
| 944 |
+
"step": 2944
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 3.3374578177727785,
|
| 948 |
+
"grad_norm": 0.1963585466146469,
|
| 949 |
+
"learning_rate": 3.3374578177727784e-05,
|
| 950 |
+
"loss": 0.4275,
|
| 951 |
+
"step": 2967
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 3.3633295838020247,
|
| 955 |
+
"grad_norm": 1.3127784729003906,
|
| 956 |
+
"learning_rate": 3.3633295838020254e-05,
|
| 957 |
+
"loss": 0.4884,
|
| 958 |
+
"step": 2990
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 3.389201349831271,
|
| 962 |
+
"grad_norm": 1.3405988216400146,
|
| 963 |
+
"learning_rate": 3.389201349831271e-05,
|
| 964 |
+
"loss": 0.2996,
|
| 965 |
+
"step": 3013
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 3.415073115860517,
|
| 969 |
+
"grad_norm": 9.742510795593262,
|
| 970 |
+
"learning_rate": 3.4150731158605174e-05,
|
| 971 |
+
"loss": 0.4061,
|
| 972 |
+
"step": 3036
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 3.440944881889764,
|
| 976 |
+
"grad_norm": 0.6711751222610474,
|
| 977 |
+
"learning_rate": 3.440944881889764e-05,
|
| 978 |
+
"loss": 0.3684,
|
| 979 |
+
"step": 3059
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 3.46681664791901,
|
| 983 |
+
"grad_norm": 9.26884937286377,
|
| 984 |
+
"learning_rate": 3.46681664791901e-05,
|
| 985 |
+
"loss": 0.4557,
|
| 986 |
+
"step": 3082
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 3.4926884139482564,
|
| 990 |
+
"grad_norm": 11.18705940246582,
|
| 991 |
+
"learning_rate": 3.492688413948257e-05,
|
| 992 |
+
"loss": 0.3843,
|
| 993 |
+
"step": 3105
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 3.518560179977503,
|
| 997 |
+
"grad_norm": 0.6143022775650024,
|
| 998 |
+
"learning_rate": 3.518560179977503e-05,
|
| 999 |
+
"loss": 0.2909,
|
| 1000 |
+
"step": 3128
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 3.5444319460067493,
|
| 1004 |
+
"grad_norm": 0.13090083003044128,
|
| 1005 |
+
"learning_rate": 3.544431946006749e-05,
|
| 1006 |
+
"loss": 0.3688,
|
| 1007 |
+
"step": 3151
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 3.5703037120359955,
|
| 1011 |
+
"grad_norm": 8.838688850402832,
|
| 1012 |
+
"learning_rate": 3.5703037120359954e-05,
|
| 1013 |
+
"loss": 0.4144,
|
| 1014 |
+
"step": 3174
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 3.5961754780652417,
|
| 1018 |
+
"grad_norm": 8.904288291931152,
|
| 1019 |
+
"learning_rate": 3.596175478065242e-05,
|
| 1020 |
+
"loss": 0.3711,
|
| 1021 |
+
"step": 3197
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 3.622047244094488,
|
| 1025 |
+
"grad_norm": 0.6677899360656738,
|
| 1026 |
+
"learning_rate": 3.622047244094489e-05,
|
| 1027 |
+
"loss": 0.5376,
|
| 1028 |
+
"step": 3220
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 3.6479190101237347,
|
| 1032 |
+
"grad_norm": 2.987168073654175,
|
| 1033 |
+
"learning_rate": 3.647919010123735e-05,
|
| 1034 |
+
"loss": 0.3642,
|
| 1035 |
+
"step": 3243
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 3.673790776152981,
|
| 1039 |
+
"grad_norm": 9.179577827453613,
|
| 1040 |
+
"learning_rate": 3.673790776152981e-05,
|
| 1041 |
+
"loss": 0.6663,
|
| 1042 |
+
"step": 3266
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 3.699662542182227,
|
| 1046 |
+
"grad_norm": 1.0456788539886475,
|
| 1047 |
+
"learning_rate": 3.699662542182227e-05,
|
| 1048 |
+
"loss": 0.478,
|
| 1049 |
+
"step": 3289
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 3.725534308211474,
|
| 1053 |
+
"grad_norm": 8.08246898651123,
|
| 1054 |
+
"learning_rate": 3.7255343082114735e-05,
|
| 1055 |
+
"loss": 0.4601,
|
| 1056 |
+
"step": 3312
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 3.75140607424072,
|
| 1060 |
+
"grad_norm": 8.33056926727295,
|
| 1061 |
+
"learning_rate": 3.7514060742407205e-05,
|
| 1062 |
+
"loss": 0.4343,
|
| 1063 |
+
"step": 3335
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 3.7772778402699663,
|
| 1067 |
+
"grad_norm": 2.994161367416382,
|
| 1068 |
+
"learning_rate": 3.777277840269967e-05,
|
| 1069 |
+
"loss": 0.3037,
|
| 1070 |
+
"step": 3358
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 3.8031496062992125,
|
| 1074 |
+
"grad_norm": 0.5124155282974243,
|
| 1075 |
+
"learning_rate": 3.8031496062992125e-05,
|
| 1076 |
+
"loss": 0.4441,
|
| 1077 |
+
"step": 3381
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 3.8290213723284587,
|
| 1081 |
+
"grad_norm": 9.554545402526855,
|
| 1082 |
+
"learning_rate": 3.829021372328459e-05,
|
| 1083 |
+
"loss": 0.4561,
|
| 1084 |
+
"step": 3404
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 3.8548931383577054,
|
| 1088 |
+
"grad_norm": 2.4750254154205322,
|
| 1089 |
+
"learning_rate": 3.854893138357705e-05,
|
| 1090 |
+
"loss": 0.5598,
|
| 1091 |
+
"step": 3427
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 3.8807649043869517,
|
| 1095 |
+
"grad_norm": 0.18071837723255157,
|
| 1096 |
+
"learning_rate": 3.880764904386952e-05,
|
| 1097 |
+
"loss": 0.392,
|
| 1098 |
+
"step": 3450
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 3.906636670416198,
|
| 1102 |
+
"grad_norm": 0.9068869948387146,
|
| 1103 |
+
"learning_rate": 3.9066366704161985e-05,
|
| 1104 |
+
"loss": 0.4358,
|
| 1105 |
+
"step": 3473
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 3.9325084364454446,
|
| 1109 |
+
"grad_norm": 2.832122325897217,
|
| 1110 |
+
"learning_rate": 3.932508436445445e-05,
|
| 1111 |
+
"loss": 0.3503,
|
| 1112 |
+
"step": 3496
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 3.958380202474691,
|
| 1116 |
+
"grad_norm": 8.372138023376465,
|
| 1117 |
+
"learning_rate": 3.9583802024746905e-05,
|
| 1118 |
+
"loss": 0.3665,
|
| 1119 |
+
"step": 3519
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 3.984251968503937,
|
| 1123 |
+
"grad_norm": 0.4025736451148987,
|
| 1124 |
+
"learning_rate": 3.984251968503937e-05,
|
| 1125 |
+
"loss": 0.3274,
|
| 1126 |
+
"step": 3542
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 4.0,
|
| 1130 |
+
"eval_accuracy": 0.7815126050420168,
|
| 1131 |
+
"eval_auc": 0.8276973891160743,
|
| 1132 |
+
"eval_f1": 0.08771929824561403,
|
| 1133 |
+
"eval_loss": 0.6576350927352905,
|
| 1134 |
+
"eval_precision": 0.4166666666666667,
|
| 1135 |
+
"eval_recall": 0.049019607843137254,
|
| 1136 |
+
"eval_runtime": 2.7248,
|
| 1137 |
+
"eval_samples_per_second": 174.693,
|
| 1138 |
+
"eval_steps_per_second": 22.02,
|
| 1139 |
+
"step": 3556
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 4.010123734533184,
|
| 1143 |
+
"grad_norm": 11.520211219787598,
|
| 1144 |
+
"learning_rate": 4.010123734533184e-05,
|
| 1145 |
+
"loss": 0.3699,
|
| 1146 |
+
"step": 3565
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 4.0359955005624295,
|
| 1150 |
+
"grad_norm": 3.6669209003448486,
|
| 1151 |
+
"learning_rate": 4.03599550056243e-05,
|
| 1152 |
+
"loss": 0.4811,
|
| 1153 |
+
"step": 3588
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 4.061867266591676,
|
| 1157 |
+
"grad_norm": 6.646744251251221,
|
| 1158 |
+
"learning_rate": 4.0618672665916766e-05,
|
| 1159 |
+
"loss": 0.3609,
|
| 1160 |
+
"step": 3611
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 4.087739032620922,
|
| 1164 |
+
"grad_norm": 0.03924195468425751,
|
| 1165 |
+
"learning_rate": 4.087739032620922e-05,
|
| 1166 |
+
"loss": 0.3344,
|
| 1167 |
+
"step": 3634
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 4.113610798650169,
|
| 1171 |
+
"grad_norm": 3.2102909088134766,
|
| 1172 |
+
"learning_rate": 4.1136107986501686e-05,
|
| 1173 |
+
"loss": 0.6133,
|
| 1174 |
+
"step": 3657
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 4.139482564679415,
|
| 1178 |
+
"grad_norm": 8.028974533081055,
|
| 1179 |
+
"learning_rate": 4.1394825646794156e-05,
|
| 1180 |
+
"loss": 0.322,
|
| 1181 |
+
"step": 3680
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 4.165354330708661,
|
| 1185 |
+
"grad_norm": 0.7968009114265442,
|
| 1186 |
+
"learning_rate": 4.165354330708662e-05,
|
| 1187 |
+
"loss": 0.337,
|
| 1188 |
+
"step": 3703
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 4.191226096737908,
|
| 1192 |
+
"grad_norm": 8.355441093444824,
|
| 1193 |
+
"learning_rate": 4.191226096737908e-05,
|
| 1194 |
+
"loss": 0.3879,
|
| 1195 |
+
"step": 3726
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 4.2170978627671545,
|
| 1199 |
+
"grad_norm": 0.24170738458633423,
|
| 1200 |
+
"learning_rate": 4.217097862767154e-05,
|
| 1201 |
+
"loss": 0.3252,
|
| 1202 |
+
"step": 3749
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 4.2429696287964,
|
| 1206 |
+
"grad_norm": 5.390313625335693,
|
| 1207 |
+
"learning_rate": 4.2429696287964e-05,
|
| 1208 |
+
"loss": 0.507,
|
| 1209 |
+
"step": 3772
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 4.268841394825647,
|
| 1213 |
+
"grad_norm": 5.854102611541748,
|
| 1214 |
+
"learning_rate": 4.268841394825647e-05,
|
| 1215 |
+
"loss": 0.3849,
|
| 1216 |
+
"step": 3795
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 4.294713160854893,
|
| 1220 |
+
"grad_norm": 8.469975471496582,
|
| 1221 |
+
"learning_rate": 4.2947131608548936e-05,
|
| 1222 |
+
"loss": 0.315,
|
| 1223 |
+
"step": 3818
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 4.320584926884139,
|
| 1227 |
+
"grad_norm": 3.3237178325653076,
|
| 1228 |
+
"learning_rate": 4.32058492688414e-05,
|
| 1229 |
+
"loss": 0.7087,
|
| 1230 |
+
"step": 3841
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 4.346456692913386,
|
| 1234 |
+
"grad_norm": 1.0640898942947388,
|
| 1235 |
+
"learning_rate": 4.346456692913386e-05,
|
| 1236 |
+
"loss": 0.4298,
|
| 1237 |
+
"step": 3864
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 4.372328458942632,
|
| 1241 |
+
"grad_norm": 0.4618189334869385,
|
| 1242 |
+
"learning_rate": 4.372328458942632e-05,
|
| 1243 |
+
"loss": 0.3469,
|
| 1244 |
+
"step": 3887
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 4.398200224971879,
|
| 1248 |
+
"grad_norm": 0.6110893487930298,
|
| 1249 |
+
"learning_rate": 4.398200224971879e-05,
|
| 1250 |
+
"loss": 0.4085,
|
| 1251 |
+
"step": 3910
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 4.424071991001125,
|
| 1255 |
+
"grad_norm": 8.723979949951172,
|
| 1256 |
+
"learning_rate": 4.424071991001125e-05,
|
| 1257 |
+
"loss": 0.263,
|
| 1258 |
+
"step": 3933
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 4.449943757030371,
|
| 1262 |
+
"grad_norm": 19.94253158569336,
|
| 1263 |
+
"learning_rate": 4.4499437570303716e-05,
|
| 1264 |
+
"loss": 0.3318,
|
| 1265 |
+
"step": 3956
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 4.475815523059618,
|
| 1269 |
+
"grad_norm": 0.9224587082862854,
|
| 1270 |
+
"learning_rate": 4.475815523059618e-05,
|
| 1271 |
+
"loss": 0.462,
|
| 1272 |
+
"step": 3979
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 4.5016872890888635,
|
| 1276 |
+
"grad_norm": 9.54886531829834,
|
| 1277 |
+
"learning_rate": 4.5016872890888636e-05,
|
| 1278 |
+
"loss": 0.4471,
|
| 1279 |
+
"step": 4002
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 4.52755905511811,
|
| 1283 |
+
"grad_norm": 3.35776686668396,
|
| 1284 |
+
"learning_rate": 4.52755905511811e-05,
|
| 1285 |
+
"loss": 0.4742,
|
| 1286 |
+
"step": 4025
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 4.553430821147357,
|
| 1290 |
+
"grad_norm": 7.887851715087891,
|
| 1291 |
+
"learning_rate": 4.553430821147357e-05,
|
| 1292 |
+
"loss": 0.4714,
|
| 1293 |
+
"step": 4048
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 4.579302587176603,
|
| 1297 |
+
"grad_norm": 3.327909469604492,
|
| 1298 |
+
"learning_rate": 4.5793025871766033e-05,
|
| 1299 |
+
"loss": 0.6452,
|
| 1300 |
+
"step": 4071
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 4.605174353205849,
|
| 1304 |
+
"grad_norm": 1.3243390321731567,
|
| 1305 |
+
"learning_rate": 4.60517435320585e-05,
|
| 1306 |
+
"loss": 0.3745,
|
| 1307 |
+
"step": 4094
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 4.631046119235096,
|
| 1311 |
+
"grad_norm": 0.7224089503288269,
|
| 1312 |
+
"learning_rate": 4.631046119235096e-05,
|
| 1313 |
+
"loss": 0.2628,
|
| 1314 |
+
"step": 4117
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 4.656917885264342,
|
| 1318 |
+
"grad_norm": 0.017284361645579338,
|
| 1319 |
+
"learning_rate": 4.656917885264342e-05,
|
| 1320 |
+
"loss": 0.317,
|
| 1321 |
+
"step": 4140
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 4.6827896512935885,
|
| 1325 |
+
"grad_norm": 14.2166166305542,
|
| 1326 |
+
"learning_rate": 4.682789651293589e-05,
|
| 1327 |
+
"loss": 0.464,
|
| 1328 |
+
"step": 4163
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 4.708661417322834,
|
| 1332 |
+
"grad_norm": 0.49798861145973206,
|
| 1333 |
+
"learning_rate": 4.708661417322835e-05,
|
| 1334 |
+
"loss": 0.3336,
|
| 1335 |
+
"step": 4186
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 4.734533183352081,
|
| 1339 |
+
"grad_norm": 1.5816570520401,
|
| 1340 |
+
"learning_rate": 4.7345331833520814e-05,
|
| 1341 |
+
"loss": 0.423,
|
| 1342 |
+
"step": 4209
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 4.760404949381328,
|
| 1346 |
+
"grad_norm": 4.431649208068848,
|
| 1347 |
+
"learning_rate": 4.760404949381328e-05,
|
| 1348 |
+
"loss": 0.4474,
|
| 1349 |
+
"step": 4232
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 4.786276715410573,
|
| 1353 |
+
"grad_norm": 7.487512588500977,
|
| 1354 |
+
"learning_rate": 4.7862767154105734e-05,
|
| 1355 |
+
"loss": 0.32,
|
| 1356 |
+
"step": 4255
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 4.81214848143982,
|
| 1360 |
+
"grad_norm": 3.283292293548584,
|
| 1361 |
+
"learning_rate": 4.8121484814398204e-05,
|
| 1362 |
+
"loss": 0.4767,
|
| 1363 |
+
"step": 4278
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 4.838020247469066,
|
| 1367 |
+
"grad_norm": 7.209409236907959,
|
| 1368 |
+
"learning_rate": 4.838020247469067e-05,
|
| 1369 |
+
"loss": 0.2593,
|
| 1370 |
+
"step": 4301
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 4.863892013498313,
|
| 1374 |
+
"grad_norm": 7.526238441467285,
|
| 1375 |
+
"learning_rate": 4.863892013498313e-05,
|
| 1376 |
+
"loss": 0.3552,
|
| 1377 |
+
"step": 4324
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 4.889763779527559,
|
| 1381 |
+
"grad_norm": 0.18729792535305023,
|
| 1382 |
+
"learning_rate": 4.8897637795275594e-05,
|
| 1383 |
+
"loss": 0.3581,
|
| 1384 |
+
"step": 4347
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 4.915635545556805,
|
| 1388 |
+
"grad_norm": 0.21440748870372772,
|
| 1389 |
+
"learning_rate": 4.915635545556805e-05,
|
| 1390 |
+
"loss": 0.2921,
|
| 1391 |
+
"step": 4370
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 4.941507311586052,
|
| 1395 |
+
"grad_norm": 12.171503067016602,
|
| 1396 |
+
"learning_rate": 4.941507311586052e-05,
|
| 1397 |
+
"loss": 0.5248,
|
| 1398 |
+
"step": 4393
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 4.967379077615298,
|
| 1402 |
+
"grad_norm": 0.14398285746574402,
|
| 1403 |
+
"learning_rate": 4.9673790776152984e-05,
|
| 1404 |
+
"loss": 0.4253,
|
| 1405 |
+
"step": 4416
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 4.993250843644544,
|
| 1409 |
+
"grad_norm": 2.316239595413208,
|
| 1410 |
+
"learning_rate": 4.993250843644545e-05,
|
| 1411 |
+
"loss": 0.4106,
|
| 1412 |
+
"step": 4439
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 5.0,
|
| 1416 |
+
"eval_accuracy": 0.7815126050420168,
|
| 1417 |
+
"eval_auc": 0.8033579742057251,
|
| 1418 |
+
"eval_f1": 0.0,
|
| 1419 |
+
"eval_loss": 0.50748610496521,
|
| 1420 |
+
"eval_precision": 0.0,
|
| 1421 |
+
"eval_recall": 0.0,
|
| 1422 |
+
"eval_runtime": 2.7592,
|
| 1423 |
+
"eval_samples_per_second": 172.516,
|
| 1424 |
+
"eval_steps_per_second": 21.746,
|
| 1425 |
+
"step": 4445
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 5.019122609673791,
|
| 1429 |
+
"grad_norm": 4.349539279937744,
|
| 1430 |
+
"learning_rate": 4.997875265591801e-05,
|
| 1431 |
+
"loss": 0.3046,
|
| 1432 |
+
"step": 4462
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 5.0449943757030375,
|
| 1436 |
+
"grad_norm": 0.3448341190814972,
|
| 1437 |
+
"learning_rate": 4.995000624921885e-05,
|
| 1438 |
+
"loss": 0.3477,
|
| 1439 |
+
"step": 4485
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 5.070866141732283,
|
| 1443 |
+
"grad_norm": 0.6105608940124512,
|
| 1444 |
+
"learning_rate": 4.9921259842519685e-05,
|
| 1445 |
+
"loss": 0.5053,
|
| 1446 |
+
"step": 4508
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 5.09673790776153,
|
| 1450 |
+
"grad_norm": 8.691744804382324,
|
| 1451 |
+
"learning_rate": 4.989251343582053e-05,
|
| 1452 |
+
"loss": 0.3679,
|
| 1453 |
+
"step": 4531
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 5.122609673790776,
|
| 1457 |
+
"grad_norm": 2.42826771736145,
|
| 1458 |
+
"learning_rate": 4.9863767029121364e-05,
|
| 1459 |
+
"loss": 0.2206,
|
| 1460 |
+
"step": 4554
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 5.1484814398200225,
|
| 1464 |
+
"grad_norm": 2.756155490875244,
|
| 1465 |
+
"learning_rate": 4.98350206224222e-05,
|
| 1466 |
+
"loss": 0.6075,
|
| 1467 |
+
"step": 4577
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 5.174353205849269,
|
| 1471 |
+
"grad_norm": 0.6143556833267212,
|
| 1472 |
+
"learning_rate": 4.9806274215723035e-05,
|
| 1473 |
+
"loss": 0.2272,
|
| 1474 |
+
"step": 4600
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 5.200224971878515,
|
| 1478 |
+
"grad_norm": 0.17892147600650787,
|
| 1479 |
+
"learning_rate": 4.977752780902387e-05,
|
| 1480 |
+
"loss": 0.5256,
|
| 1481 |
+
"step": 4623
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 5.226096737907762,
|
| 1485 |
+
"grad_norm": 0.12368874251842499,
|
| 1486 |
+
"learning_rate": 4.9748781402324714e-05,
|
| 1487 |
+
"loss": 0.5569,
|
| 1488 |
+
"step": 4646
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 5.251968503937007,
|
| 1492 |
+
"grad_norm": 0.048263806849718094,
|
| 1493 |
+
"learning_rate": 4.972003499562555e-05,
|
| 1494 |
+
"loss": 0.2298,
|
| 1495 |
+
"step": 4669
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 5.277840269966254,
|
| 1499 |
+
"grad_norm": 4.790956020355225,
|
| 1500 |
+
"learning_rate": 4.9691288588926386e-05,
|
| 1501 |
+
"loss": 0.5415,
|
| 1502 |
+
"step": 4692
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 5.303712035995501,
|
| 1506 |
+
"grad_norm": 5.684449672698975,
|
| 1507 |
+
"learning_rate": 4.966254218222722e-05,
|
| 1508 |
+
"loss": 0.3086,
|
| 1509 |
+
"step": 4715
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 5.329583802024747,
|
| 1513 |
+
"grad_norm": 0.4559657871723175,
|
| 1514 |
+
"learning_rate": 4.9633795775528064e-05,
|
| 1515 |
+
"loss": 0.3575,
|
| 1516 |
+
"step": 4738
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 5.355455568053993,
|
| 1520 |
+
"grad_norm": 7.768259048461914,
|
| 1521 |
+
"learning_rate": 4.96050493688289e-05,
|
| 1522 |
+
"loss": 0.4601,
|
| 1523 |
+
"step": 4761
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 5.38132733408324,
|
| 1527 |
+
"grad_norm": 0.44012290239334106,
|
| 1528 |
+
"learning_rate": 4.9576302962129736e-05,
|
| 1529 |
+
"loss": 0.3265,
|
| 1530 |
+
"step": 4784
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 5.407199100112486,
|
| 1534 |
+
"grad_norm": 0.22265122830867767,
|
| 1535 |
+
"learning_rate": 4.954755655543057e-05,
|
| 1536 |
+
"loss": 0.6152,
|
| 1537 |
+
"step": 4807
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 5.433070866141732,
|
| 1541 |
+
"grad_norm": 4.463795185089111,
|
| 1542 |
+
"learning_rate": 4.951881014873141e-05,
|
| 1543 |
+
"loss": 0.3046,
|
| 1544 |
+
"step": 4830
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 5.458942632170979,
|
| 1548 |
+
"grad_norm": 5.145118236541748,
|
| 1549 |
+
"learning_rate": 4.949006374203225e-05,
|
| 1550 |
+
"loss": 0.4773,
|
| 1551 |
+
"step": 4853
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 5.484814398200225,
|
| 1555 |
+
"grad_norm": 0.24177011847496033,
|
| 1556 |
+
"learning_rate": 4.9461317335333087e-05,
|
| 1557 |
+
"loss": 0.2634,
|
| 1558 |
+
"step": 4876
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 5.5106861642294716,
|
| 1562 |
+
"grad_norm": 0.8902063369750977,
|
| 1563 |
+
"learning_rate": 4.943257092863392e-05,
|
| 1564 |
+
"loss": 0.3362,
|
| 1565 |
+
"step": 4899
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 5.536557930258717,
|
| 1569 |
+
"grad_norm": 0.39589497447013855,
|
| 1570 |
+
"learning_rate": 4.940382452193476e-05,
|
| 1571 |
+
"loss": 0.6168,
|
| 1572 |
+
"step": 4922
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 5.562429696287964,
|
| 1576 |
+
"grad_norm": 1.736325979232788,
|
| 1577 |
+
"learning_rate": 4.9375078115235594e-05,
|
| 1578 |
+
"loss": 0.5466,
|
| 1579 |
+
"step": 4945
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 5.588301462317211,
|
| 1583 |
+
"grad_norm": 2.3112049102783203,
|
| 1584 |
+
"learning_rate": 4.934633170853644e-05,
|
| 1585 |
+
"loss": 0.3546,
|
| 1586 |
+
"step": 4968
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 5.6141732283464565,
|
| 1590 |
+
"grad_norm": 0.4776885509490967,
|
| 1591 |
+
"learning_rate": 4.931758530183727e-05,
|
| 1592 |
+
"loss": 0.3755,
|
| 1593 |
+
"step": 4991
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 5.640044994375703,
|
| 1597 |
+
"grad_norm": 0.09375002235174179,
|
| 1598 |
+
"learning_rate": 4.928883889513811e-05,
|
| 1599 |
+
"loss": 0.2044,
|
| 1600 |
+
"step": 5014
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 5.665916760404949,
|
| 1604 |
+
"grad_norm": 11.319817543029785,
|
| 1605 |
+
"learning_rate": 4.9260092488438945e-05,
|
| 1606 |
+
"loss": 0.5937,
|
| 1607 |
+
"step": 5037
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 5.691788526434196,
|
| 1611 |
+
"grad_norm": 1.1792347431182861,
|
| 1612 |
+
"learning_rate": 4.923134608173979e-05,
|
| 1613 |
+
"loss": 0.3856,
|
| 1614 |
+
"step": 5060
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 5.717660292463442,
|
| 1618 |
+
"grad_norm": 6.615360736846924,
|
| 1619 |
+
"learning_rate": 4.920259967504062e-05,
|
| 1620 |
+
"loss": 0.3669,
|
| 1621 |
+
"step": 5083
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"epoch": 5.743532058492688,
|
| 1625 |
+
"grad_norm": 6.705723762512207,
|
| 1626 |
+
"learning_rate": 4.917385326834146e-05,
|
| 1627 |
+
"loss": 0.2221,
|
| 1628 |
+
"step": 5106
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 5.769403824521935,
|
| 1632 |
+
"grad_norm": 2.312495470046997,
|
| 1633 |
+
"learning_rate": 4.9145106861642295e-05,
|
| 1634 |
+
"loss": 0.4308,
|
| 1635 |
+
"step": 5129
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 5.7952755905511815,
|
| 1639 |
+
"grad_norm": 11.482613563537598,
|
| 1640 |
+
"learning_rate": 4.911636045494313e-05,
|
| 1641 |
+
"loss": 0.3644,
|
| 1642 |
+
"step": 5152
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 5.821147356580427,
|
| 1646 |
+
"grad_norm": 4.5011982917785645,
|
| 1647 |
+
"learning_rate": 4.9087614048243973e-05,
|
| 1648 |
+
"loss": 0.5318,
|
| 1649 |
+
"step": 5175
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 5.847019122609674,
|
| 1653 |
+
"grad_norm": 3.1337974071502686,
|
| 1654 |
+
"learning_rate": 4.905886764154481e-05,
|
| 1655 |
+
"loss": 0.4274,
|
| 1656 |
+
"step": 5198
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 5.872890888638921,
|
| 1660 |
+
"grad_norm": 0.16391399502754211,
|
| 1661 |
+
"learning_rate": 4.9030121234845645e-05,
|
| 1662 |
+
"loss": 0.3304,
|
| 1663 |
+
"step": 5221
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 5.898762654668166,
|
| 1667 |
+
"grad_norm": 0.630847692489624,
|
| 1668 |
+
"learning_rate": 4.900137482814649e-05,
|
| 1669 |
+
"loss": 0.4878,
|
| 1670 |
+
"step": 5244
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 5.924634420697413,
|
| 1674 |
+
"grad_norm": 3.9900641441345215,
|
| 1675 |
+
"learning_rate": 4.897262842144732e-05,
|
| 1676 |
+
"loss": 0.3846,
|
| 1677 |
+
"step": 5267
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 5.950506186726659,
|
| 1681 |
+
"grad_norm": 5.799670219421387,
|
| 1682 |
+
"learning_rate": 4.894388201474816e-05,
|
| 1683 |
+
"loss": 0.3469,
|
| 1684 |
+
"step": 5290
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 5.9763779527559056,
|
| 1688 |
+
"grad_norm": 0.49845901131629944,
|
| 1689 |
+
"learning_rate": 4.8915135608048996e-05,
|
| 1690 |
+
"loss": 0.3733,
|
| 1691 |
+
"step": 5313
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 6.0,
|
| 1695 |
+
"eval_accuracy": 0.7857142857142857,
|
| 1696 |
+
"eval_auc": 0.7628709237705779,
|
| 1697 |
+
"eval_f1": 0.0,
|
| 1698 |
+
"eval_loss": 0.6193206310272217,
|
| 1699 |
+
"eval_precision": 0.0,
|
| 1700 |
+
"eval_recall": 0.0,
|
| 1701 |
+
"eval_runtime": 2.7283,
|
| 1702 |
+
"eval_samples_per_second": 174.467,
|
| 1703 |
+
"eval_steps_per_second": 21.992,
|
| 1704 |
+
"step": 5334
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 6.002249718785152,
|
| 1708 |
+
"grad_norm": 0.5700758695602417,
|
| 1709 |
+
"learning_rate": 4.888638920134984e-05,
|
| 1710 |
+
"loss": 0.3025,
|
| 1711 |
+
"step": 5336
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 6.028121484814398,
|
| 1715 |
+
"grad_norm": 0.3606770932674408,
|
| 1716 |
+
"learning_rate": 4.8857642794650674e-05,
|
| 1717 |
+
"loss": 0.383,
|
| 1718 |
+
"step": 5359
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 6.053993250843645,
|
| 1722 |
+
"grad_norm": 0.136641263961792,
|
| 1723 |
+
"learning_rate": 4.88288963879515e-05,
|
| 1724 |
+
"loss": 0.2687,
|
| 1725 |
+
"step": 5382
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 6.0798650168728905,
|
| 1729 |
+
"grad_norm": 0.3220977187156677,
|
| 1730 |
+
"learning_rate": 4.8800149981252346e-05,
|
| 1731 |
+
"loss": 0.4189,
|
| 1732 |
+
"step": 5405
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 6.105736782902137,
|
| 1736 |
+
"grad_norm": 0.8333466053009033,
|
| 1737 |
+
"learning_rate": 4.877140357455318e-05,
|
| 1738 |
+
"loss": 0.3677,
|
| 1739 |
+
"step": 5428
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 6.131608548931384,
|
| 1743 |
+
"grad_norm": 0.08014432340860367,
|
| 1744 |
+
"learning_rate": 4.8742657167854025e-05,
|
| 1745 |
+
"loss": 0.3624,
|
| 1746 |
+
"step": 5451
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 6.15748031496063,
|
| 1750 |
+
"grad_norm": 8.192437171936035,
|
| 1751 |
+
"learning_rate": 4.871391076115486e-05,
|
| 1752 |
+
"loss": 0.4036,
|
| 1753 |
+
"step": 5474
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 6.183352080989876,
|
| 1757 |
+
"grad_norm": 6.411347389221191,
|
| 1758 |
+
"learning_rate": 4.8685164354455696e-05,
|
| 1759 |
+
"loss": 0.3529,
|
| 1760 |
+
"step": 5497
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 6.209223847019123,
|
| 1764 |
+
"grad_norm": 0.006956384517252445,
|
| 1765 |
+
"learning_rate": 4.865641794775653e-05,
|
| 1766 |
+
"loss": 0.1559,
|
| 1767 |
+
"step": 5520
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 6.235095613048369,
|
| 1771 |
+
"grad_norm": 0.1893380731344223,
|
| 1772 |
+
"learning_rate": 4.862767154105737e-05,
|
| 1773 |
+
"loss": 0.5716,
|
| 1774 |
+
"step": 5543
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 6.2609673790776155,
|
| 1778 |
+
"grad_norm": 0.47687458992004395,
|
| 1779 |
+
"learning_rate": 4.859892513435821e-05,
|
| 1780 |
+
"loss": 0.399,
|
| 1781 |
+
"step": 5566
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 6.286839145106861,
|
| 1785 |
+
"grad_norm": 4.109427452087402,
|
| 1786 |
+
"learning_rate": 4.857017872765905e-05,
|
| 1787 |
+
"loss": 0.3858,
|
| 1788 |
+
"step": 5589
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 6.312710911136108,
|
| 1792 |
+
"grad_norm": 0.6130828857421875,
|
| 1793 |
+
"learning_rate": 4.854143232095988e-05,
|
| 1794 |
+
"loss": 0.3434,
|
| 1795 |
+
"step": 5612
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 6.338582677165355,
|
| 1799 |
+
"grad_norm": 2.936263084411621,
|
| 1800 |
+
"learning_rate": 4.851268591426072e-05,
|
| 1801 |
+
"loss": 0.7265,
|
| 1802 |
+
"step": 5635
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 6.3644544431946,
|
| 1806 |
+
"grad_norm": 0.4069950580596924,
|
| 1807 |
+
"learning_rate": 4.848393950756156e-05,
|
| 1808 |
+
"loss": 0.2387,
|
| 1809 |
+
"step": 5658
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 6.390326209223847,
|
| 1813 |
+
"grad_norm": 3.4261677265167236,
|
| 1814 |
+
"learning_rate": 4.84551931008624e-05,
|
| 1815 |
+
"loss": 0.4388,
|
| 1816 |
+
"step": 5681
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 6.416197975253094,
|
| 1820 |
+
"grad_norm": 4.583181381225586,
|
| 1821 |
+
"learning_rate": 4.842644669416323e-05,
|
| 1822 |
+
"loss": 0.321,
|
| 1823 |
+
"step": 5704
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 6.44206974128234,
|
| 1827 |
+
"grad_norm": 4.256218433380127,
|
| 1828 |
+
"learning_rate": 4.839770028746407e-05,
|
| 1829 |
+
"loss": 0.3261,
|
| 1830 |
+
"step": 5727
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 6.467941507311586,
|
| 1834 |
+
"grad_norm": 1.0955240726470947,
|
| 1835 |
+
"learning_rate": 4.8368953880764905e-05,
|
| 1836 |
+
"loss": 0.3753,
|
| 1837 |
+
"step": 5750
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 6.493813273340832,
|
| 1841 |
+
"grad_norm": 7.2450361251831055,
|
| 1842 |
+
"learning_rate": 4.834020747406575e-05,
|
| 1843 |
+
"loss": 0.2428,
|
| 1844 |
+
"step": 5773
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 6.519685039370079,
|
| 1848 |
+
"grad_norm": 5.181588172912598,
|
| 1849 |
+
"learning_rate": 4.831146106736658e-05,
|
| 1850 |
+
"loss": 0.3341,
|
| 1851 |
+
"step": 5796
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 6.545556805399325,
|
| 1855 |
+
"grad_norm": 2.217822790145874,
|
| 1856 |
+
"learning_rate": 4.828271466066742e-05,
|
| 1857 |
+
"loss": 0.4004,
|
| 1858 |
+
"step": 5819
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 6.571428571428571,
|
| 1862 |
+
"grad_norm": 0.3265409469604492,
|
| 1863 |
+
"learning_rate": 4.8253968253968255e-05,
|
| 1864 |
+
"loss": 0.4519,
|
| 1865 |
+
"step": 5842
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 6.597300337457818,
|
| 1869 |
+
"grad_norm": 3.4193084239959717,
|
| 1870 |
+
"learning_rate": 4.822522184726909e-05,
|
| 1871 |
+
"loss": 0.4347,
|
| 1872 |
+
"step": 5865
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 6.6231721034870645,
|
| 1876 |
+
"grad_norm": 3.1898536682128906,
|
| 1877 |
+
"learning_rate": 4.8196475440569934e-05,
|
| 1878 |
+
"loss": 0.2508,
|
| 1879 |
+
"step": 5888
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 6.64904386951631,
|
| 1883 |
+
"grad_norm": 5.070316314697266,
|
| 1884 |
+
"learning_rate": 4.816772903387077e-05,
|
| 1885 |
+
"loss": 0.4112,
|
| 1886 |
+
"step": 5911
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 6.674915635545557,
|
| 1890 |
+
"grad_norm": 0.23132477700710297,
|
| 1891 |
+
"learning_rate": 4.8138982627171605e-05,
|
| 1892 |
+
"loss": 0.2206,
|
| 1893 |
+
"step": 5934
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 6.700787401574803,
|
| 1897 |
+
"grad_norm": 4.237873554229736,
|
| 1898 |
+
"learning_rate": 4.811023622047244e-05,
|
| 1899 |
+
"loss": 0.5504,
|
| 1900 |
+
"step": 5957
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 6.7266591676040495,
|
| 1904 |
+
"grad_norm": 0.857448160648346,
|
| 1905 |
+
"learning_rate": 4.808148981377328e-05,
|
| 1906 |
+
"loss": 0.3815,
|
| 1907 |
+
"step": 5980
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 6.752530933633296,
|
| 1911 |
+
"grad_norm": 5.193442344665527,
|
| 1912 |
+
"learning_rate": 4.805274340707412e-05,
|
| 1913 |
+
"loss": 0.3129,
|
| 1914 |
+
"step": 6003
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 6.778402699662542,
|
| 1918 |
+
"grad_norm": 4.156959056854248,
|
| 1919 |
+
"learning_rate": 4.8023997000374956e-05,
|
| 1920 |
+
"loss": 0.4625,
|
| 1921 |
+
"step": 6026
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 6.804274465691789,
|
| 1925 |
+
"grad_norm": 0.4680914878845215,
|
| 1926 |
+
"learning_rate": 4.799525059367579e-05,
|
| 1927 |
+
"loss": 0.2069,
|
| 1928 |
+
"step": 6049
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 6.830146231721034,
|
| 1932 |
+
"grad_norm": 5.006599426269531,
|
| 1933 |
+
"learning_rate": 4.796650418697663e-05,
|
| 1934 |
+
"loss": 0.4084,
|
| 1935 |
+
"step": 6072
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 6.856017997750281,
|
| 1939 |
+
"grad_norm": 6.840053081512451,
|
| 1940 |
+
"learning_rate": 4.793775778027747e-05,
|
| 1941 |
+
"loss": 0.2994,
|
| 1942 |
+
"step": 6095
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 6.881889763779528,
|
| 1946 |
+
"grad_norm": 0.11373735964298248,
|
| 1947 |
+
"learning_rate": 4.7909011373578306e-05,
|
| 1948 |
+
"loss": 0.2934,
|
| 1949 |
+
"step": 6118
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 6.907761529808774,
|
| 1953 |
+
"grad_norm": 5.152923107147217,
|
| 1954 |
+
"learning_rate": 4.788026496687914e-05,
|
| 1955 |
+
"loss": 0.2687,
|
| 1956 |
+
"step": 6141
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 6.93363329583802,
|
| 1960 |
+
"grad_norm": 0.021206054836511612,
|
| 1961 |
+
"learning_rate": 4.785151856017998e-05,
|
| 1962 |
+
"loss": 0.3019,
|
| 1963 |
+
"step": 6164
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 6.959505061867267,
|
| 1967 |
+
"grad_norm": 0.06403312087059021,
|
| 1968 |
+
"learning_rate": 4.7822772153480814e-05,
|
| 1969 |
+
"loss": 0.2471,
|
| 1970 |
+
"step": 6187
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 6.985376827896513,
|
| 1974 |
+
"grad_norm": 3.1448378562927246,
|
| 1975 |
+
"learning_rate": 4.7794025746781657e-05,
|
| 1976 |
+
"loss": 0.6191,
|
| 1977 |
+
"step": 6210
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 7.0,
|
| 1981 |
+
"eval_accuracy": 0.819327731092437,
|
| 1982 |
+
"eval_auc": 0.8453785257418476,
|
| 1983 |
+
"eval_f1": 0.4625,
|
| 1984 |
+
"eval_loss": 0.3894089162349701,
|
| 1985 |
+
"eval_precision": 0.6379310344827587,
|
| 1986 |
+
"eval_recall": 0.3627450980392157,
|
| 1987 |
+
"eval_runtime": 2.7066,
|
| 1988 |
+
"eval_samples_per_second": 175.867,
|
| 1989 |
+
"eval_steps_per_second": 22.168,
|
| 1990 |
+
"step": 6223
|
| 1991 |
+
}
|
| 1992 |
+
],
|
| 1993 |
+
"logging_steps": 23,
|
| 1994 |
+
"max_steps": 44450,
|
| 1995 |
+
"num_input_tokens_seen": 0,
|
| 1996 |
+
"num_train_epochs": 50,
|
| 1997 |
+
"save_steps": 500,
|
| 1998 |
+
"stateful_callbacks": {
|
| 1999 |
+
"EarlyStoppingCallback": {
|
| 2000 |
+
"args": {
|
| 2001 |
+
"early_stopping_patience": 20,
|
| 2002 |
+
"early_stopping_threshold": 0.01
|
| 2003 |
+
},
|
| 2004 |
+
"attributes": {
|
| 2005 |
+
"early_stopping_patience_counter": 0
|
| 2006 |
+
}
|
| 2007 |
+
},
|
| 2008 |
+
"TrainerControl": {
|
| 2009 |
+
"args": {
|
| 2010 |
+
"should_epoch_stop": false,
|
| 2011 |
+
"should_evaluate": false,
|
| 2012 |
+
"should_log": false,
|
| 2013 |
+
"should_save": true,
|
| 2014 |
+
"should_training_stop": false
|
| 2015 |
+
},
|
| 2016 |
+
"attributes": {}
|
| 2017 |
+
}
|
| 2018 |
+
},
|
| 2019 |
+
"total_flos": 2.0174156175154545e+19,
|
| 2020 |
+
"train_batch_size": 4,
|
| 2021 |
+
"trial_name": null,
|
| 2022 |
+
"trial_params": null
|
| 2023 |
+
}
|
checkpoint-6223/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61b112c86bcc256ad7cbc11997c2eea749ce3d4918a385981c9c8091909d5b84
|
| 3 |
+
size 5176
|
config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-large-patch32-384",
|
| 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": 1024,
|
| 12 |
+
"id2label": {
|
| 13 |
+
"0": "no",
|
| 14 |
+
"1": "yes"
|
| 15 |
+
},
|
| 16 |
+
"image_size": 384,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 4096,
|
| 19 |
+
"label2id": {
|
| 20 |
+
"no": 0,
|
| 21 |
+
"yes": 1
|
| 22 |
+
},
|
| 23 |
+
"layer_norm_eps": 1e-12,
|
| 24 |
+
"model_type": "vit",
|
| 25 |
+
"num_attention_heads": 16,
|
| 26 |
+
"num_channels": 3,
|
| 27 |
+
"num_hidden_layers": 24,
|
| 28 |
+
"patch_size": 32,
|
| 29 |
+
"problem_type": "single_label_classification",
|
| 30 |
+
"qkv_bias": true,
|
| 31 |
+
"torch_dtype": "float32",
|
| 32 |
+
"transformers_version": "4.44.0"
|
| 33 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8c74ff4bcdac4d551a8a078b510f9615f76250ffa5bbe19830f82a09e570c7a6
|
| 3 |
+
size 1222485480
|
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": 384,
|
| 20 |
+
"width": 384
|
| 21 |
+
}
|
| 22 |
+
}
|
runs/Aug18_14-29-31_PcMasterCarrera/events.out.tfevents.1724005772.PcMasterCarrera.5764.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:afa51d0d5d5d563bbe5b5fc68870cd2f6e75f236564e76c56023858d71d9b996
|
| 3 |
+
size 240762
|
runs/Aug18_14-29-31_PcMasterCarrera/events.out.tfevents.1724008308.PcMasterCarrera.5764.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9262b50ef35817e0467dc736ac35a40d38f97eb75e65ac6916cdd82c08aa1819
|
| 3 |
+
size 617
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61b112c86bcc256ad7cbc11997c2eea749ce3d4918a385981c9c8091909d5b84
|
| 3 |
+
size 5176
|
training_params.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "prostate-mri-T2w-v05/autotrain-data",
|
| 3 |
+
"model": "google/vit-large-patch32-384",
|
| 4 |
+
"username": "Emilio407",
|
| 5 |
+
"lr": 5e-05,
|
| 6 |
+
"epochs": 50,
|
| 7 |
+
"batch_size": 4,
|
| 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": "prostate-mri-T2w-v05",
|
| 19 |
+
"auto_find_batch_size": false,
|
| 20 |
+
"mixed_precision": "bf16",
|
| 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": 20,
|
| 28 |
+
"early_stopping_threshold": 0.01
|
| 29 |
+
}
|