Hugging Face's logo

JLB-JLB
/
seizure_vit_jlb_231108_iir_adjusted

Image Classification
Transformers
TensorBoard
Safetensors
vit
seizure-detection
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
JLB-JLB commited on
Commit
3309150
·
1 Parent(s): 64c70bf

VIT_SEIZURE_231108

Browse files
Files changed (6) hide show
  1. README.md +5 -3
  2. all_results.json +13 -0
  3. eval_results.json +8 -0
  4. runs/Nov08_17-16-21_3ee6a63ee779/events.out.tfevents.1699478083.3ee6a63ee779 +3 -0
  5. train_results.json +8 -0
  6. trainer_state.json +4558 -0
README.md CHANGED
@@ -2,6 +2,8 @@
2
  license: apache-2.0
3
  base_model: google/vit-base-patch16-224-in21k
4
  tags:
 
 
5
  - generated_from_trainer
6
  model-index:
7
  - name: seizure_vit_jlb_231108_iir_adjusted
@@ -13,10 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
13
 
14
  # seizure_vit_jlb_231108_iir_adjusted
15
 
16
- This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on an unknown dataset.
17
  It achieves the following results on the evaluation set:
18
- - Loss: 0.8152
19
- - Roc Auc: 0.7587
20
 
21
  ## Model description
22
 
 
2
  license: apache-2.0
3
  base_model: google/vit-base-patch16-224-in21k
4
  tags:
5
+ - image-classification
6
+ - seizure-detection
7
  - generated_from_trainer
8
  model-index:
9
  - name: seizure_vit_jlb_231108_iir_adjusted
 
15
 
16
  # seizure_vit_jlb_231108_iir_adjusted
17
 
18
+ This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the JLB-JLB/seizure_eeg_iirFilter_greyscale_224x224_6secWindow_adjusted dataset.
19
  It achieves the following results on the evaluation set:
20
+ - Loss: 0.4198
21
+ - Roc Auc: 0.7773
22
 
23
  ## Model description
24
 
all_results.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "eval_loss": 0.4198167622089386,
4
+ "eval_roc_auc": 0.7772517014081914,
5
+ "eval_runtime": 5994.3569,
6
+ "eval_samples_per_second": 65.093,
7
+ "eval_steps_per_second": 8.137,
8
+ "total_flos": 3.6406511640831468e+19,
9
+ "train_loss": 0.21540352483164263,
10
+ "train_runtime": 8294.2837,
11
+ "train_samples_per_second": 56.643,
12
+ "train_steps_per_second": 1.77
13
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "eval_loss": 0.4198167622089386,
4
+ "eval_roc_auc": 0.7772517014081914,
5
+ "eval_runtime": 5994.3569,
6
+ "eval_samples_per_second": 65.093,
7
+ "eval_steps_per_second": 8.137
8
+ }
runs/Nov08_17-16-21_3ee6a63ee779/events.out.tfevents.1699478083.3ee6a63ee779 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6eea95a3d518570942fb4abac7ebc7f53e321f156d35e64955d3611851bf28c2
3
+ size 40
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "total_flos": 3.6406511640831468e+19,
4
+ "train_loss": 0.21540352483164263,
5
+ "train_runtime": 8294.2837,
6
+ "train_samples_per_second": 56.643,
7
+ "train_steps_per_second": 1.77
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,4558 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.47342389822006226,
3
+ "best_model_checkpoint": "seizure_vit/seizure_vit_jlb_231108_iir_adjusted/checkpoint-1000",
4
+ "epoch": 5.0,
5
+ "eval_steps": 1000,
6
+ "global_step": 14685,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01,
13
+ "learning_rate": 9.986380660537965e-06,
14
+ "loss": 0.6425,
15
+ "step": 20
16
+ },
17
+ {
18
+ "epoch": 0.01,
19
+ "learning_rate": 9.972761321075928e-06,
20
+ "loss": 0.5846,
21
+ "step": 40
22
+ },
23
+ {
24
+ "epoch": 0.02,
25
+ "learning_rate": 9.959141981613892e-06,
26
+ "loss": 0.549,
27
+ "step": 60
28
+ },
29
+ {
30
+ "epoch": 0.03,
31
+ "learning_rate": 9.945522642151857e-06,
32
+ "loss": 0.5503,
33
+ "step": 80
34
+ },
35
+ {
36
+ "epoch": 0.03,
37
+ "learning_rate": 9.931903302689821e-06,
38
+ "loss": 0.5304,
39
+ "step": 100
40
+ },
41
+ {
42
+ "epoch": 0.04,
43
+ "learning_rate": 9.918283963227783e-06,
44
+ "loss": 0.5064,
45
+ "step": 120
46
+ },
47
+ {
48
+ "epoch": 0.05,
49
+ "learning_rate": 9.904664623765749e-06,
50
+ "loss": 0.4953,
51
+ "step": 140
52
+ },
53
+ {
54
+ "epoch": 0.05,
55
+ "learning_rate": 9.891045284303712e-06,
56
+ "loss": 0.5028,
57
+ "step": 160
58
+ },
59
+ {
60
+ "epoch": 0.06,
61
+ "learning_rate": 9.877425944841676e-06,
62
+ "loss": 0.4742,
63
+ "step": 180
64
+ },
65
+ {
66
+ "epoch": 0.07,
67
+ "learning_rate": 9.86380660537964e-06,
68
+ "loss": 0.4921,
69
+ "step": 200
70
+ },
71
+ {
72
+ "epoch": 0.07,
73
+ "learning_rate": 9.850187265917604e-06,
74
+ "loss": 0.4961,
75
+ "step": 220
76
+ },
77
+ {
78
+ "epoch": 0.08,
79
+ "learning_rate": 9.836567926455567e-06,
80
+ "loss": 0.438,
81
+ "step": 240
82
+ },
83
+ {
84
+ "epoch": 0.09,
85
+ "learning_rate": 9.822948586993531e-06,
86
+ "loss": 0.4273,
87
+ "step": 260
88
+ },
89
+ {
90
+ "epoch": 0.1,
91
+ "learning_rate": 9.809329247531496e-06,
92
+ "loss": 0.4606,
93
+ "step": 280
94
+ },
95
+ {
96
+ "epoch": 0.1,
97
+ "learning_rate": 9.79570990806946e-06,
98
+ "loss": 0.4513,
99
+ "step": 300
100
+ },
101
+ {
102
+ "epoch": 0.11,
103
+ "learning_rate": 9.782090568607424e-06,
104
+ "loss": 0.4375,
105
+ "step": 320
106
+ },
107
+ {
108
+ "epoch": 0.12,
109
+ "learning_rate": 9.768471229145388e-06,
110
+ "loss": 0.4441,
111
+ "step": 340
112
+ },
113
+ {
114
+ "epoch": 0.12,
115
+ "learning_rate": 9.754851889683351e-06,
116
+ "loss": 0.4187,
117
+ "step": 360
118
+ },
119
+ {
120
+ "epoch": 0.13,
121
+ "learning_rate": 9.741232550221315e-06,
122
+ "loss": 0.4336,
123
+ "step": 380
124
+ },
125
+ {
126
+ "epoch": 0.14,
127
+ "learning_rate": 9.727613210759279e-06,
128
+ "loss": 0.4258,
129
+ "step": 400
130
+ },
131
+ {
132
+ "epoch": 0.14,
133
+ "learning_rate": 9.713993871297242e-06,
134
+ "loss": 0.4225,
135
+ "step": 420
136
+ },
137
+ {
138
+ "epoch": 0.15,
139
+ "learning_rate": 9.700374531835208e-06,
140
+ "loss": 0.4119,
141
+ "step": 440
142
+ },
143
+ {
144
+ "epoch": 0.16,
145
+ "learning_rate": 9.68675519237317e-06,
146
+ "loss": 0.4016,
147
+ "step": 460
148
+ },
149
+ {
150
+ "epoch": 0.16,
151
+ "learning_rate": 9.673135852911134e-06,
152
+ "loss": 0.43,
153
+ "step": 480
154
+ },
155
+ {
156
+ "epoch": 0.17,
157
+ "learning_rate": 9.659516513449099e-06,
158
+ "loss": 0.4397,
159
+ "step": 500
160
+ },
161
+ {
162
+ "epoch": 0.18,
163
+ "learning_rate": 9.645897173987063e-06,
164
+ "loss": 0.376,
165
+ "step": 520
166
+ },
167
+ {
168
+ "epoch": 0.18,
169
+ "learning_rate": 9.632277834525027e-06,
170
+ "loss": 0.4556,
171
+ "step": 540
172
+ },
173
+ {
174
+ "epoch": 0.19,
175
+ "learning_rate": 9.61865849506299e-06,
176
+ "loss": 0.4067,
177
+ "step": 560
178
+ },
179
+ {
180
+ "epoch": 0.2,
181
+ "learning_rate": 9.605039155600954e-06,
182
+ "loss": 0.4005,
183
+ "step": 580
184
+ },
185
+ {
186
+ "epoch": 0.2,
187
+ "learning_rate": 9.591419816138918e-06,
188
+ "loss": 0.3618,
189
+ "step": 600
190
+ },
191
+ {
192
+ "epoch": 0.21,
193
+ "learning_rate": 9.577800476676881e-06,
194
+ "loss": 0.4302,
195
+ "step": 620
196
+ },
197
+ {
198
+ "epoch": 0.22,
199
+ "learning_rate": 9.564181137214847e-06,
200
+ "loss": 0.4234,
201
+ "step": 640
202
+ },
203
+ {
204
+ "epoch": 0.22,
205
+ "learning_rate": 9.55056179775281e-06,
206
+ "loss": 0.4304,
207
+ "step": 660
208
+ },
209
+ {
210
+ "epoch": 0.23,
211
+ "learning_rate": 9.536942458290773e-06,
212
+ "loss": 0.3656,
213
+ "step": 680
214
+ },
215
+ {
216
+ "epoch": 0.24,
217
+ "learning_rate": 9.523323118828738e-06,
218
+ "loss": 0.3883,
219
+ "step": 700
220
+ },
221
+ {
222
+ "epoch": 0.25,
223
+ "learning_rate": 9.509703779366702e-06,
224
+ "loss": 0.3971,
225
+ "step": 720
226
+ },
227
+ {
228
+ "epoch": 0.25,
229
+ "learning_rate": 9.496084439904665e-06,
230
+ "loss": 0.4155,
231
+ "step": 740
232
+ },
233
+ {
234
+ "epoch": 0.26,
235
+ "learning_rate": 9.48246510044263e-06,
236
+ "loss": 0.3724,
237
+ "step": 760
238
+ },
239
+ {
240
+ "epoch": 0.27,
241
+ "learning_rate": 9.468845760980593e-06,
242
+ "loss": 0.4199,
243
+ "step": 780
244
+ },
245
+ {
246
+ "epoch": 0.27,
247
+ "learning_rate": 9.455226421518557e-06,
248
+ "loss": 0.3955,
249
+ "step": 800
250
+ },
251
+ {
252
+ "epoch": 0.28,
253
+ "learning_rate": 9.44160708205652e-06,
254
+ "loss": 0.3703,
255
+ "step": 820
256
+ },
257
+ {
258
+ "epoch": 0.29,
259
+ "learning_rate": 9.427987742594486e-06,
260
+ "loss": 0.3956,
261
+ "step": 840
262
+ },
263
+ {
264
+ "epoch": 0.29,
265
+ "learning_rate": 9.41436840313245e-06,
266
+ "loss": 0.362,
267
+ "step": 860
268
+ },
269
+ {
270
+ "epoch": 0.3,
271
+ "learning_rate": 9.400749063670412e-06,
272
+ "loss": 0.3908,
273
+ "step": 880
274
+ },
275
+ {
276
+ "epoch": 0.31,
277
+ "learning_rate": 9.387129724208377e-06,
278
+ "loss": 0.3674,
279
+ "step": 900
280
+ },
281
+ {
282
+ "epoch": 0.31,
283
+ "learning_rate": 9.37351038474634e-06,
284
+ "loss": 0.3814,
285
+ "step": 920
286
+ },
287
+ {
288
+ "epoch": 0.32,
289
+ "learning_rate": 9.359891045284304e-06,
290
+ "loss": 0.4162,
291
+ "step": 940
292
+ },
293
+ {
294
+ "epoch": 0.33,
295
+ "learning_rate": 9.346271705822268e-06,
296
+ "loss": 0.4027,
297
+ "step": 960
298
+ },
299
+ {
300
+ "epoch": 0.33,
301
+ "learning_rate": 9.332652366360232e-06,
302
+ "loss": 0.3573,
303
+ "step": 980
304
+ },
305
+ {
306
+ "epoch": 0.34,
307
+ "learning_rate": 9.319033026898197e-06,
308
+ "loss": 0.3803,
309
+ "step": 1000
310
+ },
311
+ {
312
+ "epoch": 0.34,
313
+ "eval_loss": 0.47342389822006226,
314
+ "eval_roc_auc": 0.7746478873239437,
315
+ "eval_runtime": 223.2127,
316
+ "eval_samples_per_second": 66.797,
317
+ "eval_steps_per_second": 8.351,
318
+ "step": 1000
319
+ },
320
+ {
321
+ "epoch": 0.35,
322
+ "learning_rate": 9.30541368743616e-06,
323
+ "loss": 0.3728,
324
+ "step": 1020
325
+ },
326
+ {
327
+ "epoch": 0.35,
328
+ "learning_rate": 9.291794347974123e-06,
329
+ "loss": 0.3511,
330
+ "step": 1040
331
+ },
332
+ {
333
+ "epoch": 0.36,
334
+ "learning_rate": 9.278175008512089e-06,
335
+ "loss": 0.3722,
336
+ "step": 1060
337
+ },
338
+ {
339
+ "epoch": 0.37,
340
+ "learning_rate": 9.264555669050052e-06,
341
+ "loss": 0.336,
342
+ "step": 1080
343
+ },
344
+ {
345
+ "epoch": 0.37,
346
+ "learning_rate": 9.250936329588016e-06,
347
+ "loss": 0.4234,
348
+ "step": 1100
349
+ },
350
+ {
351
+ "epoch": 0.38,
352
+ "learning_rate": 9.23731699012598e-06,
353
+ "loss": 0.3634,
354
+ "step": 1120
355
+ },
356
+ {
357
+ "epoch": 0.39,
358
+ "learning_rate": 9.223697650663943e-06,
359
+ "loss": 0.3378,
360
+ "step": 1140
361
+ },
362
+ {
363
+ "epoch": 0.39,
364
+ "learning_rate": 9.210078311201907e-06,
365
+ "loss": 0.3802,
366
+ "step": 1160
367
+ },
368
+ {
369
+ "epoch": 0.4,
370
+ "learning_rate": 9.196458971739871e-06,
371
+ "loss": 0.3998,
372
+ "step": 1180
373
+ },
374
+ {
375
+ "epoch": 0.41,
376
+ "learning_rate": 9.182839632277836e-06,
377
+ "loss": 0.3611,
378
+ "step": 1200
379
+ },
380
+ {
381
+ "epoch": 0.42,
382
+ "learning_rate": 9.169220292815798e-06,
383
+ "loss": 0.3884,
384
+ "step": 1220
385
+ },
386
+ {
387
+ "epoch": 0.42,
388
+ "learning_rate": 9.155600953353762e-06,
389
+ "loss": 0.3412,
390
+ "step": 1240
391
+ },
392
+ {
393
+ "epoch": 0.43,
394
+ "learning_rate": 9.141981613891727e-06,
395
+ "loss": 0.4026,
396
+ "step": 1260
397
+ },
398
+ {
399
+ "epoch": 0.44,
400
+ "learning_rate": 9.128362274429691e-06,
401
+ "loss": 0.3619,
402
+ "step": 1280
403
+ },
404
+ {
405
+ "epoch": 0.44,
406
+ "learning_rate": 9.114742934967655e-06,
407
+ "loss": 0.3291,
408
+ "step": 1300
409
+ },
410
+ {
411
+ "epoch": 0.45,
412
+ "learning_rate": 9.101123595505619e-06,
413
+ "loss": 0.3514,
414
+ "step": 1320
415
+ },
416
+ {
417
+ "epoch": 0.46,
418
+ "learning_rate": 9.087504256043582e-06,
419
+ "loss": 0.3353,
420
+ "step": 1340
421
+ },
422
+ {
423
+ "epoch": 0.46,
424
+ "learning_rate": 9.073884916581546e-06,
425
+ "loss": 0.3713,
426
+ "step": 1360
427
+ },
428
+ {
429
+ "epoch": 0.47,
430
+ "learning_rate": 9.06026557711951e-06,
431
+ "loss": 0.3571,
432
+ "step": 1380
433
+ },
434
+ {
435
+ "epoch": 0.48,
436
+ "learning_rate": 9.046646237657475e-06,
437
+ "loss": 0.3406,
438
+ "step": 1400
439
+ },
440
+ {
441
+ "epoch": 0.48,
442
+ "learning_rate": 9.033026898195439e-06,
443
+ "loss": 0.3785,
444
+ "step": 1420
445
+ },
446
+ {
447
+ "epoch": 0.49,
448
+ "learning_rate": 9.019407558733401e-06,
449
+ "loss": 0.3618,
450
+ "step": 1440
451
+ },
452
+ {
453
+ "epoch": 0.5,
454
+ "learning_rate": 9.005788219271366e-06,
455
+ "loss": 0.323,
456
+ "step": 1460
457
+ },
458
+ {
459
+ "epoch": 0.5,
460
+ "learning_rate": 8.99216887980933e-06,
461
+ "loss": 0.2908,
462
+ "step": 1480
463
+ },
464
+ {
465
+ "epoch": 0.51,
466
+ "learning_rate": 8.978549540347294e-06,
467
+ "loss": 0.372,
468
+ "step": 1500
469
+ },
470
+ {
471
+ "epoch": 0.52,
472
+ "learning_rate": 8.964930200885258e-06,
473
+ "loss": 0.3317,
474
+ "step": 1520
475
+ },
476
+ {
477
+ "epoch": 0.52,
478
+ "learning_rate": 8.951310861423221e-06,
479
+ "loss": 0.3953,
480
+ "step": 1540
481
+ },
482
+ {
483
+ "epoch": 0.53,
484
+ "learning_rate": 8.937691521961187e-06,
485
+ "loss": 0.3151,
486
+ "step": 1560
487
+ },
488
+ {
489
+ "epoch": 0.54,
490
+ "learning_rate": 8.924072182499149e-06,
491
+ "loss": 0.3391,
492
+ "step": 1580
493
+ },
494
+ {
495
+ "epoch": 0.54,
496
+ "learning_rate": 8.910452843037113e-06,
497
+ "loss": 0.3281,
498
+ "step": 1600
499
+ },
500
+ {
501
+ "epoch": 0.55,
502
+ "learning_rate": 8.896833503575078e-06,
503
+ "loss": 0.3613,
504
+ "step": 1620
505
+ },
506
+ {
507
+ "epoch": 0.56,
508
+ "learning_rate": 8.883214164113042e-06,
509
+ "loss": 0.3444,
510
+ "step": 1640
511
+ },
512
+ {
513
+ "epoch": 0.57,
514
+ "learning_rate": 8.869594824651005e-06,
515
+ "loss": 0.3122,
516
+ "step": 1660
517
+ },
518
+ {
519
+ "epoch": 0.57,
520
+ "learning_rate": 8.855975485188969e-06,
521
+ "loss": 0.2896,
522
+ "step": 1680
523
+ },
524
+ {
525
+ "epoch": 0.58,
526
+ "learning_rate": 8.842356145726933e-06,
527
+ "loss": 0.3603,
528
+ "step": 1700
529
+ },
530
+ {
531
+ "epoch": 0.59,
532
+ "learning_rate": 8.828736806264897e-06,
533
+ "loss": 0.3173,
534
+ "step": 1720
535
+ },
536
+ {
537
+ "epoch": 0.59,
538
+ "learning_rate": 8.81511746680286e-06,
539
+ "loss": 0.3016,
540
+ "step": 1740
541
+ },
542
+ {
543
+ "epoch": 0.6,
544
+ "learning_rate": 8.801498127340826e-06,
545
+ "loss": 0.3522,
546
+ "step": 1760
547
+ },
548
+ {
549
+ "epoch": 0.61,
550
+ "learning_rate": 8.787878787878788e-06,
551
+ "loss": 0.3482,
552
+ "step": 1780
553
+ },
554
+ {
555
+ "epoch": 0.61,
556
+ "learning_rate": 8.774259448416752e-06,
557
+ "loss": 0.2981,
558
+ "step": 1800
559
+ },
560
+ {
561
+ "epoch": 0.62,
562
+ "learning_rate": 8.760640108954717e-06,
563
+ "loss": 0.3467,
564
+ "step": 1820
565
+ },
566
+ {
567
+ "epoch": 0.63,
568
+ "learning_rate": 8.74702076949268e-06,
569
+ "loss": 0.373,
570
+ "step": 1840
571
+ },
572
+ {
573
+ "epoch": 0.63,
574
+ "learning_rate": 8.733401430030644e-06,
575
+ "loss": 0.3449,
576
+ "step": 1860
577
+ },
578
+ {
579
+ "epoch": 0.64,
580
+ "learning_rate": 8.719782090568608e-06,
581
+ "loss": 0.3541,
582
+ "step": 1880
583
+ },
584
+ {
585
+ "epoch": 0.65,
586
+ "learning_rate": 8.706162751106572e-06,
587
+ "loss": 0.3123,
588
+ "step": 1900
589
+ },
590
+ {
591
+ "epoch": 0.65,
592
+ "learning_rate": 8.692543411644536e-06,
593
+ "loss": 0.3103,
594
+ "step": 1920
595
+ },
596
+ {
597
+ "epoch": 0.66,
598
+ "learning_rate": 8.6789240721825e-06,
599
+ "loss": 0.3029,
600
+ "step": 1940
601
+ },
602
+ {
603
+ "epoch": 0.67,
604
+ "learning_rate": 8.665304732720465e-06,
605
+ "loss": 0.3077,
606
+ "step": 1960
607
+ },
608
+ {
609
+ "epoch": 0.67,
610
+ "learning_rate": 8.651685393258428e-06,
611
+ "loss": 0.3612,
612
+ "step": 1980
613
+ },
614
+ {
615
+ "epoch": 0.68,
616
+ "learning_rate": 8.63806605379639e-06,
617
+ "loss": 0.3456,
618
+ "step": 2000
619
+ },
620
+ {
621
+ "epoch": 0.68,
622
+ "eval_loss": 0.4863223433494568,
623
+ "eval_roc_auc": 0.7782025486250839,
624
+ "eval_runtime": 226.9387,
625
+ "eval_samples_per_second": 65.701,
626
+ "eval_steps_per_second": 8.214,
627
+ "step": 2000
628
+ },
629
+ {
630
+ "epoch": 0.69,
631
+ "learning_rate": 8.624446714334356e-06,
632
+ "loss": 0.308,
633
+ "step": 2020
634
+ },
635
+ {
636
+ "epoch": 0.69,
637
+ "learning_rate": 8.61082737487232e-06,
638
+ "loss": 0.314,
639
+ "step": 2040
640
+ },
641
+ {
642
+ "epoch": 0.7,
643
+ "learning_rate": 8.597208035410283e-06,
644
+ "loss": 0.3102,
645
+ "step": 2060
646
+ },
647
+ {
648
+ "epoch": 0.71,
649
+ "learning_rate": 8.583588695948247e-06,
650
+ "loss": 0.3176,
651
+ "step": 2080
652
+ },
653
+ {
654
+ "epoch": 0.72,
655
+ "learning_rate": 8.56996935648621e-06,
656
+ "loss": 0.3541,
657
+ "step": 2100
658
+ },
659
+ {
660
+ "epoch": 0.72,
661
+ "learning_rate": 8.556350017024175e-06,
662
+ "loss": 0.3057,
663
+ "step": 2120
664
+ },
665
+ {
666
+ "epoch": 0.73,
667
+ "learning_rate": 8.542730677562138e-06,
668
+ "loss": 0.2985,
669
+ "step": 2140
670
+ },
671
+ {
672
+ "epoch": 0.74,
673
+ "learning_rate": 8.529111338100104e-06,
674
+ "loss": 0.2799,
675
+ "step": 2160
676
+ },
677
+ {
678
+ "epoch": 0.74,
679
+ "learning_rate": 8.515491998638067e-06,
680
+ "loss": 0.3107,
681
+ "step": 2180
682
+ },
683
+ {
684
+ "epoch": 0.75,
685
+ "learning_rate": 8.501872659176031e-06,
686
+ "loss": 0.3512,
687
+ "step": 2200
688
+ },
689
+ {
690
+ "epoch": 0.76,
691
+ "learning_rate": 8.488253319713995e-06,
692
+ "loss": 0.311,
693
+ "step": 2220
694
+ },
695
+ {
696
+ "epoch": 0.76,
697
+ "learning_rate": 8.474633980251959e-06,
698
+ "loss": 0.3146,
699
+ "step": 2240
700
+ },
701
+ {
702
+ "epoch": 0.77,
703
+ "learning_rate": 8.461014640789922e-06,
704
+ "loss": 0.3202,
705
+ "step": 2260
706
+ },
707
+ {
708
+ "epoch": 0.78,
709
+ "learning_rate": 8.447395301327886e-06,
710
+ "loss": 0.3073,
711
+ "step": 2280
712
+ },
713
+ {
714
+ "epoch": 0.78,
715
+ "learning_rate": 8.43377596186585e-06,
716
+ "loss": 0.3144,
717
+ "step": 2300
718
+ },
719
+ {
720
+ "epoch": 0.79,
721
+ "learning_rate": 8.420156622403815e-06,
722
+ "loss": 0.3132,
723
+ "step": 2320
724
+ },
725
+ {
726
+ "epoch": 0.8,
727
+ "learning_rate": 8.406537282941777e-06,
728
+ "loss": 0.3162,
729
+ "step": 2340
730
+ },
731
+ {
732
+ "epoch": 0.8,
733
+ "learning_rate": 8.392917943479741e-06,
734
+ "loss": 0.3047,
735
+ "step": 2360
736
+ },
737
+ {
738
+ "epoch": 0.81,
739
+ "learning_rate": 8.379298604017706e-06,
740
+ "loss": 0.3362,
741
+ "step": 2380
742
+ },
743
+ {
744
+ "epoch": 0.82,
745
+ "learning_rate": 8.36567926455567e-06,
746
+ "loss": 0.3045,
747
+ "step": 2400
748
+ },
749
+ {
750
+ "epoch": 0.82,
751
+ "learning_rate": 8.352059925093634e-06,
752
+ "loss": 0.301,
753
+ "step": 2420
754
+ },
755
+ {
756
+ "epoch": 0.83,
757
+ "learning_rate": 8.338440585631598e-06,
758
+ "loss": 0.3196,
759
+ "step": 2440
760
+ },
761
+ {
762
+ "epoch": 0.84,
763
+ "learning_rate": 8.324821246169561e-06,
764
+ "loss": 0.3542,
765
+ "step": 2460
766
+ },
767
+ {
768
+ "epoch": 0.84,
769
+ "learning_rate": 8.311201906707525e-06,
770
+ "loss": 0.3279,
771
+ "step": 2480
772
+ },
773
+ {
774
+ "epoch": 0.85,
775
+ "learning_rate": 8.297582567245489e-06,
776
+ "loss": 0.3157,
777
+ "step": 2500
778
+ },
779
+ {
780
+ "epoch": 0.86,
781
+ "learning_rate": 8.283963227783454e-06,
782
+ "loss": 0.2519,
783
+ "step": 2520
784
+ },
785
+ {
786
+ "epoch": 0.86,
787
+ "learning_rate": 8.270343888321418e-06,
788
+ "loss": 0.3058,
789
+ "step": 2540
790
+ },
791
+ {
792
+ "epoch": 0.87,
793
+ "learning_rate": 8.25672454885938e-06,
794
+ "loss": 0.2764,
795
+ "step": 2560
796
+ },
797
+ {
798
+ "epoch": 0.88,
799
+ "learning_rate": 8.243105209397345e-06,
800
+ "loss": 0.2877,
801
+ "step": 2580
802
+ },
803
+ {
804
+ "epoch": 0.89,
805
+ "learning_rate": 8.229485869935309e-06,
806
+ "loss": 0.3564,
807
+ "step": 2600
808
+ },
809
+ {
810
+ "epoch": 0.89,
811
+ "learning_rate": 8.215866530473273e-06,
812
+ "loss": 0.2914,
813
+ "step": 2620
814
+ },
815
+ {
816
+ "epoch": 0.9,
817
+ "learning_rate": 8.202247191011237e-06,
818
+ "loss": 0.2847,
819
+ "step": 2640
820
+ },
821
+ {
822
+ "epoch": 0.91,
823
+ "learning_rate": 8.1886278515492e-06,
824
+ "loss": 0.3405,
825
+ "step": 2660
826
+ },
827
+ {
828
+ "epoch": 0.91,
829
+ "learning_rate": 8.175008512087164e-06,
830
+ "loss": 0.3281,
831
+ "step": 2680
832
+ },
833
+ {
834
+ "epoch": 0.92,
835
+ "learning_rate": 8.161389172625128e-06,
836
+ "loss": 0.263,
837
+ "step": 2700
838
+ },
839
+ {
840
+ "epoch": 0.93,
841
+ "learning_rate": 8.147769833163093e-06,
842
+ "loss": 0.2724,
843
+ "step": 2720
844
+ },
845
+ {
846
+ "epoch": 0.93,
847
+ "learning_rate": 8.134150493701057e-06,
848
+ "loss": 0.342,
849
+ "step": 2740
850
+ },
851
+ {
852
+ "epoch": 0.94,
853
+ "learning_rate": 8.120531154239019e-06,
854
+ "loss": 0.3156,
855
+ "step": 2760
856
+ },
857
+ {
858
+ "epoch": 0.95,
859
+ "learning_rate": 8.106911814776984e-06,
860
+ "loss": 0.3101,
861
+ "step": 2780
862
+ },
863
+ {
864
+ "epoch": 0.95,
865
+ "learning_rate": 8.093292475314948e-06,
866
+ "loss": 0.3083,
867
+ "step": 2800
868
+ },
869
+ {
870
+ "epoch": 0.96,
871
+ "learning_rate": 8.079673135852912e-06,
872
+ "loss": 0.304,
873
+ "step": 2820
874
+ },
875
+ {
876
+ "epoch": 0.97,
877
+ "learning_rate": 8.066053796390875e-06,
878
+ "loss": 0.3087,
879
+ "step": 2840
880
+ },
881
+ {
882
+ "epoch": 0.97,
883
+ "learning_rate": 8.05243445692884e-06,
884
+ "loss": 0.3681,
885
+ "step": 2860
886
+ },
887
+ {
888
+ "epoch": 0.98,
889
+ "learning_rate": 8.038815117466805e-06,
890
+ "loss": 0.2919,
891
+ "step": 2880
892
+ },
893
+ {
894
+ "epoch": 0.99,
895
+ "learning_rate": 8.025195778004767e-06,
896
+ "loss": 0.3218,
897
+ "step": 2900
898
+ },
899
+ {
900
+ "epoch": 0.99,
901
+ "learning_rate": 8.01157643854273e-06,
902
+ "loss": 0.3012,
903
+ "step": 2920
904
+ },
905
+ {
906
+ "epoch": 1.0,
907
+ "learning_rate": 7.997957099080696e-06,
908
+ "loss": 0.2952,
909
+ "step": 2940
910
+ },
911
+ {
912
+ "epoch": 1.01,
913
+ "learning_rate": 7.98433775961866e-06,
914
+ "loss": 0.2597,
915
+ "step": 2960
916
+ },
917
+ {
918
+ "epoch": 1.01,
919
+ "learning_rate": 7.970718420156623e-06,
920
+ "loss": 0.2836,
921
+ "step": 2980
922
+ },
923
+ {
924
+ "epoch": 1.02,
925
+ "learning_rate": 7.957099080694587e-06,
926
+ "loss": 0.2831,
927
+ "step": 3000
928
+ },
929
+ {
930
+ "epoch": 1.02,
931
+ "eval_loss": 0.48173409700393677,
932
+ "eval_roc_auc": 0.7896713615023474,
933
+ "eval_runtime": 226.5392,
934
+ "eval_samples_per_second": 65.816,
935
+ "eval_steps_per_second": 8.228,
936
+ "step": 3000
937
+ },
938
+ {
939
+ "epoch": 1.03,
940
+ "learning_rate": 7.94347974123255e-06,
941
+ "loss": 0.2553,
942
+ "step": 3020
943
+ },
944
+ {
945
+ "epoch": 1.04,
946
+ "learning_rate": 7.929860401770514e-06,
947
+ "loss": 0.2786,
948
+ "step": 3040
949
+ },
950
+ {
951
+ "epoch": 1.04,
952
+ "learning_rate": 7.916241062308478e-06,
953
+ "loss": 0.2636,
954
+ "step": 3060
955
+ },
956
+ {
957
+ "epoch": 1.05,
958
+ "learning_rate": 7.902621722846444e-06,
959
+ "loss": 0.2512,
960
+ "step": 3080
961
+ },
962
+ {
963
+ "epoch": 1.06,
964
+ "learning_rate": 7.889002383384407e-06,
965
+ "loss": 0.2755,
966
+ "step": 3100
967
+ },
968
+ {
969
+ "epoch": 1.06,
970
+ "learning_rate": 7.87538304392237e-06,
971
+ "loss": 0.2738,
972
+ "step": 3120
973
+ },
974
+ {
975
+ "epoch": 1.07,
976
+ "learning_rate": 7.861763704460335e-06,
977
+ "loss": 0.2653,
978
+ "step": 3140
979
+ },
980
+ {
981
+ "epoch": 1.08,
982
+ "learning_rate": 7.848144364998298e-06,
983
+ "loss": 0.2944,
984
+ "step": 3160
985
+ },
986
+ {
987
+ "epoch": 1.08,
988
+ "learning_rate": 7.834525025536262e-06,
989
+ "loss": 0.2624,
990
+ "step": 3180
991
+ },
992
+ {
993
+ "epoch": 1.09,
994
+ "learning_rate": 7.820905686074226e-06,
995
+ "loss": 0.2323,
996
+ "step": 3200
997
+ },
998
+ {
999
+ "epoch": 1.1,
1000
+ "learning_rate": 7.80728634661219e-06,
1001
+ "loss": 0.2742,
1002
+ "step": 3220
1003
+ },
1004
+ {
1005
+ "epoch": 1.1,
1006
+ "learning_rate": 7.793667007150153e-06,
1007
+ "loss": 0.2747,
1008
+ "step": 3240
1009
+ },
1010
+ {
1011
+ "epoch": 1.11,
1012
+ "learning_rate": 7.780047667688117e-06,
1013
+ "loss": 0.2377,
1014
+ "step": 3260
1015
+ },
1016
+ {
1017
+ "epoch": 1.12,
1018
+ "learning_rate": 7.766428328226083e-06,
1019
+ "loss": 0.2602,
1020
+ "step": 3280
1021
+ },
1022
+ {
1023
+ "epoch": 1.12,
1024
+ "learning_rate": 7.752808988764046e-06,
1025
+ "loss": 0.2795,
1026
+ "step": 3300
1027
+ },
1028
+ {
1029
+ "epoch": 1.13,
1030
+ "learning_rate": 7.739189649302008e-06,
1031
+ "loss": 0.2344,
1032
+ "step": 3320
1033
+ },
1034
+ {
1035
+ "epoch": 1.14,
1036
+ "learning_rate": 7.725570309839974e-06,
1037
+ "loss": 0.2519,
1038
+ "step": 3340
1039
+ },
1040
+ {
1041
+ "epoch": 1.14,
1042
+ "learning_rate": 7.711950970377937e-06,
1043
+ "loss": 0.2342,
1044
+ "step": 3360
1045
+ },
1046
+ {
1047
+ "epoch": 1.15,
1048
+ "learning_rate": 7.698331630915901e-06,
1049
+ "loss": 0.2442,
1050
+ "step": 3380
1051
+ },
1052
+ {
1053
+ "epoch": 1.16,
1054
+ "learning_rate": 7.684712291453865e-06,
1055
+ "loss": 0.2626,
1056
+ "step": 3400
1057
+ },
1058
+ {
1059
+ "epoch": 1.16,
1060
+ "learning_rate": 7.671092951991829e-06,
1061
+ "loss": 0.2381,
1062
+ "step": 3420
1063
+ },
1064
+ {
1065
+ "epoch": 1.17,
1066
+ "learning_rate": 7.657473612529794e-06,
1067
+ "loss": 0.2341,
1068
+ "step": 3440
1069
+ },
1070
+ {
1071
+ "epoch": 1.18,
1072
+ "learning_rate": 7.643854273067756e-06,
1073
+ "loss": 0.2994,
1074
+ "step": 3460
1075
+ },
1076
+ {
1077
+ "epoch": 1.18,
1078
+ "learning_rate": 7.63023493360572e-06,
1079
+ "loss": 0.2759,
1080
+ "step": 3480
1081
+ },
1082
+ {
1083
+ "epoch": 1.19,
1084
+ "learning_rate": 7.616615594143684e-06,
1085
+ "loss": 0.2927,
1086
+ "step": 3500
1087
+ },
1088
+ {
1089
+ "epoch": 1.2,
1090
+ "learning_rate": 7.602996254681648e-06,
1091
+ "loss": 0.2327,
1092
+ "step": 3520
1093
+ },
1094
+ {
1095
+ "epoch": 1.21,
1096
+ "learning_rate": 7.589376915219613e-06,
1097
+ "loss": 0.3248,
1098
+ "step": 3540
1099
+ },
1100
+ {
1101
+ "epoch": 1.21,
1102
+ "learning_rate": 7.5757575757575764e-06,
1103
+ "loss": 0.2836,
1104
+ "step": 3560
1105
+ },
1106
+ {
1107
+ "epoch": 1.22,
1108
+ "learning_rate": 7.56213823629554e-06,
1109
+ "loss": 0.2545,
1110
+ "step": 3580
1111
+ },
1112
+ {
1113
+ "epoch": 1.23,
1114
+ "learning_rate": 7.548518896833505e-06,
1115
+ "loss": 0.2596,
1116
+ "step": 3600
1117
+ },
1118
+ {
1119
+ "epoch": 1.23,
1120
+ "learning_rate": 7.534899557371468e-06,
1121
+ "loss": 0.2308,
1122
+ "step": 3620
1123
+ },
1124
+ {
1125
+ "epoch": 1.24,
1126
+ "learning_rate": 7.521280217909432e-06,
1127
+ "loss": 0.2462,
1128
+ "step": 3640
1129
+ },
1130
+ {
1131
+ "epoch": 1.25,
1132
+ "learning_rate": 7.507660878447396e-06,
1133
+ "loss": 0.2763,
1134
+ "step": 3660
1135
+ },
1136
+ {
1137
+ "epoch": 1.25,
1138
+ "learning_rate": 7.49404153898536e-06,
1139
+ "loss": 0.271,
1140
+ "step": 3680
1141
+ },
1142
+ {
1143
+ "epoch": 1.26,
1144
+ "learning_rate": 7.480422199523324e-06,
1145
+ "loss": 0.2717,
1146
+ "step": 3700
1147
+ },
1148
+ {
1149
+ "epoch": 1.27,
1150
+ "learning_rate": 7.466802860061287e-06,
1151
+ "loss": 0.2995,
1152
+ "step": 3720
1153
+ },
1154
+ {
1155
+ "epoch": 1.27,
1156
+ "learning_rate": 7.453183520599252e-06,
1157
+ "loss": 0.2812,
1158
+ "step": 3740
1159
+ },
1160
+ {
1161
+ "epoch": 1.28,
1162
+ "learning_rate": 7.439564181137215e-06,
1163
+ "loss": 0.2863,
1164
+ "step": 3760
1165
+ },
1166
+ {
1167
+ "epoch": 1.29,
1168
+ "learning_rate": 7.425944841675179e-06,
1169
+ "loss": 0.2468,
1170
+ "step": 3780
1171
+ },
1172
+ {
1173
+ "epoch": 1.29,
1174
+ "learning_rate": 7.412325502213144e-06,
1175
+ "loss": 0.2542,
1176
+ "step": 3800
1177
+ },
1178
+ {
1179
+ "epoch": 1.3,
1180
+ "learning_rate": 7.3987061627511066e-06,
1181
+ "loss": 0.2439,
1182
+ "step": 3820
1183
+ },
1184
+ {
1185
+ "epoch": 1.31,
1186
+ "learning_rate": 7.385086823289071e-06,
1187
+ "loss": 0.2923,
1188
+ "step": 3840
1189
+ },
1190
+ {
1191
+ "epoch": 1.31,
1192
+ "learning_rate": 7.371467483827035e-06,
1193
+ "loss": 0.2809,
1194
+ "step": 3860
1195
+ },
1196
+ {
1197
+ "epoch": 1.32,
1198
+ "learning_rate": 7.357848144364999e-06,
1199
+ "loss": 0.2601,
1200
+ "step": 3880
1201
+ },
1202
+ {
1203
+ "epoch": 1.33,
1204
+ "learning_rate": 7.344228804902963e-06,
1205
+ "loss": 0.2481,
1206
+ "step": 3900
1207
+ },
1208
+ {
1209
+ "epoch": 1.33,
1210
+ "learning_rate": 7.330609465440927e-06,
1211
+ "loss": 0.2886,
1212
+ "step": 3920
1213
+ },
1214
+ {
1215
+ "epoch": 1.34,
1216
+ "learning_rate": 7.3169901259788915e-06,
1217
+ "loss": 0.2546,
1218
+ "step": 3940
1219
+ },
1220
+ {
1221
+ "epoch": 1.35,
1222
+ "learning_rate": 7.303370786516854e-06,
1223
+ "loss": 0.2393,
1224
+ "step": 3960
1225
+ },
1226
+ {
1227
+ "epoch": 1.36,
1228
+ "learning_rate": 7.289751447054818e-06,
1229
+ "loss": 0.2839,
1230
+ "step": 3980
1231
+ },
1232
+ {
1233
+ "epoch": 1.36,
1234
+ "learning_rate": 7.276132107592783e-06,
1235
+ "loss": 0.2781,
1236
+ "step": 4000
1237
+ },
1238
+ {
1239
+ "epoch": 1.36,
1240
+ "eval_loss": 0.5417753458023071,
1241
+ "eval_roc_auc": 0.7655935613682091,
1242
+ "eval_runtime": 226.2172,
1243
+ "eval_samples_per_second": 65.91,
1244
+ "eval_steps_per_second": 8.24,
1245
+ "step": 4000
1246
+ },
1247
+ {
1248
+ "epoch": 1.37,
1249
+ "learning_rate": 7.262512768130746e-06,
1250
+ "loss": 0.2696,
1251
+ "step": 4020
1252
+ },
1253
+ {
1254
+ "epoch": 1.38,
1255
+ "learning_rate": 7.248893428668711e-06,
1256
+ "loss": 0.2458,
1257
+ "step": 4040
1258
+ },
1259
+ {
1260
+ "epoch": 1.38,
1261
+ "learning_rate": 7.235274089206674e-06,
1262
+ "loss": 0.269,
1263
+ "step": 4060
1264
+ },
1265
+ {
1266
+ "epoch": 1.39,
1267
+ "learning_rate": 7.2216547497446376e-06,
1268
+ "loss": 0.2798,
1269
+ "step": 4080
1270
+ },
1271
+ {
1272
+ "epoch": 1.4,
1273
+ "learning_rate": 7.208035410282602e-06,
1274
+ "loss": 0.2501,
1275
+ "step": 4100
1276
+ },
1277
+ {
1278
+ "epoch": 1.4,
1279
+ "learning_rate": 7.194416070820566e-06,
1280
+ "loss": 0.2655,
1281
+ "step": 4120
1282
+ },
1283
+ {
1284
+ "epoch": 1.41,
1285
+ "learning_rate": 7.180796731358529e-06,
1286
+ "loss": 0.2207,
1287
+ "step": 4140
1288
+ },
1289
+ {
1290
+ "epoch": 1.42,
1291
+ "learning_rate": 7.167177391896494e-06,
1292
+ "loss": 0.2381,
1293
+ "step": 4160
1294
+ },
1295
+ {
1296
+ "epoch": 1.42,
1297
+ "learning_rate": 7.153558052434457e-06,
1298
+ "loss": 0.2929,
1299
+ "step": 4180
1300
+ },
1301
+ {
1302
+ "epoch": 1.43,
1303
+ "learning_rate": 7.139938712972422e-06,
1304
+ "loss": 0.2455,
1305
+ "step": 4200
1306
+ },
1307
+ {
1308
+ "epoch": 1.44,
1309
+ "learning_rate": 7.126319373510385e-06,
1310
+ "loss": 0.2443,
1311
+ "step": 4220
1312
+ },
1313
+ {
1314
+ "epoch": 1.44,
1315
+ "learning_rate": 7.112700034048349e-06,
1316
+ "loss": 0.2439,
1317
+ "step": 4240
1318
+ },
1319
+ {
1320
+ "epoch": 1.45,
1321
+ "learning_rate": 7.099080694586314e-06,
1322
+ "loss": 0.2799,
1323
+ "step": 4260
1324
+ },
1325
+ {
1326
+ "epoch": 1.46,
1327
+ "learning_rate": 7.0854613551242765e-06,
1328
+ "loss": 0.2492,
1329
+ "step": 4280
1330
+ },
1331
+ {
1332
+ "epoch": 1.46,
1333
+ "learning_rate": 7.071842015662241e-06,
1334
+ "loss": 0.281,
1335
+ "step": 4300
1336
+ },
1337
+ {
1338
+ "epoch": 1.47,
1339
+ "learning_rate": 7.058222676200205e-06,
1340
+ "loss": 0.2539,
1341
+ "step": 4320
1342
+ },
1343
+ {
1344
+ "epoch": 1.48,
1345
+ "learning_rate": 7.0446033367381686e-06,
1346
+ "loss": 0.2464,
1347
+ "step": 4340
1348
+ },
1349
+ {
1350
+ "epoch": 1.48,
1351
+ "learning_rate": 7.030983997276133e-06,
1352
+ "loss": 0.2929,
1353
+ "step": 4360
1354
+ },
1355
+ {
1356
+ "epoch": 1.49,
1357
+ "learning_rate": 7.017364657814096e-06,
1358
+ "loss": 0.2679,
1359
+ "step": 4380
1360
+ },
1361
+ {
1362
+ "epoch": 1.5,
1363
+ "learning_rate": 7.003745318352061e-06,
1364
+ "loss": 0.2155,
1365
+ "step": 4400
1366
+ },
1367
+ {
1368
+ "epoch": 1.5,
1369
+ "learning_rate": 6.990125978890024e-06,
1370
+ "loss": 0.247,
1371
+ "step": 4420
1372
+ },
1373
+ {
1374
+ "epoch": 1.51,
1375
+ "learning_rate": 6.976506639427988e-06,
1376
+ "loss": 0.2069,
1377
+ "step": 4440
1378
+ },
1379
+ {
1380
+ "epoch": 1.52,
1381
+ "learning_rate": 6.962887299965953e-06,
1382
+ "loss": 0.2544,
1383
+ "step": 4460
1384
+ },
1385
+ {
1386
+ "epoch": 1.53,
1387
+ "learning_rate": 6.949267960503916e-06,
1388
+ "loss": 0.2162,
1389
+ "step": 4480
1390
+ },
1391
+ {
1392
+ "epoch": 1.53,
1393
+ "learning_rate": 6.935648621041881e-06,
1394
+ "loss": 0.252,
1395
+ "step": 4500
1396
+ },
1397
+ {
1398
+ "epoch": 1.54,
1399
+ "learning_rate": 6.922029281579844e-06,
1400
+ "loss": 0.2548,
1401
+ "step": 4520
1402
+ },
1403
+ {
1404
+ "epoch": 1.55,
1405
+ "learning_rate": 6.9084099421178075e-06,
1406
+ "loss": 0.27,
1407
+ "step": 4540
1408
+ },
1409
+ {
1410
+ "epoch": 1.55,
1411
+ "learning_rate": 6.894790602655772e-06,
1412
+ "loss": 0.2314,
1413
+ "step": 4560
1414
+ },
1415
+ {
1416
+ "epoch": 1.56,
1417
+ "learning_rate": 6.881171263193736e-06,
1418
+ "loss": 0.2381,
1419
+ "step": 4580
1420
+ },
1421
+ {
1422
+ "epoch": 1.57,
1423
+ "learning_rate": 6.8675519237317e-06,
1424
+ "loss": 0.2739,
1425
+ "step": 4600
1426
+ },
1427
+ {
1428
+ "epoch": 1.57,
1429
+ "learning_rate": 6.853932584269663e-06,
1430
+ "loss": 0.2185,
1431
+ "step": 4620
1432
+ },
1433
+ {
1434
+ "epoch": 1.58,
1435
+ "learning_rate": 6.840313244807627e-06,
1436
+ "loss": 0.2458,
1437
+ "step": 4640
1438
+ },
1439
+ {
1440
+ "epoch": 1.59,
1441
+ "learning_rate": 6.826693905345592e-06,
1442
+ "loss": 0.3119,
1443
+ "step": 4660
1444
+ },
1445
+ {
1446
+ "epoch": 1.59,
1447
+ "learning_rate": 6.813074565883555e-06,
1448
+ "loss": 0.2082,
1449
+ "step": 4680
1450
+ },
1451
+ {
1452
+ "epoch": 1.6,
1453
+ "learning_rate": 6.799455226421518e-06,
1454
+ "loss": 0.2772,
1455
+ "step": 4700
1456
+ },
1457
+ {
1458
+ "epoch": 1.61,
1459
+ "learning_rate": 6.785835886959483e-06,
1460
+ "loss": 0.238,
1461
+ "step": 4720
1462
+ },
1463
+ {
1464
+ "epoch": 1.61,
1465
+ "learning_rate": 6.7722165474974465e-06,
1466
+ "loss": 0.2403,
1467
+ "step": 4740
1468
+ },
1469
+ {
1470
+ "epoch": 1.62,
1471
+ "learning_rate": 6.758597208035411e-06,
1472
+ "loss": 0.2496,
1473
+ "step": 4760
1474
+ },
1475
+ {
1476
+ "epoch": 1.63,
1477
+ "learning_rate": 6.744977868573375e-06,
1478
+ "loss": 0.258,
1479
+ "step": 4780
1480
+ },
1481
+ {
1482
+ "epoch": 1.63,
1483
+ "learning_rate": 6.7313585291113385e-06,
1484
+ "loss": 0.2462,
1485
+ "step": 4800
1486
+ },
1487
+ {
1488
+ "epoch": 1.64,
1489
+ "learning_rate": 6.717739189649303e-06,
1490
+ "loss": 0.3012,
1491
+ "step": 4820
1492
+ },
1493
+ {
1494
+ "epoch": 1.65,
1495
+ "learning_rate": 6.704119850187266e-06,
1496
+ "loss": 0.2331,
1497
+ "step": 4840
1498
+ },
1499
+ {
1500
+ "epoch": 1.65,
1501
+ "learning_rate": 6.6905005107252305e-06,
1502
+ "loss": 0.2477,
1503
+ "step": 4860
1504
+ },
1505
+ {
1506
+ "epoch": 1.66,
1507
+ "learning_rate": 6.676881171263194e-06,
1508
+ "loss": 0.234,
1509
+ "step": 4880
1510
+ },
1511
+ {
1512
+ "epoch": 1.67,
1513
+ "learning_rate": 6.663261831801158e-06,
1514
+ "loss": 0.2434,
1515
+ "step": 4900
1516
+ },
1517
+ {
1518
+ "epoch": 1.68,
1519
+ "learning_rate": 6.649642492339123e-06,
1520
+ "loss": 0.3011,
1521
+ "step": 4920
1522
+ },
1523
+ {
1524
+ "epoch": 1.68,
1525
+ "learning_rate": 6.6360231528770855e-06,
1526
+ "loss": 0.2587,
1527
+ "step": 4940
1528
+ },
1529
+ {
1530
+ "epoch": 1.69,
1531
+ "learning_rate": 6.62240381341505e-06,
1532
+ "loss": 0.2252,
1533
+ "step": 4960
1534
+ },
1535
+ {
1536
+ "epoch": 1.7,
1537
+ "learning_rate": 6.608784473953014e-06,
1538
+ "loss": 0.2443,
1539
+ "step": 4980
1540
+ },
1541
+ {
1542
+ "epoch": 1.7,
1543
+ "learning_rate": 6.5951651344909775e-06,
1544
+ "loss": 0.2355,
1545
+ "step": 5000
1546
+ },
1547
+ {
1548
+ "epoch": 1.7,
1549
+ "eval_loss": 0.5398261547088623,
1550
+ "eval_roc_auc": 0.7786049631120053,
1551
+ "eval_runtime": 224.1313,
1552
+ "eval_samples_per_second": 66.524,
1553
+ "eval_steps_per_second": 8.317,
1554
+ "step": 5000
1555
+ },
1556
+ {
1557
+ "epoch": 1.71,
1558
+ "learning_rate": 6.581545795028942e-06,
1559
+ "loss": 0.2645,
1560
+ "step": 5020
1561
+ },
1562
+ {
1563
+ "epoch": 1.72,
1564
+ "learning_rate": 6.567926455566905e-06,
1565
+ "loss": 0.2447,
1566
+ "step": 5040
1567
+ },
1568
+ {
1569
+ "epoch": 1.72,
1570
+ "learning_rate": 6.5543071161048695e-06,
1571
+ "loss": 0.234,
1572
+ "step": 5060
1573
+ },
1574
+ {
1575
+ "epoch": 1.73,
1576
+ "learning_rate": 6.540687776642833e-06,
1577
+ "loss": 0.2343,
1578
+ "step": 5080
1579
+ },
1580
+ {
1581
+ "epoch": 1.74,
1582
+ "learning_rate": 6.527068437180797e-06,
1583
+ "loss": 0.2592,
1584
+ "step": 5100
1585
+ },
1586
+ {
1587
+ "epoch": 1.74,
1588
+ "learning_rate": 6.5134490977187615e-06,
1589
+ "loss": 0.2689,
1590
+ "step": 5120
1591
+ },
1592
+ {
1593
+ "epoch": 1.75,
1594
+ "learning_rate": 6.499829758256725e-06,
1595
+ "loss": 0.2535,
1596
+ "step": 5140
1597
+ },
1598
+ {
1599
+ "epoch": 1.76,
1600
+ "learning_rate": 6.48621041879469e-06,
1601
+ "loss": 0.2059,
1602
+ "step": 5160
1603
+ },
1604
+ {
1605
+ "epoch": 1.76,
1606
+ "learning_rate": 6.472591079332653e-06,
1607
+ "loss": 0.2646,
1608
+ "step": 5180
1609
+ },
1610
+ {
1611
+ "epoch": 1.77,
1612
+ "learning_rate": 6.4589717398706165e-06,
1613
+ "loss": 0.2203,
1614
+ "step": 5200
1615
+ },
1616
+ {
1617
+ "epoch": 1.78,
1618
+ "learning_rate": 6.445352400408581e-06,
1619
+ "loss": 0.2413,
1620
+ "step": 5220
1621
+ },
1622
+ {
1623
+ "epoch": 1.78,
1624
+ "learning_rate": 6.431733060946545e-06,
1625
+ "loss": 0.2399,
1626
+ "step": 5240
1627
+ },
1628
+ {
1629
+ "epoch": 1.79,
1630
+ "learning_rate": 6.418113721484509e-06,
1631
+ "loss": 0.2313,
1632
+ "step": 5260
1633
+ },
1634
+ {
1635
+ "epoch": 1.8,
1636
+ "learning_rate": 6.404494382022472e-06,
1637
+ "loss": 0.2082,
1638
+ "step": 5280
1639
+ },
1640
+ {
1641
+ "epoch": 1.8,
1642
+ "learning_rate": 6.390875042560436e-06,
1643
+ "loss": 0.2555,
1644
+ "step": 5300
1645
+ },
1646
+ {
1647
+ "epoch": 1.81,
1648
+ "learning_rate": 6.3772557030984005e-06,
1649
+ "loss": 0.2664,
1650
+ "step": 5320
1651
+ },
1652
+ {
1653
+ "epoch": 1.82,
1654
+ "learning_rate": 6.363636363636364e-06,
1655
+ "loss": 0.2367,
1656
+ "step": 5340
1657
+ },
1658
+ {
1659
+ "epoch": 1.82,
1660
+ "learning_rate": 6.350017024174327e-06,
1661
+ "loss": 0.2643,
1662
+ "step": 5360
1663
+ },
1664
+ {
1665
+ "epoch": 1.83,
1666
+ "learning_rate": 6.336397684712292e-06,
1667
+ "loss": 0.1894,
1668
+ "step": 5380
1669
+ },
1670
+ {
1671
+ "epoch": 1.84,
1672
+ "learning_rate": 6.322778345250255e-06,
1673
+ "loss": 0.2024,
1674
+ "step": 5400
1675
+ },
1676
+ {
1677
+ "epoch": 1.85,
1678
+ "learning_rate": 6.30915900578822e-06,
1679
+ "loss": 0.2293,
1680
+ "step": 5420
1681
+ },
1682
+ {
1683
+ "epoch": 1.85,
1684
+ "learning_rate": 6.295539666326184e-06,
1685
+ "loss": 0.2299,
1686
+ "step": 5440
1687
+ },
1688
+ {
1689
+ "epoch": 1.86,
1690
+ "learning_rate": 6.2819203268641474e-06,
1691
+ "loss": 0.2658,
1692
+ "step": 5460
1693
+ },
1694
+ {
1695
+ "epoch": 1.87,
1696
+ "learning_rate": 6.268300987402112e-06,
1697
+ "loss": 0.2577,
1698
+ "step": 5480
1699
+ },
1700
+ {
1701
+ "epoch": 1.87,
1702
+ "learning_rate": 6.254681647940075e-06,
1703
+ "loss": 0.2487,
1704
+ "step": 5500
1705
+ },
1706
+ {
1707
+ "epoch": 1.88,
1708
+ "learning_rate": 6.2410623084780395e-06,
1709
+ "loss": 0.2413,
1710
+ "step": 5520
1711
+ },
1712
+ {
1713
+ "epoch": 1.89,
1714
+ "learning_rate": 6.227442969016003e-06,
1715
+ "loss": 0.2544,
1716
+ "step": 5540
1717
+ },
1718
+ {
1719
+ "epoch": 1.89,
1720
+ "learning_rate": 6.213823629553967e-06,
1721
+ "loss": 0.2531,
1722
+ "step": 5560
1723
+ },
1724
+ {
1725
+ "epoch": 1.9,
1726
+ "learning_rate": 6.2002042900919315e-06,
1727
+ "loss": 0.293,
1728
+ "step": 5580
1729
+ },
1730
+ {
1731
+ "epoch": 1.91,
1732
+ "learning_rate": 6.186584950629894e-06,
1733
+ "loss": 0.2563,
1734
+ "step": 5600
1735
+ },
1736
+ {
1737
+ "epoch": 1.91,
1738
+ "learning_rate": 6.172965611167859e-06,
1739
+ "loss": 0.2549,
1740
+ "step": 5620
1741
+ },
1742
+ {
1743
+ "epoch": 1.92,
1744
+ "learning_rate": 6.159346271705823e-06,
1745
+ "loss": 0.2166,
1746
+ "step": 5640
1747
+ },
1748
+ {
1749
+ "epoch": 1.93,
1750
+ "learning_rate": 6.145726932243786e-06,
1751
+ "loss": 0.2084,
1752
+ "step": 5660
1753
+ },
1754
+ {
1755
+ "epoch": 1.93,
1756
+ "learning_rate": 6.132107592781751e-06,
1757
+ "loss": 0.2599,
1758
+ "step": 5680
1759
+ },
1760
+ {
1761
+ "epoch": 1.94,
1762
+ "learning_rate": 6.118488253319715e-06,
1763
+ "loss": 0.2266,
1764
+ "step": 5700
1765
+ },
1766
+ {
1767
+ "epoch": 1.95,
1768
+ "learning_rate": 6.104868913857679e-06,
1769
+ "loss": 0.2363,
1770
+ "step": 5720
1771
+ },
1772
+ {
1773
+ "epoch": 1.95,
1774
+ "learning_rate": 6.091249574395642e-06,
1775
+ "loss": 0.2586,
1776
+ "step": 5740
1777
+ },
1778
+ {
1779
+ "epoch": 1.96,
1780
+ "learning_rate": 6.077630234933606e-06,
1781
+ "loss": 0.2428,
1782
+ "step": 5760
1783
+ },
1784
+ {
1785
+ "epoch": 1.97,
1786
+ "learning_rate": 6.0640108954715705e-06,
1787
+ "loss": 0.269,
1788
+ "step": 5780
1789
+ },
1790
+ {
1791
+ "epoch": 1.97,
1792
+ "learning_rate": 6.050391556009534e-06,
1793
+ "loss": 0.2265,
1794
+ "step": 5800
1795
+ },
1796
+ {
1797
+ "epoch": 1.98,
1798
+ "learning_rate": 6.036772216547499e-06,
1799
+ "loss": 0.2594,
1800
+ "step": 5820
1801
+ },
1802
+ {
1803
+ "epoch": 1.99,
1804
+ "learning_rate": 6.023152877085462e-06,
1805
+ "loss": 0.2552,
1806
+ "step": 5840
1807
+ },
1808
+ {
1809
+ "epoch": 2.0,
1810
+ "learning_rate": 6.009533537623425e-06,
1811
+ "loss": 0.2279,
1812
+ "step": 5860
1813
+ },
1814
+ {
1815
+ "epoch": 2.0,
1816
+ "learning_rate": 5.99591419816139e-06,
1817
+ "loss": 0.2381,
1818
+ "step": 5880
1819
+ },
1820
+ {
1821
+ "epoch": 2.01,
1822
+ "learning_rate": 5.982294858699354e-06,
1823
+ "loss": 0.2271,
1824
+ "step": 5900
1825
+ },
1826
+ {
1827
+ "epoch": 2.02,
1828
+ "learning_rate": 5.968675519237318e-06,
1829
+ "loss": 0.1785,
1830
+ "step": 5920
1831
+ },
1832
+ {
1833
+ "epoch": 2.02,
1834
+ "learning_rate": 5.955056179775281e-06,
1835
+ "loss": 0.2066,
1836
+ "step": 5940
1837
+ },
1838
+ {
1839
+ "epoch": 2.03,
1840
+ "learning_rate": 5.941436840313245e-06,
1841
+ "loss": 0.219,
1842
+ "step": 5960
1843
+ },
1844
+ {
1845
+ "epoch": 2.04,
1846
+ "learning_rate": 5.9278175008512094e-06,
1847
+ "loss": 0.1896,
1848
+ "step": 5980
1849
+ },
1850
+ {
1851
+ "epoch": 2.04,
1852
+ "learning_rate": 5.914198161389173e-06,
1853
+ "loss": 0.1978,
1854
+ "step": 6000
1855
+ },
1856
+ {
1857
+ "epoch": 2.04,
1858
+ "eval_loss": 0.6120939254760742,
1859
+ "eval_roc_auc": 0.7648558014755198,
1860
+ "eval_runtime": 225.5238,
1861
+ "eval_samples_per_second": 66.113,
1862
+ "eval_steps_per_second": 8.265,
1863
+ "step": 6000
1864
+ },
1865
+ {
1866
+ "epoch": 2.05,
1867
+ "learning_rate": 5.900578821927137e-06,
1868
+ "loss": 0.1811,
1869
+ "step": 6020
1870
+ },
1871
+ {
1872
+ "epoch": 2.06,
1873
+ "learning_rate": 5.8869594824651015e-06,
1874
+ "loss": 0.2201,
1875
+ "step": 6040
1876
+ },
1877
+ {
1878
+ "epoch": 2.06,
1879
+ "learning_rate": 5.873340143003064e-06,
1880
+ "loss": 0.2118,
1881
+ "step": 6060
1882
+ },
1883
+ {
1884
+ "epoch": 2.07,
1885
+ "learning_rate": 5.859720803541029e-06,
1886
+ "loss": 0.1765,
1887
+ "step": 6080
1888
+ },
1889
+ {
1890
+ "epoch": 2.08,
1891
+ "learning_rate": 5.846101464078993e-06,
1892
+ "loss": 0.2063,
1893
+ "step": 6100
1894
+ },
1895
+ {
1896
+ "epoch": 2.08,
1897
+ "learning_rate": 5.832482124616956e-06,
1898
+ "loss": 0.2055,
1899
+ "step": 6120
1900
+ },
1901
+ {
1902
+ "epoch": 2.09,
1903
+ "learning_rate": 5.818862785154921e-06,
1904
+ "loss": 0.1919,
1905
+ "step": 6140
1906
+ },
1907
+ {
1908
+ "epoch": 2.1,
1909
+ "learning_rate": 5.805243445692884e-06,
1910
+ "loss": 0.1999,
1911
+ "step": 6160
1912
+ },
1913
+ {
1914
+ "epoch": 2.1,
1915
+ "learning_rate": 5.791624106230848e-06,
1916
+ "loss": 0.1761,
1917
+ "step": 6180
1918
+ },
1919
+ {
1920
+ "epoch": 2.11,
1921
+ "learning_rate": 5.778004766768812e-06,
1922
+ "loss": 0.1917,
1923
+ "step": 6200
1924
+ },
1925
+ {
1926
+ "epoch": 2.12,
1927
+ "learning_rate": 5.764385427306776e-06,
1928
+ "loss": 0.2086,
1929
+ "step": 6220
1930
+ },
1931
+ {
1932
+ "epoch": 2.12,
1933
+ "learning_rate": 5.7507660878447404e-06,
1934
+ "loss": 0.195,
1935
+ "step": 6240
1936
+ },
1937
+ {
1938
+ "epoch": 2.13,
1939
+ "learning_rate": 5.737146748382703e-06,
1940
+ "loss": 0.2244,
1941
+ "step": 6260
1942
+ },
1943
+ {
1944
+ "epoch": 2.14,
1945
+ "learning_rate": 5.723527408920668e-06,
1946
+ "loss": 0.2411,
1947
+ "step": 6280
1948
+ },
1949
+ {
1950
+ "epoch": 2.15,
1951
+ "learning_rate": 5.709908069458632e-06,
1952
+ "loss": 0.1984,
1953
+ "step": 6300
1954
+ },
1955
+ {
1956
+ "epoch": 2.15,
1957
+ "learning_rate": 5.696288729996595e-06,
1958
+ "loss": 0.1899,
1959
+ "step": 6320
1960
+ },
1961
+ {
1962
+ "epoch": 2.16,
1963
+ "learning_rate": 5.68266939053456e-06,
1964
+ "loss": 0.2241,
1965
+ "step": 6340
1966
+ },
1967
+ {
1968
+ "epoch": 2.17,
1969
+ "learning_rate": 5.669050051072524e-06,
1970
+ "loss": 0.2296,
1971
+ "step": 6360
1972
+ },
1973
+ {
1974
+ "epoch": 2.17,
1975
+ "learning_rate": 5.655430711610488e-06,
1976
+ "loss": 0.2082,
1977
+ "step": 6380
1978
+ },
1979
+ {
1980
+ "epoch": 2.18,
1981
+ "learning_rate": 5.641811372148451e-06,
1982
+ "loss": 0.1981,
1983
+ "step": 6400
1984
+ },
1985
+ {
1986
+ "epoch": 2.19,
1987
+ "learning_rate": 5.628192032686415e-06,
1988
+ "loss": 0.2306,
1989
+ "step": 6420
1990
+ },
1991
+ {
1992
+ "epoch": 2.19,
1993
+ "learning_rate": 5.614572693224379e-06,
1994
+ "loss": 0.1962,
1995
+ "step": 6440
1996
+ },
1997
+ {
1998
+ "epoch": 2.2,
1999
+ "learning_rate": 5.600953353762343e-06,
2000
+ "loss": 0.1773,
2001
+ "step": 6460
2002
+ },
2003
+ {
2004
+ "epoch": 2.21,
2005
+ "learning_rate": 5.587334014300308e-06,
2006
+ "loss": 0.2058,
2007
+ "step": 6480
2008
+ },
2009
+ {
2010
+ "epoch": 2.21,
2011
+ "learning_rate": 5.5737146748382706e-06,
2012
+ "loss": 0.1875,
2013
+ "step": 6500
2014
+ },
2015
+ {
2016
+ "epoch": 2.22,
2017
+ "learning_rate": 5.560095335376234e-06,
2018
+ "loss": 0.208,
2019
+ "step": 6520
2020
+ },
2021
+ {
2022
+ "epoch": 2.23,
2023
+ "learning_rate": 5.546475995914199e-06,
2024
+ "loss": 0.2189,
2025
+ "step": 6540
2026
+ },
2027
+ {
2028
+ "epoch": 2.23,
2029
+ "learning_rate": 5.532856656452163e-06,
2030
+ "loss": 0.1869,
2031
+ "step": 6560
2032
+ },
2033
+ {
2034
+ "epoch": 2.24,
2035
+ "learning_rate": 5.5192373169901255e-06,
2036
+ "loss": 0.1737,
2037
+ "step": 6580
2038
+ },
2039
+ {
2040
+ "epoch": 2.25,
2041
+ "learning_rate": 5.50561797752809e-06,
2042
+ "loss": 0.1807,
2043
+ "step": 6600
2044
+ },
2045
+ {
2046
+ "epoch": 2.25,
2047
+ "learning_rate": 5.491998638066054e-06,
2048
+ "loss": 0.2236,
2049
+ "step": 6620
2050
+ },
2051
+ {
2052
+ "epoch": 2.26,
2053
+ "learning_rate": 5.478379298604018e-06,
2054
+ "loss": 0.2033,
2055
+ "step": 6640
2056
+ },
2057
+ {
2058
+ "epoch": 2.27,
2059
+ "learning_rate": 5.464759959141982e-06,
2060
+ "loss": 0.177,
2061
+ "step": 6660
2062
+ },
2063
+ {
2064
+ "epoch": 2.27,
2065
+ "learning_rate": 5.451140619679946e-06,
2066
+ "loss": 0.1955,
2067
+ "step": 6680
2068
+ },
2069
+ {
2070
+ "epoch": 2.28,
2071
+ "learning_rate": 5.43752128021791e-06,
2072
+ "loss": 0.1657,
2073
+ "step": 6700
2074
+ },
2075
+ {
2076
+ "epoch": 2.29,
2077
+ "learning_rate": 5.423901940755873e-06,
2078
+ "loss": 0.2116,
2079
+ "step": 6720
2080
+ },
2081
+ {
2082
+ "epoch": 2.29,
2083
+ "learning_rate": 5.410282601293838e-06,
2084
+ "loss": 0.1991,
2085
+ "step": 6740
2086
+ },
2087
+ {
2088
+ "epoch": 2.3,
2089
+ "learning_rate": 5.3966632618318016e-06,
2090
+ "loss": 0.2085,
2091
+ "step": 6760
2092
+ },
2093
+ {
2094
+ "epoch": 2.31,
2095
+ "learning_rate": 5.383043922369765e-06,
2096
+ "loss": 0.2189,
2097
+ "step": 6780
2098
+ },
2099
+ {
2100
+ "epoch": 2.32,
2101
+ "learning_rate": 5.36942458290773e-06,
2102
+ "loss": 0.1781,
2103
+ "step": 6800
2104
+ },
2105
+ {
2106
+ "epoch": 2.32,
2107
+ "learning_rate": 5.355805243445693e-06,
2108
+ "loss": 0.192,
2109
+ "step": 6820
2110
+ },
2111
+ {
2112
+ "epoch": 2.33,
2113
+ "learning_rate": 5.342185903983657e-06,
2114
+ "loss": 0.1436,
2115
+ "step": 6840
2116
+ },
2117
+ {
2118
+ "epoch": 2.34,
2119
+ "learning_rate": 5.328566564521621e-06,
2120
+ "loss": 0.2369,
2121
+ "step": 6860
2122
+ },
2123
+ {
2124
+ "epoch": 2.34,
2125
+ "learning_rate": 5.314947225059585e-06,
2126
+ "loss": 0.1851,
2127
+ "step": 6880
2128
+ },
2129
+ {
2130
+ "epoch": 2.35,
2131
+ "learning_rate": 5.301327885597549e-06,
2132
+ "loss": 0.1813,
2133
+ "step": 6900
2134
+ },
2135
+ {
2136
+ "epoch": 2.36,
2137
+ "learning_rate": 5.287708546135513e-06,
2138
+ "loss": 0.1995,
2139
+ "step": 6920
2140
+ },
2141
+ {
2142
+ "epoch": 2.36,
2143
+ "learning_rate": 5.274089206673478e-06,
2144
+ "loss": 0.1771,
2145
+ "step": 6940
2146
+ },
2147
+ {
2148
+ "epoch": 2.37,
2149
+ "learning_rate": 5.2604698672114405e-06,
2150
+ "loss": 0.2012,
2151
+ "step": 6960
2152
+ },
2153
+ {
2154
+ "epoch": 2.38,
2155
+ "learning_rate": 5.246850527749404e-06,
2156
+ "loss": 0.2236,
2157
+ "step": 6980
2158
+ },
2159
+ {
2160
+ "epoch": 2.38,
2161
+ "learning_rate": 5.233231188287369e-06,
2162
+ "loss": 0.149,
2163
+ "step": 7000
2164
+ },
2165
+ {
2166
+ "epoch": 2.38,
2167
+ "eval_loss": 0.6401586532592773,
2168
+ "eval_roc_auc": 0.7706237424547283,
2169
+ "eval_runtime": 224.5054,
2170
+ "eval_samples_per_second": 66.413,
2171
+ "eval_steps_per_second": 8.303,
2172
+ "step": 7000
2173
+ },
2174
+ {
2175
+ "epoch": 2.39,
2176
+ "learning_rate": 5.2196118488253326e-06,
2177
+ "loss": 0.2141,
2178
+ "step": 7020
2179
+ },
2180
+ {
2181
+ "epoch": 2.4,
2182
+ "learning_rate": 5.205992509363297e-06,
2183
+ "loss": 0.1899,
2184
+ "step": 7040
2185
+ },
2186
+ {
2187
+ "epoch": 2.4,
2188
+ "learning_rate": 5.19237316990126e-06,
2189
+ "loss": 0.1737,
2190
+ "step": 7060
2191
+ },
2192
+ {
2193
+ "epoch": 2.41,
2194
+ "learning_rate": 5.178753830439224e-06,
2195
+ "loss": 0.182,
2196
+ "step": 7080
2197
+ },
2198
+ {
2199
+ "epoch": 2.42,
2200
+ "learning_rate": 5.165134490977188e-06,
2201
+ "loss": 0.1781,
2202
+ "step": 7100
2203
+ },
2204
+ {
2205
+ "epoch": 2.42,
2206
+ "learning_rate": 5.151515151515152e-06,
2207
+ "loss": 0.1758,
2208
+ "step": 7120
2209
+ },
2210
+ {
2211
+ "epoch": 2.43,
2212
+ "learning_rate": 5.137895812053117e-06,
2213
+ "loss": 0.2001,
2214
+ "step": 7140
2215
+ },
2216
+ {
2217
+ "epoch": 2.44,
2218
+ "learning_rate": 5.1242764725910795e-06,
2219
+ "loss": 0.1888,
2220
+ "step": 7160
2221
+ },
2222
+ {
2223
+ "epoch": 2.44,
2224
+ "learning_rate": 5.110657133129043e-06,
2225
+ "loss": 0.1871,
2226
+ "step": 7180
2227
+ },
2228
+ {
2229
+ "epoch": 2.45,
2230
+ "learning_rate": 5.097037793667008e-06,
2231
+ "loss": 0.2059,
2232
+ "step": 7200
2233
+ },
2234
+ {
2235
+ "epoch": 2.46,
2236
+ "learning_rate": 5.0834184542049715e-06,
2237
+ "loss": 0.2047,
2238
+ "step": 7220
2239
+ },
2240
+ {
2241
+ "epoch": 2.47,
2242
+ "learning_rate": 5.069799114742935e-06,
2243
+ "loss": 0.1985,
2244
+ "step": 7240
2245
+ },
2246
+ {
2247
+ "epoch": 2.47,
2248
+ "learning_rate": 5.0561797752809e-06,
2249
+ "loss": 0.2316,
2250
+ "step": 7260
2251
+ },
2252
+ {
2253
+ "epoch": 2.48,
2254
+ "learning_rate": 5.042560435818863e-06,
2255
+ "loss": 0.2107,
2256
+ "step": 7280
2257
+ },
2258
+ {
2259
+ "epoch": 2.49,
2260
+ "learning_rate": 5.028941096356827e-06,
2261
+ "loss": 0.1963,
2262
+ "step": 7300
2263
+ },
2264
+ {
2265
+ "epoch": 2.49,
2266
+ "learning_rate": 5.015321756894791e-06,
2267
+ "loss": 0.209,
2268
+ "step": 7320
2269
+ },
2270
+ {
2271
+ "epoch": 2.5,
2272
+ "learning_rate": 5.001702417432755e-06,
2273
+ "loss": 0.2221,
2274
+ "step": 7340
2275
+ },
2276
+ {
2277
+ "epoch": 2.51,
2278
+ "learning_rate": 4.988083077970719e-06,
2279
+ "loss": 0.1636,
2280
+ "step": 7360
2281
+ },
2282
+ {
2283
+ "epoch": 2.51,
2284
+ "learning_rate": 4.974463738508682e-06,
2285
+ "loss": 0.2369,
2286
+ "step": 7380
2287
+ },
2288
+ {
2289
+ "epoch": 2.52,
2290
+ "learning_rate": 4.960844399046647e-06,
2291
+ "loss": 0.2145,
2292
+ "step": 7400
2293
+ },
2294
+ {
2295
+ "epoch": 2.53,
2296
+ "learning_rate": 4.9472250595846105e-06,
2297
+ "loss": 0.1966,
2298
+ "step": 7420
2299
+ },
2300
+ {
2301
+ "epoch": 2.53,
2302
+ "learning_rate": 4.933605720122574e-06,
2303
+ "loss": 0.2195,
2304
+ "step": 7440
2305
+ },
2306
+ {
2307
+ "epoch": 2.54,
2308
+ "learning_rate": 4.919986380660539e-06,
2309
+ "loss": 0.1953,
2310
+ "step": 7460
2311
+ },
2312
+ {
2313
+ "epoch": 2.55,
2314
+ "learning_rate": 4.906367041198502e-06,
2315
+ "loss": 0.1959,
2316
+ "step": 7480
2317
+ },
2318
+ {
2319
+ "epoch": 2.55,
2320
+ "learning_rate": 4.892747701736466e-06,
2321
+ "loss": 0.1888,
2322
+ "step": 7500
2323
+ },
2324
+ {
2325
+ "epoch": 2.56,
2326
+ "learning_rate": 4.87912836227443e-06,
2327
+ "loss": 0.1777,
2328
+ "step": 7520
2329
+ },
2330
+ {
2331
+ "epoch": 2.57,
2332
+ "learning_rate": 4.8655090228123945e-06,
2333
+ "loss": 0.2161,
2334
+ "step": 7540
2335
+ },
2336
+ {
2337
+ "epoch": 2.57,
2338
+ "learning_rate": 4.851889683350358e-06,
2339
+ "loss": 0.2002,
2340
+ "step": 7560
2341
+ },
2342
+ {
2343
+ "epoch": 2.58,
2344
+ "learning_rate": 4.838270343888322e-06,
2345
+ "loss": 0.212,
2346
+ "step": 7580
2347
+ },
2348
+ {
2349
+ "epoch": 2.59,
2350
+ "learning_rate": 4.824651004426286e-06,
2351
+ "loss": 0.2396,
2352
+ "step": 7600
2353
+ },
2354
+ {
2355
+ "epoch": 2.59,
2356
+ "learning_rate": 4.8110316649642495e-06,
2357
+ "loss": 0.161,
2358
+ "step": 7620
2359
+ },
2360
+ {
2361
+ "epoch": 2.6,
2362
+ "learning_rate": 4.797412325502214e-06,
2363
+ "loss": 0.2192,
2364
+ "step": 7640
2365
+ },
2366
+ {
2367
+ "epoch": 2.61,
2368
+ "learning_rate": 4.783792986040177e-06,
2369
+ "loss": 0.1878,
2370
+ "step": 7660
2371
+ },
2372
+ {
2373
+ "epoch": 2.61,
2374
+ "learning_rate": 4.7701736465781415e-06,
2375
+ "loss": 0.2237,
2376
+ "step": 7680
2377
+ },
2378
+ {
2379
+ "epoch": 2.62,
2380
+ "learning_rate": 4.756554307116105e-06,
2381
+ "loss": 0.1654,
2382
+ "step": 7700
2383
+ },
2384
+ {
2385
+ "epoch": 2.63,
2386
+ "learning_rate": 4.742934967654069e-06,
2387
+ "loss": 0.1884,
2388
+ "step": 7720
2389
+ },
2390
+ {
2391
+ "epoch": 2.64,
2392
+ "learning_rate": 4.7293156281920335e-06,
2393
+ "loss": 0.2605,
2394
+ "step": 7740
2395
+ },
2396
+ {
2397
+ "epoch": 2.64,
2398
+ "learning_rate": 4.715696288729996e-06,
2399
+ "loss": 0.1947,
2400
+ "step": 7760
2401
+ },
2402
+ {
2403
+ "epoch": 2.65,
2404
+ "learning_rate": 4.702076949267961e-06,
2405
+ "loss": 0.2241,
2406
+ "step": 7780
2407
+ },
2408
+ {
2409
+ "epoch": 2.66,
2410
+ "learning_rate": 4.688457609805925e-06,
2411
+ "loss": 0.2135,
2412
+ "step": 7800
2413
+ },
2414
+ {
2415
+ "epoch": 2.66,
2416
+ "learning_rate": 4.674838270343888e-06,
2417
+ "loss": 0.1556,
2418
+ "step": 7820
2419
+ },
2420
+ {
2421
+ "epoch": 2.67,
2422
+ "learning_rate": 4.661218930881853e-06,
2423
+ "loss": 0.2006,
2424
+ "step": 7840
2425
+ },
2426
+ {
2427
+ "epoch": 2.68,
2428
+ "learning_rate": 4.647599591419817e-06,
2429
+ "loss": 0.168,
2430
+ "step": 7860
2431
+ },
2432
+ {
2433
+ "epoch": 2.68,
2434
+ "learning_rate": 4.6339802519577804e-06,
2435
+ "loss": 0.1939,
2436
+ "step": 7880
2437
+ },
2438
+ {
2439
+ "epoch": 2.69,
2440
+ "learning_rate": 4.620360912495744e-06,
2441
+ "loss": 0.1877,
2442
+ "step": 7900
2443
+ },
2444
+ {
2445
+ "epoch": 2.7,
2446
+ "learning_rate": 4.606741573033709e-06,
2447
+ "loss": 0.2254,
2448
+ "step": 7920
2449
+ },
2450
+ {
2451
+ "epoch": 2.7,
2452
+ "learning_rate": 4.5931222335716725e-06,
2453
+ "loss": 0.1725,
2454
+ "step": 7940
2455
+ },
2456
+ {
2457
+ "epoch": 2.71,
2458
+ "learning_rate": 4.579502894109636e-06,
2459
+ "loss": 0.2169,
2460
+ "step": 7960
2461
+ },
2462
+ {
2463
+ "epoch": 2.72,
2464
+ "learning_rate": 4.5658835546476e-06,
2465
+ "loss": 0.1552,
2466
+ "step": 7980
2467
+ },
2468
+ {
2469
+ "epoch": 2.72,
2470
+ "learning_rate": 4.552264215185564e-06,
2471
+ "loss": 0.1766,
2472
+ "step": 8000
2473
+ },
2474
+ {
2475
+ "epoch": 2.72,
2476
+ "eval_loss": 0.6767948865890503,
2477
+ "eval_roc_auc": 0.7610328638497652,
2478
+ "eval_runtime": 224.0971,
2479
+ "eval_samples_per_second": 66.534,
2480
+ "eval_steps_per_second": 8.318,
2481
+ "step": 8000
2482
+ },
2483
+ {
2484
+ "epoch": 2.73,
2485
+ "learning_rate": 4.538644875723528e-06,
2486
+ "loss": 0.1792,
2487
+ "step": 8020
2488
+ },
2489
+ {
2490
+ "epoch": 2.74,
2491
+ "learning_rate": 4.525025536261491e-06,
2492
+ "loss": 0.2031,
2493
+ "step": 8040
2494
+ },
2495
+ {
2496
+ "epoch": 2.74,
2497
+ "learning_rate": 4.511406196799456e-06,
2498
+ "loss": 0.1644,
2499
+ "step": 8060
2500
+ },
2501
+ {
2502
+ "epoch": 2.75,
2503
+ "learning_rate": 4.497786857337419e-06,
2504
+ "loss": 0.1895,
2505
+ "step": 8080
2506
+ },
2507
+ {
2508
+ "epoch": 2.76,
2509
+ "learning_rate": 4.484167517875383e-06,
2510
+ "loss": 0.1915,
2511
+ "step": 8100
2512
+ },
2513
+ {
2514
+ "epoch": 2.76,
2515
+ "learning_rate": 4.470548178413348e-06,
2516
+ "loss": 0.1828,
2517
+ "step": 8120
2518
+ },
2519
+ {
2520
+ "epoch": 2.77,
2521
+ "learning_rate": 4.456928838951311e-06,
2522
+ "loss": 0.2172,
2523
+ "step": 8140
2524
+ },
2525
+ {
2526
+ "epoch": 2.78,
2527
+ "learning_rate": 4.443309499489275e-06,
2528
+ "loss": 0.1817,
2529
+ "step": 8160
2530
+ },
2531
+ {
2532
+ "epoch": 2.79,
2533
+ "learning_rate": 4.429690160027239e-06,
2534
+ "loss": 0.1833,
2535
+ "step": 8180
2536
+ },
2537
+ {
2538
+ "epoch": 2.79,
2539
+ "learning_rate": 4.4160708205652035e-06,
2540
+ "loss": 0.1648,
2541
+ "step": 8200
2542
+ },
2543
+ {
2544
+ "epoch": 2.8,
2545
+ "learning_rate": 4.402451481103167e-06,
2546
+ "loss": 0.1905,
2547
+ "step": 8220
2548
+ },
2549
+ {
2550
+ "epoch": 2.81,
2551
+ "learning_rate": 4.388832141641131e-06,
2552
+ "loss": 0.1964,
2553
+ "step": 8240
2554
+ },
2555
+ {
2556
+ "epoch": 2.81,
2557
+ "learning_rate": 4.375212802179095e-06,
2558
+ "loss": 0.2237,
2559
+ "step": 8260
2560
+ },
2561
+ {
2562
+ "epoch": 2.82,
2563
+ "learning_rate": 4.361593462717058e-06,
2564
+ "loss": 0.1929,
2565
+ "step": 8280
2566
+ },
2567
+ {
2568
+ "epoch": 2.83,
2569
+ "learning_rate": 4.347974123255023e-06,
2570
+ "loss": 0.2196,
2571
+ "step": 8300
2572
+ },
2573
+ {
2574
+ "epoch": 2.83,
2575
+ "learning_rate": 4.334354783792986e-06,
2576
+ "loss": 0.1876,
2577
+ "step": 8320
2578
+ },
2579
+ {
2580
+ "epoch": 2.84,
2581
+ "learning_rate": 4.32073544433095e-06,
2582
+ "loss": 0.2213,
2583
+ "step": 8340
2584
+ },
2585
+ {
2586
+ "epoch": 2.85,
2587
+ "learning_rate": 4.307116104868914e-06,
2588
+ "loss": 0.2043,
2589
+ "step": 8360
2590
+ },
2591
+ {
2592
+ "epoch": 2.85,
2593
+ "learning_rate": 4.293496765406878e-06,
2594
+ "loss": 0.1865,
2595
+ "step": 8380
2596
+ },
2597
+ {
2598
+ "epoch": 2.86,
2599
+ "learning_rate": 4.2798774259448424e-06,
2600
+ "loss": 0.2067,
2601
+ "step": 8400
2602
+ },
2603
+ {
2604
+ "epoch": 2.87,
2605
+ "learning_rate": 4.266258086482805e-06,
2606
+ "loss": 0.1983,
2607
+ "step": 8420
2608
+ },
2609
+ {
2610
+ "epoch": 2.87,
2611
+ "learning_rate": 4.25263874702077e-06,
2612
+ "loss": 0.1724,
2613
+ "step": 8440
2614
+ },
2615
+ {
2616
+ "epoch": 2.88,
2617
+ "learning_rate": 4.239019407558734e-06,
2618
+ "loss": 0.2027,
2619
+ "step": 8460
2620
+ },
2621
+ {
2622
+ "epoch": 2.89,
2623
+ "learning_rate": 4.225400068096698e-06,
2624
+ "loss": 0.1854,
2625
+ "step": 8480
2626
+ },
2627
+ {
2628
+ "epoch": 2.89,
2629
+ "learning_rate": 4.211780728634662e-06,
2630
+ "loss": 0.1886,
2631
+ "step": 8500
2632
+ },
2633
+ {
2634
+ "epoch": 2.9,
2635
+ "learning_rate": 4.198161389172626e-06,
2636
+ "loss": 0.1882,
2637
+ "step": 8520
2638
+ },
2639
+ {
2640
+ "epoch": 2.91,
2641
+ "learning_rate": 4.184542049710589e-06,
2642
+ "loss": 0.221,
2643
+ "step": 8540
2644
+ },
2645
+ {
2646
+ "epoch": 2.91,
2647
+ "learning_rate": 4.170922710248553e-06,
2648
+ "loss": 0.1875,
2649
+ "step": 8560
2650
+ },
2651
+ {
2652
+ "epoch": 2.92,
2653
+ "learning_rate": 4.157303370786518e-06,
2654
+ "loss": 0.162,
2655
+ "step": 8580
2656
+ },
2657
+ {
2658
+ "epoch": 2.93,
2659
+ "learning_rate": 4.1436840313244805e-06,
2660
+ "loss": 0.1633,
2661
+ "step": 8600
2662
+ },
2663
+ {
2664
+ "epoch": 2.93,
2665
+ "learning_rate": 4.130064691862445e-06,
2666
+ "loss": 0.2134,
2667
+ "step": 8620
2668
+ },
2669
+ {
2670
+ "epoch": 2.94,
2671
+ "learning_rate": 4.116445352400409e-06,
2672
+ "loss": 0.2259,
2673
+ "step": 8640
2674
+ },
2675
+ {
2676
+ "epoch": 2.95,
2677
+ "learning_rate": 4.102826012938373e-06,
2678
+ "loss": 0.1672,
2679
+ "step": 8660
2680
+ },
2681
+ {
2682
+ "epoch": 2.96,
2683
+ "learning_rate": 4.089206673476337e-06,
2684
+ "loss": 0.1891,
2685
+ "step": 8680
2686
+ },
2687
+ {
2688
+ "epoch": 2.96,
2689
+ "learning_rate": 4.0755873340143e-06,
2690
+ "loss": 0.1633,
2691
+ "step": 8700
2692
+ },
2693
+ {
2694
+ "epoch": 2.97,
2695
+ "learning_rate": 4.061967994552265e-06,
2696
+ "loss": 0.192,
2697
+ "step": 8720
2698
+ },
2699
+ {
2700
+ "epoch": 2.98,
2701
+ "learning_rate": 4.048348655090228e-06,
2702
+ "loss": 0.1875,
2703
+ "step": 8740
2704
+ },
2705
+ {
2706
+ "epoch": 2.98,
2707
+ "learning_rate": 4.034729315628192e-06,
2708
+ "loss": 0.1698,
2709
+ "step": 8760
2710
+ },
2711
+ {
2712
+ "epoch": 2.99,
2713
+ "learning_rate": 4.021109976166157e-06,
2714
+ "loss": 0.1905,
2715
+ "step": 8780
2716
+ },
2717
+ {
2718
+ "epoch": 3.0,
2719
+ "learning_rate": 4.00749063670412e-06,
2720
+ "loss": 0.1735,
2721
+ "step": 8800
2722
+ },
2723
+ {
2724
+ "epoch": 3.0,
2725
+ "learning_rate": 3.993871297242084e-06,
2726
+ "loss": 0.1437,
2727
+ "step": 8820
2728
+ },
2729
+ {
2730
+ "epoch": 3.01,
2731
+ "learning_rate": 3.980251957780048e-06,
2732
+ "loss": 0.1843,
2733
+ "step": 8840
2734
+ },
2735
+ {
2736
+ "epoch": 3.02,
2737
+ "learning_rate": 3.966632618318012e-06,
2738
+ "loss": 0.1405,
2739
+ "step": 8860
2740
+ },
2741
+ {
2742
+ "epoch": 3.02,
2743
+ "learning_rate": 3.953013278855976e-06,
2744
+ "loss": 0.1279,
2745
+ "step": 8880
2746
+ },
2747
+ {
2748
+ "epoch": 3.03,
2749
+ "learning_rate": 3.93939393939394e-06,
2750
+ "loss": 0.1407,
2751
+ "step": 8900
2752
+ },
2753
+ {
2754
+ "epoch": 3.04,
2755
+ "learning_rate": 3.9257745999319036e-06,
2756
+ "loss": 0.1751,
2757
+ "step": 8920
2758
+ },
2759
+ {
2760
+ "epoch": 3.04,
2761
+ "learning_rate": 3.912155260469867e-06,
2762
+ "loss": 0.1507,
2763
+ "step": 8940
2764
+ },
2765
+ {
2766
+ "epoch": 3.05,
2767
+ "learning_rate": 3.898535921007832e-06,
2768
+ "loss": 0.1636,
2769
+ "step": 8960
2770
+ },
2771
+ {
2772
+ "epoch": 3.06,
2773
+ "learning_rate": 3.884916581545795e-06,
2774
+ "loss": 0.1631,
2775
+ "step": 8980
2776
+ },
2777
+ {
2778
+ "epoch": 3.06,
2779
+ "learning_rate": 3.871297242083759e-06,
2780
+ "loss": 0.1496,
2781
+ "step": 9000
2782
+ },
2783
+ {
2784
+ "epoch": 3.06,
2785
+ "eval_loss": 0.6239368915557861,
2786
+ "eval_roc_auc": 0.7733065057008719,
2787
+ "eval_runtime": 227.7765,
2788
+ "eval_samples_per_second": 65.459,
2789
+ "eval_steps_per_second": 8.183,
2790
+ "step": 9000
2791
+ },
2792
+ {
2793
+ "epoch": 3.07,
2794
+ "learning_rate": 3.857677902621723e-06,
2795
+ "loss": 0.1598,
2796
+ "step": 9020
2797
+ },
2798
+ {
2799
+ "epoch": 3.08,
2800
+ "learning_rate": 3.844058563159687e-06,
2801
+ "loss": 0.1247,
2802
+ "step": 9040
2803
+ },
2804
+ {
2805
+ "epoch": 3.08,
2806
+ "learning_rate": 3.830439223697651e-06,
2807
+ "loss": 0.1712,
2808
+ "step": 9060
2809
+ },
2810
+ {
2811
+ "epoch": 3.09,
2812
+ "learning_rate": 3.816819884235615e-06,
2813
+ "loss": 0.1623,
2814
+ "step": 9080
2815
+ },
2816
+ {
2817
+ "epoch": 3.1,
2818
+ "learning_rate": 3.803200544773579e-06,
2819
+ "loss": 0.1359,
2820
+ "step": 9100
2821
+ },
2822
+ {
2823
+ "epoch": 3.11,
2824
+ "learning_rate": 3.7895812053115425e-06,
2825
+ "loss": 0.1476,
2826
+ "step": 9120
2827
+ },
2828
+ {
2829
+ "epoch": 3.11,
2830
+ "learning_rate": 3.7759618658495067e-06,
2831
+ "loss": 0.1505,
2832
+ "step": 9140
2833
+ },
2834
+ {
2835
+ "epoch": 3.12,
2836
+ "learning_rate": 3.762342526387471e-06,
2837
+ "loss": 0.1779,
2838
+ "step": 9160
2839
+ },
2840
+ {
2841
+ "epoch": 3.13,
2842
+ "learning_rate": 3.748723186925434e-06,
2843
+ "loss": 0.1288,
2844
+ "step": 9180
2845
+ },
2846
+ {
2847
+ "epoch": 3.13,
2848
+ "learning_rate": 3.7351038474633983e-06,
2849
+ "loss": 0.1545,
2850
+ "step": 9200
2851
+ },
2852
+ {
2853
+ "epoch": 3.14,
2854
+ "learning_rate": 3.7214845080013624e-06,
2855
+ "loss": 0.1543,
2856
+ "step": 9220
2857
+ },
2858
+ {
2859
+ "epoch": 3.15,
2860
+ "learning_rate": 3.707865168539326e-06,
2861
+ "loss": 0.1486,
2862
+ "step": 9240
2863
+ },
2864
+ {
2865
+ "epoch": 3.15,
2866
+ "learning_rate": 3.69424582907729e-06,
2867
+ "loss": 0.1456,
2868
+ "step": 9260
2869
+ },
2870
+ {
2871
+ "epoch": 3.16,
2872
+ "learning_rate": 3.6806264896152536e-06,
2873
+ "loss": 0.1582,
2874
+ "step": 9280
2875
+ },
2876
+ {
2877
+ "epoch": 3.17,
2878
+ "learning_rate": 3.6670071501532178e-06,
2879
+ "loss": 0.1314,
2880
+ "step": 9300
2881
+ },
2882
+ {
2883
+ "epoch": 3.17,
2884
+ "learning_rate": 3.653387810691182e-06,
2885
+ "loss": 0.1784,
2886
+ "step": 9320
2887
+ },
2888
+ {
2889
+ "epoch": 3.18,
2890
+ "learning_rate": 3.639768471229146e-06,
2891
+ "loss": 0.1799,
2892
+ "step": 9340
2893
+ },
2894
+ {
2895
+ "epoch": 3.19,
2896
+ "learning_rate": 3.6261491317671094e-06,
2897
+ "loss": 0.1634,
2898
+ "step": 9360
2899
+ },
2900
+ {
2901
+ "epoch": 3.19,
2902
+ "learning_rate": 3.6125297923050735e-06,
2903
+ "loss": 0.1668,
2904
+ "step": 9380
2905
+ },
2906
+ {
2907
+ "epoch": 3.2,
2908
+ "learning_rate": 3.5989104528430373e-06,
2909
+ "loss": 0.1717,
2910
+ "step": 9400
2911
+ },
2912
+ {
2913
+ "epoch": 3.21,
2914
+ "learning_rate": 3.5852911133810014e-06,
2915
+ "loss": 0.1076,
2916
+ "step": 9420
2917
+ },
2918
+ {
2919
+ "epoch": 3.21,
2920
+ "learning_rate": 3.5716717739189656e-06,
2921
+ "loss": 0.1087,
2922
+ "step": 9440
2923
+ },
2924
+ {
2925
+ "epoch": 3.22,
2926
+ "learning_rate": 3.558052434456929e-06,
2927
+ "loss": 0.158,
2928
+ "step": 9460
2929
+ },
2930
+ {
2931
+ "epoch": 3.23,
2932
+ "learning_rate": 3.544433094994893e-06,
2933
+ "loss": 0.1229,
2934
+ "step": 9480
2935
+ },
2936
+ {
2937
+ "epoch": 3.23,
2938
+ "learning_rate": 3.530813755532857e-06,
2939
+ "loss": 0.1428,
2940
+ "step": 9500
2941
+ },
2942
+ {
2943
+ "epoch": 3.24,
2944
+ "learning_rate": 3.517194416070821e-06,
2945
+ "loss": 0.1513,
2946
+ "step": 9520
2947
+ },
2948
+ {
2949
+ "epoch": 3.25,
2950
+ "learning_rate": 3.5035750766087846e-06,
2951
+ "loss": 0.1302,
2952
+ "step": 9540
2953
+ },
2954
+ {
2955
+ "epoch": 3.26,
2956
+ "learning_rate": 3.4899557371467483e-06,
2957
+ "loss": 0.1416,
2958
+ "step": 9560
2959
+ },
2960
+ {
2961
+ "epoch": 3.26,
2962
+ "learning_rate": 3.4763363976847125e-06,
2963
+ "loss": 0.1926,
2964
+ "step": 9580
2965
+ },
2966
+ {
2967
+ "epoch": 3.27,
2968
+ "learning_rate": 3.4627170582226766e-06,
2969
+ "loss": 0.1431,
2970
+ "step": 9600
2971
+ },
2972
+ {
2973
+ "epoch": 3.28,
2974
+ "learning_rate": 3.449097718760641e-06,
2975
+ "loss": 0.1235,
2976
+ "step": 9620
2977
+ },
2978
+ {
2979
+ "epoch": 3.28,
2980
+ "learning_rate": 3.435478379298604e-06,
2981
+ "loss": 0.1897,
2982
+ "step": 9640
2983
+ },
2984
+ {
2985
+ "epoch": 3.29,
2986
+ "learning_rate": 3.4218590398365683e-06,
2987
+ "loss": 0.1464,
2988
+ "step": 9660
2989
+ },
2990
+ {
2991
+ "epoch": 3.3,
2992
+ "learning_rate": 3.408239700374532e-06,
2993
+ "loss": 0.1889,
2994
+ "step": 9680
2995
+ },
2996
+ {
2997
+ "epoch": 3.3,
2998
+ "learning_rate": 3.394620360912496e-06,
2999
+ "loss": 0.1379,
3000
+ "step": 9700
3001
+ },
3002
+ {
3003
+ "epoch": 3.31,
3004
+ "learning_rate": 3.3810010214504603e-06,
3005
+ "loss": 0.1677,
3006
+ "step": 9720
3007
+ },
3008
+ {
3009
+ "epoch": 3.32,
3010
+ "learning_rate": 3.3673816819884236e-06,
3011
+ "loss": 0.1461,
3012
+ "step": 9740
3013
+ },
3014
+ {
3015
+ "epoch": 3.32,
3016
+ "learning_rate": 3.3537623425263877e-06,
3017
+ "loss": 0.1661,
3018
+ "step": 9760
3019
+ },
3020
+ {
3021
+ "epoch": 3.33,
3022
+ "learning_rate": 3.340143003064352e-06,
3023
+ "loss": 0.1571,
3024
+ "step": 9780
3025
+ },
3026
+ {
3027
+ "epoch": 3.34,
3028
+ "learning_rate": 3.3265236636023156e-06,
3029
+ "loss": 0.157,
3030
+ "step": 9800
3031
+ },
3032
+ {
3033
+ "epoch": 3.34,
3034
+ "learning_rate": 3.3129043241402798e-06,
3035
+ "loss": 0.1774,
3036
+ "step": 9820
3037
+ },
3038
+ {
3039
+ "epoch": 3.35,
3040
+ "learning_rate": 3.299284984678243e-06,
3041
+ "loss": 0.1463,
3042
+ "step": 9840
3043
+ },
3044
+ {
3045
+ "epoch": 3.36,
3046
+ "learning_rate": 3.2856656452162072e-06,
3047
+ "loss": 0.1568,
3048
+ "step": 9860
3049
+ },
3050
+ {
3051
+ "epoch": 3.36,
3052
+ "learning_rate": 3.2720463057541714e-06,
3053
+ "loss": 0.1165,
3054
+ "step": 9880
3055
+ },
3056
+ {
3057
+ "epoch": 3.37,
3058
+ "learning_rate": 3.258426966292135e-06,
3059
+ "loss": 0.1414,
3060
+ "step": 9900
3061
+ },
3062
+ {
3063
+ "epoch": 3.38,
3064
+ "learning_rate": 3.244807626830099e-06,
3065
+ "loss": 0.133,
3066
+ "step": 9920
3067
+ },
3068
+ {
3069
+ "epoch": 3.38,
3070
+ "learning_rate": 3.231188287368063e-06,
3071
+ "loss": 0.1348,
3072
+ "step": 9940
3073
+ },
3074
+ {
3075
+ "epoch": 3.39,
3076
+ "learning_rate": 3.2175689479060267e-06,
3077
+ "loss": 0.1444,
3078
+ "step": 9960
3079
+ },
3080
+ {
3081
+ "epoch": 3.4,
3082
+ "learning_rate": 3.203949608443991e-06,
3083
+ "loss": 0.1566,
3084
+ "step": 9980
3085
+ },
3086
+ {
3087
+ "epoch": 3.4,
3088
+ "learning_rate": 3.190330268981955e-06,
3089
+ "loss": 0.155,
3090
+ "step": 10000
3091
+ },
3092
+ {
3093
+ "epoch": 3.4,
3094
+ "eval_loss": 0.7332798838615417,
3095
+ "eval_roc_auc": 0.7601609657947686,
3096
+ "eval_runtime": 222.7036,
3097
+ "eval_samples_per_second": 66.95,
3098
+ "eval_steps_per_second": 8.37,
3099
+ "step": 10000
3100
+ },
3101
+ {
3102
+ "epoch": 3.41,
3103
+ "learning_rate": 3.1767109295199183e-06,
3104
+ "loss": 0.1454,
3105
+ "step": 10020
3106
+ },
3107
+ {
3108
+ "epoch": 3.42,
3109
+ "learning_rate": 3.1630915900578825e-06,
3110
+ "loss": 0.1361,
3111
+ "step": 10040
3112
+ },
3113
+ {
3114
+ "epoch": 3.43,
3115
+ "learning_rate": 3.149472250595846e-06,
3116
+ "loss": 0.153,
3117
+ "step": 10060
3118
+ },
3119
+ {
3120
+ "epoch": 3.43,
3121
+ "learning_rate": 3.1358529111338103e-06,
3122
+ "loss": 0.1855,
3123
+ "step": 10080
3124
+ },
3125
+ {
3126
+ "epoch": 3.44,
3127
+ "learning_rate": 3.1222335716717745e-06,
3128
+ "loss": 0.1164,
3129
+ "step": 10100
3130
+ },
3131
+ {
3132
+ "epoch": 3.45,
3133
+ "learning_rate": 3.1086142322097378e-06,
3134
+ "loss": 0.1846,
3135
+ "step": 10120
3136
+ },
3137
+ {
3138
+ "epoch": 3.45,
3139
+ "learning_rate": 3.094994892747702e-06,
3140
+ "loss": 0.1596,
3141
+ "step": 10140
3142
+ },
3143
+ {
3144
+ "epoch": 3.46,
3145
+ "learning_rate": 3.081375553285666e-06,
3146
+ "loss": 0.1693,
3147
+ "step": 10160
3148
+ },
3149
+ {
3150
+ "epoch": 3.47,
3151
+ "learning_rate": 3.06775621382363e-06,
3152
+ "loss": 0.1359,
3153
+ "step": 10180
3154
+ },
3155
+ {
3156
+ "epoch": 3.47,
3157
+ "learning_rate": 3.0541368743615935e-06,
3158
+ "loss": 0.17,
3159
+ "step": 10200
3160
+ },
3161
+ {
3162
+ "epoch": 3.48,
3163
+ "learning_rate": 3.0405175348995573e-06,
3164
+ "loss": 0.1185,
3165
+ "step": 10220
3166
+ },
3167
+ {
3168
+ "epoch": 3.49,
3169
+ "learning_rate": 3.0268981954375214e-06,
3170
+ "loss": 0.111,
3171
+ "step": 10240
3172
+ },
3173
+ {
3174
+ "epoch": 3.49,
3175
+ "learning_rate": 3.0132788559754856e-06,
3176
+ "loss": 0.1632,
3177
+ "step": 10260
3178
+ },
3179
+ {
3180
+ "epoch": 3.5,
3181
+ "learning_rate": 2.9996595165134497e-06,
3182
+ "loss": 0.1217,
3183
+ "step": 10280
3184
+ },
3185
+ {
3186
+ "epoch": 3.51,
3187
+ "learning_rate": 2.986040177051413e-06,
3188
+ "loss": 0.1515,
3189
+ "step": 10300
3190
+ },
3191
+ {
3192
+ "epoch": 3.51,
3193
+ "learning_rate": 2.972420837589377e-06,
3194
+ "loss": 0.1249,
3195
+ "step": 10320
3196
+ },
3197
+ {
3198
+ "epoch": 3.52,
3199
+ "learning_rate": 2.958801498127341e-06,
3200
+ "loss": 0.1566,
3201
+ "step": 10340
3202
+ },
3203
+ {
3204
+ "epoch": 3.53,
3205
+ "learning_rate": 2.945182158665305e-06,
3206
+ "loss": 0.1707,
3207
+ "step": 10360
3208
+ },
3209
+ {
3210
+ "epoch": 3.53,
3211
+ "learning_rate": 2.931562819203269e-06,
3212
+ "loss": 0.1395,
3213
+ "step": 10380
3214
+ },
3215
+ {
3216
+ "epoch": 3.54,
3217
+ "learning_rate": 2.9179434797412325e-06,
3218
+ "loss": 0.1631,
3219
+ "step": 10400
3220
+ },
3221
+ {
3222
+ "epoch": 3.55,
3223
+ "learning_rate": 2.9043241402791967e-06,
3224
+ "loss": 0.1359,
3225
+ "step": 10420
3226
+ },
3227
+ {
3228
+ "epoch": 3.55,
3229
+ "learning_rate": 2.890704800817161e-06,
3230
+ "loss": 0.1413,
3231
+ "step": 10440
3232
+ },
3233
+ {
3234
+ "epoch": 3.56,
3235
+ "learning_rate": 2.8770854613551245e-06,
3236
+ "loss": 0.1415,
3237
+ "step": 10460
3238
+ },
3239
+ {
3240
+ "epoch": 3.57,
3241
+ "learning_rate": 2.8634661218930883e-06,
3242
+ "loss": 0.1439,
3243
+ "step": 10480
3244
+ },
3245
+ {
3246
+ "epoch": 3.58,
3247
+ "learning_rate": 2.849846782431052e-06,
3248
+ "loss": 0.1997,
3249
+ "step": 10500
3250
+ },
3251
+ {
3252
+ "epoch": 3.58,
3253
+ "learning_rate": 2.836227442969016e-06,
3254
+ "loss": 0.1727,
3255
+ "step": 10520
3256
+ },
3257
+ {
3258
+ "epoch": 3.59,
3259
+ "learning_rate": 2.8226081035069803e-06,
3260
+ "loss": 0.1468,
3261
+ "step": 10540
3262
+ },
3263
+ {
3264
+ "epoch": 3.6,
3265
+ "learning_rate": 2.8089887640449444e-06,
3266
+ "loss": 0.1451,
3267
+ "step": 10560
3268
+ },
3269
+ {
3270
+ "epoch": 3.6,
3271
+ "learning_rate": 2.7953694245829077e-06,
3272
+ "loss": 0.1466,
3273
+ "step": 10580
3274
+ },
3275
+ {
3276
+ "epoch": 3.61,
3277
+ "learning_rate": 2.781750085120872e-06,
3278
+ "loss": 0.1434,
3279
+ "step": 10600
3280
+ },
3281
+ {
3282
+ "epoch": 3.62,
3283
+ "learning_rate": 2.7681307456588356e-06,
3284
+ "loss": 0.1302,
3285
+ "step": 10620
3286
+ },
3287
+ {
3288
+ "epoch": 3.62,
3289
+ "learning_rate": 2.7545114061967998e-06,
3290
+ "loss": 0.1318,
3291
+ "step": 10640
3292
+ },
3293
+ {
3294
+ "epoch": 3.63,
3295
+ "learning_rate": 2.740892066734764e-06,
3296
+ "loss": 0.1516,
3297
+ "step": 10660
3298
+ },
3299
+ {
3300
+ "epoch": 3.64,
3301
+ "learning_rate": 2.7272727272727272e-06,
3302
+ "loss": 0.1508,
3303
+ "step": 10680
3304
+ },
3305
+ {
3306
+ "epoch": 3.64,
3307
+ "learning_rate": 2.7136533878106914e-06,
3308
+ "loss": 0.1838,
3309
+ "step": 10700
3310
+ },
3311
+ {
3312
+ "epoch": 3.65,
3313
+ "learning_rate": 2.7000340483486555e-06,
3314
+ "loss": 0.1497,
3315
+ "step": 10720
3316
+ },
3317
+ {
3318
+ "epoch": 3.66,
3319
+ "learning_rate": 2.6864147088866193e-06,
3320
+ "loss": 0.1694,
3321
+ "step": 10740
3322
+ },
3323
+ {
3324
+ "epoch": 3.66,
3325
+ "learning_rate": 2.6727953694245834e-06,
3326
+ "loss": 0.1789,
3327
+ "step": 10760
3328
+ },
3329
+ {
3330
+ "epoch": 3.67,
3331
+ "learning_rate": 2.6591760299625467e-06,
3332
+ "loss": 0.1369,
3333
+ "step": 10780
3334
+ },
3335
+ {
3336
+ "epoch": 3.68,
3337
+ "learning_rate": 2.645556690500511e-06,
3338
+ "loss": 0.1195,
3339
+ "step": 10800
3340
+ },
3341
+ {
3342
+ "epoch": 3.68,
3343
+ "learning_rate": 2.631937351038475e-06,
3344
+ "loss": 0.1486,
3345
+ "step": 10820
3346
+ },
3347
+ {
3348
+ "epoch": 3.69,
3349
+ "learning_rate": 2.6183180115764387e-06,
3350
+ "loss": 0.1374,
3351
+ "step": 10840
3352
+ },
3353
+ {
3354
+ "epoch": 3.7,
3355
+ "learning_rate": 2.6046986721144025e-06,
3356
+ "loss": 0.1454,
3357
+ "step": 10860
3358
+ },
3359
+ {
3360
+ "epoch": 3.7,
3361
+ "learning_rate": 2.5910793326523666e-06,
3362
+ "loss": 0.1969,
3363
+ "step": 10880
3364
+ },
3365
+ {
3366
+ "epoch": 3.71,
3367
+ "learning_rate": 2.5774599931903303e-06,
3368
+ "loss": 0.1388,
3369
+ "step": 10900
3370
+ },
3371
+ {
3372
+ "epoch": 3.72,
3373
+ "learning_rate": 2.5638406537282945e-06,
3374
+ "loss": 0.1714,
3375
+ "step": 10920
3376
+ },
3377
+ {
3378
+ "epoch": 3.72,
3379
+ "learning_rate": 2.5502213142662586e-06,
3380
+ "loss": 0.1339,
3381
+ "step": 10940
3382
+ },
3383
+ {
3384
+ "epoch": 3.73,
3385
+ "learning_rate": 2.536601974804222e-06,
3386
+ "loss": 0.1532,
3387
+ "step": 10960
3388
+ },
3389
+ {
3390
+ "epoch": 3.74,
3391
+ "learning_rate": 2.522982635342186e-06,
3392
+ "loss": 0.1918,
3393
+ "step": 10980
3394
+ },
3395
+ {
3396
+ "epoch": 3.75,
3397
+ "learning_rate": 2.5093632958801502e-06,
3398
+ "loss": 0.1238,
3399
+ "step": 11000
3400
+ },
3401
+ {
3402
+ "epoch": 3.75,
3403
+ "eval_loss": 0.6513485312461853,
3404
+ "eval_roc_auc": 0.7726358148893361,
3405
+ "eval_runtime": 227.7851,
3406
+ "eval_samples_per_second": 65.456,
3407
+ "eval_steps_per_second": 8.183,
3408
+ "step": 11000
3409
+ },
3410
+ {
3411
+ "epoch": 3.75,
3412
+ "learning_rate": 2.495743956418114e-06,
3413
+ "loss": 0.169,
3414
+ "step": 11020
3415
+ },
3416
+ {
3417
+ "epoch": 3.76,
3418
+ "learning_rate": 2.4821246169560777e-06,
3419
+ "loss": 0.1273,
3420
+ "step": 11040
3421
+ },
3422
+ {
3423
+ "epoch": 3.77,
3424
+ "learning_rate": 2.468505277494042e-06,
3425
+ "loss": 0.1087,
3426
+ "step": 11060
3427
+ },
3428
+ {
3429
+ "epoch": 3.77,
3430
+ "learning_rate": 2.4548859380320056e-06,
3431
+ "loss": 0.1648,
3432
+ "step": 11080
3433
+ },
3434
+ {
3435
+ "epoch": 3.78,
3436
+ "learning_rate": 2.4412665985699697e-06,
3437
+ "loss": 0.1748,
3438
+ "step": 11100
3439
+ },
3440
+ {
3441
+ "epoch": 3.79,
3442
+ "learning_rate": 2.4276472591079335e-06,
3443
+ "loss": 0.1964,
3444
+ "step": 11120
3445
+ },
3446
+ {
3447
+ "epoch": 3.79,
3448
+ "learning_rate": 2.414027919645897e-06,
3449
+ "loss": 0.1551,
3450
+ "step": 11140
3451
+ },
3452
+ {
3453
+ "epoch": 3.8,
3454
+ "learning_rate": 2.4004085801838613e-06,
3455
+ "loss": 0.175,
3456
+ "step": 11160
3457
+ },
3458
+ {
3459
+ "epoch": 3.81,
3460
+ "learning_rate": 2.386789240721825e-06,
3461
+ "loss": 0.1273,
3462
+ "step": 11180
3463
+ },
3464
+ {
3465
+ "epoch": 3.81,
3466
+ "learning_rate": 2.3731699012597892e-06,
3467
+ "loss": 0.1746,
3468
+ "step": 11200
3469
+ },
3470
+ {
3471
+ "epoch": 3.82,
3472
+ "learning_rate": 2.359550561797753e-06,
3473
+ "loss": 0.1182,
3474
+ "step": 11220
3475
+ },
3476
+ {
3477
+ "epoch": 3.83,
3478
+ "learning_rate": 2.345931222335717e-06,
3479
+ "loss": 0.1066,
3480
+ "step": 11240
3481
+ },
3482
+ {
3483
+ "epoch": 3.83,
3484
+ "learning_rate": 2.332311882873681e-06,
3485
+ "loss": 0.1269,
3486
+ "step": 11260
3487
+ },
3488
+ {
3489
+ "epoch": 3.84,
3490
+ "learning_rate": 2.3186925434116445e-06,
3491
+ "loss": 0.0901,
3492
+ "step": 11280
3493
+ },
3494
+ {
3495
+ "epoch": 3.85,
3496
+ "learning_rate": 2.3050732039496087e-06,
3497
+ "loss": 0.206,
3498
+ "step": 11300
3499
+ },
3500
+ {
3501
+ "epoch": 3.85,
3502
+ "learning_rate": 2.2914538644875724e-06,
3503
+ "loss": 0.1164,
3504
+ "step": 11320
3505
+ },
3506
+ {
3507
+ "epoch": 3.86,
3508
+ "learning_rate": 2.2778345250255366e-06,
3509
+ "loss": 0.1698,
3510
+ "step": 11340
3511
+ },
3512
+ {
3513
+ "epoch": 3.87,
3514
+ "learning_rate": 2.2642151855635003e-06,
3515
+ "loss": 0.1279,
3516
+ "step": 11360
3517
+ },
3518
+ {
3519
+ "epoch": 3.87,
3520
+ "learning_rate": 2.2505958461014645e-06,
3521
+ "loss": 0.1477,
3522
+ "step": 11380
3523
+ },
3524
+ {
3525
+ "epoch": 3.88,
3526
+ "learning_rate": 2.236976506639428e-06,
3527
+ "loss": 0.136,
3528
+ "step": 11400
3529
+ },
3530
+ {
3531
+ "epoch": 3.89,
3532
+ "learning_rate": 2.223357167177392e-06,
3533
+ "loss": 0.122,
3534
+ "step": 11420
3535
+ },
3536
+ {
3537
+ "epoch": 3.9,
3538
+ "learning_rate": 2.209737827715356e-06,
3539
+ "loss": 0.138,
3540
+ "step": 11440
3541
+ },
3542
+ {
3543
+ "epoch": 3.9,
3544
+ "learning_rate": 2.1961184882533198e-06,
3545
+ "loss": 0.161,
3546
+ "step": 11460
3547
+ },
3548
+ {
3549
+ "epoch": 3.91,
3550
+ "learning_rate": 2.182499148791284e-06,
3551
+ "loss": 0.1685,
3552
+ "step": 11480
3553
+ },
3554
+ {
3555
+ "epoch": 3.92,
3556
+ "learning_rate": 2.1688798093292477e-06,
3557
+ "loss": 0.1503,
3558
+ "step": 11500
3559
+ },
3560
+ {
3561
+ "epoch": 3.92,
3562
+ "learning_rate": 2.155260469867212e-06,
3563
+ "loss": 0.1933,
3564
+ "step": 11520
3565
+ },
3566
+ {
3567
+ "epoch": 3.93,
3568
+ "learning_rate": 2.1416411304051755e-06,
3569
+ "loss": 0.199,
3570
+ "step": 11540
3571
+ },
3572
+ {
3573
+ "epoch": 3.94,
3574
+ "learning_rate": 2.1280217909431393e-06,
3575
+ "loss": 0.1233,
3576
+ "step": 11560
3577
+ },
3578
+ {
3579
+ "epoch": 3.94,
3580
+ "learning_rate": 2.1144024514811034e-06,
3581
+ "loss": 0.1485,
3582
+ "step": 11580
3583
+ },
3584
+ {
3585
+ "epoch": 3.95,
3586
+ "learning_rate": 2.100783112019067e-06,
3587
+ "loss": 0.0955,
3588
+ "step": 11600
3589
+ },
3590
+ {
3591
+ "epoch": 3.96,
3592
+ "learning_rate": 2.0871637725570313e-06,
3593
+ "loss": 0.1369,
3594
+ "step": 11620
3595
+ },
3596
+ {
3597
+ "epoch": 3.96,
3598
+ "learning_rate": 2.073544433094995e-06,
3599
+ "loss": 0.1463,
3600
+ "step": 11640
3601
+ },
3602
+ {
3603
+ "epoch": 3.97,
3604
+ "learning_rate": 2.059925093632959e-06,
3605
+ "loss": 0.1499,
3606
+ "step": 11660
3607
+ },
3608
+ {
3609
+ "epoch": 3.98,
3610
+ "learning_rate": 2.046305754170923e-06,
3611
+ "loss": 0.1376,
3612
+ "step": 11680
3613
+ },
3614
+ {
3615
+ "epoch": 3.98,
3616
+ "learning_rate": 2.0326864147088866e-06,
3617
+ "loss": 0.1385,
3618
+ "step": 11700
3619
+ },
3620
+ {
3621
+ "epoch": 3.99,
3622
+ "learning_rate": 2.0190670752468508e-06,
3623
+ "loss": 0.1194,
3624
+ "step": 11720
3625
+ },
3626
+ {
3627
+ "epoch": 4.0,
3628
+ "learning_rate": 2.0054477357848145e-06,
3629
+ "loss": 0.1677,
3630
+ "step": 11740
3631
+ },
3632
+ {
3633
+ "epoch": 4.0,
3634
+ "learning_rate": 1.9918283963227787e-06,
3635
+ "loss": 0.1102,
3636
+ "step": 11760
3637
+ },
3638
+ {
3639
+ "epoch": 4.01,
3640
+ "learning_rate": 1.9782090568607424e-06,
3641
+ "loss": 0.114,
3642
+ "step": 11780
3643
+ },
3644
+ {
3645
+ "epoch": 4.02,
3646
+ "learning_rate": 1.9645897173987065e-06,
3647
+ "loss": 0.1493,
3648
+ "step": 11800
3649
+ },
3650
+ {
3651
+ "epoch": 4.02,
3652
+ "learning_rate": 1.9509703779366703e-06,
3653
+ "loss": 0.1083,
3654
+ "step": 11820
3655
+ },
3656
+ {
3657
+ "epoch": 4.03,
3658
+ "learning_rate": 1.937351038474634e-06,
3659
+ "loss": 0.0971,
3660
+ "step": 11840
3661
+ },
3662
+ {
3663
+ "epoch": 4.04,
3664
+ "learning_rate": 1.923731699012598e-06,
3665
+ "loss": 0.1186,
3666
+ "step": 11860
3667
+ },
3668
+ {
3669
+ "epoch": 4.04,
3670
+ "learning_rate": 1.910112359550562e-06,
3671
+ "loss": 0.0883,
3672
+ "step": 11880
3673
+ },
3674
+ {
3675
+ "epoch": 4.05,
3676
+ "learning_rate": 1.896493020088526e-06,
3677
+ "loss": 0.1268,
3678
+ "step": 11900
3679
+ },
3680
+ {
3681
+ "epoch": 4.06,
3682
+ "learning_rate": 1.8828736806264897e-06,
3683
+ "loss": 0.1449,
3684
+ "step": 11920
3685
+ },
3686
+ {
3687
+ "epoch": 4.07,
3688
+ "learning_rate": 1.8692543411644537e-06,
3689
+ "loss": 0.1092,
3690
+ "step": 11940
3691
+ },
3692
+ {
3693
+ "epoch": 4.07,
3694
+ "learning_rate": 1.8556350017024174e-06,
3695
+ "loss": 0.1058,
3696
+ "step": 11960
3697
+ },
3698
+ {
3699
+ "epoch": 4.08,
3700
+ "learning_rate": 1.8420156622403816e-06,
3701
+ "loss": 0.0781,
3702
+ "step": 11980
3703
+ },
3704
+ {
3705
+ "epoch": 4.09,
3706
+ "learning_rate": 1.8283963227783455e-06,
3707
+ "loss": 0.1054,
3708
+ "step": 12000
3709
+ },
3710
+ {
3711
+ "epoch": 4.09,
3712
+ "eval_loss": 0.7551047801971436,
3713
+ "eval_roc_auc": 0.7666666666666667,
3714
+ "eval_runtime": 225.8268,
3715
+ "eval_samples_per_second": 66.024,
3716
+ "eval_steps_per_second": 8.254,
3717
+ "step": 12000
3718
+ },
3719
+ {
3720
+ "epoch": 4.09,
3721
+ "learning_rate": 1.8147769833163092e-06,
3722
+ "loss": 0.1366,
3723
+ "step": 12020
3724
+ },
3725
+ {
3726
+ "epoch": 4.1,
3727
+ "learning_rate": 1.8011576438542734e-06,
3728
+ "loss": 0.1395,
3729
+ "step": 12040
3730
+ },
3731
+ {
3732
+ "epoch": 4.11,
3733
+ "learning_rate": 1.787538304392237e-06,
3734
+ "loss": 0.094,
3735
+ "step": 12060
3736
+ },
3737
+ {
3738
+ "epoch": 4.11,
3739
+ "learning_rate": 1.773918964930201e-06,
3740
+ "loss": 0.1077,
3741
+ "step": 12080
3742
+ },
3743
+ {
3744
+ "epoch": 4.12,
3745
+ "learning_rate": 1.760299625468165e-06,
3746
+ "loss": 0.1134,
3747
+ "step": 12100
3748
+ },
3749
+ {
3750
+ "epoch": 4.13,
3751
+ "learning_rate": 1.746680286006129e-06,
3752
+ "loss": 0.1002,
3753
+ "step": 12120
3754
+ },
3755
+ {
3756
+ "epoch": 4.13,
3757
+ "learning_rate": 1.7330609465440929e-06,
3758
+ "loss": 0.0906,
3759
+ "step": 12140
3760
+ },
3761
+ {
3762
+ "epoch": 4.14,
3763
+ "learning_rate": 1.7194416070820566e-06,
3764
+ "loss": 0.1249,
3765
+ "step": 12160
3766
+ },
3767
+ {
3768
+ "epoch": 4.15,
3769
+ "learning_rate": 1.7058222676200205e-06,
3770
+ "loss": 0.1197,
3771
+ "step": 12180
3772
+ },
3773
+ {
3774
+ "epoch": 4.15,
3775
+ "learning_rate": 1.6922029281579845e-06,
3776
+ "loss": 0.1198,
3777
+ "step": 12200
3778
+ },
3779
+ {
3780
+ "epoch": 4.16,
3781
+ "learning_rate": 1.6785835886959484e-06,
3782
+ "loss": 0.1104,
3783
+ "step": 12220
3784
+ },
3785
+ {
3786
+ "epoch": 4.17,
3787
+ "learning_rate": 1.6649642492339123e-06,
3788
+ "loss": 0.1034,
3789
+ "step": 12240
3790
+ },
3791
+ {
3792
+ "epoch": 4.17,
3793
+ "learning_rate": 1.651344909771876e-06,
3794
+ "loss": 0.0803,
3795
+ "step": 12260
3796
+ },
3797
+ {
3798
+ "epoch": 4.18,
3799
+ "learning_rate": 1.6377255703098402e-06,
3800
+ "loss": 0.1039,
3801
+ "step": 12280
3802
+ },
3803
+ {
3804
+ "epoch": 4.19,
3805
+ "learning_rate": 1.624106230847804e-06,
3806
+ "loss": 0.1119,
3807
+ "step": 12300
3808
+ },
3809
+ {
3810
+ "epoch": 4.19,
3811
+ "learning_rate": 1.6104868913857679e-06,
3812
+ "loss": 0.1463,
3813
+ "step": 12320
3814
+ },
3815
+ {
3816
+ "epoch": 4.2,
3817
+ "learning_rate": 1.5968675519237318e-06,
3818
+ "loss": 0.1201,
3819
+ "step": 12340
3820
+ },
3821
+ {
3822
+ "epoch": 4.21,
3823
+ "learning_rate": 1.5832482124616958e-06,
3824
+ "loss": 0.1104,
3825
+ "step": 12360
3826
+ },
3827
+ {
3828
+ "epoch": 4.22,
3829
+ "learning_rate": 1.5696288729996597e-06,
3830
+ "loss": 0.1164,
3831
+ "step": 12380
3832
+ },
3833
+ {
3834
+ "epoch": 4.22,
3835
+ "learning_rate": 1.5560095335376234e-06,
3836
+ "loss": 0.0739,
3837
+ "step": 12400
3838
+ },
3839
+ {
3840
+ "epoch": 4.23,
3841
+ "learning_rate": 1.5423901940755876e-06,
3842
+ "loss": 0.1345,
3843
+ "step": 12420
3844
+ },
3845
+ {
3846
+ "epoch": 4.24,
3847
+ "learning_rate": 1.5287708546135513e-06,
3848
+ "loss": 0.11,
3849
+ "step": 12440
3850
+ },
3851
+ {
3852
+ "epoch": 4.24,
3853
+ "learning_rate": 1.5151515151515152e-06,
3854
+ "loss": 0.0976,
3855
+ "step": 12460
3856
+ },
3857
+ {
3858
+ "epoch": 4.25,
3859
+ "learning_rate": 1.501532175689479e-06,
3860
+ "loss": 0.1323,
3861
+ "step": 12480
3862
+ },
3863
+ {
3864
+ "epoch": 4.26,
3865
+ "learning_rate": 1.4879128362274431e-06,
3866
+ "loss": 0.1222,
3867
+ "step": 12500
3868
+ },
3869
+ {
3870
+ "epoch": 4.26,
3871
+ "learning_rate": 1.474293496765407e-06,
3872
+ "loss": 0.1165,
3873
+ "step": 12520
3874
+ },
3875
+ {
3876
+ "epoch": 4.27,
3877
+ "learning_rate": 1.4606741573033708e-06,
3878
+ "loss": 0.1214,
3879
+ "step": 12540
3880
+ },
3881
+ {
3882
+ "epoch": 4.28,
3883
+ "learning_rate": 1.447054817841335e-06,
3884
+ "loss": 0.0871,
3885
+ "step": 12560
3886
+ },
3887
+ {
3888
+ "epoch": 4.28,
3889
+ "learning_rate": 1.4334354783792987e-06,
3890
+ "loss": 0.1088,
3891
+ "step": 12580
3892
+ },
3893
+ {
3894
+ "epoch": 4.29,
3895
+ "learning_rate": 1.4198161389172626e-06,
3896
+ "loss": 0.0892,
3897
+ "step": 12600
3898
+ },
3899
+ {
3900
+ "epoch": 4.3,
3901
+ "learning_rate": 1.4061967994552263e-06,
3902
+ "loss": 0.1352,
3903
+ "step": 12620
3904
+ },
3905
+ {
3906
+ "epoch": 4.3,
3907
+ "learning_rate": 1.3925774599931905e-06,
3908
+ "loss": 0.0977,
3909
+ "step": 12640
3910
+ },
3911
+ {
3912
+ "epoch": 4.31,
3913
+ "learning_rate": 1.3789581205311544e-06,
3914
+ "loss": 0.1374,
3915
+ "step": 12660
3916
+ },
3917
+ {
3918
+ "epoch": 4.32,
3919
+ "learning_rate": 1.3653387810691181e-06,
3920
+ "loss": 0.0864,
3921
+ "step": 12680
3922
+ },
3923
+ {
3924
+ "epoch": 4.32,
3925
+ "learning_rate": 1.3517194416070823e-06,
3926
+ "loss": 0.089,
3927
+ "step": 12700
3928
+ },
3929
+ {
3930
+ "epoch": 4.33,
3931
+ "learning_rate": 1.338100102145046e-06,
3932
+ "loss": 0.0961,
3933
+ "step": 12720
3934
+ },
3935
+ {
3936
+ "epoch": 4.34,
3937
+ "learning_rate": 1.32448076268301e-06,
3938
+ "loss": 0.1148,
3939
+ "step": 12740
3940
+ },
3941
+ {
3942
+ "epoch": 4.34,
3943
+ "learning_rate": 1.3108614232209737e-06,
3944
+ "loss": 0.1113,
3945
+ "step": 12760
3946
+ },
3947
+ {
3948
+ "epoch": 4.35,
3949
+ "learning_rate": 1.2972420837589378e-06,
3950
+ "loss": 0.0984,
3951
+ "step": 12780
3952
+ },
3953
+ {
3954
+ "epoch": 4.36,
3955
+ "learning_rate": 1.2836227442969018e-06,
3956
+ "loss": 0.1134,
3957
+ "step": 12800
3958
+ },
3959
+ {
3960
+ "epoch": 4.36,
3961
+ "learning_rate": 1.2700034048348655e-06,
3962
+ "loss": 0.1126,
3963
+ "step": 12820
3964
+ },
3965
+ {
3966
+ "epoch": 4.37,
3967
+ "learning_rate": 1.2563840653728297e-06,
3968
+ "loss": 0.1318,
3969
+ "step": 12840
3970
+ },
3971
+ {
3972
+ "epoch": 4.38,
3973
+ "learning_rate": 1.2427647259107934e-06,
3974
+ "loss": 0.1326,
3975
+ "step": 12860
3976
+ },
3977
+ {
3978
+ "epoch": 4.39,
3979
+ "learning_rate": 1.2291453864487573e-06,
3980
+ "loss": 0.0954,
3981
+ "step": 12880
3982
+ },
3983
+ {
3984
+ "epoch": 4.39,
3985
+ "learning_rate": 1.2155260469867213e-06,
3986
+ "loss": 0.1549,
3987
+ "step": 12900
3988
+ },
3989
+ {
3990
+ "epoch": 4.4,
3991
+ "learning_rate": 1.2019067075246852e-06,
3992
+ "loss": 0.1077,
3993
+ "step": 12920
3994
+ },
3995
+ {
3996
+ "epoch": 4.41,
3997
+ "learning_rate": 1.188287368062649e-06,
3998
+ "loss": 0.0598,
3999
+ "step": 12940
4000
+ },
4001
+ {
4002
+ "epoch": 4.41,
4003
+ "learning_rate": 1.1746680286006129e-06,
4004
+ "loss": 0.1317,
4005
+ "step": 12960
4006
+ },
4007
+ {
4008
+ "epoch": 4.42,
4009
+ "learning_rate": 1.161048689138577e-06,
4010
+ "loss": 0.0877,
4011
+ "step": 12980
4012
+ },
4013
+ {
4014
+ "epoch": 4.43,
4015
+ "learning_rate": 1.1474293496765407e-06,
4016
+ "loss": 0.1076,
4017
+ "step": 13000
4018
+ },
4019
+ {
4020
+ "epoch": 4.43,
4021
+ "eval_loss": 0.8132159113883972,
4022
+ "eval_roc_auc": 0.7627095908786049,
4023
+ "eval_runtime": 227.818,
4024
+ "eval_samples_per_second": 65.447,
4025
+ "eval_steps_per_second": 8.182,
4026
+ "step": 13000
4027
+ },
4028
+ {
4029
+ "epoch": 4.43,
4030
+ "learning_rate": 1.1338100102145047e-06,
4031
+ "loss": 0.1132,
4032
+ "step": 13020
4033
+ },
4034
+ {
4035
+ "epoch": 4.44,
4036
+ "learning_rate": 1.1201906707524686e-06,
4037
+ "loss": 0.0821,
4038
+ "step": 13040
4039
+ },
4040
+ {
4041
+ "epoch": 4.45,
4042
+ "learning_rate": 1.1065713312904326e-06,
4043
+ "loss": 0.0979,
4044
+ "step": 13060
4045
+ },
4046
+ {
4047
+ "epoch": 4.45,
4048
+ "learning_rate": 1.0929519918283963e-06,
4049
+ "loss": 0.1109,
4050
+ "step": 13080
4051
+ },
4052
+ {
4053
+ "epoch": 4.46,
4054
+ "learning_rate": 1.0793326523663602e-06,
4055
+ "loss": 0.1429,
4056
+ "step": 13100
4057
+ },
4058
+ {
4059
+ "epoch": 4.47,
4060
+ "learning_rate": 1.0657133129043244e-06,
4061
+ "loss": 0.1369,
4062
+ "step": 13120
4063
+ },
4064
+ {
4065
+ "epoch": 4.47,
4066
+ "learning_rate": 1.052093973442288e-06,
4067
+ "loss": 0.103,
4068
+ "step": 13140
4069
+ },
4070
+ {
4071
+ "epoch": 4.48,
4072
+ "learning_rate": 1.038474633980252e-06,
4073
+ "loss": 0.1071,
4074
+ "step": 13160
4075
+ },
4076
+ {
4077
+ "epoch": 4.49,
4078
+ "learning_rate": 1.024855294518216e-06,
4079
+ "loss": 0.0864,
4080
+ "step": 13180
4081
+ },
4082
+ {
4083
+ "epoch": 4.49,
4084
+ "learning_rate": 1.01123595505618e-06,
4085
+ "loss": 0.1063,
4086
+ "step": 13200
4087
+ },
4088
+ {
4089
+ "epoch": 4.5,
4090
+ "learning_rate": 9.976166155941436e-07,
4091
+ "loss": 0.1019,
4092
+ "step": 13220
4093
+ },
4094
+ {
4095
+ "epoch": 4.51,
4096
+ "learning_rate": 9.839972761321076e-07,
4097
+ "loss": 0.1222,
4098
+ "step": 13240
4099
+ },
4100
+ {
4101
+ "epoch": 4.51,
4102
+ "learning_rate": 9.703779366700715e-07,
4103
+ "loss": 0.0911,
4104
+ "step": 13260
4105
+ },
4106
+ {
4107
+ "epoch": 4.52,
4108
+ "learning_rate": 9.567585972080355e-07,
4109
+ "loss": 0.1774,
4110
+ "step": 13280
4111
+ },
4112
+ {
4113
+ "epoch": 4.53,
4114
+ "learning_rate": 9.431392577459994e-07,
4115
+ "loss": 0.1098,
4116
+ "step": 13300
4117
+ },
4118
+ {
4119
+ "epoch": 4.54,
4120
+ "learning_rate": 9.295199182839632e-07,
4121
+ "loss": 0.1371,
4122
+ "step": 13320
4123
+ },
4124
+ {
4125
+ "epoch": 4.54,
4126
+ "learning_rate": 9.159005788219272e-07,
4127
+ "loss": 0.1277,
4128
+ "step": 13340
4129
+ },
4130
+ {
4131
+ "epoch": 4.55,
4132
+ "learning_rate": 9.022812393598911e-07,
4133
+ "loss": 0.0962,
4134
+ "step": 13360
4135
+ },
4136
+ {
4137
+ "epoch": 4.56,
4138
+ "learning_rate": 8.886618998978551e-07,
4139
+ "loss": 0.1411,
4140
+ "step": 13380
4141
+ },
4142
+ {
4143
+ "epoch": 4.56,
4144
+ "learning_rate": 8.75042560435819e-07,
4145
+ "loss": 0.1044,
4146
+ "step": 13400
4147
+ },
4148
+ {
4149
+ "epoch": 4.57,
4150
+ "learning_rate": 8.614232209737828e-07,
4151
+ "loss": 0.0907,
4152
+ "step": 13420
4153
+ },
4154
+ {
4155
+ "epoch": 4.58,
4156
+ "learning_rate": 8.478038815117468e-07,
4157
+ "loss": 0.1084,
4158
+ "step": 13440
4159
+ },
4160
+ {
4161
+ "epoch": 4.58,
4162
+ "learning_rate": 8.341845420497106e-07,
4163
+ "loss": 0.1198,
4164
+ "step": 13460
4165
+ },
4166
+ {
4167
+ "epoch": 4.59,
4168
+ "learning_rate": 8.205652025876745e-07,
4169
+ "loss": 0.0946,
4170
+ "step": 13480
4171
+ },
4172
+ {
4173
+ "epoch": 4.6,
4174
+ "learning_rate": 8.069458631256384e-07,
4175
+ "loss": 0.1361,
4176
+ "step": 13500
4177
+ },
4178
+ {
4179
+ "epoch": 4.6,
4180
+ "learning_rate": 7.933265236636024e-07,
4181
+ "loss": 0.0859,
4182
+ "step": 13520
4183
+ },
4184
+ {
4185
+ "epoch": 4.61,
4186
+ "learning_rate": 7.797071842015664e-07,
4187
+ "loss": 0.1314,
4188
+ "step": 13540
4189
+ },
4190
+ {
4191
+ "epoch": 4.62,
4192
+ "learning_rate": 7.660878447395302e-07,
4193
+ "loss": 0.121,
4194
+ "step": 13560
4195
+ },
4196
+ {
4197
+ "epoch": 4.62,
4198
+ "learning_rate": 7.524685052774941e-07,
4199
+ "loss": 0.109,
4200
+ "step": 13580
4201
+ },
4202
+ {
4203
+ "epoch": 4.63,
4204
+ "learning_rate": 7.38849165815458e-07,
4205
+ "loss": 0.0947,
4206
+ "step": 13600
4207
+ },
4208
+ {
4209
+ "epoch": 4.64,
4210
+ "learning_rate": 7.252298263534219e-07,
4211
+ "loss": 0.105,
4212
+ "step": 13620
4213
+ },
4214
+ {
4215
+ "epoch": 4.64,
4216
+ "learning_rate": 7.116104868913857e-07,
4217
+ "loss": 0.1235,
4218
+ "step": 13640
4219
+ },
4220
+ {
4221
+ "epoch": 4.65,
4222
+ "learning_rate": 6.979911474293498e-07,
4223
+ "loss": 0.1332,
4224
+ "step": 13660
4225
+ },
4226
+ {
4227
+ "epoch": 4.66,
4228
+ "learning_rate": 6.843718079673137e-07,
4229
+ "loss": 0.0717,
4230
+ "step": 13680
4231
+ },
4232
+ {
4233
+ "epoch": 4.66,
4234
+ "learning_rate": 6.707524685052775e-07,
4235
+ "loss": 0.1095,
4236
+ "step": 13700
4237
+ },
4238
+ {
4239
+ "epoch": 4.67,
4240
+ "learning_rate": 6.571331290432415e-07,
4241
+ "loss": 0.1468,
4242
+ "step": 13720
4243
+ },
4244
+ {
4245
+ "epoch": 4.68,
4246
+ "learning_rate": 6.435137895812053e-07,
4247
+ "loss": 0.1202,
4248
+ "step": 13740
4249
+ },
4250
+ {
4251
+ "epoch": 4.69,
4252
+ "learning_rate": 6.298944501191693e-07,
4253
+ "loss": 0.0621,
4254
+ "step": 13760
4255
+ },
4256
+ {
4257
+ "epoch": 4.69,
4258
+ "learning_rate": 6.162751106571332e-07,
4259
+ "loss": 0.0852,
4260
+ "step": 13780
4261
+ },
4262
+ {
4263
+ "epoch": 4.7,
4264
+ "learning_rate": 6.02655771195097e-07,
4265
+ "loss": 0.1006,
4266
+ "step": 13800
4267
+ },
4268
+ {
4269
+ "epoch": 4.71,
4270
+ "learning_rate": 5.89036431733061e-07,
4271
+ "loss": 0.1046,
4272
+ "step": 13820
4273
+ },
4274
+ {
4275
+ "epoch": 4.71,
4276
+ "learning_rate": 5.754170922710249e-07,
4277
+ "loss": 0.1105,
4278
+ "step": 13840
4279
+ },
4280
+ {
4281
+ "epoch": 4.72,
4282
+ "learning_rate": 5.617977528089888e-07,
4283
+ "loss": 0.0842,
4284
+ "step": 13860
4285
+ },
4286
+ {
4287
+ "epoch": 4.73,
4288
+ "learning_rate": 5.481784133469528e-07,
4289
+ "loss": 0.0779,
4290
+ "step": 13880
4291
+ },
4292
+ {
4293
+ "epoch": 4.73,
4294
+ "learning_rate": 5.345590738849166e-07,
4295
+ "loss": 0.1007,
4296
+ "step": 13900
4297
+ },
4298
+ {
4299
+ "epoch": 4.74,
4300
+ "learning_rate": 5.209397344228806e-07,
4301
+ "loss": 0.0784,
4302
+ "step": 13920
4303
+ },
4304
+ {
4305
+ "epoch": 4.75,
4306
+ "learning_rate": 5.073203949608444e-07,
4307
+ "loss": 0.121,
4308
+ "step": 13940
4309
+ },
4310
+ {
4311
+ "epoch": 4.75,
4312
+ "learning_rate": 4.937010554988083e-07,
4313
+ "loss": 0.1562,
4314
+ "step": 13960
4315
+ },
4316
+ {
4317
+ "epoch": 4.76,
4318
+ "learning_rate": 4.800817160367723e-07,
4319
+ "loss": 0.1098,
4320
+ "step": 13980
4321
+ },
4322
+ {
4323
+ "epoch": 4.77,
4324
+ "learning_rate": 4.6646237657473615e-07,
4325
+ "loss": 0.1321,
4326
+ "step": 14000
4327
+ },
4328
+ {
4329
+ "epoch": 4.77,
4330
+ "eval_loss": 0.815223753452301,
4331
+ "eval_roc_auc": 0.7586854460093896,
4332
+ "eval_runtime": 225.105,
4333
+ "eval_samples_per_second": 66.236,
4334
+ "eval_steps_per_second": 8.281,
4335
+ "step": 14000
4336
+ },
4337
+ {
4338
+ "epoch": 4.77,
4339
+ "learning_rate": 4.528430371127001e-07,
4340
+ "loss": 0.1244,
4341
+ "step": 14020
4342
+ },
4343
+ {
4344
+ "epoch": 4.78,
4345
+ "learning_rate": 4.39223697650664e-07,
4346
+ "loss": 0.1415,
4347
+ "step": 14040
4348
+ },
4349
+ {
4350
+ "epoch": 4.79,
4351
+ "learning_rate": 4.2560435818862786e-07,
4352
+ "loss": 0.0982,
4353
+ "step": 14060
4354
+ },
4355
+ {
4356
+ "epoch": 4.79,
4357
+ "learning_rate": 4.1198501872659175e-07,
4358
+ "loss": 0.0874,
4359
+ "step": 14080
4360
+ },
4361
+ {
4362
+ "epoch": 4.8,
4363
+ "learning_rate": 3.9836567926455574e-07,
4364
+ "loss": 0.1736,
4365
+ "step": 14100
4366
+ },
4367
+ {
4368
+ "epoch": 4.81,
4369
+ "learning_rate": 3.8474633980251963e-07,
4370
+ "loss": 0.1213,
4371
+ "step": 14120
4372
+ },
4373
+ {
4374
+ "epoch": 4.81,
4375
+ "learning_rate": 3.711270003404835e-07,
4376
+ "loss": 0.1008,
4377
+ "step": 14140
4378
+ },
4379
+ {
4380
+ "epoch": 4.82,
4381
+ "learning_rate": 3.5750766087844745e-07,
4382
+ "loss": 0.1624,
4383
+ "step": 14160
4384
+ },
4385
+ {
4386
+ "epoch": 4.83,
4387
+ "learning_rate": 3.4388832141641134e-07,
4388
+ "loss": 0.1073,
4389
+ "step": 14180
4390
+ },
4391
+ {
4392
+ "epoch": 4.83,
4393
+ "learning_rate": 3.302689819543752e-07,
4394
+ "loss": 0.0801,
4395
+ "step": 14200
4396
+ },
4397
+ {
4398
+ "epoch": 4.84,
4399
+ "learning_rate": 3.1664964249233916e-07,
4400
+ "loss": 0.1252,
4401
+ "step": 14220
4402
+ },
4403
+ {
4404
+ "epoch": 4.85,
4405
+ "learning_rate": 3.0303030303030305e-07,
4406
+ "loss": 0.0988,
4407
+ "step": 14240
4408
+ },
4409
+ {
4410
+ "epoch": 4.86,
4411
+ "learning_rate": 2.8941096356826693e-07,
4412
+ "loss": 0.0818,
4413
+ "step": 14260
4414
+ },
4415
+ {
4416
+ "epoch": 4.86,
4417
+ "learning_rate": 2.7579162410623087e-07,
4418
+ "loss": 0.1005,
4419
+ "step": 14280
4420
+ },
4421
+ {
4422
+ "epoch": 4.87,
4423
+ "learning_rate": 2.621722846441948e-07,
4424
+ "loss": 0.1191,
4425
+ "step": 14300
4426
+ },
4427
+ {
4428
+ "epoch": 4.88,
4429
+ "learning_rate": 2.485529451821587e-07,
4430
+ "loss": 0.1062,
4431
+ "step": 14320
4432
+ },
4433
+ {
4434
+ "epoch": 4.88,
4435
+ "learning_rate": 2.349336057201226e-07,
4436
+ "loss": 0.115,
4437
+ "step": 14340
4438
+ },
4439
+ {
4440
+ "epoch": 4.89,
4441
+ "learning_rate": 2.213142662580865e-07,
4442
+ "loss": 0.1172,
4443
+ "step": 14360
4444
+ },
4445
+ {
4446
+ "epoch": 4.9,
4447
+ "learning_rate": 2.076949267960504e-07,
4448
+ "loss": 0.099,
4449
+ "step": 14380
4450
+ },
4451
+ {
4452
+ "epoch": 4.9,
4453
+ "learning_rate": 1.9407558733401432e-07,
4454
+ "loss": 0.127,
4455
+ "step": 14400
4456
+ },
4457
+ {
4458
+ "epoch": 4.91,
4459
+ "learning_rate": 1.804562478719782e-07,
4460
+ "loss": 0.0879,
4461
+ "step": 14420
4462
+ },
4463
+ {
4464
+ "epoch": 4.92,
4465
+ "learning_rate": 1.6683690840994215e-07,
4466
+ "loss": 0.1284,
4467
+ "step": 14440
4468
+ },
4469
+ {
4470
+ "epoch": 4.92,
4471
+ "learning_rate": 1.5321756894790606e-07,
4472
+ "loss": 0.1249,
4473
+ "step": 14460
4474
+ },
4475
+ {
4476
+ "epoch": 4.93,
4477
+ "learning_rate": 1.3959822948586994e-07,
4478
+ "loss": 0.1314,
4479
+ "step": 14480
4480
+ },
4481
+ {
4482
+ "epoch": 4.94,
4483
+ "learning_rate": 1.2597889002383386e-07,
4484
+ "loss": 0.1307,
4485
+ "step": 14500
4486
+ },
4487
+ {
4488
+ "epoch": 4.94,
4489
+ "learning_rate": 1.1235955056179776e-07,
4490
+ "loss": 0.092,
4491
+ "step": 14520
4492
+ },
4493
+ {
4494
+ "epoch": 4.95,
4495
+ "learning_rate": 9.874021109976167e-08,
4496
+ "loss": 0.084,
4497
+ "step": 14540
4498
+ },
4499
+ {
4500
+ "epoch": 4.96,
4501
+ "learning_rate": 8.512087163772558e-08,
4502
+ "loss": 0.1793,
4503
+ "step": 14560
4504
+ },
4505
+ {
4506
+ "epoch": 4.96,
4507
+ "learning_rate": 7.150153217568949e-08,
4508
+ "loss": 0.1071,
4509
+ "step": 14580
4510
+ },
4511
+ {
4512
+ "epoch": 4.97,
4513
+ "learning_rate": 5.788219271365339e-08,
4514
+ "loss": 0.103,
4515
+ "step": 14600
4516
+ },
4517
+ {
4518
+ "epoch": 4.98,
4519
+ "learning_rate": 4.42628532516173e-08,
4520
+ "loss": 0.1024,
4521
+ "step": 14620
4522
+ },
4523
+ {
4524
+ "epoch": 4.98,
4525
+ "learning_rate": 3.0643513789581204e-08,
4526
+ "loss": 0.0751,
4527
+ "step": 14640
4528
+ },
4529
+ {
4530
+ "epoch": 4.99,
4531
+ "learning_rate": 1.7024174327545116e-08,
4532
+ "loss": 0.0811,
4533
+ "step": 14660
4534
+ },
4535
+ {
4536
+ "epoch": 5.0,
4537
+ "learning_rate": 3.4048348655090227e-09,
4538
+ "loss": 0.0877,
4539
+ "step": 14680
4540
+ },
4541
+ {
4542
+ "epoch": 5.0,
4543
+ "step": 14685,
4544
+ "total_flos": 3.6406511640831468e+19,
4545
+ "train_loss": 0.21540352483164263,
4546
+ "train_runtime": 8294.2837,
4547
+ "train_samples_per_second": 56.643,
4548
+ "train_steps_per_second": 1.77
4549
+ }
4550
+ ],
4551
+ "logging_steps": 20,
4552
+ "max_steps": 14685,
4553
+ "num_train_epochs": 5,
4554
+ "save_steps": 1000,
4555
+ "total_flos": 3.6406511640831468e+19,
4556
+ "trial_name": null,
4557
+ "trial_params": null
4558
+ }