Commit
·
1a4b272
1
Parent(s):
50d4467
Add evaluation results and output files to model card
Browse files- README.md +2 -2
- checkpoint-414/config.json +23 -0
- checkpoint-414/model.safetensors +3 -0
- checkpoint-414/optimizer.pt +3 -0
- checkpoint-414/preprocessor_config.json +22 -0
- checkpoint-414/rng_state.pth +3 -0
- checkpoint-414/scheduler.pt +3 -0
- checkpoint-414/trainer_state.json +3483 -0
- checkpoint-414/training_args.bin +3 -0
- model.safetensors +1 -1
- pr_curve.png +0 -0
- roc_curve.png +0 -0
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -7,7 +7,7 @@ tags:
|
|
| 7 |
metrics:
|
| 8 |
- average_precision: 0.67
|
| 9 |
- roc_auc: 0.77
|
| 10 |
-
- best threshold according to F1: 0.
|
| 11 |
---
|
| 12 |
|
| 13 |
# Binary Classification Model
|
|
@@ -16,7 +16,7 @@ metrics:
|
|
| 16 |
|
| 17 |
**Average Precision:** 0.67
|
| 18 |
**ROC AUC:** 0.77
|
| 19 |
-
**best threshold according to F1: 0.
|
| 20 |
|
| 21 |
|
| 22 |
## Visualizations
|
|
|
|
| 7 |
metrics:
|
| 8 |
- average_precision: 0.67
|
| 9 |
- roc_auc: 0.77
|
| 10 |
+
- best threshold according to F1: 0.41
|
| 11 |
---
|
| 12 |
|
| 13 |
# Binary Classification Model
|
|
|
|
| 16 |
|
| 17 |
**Average Precision:** 0.67
|
| 18 |
**ROC AUC:** 0.77
|
| 19 |
+
**best threshold according to F1: 0.41
|
| 20 |
|
| 21 |
|
| 22 |
## Visualizations
|
checkpoint-414/config.json
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "GenetikaPlus/binary_classification_model_v3.1.3_spines",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"ViTForImageClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"encoder_stride": 16,
|
| 8 |
+
"hidden_act": "gelu",
|
| 9 |
+
"hidden_dropout_prob": 0.0,
|
| 10 |
+
"hidden_size": 256,
|
| 11 |
+
"image_size": 32,
|
| 12 |
+
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 512,
|
| 14 |
+
"layer_norm_eps": 1e-12,
|
| 15 |
+
"model_type": "vit",
|
| 16 |
+
"num_attention_heads": 8,
|
| 17 |
+
"num_channels": 3,
|
| 18 |
+
"num_hidden_layers": 4,
|
| 19 |
+
"patch_size": 4,
|
| 20 |
+
"qkv_bias": true,
|
| 21 |
+
"torch_dtype": "float32",
|
| 22 |
+
"transformers_version": "4.46.2"
|
| 23 |
+
}
|
checkpoint-414/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:05f698e2857ed21d2c2f654ad309d7006d406e0eecb9d46ad4a845601eda4380
|
| 3 |
+
size 8563512
|
checkpoint-414/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dcdb2576df644a99eadb2cafa6d6d81474387b6d245ae99e48d49c58c61161e0
|
| 3 |
+
size 17171514
|
checkpoint-414/preprocessor_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_rescale": true,
|
| 4 |
+
"do_resize": false,
|
| 5 |
+
"image_mean": [
|
| 6 |
+
0.5,
|
| 7 |
+
0.5,
|
| 8 |
+
0.5
|
| 9 |
+
],
|
| 10 |
+
"image_processor_type": "ViTFeatureExtractor",
|
| 11 |
+
"image_std": [
|
| 12 |
+
0.5,
|
| 13 |
+
0.5,
|
| 14 |
+
0.5
|
| 15 |
+
],
|
| 16 |
+
"resample": 2,
|
| 17 |
+
"rescale_factor": 0.00392156862745098,
|
| 18 |
+
"size": {
|
| 19 |
+
"height": 32,
|
| 20 |
+
"width": 32
|
| 21 |
+
}
|
| 22 |
+
}
|
checkpoint-414/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:23a5966e7d9aa27bf3c91ddfd2175d1f30b8f5544df504956c7ae90710a33d59
|
| 3 |
+
size 14244
|
checkpoint-414/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec2080e6c334a01834ceb7157d787a65623160fa045734644b6009c0ce8da085
|
| 3 |
+
size 1000
|
checkpoint-414/trainer_state.json
ADDED
|
@@ -0,0 +1,3483 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8273947246645071,
|
| 3 |
+
"best_model_checkpoint": "/tmp/logs/binary_classification_model_v3.1.5_spines/checkpoint-414",
|
| 4 |
+
"epoch": 46.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 414,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.1111111111111111,
|
| 13 |
+
"grad_norm": 462486.0,
|
| 14 |
+
"learning_rate": 1.148369315571888e-08,
|
| 15 |
+
"loss": 0.3999,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.2222222222222222,
|
| 20 |
+
"grad_norm": 498353.25,
|
| 21 |
+
"learning_rate": 2.296738631143776e-08,
|
| 22 |
+
"loss": 0.4304,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3333333333333333,
|
| 27 |
+
"grad_norm": 369319.8125,
|
| 28 |
+
"learning_rate": 3.4451079467156634e-08,
|
| 29 |
+
"loss": 0.4205,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.4444444444444444,
|
| 34 |
+
"grad_norm": 433521.6875,
|
| 35 |
+
"learning_rate": 4.593477262287552e-08,
|
| 36 |
+
"loss": 0.3986,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.5555555555555556,
|
| 41 |
+
"grad_norm": 341315.84375,
|
| 42 |
+
"learning_rate": 5.74184657785944e-08,
|
| 43 |
+
"loss": 0.4026,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.6666666666666666,
|
| 48 |
+
"grad_norm": 292268.09375,
|
| 49 |
+
"learning_rate": 6.890215893431327e-08,
|
| 50 |
+
"loss": 0.3945,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.7777777777777778,
|
| 55 |
+
"grad_norm": 265327.46875,
|
| 56 |
+
"learning_rate": 8.038585209003216e-08,
|
| 57 |
+
"loss": 0.3947,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.8888888888888888,
|
| 62 |
+
"grad_norm": 311655.9375,
|
| 63 |
+
"learning_rate": 9.186954524575104e-08,
|
| 64 |
+
"loss": 0.408,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.0,
|
| 69 |
+
"grad_norm": 534967.1875,
|
| 70 |
+
"learning_rate": 1.0335323840146992e-07,
|
| 71 |
+
"loss": 0.4136,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.0,
|
| 76 |
+
"eval_accuracy": 0.809600367478181,
|
| 77 |
+
"eval_f1": 0.7938323800049739,
|
| 78 |
+
"eval_loss": 0.41161027550697327,
|
| 79 |
+
"eval_precision": 0.8218331616889805,
|
| 80 |
+
"eval_recall": 0.7676767676767676,
|
| 81 |
+
"eval_runtime": 3.7209,
|
| 82 |
+
"eval_samples_per_second": 1170.135,
|
| 83 |
+
"eval_steps_per_second": 0.806,
|
| 84 |
+
"step": 9
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.1111111111111112,
|
| 88 |
+
"grad_norm": 436558.1875,
|
| 89 |
+
"learning_rate": 1.148369315571888e-07,
|
| 90 |
+
"loss": 0.4109,
|
| 91 |
+
"step": 10
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.2222222222222223,
|
| 95 |
+
"grad_norm": 183180.109375,
|
| 96 |
+
"learning_rate": 1.2632062471290768e-07,
|
| 97 |
+
"loss": 0.3964,
|
| 98 |
+
"step": 11
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.3333333333333333,
|
| 102 |
+
"grad_norm": 353782.28125,
|
| 103 |
+
"learning_rate": 1.3780431786862654e-07,
|
| 104 |
+
"loss": 0.4047,
|
| 105 |
+
"step": 12
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.4444444444444444,
|
| 109 |
+
"grad_norm": 326118.3125,
|
| 110 |
+
"learning_rate": 1.4928801102434544e-07,
|
| 111 |
+
"loss": 0.3885,
|
| 112 |
+
"step": 13
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.5555555555555556,
|
| 116 |
+
"grad_norm": 196583.015625,
|
| 117 |
+
"learning_rate": 1.6077170418006432e-07,
|
| 118 |
+
"loss": 0.3977,
|
| 119 |
+
"step": 14
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.6666666666666665,
|
| 123 |
+
"grad_norm": 88230.421875,
|
| 124 |
+
"learning_rate": 1.722553973357832e-07,
|
| 125 |
+
"loss": 0.3985,
|
| 126 |
+
"step": 15
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 1.7777777777777777,
|
| 130 |
+
"grad_norm": 102945.453125,
|
| 131 |
+
"learning_rate": 1.8373909049150207e-07,
|
| 132 |
+
"loss": 0.4188,
|
| 133 |
+
"step": 16
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 1.8888888888888888,
|
| 137 |
+
"grad_norm": 132400.3125,
|
| 138 |
+
"learning_rate": 1.9522278364722095e-07,
|
| 139 |
+
"loss": 0.3939,
|
| 140 |
+
"step": 17
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 2.0,
|
| 144 |
+
"grad_norm": 187392.875,
|
| 145 |
+
"learning_rate": 2.0670647680293983e-07,
|
| 146 |
+
"loss": 0.4189,
|
| 147 |
+
"step": 18
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 2.0,
|
| 151 |
+
"eval_accuracy": 0.8086816720257235,
|
| 152 |
+
"eval_f1": 0.7966804979253111,
|
| 153 |
+
"eval_loss": 0.41029924154281616,
|
| 154 |
+
"eval_precision": 0.8087215064420218,
|
| 155 |
+
"eval_recall": 0.784992784992785,
|
| 156 |
+
"eval_runtime": 3.8262,
|
| 157 |
+
"eval_samples_per_second": 1137.932,
|
| 158 |
+
"eval_steps_per_second": 0.784,
|
| 159 |
+
"step": 18
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 2.111111111111111,
|
| 163 |
+
"grad_norm": 165157.5,
|
| 164 |
+
"learning_rate": 2.181901699586587e-07,
|
| 165 |
+
"loss": 0.3957,
|
| 166 |
+
"step": 19
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 2.2222222222222223,
|
| 170 |
+
"grad_norm": 209579.34375,
|
| 171 |
+
"learning_rate": 2.296738631143776e-07,
|
| 172 |
+
"loss": 0.4188,
|
| 173 |
+
"step": 20
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 2.3333333333333335,
|
| 177 |
+
"grad_norm": 147438.40625,
|
| 178 |
+
"learning_rate": 2.4115755627009647e-07,
|
| 179 |
+
"loss": 0.4076,
|
| 180 |
+
"step": 21
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 2.4444444444444446,
|
| 184 |
+
"grad_norm": 164255.859375,
|
| 185 |
+
"learning_rate": 2.5264124942581537e-07,
|
| 186 |
+
"loss": 0.399,
|
| 187 |
+
"step": 22
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 2.5555555555555554,
|
| 191 |
+
"grad_norm": 68289.0234375,
|
| 192 |
+
"learning_rate": 2.641249425815342e-07,
|
| 193 |
+
"loss": 0.4097,
|
| 194 |
+
"step": 23
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.6666666666666665,
|
| 198 |
+
"grad_norm": 116692.9453125,
|
| 199 |
+
"learning_rate": 2.7560863573725307e-07,
|
| 200 |
+
"loss": 0.3804,
|
| 201 |
+
"step": 24
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 2.7777777777777777,
|
| 205 |
+
"grad_norm": 144575.59375,
|
| 206 |
+
"learning_rate": 2.87092328892972e-07,
|
| 207 |
+
"loss": 0.3957,
|
| 208 |
+
"step": 25
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 2.888888888888889,
|
| 212 |
+
"grad_norm": 69518.9375,
|
| 213 |
+
"learning_rate": 2.985760220486909e-07,
|
| 214 |
+
"loss": 0.3953,
|
| 215 |
+
"step": 26
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 3.0,
|
| 219 |
+
"grad_norm": 198278.28125,
|
| 220 |
+
"learning_rate": 3.100597152044098e-07,
|
| 221 |
+
"loss": 0.3883,
|
| 222 |
+
"step": 27
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 3.0,
|
| 226 |
+
"eval_accuracy": 0.8118971061093248,
|
| 227 |
+
"eval_f1": 0.8021261174196666,
|
| 228 |
+
"eval_loss": 0.4097851514816284,
|
| 229 |
+
"eval_precision": 0.8058252427184466,
|
| 230 |
+
"eval_recall": 0.7984607984607984,
|
| 231 |
+
"eval_runtime": 3.8584,
|
| 232 |
+
"eval_samples_per_second": 1128.45,
|
| 233 |
+
"eval_steps_per_second": 0.778,
|
| 234 |
+
"step": 27
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 3.111111111111111,
|
| 238 |
+
"grad_norm": 104717.0,
|
| 239 |
+
"learning_rate": 3.2154340836012864e-07,
|
| 240 |
+
"loss": 0.3916,
|
| 241 |
+
"step": 28
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 3.2222222222222223,
|
| 245 |
+
"grad_norm": 70841.9375,
|
| 246 |
+
"learning_rate": 3.330271015158475e-07,
|
| 247 |
+
"loss": 0.3811,
|
| 248 |
+
"step": 29
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 3.3333333333333335,
|
| 252 |
+
"grad_norm": 225365.171875,
|
| 253 |
+
"learning_rate": 3.445107946715664e-07,
|
| 254 |
+
"loss": 0.4059,
|
| 255 |
+
"step": 30
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 3.4444444444444446,
|
| 259 |
+
"grad_norm": 62955.94140625,
|
| 260 |
+
"learning_rate": 3.5599448782728525e-07,
|
| 261 |
+
"loss": 0.3893,
|
| 262 |
+
"step": 31
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 3.5555555555555554,
|
| 266 |
+
"grad_norm": 193636.390625,
|
| 267 |
+
"learning_rate": 3.6747818098300415e-07,
|
| 268 |
+
"loss": 0.4119,
|
| 269 |
+
"step": 32
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.6666666666666665,
|
| 273 |
+
"grad_norm": 166762.671875,
|
| 274 |
+
"learning_rate": 3.7896187413872305e-07,
|
| 275 |
+
"loss": 0.4034,
|
| 276 |
+
"step": 33
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 3.7777777777777777,
|
| 280 |
+
"grad_norm": 201156.375,
|
| 281 |
+
"learning_rate": 3.904455672944419e-07,
|
| 282 |
+
"loss": 0.4015,
|
| 283 |
+
"step": 34
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 3.888888888888889,
|
| 287 |
+
"grad_norm": 104657.8671875,
|
| 288 |
+
"learning_rate": 4.0192926045016076e-07,
|
| 289 |
+
"loss": 0.4104,
|
| 290 |
+
"step": 35
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 4.0,
|
| 294 |
+
"grad_norm": 176918.875,
|
| 295 |
+
"learning_rate": 4.1341295360587966e-07,
|
| 296 |
+
"loss": 0.3992,
|
| 297 |
+
"step": 36
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 4.0,
|
| 301 |
+
"eval_accuracy": 0.811437758383096,
|
| 302 |
+
"eval_f1": 0.8013549479796758,
|
| 303 |
+
"eval_loss": 0.4091891646385193,
|
| 304 |
+
"eval_precision": 0.8062317429406037,
|
| 305 |
+
"eval_recall": 0.7965367965367965,
|
| 306 |
+
"eval_runtime": 3.9227,
|
| 307 |
+
"eval_samples_per_second": 1109.959,
|
| 308 |
+
"eval_steps_per_second": 0.765,
|
| 309 |
+
"step": 36
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 4.111111111111111,
|
| 313 |
+
"grad_norm": 102350.40625,
|
| 314 |
+
"learning_rate": 4.248966467615985e-07,
|
| 315 |
+
"loss": 0.4091,
|
| 316 |
+
"step": 37
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 4.222222222222222,
|
| 320 |
+
"grad_norm": 149349.15625,
|
| 321 |
+
"learning_rate": 4.363803399173174e-07,
|
| 322 |
+
"loss": 0.4158,
|
| 323 |
+
"step": 38
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 4.333333333333333,
|
| 327 |
+
"grad_norm": 58734.1015625,
|
| 328 |
+
"learning_rate": 4.478640330730363e-07,
|
| 329 |
+
"loss": 0.3904,
|
| 330 |
+
"step": 39
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 4.444444444444445,
|
| 334 |
+
"grad_norm": 51783.84765625,
|
| 335 |
+
"learning_rate": 4.593477262287552e-07,
|
| 336 |
+
"loss": 0.3906,
|
| 337 |
+
"step": 40
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 4.555555555555555,
|
| 341 |
+
"grad_norm": 144734.21875,
|
| 342 |
+
"learning_rate": 4.70831419384474e-07,
|
| 343 |
+
"loss": 0.4004,
|
| 344 |
+
"step": 41
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.666666666666667,
|
| 348 |
+
"grad_norm": 81174.1640625,
|
| 349 |
+
"learning_rate": 4.823151125401929e-07,
|
| 350 |
+
"loss": 0.4091,
|
| 351 |
+
"step": 42
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 4.777777777777778,
|
| 355 |
+
"grad_norm": 214313.296875,
|
| 356 |
+
"learning_rate": 4.937988056959118e-07,
|
| 357 |
+
"loss": 0.4024,
|
| 358 |
+
"step": 43
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 4.888888888888889,
|
| 362 |
+
"grad_norm": 94263.2265625,
|
| 363 |
+
"learning_rate": 5.052824988516307e-07,
|
| 364 |
+
"loss": 0.3783,
|
| 365 |
+
"step": 44
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 5.0,
|
| 369 |
+
"grad_norm": 224359.484375,
|
| 370 |
+
"learning_rate": 5.167661920073495e-07,
|
| 371 |
+
"loss": 0.4167,
|
| 372 |
+
"step": 45
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 5.0,
|
| 376 |
+
"eval_accuracy": 0.8128158015617822,
|
| 377 |
+
"eval_f1": 0.8015583150718286,
|
| 378 |
+
"eval_loss": 0.40888261795043945,
|
| 379 |
+
"eval_precision": 0.8116370808678501,
|
| 380 |
+
"eval_recall": 0.7917267917267917,
|
| 381 |
+
"eval_runtime": 3.9592,
|
| 382 |
+
"eval_samples_per_second": 1099.713,
|
| 383 |
+
"eval_steps_per_second": 0.758,
|
| 384 |
+
"step": 45
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 5.111111111111111,
|
| 388 |
+
"grad_norm": 67686.3359375,
|
| 389 |
+
"learning_rate": 5.282498851630684e-07,
|
| 390 |
+
"loss": 0.4068,
|
| 391 |
+
"step": 46
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 5.222222222222222,
|
| 395 |
+
"grad_norm": 95565.5234375,
|
| 396 |
+
"learning_rate": 5.397335783187873e-07,
|
| 397 |
+
"loss": 0.3813,
|
| 398 |
+
"step": 47
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 5.333333333333333,
|
| 402 |
+
"grad_norm": 64774.375,
|
| 403 |
+
"learning_rate": 5.512172714745061e-07,
|
| 404 |
+
"loss": 0.3878,
|
| 405 |
+
"step": 48
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 5.444444444444445,
|
| 409 |
+
"grad_norm": 202234.890625,
|
| 410 |
+
"learning_rate": 5.62700964630225e-07,
|
| 411 |
+
"loss": 0.4293,
|
| 412 |
+
"step": 49
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 5.555555555555555,
|
| 416 |
+
"grad_norm": 201268.0625,
|
| 417 |
+
"learning_rate": 5.74184657785944e-07,
|
| 418 |
+
"loss": 0.3929,
|
| 419 |
+
"step": 50
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 5.666666666666667,
|
| 423 |
+
"grad_norm": 75988.9375,
|
| 424 |
+
"learning_rate": 5.856683509416628e-07,
|
| 425 |
+
"loss": 0.4009,
|
| 426 |
+
"step": 51
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 5.777777777777778,
|
| 430 |
+
"grad_norm": 96718.375,
|
| 431 |
+
"learning_rate": 5.971520440973818e-07,
|
| 432 |
+
"loss": 0.3964,
|
| 433 |
+
"step": 52
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 5.888888888888889,
|
| 437 |
+
"grad_norm": 159498.375,
|
| 438 |
+
"learning_rate": 6.086357372531007e-07,
|
| 439 |
+
"loss": 0.3998,
|
| 440 |
+
"step": 53
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 6.0,
|
| 444 |
+
"grad_norm": 182416.921875,
|
| 445 |
+
"learning_rate": 6.201194304088196e-07,
|
| 446 |
+
"loss": 0.4103,
|
| 447 |
+
"step": 54
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 6.0,
|
| 451 |
+
"eval_accuracy": 0.8109784106568673,
|
| 452 |
+
"eval_f1": 0.8012557353296306,
|
| 453 |
+
"eval_loss": 0.4089536666870117,
|
| 454 |
+
"eval_precision": 0.8045586808923375,
|
| 455 |
+
"eval_recall": 0.797979797979798,
|
| 456 |
+
"eval_runtime": 3.9918,
|
| 457 |
+
"eval_samples_per_second": 1090.745,
|
| 458 |
+
"eval_steps_per_second": 0.752,
|
| 459 |
+
"step": 54
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 6.111111111111111,
|
| 463 |
+
"grad_norm": 61874.33984375,
|
| 464 |
+
"learning_rate": 6.316031235645384e-07,
|
| 465 |
+
"loss": 0.3988,
|
| 466 |
+
"step": 55
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 6.222222222222222,
|
| 470 |
+
"grad_norm": 124532.828125,
|
| 471 |
+
"learning_rate": 6.430868167202573e-07,
|
| 472 |
+
"loss": 0.3987,
|
| 473 |
+
"step": 56
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 6.333333333333333,
|
| 477 |
+
"grad_norm": 236617.484375,
|
| 478 |
+
"learning_rate": 6.545705098759762e-07,
|
| 479 |
+
"loss": 0.4024,
|
| 480 |
+
"step": 57
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 6.444444444444445,
|
| 484 |
+
"grad_norm": 66161.015625,
|
| 485 |
+
"learning_rate": 6.66054203031695e-07,
|
| 486 |
+
"loss": 0.3784,
|
| 487 |
+
"step": 58
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 6.555555555555555,
|
| 491 |
+
"grad_norm": 266440.5,
|
| 492 |
+
"learning_rate": 6.775378961874139e-07,
|
| 493 |
+
"loss": 0.4216,
|
| 494 |
+
"step": 59
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 6.666666666666667,
|
| 498 |
+
"grad_norm": 248766.765625,
|
| 499 |
+
"learning_rate": 6.890215893431328e-07,
|
| 500 |
+
"loss": 0.3842,
|
| 501 |
+
"step": 60
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 6.777777777777778,
|
| 505 |
+
"grad_norm": 87982.4609375,
|
| 506 |
+
"learning_rate": 7.005052824988516e-07,
|
| 507 |
+
"loss": 0.4236,
|
| 508 |
+
"step": 61
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 6.888888888888889,
|
| 512 |
+
"grad_norm": 87643.5390625,
|
| 513 |
+
"learning_rate": 7.119889756545705e-07,
|
| 514 |
+
"loss": 0.4015,
|
| 515 |
+
"step": 62
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 7.0,
|
| 519 |
+
"grad_norm": 199694.375,
|
| 520 |
+
"learning_rate": 7.234726688102894e-07,
|
| 521 |
+
"loss": 0.4042,
|
| 522 |
+
"step": 63
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 7.0,
|
| 526 |
+
"eval_accuracy": 0.811437758383096,
|
| 527 |
+
"eval_f1": 0.8005829487490892,
|
| 528 |
+
"eval_loss": 0.4083126485347748,
|
| 529 |
+
"eval_precision": 0.8086359175662414,
|
| 530 |
+
"eval_recall": 0.7926887926887927,
|
| 531 |
+
"eval_runtime": 4.0191,
|
| 532 |
+
"eval_samples_per_second": 1083.34,
|
| 533 |
+
"eval_steps_per_second": 0.746,
|
| 534 |
+
"step": 63
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 7.111111111111111,
|
| 538 |
+
"grad_norm": 141664.1875,
|
| 539 |
+
"learning_rate": 7.349563619660083e-07,
|
| 540 |
+
"loss": 0.397,
|
| 541 |
+
"step": 64
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 7.222222222222222,
|
| 545 |
+
"grad_norm": 224376.9375,
|
| 546 |
+
"learning_rate": 7.464400551217272e-07,
|
| 547 |
+
"loss": 0.3918,
|
| 548 |
+
"step": 65
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 7.333333333333333,
|
| 552 |
+
"grad_norm": 53294.6640625,
|
| 553 |
+
"learning_rate": 7.579237482774461e-07,
|
| 554 |
+
"loss": 0.4015,
|
| 555 |
+
"step": 66
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 7.444444444444445,
|
| 559 |
+
"grad_norm": 104196.5859375,
|
| 560 |
+
"learning_rate": 7.694074414331649e-07,
|
| 561 |
+
"loss": 0.4077,
|
| 562 |
+
"step": 67
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 7.555555555555555,
|
| 566 |
+
"grad_norm": 242109.171875,
|
| 567 |
+
"learning_rate": 7.808911345888838e-07,
|
| 568 |
+
"loss": 0.4047,
|
| 569 |
+
"step": 68
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 7.666666666666667,
|
| 573 |
+
"grad_norm": 88400.484375,
|
| 574 |
+
"learning_rate": 7.923748277446027e-07,
|
| 575 |
+
"loss": 0.4036,
|
| 576 |
+
"step": 69
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 7.777777777777778,
|
| 580 |
+
"grad_norm": 98138.4140625,
|
| 581 |
+
"learning_rate": 8.038585209003215e-07,
|
| 582 |
+
"loss": 0.4145,
|
| 583 |
+
"step": 70
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 7.888888888888889,
|
| 587 |
+
"grad_norm": 126328.7421875,
|
| 588 |
+
"learning_rate": 8.153422140560404e-07,
|
| 589 |
+
"loss": 0.3981,
|
| 590 |
+
"step": 71
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 8.0,
|
| 594 |
+
"grad_norm": 134507.5,
|
| 595 |
+
"learning_rate": 8.268259072117593e-07,
|
| 596 |
+
"loss": 0.3705,
|
| 597 |
+
"step": 72
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 8.0,
|
| 601 |
+
"eval_accuracy": 0.8112080845199816,
|
| 602 |
+
"eval_f1": 0.7997076023391813,
|
| 603 |
+
"eval_loss": 0.40784016251564026,
|
| 604 |
+
"eval_precision": 0.8103703703703704,
|
| 605 |
+
"eval_recall": 0.7893217893217893,
|
| 606 |
+
"eval_runtime": 4.0716,
|
| 607 |
+
"eval_samples_per_second": 1069.369,
|
| 608 |
+
"eval_steps_per_second": 0.737,
|
| 609 |
+
"step": 72
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 8.11111111111111,
|
| 613 |
+
"grad_norm": 74213.140625,
|
| 614 |
+
"learning_rate": 8.383096003674781e-07,
|
| 615 |
+
"loss": 0.3839,
|
| 616 |
+
"step": 73
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 8.222222222222221,
|
| 620 |
+
"grad_norm": 123429.9375,
|
| 621 |
+
"learning_rate": 8.49793293523197e-07,
|
| 622 |
+
"loss": 0.4077,
|
| 623 |
+
"step": 74
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 8.333333333333334,
|
| 627 |
+
"grad_norm": 103865.6015625,
|
| 628 |
+
"learning_rate": 8.61276986678916e-07,
|
| 629 |
+
"loss": 0.4074,
|
| 630 |
+
"step": 75
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 8.444444444444445,
|
| 634 |
+
"grad_norm": 185123.984375,
|
| 635 |
+
"learning_rate": 8.727606798346348e-07,
|
| 636 |
+
"loss": 0.4,
|
| 637 |
+
"step": 76
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 8.555555555555555,
|
| 641 |
+
"grad_norm": 130541.5,
|
| 642 |
+
"learning_rate": 8.842443729903537e-07,
|
| 643 |
+
"loss": 0.4067,
|
| 644 |
+
"step": 77
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 8.666666666666666,
|
| 648 |
+
"grad_norm": 167941.90625,
|
| 649 |
+
"learning_rate": 8.957280661460726e-07,
|
| 650 |
+
"loss": 0.4234,
|
| 651 |
+
"step": 78
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 8.777777777777779,
|
| 655 |
+
"grad_norm": 214580.09375,
|
| 656 |
+
"learning_rate": 9.072117593017914e-07,
|
| 657 |
+
"loss": 0.382,
|
| 658 |
+
"step": 79
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 8.88888888888889,
|
| 662 |
+
"grad_norm": 253802.171875,
|
| 663 |
+
"learning_rate": 9.186954524575103e-07,
|
| 664 |
+
"loss": 0.3714,
|
| 665 |
+
"step": 80
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 9.0,
|
| 669 |
+
"grad_norm": 246684.4375,
|
| 670 |
+
"learning_rate": 9.301791456132293e-07,
|
| 671 |
+
"loss": 0.395,
|
| 672 |
+
"step": 81
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 9.0,
|
| 676 |
+
"eval_accuracy": 0.811437758383096,
|
| 677 |
+
"eval_f1": 0.8004860267314702,
|
| 678 |
+
"eval_loss": 0.40752363204956055,
|
| 679 |
+
"eval_precision": 0.8089390962671905,
|
| 680 |
+
"eval_recall": 0.7922077922077922,
|
| 681 |
+
"eval_runtime": 4.1246,
|
| 682 |
+
"eval_samples_per_second": 1055.617,
|
| 683 |
+
"eval_steps_per_second": 0.727,
|
| 684 |
+
"step": 81
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 9.11111111111111,
|
| 688 |
+
"grad_norm": 134944.796875,
|
| 689 |
+
"learning_rate": 9.41662838768948e-07,
|
| 690 |
+
"loss": 0.3812,
|
| 691 |
+
"step": 82
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 9.222222222222221,
|
| 695 |
+
"grad_norm": 76687.96875,
|
| 696 |
+
"learning_rate": 9.53146531924667e-07,
|
| 697 |
+
"loss": 0.3871,
|
| 698 |
+
"step": 83
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 9.333333333333334,
|
| 702 |
+
"grad_norm": 134176.203125,
|
| 703 |
+
"learning_rate": 9.646302250803859e-07,
|
| 704 |
+
"loss": 0.408,
|
| 705 |
+
"step": 84
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 9.444444444444445,
|
| 709 |
+
"grad_norm": 89754.578125,
|
| 710 |
+
"learning_rate": 9.761139182361047e-07,
|
| 711 |
+
"loss": 0.3973,
|
| 712 |
+
"step": 85
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 9.555555555555555,
|
| 716 |
+
"grad_norm": 195871.5,
|
| 717 |
+
"learning_rate": 9.875976113918237e-07,
|
| 718 |
+
"loss": 0.3998,
|
| 719 |
+
"step": 86
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 9.666666666666666,
|
| 723 |
+
"grad_norm": 199506.5625,
|
| 724 |
+
"learning_rate": 9.990813045475425e-07,
|
| 725 |
+
"loss": 0.4201,
|
| 726 |
+
"step": 87
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 9.777777777777779,
|
| 730 |
+
"grad_norm": 71133.1171875,
|
| 731 |
+
"learning_rate": 1.0105649977032615e-06,
|
| 732 |
+
"loss": 0.4047,
|
| 733 |
+
"step": 88
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 9.88888888888889,
|
| 737 |
+
"grad_norm": 109423.734375,
|
| 738 |
+
"learning_rate": 1.0220486908589803e-06,
|
| 739 |
+
"loss": 0.3942,
|
| 740 |
+
"step": 89
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 10.0,
|
| 744 |
+
"grad_norm": 326489.625,
|
| 745 |
+
"learning_rate": 1.033532384014699e-06,
|
| 746 |
+
"loss": 0.3836,
|
| 747 |
+
"step": 90
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 10.0,
|
| 751 |
+
"eval_accuracy": 0.8109784106568673,
|
| 752 |
+
"eval_f1": 0.8011597004107273,
|
| 753 |
+
"eval_loss": 0.4074622392654419,
|
| 754 |
+
"eval_precision": 0.8048543689320389,
|
| 755 |
+
"eval_recall": 0.7974987974987975,
|
| 756 |
+
"eval_runtime": 4.1511,
|
| 757 |
+
"eval_samples_per_second": 1048.879,
|
| 758 |
+
"eval_steps_per_second": 0.723,
|
| 759 |
+
"step": 90
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 10.11111111111111,
|
| 763 |
+
"grad_norm": 141166.140625,
|
| 764 |
+
"learning_rate": 1.045016077170418e-06,
|
| 765 |
+
"loss": 0.408,
|
| 766 |
+
"step": 91
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 10.222222222222221,
|
| 770 |
+
"grad_norm": 69169.859375,
|
| 771 |
+
"learning_rate": 1.0564997703261369e-06,
|
| 772 |
+
"loss": 0.3887,
|
| 773 |
+
"step": 92
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 10.333333333333334,
|
| 777 |
+
"grad_norm": 56820.30859375,
|
| 778 |
+
"learning_rate": 1.0679834634818557e-06,
|
| 779 |
+
"loss": 0.4045,
|
| 780 |
+
"step": 93
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 10.444444444444445,
|
| 784 |
+
"grad_norm": 82678.1875,
|
| 785 |
+
"learning_rate": 1.0794671566375747e-06,
|
| 786 |
+
"loss": 0.3867,
|
| 787 |
+
"step": 94
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 10.555555555555555,
|
| 791 |
+
"grad_norm": 66057.0703125,
|
| 792 |
+
"learning_rate": 1.0909508497932935e-06,
|
| 793 |
+
"loss": 0.3956,
|
| 794 |
+
"step": 95
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 10.666666666666666,
|
| 798 |
+
"grad_norm": 208573.453125,
|
| 799 |
+
"learning_rate": 1.1024345429490123e-06,
|
| 800 |
+
"loss": 0.4234,
|
| 801 |
+
"step": 96
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 10.777777777777779,
|
| 805 |
+
"grad_norm": 89480.4375,
|
| 806 |
+
"learning_rate": 1.1139182361047313e-06,
|
| 807 |
+
"loss": 0.3979,
|
| 808 |
+
"step": 97
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 10.88888888888889,
|
| 812 |
+
"grad_norm": 162307.484375,
|
| 813 |
+
"learning_rate": 1.12540192926045e-06,
|
| 814 |
+
"loss": 0.396,
|
| 815 |
+
"step": 98
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 11.0,
|
| 819 |
+
"grad_norm": 89120.2109375,
|
| 820 |
+
"learning_rate": 1.136885622416169e-06,
|
| 821 |
+
"loss": 0.4105,
|
| 822 |
+
"step": 99
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 11.0,
|
| 826 |
+
"eval_accuracy": 0.8123564538355535,
|
| 827 |
+
"eval_f1": 0.8010713416118822,
|
| 828 |
+
"eval_loss": 0.4068741500377655,
|
| 829 |
+
"eval_precision": 0.8111439842209073,
|
| 830 |
+
"eval_recall": 0.7912457912457912,
|
| 831 |
+
"eval_runtime": 4.1736,
|
| 832 |
+
"eval_samples_per_second": 1043.218,
|
| 833 |
+
"eval_steps_per_second": 0.719,
|
| 834 |
+
"step": 99
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 11.11111111111111,
|
| 838 |
+
"grad_norm": 93675.4296875,
|
| 839 |
+
"learning_rate": 1.148369315571888e-06,
|
| 840 |
+
"loss": 0.3865,
|
| 841 |
+
"step": 100
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 11.222222222222221,
|
| 845 |
+
"grad_norm": 172021.828125,
|
| 846 |
+
"learning_rate": 1.1598530087276067e-06,
|
| 847 |
+
"loss": 0.3964,
|
| 848 |
+
"step": 101
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 11.333333333333334,
|
| 852 |
+
"grad_norm": 123080.0546875,
|
| 853 |
+
"learning_rate": 1.1713367018833255e-06,
|
| 854 |
+
"loss": 0.4166,
|
| 855 |
+
"step": 102
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 11.444444444444445,
|
| 859 |
+
"grad_norm": 71145.3203125,
|
| 860 |
+
"learning_rate": 1.1828203950390445e-06,
|
| 861 |
+
"loss": 0.3878,
|
| 862 |
+
"step": 103
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 11.555555555555555,
|
| 866 |
+
"grad_norm": 173710.0,
|
| 867 |
+
"learning_rate": 1.1943040881947635e-06,
|
| 868 |
+
"loss": 0.3828,
|
| 869 |
+
"step": 104
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 11.666666666666666,
|
| 873 |
+
"grad_norm": 165428.84375,
|
| 874 |
+
"learning_rate": 1.2057877813504825e-06,
|
| 875 |
+
"loss": 0.3871,
|
| 876 |
+
"step": 105
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 11.777777777777779,
|
| 880 |
+
"grad_norm": 267474.3125,
|
| 881 |
+
"learning_rate": 1.2172714745062013e-06,
|
| 882 |
+
"loss": 0.4003,
|
| 883 |
+
"step": 106
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 11.88888888888889,
|
| 887 |
+
"grad_norm": 162812.609375,
|
| 888 |
+
"learning_rate": 1.2287551676619201e-06,
|
| 889 |
+
"loss": 0.4023,
|
| 890 |
+
"step": 107
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 12.0,
|
| 894 |
+
"grad_norm": 166778.921875,
|
| 895 |
+
"learning_rate": 1.2402388608176391e-06,
|
| 896 |
+
"loss": 0.4018,
|
| 897 |
+
"step": 108
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 12.0,
|
| 901 |
+
"eval_accuracy": 0.8118971061093248,
|
| 902 |
+
"eval_f1": 0.8019347037484885,
|
| 903 |
+
"eval_loss": 0.4067639112472534,
|
| 904 |
+
"eval_precision": 0.806420233463035,
|
| 905 |
+
"eval_recall": 0.7974987974987975,
|
| 906 |
+
"eval_runtime": 4.1933,
|
| 907 |
+
"eval_samples_per_second": 1038.322,
|
| 908 |
+
"eval_steps_per_second": 0.715,
|
| 909 |
+
"step": 108
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 12.11111111111111,
|
| 913 |
+
"grad_norm": 112960.7890625,
|
| 914 |
+
"learning_rate": 1.251722553973358e-06,
|
| 915 |
+
"loss": 0.3929,
|
| 916 |
+
"step": 109
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 12.222222222222221,
|
| 920 |
+
"grad_norm": 301508.375,
|
| 921 |
+
"learning_rate": 1.2632062471290767e-06,
|
| 922 |
+
"loss": 0.4061,
|
| 923 |
+
"step": 110
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 12.333333333333334,
|
| 927 |
+
"grad_norm": 254152.53125,
|
| 928 |
+
"learning_rate": 1.2746899402847958e-06,
|
| 929 |
+
"loss": 0.4029,
|
| 930 |
+
"step": 111
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 12.444444444444445,
|
| 934 |
+
"grad_norm": 104616.0859375,
|
| 935 |
+
"learning_rate": 1.2861736334405146e-06,
|
| 936 |
+
"loss": 0.3967,
|
| 937 |
+
"step": 112
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 12.555555555555555,
|
| 941 |
+
"grad_norm": 166745.109375,
|
| 942 |
+
"learning_rate": 1.2976573265962333e-06,
|
| 943 |
+
"loss": 0.4009,
|
| 944 |
+
"step": 113
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 12.666666666666666,
|
| 948 |
+
"grad_norm": 58958.81640625,
|
| 949 |
+
"learning_rate": 1.3091410197519524e-06,
|
| 950 |
+
"loss": 0.3897,
|
| 951 |
+
"step": 114
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 12.777777777777779,
|
| 955 |
+
"grad_norm": 322038.0625,
|
| 956 |
+
"learning_rate": 1.3206247129076712e-06,
|
| 957 |
+
"loss": 0.405,
|
| 958 |
+
"step": 115
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 12.88888888888889,
|
| 962 |
+
"grad_norm": 92102.4140625,
|
| 963 |
+
"learning_rate": 1.33210840606339e-06,
|
| 964 |
+
"loss": 0.4068,
|
| 965 |
+
"step": 116
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 13.0,
|
| 969 |
+
"grad_norm": 150015.234375,
|
| 970 |
+
"learning_rate": 1.343592099219109e-06,
|
| 971 |
+
"loss": 0.365,
|
| 972 |
+
"step": 117
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 13.0,
|
| 976 |
+
"eval_accuracy": 0.8118971061093248,
|
| 977 |
+
"eval_f1": 0.8020304568527918,
|
| 978 |
+
"eval_loss": 0.4065192937850952,
|
| 979 |
+
"eval_precision": 0.8061224489795918,
|
| 980 |
+
"eval_recall": 0.797979797979798,
|
| 981 |
+
"eval_runtime": 4.2544,
|
| 982 |
+
"eval_samples_per_second": 1023.409,
|
| 983 |
+
"eval_steps_per_second": 0.705,
|
| 984 |
+
"step": 117
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 13.11111111111111,
|
| 988 |
+
"grad_norm": 95103.921875,
|
| 989 |
+
"learning_rate": 1.3550757923748278e-06,
|
| 990 |
+
"loss": 0.4172,
|
| 991 |
+
"step": 118
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 13.222222222222221,
|
| 995 |
+
"grad_norm": 113910.25,
|
| 996 |
+
"learning_rate": 1.3665594855305466e-06,
|
| 997 |
+
"loss": 0.3879,
|
| 998 |
+
"step": 119
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 13.333333333333334,
|
| 1002 |
+
"grad_norm": 171457.375,
|
| 1003 |
+
"learning_rate": 1.3780431786862656e-06,
|
| 1004 |
+
"loss": 0.4148,
|
| 1005 |
+
"step": 120
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 13.444444444444445,
|
| 1009 |
+
"grad_norm": 266275.84375,
|
| 1010 |
+
"learning_rate": 1.3895268718419844e-06,
|
| 1011 |
+
"loss": 0.3913,
|
| 1012 |
+
"step": 121
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 13.555555555555555,
|
| 1016 |
+
"grad_norm": 56887.02734375,
|
| 1017 |
+
"learning_rate": 1.4010105649977032e-06,
|
| 1018 |
+
"loss": 0.4045,
|
| 1019 |
+
"step": 122
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 13.666666666666666,
|
| 1023 |
+
"grad_norm": 59645.34765625,
|
| 1024 |
+
"learning_rate": 1.4124942581534222e-06,
|
| 1025 |
+
"loss": 0.4117,
|
| 1026 |
+
"step": 123
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 13.777777777777779,
|
| 1030 |
+
"grad_norm": 236620.359375,
|
| 1031 |
+
"learning_rate": 1.423977951309141e-06,
|
| 1032 |
+
"loss": 0.3871,
|
| 1033 |
+
"step": 124
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 13.88888888888889,
|
| 1037 |
+
"grad_norm": 60322.78515625,
|
| 1038 |
+
"learning_rate": 1.4354616444648598e-06,
|
| 1039 |
+
"loss": 0.3879,
|
| 1040 |
+
"step": 125
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 14.0,
|
| 1044 |
+
"grad_norm": 121960.0,
|
| 1045 |
+
"learning_rate": 1.4469453376205788e-06,
|
| 1046 |
+
"loss": 0.3973,
|
| 1047 |
+
"step": 126
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 14.0,
|
| 1051 |
+
"eval_accuracy": 0.813275149288011,
|
| 1052 |
+
"eval_f1": 0.8043321299638989,
|
| 1053 |
+
"eval_loss": 0.4063498079776764,
|
| 1054 |
+
"eval_precision": 0.8049132947976878,
|
| 1055 |
+
"eval_recall": 0.8037518037518038,
|
| 1056 |
+
"eval_runtime": 4.2859,
|
| 1057 |
+
"eval_samples_per_second": 1015.878,
|
| 1058 |
+
"eval_steps_per_second": 0.7,
|
| 1059 |
+
"step": 126
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 14.11111111111111,
|
| 1063 |
+
"grad_norm": 207117.59375,
|
| 1064 |
+
"learning_rate": 1.4584290307762978e-06,
|
| 1065 |
+
"loss": 0.4161,
|
| 1066 |
+
"step": 127
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 14.222222222222221,
|
| 1070 |
+
"grad_norm": 126602.25,
|
| 1071 |
+
"learning_rate": 1.4699127239320166e-06,
|
| 1072 |
+
"loss": 0.4169,
|
| 1073 |
+
"step": 128
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 14.333333333333334,
|
| 1077 |
+
"grad_norm": 130595.4296875,
|
| 1078 |
+
"learning_rate": 1.4813964170877356e-06,
|
| 1079 |
+
"loss": 0.3908,
|
| 1080 |
+
"step": 129
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 14.444444444444445,
|
| 1084 |
+
"grad_norm": 112879.7578125,
|
| 1085 |
+
"learning_rate": 1.4928801102434544e-06,
|
| 1086 |
+
"loss": 0.3991,
|
| 1087 |
+
"step": 130
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 14.555555555555555,
|
| 1091 |
+
"grad_norm": 155991.609375,
|
| 1092 |
+
"learning_rate": 1.5043638033991732e-06,
|
| 1093 |
+
"loss": 0.3913,
|
| 1094 |
+
"step": 131
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 14.666666666666666,
|
| 1098 |
+
"grad_norm": 159955.5,
|
| 1099 |
+
"learning_rate": 1.5158474965548922e-06,
|
| 1100 |
+
"loss": 0.3898,
|
| 1101 |
+
"step": 132
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 14.777777777777779,
|
| 1105 |
+
"grad_norm": 118653.71875,
|
| 1106 |
+
"learning_rate": 1.527331189710611e-06,
|
| 1107 |
+
"loss": 0.3725,
|
| 1108 |
+
"step": 133
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 14.88888888888889,
|
| 1112 |
+
"grad_norm": 250835.40625,
|
| 1113 |
+
"learning_rate": 1.5388148828663298e-06,
|
| 1114 |
+
"loss": 0.384,
|
| 1115 |
+
"step": 134
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 15.0,
|
| 1119 |
+
"grad_norm": 181872.25,
|
| 1120 |
+
"learning_rate": 1.5502985760220488e-06,
|
| 1121 |
+
"loss": 0.3987,
|
| 1122 |
+
"step": 135
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 15.0,
|
| 1126 |
+
"eval_accuracy": 0.8137344970142398,
|
| 1127 |
+
"eval_f1": 0.8015659407878639,
|
| 1128 |
+
"eval_loss": 0.4055207073688507,
|
| 1129 |
+
"eval_precision": 0.8157370517928287,
|
| 1130 |
+
"eval_recall": 0.7878787878787878,
|
| 1131 |
+
"eval_runtime": 4.3323,
|
| 1132 |
+
"eval_samples_per_second": 1005.014,
|
| 1133 |
+
"eval_steps_per_second": 0.692,
|
| 1134 |
+
"step": 135
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 15.11111111111111,
|
| 1138 |
+
"grad_norm": 95328.6484375,
|
| 1139 |
+
"learning_rate": 1.5617822691777676e-06,
|
| 1140 |
+
"loss": 0.3815,
|
| 1141 |
+
"step": 136
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 15.222222222222221,
|
| 1145 |
+
"grad_norm": 119895.4375,
|
| 1146 |
+
"learning_rate": 1.5732659623334864e-06,
|
| 1147 |
+
"loss": 0.4165,
|
| 1148 |
+
"step": 137
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 15.333333333333334,
|
| 1152 |
+
"grad_norm": 248995.484375,
|
| 1153 |
+
"learning_rate": 1.5847496554892054e-06,
|
| 1154 |
+
"loss": 0.3914,
|
| 1155 |
+
"step": 138
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 15.444444444444445,
|
| 1159 |
+
"grad_norm": 89783.515625,
|
| 1160 |
+
"learning_rate": 1.5962333486449242e-06,
|
| 1161 |
+
"loss": 0.3873,
|
| 1162 |
+
"step": 139
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 15.555555555555555,
|
| 1166 |
+
"grad_norm": 129638.3671875,
|
| 1167 |
+
"learning_rate": 1.607717041800643e-06,
|
| 1168 |
+
"loss": 0.3887,
|
| 1169 |
+
"step": 140
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 15.666666666666666,
|
| 1173 |
+
"grad_norm": 103552.0859375,
|
| 1174 |
+
"learning_rate": 1.619200734956362e-06,
|
| 1175 |
+
"loss": 0.4078,
|
| 1176 |
+
"step": 141
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 15.777777777777779,
|
| 1180 |
+
"grad_norm": 274991.84375,
|
| 1181 |
+
"learning_rate": 1.6306844281120808e-06,
|
| 1182 |
+
"loss": 0.3978,
|
| 1183 |
+
"step": 142
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 15.88888888888889,
|
| 1187 |
+
"grad_norm": 289934.40625,
|
| 1188 |
+
"learning_rate": 1.6421681212677996e-06,
|
| 1189 |
+
"loss": 0.3748,
|
| 1190 |
+
"step": 143
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 16.0,
|
| 1194 |
+
"grad_norm": 113621.3359375,
|
| 1195 |
+
"learning_rate": 1.6536518144235186e-06,
|
| 1196 |
+
"loss": 0.4035,
|
| 1197 |
+
"step": 144
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 16.0,
|
| 1201 |
+
"eval_accuracy": 0.8125861276986679,
|
| 1202 |
+
"eval_f1": 0.8022297624818225,
|
| 1203 |
+
"eval_loss": 0.40545299649238586,
|
| 1204 |
+
"eval_precision": 0.8085002442598925,
|
| 1205 |
+
"eval_recall": 0.796055796055796,
|
| 1206 |
+
"eval_runtime": 4.347,
|
| 1207 |
+
"eval_samples_per_second": 1001.605,
|
| 1208 |
+
"eval_steps_per_second": 0.69,
|
| 1209 |
+
"step": 144
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 16.11111111111111,
|
| 1213 |
+
"grad_norm": 146140.265625,
|
| 1214 |
+
"learning_rate": 1.6651355075792374e-06,
|
| 1215 |
+
"loss": 0.3882,
|
| 1216 |
+
"step": 145
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 16.22222222222222,
|
| 1220 |
+
"grad_norm": 108806.140625,
|
| 1221 |
+
"learning_rate": 1.6766192007349562e-06,
|
| 1222 |
+
"loss": 0.409,
|
| 1223 |
+
"step": 146
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 16.333333333333332,
|
| 1227 |
+
"grad_norm": 157681.09375,
|
| 1228 |
+
"learning_rate": 1.6881028938906753e-06,
|
| 1229 |
+
"loss": 0.3915,
|
| 1230 |
+
"step": 147
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 16.444444444444443,
|
| 1234 |
+
"grad_norm": 100690.9375,
|
| 1235 |
+
"learning_rate": 1.699586587046394e-06,
|
| 1236 |
+
"loss": 0.3947,
|
| 1237 |
+
"step": 148
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 16.555555555555557,
|
| 1241 |
+
"grad_norm": 401684.625,
|
| 1242 |
+
"learning_rate": 1.7110702802021129e-06,
|
| 1243 |
+
"loss": 0.4007,
|
| 1244 |
+
"step": 149
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 16.666666666666668,
|
| 1248 |
+
"grad_norm": 195763.71875,
|
| 1249 |
+
"learning_rate": 1.722553973357832e-06,
|
| 1250 |
+
"loss": 0.3802,
|
| 1251 |
+
"step": 150
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 16.77777777777778,
|
| 1255 |
+
"grad_norm": 171871.40625,
|
| 1256 |
+
"learning_rate": 1.7340376665135509e-06,
|
| 1257 |
+
"loss": 0.3989,
|
| 1258 |
+
"step": 151
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 16.88888888888889,
|
| 1262 |
+
"grad_norm": 380800.71875,
|
| 1263 |
+
"learning_rate": 1.7455213596692697e-06,
|
| 1264 |
+
"loss": 0.394,
|
| 1265 |
+
"step": 152
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 17.0,
|
| 1269 |
+
"grad_norm": 131500.703125,
|
| 1270 |
+
"learning_rate": 1.7570050528249887e-06,
|
| 1271 |
+
"loss": 0.3933,
|
| 1272 |
+
"step": 153
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 17.0,
|
| 1276 |
+
"eval_accuracy": 0.8135048231511254,
|
| 1277 |
+
"eval_f1": 0.8044315992292871,
|
| 1278 |
+
"eval_loss": 0.4050571024417877,
|
| 1279 |
+
"eval_precision": 0.8055957549445248,
|
| 1280 |
+
"eval_recall": 0.8032708032708032,
|
| 1281 |
+
"eval_runtime": 4.4245,
|
| 1282 |
+
"eval_samples_per_second": 984.067,
|
| 1283 |
+
"eval_steps_per_second": 0.678,
|
| 1284 |
+
"step": 153
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 17.11111111111111,
|
| 1288 |
+
"grad_norm": 149897.84375,
|
| 1289 |
+
"learning_rate": 1.7684887459807075e-06,
|
| 1290 |
+
"loss": 0.3868,
|
| 1291 |
+
"step": 154
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 17.22222222222222,
|
| 1295 |
+
"grad_norm": 170039.859375,
|
| 1296 |
+
"learning_rate": 1.7799724391364263e-06,
|
| 1297 |
+
"loss": 0.3922,
|
| 1298 |
+
"step": 155
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 17.333333333333332,
|
| 1302 |
+
"grad_norm": 97929.5234375,
|
| 1303 |
+
"learning_rate": 1.7914561322921453e-06,
|
| 1304 |
+
"loss": 0.3879,
|
| 1305 |
+
"step": 156
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 17.444444444444443,
|
| 1309 |
+
"grad_norm": 226100.71875,
|
| 1310 |
+
"learning_rate": 1.802939825447864e-06,
|
| 1311 |
+
"loss": 0.4043,
|
| 1312 |
+
"step": 157
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 17.555555555555557,
|
| 1316 |
+
"grad_norm": 128054.578125,
|
| 1317 |
+
"learning_rate": 1.8144235186035829e-06,
|
| 1318 |
+
"loss": 0.406,
|
| 1319 |
+
"step": 158
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 17.666666666666668,
|
| 1323 |
+
"grad_norm": 319005.8125,
|
| 1324 |
+
"learning_rate": 1.825907211759302e-06,
|
| 1325 |
+
"loss": 0.4046,
|
| 1326 |
+
"step": 159
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 17.77777777777778,
|
| 1330 |
+
"grad_norm": 65652.578125,
|
| 1331 |
+
"learning_rate": 1.8373909049150207e-06,
|
| 1332 |
+
"loss": 0.4033,
|
| 1333 |
+
"step": 160
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 17.88888888888889,
|
| 1337 |
+
"grad_norm": 153639.953125,
|
| 1338 |
+
"learning_rate": 1.8488745980707395e-06,
|
| 1339 |
+
"loss": 0.3785,
|
| 1340 |
+
"step": 161
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 18.0,
|
| 1344 |
+
"grad_norm": 459349.78125,
|
| 1345 |
+
"learning_rate": 1.8603582912264585e-06,
|
| 1346 |
+
"loss": 0.3933,
|
| 1347 |
+
"step": 162
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 18.0,
|
| 1351 |
+
"eval_accuracy": 0.8137344970142398,
|
| 1352 |
+
"eval_f1": 0.801953601953602,
|
| 1353 |
+
"eval_loss": 0.40449145436286926,
|
| 1354 |
+
"eval_precision": 0.814484126984127,
|
| 1355 |
+
"eval_recall": 0.7898027898027898,
|
| 1356 |
+
"eval_runtime": 4.4226,
|
| 1357 |
+
"eval_samples_per_second": 984.498,
|
| 1358 |
+
"eval_steps_per_second": 0.678,
|
| 1359 |
+
"step": 162
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 18.11111111111111,
|
| 1363 |
+
"grad_norm": 164870.09375,
|
| 1364 |
+
"learning_rate": 1.8718419843821773e-06,
|
| 1365 |
+
"loss": 0.3948,
|
| 1366 |
+
"step": 163
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 18.22222222222222,
|
| 1370 |
+
"grad_norm": 48797.765625,
|
| 1371 |
+
"learning_rate": 1.883325677537896e-06,
|
| 1372 |
+
"loss": 0.3938,
|
| 1373 |
+
"step": 164
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 18.333333333333332,
|
| 1377 |
+
"grad_norm": 87001.015625,
|
| 1378 |
+
"learning_rate": 1.8948093706936151e-06,
|
| 1379 |
+
"loss": 0.3968,
|
| 1380 |
+
"step": 165
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 18.444444444444443,
|
| 1384 |
+
"grad_norm": 185905.75,
|
| 1385 |
+
"learning_rate": 1.906293063849334e-06,
|
| 1386 |
+
"loss": 0.3755,
|
| 1387 |
+
"step": 166
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 18.555555555555557,
|
| 1391 |
+
"grad_norm": 381337.5,
|
| 1392 |
+
"learning_rate": 1.917776757005053e-06,
|
| 1393 |
+
"loss": 0.4084,
|
| 1394 |
+
"step": 167
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 18.666666666666668,
|
| 1398 |
+
"grad_norm": 215949.28125,
|
| 1399 |
+
"learning_rate": 1.9292604501607717e-06,
|
| 1400 |
+
"loss": 0.4119,
|
| 1401 |
+
"step": 168
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 18.77777777777778,
|
| 1405 |
+
"grad_norm": 156070.921875,
|
| 1406 |
+
"learning_rate": 1.9407441433164905e-06,
|
| 1407 |
+
"loss": 0.3979,
|
| 1408 |
+
"step": 169
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 18.88888888888889,
|
| 1412 |
+
"grad_norm": 218265.953125,
|
| 1413 |
+
"learning_rate": 1.9522278364722093e-06,
|
| 1414 |
+
"loss": 0.3929,
|
| 1415 |
+
"step": 170
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 19.0,
|
| 1419 |
+
"grad_norm": 130221.5546875,
|
| 1420 |
+
"learning_rate": 1.963711529627928e-06,
|
| 1421 |
+
"loss": 0.3806,
|
| 1422 |
+
"step": 171
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 19.0,
|
| 1426 |
+
"eval_accuracy": 0.8130454754248967,
|
| 1427 |
+
"eval_f1": 0.8001963672066765,
|
| 1428 |
+
"eval_loss": 0.4040713608264923,
|
| 1429 |
+
"eval_precision": 0.8170426065162907,
|
| 1430 |
+
"eval_recall": 0.7840307840307841,
|
| 1431 |
+
"eval_runtime": 4.4758,
|
| 1432 |
+
"eval_samples_per_second": 972.78,
|
| 1433 |
+
"eval_steps_per_second": 0.67,
|
| 1434 |
+
"step": 171
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 19.11111111111111,
|
| 1438 |
+
"grad_norm": 244566.015625,
|
| 1439 |
+
"learning_rate": 1.9751952227836473e-06,
|
| 1440 |
+
"loss": 0.3735,
|
| 1441 |
+
"step": 172
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 19.22222222222222,
|
| 1445 |
+
"grad_norm": 113897.046875,
|
| 1446 |
+
"learning_rate": 1.986678915939366e-06,
|
| 1447 |
+
"loss": 0.3872,
|
| 1448 |
+
"step": 173
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 19.333333333333332,
|
| 1452 |
+
"grad_norm": 167560.140625,
|
| 1453 |
+
"learning_rate": 1.998162609095085e-06,
|
| 1454 |
+
"loss": 0.4203,
|
| 1455 |
+
"step": 174
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 19.444444444444443,
|
| 1459 |
+
"grad_norm": 234886.8125,
|
| 1460 |
+
"learning_rate": 2.0096463022508037e-06,
|
| 1461 |
+
"loss": 0.3856,
|
| 1462 |
+
"step": 175
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 19.555555555555557,
|
| 1466 |
+
"grad_norm": 378111.03125,
|
| 1467 |
+
"learning_rate": 2.021129995406523e-06,
|
| 1468 |
+
"loss": 0.4226,
|
| 1469 |
+
"step": 176
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 19.666666666666668,
|
| 1473 |
+
"grad_norm": 128344.5546875,
|
| 1474 |
+
"learning_rate": 2.0326136885622413e-06,
|
| 1475 |
+
"loss": 0.3742,
|
| 1476 |
+
"step": 177
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 19.77777777777778,
|
| 1480 |
+
"grad_norm": 741984.8125,
|
| 1481 |
+
"learning_rate": 2.0440973817179606e-06,
|
| 1482 |
+
"loss": 0.4082,
|
| 1483 |
+
"step": 178
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 19.88888888888889,
|
| 1487 |
+
"grad_norm": 383275.90625,
|
| 1488 |
+
"learning_rate": 2.0555810748736794e-06,
|
| 1489 |
+
"loss": 0.3992,
|
| 1490 |
+
"step": 179
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 20.0,
|
| 1494 |
+
"grad_norm": 242016.46875,
|
| 1495 |
+
"learning_rate": 2.067064768029398e-06,
|
| 1496 |
+
"loss": 0.3855,
|
| 1497 |
+
"step": 180
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 20.0,
|
| 1501 |
+
"eval_accuracy": 0.8174092788240698,
|
| 1502 |
+
"eval_f1": 0.8134240788547289,
|
| 1503 |
+
"eval_loss": 0.40715256333351135,
|
| 1504 |
+
"eval_precision": 0.7942254812098992,
|
| 1505 |
+
"eval_recall": 0.8335738335738335,
|
| 1506 |
+
"eval_runtime": 4.5663,
|
| 1507 |
+
"eval_samples_per_second": 953.516,
|
| 1508 |
+
"eval_steps_per_second": 0.657,
|
| 1509 |
+
"step": 180
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 20.11111111111111,
|
| 1513 |
+
"grad_norm": 443388.9375,
|
| 1514 |
+
"learning_rate": 2.078548461185117e-06,
|
| 1515 |
+
"loss": 0.4031,
|
| 1516 |
+
"step": 181
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 20.22222222222222,
|
| 1520 |
+
"grad_norm": 401805.96875,
|
| 1521 |
+
"learning_rate": 2.090032154340836e-06,
|
| 1522 |
+
"loss": 0.3979,
|
| 1523 |
+
"step": 182
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 20.333333333333332,
|
| 1527 |
+
"grad_norm": 141372.265625,
|
| 1528 |
+
"learning_rate": 2.1015158474965545e-06,
|
| 1529 |
+
"loss": 0.3798,
|
| 1530 |
+
"step": 183
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 20.444444444444443,
|
| 1534 |
+
"grad_norm": 205886.125,
|
| 1535 |
+
"learning_rate": 2.1129995406522738e-06,
|
| 1536 |
+
"loss": 0.3828,
|
| 1537 |
+
"step": 184
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 20.555555555555557,
|
| 1541 |
+
"grad_norm": 369303.71875,
|
| 1542 |
+
"learning_rate": 2.124483233807993e-06,
|
| 1543 |
+
"loss": 0.4017,
|
| 1544 |
+
"step": 185
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 20.666666666666668,
|
| 1548 |
+
"grad_norm": 126559.546875,
|
| 1549 |
+
"learning_rate": 2.1359669269637114e-06,
|
| 1550 |
+
"loss": 0.3991,
|
| 1551 |
+
"step": 186
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 20.77777777777778,
|
| 1555 |
+
"grad_norm": 251650.140625,
|
| 1556 |
+
"learning_rate": 2.1474506201194306e-06,
|
| 1557 |
+
"loss": 0.4049,
|
| 1558 |
+
"step": 187
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 20.88888888888889,
|
| 1562 |
+
"grad_norm": 539118.375,
|
| 1563 |
+
"learning_rate": 2.1589343132751494e-06,
|
| 1564 |
+
"loss": 0.3799,
|
| 1565 |
+
"step": 188
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 21.0,
|
| 1569 |
+
"grad_norm": 433059.34375,
|
| 1570 |
+
"learning_rate": 2.170418006430868e-06,
|
| 1571 |
+
"loss": 0.4132,
|
| 1572 |
+
"step": 189
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 21.0,
|
| 1576 |
+
"eval_accuracy": 0.8135048231511254,
|
| 1577 |
+
"eval_f1": 0.8007850834151129,
|
| 1578 |
+
"eval_loss": 0.40335243940353394,
|
| 1579 |
+
"eval_precision": 0.8172258387581373,
|
| 1580 |
+
"eval_recall": 0.784992784992785,
|
| 1581 |
+
"eval_runtime": 4.5819,
|
| 1582 |
+
"eval_samples_per_second": 950.264,
|
| 1583 |
+
"eval_steps_per_second": 0.655,
|
| 1584 |
+
"step": 189
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 21.11111111111111,
|
| 1588 |
+
"grad_norm": 82284.109375,
|
| 1589 |
+
"learning_rate": 2.181901699586587e-06,
|
| 1590 |
+
"loss": 0.3892,
|
| 1591 |
+
"step": 190
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 21.22222222222222,
|
| 1595 |
+
"grad_norm": 77919.21875,
|
| 1596 |
+
"learning_rate": 2.193385392742306e-06,
|
| 1597 |
+
"loss": 0.4028,
|
| 1598 |
+
"step": 191
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 21.333333333333332,
|
| 1602 |
+
"grad_norm": 357442.125,
|
| 1603 |
+
"learning_rate": 2.2048690858980246e-06,
|
| 1604 |
+
"loss": 0.3961,
|
| 1605 |
+
"step": 192
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 21.444444444444443,
|
| 1609 |
+
"grad_norm": 94285.2109375,
|
| 1610 |
+
"learning_rate": 2.216352779053744e-06,
|
| 1611 |
+
"loss": 0.4058,
|
| 1612 |
+
"step": 193
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 21.555555555555557,
|
| 1616 |
+
"grad_norm": 243661.1875,
|
| 1617 |
+
"learning_rate": 2.2278364722094626e-06,
|
| 1618 |
+
"loss": 0.3708,
|
| 1619 |
+
"step": 194
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 21.666666666666668,
|
| 1623 |
+
"grad_norm": 63565.203125,
|
| 1624 |
+
"learning_rate": 2.2393201653651814e-06,
|
| 1625 |
+
"loss": 0.4102,
|
| 1626 |
+
"step": 195
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 21.77777777777778,
|
| 1630 |
+
"grad_norm": 339940.78125,
|
| 1631 |
+
"learning_rate": 2.2508038585209e-06,
|
| 1632 |
+
"loss": 0.4072,
|
| 1633 |
+
"step": 196
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 21.88888888888889,
|
| 1637 |
+
"grad_norm": 214842.84375,
|
| 1638 |
+
"learning_rate": 2.2622875516766194e-06,
|
| 1639 |
+
"loss": 0.384,
|
| 1640 |
+
"step": 197
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 22.0,
|
| 1644 |
+
"grad_norm": 269673.40625,
|
| 1645 |
+
"learning_rate": 2.273771244832338e-06,
|
| 1646 |
+
"loss": 0.3774,
|
| 1647 |
+
"step": 198
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 22.0,
|
| 1651 |
+
"eval_accuracy": 0.8130454754248967,
|
| 1652 |
+
"eval_f1": 0.8005879470847623,
|
| 1653 |
+
"eval_loss": 0.40258845686912537,
|
| 1654 |
+
"eval_precision": 0.8157763354967549,
|
| 1655 |
+
"eval_recall": 0.785954785954786,
|
| 1656 |
+
"eval_runtime": 4.6468,
|
| 1657 |
+
"eval_samples_per_second": 936.992,
|
| 1658 |
+
"eval_steps_per_second": 0.646,
|
| 1659 |
+
"step": 198
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 22.11111111111111,
|
| 1663 |
+
"grad_norm": 147949.28125,
|
| 1664 |
+
"learning_rate": 2.285254937988057e-06,
|
| 1665 |
+
"loss": 0.3924,
|
| 1666 |
+
"step": 199
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 22.22222222222222,
|
| 1670 |
+
"grad_norm": 192279.4375,
|
| 1671 |
+
"learning_rate": 2.296738631143776e-06,
|
| 1672 |
+
"loss": 0.4075,
|
| 1673 |
+
"step": 200
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 22.333333333333332,
|
| 1677 |
+
"grad_norm": 478090.40625,
|
| 1678 |
+
"learning_rate": 2.3082223242994946e-06,
|
| 1679 |
+
"loss": 0.4006,
|
| 1680 |
+
"step": 201
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 22.444444444444443,
|
| 1684 |
+
"grad_norm": 264078.875,
|
| 1685 |
+
"learning_rate": 2.3197060174552134e-06,
|
| 1686 |
+
"loss": 0.4051,
|
| 1687 |
+
"step": 202
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 22.555555555555557,
|
| 1691 |
+
"grad_norm": 122800.484375,
|
| 1692 |
+
"learning_rate": 2.3311897106109326e-06,
|
| 1693 |
+
"loss": 0.3937,
|
| 1694 |
+
"step": 203
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 22.666666666666668,
|
| 1698 |
+
"grad_norm": 97360.9375,
|
| 1699 |
+
"learning_rate": 2.342673403766651e-06,
|
| 1700 |
+
"loss": 0.3764,
|
| 1701 |
+
"step": 204
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 22.77777777777778,
|
| 1705 |
+
"grad_norm": 108199.5546875,
|
| 1706 |
+
"learning_rate": 2.3541570969223702e-06,
|
| 1707 |
+
"loss": 0.3932,
|
| 1708 |
+
"step": 205
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 22.88888888888889,
|
| 1712 |
+
"grad_norm": 262427.3125,
|
| 1713 |
+
"learning_rate": 2.365640790078089e-06,
|
| 1714 |
+
"loss": 0.4015,
|
| 1715 |
+
"step": 206
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 23.0,
|
| 1719 |
+
"grad_norm": 412134.5625,
|
| 1720 |
+
"learning_rate": 2.3771244832338083e-06,
|
| 1721 |
+
"loss": 0.3698,
|
| 1722 |
+
"step": 207
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 23.0,
|
| 1726 |
+
"eval_accuracy": 0.8174092788240698,
|
| 1727 |
+
"eval_f1": 0.8103078024337866,
|
| 1728 |
+
"eval_loss": 0.4036799371242523,
|
| 1729 |
+
"eval_precision": 0.8039772727272727,
|
| 1730 |
+
"eval_recall": 0.8167388167388168,
|
| 1731 |
+
"eval_runtime": 4.658,
|
| 1732 |
+
"eval_samples_per_second": 934.744,
|
| 1733 |
+
"eval_steps_per_second": 0.644,
|
| 1734 |
+
"step": 207
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 23.11111111111111,
|
| 1738 |
+
"grad_norm": 265666.25,
|
| 1739 |
+
"learning_rate": 2.388608176389527e-06,
|
| 1740 |
+
"loss": 0.386,
|
| 1741 |
+
"step": 208
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 23.22222222222222,
|
| 1745 |
+
"grad_norm": 138313.109375,
|
| 1746 |
+
"learning_rate": 2.400091869545246e-06,
|
| 1747 |
+
"loss": 0.3904,
|
| 1748 |
+
"step": 209
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 23.333333333333332,
|
| 1752 |
+
"grad_norm": 116725.1171875,
|
| 1753 |
+
"learning_rate": 2.411575562700965e-06,
|
| 1754 |
+
"loss": 0.3955,
|
| 1755 |
+
"step": 210
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 23.444444444444443,
|
| 1759 |
+
"grad_norm": 187285.765625,
|
| 1760 |
+
"learning_rate": 2.4230592558566834e-06,
|
| 1761 |
+
"loss": 0.4069,
|
| 1762 |
+
"step": 211
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 23.555555555555557,
|
| 1766 |
+
"grad_norm": 121359.4765625,
|
| 1767 |
+
"learning_rate": 2.4345429490124027e-06,
|
| 1768 |
+
"loss": 0.4008,
|
| 1769 |
+
"step": 212
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 23.666666666666668,
|
| 1773 |
+
"grad_norm": 115836.03125,
|
| 1774 |
+
"learning_rate": 2.4460266421681215e-06,
|
| 1775 |
+
"loss": 0.4002,
|
| 1776 |
+
"step": 213
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 23.77777777777778,
|
| 1780 |
+
"grad_norm": 76411.34375,
|
| 1781 |
+
"learning_rate": 2.4575103353238403e-06,
|
| 1782 |
+
"loss": 0.3934,
|
| 1783 |
+
"step": 214
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 23.88888888888889,
|
| 1787 |
+
"grad_norm": 154353.21875,
|
| 1788 |
+
"learning_rate": 2.468994028479559e-06,
|
| 1789 |
+
"loss": 0.4057,
|
| 1790 |
+
"step": 215
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 24.0,
|
| 1794 |
+
"grad_norm": 486543.96875,
|
| 1795 |
+
"learning_rate": 2.4804777216352783e-06,
|
| 1796 |
+
"loss": 0.3584,
|
| 1797 |
+
"step": 216
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 24.0,
|
| 1801 |
+
"eval_accuracy": 0.8158015617822691,
|
| 1802 |
+
"eval_f1": 0.8058111380145279,
|
| 1803 |
+
"eval_loss": 0.40210822224617004,
|
| 1804 |
+
"eval_precision": 0.8113115553388591,
|
| 1805 |
+
"eval_recall": 0.8003848003848004,
|
| 1806 |
+
"eval_runtime": 5.1111,
|
| 1807 |
+
"eval_samples_per_second": 851.866,
|
| 1808 |
+
"eval_steps_per_second": 0.587,
|
| 1809 |
+
"step": 216
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 24.11111111111111,
|
| 1813 |
+
"grad_norm": 76995.2578125,
|
| 1814 |
+
"learning_rate": 2.4919614147909967e-06,
|
| 1815 |
+
"loss": 0.4137,
|
| 1816 |
+
"step": 217
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 24.22222222222222,
|
| 1820 |
+
"grad_norm": 192273.9375,
|
| 1821 |
+
"learning_rate": 2.503445107946716e-06,
|
| 1822 |
+
"loss": 0.4011,
|
| 1823 |
+
"step": 218
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 24.333333333333332,
|
| 1827 |
+
"grad_norm": 61227.8359375,
|
| 1828 |
+
"learning_rate": 2.5149288011024347e-06,
|
| 1829 |
+
"loss": 0.3909,
|
| 1830 |
+
"step": 219
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 24.444444444444443,
|
| 1834 |
+
"grad_norm": 221185.5,
|
| 1835 |
+
"learning_rate": 2.5264124942581535e-06,
|
| 1836 |
+
"loss": 0.3973,
|
| 1837 |
+
"step": 220
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 24.555555555555557,
|
| 1841 |
+
"grad_norm": 130241.0859375,
|
| 1842 |
+
"learning_rate": 2.5378961874138723e-06,
|
| 1843 |
+
"loss": 0.3877,
|
| 1844 |
+
"step": 221
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 24.666666666666668,
|
| 1848 |
+
"grad_norm": 80599.9765625,
|
| 1849 |
+
"learning_rate": 2.5493798805695915e-06,
|
| 1850 |
+
"loss": 0.4051,
|
| 1851 |
+
"step": 222
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 24.77777777777778,
|
| 1855 |
+
"grad_norm": 128989.890625,
|
| 1856 |
+
"learning_rate": 2.56086357372531e-06,
|
| 1857 |
+
"loss": 0.3691,
|
| 1858 |
+
"step": 223
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 24.88888888888889,
|
| 1862 |
+
"grad_norm": 320645.8125,
|
| 1863 |
+
"learning_rate": 2.572347266881029e-06,
|
| 1864 |
+
"loss": 0.3791,
|
| 1865 |
+
"step": 224
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 25.0,
|
| 1869 |
+
"grad_norm": 284049.625,
|
| 1870 |
+
"learning_rate": 2.583830960036748e-06,
|
| 1871 |
+
"loss": 0.3947,
|
| 1872 |
+
"step": 225
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 25.0,
|
| 1876 |
+
"eval_accuracy": 0.813275149288011,
|
| 1877 |
+
"eval_f1": 0.7983130736789879,
|
| 1878 |
+
"eval_loss": 0.40306106209754944,
|
| 1879 |
+
"eval_precision": 0.8242827868852459,
|
| 1880 |
+
"eval_recall": 0.7739297739297739,
|
| 1881 |
+
"eval_runtime": 4.7309,
|
| 1882 |
+
"eval_samples_per_second": 920.339,
|
| 1883 |
+
"eval_steps_per_second": 0.634,
|
| 1884 |
+
"step": 225
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 25.11111111111111,
|
| 1888 |
+
"grad_norm": 519498.875,
|
| 1889 |
+
"learning_rate": 2.5953146531924667e-06,
|
| 1890 |
+
"loss": 0.413,
|
| 1891 |
+
"step": 226
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 25.22222222222222,
|
| 1895 |
+
"grad_norm": 502382.40625,
|
| 1896 |
+
"learning_rate": 2.6067983463481855e-06,
|
| 1897 |
+
"loss": 0.3725,
|
| 1898 |
+
"step": 227
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 25.333333333333332,
|
| 1902 |
+
"grad_norm": 379123.5,
|
| 1903 |
+
"learning_rate": 2.6182820395039047e-06,
|
| 1904 |
+
"loss": 0.3881,
|
| 1905 |
+
"step": 228
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 25.444444444444443,
|
| 1909 |
+
"grad_norm": 190483.9375,
|
| 1910 |
+
"learning_rate": 2.629765732659623e-06,
|
| 1911 |
+
"loss": 0.4083,
|
| 1912 |
+
"step": 229
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 25.555555555555557,
|
| 1916 |
+
"grad_norm": 216956.625,
|
| 1917 |
+
"learning_rate": 2.6412494258153423e-06,
|
| 1918 |
+
"loss": 0.3987,
|
| 1919 |
+
"step": 230
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 25.666666666666668,
|
| 1923 |
+
"grad_norm": 278627.96875,
|
| 1924 |
+
"learning_rate": 2.6527331189710615e-06,
|
| 1925 |
+
"loss": 0.3936,
|
| 1926 |
+
"step": 231
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 25.77777777777778,
|
| 1930 |
+
"grad_norm": 343526.4375,
|
| 1931 |
+
"learning_rate": 2.66421681212678e-06,
|
| 1932 |
+
"loss": 0.3827,
|
| 1933 |
+
"step": 232
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 25.88888888888889,
|
| 1937 |
+
"grad_norm": 536376.0625,
|
| 1938 |
+
"learning_rate": 2.675700505282499e-06,
|
| 1939 |
+
"loss": 0.3932,
|
| 1940 |
+
"step": 233
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 26.0,
|
| 1944 |
+
"grad_norm": 188893.734375,
|
| 1945 |
+
"learning_rate": 2.687184198438218e-06,
|
| 1946 |
+
"loss": 0.3812,
|
| 1947 |
+
"step": 234
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 26.0,
|
| 1951 |
+
"eval_accuracy": 0.8180983004134129,
|
| 1952 |
+
"eval_f1": 0.8092485549132948,
|
| 1953 |
+
"eval_loss": 0.4012051820755005,
|
| 1954 |
+
"eval_precision": 0.8104196816208393,
|
| 1955 |
+
"eval_recall": 0.8080808080808081,
|
| 1956 |
+
"eval_runtime": 4.752,
|
| 1957 |
+
"eval_samples_per_second": 916.254,
|
| 1958 |
+
"eval_steps_per_second": 0.631,
|
| 1959 |
+
"step": 234
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 26.11111111111111,
|
| 1963 |
+
"grad_norm": 89162.7421875,
|
| 1964 |
+
"learning_rate": 2.6986678915939367e-06,
|
| 1965 |
+
"loss": 0.3756,
|
| 1966 |
+
"step": 235
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 26.22222222222222,
|
| 1970 |
+
"grad_norm": 246311.859375,
|
| 1971 |
+
"learning_rate": 2.7101515847496555e-06,
|
| 1972 |
+
"loss": 0.3888,
|
| 1973 |
+
"step": 236
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 26.333333333333332,
|
| 1977 |
+
"grad_norm": 127793.1015625,
|
| 1978 |
+
"learning_rate": 2.7216352779053748e-06,
|
| 1979 |
+
"loss": 0.375,
|
| 1980 |
+
"step": 237
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 26.444444444444443,
|
| 1984 |
+
"grad_norm": 67135.25,
|
| 1985 |
+
"learning_rate": 2.733118971061093e-06,
|
| 1986 |
+
"loss": 0.3977,
|
| 1987 |
+
"step": 238
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 26.555555555555557,
|
| 1991 |
+
"grad_norm": 297287.75,
|
| 1992 |
+
"learning_rate": 2.7446026642168124e-06,
|
| 1993 |
+
"loss": 0.3989,
|
| 1994 |
+
"step": 239
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 26.666666666666668,
|
| 1998 |
+
"grad_norm": 303139.96875,
|
| 1999 |
+
"learning_rate": 2.756086357372531e-06,
|
| 2000 |
+
"loss": 0.4048,
|
| 2001 |
+
"step": 240
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 26.77777777777778,
|
| 2005 |
+
"grad_norm": 606317.625,
|
| 2006 |
+
"learning_rate": 2.76757005052825e-06,
|
| 2007 |
+
"loss": 0.3873,
|
| 2008 |
+
"step": 241
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 26.88888888888889,
|
| 2012 |
+
"grad_norm": 172750.578125,
|
| 2013 |
+
"learning_rate": 2.7790537436839687e-06,
|
| 2014 |
+
"loss": 0.384,
|
| 2015 |
+
"step": 242
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 27.0,
|
| 2019 |
+
"grad_norm": 192172.140625,
|
| 2020 |
+
"learning_rate": 2.790537436839688e-06,
|
| 2021 |
+
"loss": 0.3928,
|
| 2022 |
+
"step": 243
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 27.0,
|
| 2026 |
+
"eval_accuracy": 0.8176389526871842,
|
| 2027 |
+
"eval_f1": 0.8086746987951807,
|
| 2028 |
+
"eval_loss": 0.4002959132194519,
|
| 2029 |
+
"eval_precision": 0.8102366006760019,
|
| 2030 |
+
"eval_recall": 0.8071188071188071,
|
| 2031 |
+
"eval_runtime": 4.5908,
|
| 2032 |
+
"eval_samples_per_second": 948.411,
|
| 2033 |
+
"eval_steps_per_second": 0.653,
|
| 2034 |
+
"step": 243
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 27.11111111111111,
|
| 2038 |
+
"grad_norm": 91136.6484375,
|
| 2039 |
+
"learning_rate": 2.8020211299954063e-06,
|
| 2040 |
+
"loss": 0.3889,
|
| 2041 |
+
"step": 244
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 27.22222222222222,
|
| 2045 |
+
"grad_norm": 111455.0859375,
|
| 2046 |
+
"learning_rate": 2.8135048231511256e-06,
|
| 2047 |
+
"loss": 0.3781,
|
| 2048 |
+
"step": 245
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 27.333333333333332,
|
| 2052 |
+
"grad_norm": 241420.875,
|
| 2053 |
+
"learning_rate": 2.8249885163068444e-06,
|
| 2054 |
+
"loss": 0.3868,
|
| 2055 |
+
"step": 246
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 27.444444444444443,
|
| 2059 |
+
"grad_norm": 108206.5078125,
|
| 2060 |
+
"learning_rate": 2.836472209462563e-06,
|
| 2061 |
+
"loss": 0.4015,
|
| 2062 |
+
"step": 247
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 27.555555555555557,
|
| 2066 |
+
"grad_norm": 197683.21875,
|
| 2067 |
+
"learning_rate": 2.847955902618282e-06,
|
| 2068 |
+
"loss": 0.3877,
|
| 2069 |
+
"step": 248
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 27.666666666666668,
|
| 2073 |
+
"grad_norm": 141794.296875,
|
| 2074 |
+
"learning_rate": 2.859439595774001e-06,
|
| 2075 |
+
"loss": 0.4144,
|
| 2076 |
+
"step": 249
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 27.77777777777778,
|
| 2080 |
+
"grad_norm": 256132.671875,
|
| 2081 |
+
"learning_rate": 2.8709232889297196e-06,
|
| 2082 |
+
"loss": 0.3757,
|
| 2083 |
+
"step": 250
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 27.88888888888889,
|
| 2087 |
+
"grad_norm": 245631.25,
|
| 2088 |
+
"learning_rate": 2.8824069820854388e-06,
|
| 2089 |
+
"loss": 0.3927,
|
| 2090 |
+
"step": 251
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 28.0,
|
| 2094 |
+
"grad_norm": 213683.140625,
|
| 2095 |
+
"learning_rate": 2.8938906752411576e-06,
|
| 2096 |
+
"loss": 0.3591,
|
| 2097 |
+
"step": 252
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 28.0,
|
| 2101 |
+
"eval_accuracy": 0.8183279742765274,
|
| 2102 |
+
"eval_f1": 0.8090755491189959,
|
| 2103 |
+
"eval_loss": 0.39960750937461853,
|
| 2104 |
+
"eval_precision": 0.812015503875969,
|
| 2105 |
+
"eval_recall": 0.8061568061568062,
|
| 2106 |
+
"eval_runtime": 4.6253,
|
| 2107 |
+
"eval_samples_per_second": 941.347,
|
| 2108 |
+
"eval_steps_per_second": 0.649,
|
| 2109 |
+
"step": 252
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 28.11111111111111,
|
| 2113 |
+
"grad_norm": 148696.1875,
|
| 2114 |
+
"learning_rate": 2.9053743683968764e-06,
|
| 2115 |
+
"loss": 0.3861,
|
| 2116 |
+
"step": 253
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 28.22222222222222,
|
| 2120 |
+
"grad_norm": 119950.6875,
|
| 2121 |
+
"learning_rate": 2.9168580615525956e-06,
|
| 2122 |
+
"loss": 0.3861,
|
| 2123 |
+
"step": 254
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 28.333333333333332,
|
| 2127 |
+
"grad_norm": 110785.578125,
|
| 2128 |
+
"learning_rate": 2.9283417547083144e-06,
|
| 2129 |
+
"loss": 0.3765,
|
| 2130 |
+
"step": 255
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 28.444444444444443,
|
| 2134 |
+
"grad_norm": 103968.75,
|
| 2135 |
+
"learning_rate": 2.939825447864033e-06,
|
| 2136 |
+
"loss": 0.3874,
|
| 2137 |
+
"step": 256
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 28.555555555555557,
|
| 2141 |
+
"grad_norm": 231702.34375,
|
| 2142 |
+
"learning_rate": 2.951309141019752e-06,
|
| 2143 |
+
"loss": 0.4047,
|
| 2144 |
+
"step": 257
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 28.666666666666668,
|
| 2148 |
+
"grad_norm": 127703.3359375,
|
| 2149 |
+
"learning_rate": 2.9627928341754712e-06,
|
| 2150 |
+
"loss": 0.3974,
|
| 2151 |
+
"step": 258
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 28.77777777777778,
|
| 2155 |
+
"grad_norm": 246982.40625,
|
| 2156 |
+
"learning_rate": 2.9742765273311896e-06,
|
| 2157 |
+
"loss": 0.3823,
|
| 2158 |
+
"step": 259
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 28.88888888888889,
|
| 2162 |
+
"grad_norm": 218037.109375,
|
| 2163 |
+
"learning_rate": 2.985760220486909e-06,
|
| 2164 |
+
"loss": 0.4092,
|
| 2165 |
+
"step": 260
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 29.0,
|
| 2169 |
+
"grad_norm": 91798.8125,
|
| 2170 |
+
"learning_rate": 2.9972439136426276e-06,
|
| 2171 |
+
"loss": 0.3872,
|
| 2172 |
+
"step": 261
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 29.0,
|
| 2176 |
+
"eval_accuracy": 0.8194763435920992,
|
| 2177 |
+
"eval_f1": 0.8103281853281853,
|
| 2178 |
+
"eval_loss": 0.39871689677238464,
|
| 2179 |
+
"eval_precision": 0.8130750605326876,
|
| 2180 |
+
"eval_recall": 0.8075998075998077,
|
| 2181 |
+
"eval_runtime": 4.6817,
|
| 2182 |
+
"eval_samples_per_second": 929.994,
|
| 2183 |
+
"eval_steps_per_second": 0.641,
|
| 2184 |
+
"step": 261
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 29.11111111111111,
|
| 2188 |
+
"grad_norm": 94122.96875,
|
| 2189 |
+
"learning_rate": 3.0087276067983464e-06,
|
| 2190 |
+
"loss": 0.3935,
|
| 2191 |
+
"step": 262
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 29.22222222222222,
|
| 2195 |
+
"grad_norm": 138885.125,
|
| 2196 |
+
"learning_rate": 3.0202112999540652e-06,
|
| 2197 |
+
"loss": 0.3795,
|
| 2198 |
+
"step": 263
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 29.333333333333332,
|
| 2202 |
+
"grad_norm": 433285.8125,
|
| 2203 |
+
"learning_rate": 3.0316949931097844e-06,
|
| 2204 |
+
"loss": 0.3886,
|
| 2205 |
+
"step": 264
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 29.444444444444443,
|
| 2209 |
+
"grad_norm": 320547.09375,
|
| 2210 |
+
"learning_rate": 3.043178686265503e-06,
|
| 2211 |
+
"loss": 0.3925,
|
| 2212 |
+
"step": 265
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 29.555555555555557,
|
| 2216 |
+
"grad_norm": 82943.8203125,
|
| 2217 |
+
"learning_rate": 3.054662379421222e-06,
|
| 2218 |
+
"loss": 0.3808,
|
| 2219 |
+
"step": 266
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 29.666666666666668,
|
| 2223 |
+
"grad_norm": 70880.1328125,
|
| 2224 |
+
"learning_rate": 3.066146072576941e-06,
|
| 2225 |
+
"loss": 0.3883,
|
| 2226 |
+
"step": 267
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 29.77777777777778,
|
| 2230 |
+
"grad_norm": 270253.625,
|
| 2231 |
+
"learning_rate": 3.0776297657326596e-06,
|
| 2232 |
+
"loss": 0.381,
|
| 2233 |
+
"step": 268
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 29.88888888888889,
|
| 2237 |
+
"grad_norm": 147631.03125,
|
| 2238 |
+
"learning_rate": 3.0891134588883784e-06,
|
| 2239 |
+
"loss": 0.3622,
|
| 2240 |
+
"step": 269
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 30.0,
|
| 2244 |
+
"grad_norm": 443471.0,
|
| 2245 |
+
"learning_rate": 3.1005971520440976e-06,
|
| 2246 |
+
"loss": 0.383,
|
| 2247 |
+
"step": 270
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 30.0,
|
| 2251 |
+
"eval_accuracy": 0.8208543867707855,
|
| 2252 |
+
"eval_f1": 0.8132183908045977,
|
| 2253 |
+
"eval_loss": 0.3988674283027649,
|
| 2254 |
+
"eval_precision": 0.8097281831187411,
|
| 2255 |
+
"eval_recall": 0.8167388167388168,
|
| 2256 |
+
"eval_runtime": 4.7223,
|
| 2257 |
+
"eval_samples_per_second": 922.003,
|
| 2258 |
+
"eval_steps_per_second": 0.635,
|
| 2259 |
+
"step": 270
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 30.11111111111111,
|
| 2263 |
+
"grad_norm": 141594.78125,
|
| 2264 |
+
"learning_rate": 3.112080845199816e-06,
|
| 2265 |
+
"loss": 0.3939,
|
| 2266 |
+
"step": 271
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 30.22222222222222,
|
| 2270 |
+
"grad_norm": 388412.25,
|
| 2271 |
+
"learning_rate": 3.1235645383555352e-06,
|
| 2272 |
+
"loss": 0.4021,
|
| 2273 |
+
"step": 272
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 30.333333333333332,
|
| 2277 |
+
"grad_norm": 296736.5,
|
| 2278 |
+
"learning_rate": 3.135048231511254e-06,
|
| 2279 |
+
"loss": 0.3748,
|
| 2280 |
+
"step": 273
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 30.444444444444443,
|
| 2284 |
+
"grad_norm": 245713.109375,
|
| 2285 |
+
"learning_rate": 3.146531924666973e-06,
|
| 2286 |
+
"loss": 0.3833,
|
| 2287 |
+
"step": 274
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 30.555555555555557,
|
| 2291 |
+
"grad_norm": 128015.53125,
|
| 2292 |
+
"learning_rate": 3.1580156178226916e-06,
|
| 2293 |
+
"loss": 0.382,
|
| 2294 |
+
"step": 275
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 30.666666666666668,
|
| 2298 |
+
"grad_norm": 517641.25,
|
| 2299 |
+
"learning_rate": 3.169499310978411e-06,
|
| 2300 |
+
"loss": 0.3751,
|
| 2301 |
+
"step": 276
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 30.77777777777778,
|
| 2305 |
+
"grad_norm": 620331.375,
|
| 2306 |
+
"learning_rate": 3.1809830041341297e-06,
|
| 2307 |
+
"loss": 0.385,
|
| 2308 |
+
"step": 277
|
| 2309 |
+
},
|
| 2310 |
+
{
|
| 2311 |
+
"epoch": 30.88888888888889,
|
| 2312 |
+
"grad_norm": 141372.8125,
|
| 2313 |
+
"learning_rate": 3.1924666972898485e-06,
|
| 2314 |
+
"loss": 0.3737,
|
| 2315 |
+
"step": 278
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 31.0,
|
| 2319 |
+
"grad_norm": 292104.09375,
|
| 2320 |
+
"learning_rate": 3.2039503904455677e-06,
|
| 2321 |
+
"loss": 0.3966,
|
| 2322 |
+
"step": 279
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 31.0,
|
| 2326 |
+
"eval_accuracy": 0.8135048231511254,
|
| 2327 |
+
"eval_f1": 0.8000984736582964,
|
| 2328 |
+
"eval_loss": 0.39696839451789856,
|
| 2329 |
+
"eval_precision": 0.8194654563792234,
|
| 2330 |
+
"eval_recall": 0.7816257816257817,
|
| 2331 |
+
"eval_runtime": 5.1915,
|
| 2332 |
+
"eval_samples_per_second": 838.675,
|
| 2333 |
+
"eval_steps_per_second": 0.578,
|
| 2334 |
+
"step": 279
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 31.11111111111111,
|
| 2338 |
+
"grad_norm": 268970.09375,
|
| 2339 |
+
"learning_rate": 3.215434083601286e-06,
|
| 2340 |
+
"loss": 0.3771,
|
| 2341 |
+
"step": 280
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 31.22222222222222,
|
| 2345 |
+
"grad_norm": 113689.015625,
|
| 2346 |
+
"learning_rate": 3.2269177767570053e-06,
|
| 2347 |
+
"loss": 0.3979,
|
| 2348 |
+
"step": 281
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 31.333333333333332,
|
| 2352 |
+
"grad_norm": 617523.125,
|
| 2353 |
+
"learning_rate": 3.238401469912724e-06,
|
| 2354 |
+
"loss": 0.3778,
|
| 2355 |
+
"step": 282
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 31.444444444444443,
|
| 2359 |
+
"grad_norm": 615918.875,
|
| 2360 |
+
"learning_rate": 3.249885163068443e-06,
|
| 2361 |
+
"loss": 0.393,
|
| 2362 |
+
"step": 283
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 31.555555555555557,
|
| 2366 |
+
"grad_norm": 262001.734375,
|
| 2367 |
+
"learning_rate": 3.2613688562241617e-06,
|
| 2368 |
+
"loss": 0.3789,
|
| 2369 |
+
"step": 284
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 31.666666666666668,
|
| 2373 |
+
"grad_norm": 454978.09375,
|
| 2374 |
+
"learning_rate": 3.272852549379881e-06,
|
| 2375 |
+
"loss": 0.4019,
|
| 2376 |
+
"step": 285
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 31.77777777777778,
|
| 2380 |
+
"grad_norm": 635157.0,
|
| 2381 |
+
"learning_rate": 3.2843362425355993e-06,
|
| 2382 |
+
"loss": 0.3867,
|
| 2383 |
+
"step": 286
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 31.88888888888889,
|
| 2387 |
+
"grad_norm": 180403.53125,
|
| 2388 |
+
"learning_rate": 3.2958199356913185e-06,
|
| 2389 |
+
"loss": 0.3861,
|
| 2390 |
+
"step": 287
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 32.0,
|
| 2394 |
+
"grad_norm": 741212.1875,
|
| 2395 |
+
"learning_rate": 3.3073036288470373e-06,
|
| 2396 |
+
"loss": 0.4076,
|
| 2397 |
+
"step": 288
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 32.0,
|
| 2401 |
+
"eval_accuracy": 0.8139641708773542,
|
| 2402 |
+
"eval_f1": 0.7975,
|
| 2403 |
+
"eval_loss": 0.39702439308166504,
|
| 2404 |
+
"eval_precision": 0.8302967204580948,
|
| 2405 |
+
"eval_recall": 0.7671957671957672,
|
| 2406 |
+
"eval_runtime": 4.8033,
|
| 2407 |
+
"eval_samples_per_second": 906.467,
|
| 2408 |
+
"eval_steps_per_second": 0.625,
|
| 2409 |
+
"step": 288
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 32.111111111111114,
|
| 2413 |
+
"grad_norm": 535774.0625,
|
| 2414 |
+
"learning_rate": 3.318787322002756e-06,
|
| 2415 |
+
"loss": 0.4063,
|
| 2416 |
+
"step": 289
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 32.22222222222222,
|
| 2420 |
+
"grad_norm": 334477.8125,
|
| 2421 |
+
"learning_rate": 3.330271015158475e-06,
|
| 2422 |
+
"loss": 0.3786,
|
| 2423 |
+
"step": 290
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 32.333333333333336,
|
| 2427 |
+
"grad_norm": 257520.453125,
|
| 2428 |
+
"learning_rate": 3.341754708314194e-06,
|
| 2429 |
+
"loss": 0.3938,
|
| 2430 |
+
"step": 291
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 32.44444444444444,
|
| 2434 |
+
"grad_norm": 136671.375,
|
| 2435 |
+
"learning_rate": 3.3532384014699125e-06,
|
| 2436 |
+
"loss": 0.3573,
|
| 2437 |
+
"step": 292
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 32.55555555555556,
|
| 2441 |
+
"grad_norm": 202197.96875,
|
| 2442 |
+
"learning_rate": 3.3647220946256317e-06,
|
| 2443 |
+
"loss": 0.381,
|
| 2444 |
+
"step": 293
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 32.666666666666664,
|
| 2448 |
+
"grad_norm": 415211.40625,
|
| 2449 |
+
"learning_rate": 3.3762057877813505e-06,
|
| 2450 |
+
"loss": 0.3936,
|
| 2451 |
+
"step": 294
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 32.77777777777778,
|
| 2455 |
+
"grad_norm": 508906.4375,
|
| 2456 |
+
"learning_rate": 3.3876894809370693e-06,
|
| 2457 |
+
"loss": 0.3865,
|
| 2458 |
+
"step": 295
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 32.888888888888886,
|
| 2462 |
+
"grad_norm": 526735.875,
|
| 2463 |
+
"learning_rate": 3.399173174092788e-06,
|
| 2464 |
+
"loss": 0.3889,
|
| 2465 |
+
"step": 296
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 33.0,
|
| 2469 |
+
"grad_norm": 254718.359375,
|
| 2470 |
+
"learning_rate": 3.4106568672485073e-06,
|
| 2471 |
+
"loss": 0.3876,
|
| 2472 |
+
"step": 297
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 33.0,
|
| 2476 |
+
"eval_accuracy": 0.8128158015617822,
|
| 2477 |
+
"eval_f1": 0.7953803665578709,
|
| 2478 |
+
"eval_loss": 0.3970797061920166,
|
| 2479 |
+
"eval_precision": 0.8319327731092437,
|
| 2480 |
+
"eval_recall": 0.7619047619047619,
|
| 2481 |
+
"eval_runtime": 4.8434,
|
| 2482 |
+
"eval_samples_per_second": 898.952,
|
| 2483 |
+
"eval_steps_per_second": 0.619,
|
| 2484 |
+
"step": 297
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 33.111111111111114,
|
| 2488 |
+
"grad_norm": 985664.3125,
|
| 2489 |
+
"learning_rate": 3.4221405604042257e-06,
|
| 2490 |
+
"loss": 0.393,
|
| 2491 |
+
"step": 298
|
| 2492 |
+
},
|
| 2493 |
+
{
|
| 2494 |
+
"epoch": 33.22222222222222,
|
| 2495 |
+
"grad_norm": 617586.375,
|
| 2496 |
+
"learning_rate": 3.433624253559945e-06,
|
| 2497 |
+
"loss": 0.3748,
|
| 2498 |
+
"step": 299
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 33.333333333333336,
|
| 2502 |
+
"grad_norm": 1082033.875,
|
| 2503 |
+
"learning_rate": 3.445107946715664e-06,
|
| 2504 |
+
"loss": 0.367,
|
| 2505 |
+
"step": 300
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 33.44444444444444,
|
| 2509 |
+
"grad_norm": 715623.125,
|
| 2510 |
+
"learning_rate": 3.4565916398713825e-06,
|
| 2511 |
+
"loss": 0.4088,
|
| 2512 |
+
"step": 301
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 33.55555555555556,
|
| 2516 |
+
"grad_norm": 899131.875,
|
| 2517 |
+
"learning_rate": 3.4680753330271017e-06,
|
| 2518 |
+
"loss": 0.4071,
|
| 2519 |
+
"step": 302
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 33.666666666666664,
|
| 2523 |
+
"grad_norm": 365789.3125,
|
| 2524 |
+
"learning_rate": 3.4795590261828205e-06,
|
| 2525 |
+
"loss": 0.3735,
|
| 2526 |
+
"step": 303
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 33.77777777777778,
|
| 2530 |
+
"grad_norm": 465953.4375,
|
| 2531 |
+
"learning_rate": 3.4910427193385393e-06,
|
| 2532 |
+
"loss": 0.3921,
|
| 2533 |
+
"step": 304
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 33.888888888888886,
|
| 2537 |
+
"grad_norm": 895318.9375,
|
| 2538 |
+
"learning_rate": 3.502526412494258e-06,
|
| 2539 |
+
"loss": 0.3905,
|
| 2540 |
+
"step": 305
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 34.0,
|
| 2544 |
+
"grad_norm": 879623.6875,
|
| 2545 |
+
"learning_rate": 3.5140101056499774e-06,
|
| 2546 |
+
"loss": 0.3815,
|
| 2547 |
+
"step": 306
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 34.0,
|
| 2551 |
+
"eval_accuracy": 0.8240698208543867,
|
| 2552 |
+
"eval_f1": 0.8204406938584153,
|
| 2553 |
+
"eval_loss": 0.39920851588249207,
|
| 2554 |
+
"eval_precision": 0.8001828989483311,
|
| 2555 |
+
"eval_recall": 0.8417508417508418,
|
| 2556 |
+
"eval_runtime": 4.8974,
|
| 2557 |
+
"eval_samples_per_second": 889.041,
|
| 2558 |
+
"eval_steps_per_second": 0.613,
|
| 2559 |
+
"step": 306
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 34.111111111111114,
|
| 2563 |
+
"grad_norm": 629364.25,
|
| 2564 |
+
"learning_rate": 3.5254937988056957e-06,
|
| 2565 |
+
"loss": 0.3837,
|
| 2566 |
+
"step": 307
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 34.22222222222222,
|
| 2570 |
+
"grad_norm": 289384.09375,
|
| 2571 |
+
"learning_rate": 3.536977491961415e-06,
|
| 2572 |
+
"loss": 0.3929,
|
| 2573 |
+
"step": 308
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 34.333333333333336,
|
| 2577 |
+
"grad_norm": 151251.28125,
|
| 2578 |
+
"learning_rate": 3.5484611851171338e-06,
|
| 2579 |
+
"loss": 0.397,
|
| 2580 |
+
"step": 309
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 34.44444444444444,
|
| 2584 |
+
"grad_norm": 263268.09375,
|
| 2585 |
+
"learning_rate": 3.5599448782728526e-06,
|
| 2586 |
+
"loss": 0.3895,
|
| 2587 |
+
"step": 310
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 34.55555555555556,
|
| 2591 |
+
"grad_norm": 384067.28125,
|
| 2592 |
+
"learning_rate": 3.5714285714285714e-06,
|
| 2593 |
+
"loss": 0.3933,
|
| 2594 |
+
"step": 311
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 34.666666666666664,
|
| 2598 |
+
"grad_norm": 820030.5,
|
| 2599 |
+
"learning_rate": 3.5829122645842906e-06,
|
| 2600 |
+
"loss": 0.3824,
|
| 2601 |
+
"step": 312
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 34.77777777777778,
|
| 2605 |
+
"grad_norm": 760393.3125,
|
| 2606 |
+
"learning_rate": 3.594395957740009e-06,
|
| 2607 |
+
"loss": 0.3926,
|
| 2608 |
+
"step": 313
|
| 2609 |
+
},
|
| 2610 |
+
{
|
| 2611 |
+
"epoch": 34.888888888888886,
|
| 2612 |
+
"grad_norm": 87286.8359375,
|
| 2613 |
+
"learning_rate": 3.605879650895728e-06,
|
| 2614 |
+
"loss": 0.3757,
|
| 2615 |
+
"step": 314
|
| 2616 |
+
},
|
| 2617 |
+
{
|
| 2618 |
+
"epoch": 35.0,
|
| 2619 |
+
"grad_norm": 178669.046875,
|
| 2620 |
+
"learning_rate": 3.617363344051447e-06,
|
| 2621 |
+
"loss": 0.369,
|
| 2622 |
+
"step": 315
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 35.0,
|
| 2626 |
+
"eval_accuracy": 0.8231511254019293,
|
| 2627 |
+
"eval_f1": 0.8168411037107517,
|
| 2628 |
+
"eval_loss": 0.3967457413673401,
|
| 2629 |
+
"eval_precision": 0.808,
|
| 2630 |
+
"eval_recall": 0.8258778258778259,
|
| 2631 |
+
"eval_runtime": 4.9488,
|
| 2632 |
+
"eval_samples_per_second": 879.801,
|
| 2633 |
+
"eval_steps_per_second": 0.606,
|
| 2634 |
+
"step": 315
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 35.111111111111114,
|
| 2638 |
+
"grad_norm": 358338.8125,
|
| 2639 |
+
"learning_rate": 3.6288470372071658e-06,
|
| 2640 |
+
"loss": 0.3697,
|
| 2641 |
+
"step": 316
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 35.22222222222222,
|
| 2645 |
+
"grad_norm": 202919.984375,
|
| 2646 |
+
"learning_rate": 3.6403307303628846e-06,
|
| 2647 |
+
"loss": 0.3829,
|
| 2648 |
+
"step": 317
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 35.333333333333336,
|
| 2652 |
+
"grad_norm": 600207.375,
|
| 2653 |
+
"learning_rate": 3.651814423518604e-06,
|
| 2654 |
+
"loss": 0.389,
|
| 2655 |
+
"step": 318
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 35.44444444444444,
|
| 2659 |
+
"grad_norm": 101190.6875,
|
| 2660 |
+
"learning_rate": 3.663298116674322e-06,
|
| 2661 |
+
"loss": 0.3744,
|
| 2662 |
+
"step": 319
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 35.55555555555556,
|
| 2666 |
+
"grad_norm": 161650.578125,
|
| 2667 |
+
"learning_rate": 3.6747818098300414e-06,
|
| 2668 |
+
"loss": 0.4155,
|
| 2669 |
+
"step": 320
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 35.666666666666664,
|
| 2673 |
+
"grad_norm": 748048.6875,
|
| 2674 |
+
"learning_rate": 3.68626550298576e-06,
|
| 2675 |
+
"loss": 0.4249,
|
| 2676 |
+
"step": 321
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 35.77777777777778,
|
| 2680 |
+
"grad_norm": 1179157.0,
|
| 2681 |
+
"learning_rate": 3.697749196141479e-06,
|
| 2682 |
+
"loss": 0.3645,
|
| 2683 |
+
"step": 322
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 35.888888888888886,
|
| 2687 |
+
"grad_norm": 847522.25,
|
| 2688 |
+
"learning_rate": 3.709232889297198e-06,
|
| 2689 |
+
"loss": 0.4006,
|
| 2690 |
+
"step": 323
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 36.0,
|
| 2694 |
+
"grad_norm": 1134314.125,
|
| 2695 |
+
"learning_rate": 3.720716582452917e-06,
|
| 2696 |
+
"loss": 0.3825,
|
| 2697 |
+
"step": 324
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 36.0,
|
| 2701 |
+
"eval_accuracy": 0.8167202572347267,
|
| 2702 |
+
"eval_f1": 0.8037383177570093,
|
| 2703 |
+
"eval_loss": 0.3938184976577759,
|
| 2704 |
+
"eval_precision": 0.8223452440865626,
|
| 2705 |
+
"eval_recall": 0.785954785954786,
|
| 2706 |
+
"eval_runtime": 4.9706,
|
| 2707 |
+
"eval_samples_per_second": 875.943,
|
| 2708 |
+
"eval_steps_per_second": 0.604,
|
| 2709 |
+
"step": 324
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 36.111111111111114,
|
| 2713 |
+
"grad_norm": 219243.015625,
|
| 2714 |
+
"learning_rate": 3.732200275608636e-06,
|
| 2715 |
+
"loss": 0.3819,
|
| 2716 |
+
"step": 325
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"epoch": 36.22222222222222,
|
| 2720 |
+
"grad_norm": 178601.828125,
|
| 2721 |
+
"learning_rate": 3.7436839687643546e-06,
|
| 2722 |
+
"loss": 0.3902,
|
| 2723 |
+
"step": 326
|
| 2724 |
+
},
|
| 2725 |
+
{
|
| 2726 |
+
"epoch": 36.333333333333336,
|
| 2727 |
+
"grad_norm": 770971.6875,
|
| 2728 |
+
"learning_rate": 3.755167661920074e-06,
|
| 2729 |
+
"loss": 0.3764,
|
| 2730 |
+
"step": 327
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 36.44444444444444,
|
| 2734 |
+
"grad_norm": 508608.3125,
|
| 2735 |
+
"learning_rate": 3.766651355075792e-06,
|
| 2736 |
+
"loss": 0.3837,
|
| 2737 |
+
"step": 328
|
| 2738 |
+
},
|
| 2739 |
+
{
|
| 2740 |
+
"epoch": 36.55555555555556,
|
| 2741 |
+
"grad_norm": 697210.75,
|
| 2742 |
+
"learning_rate": 3.7781350482315114e-06,
|
| 2743 |
+
"loss": 0.3894,
|
| 2744 |
+
"step": 329
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 36.666666666666664,
|
| 2748 |
+
"grad_norm": 183407.328125,
|
| 2749 |
+
"learning_rate": 3.7896187413872302e-06,
|
| 2750 |
+
"loss": 0.3919,
|
| 2751 |
+
"step": 330
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 36.77777777777778,
|
| 2755 |
+
"grad_norm": 226463.875,
|
| 2756 |
+
"learning_rate": 3.801102434542949e-06,
|
| 2757 |
+
"loss": 0.3667,
|
| 2758 |
+
"step": 331
|
| 2759 |
+
},
|
| 2760 |
+
{
|
| 2761 |
+
"epoch": 36.888888888888886,
|
| 2762 |
+
"grad_norm": 320285.21875,
|
| 2763 |
+
"learning_rate": 3.812586127698668e-06,
|
| 2764 |
+
"loss": 0.3953,
|
| 2765 |
+
"step": 332
|
| 2766 |
+
},
|
| 2767 |
+
{
|
| 2768 |
+
"epoch": 37.0,
|
| 2769 |
+
"grad_norm": 125119.46875,
|
| 2770 |
+
"learning_rate": 3.824069820854387e-06,
|
| 2771 |
+
"loss": 0.3684,
|
| 2772 |
+
"step": 333
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 37.0,
|
| 2776 |
+
"eval_accuracy": 0.815112540192926,
|
| 2777 |
+
"eval_f1": 0.8002976928801786,
|
| 2778 |
+
"eval_loss": 0.3938528597354889,
|
| 2779 |
+
"eval_precision": 0.8263319672131147,
|
| 2780 |
+
"eval_recall": 0.7758537758537759,
|
| 2781 |
+
"eval_runtime": 5.031,
|
| 2782 |
+
"eval_samples_per_second": 865.438,
|
| 2783 |
+
"eval_steps_per_second": 0.596,
|
| 2784 |
+
"step": 333
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 37.111111111111114,
|
| 2788 |
+
"grad_norm": 502826.6875,
|
| 2789 |
+
"learning_rate": 3.835553514010106e-06,
|
| 2790 |
+
"loss": 0.3702,
|
| 2791 |
+
"step": 334
|
| 2792 |
+
},
|
| 2793 |
+
{
|
| 2794 |
+
"epoch": 37.22222222222222,
|
| 2795 |
+
"grad_norm": 216404.84375,
|
| 2796 |
+
"learning_rate": 3.847037207165825e-06,
|
| 2797 |
+
"loss": 0.3821,
|
| 2798 |
+
"step": 335
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 37.333333333333336,
|
| 2802 |
+
"grad_norm": 239700.234375,
|
| 2803 |
+
"learning_rate": 3.8585209003215434e-06,
|
| 2804 |
+
"loss": 0.3714,
|
| 2805 |
+
"step": 336
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 37.44444444444444,
|
| 2809 |
+
"grad_norm": 152418.90625,
|
| 2810 |
+
"learning_rate": 3.870004593477263e-06,
|
| 2811 |
+
"loss": 0.3965,
|
| 2812 |
+
"step": 337
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 37.55555555555556,
|
| 2816 |
+
"grad_norm": 279164.09375,
|
| 2817 |
+
"learning_rate": 3.881488286632981e-06,
|
| 2818 |
+
"loss": 0.4086,
|
| 2819 |
+
"step": 338
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 37.666666666666664,
|
| 2823 |
+
"grad_norm": 179164.71875,
|
| 2824 |
+
"learning_rate": 3.8929719797887e-06,
|
| 2825 |
+
"loss": 0.3946,
|
| 2826 |
+
"step": 339
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 37.77777777777778,
|
| 2830 |
+
"grad_norm": 264468.34375,
|
| 2831 |
+
"learning_rate": 3.904455672944419e-06,
|
| 2832 |
+
"loss": 0.374,
|
| 2833 |
+
"step": 340
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 37.888888888888886,
|
| 2837 |
+
"grad_norm": 131274.546875,
|
| 2838 |
+
"learning_rate": 3.915939366100138e-06,
|
| 2839 |
+
"loss": 0.3934,
|
| 2840 |
+
"step": 341
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 38.0,
|
| 2844 |
+
"grad_norm": 277462.40625,
|
| 2845 |
+
"learning_rate": 3.927423059255856e-06,
|
| 2846 |
+
"loss": 0.3433,
|
| 2847 |
+
"step": 342
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 38.0,
|
| 2851 |
+
"eval_accuracy": 0.821773082223243,
|
| 2852 |
+
"eval_f1": 0.8121065375302663,
|
| 2853 |
+
"eval_loss": 0.3931880295276642,
|
| 2854 |
+
"eval_precision": 0.817649926864944,
|
| 2855 |
+
"eval_recall": 0.8066378066378066,
|
| 2856 |
+
"eval_runtime": 5.087,
|
| 2857 |
+
"eval_samples_per_second": 855.913,
|
| 2858 |
+
"eval_steps_per_second": 0.59,
|
| 2859 |
+
"step": 342
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 38.111111111111114,
|
| 2863 |
+
"grad_norm": 168592.34375,
|
| 2864 |
+
"learning_rate": 3.9389067524115755e-06,
|
| 2865 |
+
"loss": 0.3766,
|
| 2866 |
+
"step": 343
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 38.22222222222222,
|
| 2870 |
+
"grad_norm": 163537.234375,
|
| 2871 |
+
"learning_rate": 3.950390445567295e-06,
|
| 2872 |
+
"loss": 0.386,
|
| 2873 |
+
"step": 344
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 38.333333333333336,
|
| 2877 |
+
"grad_norm": 583295.9375,
|
| 2878 |
+
"learning_rate": 3.961874138723013e-06,
|
| 2879 |
+
"loss": 0.3867,
|
| 2880 |
+
"step": 345
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 38.44444444444444,
|
| 2884 |
+
"grad_norm": 70029.2734375,
|
| 2885 |
+
"learning_rate": 3.973357831878732e-06,
|
| 2886 |
+
"loss": 0.3735,
|
| 2887 |
+
"step": 346
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 38.55555555555556,
|
| 2891 |
+
"grad_norm": 503096.8125,
|
| 2892 |
+
"learning_rate": 3.9848415250344515e-06,
|
| 2893 |
+
"loss": 0.3764,
|
| 2894 |
+
"step": 347
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 38.666666666666664,
|
| 2898 |
+
"grad_norm": 505206.1875,
|
| 2899 |
+
"learning_rate": 3.99632521819017e-06,
|
| 2900 |
+
"loss": 0.3751,
|
| 2901 |
+
"step": 348
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 38.77777777777778,
|
| 2905 |
+
"grad_norm": 191592.875,
|
| 2906 |
+
"learning_rate": 4.007808911345889e-06,
|
| 2907 |
+
"loss": 0.4098,
|
| 2908 |
+
"step": 349
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 38.888888888888886,
|
| 2912 |
+
"grad_norm": 491039.40625,
|
| 2913 |
+
"learning_rate": 4.0192926045016075e-06,
|
| 2914 |
+
"loss": 0.3724,
|
| 2915 |
+
"step": 350
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 39.0,
|
| 2919 |
+
"grad_norm": 815383.25,
|
| 2920 |
+
"learning_rate": 4.030776297657327e-06,
|
| 2921 |
+
"loss": 0.3968,
|
| 2922 |
+
"step": 351
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 39.0,
|
| 2926 |
+
"eval_accuracy": 0.823610473128158,
|
| 2927 |
+
"eval_f1": 0.8191238813000471,
|
| 2928 |
+
"eval_loss": 0.3962981402873993,
|
| 2929 |
+
"eval_precision": 0.8024919243193355,
|
| 2930 |
+
"eval_recall": 0.8364598364598365,
|
| 2931 |
+
"eval_runtime": 5.5822,
|
| 2932 |
+
"eval_samples_per_second": 779.985,
|
| 2933 |
+
"eval_steps_per_second": 0.537,
|
| 2934 |
+
"step": 351
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 39.111111111111114,
|
| 2938 |
+
"grad_norm": 616299.0,
|
| 2939 |
+
"learning_rate": 4.042259990813046e-06,
|
| 2940 |
+
"loss": 0.4,
|
| 2941 |
+
"step": 352
|
| 2942 |
+
},
|
| 2943 |
+
{
|
| 2944 |
+
"epoch": 39.22222222222222,
|
| 2945 |
+
"grad_norm": 390517.84375,
|
| 2946 |
+
"learning_rate": 4.053743683968765e-06,
|
| 2947 |
+
"loss": 0.3874,
|
| 2948 |
+
"step": 353
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 39.333333333333336,
|
| 2952 |
+
"grad_norm": 101405.28125,
|
| 2953 |
+
"learning_rate": 4.065227377124483e-06,
|
| 2954 |
+
"loss": 0.3804,
|
| 2955 |
+
"step": 354
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 39.44444444444444,
|
| 2959 |
+
"grad_norm": 803522.0,
|
| 2960 |
+
"learning_rate": 4.076711070280202e-06,
|
| 2961 |
+
"loss": 0.4033,
|
| 2962 |
+
"step": 355
|
| 2963 |
+
},
|
| 2964 |
+
{
|
| 2965 |
+
"epoch": 39.55555555555556,
|
| 2966 |
+
"grad_norm": 676551.6875,
|
| 2967 |
+
"learning_rate": 4.088194763435921e-06,
|
| 2968 |
+
"loss": 0.3897,
|
| 2969 |
+
"step": 356
|
| 2970 |
+
},
|
| 2971 |
+
{
|
| 2972 |
+
"epoch": 39.666666666666664,
|
| 2973 |
+
"grad_norm": 1587823.0,
|
| 2974 |
+
"learning_rate": 4.09967845659164e-06,
|
| 2975 |
+
"loss": 0.3898,
|
| 2976 |
+
"step": 357
|
| 2977 |
+
},
|
| 2978 |
+
{
|
| 2979 |
+
"epoch": 39.77777777777778,
|
| 2980 |
+
"grad_norm": 1047884.4375,
|
| 2981 |
+
"learning_rate": 4.111162149747359e-06,
|
| 2982 |
+
"loss": 0.3717,
|
| 2983 |
+
"step": 358
|
| 2984 |
+
},
|
| 2985 |
+
{
|
| 2986 |
+
"epoch": 39.888888888888886,
|
| 2987 |
+
"grad_norm": 336138.5,
|
| 2988 |
+
"learning_rate": 4.122645842903078e-06,
|
| 2989 |
+
"loss": 0.3901,
|
| 2990 |
+
"step": 359
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 40.0,
|
| 2994 |
+
"grad_norm": 352737.96875,
|
| 2995 |
+
"learning_rate": 4.134129536058796e-06,
|
| 2996 |
+
"loss": 0.3728,
|
| 2997 |
+
"step": 360
|
| 2998 |
+
},
|
| 2999 |
+
{
|
| 3000 |
+
"epoch": 40.0,
|
| 3001 |
+
"eval_accuracy": 0.823610473128158,
|
| 3002 |
+
"eval_f1": 0.8176638176638177,
|
| 3003 |
+
"eval_loss": 0.39358338713645935,
|
| 3004 |
+
"eval_precision": 0.8073136427566807,
|
| 3005 |
+
"eval_recall": 0.8282828282828283,
|
| 3006 |
+
"eval_runtime": 5.1339,
|
| 3007 |
+
"eval_samples_per_second": 848.091,
|
| 3008 |
+
"eval_steps_per_second": 0.584,
|
| 3009 |
+
"step": 360
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 40.111111111111114,
|
| 3013 |
+
"grad_norm": 635885.75,
|
| 3014 |
+
"learning_rate": 4.1456132292145155e-06,
|
| 3015 |
+
"loss": 0.3848,
|
| 3016 |
+
"step": 361
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 40.22222222222222,
|
| 3020 |
+
"grad_norm": 535277.4375,
|
| 3021 |
+
"learning_rate": 4.157096922370234e-06,
|
| 3022 |
+
"loss": 0.3836,
|
| 3023 |
+
"step": 362
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 40.333333333333336,
|
| 3027 |
+
"grad_norm": 503451.34375,
|
| 3028 |
+
"learning_rate": 4.168580615525953e-06,
|
| 3029 |
+
"loss": 0.3599,
|
| 3030 |
+
"step": 363
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 40.44444444444444,
|
| 3034 |
+
"grad_norm": 177537.828125,
|
| 3035 |
+
"learning_rate": 4.180064308681672e-06,
|
| 3036 |
+
"loss": 0.3727,
|
| 3037 |
+
"step": 364
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 40.55555555555556,
|
| 3041 |
+
"grad_norm": 400331.65625,
|
| 3042 |
+
"learning_rate": 4.1915480018373916e-06,
|
| 3043 |
+
"loss": 0.3874,
|
| 3044 |
+
"step": 365
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 40.666666666666664,
|
| 3048 |
+
"grad_norm": 1095307.75,
|
| 3049 |
+
"learning_rate": 4.203031694993109e-06,
|
| 3050 |
+
"loss": 0.3961,
|
| 3051 |
+
"step": 366
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 40.77777777777778,
|
| 3055 |
+
"grad_norm": 919089.5625,
|
| 3056 |
+
"learning_rate": 4.214515388148828e-06,
|
| 3057 |
+
"loss": 0.3891,
|
| 3058 |
+
"step": 367
|
| 3059 |
+
},
|
| 3060 |
+
{
|
| 3061 |
+
"epoch": 40.888888888888886,
|
| 3062 |
+
"grad_norm": 973522.375,
|
| 3063 |
+
"learning_rate": 4.2259990813045475e-06,
|
| 3064 |
+
"loss": 0.3881,
|
| 3065 |
+
"step": 368
|
| 3066 |
+
},
|
| 3067 |
+
{
|
| 3068 |
+
"epoch": 41.0,
|
| 3069 |
+
"grad_norm": 517725.1875,
|
| 3070 |
+
"learning_rate": 4.237482774460267e-06,
|
| 3071 |
+
"loss": 0.3914,
|
| 3072 |
+
"step": 369
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 41.0,
|
| 3076 |
+
"eval_accuracy": 0.8240698208543867,
|
| 3077 |
+
"eval_f1": 0.8180522565320665,
|
| 3078 |
+
"eval_loss": 0.3933127820491791,
|
| 3079 |
+
"eval_precision": 0.8080713280150165,
|
| 3080 |
+
"eval_recall": 0.8282828282828283,
|
| 3081 |
+
"eval_runtime": 5.193,
|
| 3082 |
+
"eval_samples_per_second": 838.429,
|
| 3083 |
+
"eval_steps_per_second": 0.578,
|
| 3084 |
+
"step": 369
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 41.111111111111114,
|
| 3088 |
+
"grad_norm": 411101.21875,
|
| 3089 |
+
"learning_rate": 4.248966467615986e-06,
|
| 3090 |
+
"loss": 0.3875,
|
| 3091 |
+
"step": 370
|
| 3092 |
+
},
|
| 3093 |
+
{
|
| 3094 |
+
"epoch": 41.22222222222222,
|
| 3095 |
+
"grad_norm": 953085.1875,
|
| 3096 |
+
"learning_rate": 4.260450160771704e-06,
|
| 3097 |
+
"loss": 0.3899,
|
| 3098 |
+
"step": 371
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 41.333333333333336,
|
| 3102 |
+
"grad_norm": 1096262.375,
|
| 3103 |
+
"learning_rate": 4.271933853927423e-06,
|
| 3104 |
+
"loss": 0.3916,
|
| 3105 |
+
"step": 372
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 41.44444444444444,
|
| 3109 |
+
"grad_norm": 834261.75,
|
| 3110 |
+
"learning_rate": 4.283417547083142e-06,
|
| 3111 |
+
"loss": 0.3889,
|
| 3112 |
+
"step": 373
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 41.55555555555556,
|
| 3116 |
+
"grad_norm": 399109.46875,
|
| 3117 |
+
"learning_rate": 4.294901240238861e-06,
|
| 3118 |
+
"loss": 0.3798,
|
| 3119 |
+
"step": 374
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 41.666666666666664,
|
| 3123 |
+
"grad_norm": 370690.375,
|
| 3124 |
+
"learning_rate": 4.3063849333945796e-06,
|
| 3125 |
+
"loss": 0.3775,
|
| 3126 |
+
"step": 375
|
| 3127 |
+
},
|
| 3128 |
+
{
|
| 3129 |
+
"epoch": 41.77777777777778,
|
| 3130 |
+
"grad_norm": 660407.1875,
|
| 3131 |
+
"learning_rate": 4.317868626550299e-06,
|
| 3132 |
+
"loss": 0.3832,
|
| 3133 |
+
"step": 376
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 41.888888888888886,
|
| 3137 |
+
"grad_norm": 1159956.5,
|
| 3138 |
+
"learning_rate": 4.329352319706018e-06,
|
| 3139 |
+
"loss": 0.3639,
|
| 3140 |
+
"step": 377
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 42.0,
|
| 3144 |
+
"grad_norm": 606383.1875,
|
| 3145 |
+
"learning_rate": 4.340836012861736e-06,
|
| 3146 |
+
"loss": 0.3738,
|
| 3147 |
+
"step": 378
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 42.0,
|
| 3151 |
+
"eval_accuracy": 0.8231511254019293,
|
| 3152 |
+
"eval_f1": 0.8135593220338984,
|
| 3153 |
+
"eval_loss": 0.39093542098999023,
|
| 3154 |
+
"eval_precision": 0.8191126279863481,
|
| 3155 |
+
"eval_recall": 0.8080808080808081,
|
| 3156 |
+
"eval_runtime": 5.2157,
|
| 3157 |
+
"eval_samples_per_second": 834.794,
|
| 3158 |
+
"eval_steps_per_second": 0.575,
|
| 3159 |
+
"step": 378
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 42.111111111111114,
|
| 3163 |
+
"grad_norm": 248742.28125,
|
| 3164 |
+
"learning_rate": 4.352319706017455e-06,
|
| 3165 |
+
"loss": 0.3937,
|
| 3166 |
+
"step": 379
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 42.22222222222222,
|
| 3170 |
+
"grad_norm": 585262.625,
|
| 3171 |
+
"learning_rate": 4.363803399173174e-06,
|
| 3172 |
+
"loss": 0.3623,
|
| 3173 |
+
"step": 380
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 42.333333333333336,
|
| 3177 |
+
"grad_norm": 200463.296875,
|
| 3178 |
+
"learning_rate": 4.375287092328893e-06,
|
| 3179 |
+
"loss": 0.3749,
|
| 3180 |
+
"step": 381
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 42.44444444444444,
|
| 3184 |
+
"grad_norm": 160448.734375,
|
| 3185 |
+
"learning_rate": 4.386770785484612e-06,
|
| 3186 |
+
"loss": 0.3893,
|
| 3187 |
+
"step": 382
|
| 3188 |
+
},
|
| 3189 |
+
{
|
| 3190 |
+
"epoch": 42.55555555555556,
|
| 3191 |
+
"grad_norm": 348410.375,
|
| 3192 |
+
"learning_rate": 4.398254478640331e-06,
|
| 3193 |
+
"loss": 0.3899,
|
| 3194 |
+
"step": 383
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 42.666666666666664,
|
| 3198 |
+
"grad_norm": 115621.484375,
|
| 3199 |
+
"learning_rate": 4.409738171796049e-06,
|
| 3200 |
+
"loss": 0.3933,
|
| 3201 |
+
"step": 384
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 42.77777777777778,
|
| 3205 |
+
"grad_norm": 431595.0,
|
| 3206 |
+
"learning_rate": 4.421221864951768e-06,
|
| 3207 |
+
"loss": 0.3821,
|
| 3208 |
+
"step": 385
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 42.888888888888886,
|
| 3212 |
+
"grad_norm": 101460.4375,
|
| 3213 |
+
"learning_rate": 4.432705558107488e-06,
|
| 3214 |
+
"loss": 0.3743,
|
| 3215 |
+
"step": 386
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 43.0,
|
| 3219 |
+
"grad_norm": 563180.0625,
|
| 3220 |
+
"learning_rate": 4.444189251263206e-06,
|
| 3221 |
+
"loss": 0.3924,
|
| 3222 |
+
"step": 387
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 43.0,
|
| 3226 |
+
"eval_accuracy": 0.8252181901699587,
|
| 3227 |
+
"eval_f1": 0.8221547090441692,
|
| 3228 |
+
"eval_loss": 0.39497968554496765,
|
| 3229 |
+
"eval_precision": 0.7995454545454546,
|
| 3230 |
+
"eval_recall": 0.8460798460798461,
|
| 3231 |
+
"eval_runtime": 5.2643,
|
| 3232 |
+
"eval_samples_per_second": 827.083,
|
| 3233 |
+
"eval_steps_per_second": 0.57,
|
| 3234 |
+
"step": 387
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 43.111111111111114,
|
| 3238 |
+
"grad_norm": 456643.25,
|
| 3239 |
+
"learning_rate": 4.455672944418925e-06,
|
| 3240 |
+
"loss": 0.371,
|
| 3241 |
+
"step": 388
|
| 3242 |
+
},
|
| 3243 |
+
{
|
| 3244 |
+
"epoch": 43.22222222222222,
|
| 3245 |
+
"grad_norm": 873773.8125,
|
| 3246 |
+
"learning_rate": 4.4671566375746444e-06,
|
| 3247 |
+
"loss": 0.3962,
|
| 3248 |
+
"step": 389
|
| 3249 |
+
},
|
| 3250 |
+
{
|
| 3251 |
+
"epoch": 43.333333333333336,
|
| 3252 |
+
"grad_norm": 112325.046875,
|
| 3253 |
+
"learning_rate": 4.478640330730363e-06,
|
| 3254 |
+
"loss": 0.3756,
|
| 3255 |
+
"step": 390
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 43.44444444444444,
|
| 3259 |
+
"grad_norm": 367788.90625,
|
| 3260 |
+
"learning_rate": 4.490124023886081e-06,
|
| 3261 |
+
"loss": 0.3788,
|
| 3262 |
+
"step": 391
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 43.55555555555556,
|
| 3266 |
+
"grad_norm": 390394.09375,
|
| 3267 |
+
"learning_rate": 4.5016077170418e-06,
|
| 3268 |
+
"loss": 0.3812,
|
| 3269 |
+
"step": 392
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 43.666666666666664,
|
| 3273 |
+
"grad_norm": 325314.1875,
|
| 3274 |
+
"learning_rate": 4.51309141019752e-06,
|
| 3275 |
+
"loss": 0.3731,
|
| 3276 |
+
"step": 393
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 43.77777777777778,
|
| 3280 |
+
"grad_norm": 156517.75,
|
| 3281 |
+
"learning_rate": 4.524575103353239e-06,
|
| 3282 |
+
"loss": 0.377,
|
| 3283 |
+
"step": 394
|
| 3284 |
+
},
|
| 3285 |
+
{
|
| 3286 |
+
"epoch": 43.888888888888886,
|
| 3287 |
+
"grad_norm": 443482.53125,
|
| 3288 |
+
"learning_rate": 4.536058796508958e-06,
|
| 3289 |
+
"loss": 0.3858,
|
| 3290 |
+
"step": 395
|
| 3291 |
+
},
|
| 3292 |
+
{
|
| 3293 |
+
"epoch": 44.0,
|
| 3294 |
+
"grad_norm": 853189.375,
|
| 3295 |
+
"learning_rate": 4.547542489664676e-06,
|
| 3296 |
+
"loss": 0.3817,
|
| 3297 |
+
"step": 396
|
| 3298 |
+
},
|
| 3299 |
+
{
|
| 3300 |
+
"epoch": 44.0,
|
| 3301 |
+
"eval_accuracy": 0.8270555810748736,
|
| 3302 |
+
"eval_f1": 0.8231148696264975,
|
| 3303 |
+
"eval_loss": 0.393361896276474,
|
| 3304 |
+
"eval_precision": 0.8044077134986226,
|
| 3305 |
+
"eval_recall": 0.8427128427128427,
|
| 3306 |
+
"eval_runtime": 5.3198,
|
| 3307 |
+
"eval_samples_per_second": 818.447,
|
| 3308 |
+
"eval_steps_per_second": 0.564,
|
| 3309 |
+
"step": 396
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 44.111111111111114,
|
| 3313 |
+
"grad_norm": 568736.6875,
|
| 3314 |
+
"learning_rate": 4.559026182820395e-06,
|
| 3315 |
+
"loss": 0.3612,
|
| 3316 |
+
"step": 397
|
| 3317 |
+
},
|
| 3318 |
+
{
|
| 3319 |
+
"epoch": 44.22222222222222,
|
| 3320 |
+
"grad_norm": 603129.125,
|
| 3321 |
+
"learning_rate": 4.570509875976114e-06,
|
| 3322 |
+
"loss": 0.3805,
|
| 3323 |
+
"step": 398
|
| 3324 |
+
},
|
| 3325 |
+
{
|
| 3326 |
+
"epoch": 44.333333333333336,
|
| 3327 |
+
"grad_norm": 218879.5625,
|
| 3328 |
+
"learning_rate": 4.581993569131833e-06,
|
| 3329 |
+
"loss": 0.3866,
|
| 3330 |
+
"step": 399
|
| 3331 |
+
},
|
| 3332 |
+
{
|
| 3333 |
+
"epoch": 44.44444444444444,
|
| 3334 |
+
"grad_norm": 751706.625,
|
| 3335 |
+
"learning_rate": 4.593477262287552e-06,
|
| 3336 |
+
"loss": 0.3912,
|
| 3337 |
+
"step": 400
|
| 3338 |
+
},
|
| 3339 |
+
{
|
| 3340 |
+
"epoch": 44.55555555555556,
|
| 3341 |
+
"grad_norm": 474342.125,
|
| 3342 |
+
"learning_rate": 4.604960955443271e-06,
|
| 3343 |
+
"loss": 0.3774,
|
| 3344 |
+
"step": 401
|
| 3345 |
+
},
|
| 3346 |
+
{
|
| 3347 |
+
"epoch": 44.666666666666664,
|
| 3348 |
+
"grad_norm": 237391.578125,
|
| 3349 |
+
"learning_rate": 4.616444648598989e-06,
|
| 3350 |
+
"loss": 0.3791,
|
| 3351 |
+
"step": 402
|
| 3352 |
+
},
|
| 3353 |
+
{
|
| 3354 |
+
"epoch": 44.77777777777778,
|
| 3355 |
+
"grad_norm": 143708.234375,
|
| 3356 |
+
"learning_rate": 4.6279283417547085e-06,
|
| 3357 |
+
"loss": 0.3844,
|
| 3358 |
+
"step": 403
|
| 3359 |
+
},
|
| 3360 |
+
{
|
| 3361 |
+
"epoch": 44.888888888888886,
|
| 3362 |
+
"grad_norm": 788803.3125,
|
| 3363 |
+
"learning_rate": 4.639412034910427e-06,
|
| 3364 |
+
"loss": 0.3885,
|
| 3365 |
+
"step": 404
|
| 3366 |
+
},
|
| 3367 |
+
{
|
| 3368 |
+
"epoch": 45.0,
|
| 3369 |
+
"grad_norm": 1540259.125,
|
| 3370 |
+
"learning_rate": 4.650895728066146e-06,
|
| 3371 |
+
"loss": 0.415,
|
| 3372 |
+
"step": 405
|
| 3373 |
+
},
|
| 3374 |
+
{
|
| 3375 |
+
"epoch": 45.0,
|
| 3376 |
+
"eval_accuracy": 0.8252181901699587,
|
| 3377 |
+
"eval_f1": 0.8227346843699045,
|
| 3378 |
+
"eval_loss": 0.3959347903728485,
|
| 3379 |
+
"eval_precision": 0.7976513098464318,
|
| 3380 |
+
"eval_recall": 0.8494468494468495,
|
| 3381 |
+
"eval_runtime": 5.3462,
|
| 3382 |
+
"eval_samples_per_second": 814.415,
|
| 3383 |
+
"eval_steps_per_second": 0.561,
|
| 3384 |
+
"step": 405
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 45.111111111111114,
|
| 3388 |
+
"grad_norm": 552211.75,
|
| 3389 |
+
"learning_rate": 4.662379421221865e-06,
|
| 3390 |
+
"loss": 0.3864,
|
| 3391 |
+
"step": 406
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 45.22222222222222,
|
| 3395 |
+
"grad_norm": 177492.828125,
|
| 3396 |
+
"learning_rate": 4.6738631143775845e-06,
|
| 3397 |
+
"loss": 0.3604,
|
| 3398 |
+
"step": 407
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 45.333333333333336,
|
| 3402 |
+
"grad_norm": 196209.4375,
|
| 3403 |
+
"learning_rate": 4.685346807533302e-06,
|
| 3404 |
+
"loss": 0.3826,
|
| 3405 |
+
"step": 408
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 45.44444444444444,
|
| 3409 |
+
"grad_norm": 857919.75,
|
| 3410 |
+
"learning_rate": 4.696830500689021e-06,
|
| 3411 |
+
"loss": 0.3903,
|
| 3412 |
+
"step": 409
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 45.55555555555556,
|
| 3416 |
+
"grad_norm": 1154738.25,
|
| 3417 |
+
"learning_rate": 4.7083141938447405e-06,
|
| 3418 |
+
"loss": 0.3909,
|
| 3419 |
+
"step": 410
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 45.666666666666664,
|
| 3423 |
+
"grad_norm": 1139759.75,
|
| 3424 |
+
"learning_rate": 4.71979788700046e-06,
|
| 3425 |
+
"loss": 0.3787,
|
| 3426 |
+
"step": 411
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 45.77777777777778,
|
| 3430 |
+
"grad_norm": 564532.8125,
|
| 3431 |
+
"learning_rate": 4.731281580156178e-06,
|
| 3432 |
+
"loss": 0.3599,
|
| 3433 |
+
"step": 412
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 45.888888888888886,
|
| 3437 |
+
"grad_norm": 143203.859375,
|
| 3438 |
+
"learning_rate": 4.742765273311897e-06,
|
| 3439 |
+
"loss": 0.3973,
|
| 3440 |
+
"step": 413
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 46.0,
|
| 3444 |
+
"grad_norm": 554491.5625,
|
| 3445 |
+
"learning_rate": 4.7542489664676165e-06,
|
| 3446 |
+
"loss": 0.3882,
|
| 3447 |
+
"step": 414
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 46.0,
|
| 3451 |
+
"eval_accuracy": 0.8286632981166743,
|
| 3452 |
+
"eval_f1": 0.8273947246645071,
|
| 3453 |
+
"eval_loss": 0.3958016335964203,
|
| 3454 |
+
"eval_precision": 0.797146678555506,
|
| 3455 |
+
"eval_recall": 0.86002886002886,
|
| 3456 |
+
"eval_runtime": 5.3654,
|
| 3457 |
+
"eval_samples_per_second": 811.493,
|
| 3458 |
+
"eval_steps_per_second": 0.559,
|
| 3459 |
+
"step": 414
|
| 3460 |
+
}
|
| 3461 |
+
],
|
| 3462 |
+
"logging_steps": 1,
|
| 3463 |
+
"max_steps": 450,
|
| 3464 |
+
"num_input_tokens_seen": 0,
|
| 3465 |
+
"num_train_epochs": 50,
|
| 3466 |
+
"save_steps": 500,
|
| 3467 |
+
"stateful_callbacks": {
|
| 3468 |
+
"TrainerControl": {
|
| 3469 |
+
"args": {
|
| 3470 |
+
"should_epoch_stop": false,
|
| 3471 |
+
"should_evaluate": false,
|
| 3472 |
+
"should_log": false,
|
| 3473 |
+
"should_save": true,
|
| 3474 |
+
"should_training_stop": false
|
| 3475 |
+
},
|
| 3476 |
+
"attributes": {}
|
| 3477 |
+
}
|
| 3478 |
+
},
|
| 3479 |
+
"total_flos": 3.1583883938955264e+16,
|
| 3480 |
+
"train_batch_size": 2048,
|
| 3481 |
+
"trial_name": null,
|
| 3482 |
+
"trial_params": null
|
| 3483 |
+
}
|
checkpoint-414/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:efb0fa4515340c52f02e80c6b7178ae61ffec5d2b093f61094126b6d6d022829
|
| 3 |
+
size 5368
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8563512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8c28a6bfc84261310d472d8c3e4c3f619e2c02b37dc643c7451efe5896b9d54
|
| 3 |
size 8563512
|
pr_curve.png
ADDED
|
roc_curve.png
ADDED
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5368
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:efb0fa4515340c52f02e80c6b7178ae61ffec5d2b093f61094126b6d6d022829
|
| 3 |
size 5368
|