Commit
·
50d4467
1
Parent(s):
076c454
Add evaluation results and output files to model card
Browse files- checkpoint-432/config.json +23 -0
- checkpoint-432/model.safetensors +3 -0
- checkpoint-432/optimizer.pt +3 -0
- checkpoint-432/preprocessor_config.json +22 -0
- checkpoint-432/rng_state.pth +3 -0
- checkpoint-432/scheduler.pt +3 -0
- checkpoint-432/trainer_state.json +3633 -0
- checkpoint-432/training_args.bin +3 -0
- model.safetensors +1 -1
- training_args.bin +1 -1
checkpoint-432/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-432/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:644d5a596e90f66fd3182460217969f710bd1c381d54989fe396eebc0b89317a
|
| 3 |
+
size 8563512
|
checkpoint-432/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7edc1d89b07e4cbb1ceb7ace817fbc50db69f700522cbfcf55a0f84bcc2661fe
|
| 3 |
+
size 17171514
|
checkpoint-432/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-432/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:648dcfb444ad7fb98902cb925e20a9c5a134b22adbcacacee667b1a501d4033e
|
| 3 |
+
size 14244
|
checkpoint-432/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bb858fbb0baebe5a9c0da911e2ec407ba692026ef0af136cace00a95b5381bd3
|
| 3 |
+
size 1000
|
checkpoint-432/trainer_state.json
ADDED
|
@@ -0,0 +1,3633 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.806949806949807,
|
| 3 |
+
"best_model_checkpoint": "/tmp/logs/binary_classification_model_v3.1.5_spines/checkpoint-432",
|
| 4 |
+
"epoch": 48.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 432,
|
| 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": 482914.5625,
|
| 14 |
+
"learning_rate": 1.148369315571888e-09,
|
| 15 |
+
"loss": 0.4045,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.2222222222222222,
|
| 20 |
+
"grad_norm": 435595.6875,
|
| 21 |
+
"learning_rate": 2.296738631143776e-09,
|
| 22 |
+
"loss": 0.4188,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3333333333333333,
|
| 27 |
+
"grad_norm": 351966.96875,
|
| 28 |
+
"learning_rate": 3.4451079467156637e-09,
|
| 29 |
+
"loss": 0.4189,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.4444444444444444,
|
| 34 |
+
"grad_norm": 439417.3125,
|
| 35 |
+
"learning_rate": 4.593477262287552e-09,
|
| 36 |
+
"loss": 0.3916,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.5555555555555556,
|
| 41 |
+
"grad_norm": 207262.421875,
|
| 42 |
+
"learning_rate": 5.74184657785944e-09,
|
| 43 |
+
"loss": 0.3841,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.6666666666666666,
|
| 48 |
+
"grad_norm": 313504.625,
|
| 49 |
+
"learning_rate": 6.8902158934313275e-09,
|
| 50 |
+
"loss": 0.392,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.7777777777777778,
|
| 55 |
+
"grad_norm": 291762.28125,
|
| 56 |
+
"learning_rate": 8.038585209003216e-09,
|
| 57 |
+
"loss": 0.3982,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.8888888888888888,
|
| 62 |
+
"grad_norm": 359227.53125,
|
| 63 |
+
"learning_rate": 9.186954524575104e-09,
|
| 64 |
+
"loss": 0.4042,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.0,
|
| 69 |
+
"grad_norm": 620293.5625,
|
| 70 |
+
"learning_rate": 1.0335323840146991e-08,
|
| 71 |
+
"loss": 0.4047,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.0,
|
| 76 |
+
"eval_accuracy": 0.8070739549839229,
|
| 77 |
+
"eval_f1": 0.7889447236180904,
|
| 78 |
+
"eval_loss": 0.40964627265930176,
|
| 79 |
+
"eval_precision": 0.8258811152025249,
|
| 80 |
+
"eval_recall": 0.7551707551707552,
|
| 81 |
+
"eval_runtime": 3.8706,
|
| 82 |
+
"eval_samples_per_second": 1124.894,
|
| 83 |
+
"eval_steps_per_second": 0.775,
|
| 84 |
+
"step": 9
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.1111111111111112,
|
| 88 |
+
"grad_norm": 584924.0,
|
| 89 |
+
"learning_rate": 1.148369315571888e-08,
|
| 90 |
+
"loss": 0.4252,
|
| 91 |
+
"step": 10
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.2222222222222223,
|
| 95 |
+
"grad_norm": 242005.0,
|
| 96 |
+
"learning_rate": 1.2632062471290768e-08,
|
| 97 |
+
"loss": 0.3976,
|
| 98 |
+
"step": 11
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.3333333333333333,
|
| 102 |
+
"grad_norm": 403620.0625,
|
| 103 |
+
"learning_rate": 1.3780431786862655e-08,
|
| 104 |
+
"loss": 0.4065,
|
| 105 |
+
"step": 12
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.4444444444444444,
|
| 109 |
+
"grad_norm": 337349.9375,
|
| 110 |
+
"learning_rate": 1.4928801102434543e-08,
|
| 111 |
+
"loss": 0.3853,
|
| 112 |
+
"step": 13
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.5555555555555556,
|
| 116 |
+
"grad_norm": 381869.15625,
|
| 117 |
+
"learning_rate": 1.6077170418006432e-08,
|
| 118 |
+
"loss": 0.3902,
|
| 119 |
+
"step": 14
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.6666666666666665,
|
| 123 |
+
"grad_norm": 196447.03125,
|
| 124 |
+
"learning_rate": 1.722553973357832e-08,
|
| 125 |
+
"loss": 0.4081,
|
| 126 |
+
"step": 15
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 1.7777777777777777,
|
| 130 |
+
"grad_norm": 276488.40625,
|
| 131 |
+
"learning_rate": 1.837390904915021e-08,
|
| 132 |
+
"loss": 0.4209,
|
| 133 |
+
"step": 16
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 1.8888888888888888,
|
| 137 |
+
"grad_norm": 402769.25,
|
| 138 |
+
"learning_rate": 1.9522278364722097e-08,
|
| 139 |
+
"loss": 0.4025,
|
| 140 |
+
"step": 17
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 2.0,
|
| 144 |
+
"grad_norm": 232493.25,
|
| 145 |
+
"learning_rate": 2.0670647680293982e-08,
|
| 146 |
+
"loss": 0.4144,
|
| 147 |
+
"step": 18
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 2.0,
|
| 151 |
+
"eval_accuracy": 0.808451998162609,
|
| 152 |
+
"eval_f1": 0.7910821643286573,
|
| 153 |
+
"eval_loss": 0.40929052233695984,
|
| 154 |
+
"eval_precision": 0.825405122843701,
|
| 155 |
+
"eval_recall": 0.7594997594997595,
|
| 156 |
+
"eval_runtime": 3.9399,
|
| 157 |
+
"eval_samples_per_second": 1105.093,
|
| 158 |
+
"eval_steps_per_second": 0.761,
|
| 159 |
+
"step": 18
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 2.111111111111111,
|
| 163 |
+
"grad_norm": 534834.125,
|
| 164 |
+
"learning_rate": 2.1819016995865874e-08,
|
| 165 |
+
"loss": 0.4073,
|
| 166 |
+
"step": 19
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 2.2222222222222223,
|
| 170 |
+
"grad_norm": 271332.875,
|
| 171 |
+
"learning_rate": 2.296738631143776e-08,
|
| 172 |
+
"loss": 0.4121,
|
| 173 |
+
"step": 20
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 2.3333333333333335,
|
| 177 |
+
"grad_norm": 489498.5625,
|
| 178 |
+
"learning_rate": 2.4115755627009648e-08,
|
| 179 |
+
"loss": 0.4261,
|
| 180 |
+
"step": 21
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 2.4444444444444446,
|
| 184 |
+
"grad_norm": 242193.1875,
|
| 185 |
+
"learning_rate": 2.5264124942581536e-08,
|
| 186 |
+
"loss": 0.3974,
|
| 187 |
+
"step": 22
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 2.5555555555555554,
|
| 191 |
+
"grad_norm": 318717.34375,
|
| 192 |
+
"learning_rate": 2.6412494258153425e-08,
|
| 193 |
+
"loss": 0.4082,
|
| 194 |
+
"step": 23
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.6666666666666665,
|
| 198 |
+
"grad_norm": 396368.1875,
|
| 199 |
+
"learning_rate": 2.756086357372531e-08,
|
| 200 |
+
"loss": 0.3815,
|
| 201 |
+
"step": 24
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 2.7777777777777777,
|
| 205 |
+
"grad_norm": 247636.09375,
|
| 206 |
+
"learning_rate": 2.87092328892972e-08,
|
| 207 |
+
"loss": 0.3953,
|
| 208 |
+
"step": 25
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 2.888888888888889,
|
| 212 |
+
"grad_norm": 294976.5,
|
| 213 |
+
"learning_rate": 2.985760220486909e-08,
|
| 214 |
+
"loss": 0.4046,
|
| 215 |
+
"step": 26
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 3.0,
|
| 219 |
+
"grad_norm": 161600.75,
|
| 220 |
+
"learning_rate": 3.100597152044098e-08,
|
| 221 |
+
"loss": 0.3723,
|
| 222 |
+
"step": 27
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 3.0,
|
| 226 |
+
"eval_accuracy": 0.8091410197519522,
|
| 227 |
+
"eval_f1": 0.7928197456993269,
|
| 228 |
+
"eval_loss": 0.4087640643119812,
|
| 229 |
+
"eval_precision": 0.8229813664596274,
|
| 230 |
+
"eval_recall": 0.7647907647907648,
|
| 231 |
+
"eval_runtime": 3.9741,
|
| 232 |
+
"eval_samples_per_second": 1095.607,
|
| 233 |
+
"eval_steps_per_second": 0.755,
|
| 234 |
+
"step": 27
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 3.111111111111111,
|
| 238 |
+
"grad_norm": 305851.40625,
|
| 239 |
+
"learning_rate": 3.2154340836012864e-08,
|
| 240 |
+
"loss": 0.3942,
|
| 241 |
+
"step": 28
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 3.2222222222222223,
|
| 245 |
+
"grad_norm": 380446.75,
|
| 246 |
+
"learning_rate": 3.330271015158475e-08,
|
| 247 |
+
"loss": 0.3954,
|
| 248 |
+
"step": 29
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 3.3333333333333335,
|
| 252 |
+
"grad_norm": 160068.890625,
|
| 253 |
+
"learning_rate": 3.445107946715664e-08,
|
| 254 |
+
"loss": 0.4057,
|
| 255 |
+
"step": 30
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 3.4444444444444446,
|
| 259 |
+
"grad_norm": 351186.96875,
|
| 260 |
+
"learning_rate": 3.5599448782728526e-08,
|
| 261 |
+
"loss": 0.4015,
|
| 262 |
+
"step": 31
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 3.5555555555555554,
|
| 266 |
+
"grad_norm": 283727.625,
|
| 267 |
+
"learning_rate": 3.674781809830042e-08,
|
| 268 |
+
"loss": 0.4116,
|
| 269 |
+
"step": 32
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.6666666666666665,
|
| 273 |
+
"grad_norm": 348961.09375,
|
| 274 |
+
"learning_rate": 3.78961874138723e-08,
|
| 275 |
+
"loss": 0.403,
|
| 276 |
+
"step": 33
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 3.7777777777777777,
|
| 280 |
+
"grad_norm": 474216.40625,
|
| 281 |
+
"learning_rate": 3.9044556729444195e-08,
|
| 282 |
+
"loss": 0.4025,
|
| 283 |
+
"step": 34
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 3.888888888888889,
|
| 287 |
+
"grad_norm": 228167.640625,
|
| 288 |
+
"learning_rate": 4.019292604501608e-08,
|
| 289 |
+
"loss": 0.4208,
|
| 290 |
+
"step": 35
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 4.0,
|
| 294 |
+
"grad_norm": 195539.109375,
|
| 295 |
+
"learning_rate": 4.1341295360587965e-08,
|
| 296 |
+
"loss": 0.4127,
|
| 297 |
+
"step": 36
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 4.0,
|
| 301 |
+
"eval_accuracy": 0.8089113458888378,
|
| 302 |
+
"eval_f1": 0.7935483870967742,
|
| 303 |
+
"eval_loss": 0.4081500768661499,
|
| 304 |
+
"eval_precision": 0.8195797027165556,
|
| 305 |
+
"eval_recall": 0.7691197691197691,
|
| 306 |
+
"eval_runtime": 4.0357,
|
| 307 |
+
"eval_samples_per_second": 1078.877,
|
| 308 |
+
"eval_steps_per_second": 0.743,
|
| 309 |
+
"step": 36
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 4.111111111111111,
|
| 313 |
+
"grad_norm": 157376.421875,
|
| 314 |
+
"learning_rate": 4.248966467615985e-08,
|
| 315 |
+
"loss": 0.4016,
|
| 316 |
+
"step": 37
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 4.222222222222222,
|
| 320 |
+
"grad_norm": 114119.765625,
|
| 321 |
+
"learning_rate": 4.363803399173175e-08,
|
| 322 |
+
"loss": 0.4224,
|
| 323 |
+
"step": 38
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 4.333333333333333,
|
| 327 |
+
"grad_norm": 240913.140625,
|
| 328 |
+
"learning_rate": 4.4786403307303634e-08,
|
| 329 |
+
"loss": 0.3962,
|
| 330 |
+
"step": 39
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 4.444444444444445,
|
| 334 |
+
"grad_norm": 243501.609375,
|
| 335 |
+
"learning_rate": 4.593477262287552e-08,
|
| 336 |
+
"loss": 0.3955,
|
| 337 |
+
"step": 40
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 4.555555555555555,
|
| 341 |
+
"grad_norm": 291720.53125,
|
| 342 |
+
"learning_rate": 4.7083141938447404e-08,
|
| 343 |
+
"loss": 0.401,
|
| 344 |
+
"step": 41
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.666666666666667,
|
| 348 |
+
"grad_norm": 194599.9375,
|
| 349 |
+
"learning_rate": 4.8231511254019296e-08,
|
| 350 |
+
"loss": 0.4012,
|
| 351 |
+
"step": 42
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 4.777777777777778,
|
| 355 |
+
"grad_norm": 337977.96875,
|
| 356 |
+
"learning_rate": 4.937988056959119e-08,
|
| 357 |
+
"loss": 0.4027,
|
| 358 |
+
"step": 43
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 4.888888888888889,
|
| 362 |
+
"grad_norm": 159864.734375,
|
| 363 |
+
"learning_rate": 5.052824988516307e-08,
|
| 364 |
+
"loss": 0.3815,
|
| 365 |
+
"step": 44
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 5.0,
|
| 369 |
+
"grad_norm": 276918.53125,
|
| 370 |
+
"learning_rate": 5.167661920073496e-08,
|
| 371 |
+
"loss": 0.4277,
|
| 372 |
+
"step": 45
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 5.0,
|
| 376 |
+
"eval_accuracy": 0.8109784106568673,
|
| 377 |
+
"eval_f1": 0.7970406905055487,
|
| 378 |
+
"eval_loss": 0.4075668752193451,
|
| 379 |
+
"eval_precision": 0.8178137651821862,
|
| 380 |
+
"eval_recall": 0.7772967772967773,
|
| 381 |
+
"eval_runtime": 4.127,
|
| 382 |
+
"eval_samples_per_second": 1055.009,
|
| 383 |
+
"eval_steps_per_second": 0.727,
|
| 384 |
+
"step": 45
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 5.111111111111111,
|
| 388 |
+
"grad_norm": 173940.5625,
|
| 389 |
+
"learning_rate": 5.282498851630685e-08,
|
| 390 |
+
"loss": 0.4088,
|
| 391 |
+
"step": 46
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 5.222222222222222,
|
| 395 |
+
"grad_norm": 146474.640625,
|
| 396 |
+
"learning_rate": 5.3973357831878735e-08,
|
| 397 |
+
"loss": 0.3768,
|
| 398 |
+
"step": 47
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 5.333333333333333,
|
| 402 |
+
"grad_norm": 108079.078125,
|
| 403 |
+
"learning_rate": 5.512172714745062e-08,
|
| 404 |
+
"loss": 0.3926,
|
| 405 |
+
"step": 48
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 5.444444444444445,
|
| 409 |
+
"grad_norm": 336127.96875,
|
| 410 |
+
"learning_rate": 5.6270096463022505e-08,
|
| 411 |
+
"loss": 0.4333,
|
| 412 |
+
"step": 49
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 5.555555555555555,
|
| 416 |
+
"grad_norm": 110071.4296875,
|
| 417 |
+
"learning_rate": 5.74184657785944e-08,
|
| 418 |
+
"loss": 0.3931,
|
| 419 |
+
"step": 50
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 5.666666666666667,
|
| 423 |
+
"grad_norm": 160531.78125,
|
| 424 |
+
"learning_rate": 5.856683509416628e-08,
|
| 425 |
+
"loss": 0.4092,
|
| 426 |
+
"step": 51
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 5.777777777777778,
|
| 430 |
+
"grad_norm": 329436.15625,
|
| 431 |
+
"learning_rate": 5.971520440973817e-08,
|
| 432 |
+
"loss": 0.4036,
|
| 433 |
+
"step": 52
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 5.888888888888889,
|
| 437 |
+
"grad_norm": 138796.90625,
|
| 438 |
+
"learning_rate": 6.086357372531007e-08,
|
| 439 |
+
"loss": 0.4063,
|
| 440 |
+
"step": 53
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 6.0,
|
| 444 |
+
"grad_norm": 289409.84375,
|
| 445 |
+
"learning_rate": 6.201194304088196e-08,
|
| 446 |
+
"loss": 0.3986,
|
| 447 |
+
"step": 54
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 6.0,
|
| 451 |
+
"eval_accuracy": 0.8116674322462104,
|
| 452 |
+
"eval_f1": 0.7988223748773308,
|
| 453 |
+
"eval_loss": 0.4071072041988373,
|
| 454 |
+
"eval_precision": 0.8152228342513771,
|
| 455 |
+
"eval_recall": 0.783068783068783,
|
| 456 |
+
"eval_runtime": 4.1104,
|
| 457 |
+
"eval_samples_per_second": 1059.253,
|
| 458 |
+
"eval_steps_per_second": 0.73,
|
| 459 |
+
"step": 54
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 6.111111111111111,
|
| 463 |
+
"grad_norm": 177895.421875,
|
| 464 |
+
"learning_rate": 6.316031235645384e-08,
|
| 465 |
+
"loss": 0.4005,
|
| 466 |
+
"step": 55
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 6.222222222222222,
|
| 470 |
+
"grad_norm": 221305.015625,
|
| 471 |
+
"learning_rate": 6.430868167202573e-08,
|
| 472 |
+
"loss": 0.4065,
|
| 473 |
+
"step": 56
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 6.333333333333333,
|
| 477 |
+
"grad_norm": 85639.6171875,
|
| 478 |
+
"learning_rate": 6.545705098759761e-08,
|
| 479 |
+
"loss": 0.3974,
|
| 480 |
+
"step": 57
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 6.444444444444445,
|
| 484 |
+
"grad_norm": 231708.6875,
|
| 485 |
+
"learning_rate": 6.66054203031695e-08,
|
| 486 |
+
"loss": 0.3898,
|
| 487 |
+
"step": 58
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 6.555555555555555,
|
| 491 |
+
"grad_norm": 149632.109375,
|
| 492 |
+
"learning_rate": 6.775378961874138e-08,
|
| 493 |
+
"loss": 0.4166,
|
| 494 |
+
"step": 59
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 6.666666666666667,
|
| 498 |
+
"grad_norm": 117517.3515625,
|
| 499 |
+
"learning_rate": 6.890215893431328e-08,
|
| 500 |
+
"loss": 0.3889,
|
| 501 |
+
"step": 60
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 6.777777777777778,
|
| 505 |
+
"grad_norm": 138410.859375,
|
| 506 |
+
"learning_rate": 7.005052824988517e-08,
|
| 507 |
+
"loss": 0.4241,
|
| 508 |
+
"step": 61
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 6.888888888888889,
|
| 512 |
+
"grad_norm": 227332.6875,
|
| 513 |
+
"learning_rate": 7.119889756545705e-08,
|
| 514 |
+
"loss": 0.4088,
|
| 515 |
+
"step": 62
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 7.0,
|
| 519 |
+
"grad_norm": 88005.5234375,
|
| 520 |
+
"learning_rate": 7.234726688102894e-08,
|
| 521 |
+
"loss": 0.4032,
|
| 522 |
+
"step": 63
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 7.0,
|
| 526 |
+
"eval_accuracy": 0.8137344970142398,
|
| 527 |
+
"eval_f1": 0.8020502806931902,
|
| 528 |
+
"eval_loss": 0.4067862927913666,
|
| 529 |
+
"eval_precision": 0.8141724479682855,
|
| 530 |
+
"eval_recall": 0.7902837902837903,
|
| 531 |
+
"eval_runtime": 4.1376,
|
| 532 |
+
"eval_samples_per_second": 1052.312,
|
| 533 |
+
"eval_steps_per_second": 0.725,
|
| 534 |
+
"step": 63
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 7.111111111111111,
|
| 538 |
+
"grad_norm": 66663.6328125,
|
| 539 |
+
"learning_rate": 7.349563619660084e-08,
|
| 540 |
+
"loss": 0.4037,
|
| 541 |
+
"step": 64
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 7.222222222222222,
|
| 545 |
+
"grad_norm": 195190.40625,
|
| 546 |
+
"learning_rate": 7.464400551217272e-08,
|
| 547 |
+
"loss": 0.4011,
|
| 548 |
+
"step": 65
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 7.333333333333333,
|
| 552 |
+
"grad_norm": 77268.6875,
|
| 553 |
+
"learning_rate": 7.57923748277446e-08,
|
| 554 |
+
"loss": 0.3993,
|
| 555 |
+
"step": 66
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 7.444444444444445,
|
| 559 |
+
"grad_norm": 123770.078125,
|
| 560 |
+
"learning_rate": 7.694074414331649e-08,
|
| 561 |
+
"loss": 0.4058,
|
| 562 |
+
"step": 67
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 7.555555555555555,
|
| 566 |
+
"grad_norm": 206524.921875,
|
| 567 |
+
"learning_rate": 7.808911345888839e-08,
|
| 568 |
+
"loss": 0.4068,
|
| 569 |
+
"step": 68
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 7.666666666666667,
|
| 573 |
+
"grad_norm": 93597.125,
|
| 574 |
+
"learning_rate": 7.923748277446027e-08,
|
| 575 |
+
"loss": 0.4005,
|
| 576 |
+
"step": 69
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 7.777777777777778,
|
| 580 |
+
"grad_norm": 137944.546875,
|
| 581 |
+
"learning_rate": 8.038585209003216e-08,
|
| 582 |
+
"loss": 0.4114,
|
| 583 |
+
"step": 70
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 7.888888888888889,
|
| 587 |
+
"grad_norm": 152527.265625,
|
| 588 |
+
"learning_rate": 8.153422140560404e-08,
|
| 589 |
+
"loss": 0.4002,
|
| 590 |
+
"step": 71
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 8.0,
|
| 594 |
+
"grad_norm": 189652.96875,
|
| 595 |
+
"learning_rate": 8.268259072117593e-08,
|
| 596 |
+
"loss": 0.3697,
|
| 597 |
+
"step": 72
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 8.0,
|
| 601 |
+
"eval_accuracy": 0.815112540192926,
|
| 602 |
+
"eval_f1": 0.8042791150012156,
|
| 603 |
+
"eval_loss": 0.4065553843975067,
|
| 604 |
+
"eval_precision": 0.8131760078662733,
|
| 605 |
+
"eval_recall": 0.7955747955747956,
|
| 606 |
+
"eval_runtime": 4.1923,
|
| 607 |
+
"eval_samples_per_second": 1038.578,
|
| 608 |
+
"eval_steps_per_second": 0.716,
|
| 609 |
+
"step": 72
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 8.11111111111111,
|
| 613 |
+
"grad_norm": 118062.1328125,
|
| 614 |
+
"learning_rate": 8.383096003674781e-08,
|
| 615 |
+
"loss": 0.3995,
|
| 616 |
+
"step": 73
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 8.222222222222221,
|
| 620 |
+
"grad_norm": 118298.40625,
|
| 621 |
+
"learning_rate": 8.49793293523197e-08,
|
| 622 |
+
"loss": 0.4072,
|
| 623 |
+
"step": 74
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 8.333333333333334,
|
| 627 |
+
"grad_norm": 165875.65625,
|
| 628 |
+
"learning_rate": 8.61276986678916e-08,
|
| 629 |
+
"loss": 0.4195,
|
| 630 |
+
"step": 75
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 8.444444444444445,
|
| 634 |
+
"grad_norm": 141588.953125,
|
| 635 |
+
"learning_rate": 8.72760679834635e-08,
|
| 636 |
+
"loss": 0.4065,
|
| 637 |
+
"step": 76
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 8.555555555555555,
|
| 641 |
+
"grad_norm": 118747.3046875,
|
| 642 |
+
"learning_rate": 8.842443729903538e-08,
|
| 643 |
+
"loss": 0.4114,
|
| 644 |
+
"step": 77
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 8.666666666666666,
|
| 648 |
+
"grad_norm": 217602.65625,
|
| 649 |
+
"learning_rate": 8.957280661460727e-08,
|
| 650 |
+
"loss": 0.4167,
|
| 651 |
+
"step": 78
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 8.777777777777779,
|
| 655 |
+
"grad_norm": 130492.5390625,
|
| 656 |
+
"learning_rate": 9.072117593017915e-08,
|
| 657 |
+
"loss": 0.3948,
|
| 658 |
+
"step": 79
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 8.88888888888889,
|
| 662 |
+
"grad_norm": 157410.8125,
|
| 663 |
+
"learning_rate": 9.186954524575104e-08,
|
| 664 |
+
"loss": 0.382,
|
| 665 |
+
"step": 80
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 9.0,
|
| 669 |
+
"grad_norm": 297632.4375,
|
| 670 |
+
"learning_rate": 9.301791456132292e-08,
|
| 671 |
+
"loss": 0.3903,
|
| 672 |
+
"step": 81
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 9.0,
|
| 676 |
+
"eval_accuracy": 0.8164905833716123,
|
| 677 |
+
"eval_f1": 0.8062090710647587,
|
| 678 |
+
"eval_loss": 0.40632495284080505,
|
| 679 |
+
"eval_precision": 0.8131115459882583,
|
| 680 |
+
"eval_recall": 0.7994227994227994,
|
| 681 |
+
"eval_runtime": 4.2297,
|
| 682 |
+
"eval_samples_per_second": 1029.392,
|
| 683 |
+
"eval_steps_per_second": 0.709,
|
| 684 |
+
"step": 81
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 9.11111111111111,
|
| 688 |
+
"grad_norm": 78985.640625,
|
| 689 |
+
"learning_rate": 9.416628387689481e-08,
|
| 690 |
+
"loss": 0.3857,
|
| 691 |
+
"step": 82
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 9.222222222222221,
|
| 695 |
+
"grad_norm": 63331.29296875,
|
| 696 |
+
"learning_rate": 9.53146531924667e-08,
|
| 697 |
+
"loss": 0.398,
|
| 698 |
+
"step": 83
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 9.333333333333334,
|
| 702 |
+
"grad_norm": 234827.0,
|
| 703 |
+
"learning_rate": 9.646302250803859e-08,
|
| 704 |
+
"loss": 0.4111,
|
| 705 |
+
"step": 84
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 9.444444444444445,
|
| 709 |
+
"grad_norm": 84457.109375,
|
| 710 |
+
"learning_rate": 9.761139182361048e-08,
|
| 711 |
+
"loss": 0.3911,
|
| 712 |
+
"step": 85
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 9.555555555555555,
|
| 716 |
+
"grad_norm": 111777.2109375,
|
| 717 |
+
"learning_rate": 9.875976113918237e-08,
|
| 718 |
+
"loss": 0.3972,
|
| 719 |
+
"step": 86
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 9.666666666666666,
|
| 723 |
+
"grad_norm": 114194.203125,
|
| 724 |
+
"learning_rate": 9.990813045475426e-08,
|
| 725 |
+
"loss": 0.407,
|
| 726 |
+
"step": 87
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 9.777777777777779,
|
| 730 |
+
"grad_norm": 99612.53125,
|
| 731 |
+
"learning_rate": 1.0105649977032615e-07,
|
| 732 |
+
"loss": 0.4061,
|
| 733 |
+
"step": 88
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 9.88888888888889,
|
| 737 |
+
"grad_norm": 79423.015625,
|
| 738 |
+
"learning_rate": 1.0220486908589803e-07,
|
| 739 |
+
"loss": 0.4026,
|
| 740 |
+
"step": 89
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 10.0,
|
| 744 |
+
"grad_norm": 290108.0,
|
| 745 |
+
"learning_rate": 1.0335323840146992e-07,
|
| 746 |
+
"loss": 0.3937,
|
| 747 |
+
"step": 90
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 10.0,
|
| 751 |
+
"eval_accuracy": 0.8153422140560405,
|
| 752 |
+
"eval_f1": 0.8045697617890131,
|
| 753 |
+
"eval_loss": 0.40608739852905273,
|
| 754 |
+
"eval_precision": 0.8132678132678133,
|
| 755 |
+
"eval_recall": 0.796055796055796,
|
| 756 |
+
"eval_runtime": 4.27,
|
| 757 |
+
"eval_samples_per_second": 1019.676,
|
| 758 |
+
"eval_steps_per_second": 0.703,
|
| 759 |
+
"step": 90
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 10.11111111111111,
|
| 763 |
+
"grad_norm": 139479.40625,
|
| 764 |
+
"learning_rate": 1.045016077170418e-07,
|
| 765 |
+
"loss": 0.399,
|
| 766 |
+
"step": 91
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 10.222222222222221,
|
| 770 |
+
"grad_norm": 72552.421875,
|
| 771 |
+
"learning_rate": 1.056499770326137e-07,
|
| 772 |
+
"loss": 0.3969,
|
| 773 |
+
"step": 92
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 10.333333333333334,
|
| 777 |
+
"grad_norm": 234034.046875,
|
| 778 |
+
"learning_rate": 1.0679834634818558e-07,
|
| 779 |
+
"loss": 0.4061,
|
| 780 |
+
"step": 93
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 10.444444444444445,
|
| 784 |
+
"grad_norm": 110617.3671875,
|
| 785 |
+
"learning_rate": 1.0794671566375747e-07,
|
| 786 |
+
"loss": 0.3898,
|
| 787 |
+
"step": 94
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 10.555555555555555,
|
| 791 |
+
"grad_norm": 96216.1484375,
|
| 792 |
+
"learning_rate": 1.0909508497932935e-07,
|
| 793 |
+
"loss": 0.3988,
|
| 794 |
+
"step": 95
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 10.666666666666666,
|
| 798 |
+
"grad_norm": 188674.1875,
|
| 799 |
+
"learning_rate": 1.1024345429490124e-07,
|
| 800 |
+
"loss": 0.4165,
|
| 801 |
+
"step": 96
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 10.777777777777779,
|
| 805 |
+
"grad_norm": 136938.84375,
|
| 806 |
+
"learning_rate": 1.1139182361047312e-07,
|
| 807 |
+
"loss": 0.3956,
|
| 808 |
+
"step": 97
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 10.88888888888889,
|
| 812 |
+
"grad_norm": 144631.921875,
|
| 813 |
+
"learning_rate": 1.1254019292604501e-07,
|
| 814 |
+
"loss": 0.4044,
|
| 815 |
+
"step": 98
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 11.0,
|
| 819 |
+
"grad_norm": 152543.46875,
|
| 820 |
+
"learning_rate": 1.1368856224161691e-07,
|
| 821 |
+
"loss": 0.4142,
|
| 822 |
+
"step": 99
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 11.0,
|
| 826 |
+
"eval_accuracy": 0.8148828663298117,
|
| 827 |
+
"eval_f1": 0.8036062378167641,
|
| 828 |
+
"eval_loss": 0.4058963358402252,
|
| 829 |
+
"eval_precision": 0.814320987654321,
|
| 830 |
+
"eval_recall": 0.7931697931697932,
|
| 831 |
+
"eval_runtime": 4.9658,
|
| 832 |
+
"eval_samples_per_second": 876.79,
|
| 833 |
+
"eval_steps_per_second": 0.604,
|
| 834 |
+
"step": 99
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 11.11111111111111,
|
| 838 |
+
"grad_norm": 154895.921875,
|
| 839 |
+
"learning_rate": 1.148369315571888e-07,
|
| 840 |
+
"loss": 0.3944,
|
| 841 |
+
"step": 100
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 11.222222222222221,
|
| 845 |
+
"grad_norm": 135768.296875,
|
| 846 |
+
"learning_rate": 1.1598530087276068e-07,
|
| 847 |
+
"loss": 0.4035,
|
| 848 |
+
"step": 101
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 11.333333333333334,
|
| 852 |
+
"grad_norm": 159888.703125,
|
| 853 |
+
"learning_rate": 1.1713367018833256e-07,
|
| 854 |
+
"loss": 0.4137,
|
| 855 |
+
"step": 102
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 11.444444444444445,
|
| 859 |
+
"grad_norm": 90339.2109375,
|
| 860 |
+
"learning_rate": 1.1828203950390445e-07,
|
| 861 |
+
"loss": 0.389,
|
| 862 |
+
"step": 103
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 11.555555555555555,
|
| 866 |
+
"grad_norm": 104756.0078125,
|
| 867 |
+
"learning_rate": 1.1943040881947635e-07,
|
| 868 |
+
"loss": 0.3865,
|
| 869 |
+
"step": 104
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 11.666666666666666,
|
| 873 |
+
"grad_norm": 115121.1953125,
|
| 874 |
+
"learning_rate": 1.2057877813504826e-07,
|
| 875 |
+
"loss": 0.3901,
|
| 876 |
+
"step": 105
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 11.777777777777779,
|
| 880 |
+
"grad_norm": 295445.125,
|
| 881 |
+
"learning_rate": 1.2172714745062014e-07,
|
| 882 |
+
"loss": 0.4113,
|
| 883 |
+
"step": 106
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 11.88888888888889,
|
| 887 |
+
"grad_norm": 156321.390625,
|
| 888 |
+
"learning_rate": 1.2287551676619203e-07,
|
| 889 |
+
"loss": 0.3993,
|
| 890 |
+
"step": 107
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 12.0,
|
| 894 |
+
"grad_norm": 152413.09375,
|
| 895 |
+
"learning_rate": 1.2402388608176391e-07,
|
| 896 |
+
"loss": 0.4051,
|
| 897 |
+
"step": 108
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 12.0,
|
| 901 |
+
"eval_accuracy": 0.8144235186035829,
|
| 902 |
+
"eval_f1": 0.8031189083820662,
|
| 903 |
+
"eval_loss": 0.40576180815696716,
|
| 904 |
+
"eval_precision": 0.8138271604938272,
|
| 905 |
+
"eval_recall": 0.7926887926887927,
|
| 906 |
+
"eval_runtime": 4.3329,
|
| 907 |
+
"eval_samples_per_second": 1004.878,
|
| 908 |
+
"eval_steps_per_second": 0.692,
|
| 909 |
+
"step": 108
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 12.11111111111111,
|
| 913 |
+
"grad_norm": 59771.40625,
|
| 914 |
+
"learning_rate": 1.251722553973358e-07,
|
| 915 |
+
"loss": 0.3913,
|
| 916 |
+
"step": 109
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 12.222222222222221,
|
| 920 |
+
"grad_norm": 129880.8046875,
|
| 921 |
+
"learning_rate": 1.2632062471290768e-07,
|
| 922 |
+
"loss": 0.4113,
|
| 923 |
+
"step": 110
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 12.333333333333334,
|
| 927 |
+
"grad_norm": 127175.734375,
|
| 928 |
+
"learning_rate": 1.2746899402847957e-07,
|
| 929 |
+
"loss": 0.4003,
|
| 930 |
+
"step": 111
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 12.444444444444445,
|
| 934 |
+
"grad_norm": 96643.9140625,
|
| 935 |
+
"learning_rate": 1.2861736334405146e-07,
|
| 936 |
+
"loss": 0.4086,
|
| 937 |
+
"step": 112
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 12.555555555555555,
|
| 941 |
+
"grad_norm": 104994.5078125,
|
| 942 |
+
"learning_rate": 1.2976573265962334e-07,
|
| 943 |
+
"loss": 0.3986,
|
| 944 |
+
"step": 113
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 12.666666666666666,
|
| 948 |
+
"grad_norm": 54012.76953125,
|
| 949 |
+
"learning_rate": 1.3091410197519523e-07,
|
| 950 |
+
"loss": 0.3907,
|
| 951 |
+
"step": 114
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 12.777777777777779,
|
| 955 |
+
"grad_norm": 270469.09375,
|
| 956 |
+
"learning_rate": 1.320624712907671e-07,
|
| 957 |
+
"loss": 0.404,
|
| 958 |
+
"step": 115
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 12.88888888888889,
|
| 962 |
+
"grad_norm": 102465.6796875,
|
| 963 |
+
"learning_rate": 1.33210840606339e-07,
|
| 964 |
+
"loss": 0.4069,
|
| 965 |
+
"step": 116
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 13.0,
|
| 969 |
+
"grad_norm": 215163.171875,
|
| 970 |
+
"learning_rate": 1.3435920992191088e-07,
|
| 971 |
+
"loss": 0.3644,
|
| 972 |
+
"step": 117
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 13.0,
|
| 976 |
+
"eval_accuracy": 0.8146531924666973,
|
| 977 |
+
"eval_f1": 0.803218727139722,
|
| 978 |
+
"eval_loss": 0.4056652784347534,
|
| 979 |
+
"eval_precision": 0.814540059347181,
|
| 980 |
+
"eval_recall": 0.7922077922077922,
|
| 981 |
+
"eval_runtime": 3.9431,
|
| 982 |
+
"eval_samples_per_second": 1104.204,
|
| 983 |
+
"eval_steps_per_second": 0.761,
|
| 984 |
+
"step": 117
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 13.11111111111111,
|
| 988 |
+
"grad_norm": 101811.140625,
|
| 989 |
+
"learning_rate": 1.3550757923748277e-07,
|
| 990 |
+
"loss": 0.417,
|
| 991 |
+
"step": 118
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 13.222222222222221,
|
| 995 |
+
"grad_norm": 70086.578125,
|
| 996 |
+
"learning_rate": 1.3665594855305468e-07,
|
| 997 |
+
"loss": 0.392,
|
| 998 |
+
"step": 119
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 13.333333333333334,
|
| 1002 |
+
"grad_norm": 143038.09375,
|
| 1003 |
+
"learning_rate": 1.3780431786862656e-07,
|
| 1004 |
+
"loss": 0.4184,
|
| 1005 |
+
"step": 120
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 13.444444444444445,
|
| 1009 |
+
"grad_norm": 219920.734375,
|
| 1010 |
+
"learning_rate": 1.3895268718419845e-07,
|
| 1011 |
+
"loss": 0.3875,
|
| 1012 |
+
"step": 121
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 13.555555555555555,
|
| 1016 |
+
"grad_norm": 57099.09375,
|
| 1017 |
+
"learning_rate": 1.4010105649977033e-07,
|
| 1018 |
+
"loss": 0.4044,
|
| 1019 |
+
"step": 122
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 13.666666666666666,
|
| 1023 |
+
"grad_norm": 128275.0390625,
|
| 1024 |
+
"learning_rate": 1.4124942581534222e-07,
|
| 1025 |
+
"loss": 0.4059,
|
| 1026 |
+
"step": 123
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 13.777777777777779,
|
| 1030 |
+
"grad_norm": 235324.9375,
|
| 1031 |
+
"learning_rate": 1.423977951309141e-07,
|
| 1032 |
+
"loss": 0.3892,
|
| 1033 |
+
"step": 124
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 13.88888888888889,
|
| 1037 |
+
"grad_norm": 81514.5078125,
|
| 1038 |
+
"learning_rate": 1.43546164446486e-07,
|
| 1039 |
+
"loss": 0.3915,
|
| 1040 |
+
"step": 125
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 14.0,
|
| 1044 |
+
"grad_norm": 148182.484375,
|
| 1045 |
+
"learning_rate": 1.4469453376205787e-07,
|
| 1046 |
+
"loss": 0.4034,
|
| 1047 |
+
"step": 126
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 14.0,
|
| 1051 |
+
"eval_accuracy": 0.8146531924666973,
|
| 1052 |
+
"eval_f1": 0.8037928519328957,
|
| 1053 |
+
"eval_loss": 0.40557122230529785,
|
| 1054 |
+
"eval_precision": 0.8126843657817109,
|
| 1055 |
+
"eval_recall": 0.7950937950937951,
|
| 1056 |
+
"eval_runtime": 3.9983,
|
| 1057 |
+
"eval_samples_per_second": 1088.959,
|
| 1058 |
+
"eval_steps_per_second": 0.75,
|
| 1059 |
+
"step": 126
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 14.11111111111111,
|
| 1063 |
+
"grad_norm": 142549.59375,
|
| 1064 |
+
"learning_rate": 1.4584290307762979e-07,
|
| 1065 |
+
"loss": 0.4255,
|
| 1066 |
+
"step": 127
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 14.222222222222221,
|
| 1070 |
+
"grad_norm": 60380.79296875,
|
| 1071 |
+
"learning_rate": 1.4699127239320167e-07,
|
| 1072 |
+
"loss": 0.4101,
|
| 1073 |
+
"step": 128
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 14.333333333333334,
|
| 1077 |
+
"grad_norm": 58979.21484375,
|
| 1078 |
+
"learning_rate": 1.4813964170877356e-07,
|
| 1079 |
+
"loss": 0.3949,
|
| 1080 |
+
"step": 129
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 14.444444444444445,
|
| 1084 |
+
"grad_norm": 73793.7734375,
|
| 1085 |
+
"learning_rate": 1.4928801102434544e-07,
|
| 1086 |
+
"loss": 0.4167,
|
| 1087 |
+
"step": 130
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 14.555555555555555,
|
| 1091 |
+
"grad_norm": 108768.390625,
|
| 1092 |
+
"learning_rate": 1.5043638033991733e-07,
|
| 1093 |
+
"loss": 0.3798,
|
| 1094 |
+
"step": 131
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 14.666666666666666,
|
| 1098 |
+
"grad_norm": 79237.4765625,
|
| 1099 |
+
"learning_rate": 1.515847496554892e-07,
|
| 1100 |
+
"loss": 0.3917,
|
| 1101 |
+
"step": 132
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 14.777777777777779,
|
| 1105 |
+
"grad_norm": 286473.53125,
|
| 1106 |
+
"learning_rate": 1.527331189710611e-07,
|
| 1107 |
+
"loss": 0.3752,
|
| 1108 |
+
"step": 133
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 14.88888888888889,
|
| 1112 |
+
"grad_norm": 114565.21875,
|
| 1113 |
+
"learning_rate": 1.5388148828663298e-07,
|
| 1114 |
+
"loss": 0.3787,
|
| 1115 |
+
"step": 134
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 15.0,
|
| 1119 |
+
"grad_norm": 102380.3359375,
|
| 1120 |
+
"learning_rate": 1.5502985760220487e-07,
|
| 1121 |
+
"loss": 0.4162,
|
| 1122 |
+
"step": 135
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 15.0,
|
| 1126 |
+
"eval_accuracy": 0.8148828663298117,
|
| 1127 |
+
"eval_f1": 0.804368932038835,
|
| 1128 |
+
"eval_loss": 0.4054870307445526,
|
| 1129 |
+
"eval_precision": 0.8118569328760411,
|
| 1130 |
+
"eval_recall": 0.7970177970177971,
|
| 1131 |
+
"eval_runtime": 4.0281,
|
| 1132 |
+
"eval_samples_per_second": 1080.894,
|
| 1133 |
+
"eval_steps_per_second": 0.745,
|
| 1134 |
+
"step": 135
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 15.11111111111111,
|
| 1138 |
+
"grad_norm": 116745.8515625,
|
| 1139 |
+
"learning_rate": 1.5617822691777678e-07,
|
| 1140 |
+
"loss": 0.385,
|
| 1141 |
+
"step": 136
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 15.222222222222221,
|
| 1145 |
+
"grad_norm": 167835.21875,
|
| 1146 |
+
"learning_rate": 1.5732659623334866e-07,
|
| 1147 |
+
"loss": 0.4247,
|
| 1148 |
+
"step": 137
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 15.333333333333334,
|
| 1152 |
+
"grad_norm": 289577.75,
|
| 1153 |
+
"learning_rate": 1.5847496554892055e-07,
|
| 1154 |
+
"loss": 0.4071,
|
| 1155 |
+
"step": 138
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 15.444444444444445,
|
| 1159 |
+
"grad_norm": 99801.046875,
|
| 1160 |
+
"learning_rate": 1.5962333486449243e-07,
|
| 1161 |
+
"loss": 0.3914,
|
| 1162 |
+
"step": 139
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 15.555555555555555,
|
| 1166 |
+
"grad_norm": 73735.4609375,
|
| 1167 |
+
"learning_rate": 1.6077170418006432e-07,
|
| 1168 |
+
"loss": 0.4035,
|
| 1169 |
+
"step": 140
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 15.666666666666666,
|
| 1173 |
+
"grad_norm": 110641.0625,
|
| 1174 |
+
"learning_rate": 1.619200734956362e-07,
|
| 1175 |
+
"loss": 0.4105,
|
| 1176 |
+
"step": 141
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 15.777777777777779,
|
| 1180 |
+
"grad_norm": 187587.828125,
|
| 1181 |
+
"learning_rate": 1.630684428112081e-07,
|
| 1182 |
+
"loss": 0.4045,
|
| 1183 |
+
"step": 142
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 15.88888888888889,
|
| 1187 |
+
"grad_norm": 219092.484375,
|
| 1188 |
+
"learning_rate": 1.6421681212677997e-07,
|
| 1189 |
+
"loss": 0.3789,
|
| 1190 |
+
"step": 143
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 16.0,
|
| 1194 |
+
"grad_norm": 97330.953125,
|
| 1195 |
+
"learning_rate": 1.6536518144235186e-07,
|
| 1196 |
+
"loss": 0.4096,
|
| 1197 |
+
"step": 144
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 16.0,
|
| 1201 |
+
"eval_accuracy": 0.8139641708773542,
|
| 1202 |
+
"eval_f1": 0.8033025740650801,
|
| 1203 |
+
"eval_loss": 0.4053776264190674,
|
| 1204 |
+
"eval_precision": 0.8111819519372241,
|
| 1205 |
+
"eval_recall": 0.7955747955747956,
|
| 1206 |
+
"eval_runtime": 4.0824,
|
| 1207 |
+
"eval_samples_per_second": 1066.522,
|
| 1208 |
+
"eval_steps_per_second": 0.735,
|
| 1209 |
+
"step": 144
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 16.11111111111111,
|
| 1213 |
+
"grad_norm": 195703.734375,
|
| 1214 |
+
"learning_rate": 1.6651355075792374e-07,
|
| 1215 |
+
"loss": 0.389,
|
| 1216 |
+
"step": 145
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 16.22222222222222,
|
| 1220 |
+
"grad_norm": 63109.23046875,
|
| 1221 |
+
"learning_rate": 1.6766192007349563e-07,
|
| 1222 |
+
"loss": 0.4151,
|
| 1223 |
+
"step": 146
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 16.333333333333332,
|
| 1227 |
+
"grad_norm": 97057.3046875,
|
| 1228 |
+
"learning_rate": 1.6881028938906751e-07,
|
| 1229 |
+
"loss": 0.3928,
|
| 1230 |
+
"step": 147
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 16.444444444444443,
|
| 1234 |
+
"grad_norm": 225292.15625,
|
| 1235 |
+
"learning_rate": 1.699586587046394e-07,
|
| 1236 |
+
"loss": 0.3984,
|
| 1237 |
+
"step": 148
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 16.555555555555557,
|
| 1241 |
+
"grad_norm": 246126.90625,
|
| 1242 |
+
"learning_rate": 1.7110702802021129e-07,
|
| 1243 |
+
"loss": 0.4032,
|
| 1244 |
+
"step": 149
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 16.666666666666668,
|
| 1248 |
+
"grad_norm": 65869.1640625,
|
| 1249 |
+
"learning_rate": 1.722553973357832e-07,
|
| 1250 |
+
"loss": 0.3859,
|
| 1251 |
+
"step": 150
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 16.77777777777778,
|
| 1255 |
+
"grad_norm": 87636.9921875,
|
| 1256 |
+
"learning_rate": 1.7340376665135508e-07,
|
| 1257 |
+
"loss": 0.4103,
|
| 1258 |
+
"step": 151
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 16.88888888888889,
|
| 1262 |
+
"grad_norm": 285831.25,
|
| 1263 |
+
"learning_rate": 1.74552135966927e-07,
|
| 1264 |
+
"loss": 0.3969,
|
| 1265 |
+
"step": 152
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 17.0,
|
| 1269 |
+
"grad_norm": 154354.15625,
|
| 1270 |
+
"learning_rate": 1.7570050528249888e-07,
|
| 1271 |
+
"loss": 0.393,
|
| 1272 |
+
"step": 153
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 17.0,
|
| 1276 |
+
"eval_accuracy": 0.815112540192926,
|
| 1277 |
+
"eval_f1": 0.8036106367406685,
|
| 1278 |
+
"eval_loss": 0.4052949547767639,
|
| 1279 |
+
"eval_precision": 0.8153465346534653,
|
| 1280 |
+
"eval_recall": 0.7922077922077922,
|
| 1281 |
+
"eval_runtime": 4.1342,
|
| 1282 |
+
"eval_samples_per_second": 1053.169,
|
| 1283 |
+
"eval_steps_per_second": 0.726,
|
| 1284 |
+
"step": 153
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 17.11111111111111,
|
| 1288 |
+
"grad_norm": 49760.75,
|
| 1289 |
+
"learning_rate": 1.7684887459807076e-07,
|
| 1290 |
+
"loss": 0.3866,
|
| 1291 |
+
"step": 154
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 17.22222222222222,
|
| 1295 |
+
"grad_norm": 125308.21875,
|
| 1296 |
+
"learning_rate": 1.7799724391364265e-07,
|
| 1297 |
+
"loss": 0.4089,
|
| 1298 |
+
"step": 155
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 17.333333333333332,
|
| 1302 |
+
"grad_norm": 339002.8125,
|
| 1303 |
+
"learning_rate": 1.7914561322921453e-07,
|
| 1304 |
+
"loss": 0.4052,
|
| 1305 |
+
"step": 156
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 17.444444444444443,
|
| 1309 |
+
"grad_norm": 58521.89453125,
|
| 1310 |
+
"learning_rate": 1.8029398254478642e-07,
|
| 1311 |
+
"loss": 0.4052,
|
| 1312 |
+
"step": 157
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 17.555555555555557,
|
| 1316 |
+
"grad_norm": 150677.5,
|
| 1317 |
+
"learning_rate": 1.814423518603583e-07,
|
| 1318 |
+
"loss": 0.4085,
|
| 1319 |
+
"step": 158
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 17.666666666666668,
|
| 1323 |
+
"grad_norm": 204164.671875,
|
| 1324 |
+
"learning_rate": 1.825907211759302e-07,
|
| 1325 |
+
"loss": 0.4093,
|
| 1326 |
+
"step": 159
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 17.77777777777778,
|
| 1330 |
+
"grad_norm": 54231.87890625,
|
| 1331 |
+
"learning_rate": 1.8373909049150207e-07,
|
| 1332 |
+
"loss": 0.4017,
|
| 1333 |
+
"step": 160
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 17.88888888888889,
|
| 1337 |
+
"grad_norm": 104072.015625,
|
| 1338 |
+
"learning_rate": 1.8488745980707396e-07,
|
| 1339 |
+
"loss": 0.3916,
|
| 1340 |
+
"step": 161
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 18.0,
|
| 1344 |
+
"grad_norm": 271511.78125,
|
| 1345 |
+
"learning_rate": 1.8603582912264585e-07,
|
| 1346 |
+
"loss": 0.3912,
|
| 1347 |
+
"step": 162
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 18.0,
|
| 1351 |
+
"eval_accuracy": 0.8148828663298117,
|
| 1352 |
+
"eval_f1": 0.8037018996590356,
|
| 1353 |
+
"eval_loss": 0.4052158296108246,
|
| 1354 |
+
"eval_precision": 0.8140108534780464,
|
| 1355 |
+
"eval_recall": 0.7936507936507936,
|
| 1356 |
+
"eval_runtime": 4.1677,
|
| 1357 |
+
"eval_samples_per_second": 1044.694,
|
| 1358 |
+
"eval_steps_per_second": 0.72,
|
| 1359 |
+
"step": 162
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 18.11111111111111,
|
| 1363 |
+
"grad_norm": 95303.6796875,
|
| 1364 |
+
"learning_rate": 1.8718419843821773e-07,
|
| 1365 |
+
"loss": 0.4009,
|
| 1366 |
+
"step": 163
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 18.22222222222222,
|
| 1370 |
+
"grad_norm": 71401.6484375,
|
| 1371 |
+
"learning_rate": 1.8833256775378962e-07,
|
| 1372 |
+
"loss": 0.3918,
|
| 1373 |
+
"step": 164
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 18.333333333333332,
|
| 1377 |
+
"grad_norm": 107687.984375,
|
| 1378 |
+
"learning_rate": 1.894809370693615e-07,
|
| 1379 |
+
"loss": 0.4,
|
| 1380 |
+
"step": 165
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 18.444444444444443,
|
| 1384 |
+
"grad_norm": 102388.0703125,
|
| 1385 |
+
"learning_rate": 1.906293063849334e-07,
|
| 1386 |
+
"loss": 0.3796,
|
| 1387 |
+
"step": 166
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 18.555555555555557,
|
| 1391 |
+
"grad_norm": 112113.8671875,
|
| 1392 |
+
"learning_rate": 1.917776757005053e-07,
|
| 1393 |
+
"loss": 0.3953,
|
| 1394 |
+
"step": 167
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 18.666666666666668,
|
| 1398 |
+
"grad_norm": 113897.875,
|
| 1399 |
+
"learning_rate": 1.9292604501607718e-07,
|
| 1400 |
+
"loss": 0.4134,
|
| 1401 |
+
"step": 168
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 18.77777777777778,
|
| 1405 |
+
"grad_norm": 86523.8203125,
|
| 1406 |
+
"learning_rate": 1.9407441433164907e-07,
|
| 1407 |
+
"loss": 0.4177,
|
| 1408 |
+
"step": 169
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 18.88888888888889,
|
| 1412 |
+
"grad_norm": 138922.859375,
|
| 1413 |
+
"learning_rate": 1.9522278364722095e-07,
|
| 1414 |
+
"loss": 0.3887,
|
| 1415 |
+
"step": 170
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 19.0,
|
| 1419 |
+
"grad_norm": 121278.3671875,
|
| 1420 |
+
"learning_rate": 1.9637115296279284e-07,
|
| 1421 |
+
"loss": 0.3955,
|
| 1422 |
+
"step": 171
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 19.0,
|
| 1426 |
+
"eval_accuracy": 0.8144235186035829,
|
| 1427 |
+
"eval_f1": 0.8037882467217096,
|
| 1428 |
+
"eval_loss": 0.4051366150379181,
|
| 1429 |
+
"eval_precision": 0.8116723884256989,
|
| 1430 |
+
"eval_recall": 0.796055796055796,
|
| 1431 |
+
"eval_runtime": 4.1949,
|
| 1432 |
+
"eval_samples_per_second": 1037.921,
|
| 1433 |
+
"eval_steps_per_second": 0.715,
|
| 1434 |
+
"step": 171
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 19.11111111111111,
|
| 1438 |
+
"grad_norm": 61011.9609375,
|
| 1439 |
+
"learning_rate": 1.9751952227836475e-07,
|
| 1440 |
+
"loss": 0.3844,
|
| 1441 |
+
"step": 172
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 19.22222222222222,
|
| 1445 |
+
"grad_norm": 196222.4375,
|
| 1446 |
+
"learning_rate": 1.986678915939366e-07,
|
| 1447 |
+
"loss": 0.3977,
|
| 1448 |
+
"step": 173
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 19.333333333333332,
|
| 1452 |
+
"grad_norm": 91308.3203125,
|
| 1453 |
+
"learning_rate": 1.9981626090950852e-07,
|
| 1454 |
+
"loss": 0.4267,
|
| 1455 |
+
"step": 174
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 19.444444444444443,
|
| 1459 |
+
"grad_norm": 205705.5625,
|
| 1460 |
+
"learning_rate": 2.0096463022508038e-07,
|
| 1461 |
+
"loss": 0.4008,
|
| 1462 |
+
"step": 175
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 19.555555555555557,
|
| 1466 |
+
"grad_norm": 127608.0859375,
|
| 1467 |
+
"learning_rate": 2.021129995406523e-07,
|
| 1468 |
+
"loss": 0.4118,
|
| 1469 |
+
"step": 176
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 19.666666666666668,
|
| 1473 |
+
"grad_norm": 229128.703125,
|
| 1474 |
+
"learning_rate": 2.0326136885622415e-07,
|
| 1475 |
+
"loss": 0.3761,
|
| 1476 |
+
"step": 177
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 19.77777777777778,
|
| 1480 |
+
"grad_norm": 425327.84375,
|
| 1481 |
+
"learning_rate": 2.0440973817179606e-07,
|
| 1482 |
+
"loss": 0.4085,
|
| 1483 |
+
"step": 178
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 19.88888888888889,
|
| 1487 |
+
"grad_norm": 65994.8984375,
|
| 1488 |
+
"learning_rate": 2.0555810748736792e-07,
|
| 1489 |
+
"loss": 0.4058,
|
| 1490 |
+
"step": 179
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 20.0,
|
| 1494 |
+
"grad_norm": 138098.46875,
|
| 1495 |
+
"learning_rate": 2.0670647680293983e-07,
|
| 1496 |
+
"loss": 0.3767,
|
| 1497 |
+
"step": 180
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 20.0,
|
| 1501 |
+
"eval_accuracy": 0.8148828663298117,
|
| 1502 |
+
"eval_f1": 0.8046534173533689,
|
| 1503 |
+
"eval_loss": 0.405115008354187,
|
| 1504 |
+
"eval_precision": 0.810942843185149,
|
| 1505 |
+
"eval_recall": 0.7984607984607984,
|
| 1506 |
+
"eval_runtime": 4.2367,
|
| 1507 |
+
"eval_samples_per_second": 1027.696,
|
| 1508 |
+
"eval_steps_per_second": 0.708,
|
| 1509 |
+
"step": 180
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 20.11111111111111,
|
| 1513 |
+
"grad_norm": 74958.46875,
|
| 1514 |
+
"learning_rate": 2.0785484611851172e-07,
|
| 1515 |
+
"loss": 0.3977,
|
| 1516 |
+
"step": 181
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 20.22222222222222,
|
| 1520 |
+
"grad_norm": 82404.0234375,
|
| 1521 |
+
"learning_rate": 2.090032154340836e-07,
|
| 1522 |
+
"loss": 0.4025,
|
| 1523 |
+
"step": 182
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 20.333333333333332,
|
| 1527 |
+
"grad_norm": 67746.3046875,
|
| 1528 |
+
"learning_rate": 2.1015158474965549e-07,
|
| 1529 |
+
"loss": 0.3838,
|
| 1530 |
+
"step": 183
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 20.444444444444443,
|
| 1534 |
+
"grad_norm": 92680.1796875,
|
| 1535 |
+
"learning_rate": 2.112999540652274e-07,
|
| 1536 |
+
"loss": 0.3965,
|
| 1537 |
+
"step": 184
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 20.555555555555557,
|
| 1541 |
+
"grad_norm": 115058.3203125,
|
| 1542 |
+
"learning_rate": 2.1244832338079928e-07,
|
| 1543 |
+
"loss": 0.4089,
|
| 1544 |
+
"step": 185
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 20.666666666666668,
|
| 1548 |
+
"grad_norm": 128900.953125,
|
| 1549 |
+
"learning_rate": 2.1359669269637117e-07,
|
| 1550 |
+
"loss": 0.4074,
|
| 1551 |
+
"step": 186
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 20.77777777777778,
|
| 1555 |
+
"grad_norm": 149284.90625,
|
| 1556 |
+
"learning_rate": 2.1474506201194308e-07,
|
| 1557 |
+
"loss": 0.3924,
|
| 1558 |
+
"step": 187
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 20.88888888888889,
|
| 1562 |
+
"grad_norm": 84672.4921875,
|
| 1563 |
+
"learning_rate": 2.1589343132751494e-07,
|
| 1564 |
+
"loss": 0.3824,
|
| 1565 |
+
"step": 188
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 21.0,
|
| 1569 |
+
"grad_norm": 92032.4140625,
|
| 1570 |
+
"learning_rate": 2.1704180064308685e-07,
|
| 1571 |
+
"loss": 0.4127,
|
| 1572 |
+
"step": 189
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 21.0,
|
| 1576 |
+
"eval_accuracy": 0.8141938447404685,
|
| 1577 |
+
"eval_f1": 0.8038787878787879,
|
| 1578 |
+
"eval_loss": 0.40498441457748413,
|
| 1579 |
+
"eval_precision": 0.8103616813294232,
|
| 1580 |
+
"eval_recall": 0.7974987974987975,
|
| 1581 |
+
"eval_runtime": 4.3154,
|
| 1582 |
+
"eval_samples_per_second": 1008.943,
|
| 1583 |
+
"eval_steps_per_second": 0.695,
|
| 1584 |
+
"step": 189
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 21.11111111111111,
|
| 1588 |
+
"grad_norm": 93812.4296875,
|
| 1589 |
+
"learning_rate": 2.181901699586587e-07,
|
| 1590 |
+
"loss": 0.3964,
|
| 1591 |
+
"step": 190
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 21.22222222222222,
|
| 1595 |
+
"grad_norm": 77012.28125,
|
| 1596 |
+
"learning_rate": 2.1933853927423062e-07,
|
| 1597 |
+
"loss": 0.4109,
|
| 1598 |
+
"step": 191
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 21.333333333333332,
|
| 1602 |
+
"grad_norm": 177500.828125,
|
| 1603 |
+
"learning_rate": 2.2048690858980248e-07,
|
| 1604 |
+
"loss": 0.3954,
|
| 1605 |
+
"step": 192
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 21.444444444444443,
|
| 1609 |
+
"grad_norm": 149028.421875,
|
| 1610 |
+
"learning_rate": 2.216352779053744e-07,
|
| 1611 |
+
"loss": 0.4003,
|
| 1612 |
+
"step": 193
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 21.555555555555557,
|
| 1616 |
+
"grad_norm": 96760.171875,
|
| 1617 |
+
"learning_rate": 2.2278364722094625e-07,
|
| 1618 |
+
"loss": 0.3827,
|
| 1619 |
+
"step": 194
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 21.666666666666668,
|
| 1623 |
+
"grad_norm": 123675.0,
|
| 1624 |
+
"learning_rate": 2.2393201653651816e-07,
|
| 1625 |
+
"loss": 0.4047,
|
| 1626 |
+
"step": 195
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 21.77777777777778,
|
| 1630 |
+
"grad_norm": 112310.28125,
|
| 1631 |
+
"learning_rate": 2.2508038585209002e-07,
|
| 1632 |
+
"loss": 0.4085,
|
| 1633 |
+
"step": 196
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 21.88888888888889,
|
| 1637 |
+
"grad_norm": 88293.609375,
|
| 1638 |
+
"learning_rate": 2.2622875516766193e-07,
|
| 1639 |
+
"loss": 0.3874,
|
| 1640 |
+
"step": 197
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 22.0,
|
| 1644 |
+
"grad_norm": 191053.4375,
|
| 1645 |
+
"learning_rate": 2.2737712448323382e-07,
|
| 1646 |
+
"loss": 0.3947,
|
| 1647 |
+
"step": 198
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 22.0,
|
| 1651 |
+
"eval_accuracy": 0.8144235186035829,
|
| 1652 |
+
"eval_f1": 0.8037882467217096,
|
| 1653 |
+
"eval_loss": 0.4048587381839752,
|
| 1654 |
+
"eval_precision": 0.8116723884256989,
|
| 1655 |
+
"eval_recall": 0.796055796055796,
|
| 1656 |
+
"eval_runtime": 4.4003,
|
| 1657 |
+
"eval_samples_per_second": 989.472,
|
| 1658 |
+
"eval_steps_per_second": 0.682,
|
| 1659 |
+
"step": 198
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 22.11111111111111,
|
| 1663 |
+
"grad_norm": 170156.15625,
|
| 1664 |
+
"learning_rate": 2.285254937988057e-07,
|
| 1665 |
+
"loss": 0.3962,
|
| 1666 |
+
"step": 199
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 22.22222222222222,
|
| 1670 |
+
"grad_norm": 127575.1328125,
|
| 1671 |
+
"learning_rate": 2.296738631143776e-07,
|
| 1672 |
+
"loss": 0.4036,
|
| 1673 |
+
"step": 200
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 22.333333333333332,
|
| 1677 |
+
"grad_norm": 236836.921875,
|
| 1678 |
+
"learning_rate": 2.308222324299495e-07,
|
| 1679 |
+
"loss": 0.3971,
|
| 1680 |
+
"step": 201
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 22.444444444444443,
|
| 1684 |
+
"grad_norm": 77587.4453125,
|
| 1685 |
+
"learning_rate": 2.3197060174552136e-07,
|
| 1686 |
+
"loss": 0.4044,
|
| 1687 |
+
"step": 202
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 22.555555555555557,
|
| 1691 |
+
"grad_norm": 45806.18359375,
|
| 1692 |
+
"learning_rate": 2.3311897106109327e-07,
|
| 1693 |
+
"loss": 0.3936,
|
| 1694 |
+
"step": 203
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 22.666666666666668,
|
| 1698 |
+
"grad_norm": 97516.109375,
|
| 1699 |
+
"learning_rate": 2.3426734037666513e-07,
|
| 1700 |
+
"loss": 0.3851,
|
| 1701 |
+
"step": 204
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 22.77777777777778,
|
| 1705 |
+
"grad_norm": 174627.4375,
|
| 1706 |
+
"learning_rate": 2.3541570969223704e-07,
|
| 1707 |
+
"loss": 0.3926,
|
| 1708 |
+
"step": 205
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 22.88888888888889,
|
| 1712 |
+
"grad_norm": 196857.96875,
|
| 1713 |
+
"learning_rate": 2.365640790078089e-07,
|
| 1714 |
+
"loss": 0.4156,
|
| 1715 |
+
"step": 206
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 23.0,
|
| 1719 |
+
"grad_norm": 133552.34375,
|
| 1720 |
+
"learning_rate": 2.377124483233808e-07,
|
| 1721 |
+
"loss": 0.3795,
|
| 1722 |
+
"step": 207
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 23.0,
|
| 1726 |
+
"eval_accuracy": 0.8139641708773542,
|
| 1727 |
+
"eval_f1": 0.8034934497816594,
|
| 1728 |
+
"eval_loss": 0.4047852158546448,
|
| 1729 |
+
"eval_precision": 0.8105726872246696,
|
| 1730 |
+
"eval_recall": 0.7965367965367965,
|
| 1731 |
+
"eval_runtime": 4.8006,
|
| 1732 |
+
"eval_samples_per_second": 906.965,
|
| 1733 |
+
"eval_steps_per_second": 0.625,
|
| 1734 |
+
"step": 207
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 23.11111111111111,
|
| 1738 |
+
"grad_norm": 62008.26171875,
|
| 1739 |
+
"learning_rate": 2.388608176389527e-07,
|
| 1740 |
+
"loss": 0.3883,
|
| 1741 |
+
"step": 208
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 23.22222222222222,
|
| 1745 |
+
"grad_norm": 70936.859375,
|
| 1746 |
+
"learning_rate": 2.400091869545246e-07,
|
| 1747 |
+
"loss": 0.3965,
|
| 1748 |
+
"step": 209
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 23.333333333333332,
|
| 1752 |
+
"grad_norm": 156648.625,
|
| 1753 |
+
"learning_rate": 2.411575562700965e-07,
|
| 1754 |
+
"loss": 0.4019,
|
| 1755 |
+
"step": 210
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 23.444444444444443,
|
| 1759 |
+
"grad_norm": 112089.359375,
|
| 1760 |
+
"learning_rate": 2.423059255856684e-07,
|
| 1761 |
+
"loss": 0.4071,
|
| 1762 |
+
"step": 211
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 23.555555555555557,
|
| 1766 |
+
"grad_norm": 79252.5859375,
|
| 1767 |
+
"learning_rate": 2.434542949012403e-07,
|
| 1768 |
+
"loss": 0.4041,
|
| 1769 |
+
"step": 212
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 23.666666666666668,
|
| 1773 |
+
"grad_norm": 277307.15625,
|
| 1774 |
+
"learning_rate": 2.4460266421681215e-07,
|
| 1775 |
+
"loss": 0.3978,
|
| 1776 |
+
"step": 213
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 23.77777777777778,
|
| 1780 |
+
"grad_norm": 79551.9765625,
|
| 1781 |
+
"learning_rate": 2.4575103353238406e-07,
|
| 1782 |
+
"loss": 0.3963,
|
| 1783 |
+
"step": 214
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 23.88888888888889,
|
| 1787 |
+
"grad_norm": 88835.6484375,
|
| 1788 |
+
"learning_rate": 2.468994028479559e-07,
|
| 1789 |
+
"loss": 0.4016,
|
| 1790 |
+
"step": 215
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 24.0,
|
| 1794 |
+
"grad_norm": 357092.75,
|
| 1795 |
+
"learning_rate": 2.4804777216352783e-07,
|
| 1796 |
+
"loss": 0.3643,
|
| 1797 |
+
"step": 216
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 24.0,
|
| 1801 |
+
"eval_accuracy": 0.815112540192926,
|
| 1802 |
+
"eval_f1": 0.8052262279216066,
|
| 1803 |
+
"eval_loss": 0.4047646224498749,
|
| 1804 |
+
"eval_precision": 0.810126582278481,
|
| 1805 |
+
"eval_recall": 0.8003848003848004,
|
| 1806 |
+
"eval_runtime": 4.4662,
|
| 1807 |
+
"eval_samples_per_second": 974.88,
|
| 1808 |
+
"eval_steps_per_second": 0.672,
|
| 1809 |
+
"step": 216
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 24.11111111111111,
|
| 1813 |
+
"grad_norm": 68085.9453125,
|
| 1814 |
+
"learning_rate": 2.491961414790997e-07,
|
| 1815 |
+
"loss": 0.4226,
|
| 1816 |
+
"step": 217
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 24.22222222222222,
|
| 1820 |
+
"grad_norm": 100080.5703125,
|
| 1821 |
+
"learning_rate": 2.503445107946716e-07,
|
| 1822 |
+
"loss": 0.4022,
|
| 1823 |
+
"step": 218
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 24.333333333333332,
|
| 1827 |
+
"grad_norm": 44949.98828125,
|
| 1828 |
+
"learning_rate": 2.5149288011024346e-07,
|
| 1829 |
+
"loss": 0.3913,
|
| 1830 |
+
"step": 219
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 24.444444444444443,
|
| 1834 |
+
"grad_norm": 129354.359375,
|
| 1835 |
+
"learning_rate": 2.5264124942581537e-07,
|
| 1836 |
+
"loss": 0.4022,
|
| 1837 |
+
"step": 220
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 24.555555555555557,
|
| 1841 |
+
"grad_norm": 111662.390625,
|
| 1842 |
+
"learning_rate": 2.5378961874138723e-07,
|
| 1843 |
+
"loss": 0.3831,
|
| 1844 |
+
"step": 221
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 24.666666666666668,
|
| 1848 |
+
"grad_norm": 92629.7421875,
|
| 1849 |
+
"learning_rate": 2.5493798805695914e-07,
|
| 1850 |
+
"loss": 0.4062,
|
| 1851 |
+
"step": 222
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 24.77777777777778,
|
| 1855 |
+
"grad_norm": 65400.94140625,
|
| 1856 |
+
"learning_rate": 2.56086357372531e-07,
|
| 1857 |
+
"loss": 0.3813,
|
| 1858 |
+
"step": 223
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 24.88888888888889,
|
| 1862 |
+
"grad_norm": 110438.328125,
|
| 1863 |
+
"learning_rate": 2.572347266881029e-07,
|
| 1864 |
+
"loss": 0.3893,
|
| 1865 |
+
"step": 224
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 25.0,
|
| 1869 |
+
"grad_norm": 377074.5,
|
| 1870 |
+
"learning_rate": 2.5838309600367477e-07,
|
| 1871 |
+
"loss": 0.413,
|
| 1872 |
+
"step": 225
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 25.0,
|
| 1876 |
+
"eval_accuracy": 0.8144235186035829,
|
| 1877 |
+
"eval_f1": 0.8039786511402232,
|
| 1878 |
+
"eval_loss": 0.4046037197113037,
|
| 1879 |
+
"eval_precision": 0.811062163485071,
|
| 1880 |
+
"eval_recall": 0.7970177970177971,
|
| 1881 |
+
"eval_runtime": 4.5221,
|
| 1882 |
+
"eval_samples_per_second": 962.818,
|
| 1883 |
+
"eval_steps_per_second": 0.663,
|
| 1884 |
+
"step": 225
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 25.11111111111111,
|
| 1888 |
+
"grad_norm": 67365.15625,
|
| 1889 |
+
"learning_rate": 2.595314653192467e-07,
|
| 1890 |
+
"loss": 0.4213,
|
| 1891 |
+
"step": 226
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 25.22222222222222,
|
| 1895 |
+
"grad_norm": 130178.484375,
|
| 1896 |
+
"learning_rate": 2.6067983463481854e-07,
|
| 1897 |
+
"loss": 0.3799,
|
| 1898 |
+
"step": 227
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 25.333333333333332,
|
| 1902 |
+
"grad_norm": 38395.203125,
|
| 1903 |
+
"learning_rate": 2.6182820395039045e-07,
|
| 1904 |
+
"loss": 0.3885,
|
| 1905 |
+
"step": 228
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 25.444444444444443,
|
| 1909 |
+
"grad_norm": 82970.3359375,
|
| 1910 |
+
"learning_rate": 2.629765732659623e-07,
|
| 1911 |
+
"loss": 0.4155,
|
| 1912 |
+
"step": 229
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 25.555555555555557,
|
| 1916 |
+
"grad_norm": 178850.171875,
|
| 1917 |
+
"learning_rate": 2.641249425815342e-07,
|
| 1918 |
+
"loss": 0.4066,
|
| 1919 |
+
"step": 230
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 25.666666666666668,
|
| 1923 |
+
"grad_norm": 143966.3125,
|
| 1924 |
+
"learning_rate": 2.6527331189710613e-07,
|
| 1925 |
+
"loss": 0.3939,
|
| 1926 |
+
"step": 231
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 25.77777777777778,
|
| 1930 |
+
"grad_norm": 84673.3203125,
|
| 1931 |
+
"learning_rate": 2.66421681212678e-07,
|
| 1932 |
+
"loss": 0.3885,
|
| 1933 |
+
"step": 232
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 25.88888888888889,
|
| 1937 |
+
"grad_norm": 230422.8125,
|
| 1938 |
+
"learning_rate": 2.675700505282499e-07,
|
| 1939 |
+
"loss": 0.3977,
|
| 1940 |
+
"step": 233
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 26.0,
|
| 1944 |
+
"grad_norm": 273963.78125,
|
| 1945 |
+
"learning_rate": 2.6871841984382176e-07,
|
| 1946 |
+
"loss": 0.3802,
|
| 1947 |
+
"step": 234
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 26.0,
|
| 1951 |
+
"eval_accuracy": 0.8139641708773542,
|
| 1952 |
+
"eval_f1": 0.802823758519961,
|
| 1953 |
+
"eval_loss": 0.4044816493988037,
|
| 1954 |
+
"eval_precision": 0.8127156234598324,
|
| 1955 |
+
"eval_recall": 0.7931697931697932,
|
| 1956 |
+
"eval_runtime": 4.5407,
|
| 1957 |
+
"eval_samples_per_second": 958.89,
|
| 1958 |
+
"eval_steps_per_second": 0.661,
|
| 1959 |
+
"step": 234
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 26.11111111111111,
|
| 1963 |
+
"grad_norm": 191970.5,
|
| 1964 |
+
"learning_rate": 2.6986678915939367e-07,
|
| 1965 |
+
"loss": 0.3837,
|
| 1966 |
+
"step": 235
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 26.22222222222222,
|
| 1970 |
+
"grad_norm": 53745.9453125,
|
| 1971 |
+
"learning_rate": 2.7101515847496553e-07,
|
| 1972 |
+
"loss": 0.3891,
|
| 1973 |
+
"step": 236
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 26.333333333333332,
|
| 1977 |
+
"grad_norm": 119084.453125,
|
| 1978 |
+
"learning_rate": 2.7216352779053744e-07,
|
| 1979 |
+
"loss": 0.382,
|
| 1980 |
+
"step": 237
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 26.444444444444443,
|
| 1984 |
+
"grad_norm": 54897.70703125,
|
| 1985 |
+
"learning_rate": 2.7331189710610936e-07,
|
| 1986 |
+
"loss": 0.4096,
|
| 1987 |
+
"step": 238
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 26.555555555555557,
|
| 1991 |
+
"grad_norm": 61443.4375,
|
| 1992 |
+
"learning_rate": 2.744602664216812e-07,
|
| 1993 |
+
"loss": 0.4107,
|
| 1994 |
+
"step": 239
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 26.666666666666668,
|
| 1998 |
+
"grad_norm": 60475.47265625,
|
| 1999 |
+
"learning_rate": 2.756086357372531e-07,
|
| 2000 |
+
"loss": 0.4117,
|
| 2001 |
+
"step": 240
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 26.77777777777778,
|
| 2005 |
+
"grad_norm": 365218.84375,
|
| 2006 |
+
"learning_rate": 2.7675700505282504e-07,
|
| 2007 |
+
"loss": 0.4008,
|
| 2008 |
+
"step": 241
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 26.88888888888889,
|
| 2012 |
+
"grad_norm": 66929.9453125,
|
| 2013 |
+
"learning_rate": 2.779053743683969e-07,
|
| 2014 |
+
"loss": 0.3931,
|
| 2015 |
+
"step": 242
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 27.0,
|
| 2019 |
+
"grad_norm": 161915.15625,
|
| 2020 |
+
"learning_rate": 2.790537436839688e-07,
|
| 2021 |
+
"loss": 0.4034,
|
| 2022 |
+
"step": 243
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 27.0,
|
| 2026 |
+
"eval_accuracy": 0.8139641708773542,
|
| 2027 |
+
"eval_f1": 0.8033025740650801,
|
| 2028 |
+
"eval_loss": 0.40431469678878784,
|
| 2029 |
+
"eval_precision": 0.8111819519372241,
|
| 2030 |
+
"eval_recall": 0.7955747955747956,
|
| 2031 |
+
"eval_runtime": 4.5527,
|
| 2032 |
+
"eval_samples_per_second": 956.365,
|
| 2033 |
+
"eval_steps_per_second": 0.659,
|
| 2034 |
+
"step": 243
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 27.11111111111111,
|
| 2038 |
+
"grad_norm": 125099.7890625,
|
| 2039 |
+
"learning_rate": 2.8020211299954067e-07,
|
| 2040 |
+
"loss": 0.4098,
|
| 2041 |
+
"step": 244
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 27.22222222222222,
|
| 2045 |
+
"grad_norm": 129365.34375,
|
| 2046 |
+
"learning_rate": 2.813504823151126e-07,
|
| 2047 |
+
"loss": 0.384,
|
| 2048 |
+
"step": 245
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 27.333333333333332,
|
| 2052 |
+
"grad_norm": 76045.6953125,
|
| 2053 |
+
"learning_rate": 2.8249885163068444e-07,
|
| 2054 |
+
"loss": 0.3974,
|
| 2055 |
+
"step": 246
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 27.444444444444443,
|
| 2059 |
+
"grad_norm": 95814.2109375,
|
| 2060 |
+
"learning_rate": 2.8364722094625635e-07,
|
| 2061 |
+
"loss": 0.4016,
|
| 2062 |
+
"step": 247
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 27.555555555555557,
|
| 2066 |
+
"grad_norm": 131993.0,
|
| 2067 |
+
"learning_rate": 2.847955902618282e-07,
|
| 2068 |
+
"loss": 0.3872,
|
| 2069 |
+
"step": 248
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 27.666666666666668,
|
| 2073 |
+
"grad_norm": 151341.3125,
|
| 2074 |
+
"learning_rate": 2.859439595774001e-07,
|
| 2075 |
+
"loss": 0.4272,
|
| 2076 |
+
"step": 249
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 27.77777777777778,
|
| 2080 |
+
"grad_norm": 74917.6796875,
|
| 2081 |
+
"learning_rate": 2.87092328892972e-07,
|
| 2082 |
+
"loss": 0.3899,
|
| 2083 |
+
"step": 250
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 27.88888888888889,
|
| 2087 |
+
"grad_norm": 60322.203125,
|
| 2088 |
+
"learning_rate": 2.882406982085439e-07,
|
| 2089 |
+
"loss": 0.3942,
|
| 2090 |
+
"step": 251
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 28.0,
|
| 2094 |
+
"grad_norm": 174080.96875,
|
| 2095 |
+
"learning_rate": 2.8938906752411575e-07,
|
| 2096 |
+
"loss": 0.3687,
|
| 2097 |
+
"step": 252
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 28.0,
|
| 2101 |
+
"eval_accuracy": 0.8135048231511254,
|
| 2102 |
+
"eval_f1": 0.8031037827352085,
|
| 2103 |
+
"eval_loss": 0.40424075722694397,
|
| 2104 |
+
"eval_precision": 0.8097799511002445,
|
| 2105 |
+
"eval_recall": 0.7965367965367965,
|
| 2106 |
+
"eval_runtime": 4.5972,
|
| 2107 |
+
"eval_samples_per_second": 947.097,
|
| 2108 |
+
"eval_steps_per_second": 0.653,
|
| 2109 |
+
"step": 252
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 28.11111111111111,
|
| 2113 |
+
"grad_norm": 74456.140625,
|
| 2114 |
+
"learning_rate": 2.9053743683968766e-07,
|
| 2115 |
+
"loss": 0.4021,
|
| 2116 |
+
"step": 253
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 28.22222222222222,
|
| 2120 |
+
"grad_norm": 67092.3671875,
|
| 2121 |
+
"learning_rate": 2.9168580615525957e-07,
|
| 2122 |
+
"loss": 0.3933,
|
| 2123 |
+
"step": 254
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 28.333333333333332,
|
| 2127 |
+
"grad_norm": 53798.96875,
|
| 2128 |
+
"learning_rate": 2.9283417547083143e-07,
|
| 2129 |
+
"loss": 0.3938,
|
| 2130 |
+
"step": 255
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 28.444444444444443,
|
| 2134 |
+
"grad_norm": 82221.28125,
|
| 2135 |
+
"learning_rate": 2.9398254478640334e-07,
|
| 2136 |
+
"loss": 0.3966,
|
| 2137 |
+
"step": 256
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 28.555555555555557,
|
| 2141 |
+
"grad_norm": 99482.171875,
|
| 2142 |
+
"learning_rate": 2.951309141019752e-07,
|
| 2143 |
+
"loss": 0.4029,
|
| 2144 |
+
"step": 257
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 28.666666666666668,
|
| 2148 |
+
"grad_norm": 99727.8203125,
|
| 2149 |
+
"learning_rate": 2.962792834175471e-07,
|
| 2150 |
+
"loss": 0.3999,
|
| 2151 |
+
"step": 258
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 28.77777777777778,
|
| 2155 |
+
"grad_norm": 97492.7890625,
|
| 2156 |
+
"learning_rate": 2.9742765273311897e-07,
|
| 2157 |
+
"loss": 0.3852,
|
| 2158 |
+
"step": 259
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 28.88888888888889,
|
| 2162 |
+
"grad_norm": 142282.453125,
|
| 2163 |
+
"learning_rate": 2.985760220486909e-07,
|
| 2164 |
+
"loss": 0.4204,
|
| 2165 |
+
"step": 260
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 29.0,
|
| 2169 |
+
"grad_norm": 68024.1640625,
|
| 2170 |
+
"learning_rate": 2.9972439136426274e-07,
|
| 2171 |
+
"loss": 0.4046,
|
| 2172 |
+
"step": 261
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 29.0,
|
| 2176 |
+
"eval_accuracy": 0.8139641708773542,
|
| 2177 |
+
"eval_f1": 0.8035887487875849,
|
| 2178 |
+
"eval_loss": 0.40411704778671265,
|
| 2179 |
+
"eval_precision": 0.8102689486552567,
|
| 2180 |
+
"eval_recall": 0.7970177970177971,
|
| 2181 |
+
"eval_runtime": 4.6334,
|
| 2182 |
+
"eval_samples_per_second": 939.7,
|
| 2183 |
+
"eval_steps_per_second": 0.647,
|
| 2184 |
+
"step": 261
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 29.11111111111111,
|
| 2188 |
+
"grad_norm": 130296.609375,
|
| 2189 |
+
"learning_rate": 3.0087276067983465e-07,
|
| 2190 |
+
"loss": 0.4018,
|
| 2191 |
+
"step": 262
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 29.22222222222222,
|
| 2195 |
+
"grad_norm": 138968.453125,
|
| 2196 |
+
"learning_rate": 3.020211299954065e-07,
|
| 2197 |
+
"loss": 0.3902,
|
| 2198 |
+
"step": 263
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 29.333333333333332,
|
| 2202 |
+
"grad_norm": 179395.015625,
|
| 2203 |
+
"learning_rate": 3.031694993109784e-07,
|
| 2204 |
+
"loss": 0.4034,
|
| 2205 |
+
"step": 264
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 29.444444444444443,
|
| 2209 |
+
"grad_norm": 128004.46875,
|
| 2210 |
+
"learning_rate": 3.043178686265503e-07,
|
| 2211 |
+
"loss": 0.4124,
|
| 2212 |
+
"step": 265
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 29.555555555555557,
|
| 2216 |
+
"grad_norm": 52466.47265625,
|
| 2217 |
+
"learning_rate": 3.054662379421222e-07,
|
| 2218 |
+
"loss": 0.3917,
|
| 2219 |
+
"step": 266
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 29.666666666666668,
|
| 2223 |
+
"grad_norm": 71274.9765625,
|
| 2224 |
+
"learning_rate": 3.0661460725769405e-07,
|
| 2225 |
+
"loss": 0.4164,
|
| 2226 |
+
"step": 267
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 29.77777777777778,
|
| 2230 |
+
"grad_norm": 92328.015625,
|
| 2231 |
+
"learning_rate": 3.0776297657326596e-07,
|
| 2232 |
+
"loss": 0.3953,
|
| 2233 |
+
"step": 268
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 29.88888888888889,
|
| 2237 |
+
"grad_norm": 137363.90625,
|
| 2238 |
+
"learning_rate": 3.089113458888379e-07,
|
| 2239 |
+
"loss": 0.3776,
|
| 2240 |
+
"step": 269
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 30.0,
|
| 2244 |
+
"grad_norm": 103414.34375,
|
| 2245 |
+
"learning_rate": 3.1005971520440973e-07,
|
| 2246 |
+
"loss": 0.3923,
|
| 2247 |
+
"step": 270
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 30.0,
|
| 2251 |
+
"eval_accuracy": 0.8153422140560405,
|
| 2252 |
+
"eval_f1": 0.805703238279362,
|
| 2253 |
+
"eval_loss": 0.4040597975254059,
|
| 2254 |
+
"eval_precision": 0.8096163186012627,
|
| 2255 |
+
"eval_recall": 0.8018278018278018,
|
| 2256 |
+
"eval_runtime": 5.5187,
|
| 2257 |
+
"eval_samples_per_second": 788.953,
|
| 2258 |
+
"eval_steps_per_second": 0.544,
|
| 2259 |
+
"step": 270
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 30.11111111111111,
|
| 2263 |
+
"grad_norm": 61169.80078125,
|
| 2264 |
+
"learning_rate": 3.1120808451998164e-07,
|
| 2265 |
+
"loss": 0.4052,
|
| 2266 |
+
"step": 271
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 30.22222222222222,
|
| 2270 |
+
"grad_norm": 58287.30859375,
|
| 2271 |
+
"learning_rate": 3.1235645383555356e-07,
|
| 2272 |
+
"loss": 0.4189,
|
| 2273 |
+
"step": 272
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 30.333333333333332,
|
| 2277 |
+
"grad_norm": 130690.0078125,
|
| 2278 |
+
"learning_rate": 3.135048231511254e-07,
|
| 2279 |
+
"loss": 0.3915,
|
| 2280 |
+
"step": 273
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 30.444444444444443,
|
| 2284 |
+
"grad_norm": 78648.6953125,
|
| 2285 |
+
"learning_rate": 3.1465319246669733e-07,
|
| 2286 |
+
"loss": 0.393,
|
| 2287 |
+
"step": 274
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 30.555555555555557,
|
| 2291 |
+
"grad_norm": 208475.90625,
|
| 2292 |
+
"learning_rate": 3.158015617822692e-07,
|
| 2293 |
+
"loss": 0.3973,
|
| 2294 |
+
"step": 275
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 30.666666666666668,
|
| 2298 |
+
"grad_norm": 87458.609375,
|
| 2299 |
+
"learning_rate": 3.169499310978411e-07,
|
| 2300 |
+
"loss": 0.3835,
|
| 2301 |
+
"step": 276
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 30.77777777777778,
|
| 2305 |
+
"grad_norm": 77466.1640625,
|
| 2306 |
+
"learning_rate": 3.18098300413413e-07,
|
| 2307 |
+
"loss": 0.402,
|
| 2308 |
+
"step": 277
|
| 2309 |
+
},
|
| 2310 |
+
{
|
| 2311 |
+
"epoch": 30.88888888888889,
|
| 2312 |
+
"grad_norm": 359230.78125,
|
| 2313 |
+
"learning_rate": 3.1924666972898487e-07,
|
| 2314 |
+
"loss": 0.3877,
|
| 2315 |
+
"step": 278
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 31.0,
|
| 2319 |
+
"grad_norm": 199919.71875,
|
| 2320 |
+
"learning_rate": 3.203950390445568e-07,
|
| 2321 |
+
"loss": 0.4064,
|
| 2322 |
+
"step": 279
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 31.0,
|
| 2326 |
+
"eval_accuracy": 0.815112540192926,
|
| 2327 |
+
"eval_f1": 0.8052262279216066,
|
| 2328 |
+
"eval_loss": 0.4039256274700165,
|
| 2329 |
+
"eval_precision": 0.810126582278481,
|
| 2330 |
+
"eval_recall": 0.8003848003848004,
|
| 2331 |
+
"eval_runtime": 4.7082,
|
| 2332 |
+
"eval_samples_per_second": 924.771,
|
| 2333 |
+
"eval_steps_per_second": 0.637,
|
| 2334 |
+
"step": 279
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 31.11111111111111,
|
| 2338 |
+
"grad_norm": 73082.28125,
|
| 2339 |
+
"learning_rate": 3.2154340836012864e-07,
|
| 2340 |
+
"loss": 0.3972,
|
| 2341 |
+
"step": 280
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 31.22222222222222,
|
| 2345 |
+
"grad_norm": 77317.6640625,
|
| 2346 |
+
"learning_rate": 3.2269177767570055e-07,
|
| 2347 |
+
"loss": 0.4033,
|
| 2348 |
+
"step": 281
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 31.333333333333332,
|
| 2352 |
+
"grad_norm": 139156.5625,
|
| 2353 |
+
"learning_rate": 3.238401469912724e-07,
|
| 2354 |
+
"loss": 0.4,
|
| 2355 |
+
"step": 282
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 31.444444444444443,
|
| 2359 |
+
"grad_norm": 70580.640625,
|
| 2360 |
+
"learning_rate": 3.249885163068443e-07,
|
| 2361 |
+
"loss": 0.3989,
|
| 2362 |
+
"step": 283
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 31.555555555555557,
|
| 2366 |
+
"grad_norm": 261587.0625,
|
| 2367 |
+
"learning_rate": 3.261368856224162e-07,
|
| 2368 |
+
"loss": 0.3851,
|
| 2369 |
+
"step": 284
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 31.666666666666668,
|
| 2373 |
+
"grad_norm": 108566.5859375,
|
| 2374 |
+
"learning_rate": 3.272852549379881e-07,
|
| 2375 |
+
"loss": 0.4106,
|
| 2376 |
+
"step": 285
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 31.77777777777778,
|
| 2380 |
+
"grad_norm": 152543.453125,
|
| 2381 |
+
"learning_rate": 3.2843362425355995e-07,
|
| 2382 |
+
"loss": 0.4042,
|
| 2383 |
+
"step": 286
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 31.88888888888889,
|
| 2387 |
+
"grad_norm": 69462.53125,
|
| 2388 |
+
"learning_rate": 3.2958199356913186e-07,
|
| 2389 |
+
"loss": 0.3991,
|
| 2390 |
+
"step": 287
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 32.0,
|
| 2394 |
+
"grad_norm": 251283.9375,
|
| 2395 |
+
"learning_rate": 3.307303628847037e-07,
|
| 2396 |
+
"loss": 0.3866,
|
| 2397 |
+
"step": 288
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 32.0,
|
| 2401 |
+
"eval_accuracy": 0.8144235186035829,
|
| 2402 |
+
"eval_f1": 0.803022915650902,
|
| 2403 |
+
"eval_loss": 0.40380600094795227,
|
| 2404 |
+
"eval_precision": 0.8141374196737519,
|
| 2405 |
+
"eval_recall": 0.7922077922077922,
|
| 2406 |
+
"eval_runtime": 4.7491,
|
| 2407 |
+
"eval_samples_per_second": 916.813,
|
| 2408 |
+
"eval_steps_per_second": 0.632,
|
| 2409 |
+
"step": 288
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 32.111111111111114,
|
| 2413 |
+
"grad_norm": 137076.40625,
|
| 2414 |
+
"learning_rate": 3.3187873220027563e-07,
|
| 2415 |
+
"loss": 0.4108,
|
| 2416 |
+
"step": 289
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 32.22222222222222,
|
| 2420 |
+
"grad_norm": 93470.3203125,
|
| 2421 |
+
"learning_rate": 3.330271015158475e-07,
|
| 2422 |
+
"loss": 0.3773,
|
| 2423 |
+
"step": 290
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 32.333333333333336,
|
| 2427 |
+
"grad_norm": 112162.03125,
|
| 2428 |
+
"learning_rate": 3.341754708314194e-07,
|
| 2429 |
+
"loss": 0.4037,
|
| 2430 |
+
"step": 291
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 32.44444444444444,
|
| 2434 |
+
"grad_norm": 100211.0,
|
| 2435 |
+
"learning_rate": 3.3532384014699126e-07,
|
| 2436 |
+
"loss": 0.3674,
|
| 2437 |
+
"step": 292
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 32.55555555555556,
|
| 2441 |
+
"grad_norm": 151129.640625,
|
| 2442 |
+
"learning_rate": 3.3647220946256317e-07,
|
| 2443 |
+
"loss": 0.3921,
|
| 2444 |
+
"step": 293
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 32.666666666666664,
|
| 2448 |
+
"grad_norm": 70404.953125,
|
| 2449 |
+
"learning_rate": 3.3762057877813503e-07,
|
| 2450 |
+
"loss": 0.3959,
|
| 2451 |
+
"step": 294
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 32.77777777777778,
|
| 2455 |
+
"grad_norm": 63373.8515625,
|
| 2456 |
+
"learning_rate": 3.3876894809370694e-07,
|
| 2457 |
+
"loss": 0.4045,
|
| 2458 |
+
"step": 295
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 32.888888888888886,
|
| 2462 |
+
"grad_norm": 311372.9375,
|
| 2463 |
+
"learning_rate": 3.399173174092788e-07,
|
| 2464 |
+
"loss": 0.4015,
|
| 2465 |
+
"step": 296
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 33.0,
|
| 2469 |
+
"grad_norm": 202328.421875,
|
| 2470 |
+
"learning_rate": 3.410656867248507e-07,
|
| 2471 |
+
"loss": 0.3996,
|
| 2472 |
+
"step": 297
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 33.0,
|
| 2476 |
+
"eval_accuracy": 0.8158015617822691,
|
| 2477 |
+
"eval_f1": 0.8063737324963786,
|
| 2478 |
+
"eval_loss": 0.40382614731788635,
|
| 2479 |
+
"eval_precision": 0.8095007270964615,
|
| 2480 |
+
"eval_recall": 0.8032708032708032,
|
| 2481 |
+
"eval_runtime": 4.7927,
|
| 2482 |
+
"eval_samples_per_second": 908.458,
|
| 2483 |
+
"eval_steps_per_second": 0.626,
|
| 2484 |
+
"step": 297
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 33.111111111111114,
|
| 2488 |
+
"grad_norm": 133244.21875,
|
| 2489 |
+
"learning_rate": 3.4221405604042257e-07,
|
| 2490 |
+
"loss": 0.4102,
|
| 2491 |
+
"step": 298
|
| 2492 |
+
},
|
| 2493 |
+
{
|
| 2494 |
+
"epoch": 33.22222222222222,
|
| 2495 |
+
"grad_norm": 212747.078125,
|
| 2496 |
+
"learning_rate": 3.433624253559945e-07,
|
| 2497 |
+
"loss": 0.4053,
|
| 2498 |
+
"step": 299
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 33.333333333333336,
|
| 2502 |
+
"grad_norm": 158348.46875,
|
| 2503 |
+
"learning_rate": 3.445107946715664e-07,
|
| 2504 |
+
"loss": 0.3743,
|
| 2505 |
+
"step": 300
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 33.44444444444444,
|
| 2509 |
+
"grad_norm": 157230.78125,
|
| 2510 |
+
"learning_rate": 3.4565916398713825e-07,
|
| 2511 |
+
"loss": 0.4139,
|
| 2512 |
+
"step": 301
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 33.55555555555556,
|
| 2516 |
+
"grad_norm": 127684.703125,
|
| 2517 |
+
"learning_rate": 3.4680753330271016e-07,
|
| 2518 |
+
"loss": 0.406,
|
| 2519 |
+
"step": 302
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 33.666666666666664,
|
| 2523 |
+
"grad_norm": 103921.953125,
|
| 2524 |
+
"learning_rate": 3.479559026182821e-07,
|
| 2525 |
+
"loss": 0.3903,
|
| 2526 |
+
"step": 303
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 33.77777777777778,
|
| 2530 |
+
"grad_norm": 100353.8828125,
|
| 2531 |
+
"learning_rate": 3.49104271933854e-07,
|
| 2532 |
+
"loss": 0.3976,
|
| 2533 |
+
"step": 304
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 33.888888888888886,
|
| 2537 |
+
"grad_norm": 243526.859375,
|
| 2538 |
+
"learning_rate": 3.5025264124942585e-07,
|
| 2539 |
+
"loss": 0.4017,
|
| 2540 |
+
"step": 305
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 34.0,
|
| 2544 |
+
"grad_norm": 158435.421875,
|
| 2545 |
+
"learning_rate": 3.5140101056499776e-07,
|
| 2546 |
+
"loss": 0.3909,
|
| 2547 |
+
"step": 306
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 34.0,
|
| 2551 |
+
"eval_accuracy": 0.815112540192926,
|
| 2552 |
+
"eval_f1": 0.8052262279216066,
|
| 2553 |
+
"eval_loss": 0.40362977981567383,
|
| 2554 |
+
"eval_precision": 0.810126582278481,
|
| 2555 |
+
"eval_recall": 0.8003848003848004,
|
| 2556 |
+
"eval_runtime": 4.7975,
|
| 2557 |
+
"eval_samples_per_second": 907.551,
|
| 2558 |
+
"eval_steps_per_second": 0.625,
|
| 2559 |
+
"step": 306
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 34.111111111111114,
|
| 2563 |
+
"grad_norm": 100413.4609375,
|
| 2564 |
+
"learning_rate": 3.525493798805696e-07,
|
| 2565 |
+
"loss": 0.3933,
|
| 2566 |
+
"step": 307
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 34.22222222222222,
|
| 2570 |
+
"grad_norm": 150620.953125,
|
| 2571 |
+
"learning_rate": 3.5369774919614153e-07,
|
| 2572 |
+
"loss": 0.4032,
|
| 2573 |
+
"step": 308
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 34.333333333333336,
|
| 2577 |
+
"grad_norm": 78366.6328125,
|
| 2578 |
+
"learning_rate": 3.548461185117134e-07,
|
| 2579 |
+
"loss": 0.422,
|
| 2580 |
+
"step": 309
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 34.44444444444444,
|
| 2584 |
+
"grad_norm": 116747.1875,
|
| 2585 |
+
"learning_rate": 3.559944878272853e-07,
|
| 2586 |
+
"loss": 0.4149,
|
| 2587 |
+
"step": 310
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 34.55555555555556,
|
| 2591 |
+
"grad_norm": 127114.5078125,
|
| 2592 |
+
"learning_rate": 3.5714285714285716e-07,
|
| 2593 |
+
"loss": 0.3926,
|
| 2594 |
+
"step": 311
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 34.666666666666664,
|
| 2598 |
+
"grad_norm": 156098.375,
|
| 2599 |
+
"learning_rate": 3.5829122645842907e-07,
|
| 2600 |
+
"loss": 0.3817,
|
| 2601 |
+
"step": 312
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 34.77777777777778,
|
| 2605 |
+
"grad_norm": 83805.4765625,
|
| 2606 |
+
"learning_rate": 3.5943959577400093e-07,
|
| 2607 |
+
"loss": 0.385,
|
| 2608 |
+
"step": 313
|
| 2609 |
+
},
|
| 2610 |
+
{
|
| 2611 |
+
"epoch": 34.888888888888886,
|
| 2612 |
+
"grad_norm": 71467.9375,
|
| 2613 |
+
"learning_rate": 3.6058796508957284e-07,
|
| 2614 |
+
"loss": 0.3905,
|
| 2615 |
+
"step": 314
|
| 2616 |
+
},
|
| 2617 |
+
{
|
| 2618 |
+
"epoch": 35.0,
|
| 2619 |
+
"grad_norm": 124766.5625,
|
| 2620 |
+
"learning_rate": 3.617363344051447e-07,
|
| 2621 |
+
"loss": 0.3737,
|
| 2622 |
+
"step": 315
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 35.0,
|
| 2626 |
+
"eval_accuracy": 0.8139641708773542,
|
| 2627 |
+
"eval_f1": 0.802439024390244,
|
| 2628 |
+
"eval_loss": 0.40355560183525085,
|
| 2629 |
+
"eval_precision": 0.813953488372093,
|
| 2630 |
+
"eval_recall": 0.7912457912457912,
|
| 2631 |
+
"eval_runtime": 4.8433,
|
| 2632 |
+
"eval_samples_per_second": 898.973,
|
| 2633 |
+
"eval_steps_per_second": 0.619,
|
| 2634 |
+
"step": 315
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 35.111111111111114,
|
| 2638 |
+
"grad_norm": 144598.78125,
|
| 2639 |
+
"learning_rate": 3.628847037207166e-07,
|
| 2640 |
+
"loss": 0.3868,
|
| 2641 |
+
"step": 316
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 35.22222222222222,
|
| 2645 |
+
"grad_norm": 148750.015625,
|
| 2646 |
+
"learning_rate": 3.6403307303628847e-07,
|
| 2647 |
+
"loss": 0.3853,
|
| 2648 |
+
"step": 317
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 35.333333333333336,
|
| 2652 |
+
"grad_norm": 214067.65625,
|
| 2653 |
+
"learning_rate": 3.651814423518604e-07,
|
| 2654 |
+
"loss": 0.3995,
|
| 2655 |
+
"step": 318
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 35.44444444444444,
|
| 2659 |
+
"grad_norm": 116985.1484375,
|
| 2660 |
+
"learning_rate": 3.6632981166743224e-07,
|
| 2661 |
+
"loss": 0.3914,
|
| 2662 |
+
"step": 319
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 35.55555555555556,
|
| 2666 |
+
"grad_norm": 214030.203125,
|
| 2667 |
+
"learning_rate": 3.6747818098300415e-07,
|
| 2668 |
+
"loss": 0.4258,
|
| 2669 |
+
"step": 320
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 35.666666666666664,
|
| 2673 |
+
"grad_norm": 135642.34375,
|
| 2674 |
+
"learning_rate": 3.68626550298576e-07,
|
| 2675 |
+
"loss": 0.4242,
|
| 2676 |
+
"step": 321
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 35.77777777777778,
|
| 2680 |
+
"grad_norm": 312180.5,
|
| 2681 |
+
"learning_rate": 3.697749196141479e-07,
|
| 2682 |
+
"loss": 0.3784,
|
| 2683 |
+
"step": 322
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 35.888888888888886,
|
| 2687 |
+
"grad_norm": 98129.53125,
|
| 2688 |
+
"learning_rate": 3.7092328892971983e-07,
|
| 2689 |
+
"loss": 0.4089,
|
| 2690 |
+
"step": 323
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 36.0,
|
| 2694 |
+
"grad_norm": 576705.5625,
|
| 2695 |
+
"learning_rate": 3.720716582452917e-07,
|
| 2696 |
+
"loss": 0.3819,
|
| 2697 |
+
"step": 324
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 36.0,
|
| 2701 |
+
"eval_accuracy": 0.8146531924666973,
|
| 2702 |
+
"eval_f1": 0.804553160571567,
|
| 2703 |
+
"eval_loss": 0.40337246656417847,
|
| 2704 |
+
"eval_precision": 0.8102439024390244,
|
| 2705 |
+
"eval_recall": 0.798941798941799,
|
| 2706 |
+
"eval_runtime": 4.8778,
|
| 2707 |
+
"eval_samples_per_second": 892.62,
|
| 2708 |
+
"eval_steps_per_second": 0.615,
|
| 2709 |
+
"step": 324
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 36.111111111111114,
|
| 2713 |
+
"grad_norm": 60154.80859375,
|
| 2714 |
+
"learning_rate": 3.732200275608636e-07,
|
| 2715 |
+
"loss": 0.3863,
|
| 2716 |
+
"step": 325
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"epoch": 36.22222222222222,
|
| 2720 |
+
"grad_norm": 172251.390625,
|
| 2721 |
+
"learning_rate": 3.7436839687643546e-07,
|
| 2722 |
+
"loss": 0.4075,
|
| 2723 |
+
"step": 326
|
| 2724 |
+
},
|
| 2725 |
+
{
|
| 2726 |
+
"epoch": 36.333333333333336,
|
| 2727 |
+
"grad_norm": 138052.640625,
|
| 2728 |
+
"learning_rate": 3.7551676619200737e-07,
|
| 2729 |
+
"loss": 0.3931,
|
| 2730 |
+
"step": 327
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 36.44444444444444,
|
| 2734 |
+
"grad_norm": 90786.671875,
|
| 2735 |
+
"learning_rate": 3.7666513550757923e-07,
|
| 2736 |
+
"loss": 0.4035,
|
| 2737 |
+
"step": 328
|
| 2738 |
+
},
|
| 2739 |
+
{
|
| 2740 |
+
"epoch": 36.55555555555556,
|
| 2741 |
+
"grad_norm": 302897.8125,
|
| 2742 |
+
"learning_rate": 3.7781350482315114e-07,
|
| 2743 |
+
"loss": 0.3998,
|
| 2744 |
+
"step": 329
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 36.666666666666664,
|
| 2748 |
+
"grad_norm": 68080.4765625,
|
| 2749 |
+
"learning_rate": 3.78961874138723e-07,
|
| 2750 |
+
"loss": 0.3975,
|
| 2751 |
+
"step": 330
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 36.77777777777778,
|
| 2755 |
+
"grad_norm": 77021.703125,
|
| 2756 |
+
"learning_rate": 3.801102434542949e-07,
|
| 2757 |
+
"loss": 0.3808,
|
| 2758 |
+
"step": 331
|
| 2759 |
+
},
|
| 2760 |
+
{
|
| 2761 |
+
"epoch": 36.888888888888886,
|
| 2762 |
+
"grad_norm": 167419.8125,
|
| 2763 |
+
"learning_rate": 3.812586127698668e-07,
|
| 2764 |
+
"loss": 0.4127,
|
| 2765 |
+
"step": 332
|
| 2766 |
+
},
|
| 2767 |
+
{
|
| 2768 |
+
"epoch": 37.0,
|
| 2769 |
+
"grad_norm": 214871.546875,
|
| 2770 |
+
"learning_rate": 3.824069820854387e-07,
|
| 2771 |
+
"loss": 0.3832,
|
| 2772 |
+
"step": 333
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 37.0,
|
| 2776 |
+
"eval_accuracy": 0.8158015617822691,
|
| 2777 |
+
"eval_f1": 0.8067469879518072,
|
| 2778 |
+
"eval_loss": 0.40353327989578247,
|
| 2779 |
+
"eval_precision": 0.8083051665861902,
|
| 2780 |
+
"eval_recall": 0.8051948051948052,
|
| 2781 |
+
"eval_runtime": 5.3516,
|
| 2782 |
+
"eval_samples_per_second": 813.587,
|
| 2783 |
+
"eval_steps_per_second": 0.561,
|
| 2784 |
+
"step": 333
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 37.111111111111114,
|
| 2788 |
+
"grad_norm": 93671.765625,
|
| 2789 |
+
"learning_rate": 3.835553514010106e-07,
|
| 2790 |
+
"loss": 0.3822,
|
| 2791 |
+
"step": 334
|
| 2792 |
+
},
|
| 2793 |
+
{
|
| 2794 |
+
"epoch": 37.22222222222222,
|
| 2795 |
+
"grad_norm": 174590.234375,
|
| 2796 |
+
"learning_rate": 3.847037207165825e-07,
|
| 2797 |
+
"loss": 0.4036,
|
| 2798 |
+
"step": 335
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 37.333333333333336,
|
| 2802 |
+
"grad_norm": 103448.203125,
|
| 2803 |
+
"learning_rate": 3.8585209003215437e-07,
|
| 2804 |
+
"loss": 0.3843,
|
| 2805 |
+
"step": 336
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 37.44444444444444,
|
| 2809 |
+
"grad_norm": 85651.0546875,
|
| 2810 |
+
"learning_rate": 3.870004593477263e-07,
|
| 2811 |
+
"loss": 0.4012,
|
| 2812 |
+
"step": 337
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 37.55555555555556,
|
| 2816 |
+
"grad_norm": 104460.265625,
|
| 2817 |
+
"learning_rate": 3.8814882866329814e-07,
|
| 2818 |
+
"loss": 0.4183,
|
| 2819 |
+
"step": 338
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 37.666666666666664,
|
| 2823 |
+
"grad_norm": 88310.65625,
|
| 2824 |
+
"learning_rate": 3.8929719797887005e-07,
|
| 2825 |
+
"loss": 0.4154,
|
| 2826 |
+
"step": 339
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 37.77777777777778,
|
| 2830 |
+
"grad_norm": 103495.4375,
|
| 2831 |
+
"learning_rate": 3.904455672944419e-07,
|
| 2832 |
+
"loss": 0.387,
|
| 2833 |
+
"step": 340
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 37.888888888888886,
|
| 2837 |
+
"grad_norm": 103231.75,
|
| 2838 |
+
"learning_rate": 3.9159393661001376e-07,
|
| 2839 |
+
"loss": 0.3945,
|
| 2840 |
+
"step": 341
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 38.0,
|
| 2844 |
+
"grad_norm": 168905.34375,
|
| 2845 |
+
"learning_rate": 3.927423059255857e-07,
|
| 2846 |
+
"loss": 0.3614,
|
| 2847 |
+
"step": 342
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 38.0,
|
| 2851 |
+
"eval_accuracy": 0.8144235186035829,
|
| 2852 |
+
"eval_f1": 0.802638006839277,
|
| 2853 |
+
"eval_loss": 0.4033564627170563,
|
| 2854 |
+
"eval_precision": 0.8153846153846154,
|
| 2855 |
+
"eval_recall": 0.7902837902837903,
|
| 2856 |
+
"eval_runtime": 4.97,
|
| 2857 |
+
"eval_samples_per_second": 876.054,
|
| 2858 |
+
"eval_steps_per_second": 0.604,
|
| 2859 |
+
"step": 342
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 38.111111111111114,
|
| 2863 |
+
"grad_norm": 189091.21875,
|
| 2864 |
+
"learning_rate": 3.938906752411576e-07,
|
| 2865 |
+
"loss": 0.3928,
|
| 2866 |
+
"step": 343
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 38.22222222222222,
|
| 2870 |
+
"grad_norm": 92536.6875,
|
| 2871 |
+
"learning_rate": 3.950390445567295e-07,
|
| 2872 |
+
"loss": 0.405,
|
| 2873 |
+
"step": 344
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 38.333333333333336,
|
| 2877 |
+
"grad_norm": 218180.59375,
|
| 2878 |
+
"learning_rate": 3.961874138723013e-07,
|
| 2879 |
+
"loss": 0.3978,
|
| 2880 |
+
"step": 345
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 38.44444444444444,
|
| 2884 |
+
"grad_norm": 94875.9765625,
|
| 2885 |
+
"learning_rate": 3.973357831878732e-07,
|
| 2886 |
+
"loss": 0.4052,
|
| 2887 |
+
"step": 346
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 38.55555555555556,
|
| 2891 |
+
"grad_norm": 178712.5,
|
| 2892 |
+
"learning_rate": 3.9848415250344513e-07,
|
| 2893 |
+
"loss": 0.3984,
|
| 2894 |
+
"step": 347
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 38.666666666666664,
|
| 2898 |
+
"grad_norm": 173569.21875,
|
| 2899 |
+
"learning_rate": 3.9963252181901704e-07,
|
| 2900 |
+
"loss": 0.3975,
|
| 2901 |
+
"step": 348
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 38.77777777777778,
|
| 2905 |
+
"grad_norm": 251874.828125,
|
| 2906 |
+
"learning_rate": 4.0078089113458895e-07,
|
| 2907 |
+
"loss": 0.4289,
|
| 2908 |
+
"step": 349
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 38.888888888888886,
|
| 2912 |
+
"grad_norm": 160366.84375,
|
| 2913 |
+
"learning_rate": 4.0192926045016076e-07,
|
| 2914 |
+
"loss": 0.3857,
|
| 2915 |
+
"step": 350
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 39.0,
|
| 2919 |
+
"grad_norm": 186624.96875,
|
| 2920 |
+
"learning_rate": 4.0307762976573267e-07,
|
| 2921 |
+
"loss": 0.3968,
|
| 2922 |
+
"step": 351
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 39.0,
|
| 2926 |
+
"eval_accuracy": 0.8144235186035829,
|
| 2927 |
+
"eval_f1": 0.8035974720466699,
|
| 2928 |
+
"eval_loss": 0.4032462537288666,
|
| 2929 |
+
"eval_precision": 0.8122850122850123,
|
| 2930 |
+
"eval_recall": 0.7950937950937951,
|
| 2931 |
+
"eval_runtime": 4.9901,
|
| 2932 |
+
"eval_samples_per_second": 872.532,
|
| 2933 |
+
"eval_steps_per_second": 0.601,
|
| 2934 |
+
"step": 351
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 39.111111111111114,
|
| 2938 |
+
"grad_norm": 84011.5703125,
|
| 2939 |
+
"learning_rate": 4.042259990813046e-07,
|
| 2940 |
+
"loss": 0.3985,
|
| 2941 |
+
"step": 352
|
| 2942 |
+
},
|
| 2943 |
+
{
|
| 2944 |
+
"epoch": 39.22222222222222,
|
| 2945 |
+
"grad_norm": 45853.99609375,
|
| 2946 |
+
"learning_rate": 4.053743683968765e-07,
|
| 2947 |
+
"loss": 0.3948,
|
| 2948 |
+
"step": 353
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 39.333333333333336,
|
| 2952 |
+
"grad_norm": 71303.7109375,
|
| 2953 |
+
"learning_rate": 4.065227377124483e-07,
|
| 2954 |
+
"loss": 0.4027,
|
| 2955 |
+
"step": 354
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 39.44444444444444,
|
| 2959 |
+
"grad_norm": 116944.75,
|
| 2960 |
+
"learning_rate": 4.076711070280202e-07,
|
| 2961 |
+
"loss": 0.4141,
|
| 2962 |
+
"step": 355
|
| 2963 |
+
},
|
| 2964 |
+
{
|
| 2965 |
+
"epoch": 39.55555555555556,
|
| 2966 |
+
"grad_norm": 168545.234375,
|
| 2967 |
+
"learning_rate": 4.088194763435921e-07,
|
| 2968 |
+
"loss": 0.4021,
|
| 2969 |
+
"step": 356
|
| 2970 |
+
},
|
| 2971 |
+
{
|
| 2972 |
+
"epoch": 39.666666666666664,
|
| 2973 |
+
"grad_norm": 163324.765625,
|
| 2974 |
+
"learning_rate": 4.0996784565916403e-07,
|
| 2975 |
+
"loss": 0.3955,
|
| 2976 |
+
"step": 357
|
| 2977 |
+
},
|
| 2978 |
+
{
|
| 2979 |
+
"epoch": 39.77777777777778,
|
| 2980 |
+
"grad_norm": 77927.3046875,
|
| 2981 |
+
"learning_rate": 4.1111621497473584e-07,
|
| 2982 |
+
"loss": 0.3784,
|
| 2983 |
+
"step": 358
|
| 2984 |
+
},
|
| 2985 |
+
{
|
| 2986 |
+
"epoch": 39.888888888888886,
|
| 2987 |
+
"grad_norm": 190481.09375,
|
| 2988 |
+
"learning_rate": 4.1226458429030775e-07,
|
| 2989 |
+
"loss": 0.3918,
|
| 2990 |
+
"step": 359
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 40.0,
|
| 2994 |
+
"grad_norm": 201354.09375,
|
| 2995 |
+
"learning_rate": 4.1341295360587966e-07,
|
| 2996 |
+
"loss": 0.381,
|
| 2997 |
+
"step": 360
|
| 2998 |
+
},
|
| 2999 |
+
{
|
| 3000 |
+
"epoch": 40.0,
|
| 3001 |
+
"eval_accuracy": 0.8155718879191548,
|
| 3002 |
+
"eval_f1": 0.8058979937152526,
|
| 3003 |
+
"eval_loss": 0.4031665623188019,
|
| 3004 |
+
"eval_precision": 0.8100097181729835,
|
| 3005 |
+
"eval_recall": 0.8018278018278018,
|
| 3006 |
+
"eval_runtime": 5.0443,
|
| 3007 |
+
"eval_samples_per_second": 863.152,
|
| 3008 |
+
"eval_steps_per_second": 0.595,
|
| 3009 |
+
"step": 360
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 40.111111111111114,
|
| 3013 |
+
"grad_norm": 82638.6171875,
|
| 3014 |
+
"learning_rate": 4.145613229214516e-07,
|
| 3015 |
+
"loss": 0.4075,
|
| 3016 |
+
"step": 361
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 40.22222222222222,
|
| 3020 |
+
"grad_norm": 55737.82421875,
|
| 3021 |
+
"learning_rate": 4.1570969223702343e-07,
|
| 3022 |
+
"loss": 0.3949,
|
| 3023 |
+
"step": 362
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 40.333333333333336,
|
| 3027 |
+
"grad_norm": 156323.96875,
|
| 3028 |
+
"learning_rate": 4.1685806155259534e-07,
|
| 3029 |
+
"loss": 0.3846,
|
| 3030 |
+
"step": 363
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 40.44444444444444,
|
| 3034 |
+
"grad_norm": 157295.421875,
|
| 3035 |
+
"learning_rate": 4.180064308681672e-07,
|
| 3036 |
+
"loss": 0.3918,
|
| 3037 |
+
"step": 364
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 40.55555555555556,
|
| 3041 |
+
"grad_norm": 182490.3125,
|
| 3042 |
+
"learning_rate": 4.191548001837391e-07,
|
| 3043 |
+
"loss": 0.3988,
|
| 3044 |
+
"step": 365
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 40.666666666666664,
|
| 3048 |
+
"grad_norm": 214982.359375,
|
| 3049 |
+
"learning_rate": 4.2030316949931097e-07,
|
| 3050 |
+
"loss": 0.408,
|
| 3051 |
+
"step": 366
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 40.77777777777778,
|
| 3055 |
+
"grad_norm": 155003.234375,
|
| 3056 |
+
"learning_rate": 4.214515388148829e-07,
|
| 3057 |
+
"loss": 0.4078,
|
| 3058 |
+
"step": 367
|
| 3059 |
+
},
|
| 3060 |
+
{
|
| 3061 |
+
"epoch": 40.888888888888886,
|
| 3062 |
+
"grad_norm": 308192.21875,
|
| 3063 |
+
"learning_rate": 4.225999081304548e-07,
|
| 3064 |
+
"loss": 0.4031,
|
| 3065 |
+
"step": 368
|
| 3066 |
+
},
|
| 3067 |
+
{
|
| 3068 |
+
"epoch": 41.0,
|
| 3069 |
+
"grad_norm": 313317.875,
|
| 3070 |
+
"learning_rate": 4.237482774460267e-07,
|
| 3071 |
+
"loss": 0.4007,
|
| 3072 |
+
"step": 369
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 41.0,
|
| 3076 |
+
"eval_accuracy": 0.8139641708773542,
|
| 3077 |
+
"eval_f1": 0.8029197080291971,
|
| 3078 |
+
"eval_loss": 0.40287867188453674,
|
| 3079 |
+
"eval_precision": 0.8124076809453471,
|
| 3080 |
+
"eval_recall": 0.7936507936507936,
|
| 3081 |
+
"eval_runtime": 5.0875,
|
| 3082 |
+
"eval_samples_per_second": 855.818,
|
| 3083 |
+
"eval_steps_per_second": 0.59,
|
| 3084 |
+
"step": 369
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 41.111111111111114,
|
| 3088 |
+
"grad_norm": 116586.2890625,
|
| 3089 |
+
"learning_rate": 4.2489664676159857e-07,
|
| 3090 |
+
"loss": 0.3991,
|
| 3091 |
+
"step": 370
|
| 3092 |
+
},
|
| 3093 |
+
{
|
| 3094 |
+
"epoch": 41.22222222222222,
|
| 3095 |
+
"grad_norm": 55302.421875,
|
| 3096 |
+
"learning_rate": 4.260450160771704e-07,
|
| 3097 |
+
"loss": 0.3955,
|
| 3098 |
+
"step": 371
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 41.333333333333336,
|
| 3102 |
+
"grad_norm": 181413.859375,
|
| 3103 |
+
"learning_rate": 4.2719338539274234e-07,
|
| 3104 |
+
"loss": 0.3992,
|
| 3105 |
+
"step": 372
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 41.44444444444444,
|
| 3109 |
+
"grad_norm": 134792.09375,
|
| 3110 |
+
"learning_rate": 4.2834175470831425e-07,
|
| 3111 |
+
"loss": 0.4019,
|
| 3112 |
+
"step": 373
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 41.55555555555556,
|
| 3116 |
+
"grad_norm": 86486.609375,
|
| 3117 |
+
"learning_rate": 4.2949012402388616e-07,
|
| 3118 |
+
"loss": 0.4,
|
| 3119 |
+
"step": 374
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 41.666666666666664,
|
| 3123 |
+
"grad_norm": 69885.5703125,
|
| 3124 |
+
"learning_rate": 4.3063849333945797e-07,
|
| 3125 |
+
"loss": 0.3926,
|
| 3126 |
+
"step": 375
|
| 3127 |
+
},
|
| 3128 |
+
{
|
| 3129 |
+
"epoch": 41.77777777777778,
|
| 3130 |
+
"grad_norm": 65585.46875,
|
| 3131 |
+
"learning_rate": 4.317868626550299e-07,
|
| 3132 |
+
"loss": 0.4004,
|
| 3133 |
+
"step": 376
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 41.888888888888886,
|
| 3137 |
+
"grad_norm": 195587.671875,
|
| 3138 |
+
"learning_rate": 4.329352319706018e-07,
|
| 3139 |
+
"loss": 0.3899,
|
| 3140 |
+
"step": 377
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 42.0,
|
| 3144 |
+
"grad_norm": 138780.78125,
|
| 3145 |
+
"learning_rate": 4.340836012861737e-07,
|
| 3146 |
+
"loss": 0.3884,
|
| 3147 |
+
"step": 378
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 42.0,
|
| 3151 |
+
"eval_accuracy": 0.8139641708773542,
|
| 3152 |
+
"eval_f1": 0.8035887487875849,
|
| 3153 |
+
"eval_loss": 0.40278443694114685,
|
| 3154 |
+
"eval_precision": 0.8102689486552567,
|
| 3155 |
+
"eval_recall": 0.7970177970177971,
|
| 3156 |
+
"eval_runtime": 5.1036,
|
| 3157 |
+
"eval_samples_per_second": 853.128,
|
| 3158 |
+
"eval_steps_per_second": 0.588,
|
| 3159 |
+
"step": 378
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 42.111111111111114,
|
| 3163 |
+
"grad_norm": 139797.984375,
|
| 3164 |
+
"learning_rate": 4.352319706017455e-07,
|
| 3165 |
+
"loss": 0.4003,
|
| 3166 |
+
"step": 379
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 42.22222222222222,
|
| 3170 |
+
"grad_norm": 167927.25,
|
| 3171 |
+
"learning_rate": 4.363803399173174e-07,
|
| 3172 |
+
"loss": 0.3784,
|
| 3173 |
+
"step": 380
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 42.333333333333336,
|
| 3177 |
+
"grad_norm": 64973.1328125,
|
| 3178 |
+
"learning_rate": 4.3752870923288933e-07,
|
| 3179 |
+
"loss": 0.3957,
|
| 3180 |
+
"step": 381
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 42.44444444444444,
|
| 3184 |
+
"grad_norm": 177692.0,
|
| 3185 |
+
"learning_rate": 4.3867707854846124e-07,
|
| 3186 |
+
"loss": 0.4117,
|
| 3187 |
+
"step": 382
|
| 3188 |
+
},
|
| 3189 |
+
{
|
| 3190 |
+
"epoch": 42.55555555555556,
|
| 3191 |
+
"grad_norm": 100954.90625,
|
| 3192 |
+
"learning_rate": 4.3982544786403305e-07,
|
| 3193 |
+
"loss": 0.4032,
|
| 3194 |
+
"step": 383
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 42.666666666666664,
|
| 3198 |
+
"grad_norm": 72637.5859375,
|
| 3199 |
+
"learning_rate": 4.4097381717960496e-07,
|
| 3200 |
+
"loss": 0.404,
|
| 3201 |
+
"step": 384
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 42.77777777777778,
|
| 3205 |
+
"grad_norm": 201453.28125,
|
| 3206 |
+
"learning_rate": 4.4212218649517687e-07,
|
| 3207 |
+
"loss": 0.3859,
|
| 3208 |
+
"step": 385
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 42.888888888888886,
|
| 3212 |
+
"grad_norm": 219890.0,
|
| 3213 |
+
"learning_rate": 4.432705558107488e-07,
|
| 3214 |
+
"loss": 0.3853,
|
| 3215 |
+
"step": 386
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 43.0,
|
| 3219 |
+
"grad_norm": 101513.015625,
|
| 3220 |
+
"learning_rate": 4.444189251263206e-07,
|
| 3221 |
+
"loss": 0.4054,
|
| 3222 |
+
"step": 387
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 43.0,
|
| 3226 |
+
"eval_accuracy": 0.8144235186035829,
|
| 3227 |
+
"eval_f1": 0.8039786511402232,
|
| 3228 |
+
"eval_loss": 0.40269890427589417,
|
| 3229 |
+
"eval_precision": 0.811062163485071,
|
| 3230 |
+
"eval_recall": 0.7970177970177971,
|
| 3231 |
+
"eval_runtime": 5.1477,
|
| 3232 |
+
"eval_samples_per_second": 845.813,
|
| 3233 |
+
"eval_steps_per_second": 0.583,
|
| 3234 |
+
"step": 387
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 43.111111111111114,
|
| 3238 |
+
"grad_norm": 225872.515625,
|
| 3239 |
+
"learning_rate": 4.455672944418925e-07,
|
| 3240 |
+
"loss": 0.3915,
|
| 3241 |
+
"step": 388
|
| 3242 |
+
},
|
| 3243 |
+
{
|
| 3244 |
+
"epoch": 43.22222222222222,
|
| 3245 |
+
"grad_norm": 70304.9609375,
|
| 3246 |
+
"learning_rate": 4.467156637574644e-07,
|
| 3247 |
+
"loss": 0.4047,
|
| 3248 |
+
"step": 389
|
| 3249 |
+
},
|
| 3250 |
+
{
|
| 3251 |
+
"epoch": 43.333333333333336,
|
| 3252 |
+
"grad_norm": 95358.1015625,
|
| 3253 |
+
"learning_rate": 4.478640330730363e-07,
|
| 3254 |
+
"loss": 0.398,
|
| 3255 |
+
"step": 390
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 43.44444444444444,
|
| 3259 |
+
"grad_norm": 71124.046875,
|
| 3260 |
+
"learning_rate": 4.490124023886082e-07,
|
| 3261 |
+
"loss": 0.4073,
|
| 3262 |
+
"step": 391
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 43.55555555555556,
|
| 3266 |
+
"grad_norm": 156171.65625,
|
| 3267 |
+
"learning_rate": 4.5016077170418004e-07,
|
| 3268 |
+
"loss": 0.391,
|
| 3269 |
+
"step": 392
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 43.666666666666664,
|
| 3273 |
+
"grad_norm": 225986.453125,
|
| 3274 |
+
"learning_rate": 4.5130914101975195e-07,
|
| 3275 |
+
"loss": 0.386,
|
| 3276 |
+
"step": 393
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 43.77777777777778,
|
| 3280 |
+
"grad_norm": 129069.015625,
|
| 3281 |
+
"learning_rate": 4.5245751033532386e-07,
|
| 3282 |
+
"loss": 0.3941,
|
| 3283 |
+
"step": 394
|
| 3284 |
+
},
|
| 3285 |
+
{
|
| 3286 |
+
"epoch": 43.888888888888886,
|
| 3287 |
+
"grad_norm": 96861.7890625,
|
| 3288 |
+
"learning_rate": 4.536058796508958e-07,
|
| 3289 |
+
"loss": 0.3997,
|
| 3290 |
+
"step": 395
|
| 3291 |
+
},
|
| 3292 |
+
{
|
| 3293 |
+
"epoch": 44.0,
|
| 3294 |
+
"grad_norm": 248732.328125,
|
| 3295 |
+
"learning_rate": 4.5475424896646763e-07,
|
| 3296 |
+
"loss": 0.3731,
|
| 3297 |
+
"step": 396
|
| 3298 |
+
},
|
| 3299 |
+
{
|
| 3300 |
+
"epoch": 44.0,
|
| 3301 |
+
"eval_accuracy": 0.8148828663298117,
|
| 3302 |
+
"eval_f1": 0.8045586808923375,
|
| 3303 |
+
"eval_loss": 0.40265989303588867,
|
| 3304 |
+
"eval_precision": 0.8112469437652812,
|
| 3305 |
+
"eval_recall": 0.797979797979798,
|
| 3306 |
+
"eval_runtime": 5.1966,
|
| 3307 |
+
"eval_samples_per_second": 837.848,
|
| 3308 |
+
"eval_steps_per_second": 0.577,
|
| 3309 |
+
"step": 396
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 44.111111111111114,
|
| 3313 |
+
"grad_norm": 62515.44140625,
|
| 3314 |
+
"learning_rate": 4.5590261828203955e-07,
|
| 3315 |
+
"loss": 0.3753,
|
| 3316 |
+
"step": 397
|
| 3317 |
+
},
|
| 3318 |
+
{
|
| 3319 |
+
"epoch": 44.22222222222222,
|
| 3320 |
+
"grad_norm": 144591.4375,
|
| 3321 |
+
"learning_rate": 4.570509875976114e-07,
|
| 3322 |
+
"loss": 0.3881,
|
| 3323 |
+
"step": 398
|
| 3324 |
+
},
|
| 3325 |
+
{
|
| 3326 |
+
"epoch": 44.333333333333336,
|
| 3327 |
+
"grad_norm": 101405.015625,
|
| 3328 |
+
"learning_rate": 4.581993569131833e-07,
|
| 3329 |
+
"loss": 0.3969,
|
| 3330 |
+
"step": 399
|
| 3331 |
+
},
|
| 3332 |
+
{
|
| 3333 |
+
"epoch": 44.44444444444444,
|
| 3334 |
+
"grad_norm": 119058.875,
|
| 3335 |
+
"learning_rate": 4.593477262287552e-07,
|
| 3336 |
+
"loss": 0.4085,
|
| 3337 |
+
"step": 400
|
| 3338 |
+
},
|
| 3339 |
+
{
|
| 3340 |
+
"epoch": 44.55555555555556,
|
| 3341 |
+
"grad_norm": 112507.6484375,
|
| 3342 |
+
"learning_rate": 4.604960955443271e-07,
|
| 3343 |
+
"loss": 0.3891,
|
| 3344 |
+
"step": 401
|
| 3345 |
+
},
|
| 3346 |
+
{
|
| 3347 |
+
"epoch": 44.666666666666664,
|
| 3348 |
+
"grad_norm": 56403.96875,
|
| 3349 |
+
"learning_rate": 4.61644464859899e-07,
|
| 3350 |
+
"loss": 0.3913,
|
| 3351 |
+
"step": 402
|
| 3352 |
+
},
|
| 3353 |
+
{
|
| 3354 |
+
"epoch": 44.77777777777778,
|
| 3355 |
+
"grad_norm": 289250.875,
|
| 3356 |
+
"learning_rate": 4.627928341754709e-07,
|
| 3357 |
+
"loss": 0.3999,
|
| 3358 |
+
"step": 403
|
| 3359 |
+
},
|
| 3360 |
+
{
|
| 3361 |
+
"epoch": 44.888888888888886,
|
| 3362 |
+
"grad_norm": 69331.8984375,
|
| 3363 |
+
"learning_rate": 4.639412034910427e-07,
|
| 3364 |
+
"loss": 0.4067,
|
| 3365 |
+
"step": 404
|
| 3366 |
+
},
|
| 3367 |
+
{
|
| 3368 |
+
"epoch": 45.0,
|
| 3369 |
+
"grad_norm": 97999.0234375,
|
| 3370 |
+
"learning_rate": 4.650895728066146e-07,
|
| 3371 |
+
"loss": 0.4046,
|
| 3372 |
+
"step": 405
|
| 3373 |
+
},
|
| 3374 |
+
{
|
| 3375 |
+
"epoch": 45.0,
|
| 3376 |
+
"eval_accuracy": 0.8141938447404685,
|
| 3377 |
+
"eval_f1": 0.8031151131662205,
|
| 3378 |
+
"eval_loss": 0.40241068601608276,
|
| 3379 |
+
"eval_precision": 0.812807881773399,
|
| 3380 |
+
"eval_recall": 0.7936507936507936,
|
| 3381 |
+
"eval_runtime": 5.7031,
|
| 3382 |
+
"eval_samples_per_second": 763.441,
|
| 3383 |
+
"eval_steps_per_second": 0.526,
|
| 3384 |
+
"step": 405
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 45.111111111111114,
|
| 3388 |
+
"grad_norm": 301622.4375,
|
| 3389 |
+
"learning_rate": 4.6623794212218654e-07,
|
| 3390 |
+
"loss": 0.3995,
|
| 3391 |
+
"step": 406
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 45.22222222222222,
|
| 3395 |
+
"grad_norm": 299197.59375,
|
| 3396 |
+
"learning_rate": 4.6738631143775845e-07,
|
| 3397 |
+
"loss": 0.3758,
|
| 3398 |
+
"step": 407
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 45.333333333333336,
|
| 3402 |
+
"grad_norm": 171744.8125,
|
| 3403 |
+
"learning_rate": 4.6853468075333026e-07,
|
| 3404 |
+
"loss": 0.3939,
|
| 3405 |
+
"step": 408
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 45.44444444444444,
|
| 3409 |
+
"grad_norm": 135151.921875,
|
| 3410 |
+
"learning_rate": 4.6968305006890217e-07,
|
| 3411 |
+
"loss": 0.4104,
|
| 3412 |
+
"step": 409
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 45.55555555555556,
|
| 3416 |
+
"grad_norm": 152822.78125,
|
| 3417 |
+
"learning_rate": 4.708314193844741e-07,
|
| 3418 |
+
"loss": 0.4051,
|
| 3419 |
+
"step": 410
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 45.666666666666664,
|
| 3423 |
+
"grad_norm": 97492.375,
|
| 3424 |
+
"learning_rate": 4.71979788700046e-07,
|
| 3425 |
+
"loss": 0.39,
|
| 3426 |
+
"step": 411
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 45.77777777777778,
|
| 3430 |
+
"grad_norm": 184670.78125,
|
| 3431 |
+
"learning_rate": 4.731281580156178e-07,
|
| 3432 |
+
"loss": 0.3944,
|
| 3433 |
+
"step": 412
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 45.888888888888886,
|
| 3437 |
+
"grad_norm": 78265.125,
|
| 3438 |
+
"learning_rate": 4.742765273311897e-07,
|
| 3439 |
+
"loss": 0.4159,
|
| 3440 |
+
"step": 413
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 46.0,
|
| 3444 |
+
"grad_norm": 198317.984375,
|
| 3445 |
+
"learning_rate": 4.754248966467616e-07,
|
| 3446 |
+
"loss": 0.3959,
|
| 3447 |
+
"step": 414
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 46.0,
|
| 3451 |
+
"eval_accuracy": 0.8148828663298117,
|
| 3452 |
+
"eval_f1": 0.804368932038835,
|
| 3453 |
+
"eval_loss": 0.4022214710712433,
|
| 3454 |
+
"eval_precision": 0.8118569328760411,
|
| 3455 |
+
"eval_recall": 0.7970177970177971,
|
| 3456 |
+
"eval_runtime": 5.2348,
|
| 3457 |
+
"eval_samples_per_second": 831.737,
|
| 3458 |
+
"eval_steps_per_second": 0.573,
|
| 3459 |
+
"step": 414
|
| 3460 |
+
},
|
| 3461 |
+
{
|
| 3462 |
+
"epoch": 46.111111111111114,
|
| 3463 |
+
"grad_norm": 149671.578125,
|
| 3464 |
+
"learning_rate": 4.7657326596233353e-07,
|
| 3465 |
+
"loss": 0.4225,
|
| 3466 |
+
"step": 415
|
| 3467 |
+
},
|
| 3468 |
+
{
|
| 3469 |
+
"epoch": 46.22222222222222,
|
| 3470 |
+
"grad_norm": 69658.734375,
|
| 3471 |
+
"learning_rate": 4.777216352779054e-07,
|
| 3472 |
+
"loss": 0.3941,
|
| 3473 |
+
"step": 416
|
| 3474 |
+
},
|
| 3475 |
+
{
|
| 3476 |
+
"epoch": 46.333333333333336,
|
| 3477 |
+
"grad_norm": 90003.7890625,
|
| 3478 |
+
"learning_rate": 4.788700045934772e-07,
|
| 3479 |
+
"loss": 0.4081,
|
| 3480 |
+
"step": 417
|
| 3481 |
+
},
|
| 3482 |
+
{
|
| 3483 |
+
"epoch": 46.44444444444444,
|
| 3484 |
+
"grad_norm": 104023.953125,
|
| 3485 |
+
"learning_rate": 4.800183739090492e-07,
|
| 3486 |
+
"loss": 0.4011,
|
| 3487 |
+
"step": 418
|
| 3488 |
+
},
|
| 3489 |
+
{
|
| 3490 |
+
"epoch": 46.55555555555556,
|
| 3491 |
+
"grad_norm": 110850.140625,
|
| 3492 |
+
"learning_rate": 4.811667432246211e-07,
|
| 3493 |
+
"loss": 0.3945,
|
| 3494 |
+
"step": 419
|
| 3495 |
+
},
|
| 3496 |
+
{
|
| 3497 |
+
"epoch": 46.666666666666664,
|
| 3498 |
+
"grad_norm": 158396.875,
|
| 3499 |
+
"learning_rate": 4.82315112540193e-07,
|
| 3500 |
+
"loss": 0.3881,
|
| 3501 |
+
"step": 420
|
| 3502 |
+
},
|
| 3503 |
+
{
|
| 3504 |
+
"epoch": 46.77777777777778,
|
| 3505 |
+
"grad_norm": 117995.6875,
|
| 3506 |
+
"learning_rate": 4.834634818557648e-07,
|
| 3507 |
+
"loss": 0.3887,
|
| 3508 |
+
"step": 421
|
| 3509 |
+
},
|
| 3510 |
+
{
|
| 3511 |
+
"epoch": 46.888888888888886,
|
| 3512 |
+
"grad_norm": 94518.578125,
|
| 3513 |
+
"learning_rate": 4.846118511713368e-07,
|
| 3514 |
+
"loss": 0.3897,
|
| 3515 |
+
"step": 422
|
| 3516 |
+
},
|
| 3517 |
+
{
|
| 3518 |
+
"epoch": 47.0,
|
| 3519 |
+
"grad_norm": 450391.875,
|
| 3520 |
+
"learning_rate": 4.857602204869086e-07,
|
| 3521 |
+
"loss": 0.3995,
|
| 3522 |
+
"step": 423
|
| 3523 |
+
},
|
| 3524 |
+
{
|
| 3525 |
+
"epoch": 47.0,
|
| 3526 |
+
"eval_accuracy": 0.8137344970142398,
|
| 3527 |
+
"eval_f1": 0.8030119018702939,
|
| 3528 |
+
"eval_loss": 0.402113676071167,
|
| 3529 |
+
"eval_precision": 0.8110893032384691,
|
| 3530 |
+
"eval_recall": 0.7950937950937951,
|
| 3531 |
+
"eval_runtime": 5.312,
|
| 3532 |
+
"eval_samples_per_second": 819.647,
|
| 3533 |
+
"eval_steps_per_second": 0.565,
|
| 3534 |
+
"step": 423
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 47.111111111111114,
|
| 3538 |
+
"grad_norm": 147506.234375,
|
| 3539 |
+
"learning_rate": 4.869085898024806e-07,
|
| 3540 |
+
"loss": 0.4033,
|
| 3541 |
+
"step": 424
|
| 3542 |
+
},
|
| 3543 |
+
{
|
| 3544 |
+
"epoch": 47.22222222222222,
|
| 3545 |
+
"grad_norm": 64057.921875,
|
| 3546 |
+
"learning_rate": 4.880569591180523e-07,
|
| 3547 |
+
"loss": 0.3938,
|
| 3548 |
+
"step": 425
|
| 3549 |
+
},
|
| 3550 |
+
{
|
| 3551 |
+
"epoch": 47.333333333333336,
|
| 3552 |
+
"grad_norm": 72932.4296875,
|
| 3553 |
+
"learning_rate": 4.892053284336243e-07,
|
| 3554 |
+
"loss": 0.3786,
|
| 3555 |
+
"step": 426
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 47.44444444444444,
|
| 3559 |
+
"grad_norm": 180486.53125,
|
| 3560 |
+
"learning_rate": 4.903536977491962e-07,
|
| 3561 |
+
"loss": 0.405,
|
| 3562 |
+
"step": 427
|
| 3563 |
+
},
|
| 3564 |
+
{
|
| 3565 |
+
"epoch": 47.55555555555556,
|
| 3566 |
+
"grad_norm": 169827.8125,
|
| 3567 |
+
"learning_rate": 4.915020670647681e-07,
|
| 3568 |
+
"loss": 0.3964,
|
| 3569 |
+
"step": 428
|
| 3570 |
+
},
|
| 3571 |
+
{
|
| 3572 |
+
"epoch": 47.666666666666664,
|
| 3573 |
+
"grad_norm": 90484.8046875,
|
| 3574 |
+
"learning_rate": 4.926504363803399e-07,
|
| 3575 |
+
"loss": 0.4097,
|
| 3576 |
+
"step": 429
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 47.77777777777778,
|
| 3580 |
+
"grad_norm": 135750.921875,
|
| 3581 |
+
"learning_rate": 4.937988056959118e-07,
|
| 3582 |
+
"loss": 0.3741,
|
| 3583 |
+
"step": 430
|
| 3584 |
+
},
|
| 3585 |
+
{
|
| 3586 |
+
"epoch": 47.888888888888886,
|
| 3587 |
+
"grad_norm": 182169.15625,
|
| 3588 |
+
"learning_rate": 4.949471750114837e-07,
|
| 3589 |
+
"loss": 0.3864,
|
| 3590 |
+
"step": 431
|
| 3591 |
+
},
|
| 3592 |
+
{
|
| 3593 |
+
"epoch": 48.0,
|
| 3594 |
+
"grad_norm": 320654.59375,
|
| 3595 |
+
"learning_rate": 4.960955443270557e-07,
|
| 3596 |
+
"loss": 0.3996,
|
| 3597 |
+
"step": 432
|
| 3598 |
+
},
|
| 3599 |
+
{
|
| 3600 |
+
"epoch": 48.0,
|
| 3601 |
+
"eval_accuracy": 0.816260909508498,
|
| 3602 |
+
"eval_f1": 0.806949806949807,
|
| 3603 |
+
"eval_loss": 0.40212130546569824,
|
| 3604 |
+
"eval_precision": 0.8096852300242131,
|
| 3605 |
+
"eval_recall": 0.8042328042328042,
|
| 3606 |
+
"eval_runtime": 5.311,
|
| 3607 |
+
"eval_samples_per_second": 819.808,
|
| 3608 |
+
"eval_steps_per_second": 0.565,
|
| 3609 |
+
"step": 432
|
| 3610 |
+
}
|
| 3611 |
+
],
|
| 3612 |
+
"logging_steps": 1,
|
| 3613 |
+
"max_steps": 450,
|
| 3614 |
+
"num_input_tokens_seen": 0,
|
| 3615 |
+
"num_train_epochs": 50,
|
| 3616 |
+
"save_steps": 500,
|
| 3617 |
+
"stateful_callbacks": {
|
| 3618 |
+
"TrainerControl": {
|
| 3619 |
+
"args": {
|
| 3620 |
+
"should_epoch_stop": false,
|
| 3621 |
+
"should_evaluate": false,
|
| 3622 |
+
"should_log": false,
|
| 3623 |
+
"should_save": true,
|
| 3624 |
+
"should_training_stop": false
|
| 3625 |
+
},
|
| 3626 |
+
"attributes": {}
|
| 3627 |
+
}
|
| 3628 |
+
},
|
| 3629 |
+
"total_flos": 3.295709628412723e+16,
|
| 3630 |
+
"train_batch_size": 2048,
|
| 3631 |
+
"trial_name": null,
|
| 3632 |
+
"trial_params": null
|
| 3633 |
+
}
|
checkpoint-432/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:84dd02cddcded42bf354e8633385d35d0e79606bbb5580f4503cf06d25f922ea
|
| 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:e93cd22a08f1a336b184d558e7b03b78b55971c285b7447a591b30640cb89618
|
| 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 sha256:84dd02cddcded42bf354e8633385d35d0e79606bbb5580f4503cf06d25f922ea
|
| 3 |
size 5368
|