Commit
·
076c454
1
Parent(s):
c29ff1a
Add evaluation results and output files to model card
Browse files- README.md +2 -2
- checkpoint-252/config.json +23 -0
- checkpoint-252/model.safetensors +3 -0
- checkpoint-252/optimizer.pt +3 -0
- checkpoint-252/preprocessor_config.json +22 -0
- checkpoint-252/rng_state.pth +3 -0
- checkpoint-252/scheduler.pt +3 -0
- checkpoint-252/trainer_state.json +2133 -0
- checkpoint-252/training_args.bin +3 -0
- model.safetensors +1 -1
- 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: 0.40
|
| 11 |
---
|
| 12 |
|
| 13 |
# Binary Classification Model
|
|
|
|
| 16 |
|
| 17 |
**Average Precision:** 0.67
|
| 18 |
**ROC AUC:** 0.77
|
| 19 |
+
**best threshold according to F1: 0.40
|
| 20 |
|
| 21 |
|
| 22 |
## Visualizations
|
checkpoint-252/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-252/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dbe878e912b9c485efa500f04a68eee9581442a6da571a687020bd06096011d6
|
| 3 |
+
size 8563512
|
checkpoint-252/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5895f266004da80d54c48eeb719a0d9cce3244a59370d4d37e6db6f16574b69b
|
| 3 |
+
size 17171514
|
checkpoint-252/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-252/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5773da530e74b4eb1c762c1159a5b4b42f867e42025fb298db7543766e0d2da1
|
| 3 |
+
size 14244
|
checkpoint-252/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7b6cfabc855b726101d507f53934d36b2a56d5794a0a749a646e51b3f7bf4e26
|
| 3 |
+
size 1000
|
checkpoint-252/trainer_state.json
ADDED
|
@@ -0,0 +1,2133 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8,
|
| 3 |
+
"best_model_checkpoint": "/tmp/logs/binary_classification_model_v3.1.5_spines/checkpoint-252",
|
| 4 |
+
"epoch": 28.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 252,
|
| 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": 456393.71875,
|
| 14 |
+
"learning_rate": 1.148369315571888e-10,
|
| 15 |
+
"loss": 0.397,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.2222222222222222,
|
| 20 |
+
"grad_norm": 507804.75,
|
| 21 |
+
"learning_rate": 2.296738631143776e-10,
|
| 22 |
+
"loss": 0.4283,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3333333333333333,
|
| 27 |
+
"grad_norm": 342309.34375,
|
| 28 |
+
"learning_rate": 3.445107946715664e-10,
|
| 29 |
+
"loss": 0.4275,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.4444444444444444,
|
| 34 |
+
"grad_norm": 458123.6875,
|
| 35 |
+
"learning_rate": 4.593477262287552e-10,
|
| 36 |
+
"loss": 0.4012,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.5555555555555556,
|
| 41 |
+
"grad_norm": 295167.625,
|
| 42 |
+
"learning_rate": 5.74184657785944e-10,
|
| 43 |
+
"loss": 0.3851,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.6666666666666666,
|
| 48 |
+
"grad_norm": 357707.125,
|
| 49 |
+
"learning_rate": 6.890215893431328e-10,
|
| 50 |
+
"loss": 0.397,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.7777777777777778,
|
| 55 |
+
"grad_norm": 307618.03125,
|
| 56 |
+
"learning_rate": 8.038585209003217e-10,
|
| 57 |
+
"loss": 0.3935,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.8888888888888888,
|
| 62 |
+
"grad_norm": 311204.9375,
|
| 63 |
+
"learning_rate": 9.186954524575104e-10,
|
| 64 |
+
"loss": 0.4009,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.0,
|
| 69 |
+
"grad_norm": 720027.375,
|
| 70 |
+
"learning_rate": 1.033532384014699e-09,
|
| 71 |
+
"loss": 0.404,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.0,
|
| 76 |
+
"eval_accuracy": 0.8066146072576941,
|
| 77 |
+
"eval_f1": 0.7878024193548387,
|
| 78 |
+
"eval_loss": 0.41061264276504517,
|
| 79 |
+
"eval_precision": 0.8274219163578613,
|
| 80 |
+
"eval_recall": 0.7518037518037518,
|
| 81 |
+
"eval_runtime": 2.9961,
|
| 82 |
+
"eval_samples_per_second": 1453.223,
|
| 83 |
+
"eval_steps_per_second": 1.001,
|
| 84 |
+
"step": 9
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.1111111111111112,
|
| 88 |
+
"grad_norm": 498179.78125,
|
| 89 |
+
"learning_rate": 1.148369315571888e-09,
|
| 90 |
+
"loss": 0.4095,
|
| 91 |
+
"step": 10
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.2222222222222223,
|
| 95 |
+
"grad_norm": 290873.78125,
|
| 96 |
+
"learning_rate": 1.2632062471290768e-09,
|
| 97 |
+
"loss": 0.4043,
|
| 98 |
+
"step": 11
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.3333333333333333,
|
| 102 |
+
"grad_norm": 565174.3125,
|
| 103 |
+
"learning_rate": 1.3780431786862656e-09,
|
| 104 |
+
"loss": 0.4174,
|
| 105 |
+
"step": 12
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.4444444444444444,
|
| 109 |
+
"grad_norm": 415411.8125,
|
| 110 |
+
"learning_rate": 1.4928801102434545e-09,
|
| 111 |
+
"loss": 0.3997,
|
| 112 |
+
"step": 13
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.5555555555555556,
|
| 116 |
+
"grad_norm": 369371.40625,
|
| 117 |
+
"learning_rate": 1.6077170418006433e-09,
|
| 118 |
+
"loss": 0.387,
|
| 119 |
+
"step": 14
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.6666666666666665,
|
| 123 |
+
"grad_norm": 250896.09375,
|
| 124 |
+
"learning_rate": 1.7225539733578319e-09,
|
| 125 |
+
"loss": 0.4132,
|
| 126 |
+
"step": 15
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 1.7777777777777777,
|
| 130 |
+
"grad_norm": 308713.34375,
|
| 131 |
+
"learning_rate": 1.8373909049150208e-09,
|
| 132 |
+
"loss": 0.4175,
|
| 133 |
+
"step": 16
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 1.8888888888888888,
|
| 137 |
+
"grad_norm": 381976.65625,
|
| 138 |
+
"learning_rate": 1.95222783647221e-09,
|
| 139 |
+
"loss": 0.3941,
|
| 140 |
+
"step": 17
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 2.0,
|
| 144 |
+
"grad_norm": 236604.359375,
|
| 145 |
+
"learning_rate": 2.067064768029398e-09,
|
| 146 |
+
"loss": 0.4156,
|
| 147 |
+
"step": 18
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 2.0,
|
| 151 |
+
"eval_accuracy": 0.8068442811208084,
|
| 152 |
+
"eval_f1": 0.7881078357268834,
|
| 153 |
+
"eval_loss": 0.41057315468788147,
|
| 154 |
+
"eval_precision": 0.8275132275132275,
|
| 155 |
+
"eval_recall": 0.7522847522847523,
|
| 156 |
+
"eval_runtime": 3.1648,
|
| 157 |
+
"eval_samples_per_second": 1375.763,
|
| 158 |
+
"eval_steps_per_second": 0.948,
|
| 159 |
+
"step": 18
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 2.111111111111111,
|
| 163 |
+
"grad_norm": 490775.53125,
|
| 164 |
+
"learning_rate": 2.1819016995865873e-09,
|
| 165 |
+
"loss": 0.4048,
|
| 166 |
+
"step": 19
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 2.2222222222222223,
|
| 170 |
+
"grad_norm": 272529.0625,
|
| 171 |
+
"learning_rate": 2.296738631143776e-09,
|
| 172 |
+
"loss": 0.419,
|
| 173 |
+
"step": 20
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 2.3333333333333335,
|
| 177 |
+
"grad_norm": 489697.5,
|
| 178 |
+
"learning_rate": 2.411575562700965e-09,
|
| 179 |
+
"loss": 0.4171,
|
| 180 |
+
"step": 21
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 2.4444444444444446,
|
| 184 |
+
"grad_norm": 312067.78125,
|
| 185 |
+
"learning_rate": 2.5264124942581536e-09,
|
| 186 |
+
"loss": 0.3963,
|
| 187 |
+
"step": 22
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 2.5555555555555554,
|
| 191 |
+
"grad_norm": 387958.1875,
|
| 192 |
+
"learning_rate": 2.6412494258153424e-09,
|
| 193 |
+
"loss": 0.4141,
|
| 194 |
+
"step": 23
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.6666666666666665,
|
| 198 |
+
"grad_norm": 421681.21875,
|
| 199 |
+
"learning_rate": 2.756086357372531e-09,
|
| 200 |
+
"loss": 0.3882,
|
| 201 |
+
"step": 24
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 2.7777777777777777,
|
| 205 |
+
"grad_norm": 357250.75,
|
| 206 |
+
"learning_rate": 2.87092328892972e-09,
|
| 207 |
+
"loss": 0.4031,
|
| 208 |
+
"step": 25
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 2.888888888888889,
|
| 212 |
+
"grad_norm": 423132.46875,
|
| 213 |
+
"learning_rate": 2.985760220486909e-09,
|
| 214 |
+
"loss": 0.414,
|
| 215 |
+
"step": 26
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 3.0,
|
| 219 |
+
"grad_norm": 283076.28125,
|
| 220 |
+
"learning_rate": 3.100597152044098e-09,
|
| 221 |
+
"loss": 0.3811,
|
| 222 |
+
"step": 27
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 3.0,
|
| 226 |
+
"eval_accuracy": 0.8063849333945797,
|
| 227 |
+
"eval_f1": 0.7878177699471433,
|
| 228 |
+
"eval_loss": 0.4105064272880554,
|
| 229 |
+
"eval_precision": 0.8262935586061246,
|
| 230 |
+
"eval_recall": 0.7527657527657527,
|
| 231 |
+
"eval_runtime": 3.2358,
|
| 232 |
+
"eval_samples_per_second": 1345.576,
|
| 233 |
+
"eval_steps_per_second": 0.927,
|
| 234 |
+
"step": 27
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 3.111111111111111,
|
| 238 |
+
"grad_norm": 409029.15625,
|
| 239 |
+
"learning_rate": 3.2154340836012866e-09,
|
| 240 |
+
"loss": 0.4026,
|
| 241 |
+
"step": 28
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 3.2222222222222223,
|
| 245 |
+
"grad_norm": 472187.0625,
|
| 246 |
+
"learning_rate": 3.3302710151584754e-09,
|
| 247 |
+
"loss": 0.3859,
|
| 248 |
+
"step": 29
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 3.3333333333333335,
|
| 252 |
+
"grad_norm": 229800.890625,
|
| 253 |
+
"learning_rate": 3.4451079467156637e-09,
|
| 254 |
+
"loss": 0.4057,
|
| 255 |
+
"step": 30
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 3.4444444444444446,
|
| 259 |
+
"grad_norm": 386687.78125,
|
| 260 |
+
"learning_rate": 3.5599448782728525e-09,
|
| 261 |
+
"loss": 0.397,
|
| 262 |
+
"step": 31
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 3.5555555555555554,
|
| 266 |
+
"grad_norm": 280445.84375,
|
| 267 |
+
"learning_rate": 3.6747818098300417e-09,
|
| 268 |
+
"loss": 0.4133,
|
| 269 |
+
"step": 32
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.6666666666666665,
|
| 273 |
+
"grad_norm": 490347.75,
|
| 274 |
+
"learning_rate": 3.7896187413872304e-09,
|
| 275 |
+
"loss": 0.4061,
|
| 276 |
+
"step": 33
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 3.7777777777777777,
|
| 280 |
+
"grad_norm": 484259.71875,
|
| 281 |
+
"learning_rate": 3.90445567294442e-09,
|
| 282 |
+
"loss": 0.3872,
|
| 283 |
+
"step": 34
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 3.888888888888889,
|
| 287 |
+
"grad_norm": 415715.5625,
|
| 288 |
+
"learning_rate": 4.019292604501608e-09,
|
| 289 |
+
"loss": 0.419,
|
| 290 |
+
"step": 35
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 4.0,
|
| 294 |
+
"grad_norm": 256996.140625,
|
| 295 |
+
"learning_rate": 4.134129536058796e-09,
|
| 296 |
+
"loss": 0.4032,
|
| 297 |
+
"step": 36
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 4.0,
|
| 301 |
+
"eval_accuracy": 0.8063849333945797,
|
| 302 |
+
"eval_f1": 0.7879245283018868,
|
| 303 |
+
"eval_loss": 0.4104115068912506,
|
| 304 |
+
"eval_precision": 0.8259493670886076,
|
| 305 |
+
"eval_recall": 0.7532467532467533,
|
| 306 |
+
"eval_runtime": 3.3192,
|
| 307 |
+
"eval_samples_per_second": 1311.755,
|
| 308 |
+
"eval_steps_per_second": 0.904,
|
| 309 |
+
"step": 36
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 4.111111111111111,
|
| 313 |
+
"grad_norm": 315028.59375,
|
| 314 |
+
"learning_rate": 4.2489664676159855e-09,
|
| 315 |
+
"loss": 0.4157,
|
| 316 |
+
"step": 37
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 4.222222222222222,
|
| 320 |
+
"grad_norm": 308553.3125,
|
| 321 |
+
"learning_rate": 4.363803399173175e-09,
|
| 322 |
+
"loss": 0.4248,
|
| 323 |
+
"step": 38
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 4.333333333333333,
|
| 327 |
+
"grad_norm": 313868.625,
|
| 328 |
+
"learning_rate": 4.478640330730363e-09,
|
| 329 |
+
"loss": 0.3864,
|
| 330 |
+
"step": 39
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 4.444444444444445,
|
| 334 |
+
"grad_norm": 337871.21875,
|
| 335 |
+
"learning_rate": 4.593477262287552e-09,
|
| 336 |
+
"loss": 0.4008,
|
| 337 |
+
"step": 40
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 4.555555555555555,
|
| 341 |
+
"grad_norm": 510693.46875,
|
| 342 |
+
"learning_rate": 4.7083141938447406e-09,
|
| 343 |
+
"loss": 0.4053,
|
| 344 |
+
"step": 41
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.666666666666667,
|
| 348 |
+
"grad_norm": 265638.21875,
|
| 349 |
+
"learning_rate": 4.82315112540193e-09,
|
| 350 |
+
"loss": 0.4019,
|
| 351 |
+
"step": 42
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 4.777777777777778,
|
| 355 |
+
"grad_norm": 454671.5,
|
| 356 |
+
"learning_rate": 4.937988056959119e-09,
|
| 357 |
+
"loss": 0.4004,
|
| 358 |
+
"step": 43
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 4.888888888888889,
|
| 362 |
+
"grad_norm": 384609.46875,
|
| 363 |
+
"learning_rate": 5.052824988516307e-09,
|
| 364 |
+
"loss": 0.3831,
|
| 365 |
+
"step": 44
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 5.0,
|
| 369 |
+
"grad_norm": 576690.625,
|
| 370 |
+
"learning_rate": 5.1676619200734964e-09,
|
| 371 |
+
"loss": 0.4544,
|
| 372 |
+
"step": 45
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 5.0,
|
| 376 |
+
"eval_accuracy": 0.8068442811208084,
|
| 377 |
+
"eval_f1": 0.7886403618999749,
|
| 378 |
+
"eval_loss": 0.410297155380249,
|
| 379 |
+
"eval_precision": 0.8257894736842105,
|
| 380 |
+
"eval_recall": 0.7546897546897547,
|
| 381 |
+
"eval_runtime": 3.6141,
|
| 382 |
+
"eval_samples_per_second": 1204.736,
|
| 383 |
+
"eval_steps_per_second": 0.83,
|
| 384 |
+
"step": 45
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 5.111111111111111,
|
| 388 |
+
"grad_norm": 268056.25,
|
| 389 |
+
"learning_rate": 5.282498851630685e-09,
|
| 390 |
+
"loss": 0.4078,
|
| 391 |
+
"step": 46
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 5.222222222222222,
|
| 395 |
+
"grad_norm": 349917.96875,
|
| 396 |
+
"learning_rate": 5.397335783187874e-09,
|
| 397 |
+
"loss": 0.3682,
|
| 398 |
+
"step": 47
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 5.333333333333333,
|
| 402 |
+
"grad_norm": 334901.15625,
|
| 403 |
+
"learning_rate": 5.512172714745062e-09,
|
| 404 |
+
"loss": 0.3995,
|
| 405 |
+
"step": 48
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 5.444444444444445,
|
| 409 |
+
"grad_norm": 469042.4375,
|
| 410 |
+
"learning_rate": 5.627009646302251e-09,
|
| 411 |
+
"loss": 0.4276,
|
| 412 |
+
"step": 49
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 5.555555555555555,
|
| 416 |
+
"grad_norm": 250734.765625,
|
| 417 |
+
"learning_rate": 5.74184657785944e-09,
|
| 418 |
+
"loss": 0.3969,
|
| 419 |
+
"step": 50
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 5.666666666666667,
|
| 423 |
+
"grad_norm": 375891.0,
|
| 424 |
+
"learning_rate": 5.856683509416628e-09,
|
| 425 |
+
"loss": 0.397,
|
| 426 |
+
"step": 51
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 5.777777777777778,
|
| 430 |
+
"grad_norm": 478341.78125,
|
| 431 |
+
"learning_rate": 5.971520440973818e-09,
|
| 432 |
+
"loss": 0.4009,
|
| 433 |
+
"step": 52
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 5.888888888888889,
|
| 437 |
+
"grad_norm": 276739.6875,
|
| 438 |
+
"learning_rate": 6.0863573725310065e-09,
|
| 439 |
+
"loss": 0.4079,
|
| 440 |
+
"step": 53
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 6.0,
|
| 444 |
+
"grad_norm": 506429.59375,
|
| 445 |
+
"learning_rate": 6.201194304088196e-09,
|
| 446 |
+
"loss": 0.4062,
|
| 447 |
+
"step": 54
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 6.0,
|
| 451 |
+
"eval_accuracy": 0.8068442811208084,
|
| 452 |
+
"eval_f1": 0.7887465460939462,
|
| 453 |
+
"eval_loss": 0.41015708446502686,
|
| 454 |
+
"eval_precision": 0.825446898002103,
|
| 455 |
+
"eval_recall": 0.7551707551707552,
|
| 456 |
+
"eval_runtime": 3.3917,
|
| 457 |
+
"eval_samples_per_second": 1283.727,
|
| 458 |
+
"eval_steps_per_second": 0.885,
|
| 459 |
+
"step": 54
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 6.111111111111111,
|
| 463 |
+
"grad_norm": 411659.09375,
|
| 464 |
+
"learning_rate": 6.316031235645384e-09,
|
| 465 |
+
"loss": 0.3962,
|
| 466 |
+
"step": 55
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 6.222222222222222,
|
| 470 |
+
"grad_norm": 371713.75,
|
| 471 |
+
"learning_rate": 6.430868167202573e-09,
|
| 472 |
+
"loss": 0.4098,
|
| 473 |
+
"step": 56
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 6.333333333333333,
|
| 477 |
+
"grad_norm": 326940.0,
|
| 478 |
+
"learning_rate": 6.545705098759762e-09,
|
| 479 |
+
"loss": 0.3976,
|
| 480 |
+
"step": 57
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 6.444444444444445,
|
| 484 |
+
"grad_norm": 485655.625,
|
| 485 |
+
"learning_rate": 6.660542030316951e-09,
|
| 486 |
+
"loss": 0.3864,
|
| 487 |
+
"step": 58
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 6.555555555555555,
|
| 491 |
+
"grad_norm": 226484.421875,
|
| 492 |
+
"learning_rate": 6.775378961874139e-09,
|
| 493 |
+
"loss": 0.4179,
|
| 494 |
+
"step": 59
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 6.666666666666667,
|
| 498 |
+
"grad_norm": 236230.21875,
|
| 499 |
+
"learning_rate": 6.8902158934313275e-09,
|
| 500 |
+
"loss": 0.3837,
|
| 501 |
+
"step": 60
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 6.777777777777778,
|
| 505 |
+
"grad_norm": 397832.65625,
|
| 506 |
+
"learning_rate": 7.005052824988517e-09,
|
| 507 |
+
"loss": 0.4243,
|
| 508 |
+
"step": 61
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 6.888888888888889,
|
| 512 |
+
"grad_norm": 426641.34375,
|
| 513 |
+
"learning_rate": 7.119889756545705e-09,
|
| 514 |
+
"loss": 0.4059,
|
| 515 |
+
"step": 62
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 7.0,
|
| 519 |
+
"grad_norm": 453001.25,
|
| 520 |
+
"learning_rate": 7.234726688102894e-09,
|
| 521 |
+
"loss": 0.3995,
|
| 522 |
+
"step": 63
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 7.0,
|
| 526 |
+
"eval_accuracy": 0.8086816720257235,
|
| 527 |
+
"eval_f1": 0.7911757332664828,
|
| 528 |
+
"eval_loss": 0.4099947512149811,
|
| 529 |
+
"eval_precision": 0.8261780104712042,
|
| 530 |
+
"eval_recall": 0.759018759018759,
|
| 531 |
+
"eval_runtime": 3.4242,
|
| 532 |
+
"eval_samples_per_second": 1271.531,
|
| 533 |
+
"eval_steps_per_second": 0.876,
|
| 534 |
+
"step": 63
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 7.111111111111111,
|
| 538 |
+
"grad_norm": 393810.0625,
|
| 539 |
+
"learning_rate": 7.349563619660083e-09,
|
| 540 |
+
"loss": 0.4114,
|
| 541 |
+
"step": 64
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 7.222222222222222,
|
| 545 |
+
"grad_norm": 564651.5,
|
| 546 |
+
"learning_rate": 7.464400551217272e-09,
|
| 547 |
+
"loss": 0.4045,
|
| 548 |
+
"step": 65
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 7.333333333333333,
|
| 552 |
+
"grad_norm": 334889.15625,
|
| 553 |
+
"learning_rate": 7.579237482774461e-09,
|
| 554 |
+
"loss": 0.3991,
|
| 555 |
+
"step": 66
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 7.444444444444445,
|
| 559 |
+
"grad_norm": 442494.65625,
|
| 560 |
+
"learning_rate": 7.69407441433165e-09,
|
| 561 |
+
"loss": 0.4092,
|
| 562 |
+
"step": 67
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 7.555555555555555,
|
| 566 |
+
"grad_norm": 306049.78125,
|
| 567 |
+
"learning_rate": 7.80891134588884e-09,
|
| 568 |
+
"loss": 0.4193,
|
| 569 |
+
"step": 68
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 7.666666666666667,
|
| 573 |
+
"grad_norm": 352232.0,
|
| 574 |
+
"learning_rate": 7.923748277446027e-09,
|
| 575 |
+
"loss": 0.4045,
|
| 576 |
+
"step": 69
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 7.777777777777778,
|
| 580 |
+
"grad_norm": 307979.75,
|
| 581 |
+
"learning_rate": 8.038585209003216e-09,
|
| 582 |
+
"loss": 0.4155,
|
| 583 |
+
"step": 70
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 7.888888888888889,
|
| 587 |
+
"grad_norm": 303762.40625,
|
| 588 |
+
"learning_rate": 8.153422140560405e-09,
|
| 589 |
+
"loss": 0.3984,
|
| 590 |
+
"step": 71
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 8.0,
|
| 594 |
+
"grad_norm": 312884.09375,
|
| 595 |
+
"learning_rate": 8.268259072117593e-09,
|
| 596 |
+
"loss": 0.3727,
|
| 597 |
+
"step": 72
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 8.0,
|
| 601 |
+
"eval_accuracy": 0.808451998162609,
|
| 602 |
+
"eval_f1": 0.7910821643286573,
|
| 603 |
+
"eval_loss": 0.40981706976890564,
|
| 604 |
+
"eval_precision": 0.825405122843701,
|
| 605 |
+
"eval_recall": 0.7594997594997595,
|
| 606 |
+
"eval_runtime": 3.4777,
|
| 607 |
+
"eval_samples_per_second": 1251.968,
|
| 608 |
+
"eval_steps_per_second": 0.863,
|
| 609 |
+
"step": 72
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 8.11111111111111,
|
| 613 |
+
"grad_norm": 440070.125,
|
| 614 |
+
"learning_rate": 8.383096003674782e-09,
|
| 615 |
+
"loss": 0.3895,
|
| 616 |
+
"step": 73
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 8.222222222222221,
|
| 620 |
+
"grad_norm": 386623.9375,
|
| 621 |
+
"learning_rate": 8.497932935231971e-09,
|
| 622 |
+
"loss": 0.4174,
|
| 623 |
+
"step": 74
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 8.333333333333334,
|
| 627 |
+
"grad_norm": 328681.78125,
|
| 628 |
+
"learning_rate": 8.61276986678916e-09,
|
| 629 |
+
"loss": 0.4159,
|
| 630 |
+
"step": 75
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 8.444444444444445,
|
| 634 |
+
"grad_norm": 461431.71875,
|
| 635 |
+
"learning_rate": 8.72760679834635e-09,
|
| 636 |
+
"loss": 0.3981,
|
| 637 |
+
"step": 76
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 8.555555555555555,
|
| 641 |
+
"grad_norm": 348082.03125,
|
| 642 |
+
"learning_rate": 8.842443729903539e-09,
|
| 643 |
+
"loss": 0.4273,
|
| 644 |
+
"step": 77
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 8.666666666666666,
|
| 648 |
+
"grad_norm": 242071.75,
|
| 649 |
+
"learning_rate": 8.957280661460726e-09,
|
| 650 |
+
"loss": 0.4121,
|
| 651 |
+
"step": 78
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 8.777777777777779,
|
| 655 |
+
"grad_norm": 213607.15625,
|
| 656 |
+
"learning_rate": 9.072117593017915e-09,
|
| 657 |
+
"loss": 0.3839,
|
| 658 |
+
"step": 79
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 8.88888888888889,
|
| 662 |
+
"grad_norm": 206478.359375,
|
| 663 |
+
"learning_rate": 9.186954524575104e-09,
|
| 664 |
+
"loss": 0.3753,
|
| 665 |
+
"step": 80
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 9.0,
|
| 669 |
+
"grad_norm": 194609.3125,
|
| 670 |
+
"learning_rate": 9.301791456132294e-09,
|
| 671 |
+
"loss": 0.3988,
|
| 672 |
+
"step": 81
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 9.0,
|
| 676 |
+
"eval_accuracy": 0.808451998162609,
|
| 677 |
+
"eval_f1": 0.7912912912912913,
|
| 678 |
+
"eval_loss": 0.409627765417099,
|
| 679 |
+
"eval_precision": 0.8247261345852895,
|
| 680 |
+
"eval_recall": 0.7604617604617605,
|
| 681 |
+
"eval_runtime": 3.5371,
|
| 682 |
+
"eval_samples_per_second": 1230.936,
|
| 683 |
+
"eval_steps_per_second": 0.848,
|
| 684 |
+
"step": 81
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 9.11111111111111,
|
| 688 |
+
"grad_norm": 561483.3125,
|
| 689 |
+
"learning_rate": 9.416628387689481e-09,
|
| 690 |
+
"loss": 0.4005,
|
| 691 |
+
"step": 82
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 9.222222222222221,
|
| 695 |
+
"grad_norm": 225070.078125,
|
| 696 |
+
"learning_rate": 9.53146531924667e-09,
|
| 697 |
+
"loss": 0.3984,
|
| 698 |
+
"step": 83
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 9.333333333333334,
|
| 702 |
+
"grad_norm": 327914.65625,
|
| 703 |
+
"learning_rate": 9.64630225080386e-09,
|
| 704 |
+
"loss": 0.4252,
|
| 705 |
+
"step": 84
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 9.444444444444445,
|
| 709 |
+
"grad_norm": 334806.53125,
|
| 710 |
+
"learning_rate": 9.761139182361047e-09,
|
| 711 |
+
"loss": 0.3961,
|
| 712 |
+
"step": 85
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 9.555555555555555,
|
| 716 |
+
"grad_norm": 422189.59375,
|
| 717 |
+
"learning_rate": 9.875976113918238e-09,
|
| 718 |
+
"loss": 0.3975,
|
| 719 |
+
"step": 86
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 9.666666666666666,
|
| 723 |
+
"grad_norm": 425266.5625,
|
| 724 |
+
"learning_rate": 9.990813045475427e-09,
|
| 725 |
+
"loss": 0.4352,
|
| 726 |
+
"step": 87
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 9.777777777777779,
|
| 730 |
+
"grad_norm": 312599.34375,
|
| 731 |
+
"learning_rate": 1.0105649977032615e-08,
|
| 732 |
+
"loss": 0.4063,
|
| 733 |
+
"step": 88
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 9.88888888888889,
|
| 737 |
+
"grad_norm": 318109.625,
|
| 738 |
+
"learning_rate": 1.0220486908589804e-08,
|
| 739 |
+
"loss": 0.4041,
|
| 740 |
+
"step": 89
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 10.0,
|
| 744 |
+
"grad_norm": 132810.046875,
|
| 745 |
+
"learning_rate": 1.0335323840146993e-08,
|
| 746 |
+
"loss": 0.3777,
|
| 747 |
+
"step": 90
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 10.0,
|
| 751 |
+
"eval_accuracy": 0.8091410197519522,
|
| 752 |
+
"eval_f1": 0.7924056957282039,
|
| 753 |
+
"eval_loss": 0.40942761301994324,
|
| 754 |
+
"eval_precision": 0.8243243243243243,
|
| 755 |
+
"eval_recall": 0.7628667628667629,
|
| 756 |
+
"eval_runtime": 3.8583,
|
| 757 |
+
"eval_samples_per_second": 1128.481,
|
| 758 |
+
"eval_steps_per_second": 0.778,
|
| 759 |
+
"step": 90
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 10.11111111111111,
|
| 763 |
+
"grad_norm": 166400.25,
|
| 764 |
+
"learning_rate": 1.045016077170418e-08,
|
| 765 |
+
"loss": 0.401,
|
| 766 |
+
"step": 91
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 10.222222222222221,
|
| 770 |
+
"grad_norm": 305970.1875,
|
| 771 |
+
"learning_rate": 1.056499770326137e-08,
|
| 772 |
+
"loss": 0.3924,
|
| 773 |
+
"step": 92
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 10.333333333333334,
|
| 777 |
+
"grad_norm": 409570.40625,
|
| 778 |
+
"learning_rate": 1.0679834634818559e-08,
|
| 779 |
+
"loss": 0.4085,
|
| 780 |
+
"step": 93
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 10.444444444444445,
|
| 784 |
+
"grad_norm": 282469.40625,
|
| 785 |
+
"learning_rate": 1.0794671566375748e-08,
|
| 786 |
+
"loss": 0.3878,
|
| 787 |
+
"step": 94
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 10.555555555555555,
|
| 791 |
+
"grad_norm": 223962.640625,
|
| 792 |
+
"learning_rate": 1.0909508497932935e-08,
|
| 793 |
+
"loss": 0.3986,
|
| 794 |
+
"step": 95
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 10.666666666666666,
|
| 798 |
+
"grad_norm": 408710.5625,
|
| 799 |
+
"learning_rate": 1.1024345429490125e-08,
|
| 800 |
+
"loss": 0.4301,
|
| 801 |
+
"step": 96
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 10.777777777777779,
|
| 805 |
+
"grad_norm": 156632.640625,
|
| 806 |
+
"learning_rate": 1.1139182361047314e-08,
|
| 807 |
+
"loss": 0.3915,
|
| 808 |
+
"step": 97
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 10.88888888888889,
|
| 812 |
+
"grad_norm": 376189.65625,
|
| 813 |
+
"learning_rate": 1.1254019292604501e-08,
|
| 814 |
+
"loss": 0.4031,
|
| 815 |
+
"step": 98
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 11.0,
|
| 819 |
+
"grad_norm": 119827.9453125,
|
| 820 |
+
"learning_rate": 1.136885622416169e-08,
|
| 821 |
+
"loss": 0.4026,
|
| 822 |
+
"step": 99
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 11.0,
|
| 826 |
+
"eval_accuracy": 0.8098300413412953,
|
| 827 |
+
"eval_f1": 0.7933100349475787,
|
| 828 |
+
"eval_loss": 0.4092212915420532,
|
| 829 |
+
"eval_precision": 0.8245978204462896,
|
| 830 |
+
"eval_recall": 0.7643097643097643,
|
| 831 |
+
"eval_runtime": 3.6143,
|
| 832 |
+
"eval_samples_per_second": 1204.659,
|
| 833 |
+
"eval_steps_per_second": 0.83,
|
| 834 |
+
"step": 99
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 11.11111111111111,
|
| 838 |
+
"grad_norm": 289736.1875,
|
| 839 |
+
"learning_rate": 1.148369315571888e-08,
|
| 840 |
+
"loss": 0.4042,
|
| 841 |
+
"step": 100
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 11.222222222222221,
|
| 845 |
+
"grad_norm": 194127.875,
|
| 846 |
+
"learning_rate": 1.1598530087276069e-08,
|
| 847 |
+
"loss": 0.4082,
|
| 848 |
+
"step": 101
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 11.333333333333334,
|
| 852 |
+
"grad_norm": 383586.8125,
|
| 853 |
+
"learning_rate": 1.1713367018833256e-08,
|
| 854 |
+
"loss": 0.4196,
|
| 855 |
+
"step": 102
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 11.444444444444445,
|
| 859 |
+
"grad_norm": 309528.40625,
|
| 860 |
+
"learning_rate": 1.1828203950390446e-08,
|
| 861 |
+
"loss": 0.4009,
|
| 862 |
+
"step": 103
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 11.555555555555555,
|
| 866 |
+
"grad_norm": 275662.09375,
|
| 867 |
+
"learning_rate": 1.1943040881947636e-08,
|
| 868 |
+
"loss": 0.3864,
|
| 869 |
+
"step": 104
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 11.666666666666666,
|
| 873 |
+
"grad_norm": 240105.71875,
|
| 874 |
+
"learning_rate": 1.2057877813504826e-08,
|
| 875 |
+
"loss": 0.4044,
|
| 876 |
+
"step": 105
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 11.777777777777779,
|
| 880 |
+
"grad_norm": 449321.34375,
|
| 881 |
+
"learning_rate": 1.2172714745062013e-08,
|
| 882 |
+
"loss": 0.3972,
|
| 883 |
+
"step": 106
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 11.88888888888889,
|
| 887 |
+
"grad_norm": 420052.9375,
|
| 888 |
+
"learning_rate": 1.2287551676619202e-08,
|
| 889 |
+
"loss": 0.4077,
|
| 890 |
+
"step": 107
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 12.0,
|
| 894 |
+
"grad_norm": 200356.359375,
|
| 895 |
+
"learning_rate": 1.2402388608176391e-08,
|
| 896 |
+
"loss": 0.4241,
|
| 897 |
+
"step": 108
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 12.0,
|
| 901 |
+
"eval_accuracy": 0.8107487367937529,
|
| 902 |
+
"eval_f1": 0.7948207171314741,
|
| 903 |
+
"eval_loss": 0.4090159237384796,
|
| 904 |
+
"eval_precision": 0.8239545689210118,
|
| 905 |
+
"eval_recall": 0.7676767676767676,
|
| 906 |
+
"eval_runtime": 3.6992,
|
| 907 |
+
"eval_samples_per_second": 1177.009,
|
| 908 |
+
"eval_steps_per_second": 0.811,
|
| 909 |
+
"step": 108
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 12.11111111111111,
|
| 913 |
+
"grad_norm": 208306.484375,
|
| 914 |
+
"learning_rate": 1.251722553973358e-08,
|
| 915 |
+
"loss": 0.3955,
|
| 916 |
+
"step": 109
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 12.222222222222221,
|
| 920 |
+
"grad_norm": 58767.4453125,
|
| 921 |
+
"learning_rate": 1.2632062471290768e-08,
|
| 922 |
+
"loss": 0.4038,
|
| 923 |
+
"step": 110
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 12.333333333333334,
|
| 927 |
+
"grad_norm": 176059.9375,
|
| 928 |
+
"learning_rate": 1.2746899402847957e-08,
|
| 929 |
+
"loss": 0.4034,
|
| 930 |
+
"step": 111
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 12.444444444444445,
|
| 934 |
+
"grad_norm": 290994.40625,
|
| 935 |
+
"learning_rate": 1.2861736334405147e-08,
|
| 936 |
+
"loss": 0.4001,
|
| 937 |
+
"step": 112
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 12.555555555555555,
|
| 941 |
+
"grad_norm": 211688.546875,
|
| 942 |
+
"learning_rate": 1.2976573265962334e-08,
|
| 943 |
+
"loss": 0.3999,
|
| 944 |
+
"step": 113
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 12.666666666666666,
|
| 948 |
+
"grad_norm": 219369.5,
|
| 949 |
+
"learning_rate": 1.3091410197519523e-08,
|
| 950 |
+
"loss": 0.3917,
|
| 951 |
+
"step": 114
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 12.777777777777779,
|
| 955 |
+
"grad_norm": 401543.40625,
|
| 956 |
+
"learning_rate": 1.3206247129076712e-08,
|
| 957 |
+
"loss": 0.4041,
|
| 958 |
+
"step": 115
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 12.88888888888889,
|
| 962 |
+
"grad_norm": 362102.40625,
|
| 963 |
+
"learning_rate": 1.3321084060633902e-08,
|
| 964 |
+
"loss": 0.4154,
|
| 965 |
+
"step": 116
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 13.0,
|
| 969 |
+
"grad_norm": 477879.5,
|
| 970 |
+
"learning_rate": 1.3435920992191089e-08,
|
| 971 |
+
"loss": 0.3709,
|
| 972 |
+
"step": 117
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 13.0,
|
| 976 |
+
"eval_accuracy": 0.8107487367937529,
|
| 977 |
+
"eval_f1": 0.7950248756218905,
|
| 978 |
+
"eval_loss": 0.4088153839111328,
|
| 979 |
+
"eval_precision": 0.8232869654817104,
|
| 980 |
+
"eval_recall": 0.7686387686387687,
|
| 981 |
+
"eval_runtime": 3.7239,
|
| 982 |
+
"eval_samples_per_second": 1169.192,
|
| 983 |
+
"eval_steps_per_second": 0.806,
|
| 984 |
+
"step": 117
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 13.11111111111111,
|
| 988 |
+
"grad_norm": 235991.40625,
|
| 989 |
+
"learning_rate": 1.3550757923748278e-08,
|
| 990 |
+
"loss": 0.414,
|
| 991 |
+
"step": 118
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 13.222222222222221,
|
| 995 |
+
"grad_norm": 95655.484375,
|
| 996 |
+
"learning_rate": 1.3665594855305467e-08,
|
| 997 |
+
"loss": 0.3927,
|
| 998 |
+
"step": 119
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 13.333333333333334,
|
| 1002 |
+
"grad_norm": 180109.375,
|
| 1003 |
+
"learning_rate": 1.3780431786862655e-08,
|
| 1004 |
+
"loss": 0.4221,
|
| 1005 |
+
"step": 120
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 13.444444444444445,
|
| 1009 |
+
"grad_norm": 114767.703125,
|
| 1010 |
+
"learning_rate": 1.3895268718419844e-08,
|
| 1011 |
+
"loss": 0.3897,
|
| 1012 |
+
"step": 121
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 13.555555555555555,
|
| 1016 |
+
"grad_norm": 261420.796875,
|
| 1017 |
+
"learning_rate": 1.4010105649977033e-08,
|
| 1018 |
+
"loss": 0.4083,
|
| 1019 |
+
"step": 122
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 13.666666666666666,
|
| 1023 |
+
"grad_norm": 267185.46875,
|
| 1024 |
+
"learning_rate": 1.4124942581534222e-08,
|
| 1025 |
+
"loss": 0.4178,
|
| 1026 |
+
"step": 123
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 13.777777777777779,
|
| 1030 |
+
"grad_norm": 433391.46875,
|
| 1031 |
+
"learning_rate": 1.423977951309141e-08,
|
| 1032 |
+
"loss": 0.3946,
|
| 1033 |
+
"step": 124
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 13.88888888888889,
|
| 1037 |
+
"grad_norm": 240314.125,
|
| 1038 |
+
"learning_rate": 1.43546164446486e-08,
|
| 1039 |
+
"loss": 0.3969,
|
| 1040 |
+
"step": 125
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 14.0,
|
| 1044 |
+
"grad_norm": 494791.96875,
|
| 1045 |
+
"learning_rate": 1.4469453376205788e-08,
|
| 1046 |
+
"loss": 0.422,
|
| 1047 |
+
"step": 126
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 14.0,
|
| 1051 |
+
"eval_accuracy": 0.8112080845199816,
|
| 1052 |
+
"eval_f1": 0.7958271236959762,
|
| 1053 |
+
"eval_loss": 0.40861907601356506,
|
| 1054 |
+
"eval_precision": 0.8228043143297381,
|
| 1055 |
+
"eval_recall": 0.7705627705627706,
|
| 1056 |
+
"eval_runtime": 3.7622,
|
| 1057 |
+
"eval_samples_per_second": 1157.3,
|
| 1058 |
+
"eval_steps_per_second": 0.797,
|
| 1059 |
+
"step": 126
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 14.11111111111111,
|
| 1063 |
+
"grad_norm": 230925.59375,
|
| 1064 |
+
"learning_rate": 1.458429030776298e-08,
|
| 1065 |
+
"loss": 0.4406,
|
| 1066 |
+
"step": 127
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 14.222222222222221,
|
| 1070 |
+
"grad_norm": 248435.78125,
|
| 1071 |
+
"learning_rate": 1.4699127239320167e-08,
|
| 1072 |
+
"loss": 0.4152,
|
| 1073 |
+
"step": 128
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 14.333333333333334,
|
| 1077 |
+
"grad_norm": 276681.65625,
|
| 1078 |
+
"learning_rate": 1.4813964170877356e-08,
|
| 1079 |
+
"loss": 0.4062,
|
| 1080 |
+
"step": 129
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 14.444444444444445,
|
| 1084 |
+
"grad_norm": 119796.078125,
|
| 1085 |
+
"learning_rate": 1.4928801102434543e-08,
|
| 1086 |
+
"loss": 0.4135,
|
| 1087 |
+
"step": 130
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 14.555555555555555,
|
| 1091 |
+
"grad_norm": 329935.125,
|
| 1092 |
+
"learning_rate": 1.5043638033991733e-08,
|
| 1093 |
+
"loss": 0.3937,
|
| 1094 |
+
"step": 131
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 14.666666666666666,
|
| 1098 |
+
"grad_norm": 262459.5625,
|
| 1099 |
+
"learning_rate": 1.5158474965548922e-08,
|
| 1100 |
+
"loss": 0.3941,
|
| 1101 |
+
"step": 132
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 14.777777777777779,
|
| 1105 |
+
"grad_norm": 105897.59375,
|
| 1106 |
+
"learning_rate": 1.527331189710611e-08,
|
| 1107 |
+
"loss": 0.3796,
|
| 1108 |
+
"step": 133
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 14.88888888888889,
|
| 1112 |
+
"grad_norm": 225864.171875,
|
| 1113 |
+
"learning_rate": 1.53881488286633e-08,
|
| 1114 |
+
"loss": 0.3803,
|
| 1115 |
+
"step": 134
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 15.0,
|
| 1119 |
+
"grad_norm": 341410.21875,
|
| 1120 |
+
"learning_rate": 1.550298576022049e-08,
|
| 1121 |
+
"loss": 0.4158,
|
| 1122 |
+
"step": 135
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 15.0,
|
| 1126 |
+
"eval_accuracy": 0.8109784106568673,
|
| 1127 |
+
"eval_f1": 0.7957309506080913,
|
| 1128 |
+
"eval_loss": 0.40841957926750183,
|
| 1129 |
+
"eval_precision": 0.8220512820512821,
|
| 1130 |
+
"eval_recall": 0.7710437710437711,
|
| 1131 |
+
"eval_runtime": 3.8028,
|
| 1132 |
+
"eval_samples_per_second": 1144.947,
|
| 1133 |
+
"eval_steps_per_second": 0.789,
|
| 1134 |
+
"step": 135
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 15.11111111111111,
|
| 1138 |
+
"grad_norm": 132152.578125,
|
| 1139 |
+
"learning_rate": 1.561782269177768e-08,
|
| 1140 |
+
"loss": 0.3878,
|
| 1141 |
+
"step": 136
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 15.222222222222221,
|
| 1145 |
+
"grad_norm": 72833.984375,
|
| 1146 |
+
"learning_rate": 1.5732659623334864e-08,
|
| 1147 |
+
"loss": 0.4254,
|
| 1148 |
+
"step": 137
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 15.333333333333334,
|
| 1152 |
+
"grad_norm": 510368.78125,
|
| 1153 |
+
"learning_rate": 1.5847496554892054e-08,
|
| 1154 |
+
"loss": 0.415,
|
| 1155 |
+
"step": 138
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 15.444444444444445,
|
| 1159 |
+
"grad_norm": 288746.75,
|
| 1160 |
+
"learning_rate": 1.5962333486449243e-08,
|
| 1161 |
+
"loss": 0.3957,
|
| 1162 |
+
"step": 139
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 15.555555555555555,
|
| 1166 |
+
"grad_norm": 155515.0625,
|
| 1167 |
+
"learning_rate": 1.6077170418006432e-08,
|
| 1168 |
+
"loss": 0.4078,
|
| 1169 |
+
"step": 140
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 15.666666666666666,
|
| 1173 |
+
"grad_norm": 293017.9375,
|
| 1174 |
+
"learning_rate": 1.619200734956362e-08,
|
| 1175 |
+
"loss": 0.4113,
|
| 1176 |
+
"step": 141
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 15.777777777777779,
|
| 1180 |
+
"grad_norm": 145871.765625,
|
| 1181 |
+
"learning_rate": 1.630684428112081e-08,
|
| 1182 |
+
"loss": 0.4132,
|
| 1183 |
+
"step": 142
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 15.88888888888889,
|
| 1187 |
+
"grad_norm": 153809.59375,
|
| 1188 |
+
"learning_rate": 1.6421681212678e-08,
|
| 1189 |
+
"loss": 0.3727,
|
| 1190 |
+
"step": 143
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 16.0,
|
| 1194 |
+
"grad_norm": 203934.796875,
|
| 1195 |
+
"learning_rate": 1.6536518144235185e-08,
|
| 1196 |
+
"loss": 0.4209,
|
| 1197 |
+
"step": 144
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 16.0,
|
| 1201 |
+
"eval_accuracy": 0.8102893890675241,
|
| 1202 |
+
"eval_f1": 0.7953419226957383,
|
| 1203 |
+
"eval_loss": 0.40823647379875183,
|
| 1204 |
+
"eval_precision": 0.8201328564128768,
|
| 1205 |
+
"eval_recall": 0.772005772005772,
|
| 1206 |
+
"eval_runtime": 4.1397,
|
| 1207 |
+
"eval_samples_per_second": 1051.768,
|
| 1208 |
+
"eval_steps_per_second": 0.725,
|
| 1209 |
+
"step": 144
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 16.11111111111111,
|
| 1213 |
+
"grad_norm": 201536.859375,
|
| 1214 |
+
"learning_rate": 1.6651355075792374e-08,
|
| 1215 |
+
"loss": 0.3897,
|
| 1216 |
+
"step": 145
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 16.22222222222222,
|
| 1220 |
+
"grad_norm": 157509.171875,
|
| 1221 |
+
"learning_rate": 1.6766192007349564e-08,
|
| 1222 |
+
"loss": 0.4073,
|
| 1223 |
+
"step": 146
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 16.333333333333332,
|
| 1227 |
+
"grad_norm": 91640.9921875,
|
| 1228 |
+
"learning_rate": 1.6881028938906753e-08,
|
| 1229 |
+
"loss": 0.3989,
|
| 1230 |
+
"step": 147
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 16.444444444444443,
|
| 1234 |
+
"grad_norm": 162518.9375,
|
| 1235 |
+
"learning_rate": 1.6995865870463942e-08,
|
| 1236 |
+
"loss": 0.4077,
|
| 1237 |
+
"step": 148
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 16.555555555555557,
|
| 1241 |
+
"grad_norm": 312546.71875,
|
| 1242 |
+
"learning_rate": 1.711070280202113e-08,
|
| 1243 |
+
"loss": 0.4016,
|
| 1244 |
+
"step": 149
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 16.666666666666668,
|
| 1248 |
+
"grad_norm": 133448.515625,
|
| 1249 |
+
"learning_rate": 1.722553973357832e-08,
|
| 1250 |
+
"loss": 0.3874,
|
| 1251 |
+
"step": 150
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 16.77777777777778,
|
| 1255 |
+
"grad_norm": 228277.046875,
|
| 1256 |
+
"learning_rate": 1.734037666513551e-08,
|
| 1257 |
+
"loss": 0.4197,
|
| 1258 |
+
"step": 151
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 16.88888888888889,
|
| 1262 |
+
"grad_norm": 410443.84375,
|
| 1263 |
+
"learning_rate": 1.74552135966927e-08,
|
| 1264 |
+
"loss": 0.4011,
|
| 1265 |
+
"step": 152
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 17.0,
|
| 1269 |
+
"grad_norm": 142888.328125,
|
| 1270 |
+
"learning_rate": 1.7570050528249888e-08,
|
| 1271 |
+
"loss": 0.3922,
|
| 1272 |
+
"step": 153
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 17.0,
|
| 1276 |
+
"eval_accuracy": 0.811437758383096,
|
| 1277 |
+
"eval_f1": 0.7969329705664111,
|
| 1278 |
+
"eval_loss": 0.40807411074638367,
|
| 1279 |
+
"eval_precision": 0.820264765784114,
|
| 1280 |
+
"eval_recall": 0.7748917748917749,
|
| 1281 |
+
"eval_runtime": 3.897,
|
| 1282 |
+
"eval_samples_per_second": 1117.267,
|
| 1283 |
+
"eval_steps_per_second": 0.77,
|
| 1284 |
+
"step": 153
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 17.11111111111111,
|
| 1288 |
+
"grad_norm": 114717.609375,
|
| 1289 |
+
"learning_rate": 1.7684887459807077e-08,
|
| 1290 |
+
"loss": 0.3872,
|
| 1291 |
+
"step": 154
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 17.22222222222222,
|
| 1295 |
+
"grad_norm": 176862.40625,
|
| 1296 |
+
"learning_rate": 1.7799724391364266e-08,
|
| 1297 |
+
"loss": 0.4056,
|
| 1298 |
+
"step": 155
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 17.333333333333332,
|
| 1302 |
+
"grad_norm": 275508.40625,
|
| 1303 |
+
"learning_rate": 1.7914561322921452e-08,
|
| 1304 |
+
"loss": 0.3976,
|
| 1305 |
+
"step": 156
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 17.444444444444443,
|
| 1309 |
+
"grad_norm": 155124.390625,
|
| 1310 |
+
"learning_rate": 1.802939825447864e-08,
|
| 1311 |
+
"loss": 0.4152,
|
| 1312 |
+
"step": 157
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 17.555555555555557,
|
| 1316 |
+
"grad_norm": 158142.9375,
|
| 1317 |
+
"learning_rate": 1.814423518603583e-08,
|
| 1318 |
+
"loss": 0.4046,
|
| 1319 |
+
"step": 158
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 17.666666666666668,
|
| 1323 |
+
"grad_norm": 175609.71875,
|
| 1324 |
+
"learning_rate": 1.825907211759302e-08,
|
| 1325 |
+
"loss": 0.4145,
|
| 1326 |
+
"step": 159
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 17.77777777777778,
|
| 1330 |
+
"grad_norm": 80656.234375,
|
| 1331 |
+
"learning_rate": 1.837390904915021e-08,
|
| 1332 |
+
"loss": 0.407,
|
| 1333 |
+
"step": 160
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 17.88888888888889,
|
| 1337 |
+
"grad_norm": 82560.1171875,
|
| 1338 |
+
"learning_rate": 1.8488745980707398e-08,
|
| 1339 |
+
"loss": 0.389,
|
| 1340 |
+
"step": 161
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 18.0,
|
| 1344 |
+
"grad_norm": 374558.875,
|
| 1345 |
+
"learning_rate": 1.8603582912264587e-08,
|
| 1346 |
+
"loss": 0.3957,
|
| 1347 |
+
"step": 162
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 18.0,
|
| 1351 |
+
"eval_accuracy": 0.811437758383096,
|
| 1352 |
+
"eval_f1": 0.797133679268594,
|
| 1353 |
+
"eval_loss": 0.4079172611236572,
|
| 1354 |
+
"eval_precision": 0.8196138211382114,
|
| 1355 |
+
"eval_recall": 0.7758537758537759,
|
| 1356 |
+
"eval_runtime": 3.9317,
|
| 1357 |
+
"eval_samples_per_second": 1107.398,
|
| 1358 |
+
"eval_steps_per_second": 0.763,
|
| 1359 |
+
"step": 162
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 18.11111111111111,
|
| 1363 |
+
"grad_norm": 191892.828125,
|
| 1364 |
+
"learning_rate": 1.8718419843821773e-08,
|
| 1365 |
+
"loss": 0.4017,
|
| 1366 |
+
"step": 163
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 18.22222222222222,
|
| 1370 |
+
"grad_norm": 118981.9375,
|
| 1371 |
+
"learning_rate": 1.8833256775378962e-08,
|
| 1372 |
+
"loss": 0.4055,
|
| 1373 |
+
"step": 164
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 18.333333333333332,
|
| 1377 |
+
"grad_norm": 253416.578125,
|
| 1378 |
+
"learning_rate": 1.894809370693615e-08,
|
| 1379 |
+
"loss": 0.3939,
|
| 1380 |
+
"step": 165
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 18.444444444444443,
|
| 1384 |
+
"grad_norm": 238575.578125,
|
| 1385 |
+
"learning_rate": 1.906293063849334e-08,
|
| 1386 |
+
"loss": 0.3826,
|
| 1387 |
+
"step": 166
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 18.555555555555557,
|
| 1391 |
+
"grad_norm": 124402.765625,
|
| 1392 |
+
"learning_rate": 1.917776757005053e-08,
|
| 1393 |
+
"loss": 0.4123,
|
| 1394 |
+
"step": 167
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 18.666666666666668,
|
| 1398 |
+
"grad_norm": 191731.640625,
|
| 1399 |
+
"learning_rate": 1.929260450160772e-08,
|
| 1400 |
+
"loss": 0.4203,
|
| 1401 |
+
"step": 168
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 18.77777777777778,
|
| 1405 |
+
"grad_norm": 133824.984375,
|
| 1406 |
+
"learning_rate": 1.9407441433164908e-08,
|
| 1407 |
+
"loss": 0.4145,
|
| 1408 |
+
"step": 169
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 18.88888888888889,
|
| 1412 |
+
"grad_norm": 109809.515625,
|
| 1413 |
+
"learning_rate": 1.9522278364722094e-08,
|
| 1414 |
+
"loss": 0.3911,
|
| 1415 |
+
"step": 170
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 19.0,
|
| 1419 |
+
"grad_norm": 285880.6875,
|
| 1420 |
+
"learning_rate": 1.9637115296279283e-08,
|
| 1421 |
+
"loss": 0.3963,
|
| 1422 |
+
"step": 171
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 19.0,
|
| 1426 |
+
"eval_accuracy": 0.8112080845199816,
|
| 1427 |
+
"eval_f1": 0.7972372964972866,
|
| 1428 |
+
"eval_loss": 0.4077652096748352,
|
| 1429 |
+
"eval_precision": 0.8182278481012658,
|
| 1430 |
+
"eval_recall": 0.7772967772967773,
|
| 1431 |
+
"eval_runtime": 3.9878,
|
| 1432 |
+
"eval_samples_per_second": 1091.836,
|
| 1433 |
+
"eval_steps_per_second": 0.752,
|
| 1434 |
+
"step": 171
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 19.11111111111111,
|
| 1438 |
+
"grad_norm": 261054.515625,
|
| 1439 |
+
"learning_rate": 1.9751952227836476e-08,
|
| 1440 |
+
"loss": 0.3847,
|
| 1441 |
+
"step": 172
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 19.22222222222222,
|
| 1445 |
+
"grad_norm": 158484.296875,
|
| 1446 |
+
"learning_rate": 1.986678915939366e-08,
|
| 1447 |
+
"loss": 0.401,
|
| 1448 |
+
"step": 173
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 19.333333333333332,
|
| 1452 |
+
"grad_norm": 85237.9765625,
|
| 1453 |
+
"learning_rate": 1.9981626090950854e-08,
|
| 1454 |
+
"loss": 0.4312,
|
| 1455 |
+
"step": 174
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 19.444444444444443,
|
| 1459 |
+
"grad_norm": 280998.5,
|
| 1460 |
+
"learning_rate": 2.009646302250804e-08,
|
| 1461 |
+
"loss": 0.4048,
|
| 1462 |
+
"step": 175
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 19.555555555555557,
|
| 1466 |
+
"grad_norm": 219749.75,
|
| 1467 |
+
"learning_rate": 2.021129995406523e-08,
|
| 1468 |
+
"loss": 0.4207,
|
| 1469 |
+
"step": 176
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 19.666666666666668,
|
| 1473 |
+
"grad_norm": 238937.890625,
|
| 1474 |
+
"learning_rate": 2.0326136885622415e-08,
|
| 1475 |
+
"loss": 0.3747,
|
| 1476 |
+
"step": 177
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 19.77777777777778,
|
| 1480 |
+
"grad_norm": 245600.234375,
|
| 1481 |
+
"learning_rate": 2.0440973817179607e-08,
|
| 1482 |
+
"loss": 0.4028,
|
| 1483 |
+
"step": 178
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 19.88888888888889,
|
| 1487 |
+
"grad_norm": 144331.484375,
|
| 1488 |
+
"learning_rate": 2.0555810748736793e-08,
|
| 1489 |
+
"loss": 0.4067,
|
| 1490 |
+
"step": 179
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 20.0,
|
| 1494 |
+
"grad_norm": 245784.15625,
|
| 1495 |
+
"learning_rate": 2.0670647680293986e-08,
|
| 1496 |
+
"loss": 0.3778,
|
| 1497 |
+
"step": 180
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 20.0,
|
| 1501 |
+
"eval_accuracy": 0.8112080845199816,
|
| 1502 |
+
"eval_f1": 0.7974371611631346,
|
| 1503 |
+
"eval_loss": 0.4076385796070099,
|
| 1504 |
+
"eval_precision": 0.8175846387064174,
|
| 1505 |
+
"eval_recall": 0.7782587782587782,
|
| 1506 |
+
"eval_runtime": 4.0155,
|
| 1507 |
+
"eval_samples_per_second": 1084.293,
|
| 1508 |
+
"eval_steps_per_second": 0.747,
|
| 1509 |
+
"step": 180
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 20.11111111111111,
|
| 1513 |
+
"grad_norm": 110038.1328125,
|
| 1514 |
+
"learning_rate": 2.078548461185117e-08,
|
| 1515 |
+
"loss": 0.4099,
|
| 1516 |
+
"step": 181
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 20.22222222222222,
|
| 1520 |
+
"grad_norm": 113520.578125,
|
| 1521 |
+
"learning_rate": 2.090032154340836e-08,
|
| 1522 |
+
"loss": 0.4048,
|
| 1523 |
+
"step": 182
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 20.333333333333332,
|
| 1527 |
+
"grad_norm": 132782.125,
|
| 1528 |
+
"learning_rate": 2.101515847496555e-08,
|
| 1529 |
+
"loss": 0.3995,
|
| 1530 |
+
"step": 183
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 20.444444444444443,
|
| 1534 |
+
"grad_norm": 98626.984375,
|
| 1535 |
+
"learning_rate": 2.112999540652274e-08,
|
| 1536 |
+
"loss": 0.3966,
|
| 1537 |
+
"step": 184
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 20.555555555555557,
|
| 1541 |
+
"grad_norm": 296527.78125,
|
| 1542 |
+
"learning_rate": 2.124483233807993e-08,
|
| 1543 |
+
"loss": 0.4215,
|
| 1544 |
+
"step": 185
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 20.666666666666668,
|
| 1548 |
+
"grad_norm": 109782.6171875,
|
| 1549 |
+
"learning_rate": 2.1359669269637117e-08,
|
| 1550 |
+
"loss": 0.4089,
|
| 1551 |
+
"step": 186
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 20.77777777777778,
|
| 1555 |
+
"grad_norm": 131173.890625,
|
| 1556 |
+
"learning_rate": 2.1474506201194307e-08,
|
| 1557 |
+
"loss": 0.4165,
|
| 1558 |
+
"step": 187
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 20.88888888888889,
|
| 1562 |
+
"grad_norm": 186916.5625,
|
| 1563 |
+
"learning_rate": 2.1589343132751496e-08,
|
| 1564 |
+
"loss": 0.3851,
|
| 1565 |
+
"step": 188
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 21.0,
|
| 1569 |
+
"grad_norm": 198655.515625,
|
| 1570 |
+
"learning_rate": 2.1704180064308685e-08,
|
| 1571 |
+
"loss": 0.4147,
|
| 1572 |
+
"step": 189
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 21.0,
|
| 1576 |
+
"eval_accuracy": 0.8112080845199816,
|
| 1577 |
+
"eval_f1": 0.7975369458128079,
|
| 1578 |
+
"eval_loss": 0.40751826763153076,
|
| 1579 |
+
"eval_precision": 0.8172640080767289,
|
| 1580 |
+
"eval_recall": 0.7787397787397787,
|
| 1581 |
+
"eval_runtime": 4.0653,
|
| 1582 |
+
"eval_samples_per_second": 1071.029,
|
| 1583 |
+
"eval_steps_per_second": 0.738,
|
| 1584 |
+
"step": 189
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 21.11111111111111,
|
| 1588 |
+
"grad_norm": 119997.5234375,
|
| 1589 |
+
"learning_rate": 2.181901699586587e-08,
|
| 1590 |
+
"loss": 0.3979,
|
| 1591 |
+
"step": 190
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 21.22222222222222,
|
| 1595 |
+
"grad_norm": 55222.81640625,
|
| 1596 |
+
"learning_rate": 2.1933853927423063e-08,
|
| 1597 |
+
"loss": 0.4095,
|
| 1598 |
+
"step": 191
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 21.333333333333332,
|
| 1602 |
+
"grad_norm": 120158.1484375,
|
| 1603 |
+
"learning_rate": 2.204869085898025e-08,
|
| 1604 |
+
"loss": 0.4031,
|
| 1605 |
+
"step": 192
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 21.444444444444443,
|
| 1609 |
+
"grad_norm": 228876.90625,
|
| 1610 |
+
"learning_rate": 2.216352779053744e-08,
|
| 1611 |
+
"loss": 0.416,
|
| 1612 |
+
"step": 193
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 21.555555555555557,
|
| 1616 |
+
"grad_norm": 139059.484375,
|
| 1617 |
+
"learning_rate": 2.2278364722094628e-08,
|
| 1618 |
+
"loss": 0.3891,
|
| 1619 |
+
"step": 194
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 21.666666666666668,
|
| 1623 |
+
"grad_norm": 253780.5,
|
| 1624 |
+
"learning_rate": 2.2393201653651817e-08,
|
| 1625 |
+
"loss": 0.4171,
|
| 1626 |
+
"step": 195
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 21.77777777777778,
|
| 1630 |
+
"grad_norm": 143734.296875,
|
| 1631 |
+
"learning_rate": 2.2508038585209003e-08,
|
| 1632 |
+
"loss": 0.4165,
|
| 1633 |
+
"step": 196
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 21.88888888888889,
|
| 1637 |
+
"grad_norm": 114648.3046875,
|
| 1638 |
+
"learning_rate": 2.2622875516766195e-08,
|
| 1639 |
+
"loss": 0.3873,
|
| 1640 |
+
"step": 197
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 22.0,
|
| 1644 |
+
"grad_norm": 176826.3125,
|
| 1645 |
+
"learning_rate": 2.273771244832338e-08,
|
| 1646 |
+
"loss": 0.3808,
|
| 1647 |
+
"step": 198
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 22.0,
|
| 1651 |
+
"eval_accuracy": 0.8112080845199816,
|
| 1652 |
+
"eval_f1": 0.7978357107722578,
|
| 1653 |
+
"eval_loss": 0.40740326046943665,
|
| 1654 |
+
"eval_precision": 0.8163059889280322,
|
| 1655 |
+
"eval_recall": 0.7801827801827802,
|
| 1656 |
+
"eval_runtime": 4.4548,
|
| 1657 |
+
"eval_samples_per_second": 977.363,
|
| 1658 |
+
"eval_steps_per_second": 0.673,
|
| 1659 |
+
"step": 198
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 22.11111111111111,
|
| 1663 |
+
"grad_norm": 195394.6875,
|
| 1664 |
+
"learning_rate": 2.2852549379880573e-08,
|
| 1665 |
+
"loss": 0.4074,
|
| 1666 |
+
"step": 199
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 22.22222222222222,
|
| 1670 |
+
"grad_norm": 125695.6328125,
|
| 1671 |
+
"learning_rate": 2.296738631143776e-08,
|
| 1672 |
+
"loss": 0.4051,
|
| 1673 |
+
"step": 200
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 22.333333333333332,
|
| 1677 |
+
"grad_norm": 167733.265625,
|
| 1678 |
+
"learning_rate": 2.308222324299495e-08,
|
| 1679 |
+
"loss": 0.4093,
|
| 1680 |
+
"step": 201
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 22.444444444444443,
|
| 1684 |
+
"grad_norm": 100189.2890625,
|
| 1685 |
+
"learning_rate": 2.3197060174552138e-08,
|
| 1686 |
+
"loss": 0.4165,
|
| 1687 |
+
"step": 202
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 22.555555555555557,
|
| 1691 |
+
"grad_norm": 173090.84375,
|
| 1692 |
+
"learning_rate": 2.3311897106109327e-08,
|
| 1693 |
+
"loss": 0.3996,
|
| 1694 |
+
"step": 203
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 22.666666666666668,
|
| 1698 |
+
"grad_norm": 83075.390625,
|
| 1699 |
+
"learning_rate": 2.3426734037666513e-08,
|
| 1700 |
+
"loss": 0.3813,
|
| 1701 |
+
"step": 204
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 22.77777777777778,
|
| 1705 |
+
"grad_norm": 200632.8125,
|
| 1706 |
+
"learning_rate": 2.3541570969223705e-08,
|
| 1707 |
+
"loss": 0.3937,
|
| 1708 |
+
"step": 205
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 22.88888888888889,
|
| 1712 |
+
"grad_norm": 146405.703125,
|
| 1713 |
+
"learning_rate": 2.365640790078089e-08,
|
| 1714 |
+
"loss": 0.4075,
|
| 1715 |
+
"step": 206
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 23.0,
|
| 1719 |
+
"grad_norm": 145782.46875,
|
| 1720 |
+
"learning_rate": 2.377124483233808e-08,
|
| 1721 |
+
"loss": 0.3625,
|
| 1722 |
+
"step": 207
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 23.0,
|
| 1726 |
+
"eval_accuracy": 0.811437758383096,
|
| 1727 |
+
"eval_f1": 0.7983296487349546,
|
| 1728 |
+
"eval_loss": 0.407308429479599,
|
| 1729 |
+
"eval_precision": 0.8157630522088354,
|
| 1730 |
+
"eval_recall": 0.7816257816257817,
|
| 1731 |
+
"eval_runtime": 4.1931,
|
| 1732 |
+
"eval_samples_per_second": 1038.381,
|
| 1733 |
+
"eval_steps_per_second": 0.715,
|
| 1734 |
+
"step": 207
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 23.11111111111111,
|
| 1738 |
+
"grad_norm": 80425.2578125,
|
| 1739 |
+
"learning_rate": 2.3886081763895273e-08,
|
| 1740 |
+
"loss": 0.3928,
|
| 1741 |
+
"step": 208
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 23.22222222222222,
|
| 1745 |
+
"grad_norm": 138923.78125,
|
| 1746 |
+
"learning_rate": 2.400091869545246e-08,
|
| 1747 |
+
"loss": 0.401,
|
| 1748 |
+
"step": 209
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 23.333333333333332,
|
| 1752 |
+
"grad_norm": 137138.015625,
|
| 1753 |
+
"learning_rate": 2.411575562700965e-08,
|
| 1754 |
+
"loss": 0.4047,
|
| 1755 |
+
"step": 210
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 23.444444444444443,
|
| 1759 |
+
"grad_norm": 120477.125,
|
| 1760 |
+
"learning_rate": 2.4230592558566837e-08,
|
| 1761 |
+
"loss": 0.4139,
|
| 1762 |
+
"step": 211
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 23.555555555555557,
|
| 1766 |
+
"grad_norm": 87321.953125,
|
| 1767 |
+
"learning_rate": 2.4345429490124026e-08,
|
| 1768 |
+
"loss": 0.4152,
|
| 1769 |
+
"step": 212
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 23.666666666666668,
|
| 1773 |
+
"grad_norm": 201660.828125,
|
| 1774 |
+
"learning_rate": 2.4460266421681215e-08,
|
| 1775 |
+
"loss": 0.4057,
|
| 1776 |
+
"step": 213
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 23.77777777777778,
|
| 1780 |
+
"grad_norm": 76530.015625,
|
| 1781 |
+
"learning_rate": 2.4575103353238405e-08,
|
| 1782 |
+
"loss": 0.402,
|
| 1783 |
+
"step": 214
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 23.88888888888889,
|
| 1787 |
+
"grad_norm": 136255.46875,
|
| 1788 |
+
"learning_rate": 2.468994028479559e-08,
|
| 1789 |
+
"loss": 0.3999,
|
| 1790 |
+
"step": 215
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 24.0,
|
| 1794 |
+
"grad_norm": 292040.84375,
|
| 1795 |
+
"learning_rate": 2.4804777216352783e-08,
|
| 1796 |
+
"loss": 0.3615,
|
| 1797 |
+
"step": 216
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 24.0,
|
| 1801 |
+
"eval_accuracy": 0.8116674322462104,
|
| 1802 |
+
"eval_f1": 0.7987236131566029,
|
| 1803 |
+
"eval_loss": 0.4072152376174927,
|
| 1804 |
+
"eval_precision": 0.8155388471177945,
|
| 1805 |
+
"eval_recall": 0.7825877825877826,
|
| 1806 |
+
"eval_runtime": 4.25,
|
| 1807 |
+
"eval_samples_per_second": 1024.465,
|
| 1808 |
+
"eval_steps_per_second": 0.706,
|
| 1809 |
+
"step": 216
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 24.11111111111111,
|
| 1813 |
+
"grad_norm": 99101.1328125,
|
| 1814 |
+
"learning_rate": 2.491961414790997e-08,
|
| 1815 |
+
"loss": 0.4274,
|
| 1816 |
+
"step": 217
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 24.22222222222222,
|
| 1820 |
+
"grad_norm": 64315.54296875,
|
| 1821 |
+
"learning_rate": 2.503445107946716e-08,
|
| 1822 |
+
"loss": 0.3936,
|
| 1823 |
+
"step": 218
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 24.333333333333332,
|
| 1827 |
+
"grad_norm": 133974.671875,
|
| 1828 |
+
"learning_rate": 2.5149288011024347e-08,
|
| 1829 |
+
"loss": 0.3976,
|
| 1830 |
+
"step": 219
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 24.444444444444443,
|
| 1834 |
+
"grad_norm": 181377.90625,
|
| 1835 |
+
"learning_rate": 2.5264124942581536e-08,
|
| 1836 |
+
"loss": 0.4097,
|
| 1837 |
+
"step": 220
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 24.555555555555557,
|
| 1841 |
+
"grad_norm": 117327.421875,
|
| 1842 |
+
"learning_rate": 2.5378961874138722e-08,
|
| 1843 |
+
"loss": 0.4014,
|
| 1844 |
+
"step": 221
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 24.666666666666668,
|
| 1848 |
+
"grad_norm": 131969.5625,
|
| 1849 |
+
"learning_rate": 2.5493798805695915e-08,
|
| 1850 |
+
"loss": 0.4142,
|
| 1851 |
+
"step": 222
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 24.77777777777778,
|
| 1855 |
+
"grad_norm": 73783.3046875,
|
| 1856 |
+
"learning_rate": 2.56086357372531e-08,
|
| 1857 |
+
"loss": 0.3713,
|
| 1858 |
+
"step": 223
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 24.88888888888889,
|
| 1862 |
+
"grad_norm": 146125.0625,
|
| 1863 |
+
"learning_rate": 2.5723472668810293e-08,
|
| 1864 |
+
"loss": 0.3838,
|
| 1865 |
+
"step": 224
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 25.0,
|
| 1869 |
+
"grad_norm": 397248.34375,
|
| 1870 |
+
"learning_rate": 2.583830960036748e-08,
|
| 1871 |
+
"loss": 0.4203,
|
| 1872 |
+
"step": 225
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 25.0,
|
| 1876 |
+
"eval_accuracy": 0.8116674322462104,
|
| 1877 |
+
"eval_f1": 0.7989210397253556,
|
| 1878 |
+
"eval_loss": 0.4071270823478699,
|
| 1879 |
+
"eval_precision": 0.8149074537268635,
|
| 1880 |
+
"eval_recall": 0.7835497835497836,
|
| 1881 |
+
"eval_runtime": 4.2738,
|
| 1882 |
+
"eval_samples_per_second": 1018.776,
|
| 1883 |
+
"eval_steps_per_second": 0.702,
|
| 1884 |
+
"step": 225
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 25.11111111111111,
|
| 1888 |
+
"grad_norm": 76815.546875,
|
| 1889 |
+
"learning_rate": 2.5953146531924668e-08,
|
| 1890 |
+
"loss": 0.4168,
|
| 1891 |
+
"step": 226
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 25.22222222222222,
|
| 1895 |
+
"grad_norm": 115688.4140625,
|
| 1896 |
+
"learning_rate": 2.6067983463481857e-08,
|
| 1897 |
+
"loss": 0.3871,
|
| 1898 |
+
"step": 227
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 25.333333333333332,
|
| 1902 |
+
"grad_norm": 173911.359375,
|
| 1903 |
+
"learning_rate": 2.6182820395039046e-08,
|
| 1904 |
+
"loss": 0.3962,
|
| 1905 |
+
"step": 228
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 25.444444444444443,
|
| 1909 |
+
"grad_norm": 99825.3984375,
|
| 1910 |
+
"learning_rate": 2.6297657326596232e-08,
|
| 1911 |
+
"loss": 0.4211,
|
| 1912 |
+
"step": 229
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 25.555555555555557,
|
| 1916 |
+
"grad_norm": 172674.21875,
|
| 1917 |
+
"learning_rate": 2.6412494258153425e-08,
|
| 1918 |
+
"loss": 0.4087,
|
| 1919 |
+
"step": 230
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 25.666666666666668,
|
| 1923 |
+
"grad_norm": 150575.75,
|
| 1924 |
+
"learning_rate": 2.6527331189710614e-08,
|
| 1925 |
+
"loss": 0.4029,
|
| 1926 |
+
"step": 231
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 25.77777777777778,
|
| 1930 |
+
"grad_norm": 112252.109375,
|
| 1931 |
+
"learning_rate": 2.6642168121267803e-08,
|
| 1932 |
+
"loss": 0.3987,
|
| 1933 |
+
"step": 232
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 25.88888888888889,
|
| 1937 |
+
"grad_norm": 182919.984375,
|
| 1938 |
+
"learning_rate": 2.6757005052824992e-08,
|
| 1939 |
+
"loss": 0.3986,
|
| 1940 |
+
"step": 233
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 26.0,
|
| 1944 |
+
"grad_norm": 342678.03125,
|
| 1945 |
+
"learning_rate": 2.6871841984382178e-08,
|
| 1946 |
+
"loss": 0.3989,
|
| 1947 |
+
"step": 234
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 26.0,
|
| 1951 |
+
"eval_accuracy": 0.8121267799724391,
|
| 1952 |
+
"eval_f1": 0.7995098039215687,
|
| 1953 |
+
"eval_loss": 0.40705057978630066,
|
| 1954 |
+
"eval_precision": 0.8150924537731135,
|
| 1955 |
+
"eval_recall": 0.7845117845117845,
|
| 1956 |
+
"eval_runtime": 4.3118,
|
| 1957 |
+
"eval_samples_per_second": 1009.793,
|
| 1958 |
+
"eval_steps_per_second": 0.696,
|
| 1959 |
+
"step": 234
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 26.11111111111111,
|
| 1963 |
+
"grad_norm": 234523.734375,
|
| 1964 |
+
"learning_rate": 2.698667891593937e-08,
|
| 1965 |
+
"loss": 0.3909,
|
| 1966 |
+
"step": 235
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 26.22222222222222,
|
| 1970 |
+
"grad_norm": 134018.859375,
|
| 1971 |
+
"learning_rate": 2.7101515847496557e-08,
|
| 1972 |
+
"loss": 0.3957,
|
| 1973 |
+
"step": 236
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 26.333333333333332,
|
| 1977 |
+
"grad_norm": 84843.9765625,
|
| 1978 |
+
"learning_rate": 2.721635277905375e-08,
|
| 1979 |
+
"loss": 0.385,
|
| 1980 |
+
"step": 237
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 26.444444444444443,
|
| 1984 |
+
"grad_norm": 127592.4140625,
|
| 1985 |
+
"learning_rate": 2.7331189710610935e-08,
|
| 1986 |
+
"loss": 0.4155,
|
| 1987 |
+
"step": 238
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 26.555555555555557,
|
| 1991 |
+
"grad_norm": 101331.8828125,
|
| 1992 |
+
"learning_rate": 2.7446026642168124e-08,
|
| 1993 |
+
"loss": 0.3954,
|
| 1994 |
+
"step": 239
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 26.666666666666668,
|
| 1998 |
+
"grad_norm": 97625.59375,
|
| 1999 |
+
"learning_rate": 2.756086357372531e-08,
|
| 2000 |
+
"loss": 0.4154,
|
| 2001 |
+
"step": 240
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 26.77777777777778,
|
| 2005 |
+
"grad_norm": 268018.65625,
|
| 2006 |
+
"learning_rate": 2.7675700505282502e-08,
|
| 2007 |
+
"loss": 0.4041,
|
| 2008 |
+
"step": 241
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 26.88888888888889,
|
| 2012 |
+
"grad_norm": 121827.4140625,
|
| 2013 |
+
"learning_rate": 2.7790537436839688e-08,
|
| 2014 |
+
"loss": 0.3923,
|
| 2015 |
+
"step": 242
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 27.0,
|
| 2019 |
+
"grad_norm": 137969.9375,
|
| 2020 |
+
"learning_rate": 2.790537436839688e-08,
|
| 2021 |
+
"loss": 0.395,
|
| 2022 |
+
"step": 243
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 27.0,
|
| 2026 |
+
"eval_accuracy": 0.8123564538355535,
|
| 2027 |
+
"eval_f1": 0.7999020328190056,
|
| 2028 |
+
"eval_loss": 0.4069819748401642,
|
| 2029 |
+
"eval_precision": 0.814870259481038,
|
| 2030 |
+
"eval_recall": 0.7854737854737854,
|
| 2031 |
+
"eval_runtime": 4.3601,
|
| 2032 |
+
"eval_samples_per_second": 998.609,
|
| 2033 |
+
"eval_steps_per_second": 0.688,
|
| 2034 |
+
"step": 243
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 27.11111111111111,
|
| 2038 |
+
"grad_norm": 94782.7109375,
|
| 2039 |
+
"learning_rate": 2.8020211299954067e-08,
|
| 2040 |
+
"loss": 0.3945,
|
| 2041 |
+
"step": 244
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 27.22222222222222,
|
| 2045 |
+
"grad_norm": 110734.5703125,
|
| 2046 |
+
"learning_rate": 2.8135048231511256e-08,
|
| 2047 |
+
"loss": 0.3908,
|
| 2048 |
+
"step": 245
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 27.333333333333332,
|
| 2052 |
+
"grad_norm": 98079.078125,
|
| 2053 |
+
"learning_rate": 2.8249885163068445e-08,
|
| 2054 |
+
"loss": 0.401,
|
| 2055 |
+
"step": 246
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 27.444444444444443,
|
| 2059 |
+
"grad_norm": 62154.65625,
|
| 2060 |
+
"learning_rate": 2.8364722094625634e-08,
|
| 2061 |
+
"loss": 0.4173,
|
| 2062 |
+
"step": 247
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 27.555555555555557,
|
| 2066 |
+
"grad_norm": 100930.515625,
|
| 2067 |
+
"learning_rate": 2.847955902618282e-08,
|
| 2068 |
+
"loss": 0.3957,
|
| 2069 |
+
"step": 248
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 27.666666666666668,
|
| 2073 |
+
"grad_norm": 129585.421875,
|
| 2074 |
+
"learning_rate": 2.8594395957740013e-08,
|
| 2075 |
+
"loss": 0.4201,
|
| 2076 |
+
"step": 249
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 27.77777777777778,
|
| 2080 |
+
"grad_norm": 106040.046875,
|
| 2081 |
+
"learning_rate": 2.87092328892972e-08,
|
| 2082 |
+
"loss": 0.399,
|
| 2083 |
+
"step": 250
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 27.88888888888889,
|
| 2087 |
+
"grad_norm": 101252.2265625,
|
| 2088 |
+
"learning_rate": 2.882406982085439e-08,
|
| 2089 |
+
"loss": 0.4012,
|
| 2090 |
+
"step": 251
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 28.0,
|
| 2094 |
+
"grad_norm": 204035.765625,
|
| 2095 |
+
"learning_rate": 2.8938906752411577e-08,
|
| 2096 |
+
"loss": 0.3682,
|
| 2097 |
+
"step": 252
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 28.0,
|
| 2101 |
+
"eval_accuracy": 0.8123564538355535,
|
| 2102 |
+
"eval_f1": 0.8,
|
| 2103 |
+
"eval_loss": 0.40691375732421875,
|
| 2104 |
+
"eval_precision": 0.8145563310069791,
|
| 2105 |
+
"eval_recall": 0.785954785954786,
|
| 2106 |
+
"eval_runtime": 4.3908,
|
| 2107 |
+
"eval_samples_per_second": 991.612,
|
| 2108 |
+
"eval_steps_per_second": 0.683,
|
| 2109 |
+
"step": 252
|
| 2110 |
+
}
|
| 2111 |
+
],
|
| 2112 |
+
"logging_steps": 1,
|
| 2113 |
+
"max_steps": 450,
|
| 2114 |
+
"num_input_tokens_seen": 0,
|
| 2115 |
+
"num_train_epochs": 50,
|
| 2116 |
+
"save_steps": 500,
|
| 2117 |
+
"stateful_callbacks": {
|
| 2118 |
+
"TrainerControl": {
|
| 2119 |
+
"args": {
|
| 2120 |
+
"should_epoch_stop": false,
|
| 2121 |
+
"should_evaluate": false,
|
| 2122 |
+
"should_log": false,
|
| 2123 |
+
"should_save": true,
|
| 2124 |
+
"should_training_stop": false
|
| 2125 |
+
},
|
| 2126 |
+
"attributes": {}
|
| 2127 |
+
}
|
| 2128 |
+
},
|
| 2129 |
+
"total_flos": 1.922497283240755e+16,
|
| 2130 |
+
"train_batch_size": 2048,
|
| 2131 |
+
"trial_name": null,
|
| 2132 |
+
"trial_params": null
|
| 2133 |
+
}
|
checkpoint-252/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:74dac413ac07c27dc9943e74eae0633252d6bd133e88b42bc5e86ef037ddc678
|
| 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 sha256:1cd3c11b6cd894c113178f0338d6a9fe1b56741a6767848ff5601040a1eb3837
|
| 3 |
size 8563512
|
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 sha256:74dac413ac07c27dc9943e74eae0633252d6bd133e88b42bc5e86ef037ddc678
|
| 3 |
size 5368
|