Upload folder using huggingface_hub
Browse files- README.md +31 -0
- checkpoint-11988/config.json +33 -0
- checkpoint-11988/model.safetensors +3 -0
- checkpoint-11988/optimizer.pt +3 -0
- checkpoint-11988/rng_state.pth +3 -0
- checkpoint-11988/scheduler.pt +3 -0
- checkpoint-11988/trainer_state.json +3447 -0
- checkpoint-11988/training_args.bin +3 -0
- config.json +33 -0
- model.safetensors +3 -0
- preprocessor_config.json +22 -0
- runs/Oct21_02-57-43_r-haiefff-sjstuff-xmuxso1w-882ba-zoike/events.out.tfevents.1729479465.r-haiefff-sjstuff-xmuxso1w-882ba-zoike.172.0 +2 -2
- runs/Oct21_02-57-43_r-haiefff-sjstuff-xmuxso1w-882ba-zoike/events.out.tfevents.1729483109.r-haiefff-sjstuff-xmuxso1w-882ba-zoike.172.1 +3 -0
- training_args.bin +3 -0
- training_params.json +29 -0
README.md
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
---
|
| 3 |
+
tags:
|
| 4 |
+
- autotrain
|
| 5 |
+
- image-classification
|
| 6 |
+
base_model: google/vit-base-patch16-384
|
| 7 |
+
widget:
|
| 8 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/tiger.jpg
|
| 9 |
+
example_title: Tiger
|
| 10 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/teapot.jpg
|
| 11 |
+
example_title: Teapot
|
| 12 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/palace.jpg
|
| 13 |
+
example_title: Palace
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
# Model Trained Using AutoTrain
|
| 17 |
+
|
| 18 |
+
- Problem type: Image Classification
|
| 19 |
+
|
| 20 |
+
## Validation Metrics
|
| 21 |
+
loss: 0.002970785601064563
|
| 22 |
+
|
| 23 |
+
f1: 0.9659863945578231
|
| 24 |
+
|
| 25 |
+
precision: 0.9726027397260274
|
| 26 |
+
|
| 27 |
+
recall: 0.9594594594594594
|
| 28 |
+
|
| 29 |
+
auc: 0.9999155405405404
|
| 30 |
+
|
| 31 |
+
accuracy: 0.9991658324991658
|
checkpoint-11988/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-base-patch16-384",
|
| 3 |
+
"_num_labels": 2,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"ViTForImageClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_probs_dropout_prob": 0.0,
|
| 8 |
+
"encoder_stride": 16,
|
| 9 |
+
"hidden_act": "gelu",
|
| 10 |
+
"hidden_dropout_prob": 0.0,
|
| 11 |
+
"hidden_size": 768,
|
| 12 |
+
"id2label": {
|
| 13 |
+
"0": "not_soyjak",
|
| 14 |
+
"1": "soyjak"
|
| 15 |
+
},
|
| 16 |
+
"image_size": 384,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 3072,
|
| 19 |
+
"label2id": {
|
| 20 |
+
"not_soyjak": 0,
|
| 21 |
+
"soyjak": 1
|
| 22 |
+
},
|
| 23 |
+
"layer_norm_eps": 1e-12,
|
| 24 |
+
"model_type": "vit",
|
| 25 |
+
"num_attention_heads": 12,
|
| 26 |
+
"num_channels": 3,
|
| 27 |
+
"num_hidden_layers": 12,
|
| 28 |
+
"patch_size": 16,
|
| 29 |
+
"problem_type": "single_label_classification",
|
| 30 |
+
"qkv_bias": true,
|
| 31 |
+
"torch_dtype": "float32",
|
| 32 |
+
"transformers_version": "4.45.0"
|
| 33 |
+
}
|
checkpoint-11988/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:79c52f74f6defcd2becb5f8c8d19d415c07a5ce8587e929a351d1a886356b70f
|
| 3 |
+
size 344391328
|
checkpoint-11988/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a10ce4e25732dba7ddf463c525aca67cc3d7cda8b5202a2d6e0a9c2f733752d9
|
| 3 |
+
size 688903610
|
checkpoint-11988/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e32df0883752eb9da8c5aa8e999c457883a52936228ea8d6f38b1bd5a3b11ba
|
| 3 |
+
size 14244
|
checkpoint-11988/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a430dae9e7360b02a463d96313dcfc15bbf2be1f4932fc91e49d0affad5b66f2
|
| 3 |
+
size 1064
|
checkpoint-11988/trainer_state.json
ADDED
|
@@ -0,0 +1,3447 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.002970785601064563,
|
| 3 |
+
"best_model_checkpoint": "sj-det-3/checkpoint-11988",
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 11988,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.008341675008341674,
|
| 13 |
+
"grad_norm": 14.153964042663574,
|
| 14 |
+
"learning_rate": 1.0008340283569642e-06,
|
| 15 |
+
"loss": 0.7821,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.01668335001668335,
|
| 20 |
+
"grad_norm": 11.691067695617676,
|
| 21 |
+
"learning_rate": 2.0433694745621353e-06,
|
| 22 |
+
"loss": 0.6237,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.025025025025025027,
|
| 27 |
+
"grad_norm": 6.019664764404297,
|
| 28 |
+
"learning_rate": 3.0859049207673064e-06,
|
| 29 |
+
"loss": 0.4143,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0333667000333667,
|
| 34 |
+
"grad_norm": 2.7289156913757324,
|
| 35 |
+
"learning_rate": 4.128440366972477e-06,
|
| 36 |
+
"loss": 0.1794,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.04170837504170837,
|
| 41 |
+
"grad_norm": 0.4015839397907257,
|
| 42 |
+
"learning_rate": 5.170975813177649e-06,
|
| 43 |
+
"loss": 0.0836,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.05005005005005005,
|
| 48 |
+
"grad_norm": 0.41425979137420654,
|
| 49 |
+
"learning_rate": 6.213511259382819e-06,
|
| 50 |
+
"loss": 0.0735,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.05839172505839173,
|
| 55 |
+
"grad_norm": 0.04649379849433899,
|
| 56 |
+
"learning_rate": 7.256046705587991e-06,
|
| 57 |
+
"loss": 0.0196,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0667334000667334,
|
| 62 |
+
"grad_norm": 0.03922989219427109,
|
| 63 |
+
"learning_rate": 8.298582151793162e-06,
|
| 64 |
+
"loss": 0.0703,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.07507507507507508,
|
| 69 |
+
"grad_norm": 0.09923669695854187,
|
| 70 |
+
"learning_rate": 9.341117597998331e-06,
|
| 71 |
+
"loss": 0.0382,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.08341675008341674,
|
| 76 |
+
"grad_norm": 0.41646531224250793,
|
| 77 |
+
"learning_rate": 1.0383653044203504e-05,
|
| 78 |
+
"loss": 0.0306,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.09175842509175842,
|
| 83 |
+
"grad_norm": 0.030469655990600586,
|
| 84 |
+
"learning_rate": 1.1426188490408674e-05,
|
| 85 |
+
"loss": 0.0672,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.1001001001001001,
|
| 90 |
+
"grad_norm": 6.139395713806152,
|
| 91 |
+
"learning_rate": 1.2468723936613845e-05,
|
| 92 |
+
"loss": 0.0419,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.10844177510844177,
|
| 97 |
+
"grad_norm": 1.2067941427230835,
|
| 98 |
+
"learning_rate": 1.3511259382819017e-05,
|
| 99 |
+
"loss": 0.0047,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.11678345011678345,
|
| 104 |
+
"grad_norm": 0.019348450005054474,
|
| 105 |
+
"learning_rate": 1.4553794829024186e-05,
|
| 106 |
+
"loss": 0.001,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.12512512512512514,
|
| 111 |
+
"grad_norm": 0.024063795804977417,
|
| 112 |
+
"learning_rate": 1.559633027522936e-05,
|
| 113 |
+
"loss": 0.048,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.1334668001334668,
|
| 118 |
+
"grad_norm": 0.01152120716869831,
|
| 119 |
+
"learning_rate": 1.663886572143453e-05,
|
| 120 |
+
"loss": 0.0275,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.14180847514180847,
|
| 125 |
+
"grad_norm": 0.007050159387290478,
|
| 126 |
+
"learning_rate": 1.7681401167639698e-05,
|
| 127 |
+
"loss": 0.0109,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.15015015015015015,
|
| 132 |
+
"grad_norm": 0.007726074196398258,
|
| 133 |
+
"learning_rate": 1.8723936613844872e-05,
|
| 134 |
+
"loss": 0.0007,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.15849182515849183,
|
| 139 |
+
"grad_norm": 0.008293859660625458,
|
| 140 |
+
"learning_rate": 1.9766472060050043e-05,
|
| 141 |
+
"loss": 0.028,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.1668335001668335,
|
| 146 |
+
"grad_norm": 0.0058615542948246,
|
| 147 |
+
"learning_rate": 2.0809007506255214e-05,
|
| 148 |
+
"loss": 0.0025,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.17517517517517517,
|
| 153 |
+
"grad_norm": 0.028854593634605408,
|
| 154 |
+
"learning_rate": 2.1851542952460384e-05,
|
| 155 |
+
"loss": 0.0685,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.18351685018351685,
|
| 160 |
+
"grad_norm": 0.042628563940525055,
|
| 161 |
+
"learning_rate": 2.2894078398665555e-05,
|
| 162 |
+
"loss": 0.0401,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.19185852519185853,
|
| 167 |
+
"grad_norm": 0.378325879573822,
|
| 168 |
+
"learning_rate": 2.3936613844870726e-05,
|
| 169 |
+
"loss": 0.0014,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.2002002002002002,
|
| 174 |
+
"grad_norm": 0.007751138880848885,
|
| 175 |
+
"learning_rate": 2.49791492910759e-05,
|
| 176 |
+
"loss": 0.0445,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.20854187520854187,
|
| 181 |
+
"grad_norm": 0.008357309736311436,
|
| 182 |
+
"learning_rate": 2.602168473728107e-05,
|
| 183 |
+
"loss": 0.0079,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.21688355021688355,
|
| 188 |
+
"grad_norm": 0.17902156710624695,
|
| 189 |
+
"learning_rate": 2.7064220183486238e-05,
|
| 190 |
+
"loss": 0.026,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.22522522522522523,
|
| 195 |
+
"grad_norm": 0.030053455382585526,
|
| 196 |
+
"learning_rate": 2.810675562969141e-05,
|
| 197 |
+
"loss": 0.1212,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.2335669002335669,
|
| 202 |
+
"grad_norm": 0.0116727901622653,
|
| 203 |
+
"learning_rate": 2.9149291075896586e-05,
|
| 204 |
+
"loss": 0.0459,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.24190857524190856,
|
| 209 |
+
"grad_norm": 0.005300463642925024,
|
| 210 |
+
"learning_rate": 3.0191826522101753e-05,
|
| 211 |
+
"loss": 0.0004,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.2502502502502503,
|
| 216 |
+
"grad_norm": 0.005538405384868383,
|
| 217 |
+
"learning_rate": 3.1234361968306924e-05,
|
| 218 |
+
"loss": 0.0205,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.2585919252585919,
|
| 223 |
+
"grad_norm": 0.009918354451656342,
|
| 224 |
+
"learning_rate": 3.2276897414512094e-05,
|
| 225 |
+
"loss": 0.0025,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.2669336002669336,
|
| 230 |
+
"grad_norm": 0.018012989312410355,
|
| 231 |
+
"learning_rate": 3.3319432860717265e-05,
|
| 232 |
+
"loss": 0.082,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.2752752752752753,
|
| 237 |
+
"grad_norm": 0.0018511965172365308,
|
| 238 |
+
"learning_rate": 3.4361968306922436e-05,
|
| 239 |
+
"loss": 0.0071,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.28361695028361694,
|
| 244 |
+
"grad_norm": 0.0468880757689476,
|
| 245 |
+
"learning_rate": 3.540450375312761e-05,
|
| 246 |
+
"loss": 0.0281,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.29195862529195865,
|
| 251 |
+
"grad_norm": 0.004004584159702063,
|
| 252 |
+
"learning_rate": 3.644703919933278e-05,
|
| 253 |
+
"loss": 0.0662,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.3003003003003003,
|
| 258 |
+
"grad_norm": 0.006573744583874941,
|
| 259 |
+
"learning_rate": 3.748957464553795e-05,
|
| 260 |
+
"loss": 0.0017,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.30864197530864196,
|
| 265 |
+
"grad_norm": 0.00465534208342433,
|
| 266 |
+
"learning_rate": 3.8532110091743125e-05,
|
| 267 |
+
"loss": 0.0159,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.31698365031698367,
|
| 272 |
+
"grad_norm": 0.6863465905189514,
|
| 273 |
+
"learning_rate": 3.957464553794829e-05,
|
| 274 |
+
"loss": 0.0037,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.3253253253253253,
|
| 279 |
+
"grad_norm": 0.0018059064168483019,
|
| 280 |
+
"learning_rate": 4.0617180984153466e-05,
|
| 281 |
+
"loss": 0.0001,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.333667000333667,
|
| 286 |
+
"grad_norm": 0.001327801262959838,
|
| 287 |
+
"learning_rate": 4.165971643035864e-05,
|
| 288 |
+
"loss": 0.0001,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.3420086753420087,
|
| 293 |
+
"grad_norm": 0.0013706604950129986,
|
| 294 |
+
"learning_rate": 4.27022518765638e-05,
|
| 295 |
+
"loss": 0.0194,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.35035035035035034,
|
| 300 |
+
"grad_norm": 0.0007294805254787207,
|
| 301 |
+
"learning_rate": 4.374478732276898e-05,
|
| 302 |
+
"loss": 0.0001,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.35869202535869205,
|
| 307 |
+
"grad_norm": 0.018524976447224617,
|
| 308 |
+
"learning_rate": 4.478732276897415e-05,
|
| 309 |
+
"loss": 0.0481,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.3670337003670337,
|
| 314 |
+
"grad_norm": 0.009606343694031239,
|
| 315 |
+
"learning_rate": 4.582985821517932e-05,
|
| 316 |
+
"loss": 0.0586,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.37537537537537535,
|
| 321 |
+
"grad_norm": 0.0046248408034443855,
|
| 322 |
+
"learning_rate": 4.687239366138449e-05,
|
| 323 |
+
"loss": 0.0105,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.38371705038371706,
|
| 328 |
+
"grad_norm": 0.017802894115447998,
|
| 329 |
+
"learning_rate": 4.791492910758966e-05,
|
| 330 |
+
"loss": 0.0019,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.3920587253920587,
|
| 335 |
+
"grad_norm": 14.872435569763184,
|
| 336 |
+
"learning_rate": 4.895746455379483e-05,
|
| 337 |
+
"loss": 0.0849,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.4004004004004004,
|
| 342 |
+
"grad_norm": 0.001058954163454473,
|
| 343 |
+
"learning_rate": 5e-05,
|
| 344 |
+
"loss": 0.0395,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.4087420754087421,
|
| 349 |
+
"grad_norm": 0.0007220833795145154,
|
| 350 |
+
"learning_rate": 4.988414125498193e-05,
|
| 351 |
+
"loss": 0.0001,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.41708375041708373,
|
| 356 |
+
"grad_norm": 0.0015725018456578255,
|
| 357 |
+
"learning_rate": 4.9768282509963857e-05,
|
| 358 |
+
"loss": 0.0,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.42542542542542544,
|
| 363 |
+
"grad_norm": 0.00035438829218037426,
|
| 364 |
+
"learning_rate": 4.965242376494578e-05,
|
| 365 |
+
"loss": 0.0001,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.4337671004337671,
|
| 370 |
+
"grad_norm": 0.004498378373682499,
|
| 371 |
+
"learning_rate": 4.9536565019927704e-05,
|
| 372 |
+
"loss": 0.0617,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.44210877544210875,
|
| 377 |
+
"grad_norm": 0.0019637339282780886,
|
| 378 |
+
"learning_rate": 4.9420706274909634e-05,
|
| 379 |
+
"loss": 0.0452,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.45045045045045046,
|
| 384 |
+
"grad_norm": 0.0009583417559042573,
|
| 385 |
+
"learning_rate": 4.930484752989156e-05,
|
| 386 |
+
"loss": 0.0004,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.4587921254587921,
|
| 391 |
+
"grad_norm": 0.011658573523163795,
|
| 392 |
+
"learning_rate": 4.918898878487349e-05,
|
| 393 |
+
"loss": 0.0626,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.4671338004671338,
|
| 398 |
+
"grad_norm": 0.002904647495597601,
|
| 399 |
+
"learning_rate": 4.9073130039855406e-05,
|
| 400 |
+
"loss": 0.0067,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.4754754754754755,
|
| 405 |
+
"grad_norm": 0.004803840536624193,
|
| 406 |
+
"learning_rate": 4.8957271294837336e-05,
|
| 407 |
+
"loss": 0.0453,
|
| 408 |
+
"step": 1425
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.4838171504838171,
|
| 412 |
+
"grad_norm": 0.0012352573685348034,
|
| 413 |
+
"learning_rate": 4.8841412549819267e-05,
|
| 414 |
+
"loss": 0.0003,
|
| 415 |
+
"step": 1450
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.49215882549215884,
|
| 419 |
+
"grad_norm": 0.0022556546609848738,
|
| 420 |
+
"learning_rate": 4.872555380480119e-05,
|
| 421 |
+
"loss": 0.0002,
|
| 422 |
+
"step": 1475
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.5005005005005005,
|
| 426 |
+
"grad_norm": 0.01666790060698986,
|
| 427 |
+
"learning_rate": 4.8609695059783114e-05,
|
| 428 |
+
"loss": 0.0142,
|
| 429 |
+
"step": 1500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.5088421755088421,
|
| 433 |
+
"grad_norm": 0.0008659372106194496,
|
| 434 |
+
"learning_rate": 4.849383631476504e-05,
|
| 435 |
+
"loss": 0.0003,
|
| 436 |
+
"step": 1525
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.5171838505171839,
|
| 440 |
+
"grad_norm": 0.009269647300243378,
|
| 441 |
+
"learning_rate": 4.837797756974697e-05,
|
| 442 |
+
"loss": 0.0001,
|
| 443 |
+
"step": 1550
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.5255255255255256,
|
| 447 |
+
"grad_norm": 0.00042679469333961606,
|
| 448 |
+
"learning_rate": 4.826211882472889e-05,
|
| 449 |
+
"loss": 0.0,
|
| 450 |
+
"step": 1575
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.5338672005338672,
|
| 454 |
+
"grad_norm": 0.0003430478391237557,
|
| 455 |
+
"learning_rate": 4.814626007971082e-05,
|
| 456 |
+
"loss": 0.0,
|
| 457 |
+
"step": 1600
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.5422088755422089,
|
| 461 |
+
"grad_norm": 0.002377784810960293,
|
| 462 |
+
"learning_rate": 4.8030401334692746e-05,
|
| 463 |
+
"loss": 0.0926,
|
| 464 |
+
"step": 1625
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.5505505505505506,
|
| 468 |
+
"grad_norm": 0.003815292613580823,
|
| 469 |
+
"learning_rate": 4.791454258967467e-05,
|
| 470 |
+
"loss": 0.0001,
|
| 471 |
+
"step": 1650
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.5588922255588922,
|
| 475 |
+
"grad_norm": 0.0007572752656415105,
|
| 476 |
+
"learning_rate": 4.779868384465659e-05,
|
| 477 |
+
"loss": 0.0001,
|
| 478 |
+
"step": 1675
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.5672339005672339,
|
| 482 |
+
"grad_norm": 0.003480930346995592,
|
| 483 |
+
"learning_rate": 4.7682825099638524e-05,
|
| 484 |
+
"loss": 0.023,
|
| 485 |
+
"step": 1700
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.5755755755755756,
|
| 489 |
+
"grad_norm": 0.0046263947151601315,
|
| 490 |
+
"learning_rate": 4.7566966354620454e-05,
|
| 491 |
+
"loss": 0.0219,
|
| 492 |
+
"step": 1725
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.5839172505839173,
|
| 496 |
+
"grad_norm": 0.3202002942562103,
|
| 497 |
+
"learning_rate": 4.745110760960237e-05,
|
| 498 |
+
"loss": 0.0298,
|
| 499 |
+
"step": 1750
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.5922589255922589,
|
| 503 |
+
"grad_norm": 6.295952320098877,
|
| 504 |
+
"learning_rate": 4.73352488645843e-05,
|
| 505 |
+
"loss": 0.0546,
|
| 506 |
+
"step": 1775
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.6006006006006006,
|
| 510 |
+
"grad_norm": 0.08465034514665604,
|
| 511 |
+
"learning_rate": 4.7219390119566225e-05,
|
| 512 |
+
"loss": 0.0376,
|
| 513 |
+
"step": 1800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.6089422756089423,
|
| 517 |
+
"grad_norm": 0.003993997350335121,
|
| 518 |
+
"learning_rate": 4.7103531374548156e-05,
|
| 519 |
+
"loss": 0.0541,
|
| 520 |
+
"step": 1825
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.6172839506172839,
|
| 524 |
+
"grad_norm": 2.8580386638641357,
|
| 525 |
+
"learning_rate": 4.698767262953008e-05,
|
| 526 |
+
"loss": 0.0475,
|
| 527 |
+
"step": 1850
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.6256256256256256,
|
| 531 |
+
"grad_norm": 0.0002566488110460341,
|
| 532 |
+
"learning_rate": 4.6871813884512e-05,
|
| 533 |
+
"loss": 0.0155,
|
| 534 |
+
"step": 1875
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.6339673006339673,
|
| 538 |
+
"grad_norm": 0.004632447846233845,
|
| 539 |
+
"learning_rate": 4.675595513949393e-05,
|
| 540 |
+
"loss": 0.0,
|
| 541 |
+
"step": 1900
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.6423089756423089,
|
| 545 |
+
"grad_norm": 0.0001871311542345211,
|
| 546 |
+
"learning_rate": 4.664009639447586e-05,
|
| 547 |
+
"loss": 0.0,
|
| 548 |
+
"step": 1925
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.6506506506506506,
|
| 552 |
+
"grad_norm": 0.0010421440238133073,
|
| 553 |
+
"learning_rate": 4.652423764945779e-05,
|
| 554 |
+
"loss": 0.0416,
|
| 555 |
+
"step": 1950
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.6589923256589924,
|
| 559 |
+
"grad_norm": 0.005437193904072046,
|
| 560 |
+
"learning_rate": 4.640837890443971e-05,
|
| 561 |
+
"loss": 0.0564,
|
| 562 |
+
"step": 1975
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.667334000667334,
|
| 566 |
+
"grad_norm": 0.0026402724906802177,
|
| 567 |
+
"learning_rate": 4.6292520159421635e-05,
|
| 568 |
+
"loss": 0.0209,
|
| 569 |
+
"step": 2000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.6756756756756757,
|
| 573 |
+
"grad_norm": 0.0003550160035956651,
|
| 574 |
+
"learning_rate": 4.617666141440356e-05,
|
| 575 |
+
"loss": 0.0012,
|
| 576 |
+
"step": 2025
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.6840173506840174,
|
| 580 |
+
"grad_norm": 0.00020284796482883394,
|
| 581 |
+
"learning_rate": 4.606080266938549e-05,
|
| 582 |
+
"loss": 0.0002,
|
| 583 |
+
"step": 2050
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.6923590256923591,
|
| 587 |
+
"grad_norm": 0.0005185810732655227,
|
| 588 |
+
"learning_rate": 4.594494392436741e-05,
|
| 589 |
+
"loss": 0.0001,
|
| 590 |
+
"step": 2075
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.7007007007007007,
|
| 594 |
+
"grad_norm": 0.29189226031303406,
|
| 595 |
+
"learning_rate": 4.582908517934934e-05,
|
| 596 |
+
"loss": 0.0002,
|
| 597 |
+
"step": 2100
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.7090423757090424,
|
| 601 |
+
"grad_norm": 0.006394149269908667,
|
| 602 |
+
"learning_rate": 4.571322643433126e-05,
|
| 603 |
+
"loss": 0.0,
|
| 604 |
+
"step": 2125
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.7173840507173841,
|
| 608 |
+
"grad_norm": 0.003997980151325464,
|
| 609 |
+
"learning_rate": 4.559736768931319e-05,
|
| 610 |
+
"loss": 0.0431,
|
| 611 |
+
"step": 2150
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.7257257257257257,
|
| 615 |
+
"grad_norm": 0.007415532600134611,
|
| 616 |
+
"learning_rate": 4.548150894429512e-05,
|
| 617 |
+
"loss": 0.0193,
|
| 618 |
+
"step": 2175
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.7340674007340674,
|
| 622 |
+
"grad_norm": 0.0035364318173378706,
|
| 623 |
+
"learning_rate": 4.5365650199277045e-05,
|
| 624 |
+
"loss": 0.0631,
|
| 625 |
+
"step": 2200
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.7424090757424091,
|
| 629 |
+
"grad_norm": 0.002615898149088025,
|
| 630 |
+
"learning_rate": 4.524979145425897e-05,
|
| 631 |
+
"loss": 0.0349,
|
| 632 |
+
"step": 2225
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.7507507507507507,
|
| 636 |
+
"grad_norm": 0.0017047699075192213,
|
| 637 |
+
"learning_rate": 4.513393270924089e-05,
|
| 638 |
+
"loss": 0.021,
|
| 639 |
+
"step": 2250
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.7590924257590924,
|
| 643 |
+
"grad_norm": 0.001677798223681748,
|
| 644 |
+
"learning_rate": 4.501807396422282e-05,
|
| 645 |
+
"loss": 0.0209,
|
| 646 |
+
"step": 2275
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.7674341007674341,
|
| 650 |
+
"grad_norm": 0.0003330695035401732,
|
| 651 |
+
"learning_rate": 4.490221521920475e-05,
|
| 652 |
+
"loss": 0.0005,
|
| 653 |
+
"step": 2300
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.7757757757757757,
|
| 657 |
+
"grad_norm": 9.45928695728071e-05,
|
| 658 |
+
"learning_rate": 4.478635647418668e-05,
|
| 659 |
+
"loss": 0.026,
|
| 660 |
+
"step": 2325
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.7841174507841174,
|
| 664 |
+
"grad_norm": 6.650515556335449,
|
| 665 |
+
"learning_rate": 4.46704977291686e-05,
|
| 666 |
+
"loss": 0.018,
|
| 667 |
+
"step": 2350
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.7924591257924591,
|
| 671 |
+
"grad_norm": 0.0031244370620697737,
|
| 672 |
+
"learning_rate": 4.4554638984150525e-05,
|
| 673 |
+
"loss": 0.0001,
|
| 674 |
+
"step": 2375
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.8008008008008008,
|
| 678 |
+
"grad_norm": 0.0006237108027562499,
|
| 679 |
+
"learning_rate": 4.443878023913245e-05,
|
| 680 |
+
"loss": 0.0145,
|
| 681 |
+
"step": 2400
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.8091424758091424,
|
| 685 |
+
"grad_norm": 0.0002782900701276958,
|
| 686 |
+
"learning_rate": 4.432292149411438e-05,
|
| 687 |
+
"loss": 0.0677,
|
| 688 |
+
"step": 2425
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.8174841508174842,
|
| 692 |
+
"grad_norm": 0.0003246616688556969,
|
| 693 |
+
"learning_rate": 4.420706274909631e-05,
|
| 694 |
+
"loss": 0.0224,
|
| 695 |
+
"step": 2450
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.8258258258258259,
|
| 699 |
+
"grad_norm": 0.0015066479099914432,
|
| 700 |
+
"learning_rate": 4.4091204004078226e-05,
|
| 701 |
+
"loss": 0.0112,
|
| 702 |
+
"step": 2475
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.8341675008341675,
|
| 706 |
+
"grad_norm": 0.0002765478566288948,
|
| 707 |
+
"learning_rate": 4.397534525906016e-05,
|
| 708 |
+
"loss": 0.0078,
|
| 709 |
+
"step": 2500
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.8425091758425092,
|
| 713 |
+
"grad_norm": 0.002668420784175396,
|
| 714 |
+
"learning_rate": 4.385948651404208e-05,
|
| 715 |
+
"loss": 0.0011,
|
| 716 |
+
"step": 2525
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.8508508508508509,
|
| 720 |
+
"grad_norm": 0.00015687810082454234,
|
| 721 |
+
"learning_rate": 4.374362776902401e-05,
|
| 722 |
+
"loss": 0.0004,
|
| 723 |
+
"step": 2550
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.8591925258591925,
|
| 727 |
+
"grad_norm": 1.4734420776367188,
|
| 728 |
+
"learning_rate": 4.3627769024005935e-05,
|
| 729 |
+
"loss": 0.0023,
|
| 730 |
+
"step": 2575
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.8675342008675342,
|
| 734 |
+
"grad_norm": 0.0018520369194447994,
|
| 735 |
+
"learning_rate": 4.351191027898786e-05,
|
| 736 |
+
"loss": 0.0001,
|
| 737 |
+
"step": 2600
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.8758758758758759,
|
| 741 |
+
"grad_norm": 0.012340963818132877,
|
| 742 |
+
"learning_rate": 4.339605153396978e-05,
|
| 743 |
+
"loss": 0.0488,
|
| 744 |
+
"step": 2625
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.8842175508842175,
|
| 748 |
+
"grad_norm": 0.00010058472980745137,
|
| 749 |
+
"learning_rate": 4.328019278895171e-05,
|
| 750 |
+
"loss": 0.0123,
|
| 751 |
+
"step": 2650
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.8925592258925592,
|
| 755 |
+
"grad_norm": 0.022818908095359802,
|
| 756 |
+
"learning_rate": 4.316433404393364e-05,
|
| 757 |
+
"loss": 0.1159,
|
| 758 |
+
"step": 2675
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.9009009009009009,
|
| 762 |
+
"grad_norm": 0.006666264962404966,
|
| 763 |
+
"learning_rate": 4.304847529891557e-05,
|
| 764 |
+
"loss": 0.0008,
|
| 765 |
+
"step": 2700
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.9092425759092426,
|
| 769 |
+
"grad_norm": 11.06509780883789,
|
| 770 |
+
"learning_rate": 4.293261655389749e-05,
|
| 771 |
+
"loss": 0.0061,
|
| 772 |
+
"step": 2725
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.9175842509175842,
|
| 776 |
+
"grad_norm": 0.011369475163519382,
|
| 777 |
+
"learning_rate": 4.2816757808879414e-05,
|
| 778 |
+
"loss": 0.0413,
|
| 779 |
+
"step": 2750
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.9259259259259259,
|
| 783 |
+
"grad_norm": 7.088267803192139,
|
| 784 |
+
"learning_rate": 4.2700899063861345e-05,
|
| 785 |
+
"loss": 0.0311,
|
| 786 |
+
"step": 2775
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.9342676009342676,
|
| 790 |
+
"grad_norm": 4.856973171234131,
|
| 791 |
+
"learning_rate": 4.258504031884327e-05,
|
| 792 |
+
"loss": 0.0024,
|
| 793 |
+
"step": 2800
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.9426092759426092,
|
| 797 |
+
"grad_norm": 0.0006875558174215257,
|
| 798 |
+
"learning_rate": 4.246918157382519e-05,
|
| 799 |
+
"loss": 0.0133,
|
| 800 |
+
"step": 2825
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.950950950950951,
|
| 804 |
+
"grad_norm": 0.008509138599038124,
|
| 805 |
+
"learning_rate": 4.2353322828807116e-05,
|
| 806 |
+
"loss": 0.0,
|
| 807 |
+
"step": 2850
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.9592926259592927,
|
| 811 |
+
"grad_norm": 0.0004129385342821479,
|
| 812 |
+
"learning_rate": 4.2237464083789046e-05,
|
| 813 |
+
"loss": 0.0025,
|
| 814 |
+
"step": 2875
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.9676343009676343,
|
| 818 |
+
"grad_norm": 0.0008459310047328472,
|
| 819 |
+
"learning_rate": 4.212160533877097e-05,
|
| 820 |
+
"loss": 0.0,
|
| 821 |
+
"step": 2900
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.975975975975976,
|
| 825 |
+
"grad_norm": 0.01643763668835163,
|
| 826 |
+
"learning_rate": 4.20057465937529e-05,
|
| 827 |
+
"loss": 0.0981,
|
| 828 |
+
"step": 2925
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.9843176509843177,
|
| 832 |
+
"grad_norm": 0.0025746412575244904,
|
| 833 |
+
"learning_rate": 4.1889887848734824e-05,
|
| 834 |
+
"loss": 0.0006,
|
| 835 |
+
"step": 2950
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.9926593259926594,
|
| 839 |
+
"grad_norm": 0.003589796368032694,
|
| 840 |
+
"learning_rate": 4.177402910371675e-05,
|
| 841 |
+
"loss": 0.0022,
|
| 842 |
+
"step": 2975
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.0,
|
| 846 |
+
"eval_accuracy": 0.995995995995996,
|
| 847 |
+
"eval_auc": 0.9962769357195034,
|
| 848 |
+
"eval_f1": 0.8518518518518519,
|
| 849 |
+
"eval_loss": 0.014670311473309994,
|
| 850 |
+
"eval_precision": 0.7840909090909091,
|
| 851 |
+
"eval_recall": 0.9324324324324325,
|
| 852 |
+
"eval_runtime": 78.4692,
|
| 853 |
+
"eval_samples_per_second": 76.387,
|
| 854 |
+
"eval_steps_per_second": 4.779,
|
| 855 |
+
"step": 2997
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 1.001001001001001,
|
| 859 |
+
"grad_norm": 0.003225616877898574,
|
| 860 |
+
"learning_rate": 4.165817035869868e-05,
|
| 861 |
+
"loss": 0.0324,
|
| 862 |
+
"step": 3000
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 1.0093426760093427,
|
| 866 |
+
"grad_norm": 0.23891094326972961,
|
| 867 |
+
"learning_rate": 4.15423116136806e-05,
|
| 868 |
+
"loss": 0.0233,
|
| 869 |
+
"step": 3025
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 1.0176843510176843,
|
| 873 |
+
"grad_norm": 0.004559539258480072,
|
| 874 |
+
"learning_rate": 4.142645286866253e-05,
|
| 875 |
+
"loss": 0.0402,
|
| 876 |
+
"step": 3050
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.026026026026026,
|
| 880 |
+
"grad_norm": 0.005625482648611069,
|
| 881 |
+
"learning_rate": 4.131059412364445e-05,
|
| 882 |
+
"loss": 0.0003,
|
| 883 |
+
"step": 3075
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.0343677010343677,
|
| 887 |
+
"grad_norm": 0.004497454967349768,
|
| 888 |
+
"learning_rate": 4.119473537862638e-05,
|
| 889 |
+
"loss": 0.0004,
|
| 890 |
+
"step": 3100
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.0427093760427093,
|
| 894 |
+
"grad_norm": 0.0791538655757904,
|
| 895 |
+
"learning_rate": 4.1078876633608304e-05,
|
| 896 |
+
"loss": 0.0057,
|
| 897 |
+
"step": 3125
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.0510510510510511,
|
| 901 |
+
"grad_norm": 0.0016846450744196773,
|
| 902 |
+
"learning_rate": 4.0963017888590234e-05,
|
| 903 |
+
"loss": 0.0003,
|
| 904 |
+
"step": 3150
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.0593927260593927,
|
| 908 |
+
"grad_norm": 0.0009143862989731133,
|
| 909 |
+
"learning_rate": 4.0847159143572165e-05,
|
| 910 |
+
"loss": 0.0001,
|
| 911 |
+
"step": 3175
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.0677344010677343,
|
| 915 |
+
"grad_norm": 0.0010049792472273111,
|
| 916 |
+
"learning_rate": 4.073130039855408e-05,
|
| 917 |
+
"loss": 0.017,
|
| 918 |
+
"step": 3200
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.0760760760760761,
|
| 922 |
+
"grad_norm": 0.003489920636638999,
|
| 923 |
+
"learning_rate": 4.061544165353601e-05,
|
| 924 |
+
"loss": 0.0001,
|
| 925 |
+
"step": 3225
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 1.0844177510844177,
|
| 929 |
+
"grad_norm": 1.6741670370101929,
|
| 930 |
+
"learning_rate": 4.0499582908517936e-05,
|
| 931 |
+
"loss": 0.0237,
|
| 932 |
+
"step": 3250
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 1.0927594260927593,
|
| 936 |
+
"grad_norm": 0.000581038068048656,
|
| 937 |
+
"learning_rate": 4.0383724163499866e-05,
|
| 938 |
+
"loss": 0.0001,
|
| 939 |
+
"step": 3275
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1.1011011011011012,
|
| 943 |
+
"grad_norm": 0.0003763669228646904,
|
| 944 |
+
"learning_rate": 4.026786541848179e-05,
|
| 945 |
+
"loss": 0.0001,
|
| 946 |
+
"step": 3300
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 1.1094427761094428,
|
| 950 |
+
"grad_norm": 0.006402493920177221,
|
| 951 |
+
"learning_rate": 4.0152006673463714e-05,
|
| 952 |
+
"loss": 0.0147,
|
| 953 |
+
"step": 3325
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 1.1177844511177844,
|
| 957 |
+
"grad_norm": 0.022542010992765427,
|
| 958 |
+
"learning_rate": 4.003614792844564e-05,
|
| 959 |
+
"loss": 0.0507,
|
| 960 |
+
"step": 3350
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 1.1261261261261262,
|
| 964 |
+
"grad_norm": 0.0030998587608337402,
|
| 965 |
+
"learning_rate": 3.992028918342757e-05,
|
| 966 |
+
"loss": 0.0103,
|
| 967 |
+
"step": 3375
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 1.1344678011344678,
|
| 971 |
+
"grad_norm": 0.001442295964807272,
|
| 972 |
+
"learning_rate": 3.98044304384095e-05,
|
| 973 |
+
"loss": 0.0003,
|
| 974 |
+
"step": 3400
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.1428094761428094,
|
| 978 |
+
"grad_norm": 0.007275531068444252,
|
| 979 |
+
"learning_rate": 3.968857169339142e-05,
|
| 980 |
+
"loss": 0.0001,
|
| 981 |
+
"step": 3425
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1.1511511511511512,
|
| 985 |
+
"grad_norm": 0.000654282805044204,
|
| 986 |
+
"learning_rate": 3.9572712948373346e-05,
|
| 987 |
+
"loss": 0.0185,
|
| 988 |
+
"step": 3450
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 1.1594928261594928,
|
| 992 |
+
"grad_norm": 0.016119500622153282,
|
| 993 |
+
"learning_rate": 3.945685420335527e-05,
|
| 994 |
+
"loss": 0.0052,
|
| 995 |
+
"step": 3475
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 1.1678345011678344,
|
| 999 |
+
"grad_norm": 0.0006950986571609974,
|
| 1000 |
+
"learning_rate": 3.93409954583372e-05,
|
| 1001 |
+
"loss": 0.0002,
|
| 1002 |
+
"step": 3500
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.1761761761761762,
|
| 1006 |
+
"grad_norm": 0.05980084836483002,
|
| 1007 |
+
"learning_rate": 3.9225136713319123e-05,
|
| 1008 |
+
"loss": 0.0001,
|
| 1009 |
+
"step": 3525
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 1.1845178511845178,
|
| 1013 |
+
"grad_norm": 0.0003822998551186174,
|
| 1014 |
+
"learning_rate": 3.910927796830105e-05,
|
| 1015 |
+
"loss": 0.0011,
|
| 1016 |
+
"step": 3550
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 1.1928595261928596,
|
| 1020 |
+
"grad_norm": 0.0003051291569136083,
|
| 1021 |
+
"learning_rate": 3.899341922328297e-05,
|
| 1022 |
+
"loss": 0.0004,
|
| 1023 |
+
"step": 3575
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 1.2012012012012012,
|
| 1027 |
+
"grad_norm": 0.02143264003098011,
|
| 1028 |
+
"learning_rate": 3.88775604782649e-05,
|
| 1029 |
+
"loss": 0.0151,
|
| 1030 |
+
"step": 3600
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 1.2095428762095428,
|
| 1034 |
+
"grad_norm": 0.012560728006064892,
|
| 1035 |
+
"learning_rate": 3.8761701733246825e-05,
|
| 1036 |
+
"loss": 0.1055,
|
| 1037 |
+
"step": 3625
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 1.2178845512178846,
|
| 1041 |
+
"grad_norm": 0.06956183165311813,
|
| 1042 |
+
"learning_rate": 3.8645842988228756e-05,
|
| 1043 |
+
"loss": 0.0288,
|
| 1044 |
+
"step": 3650
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.2262262262262262,
|
| 1048 |
+
"grad_norm": 0.019330767914652824,
|
| 1049 |
+
"learning_rate": 3.852998424321068e-05,
|
| 1050 |
+
"loss": 0.0564,
|
| 1051 |
+
"step": 3675
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.2345679012345678,
|
| 1055 |
+
"grad_norm": 0.022493431344628334,
|
| 1056 |
+
"learning_rate": 3.84141254981926e-05,
|
| 1057 |
+
"loss": 0.0188,
|
| 1058 |
+
"step": 3700
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.2429095762429097,
|
| 1062 |
+
"grad_norm": 0.0016129710711538792,
|
| 1063 |
+
"learning_rate": 3.8298266753174533e-05,
|
| 1064 |
+
"loss": 0.0001,
|
| 1065 |
+
"step": 3725
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 1.2512512512512513,
|
| 1069 |
+
"grad_norm": 0.0006044628680683672,
|
| 1070 |
+
"learning_rate": 3.818240800815646e-05,
|
| 1071 |
+
"loss": 0.0001,
|
| 1072 |
+
"step": 3750
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 1.2595929262595928,
|
| 1076 |
+
"grad_norm": 0.0004552229365799576,
|
| 1077 |
+
"learning_rate": 3.806654926313839e-05,
|
| 1078 |
+
"loss": 0.0005,
|
| 1079 |
+
"step": 3775
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 1.2679346012679347,
|
| 1083 |
+
"grad_norm": 0.0007301606819964945,
|
| 1084 |
+
"learning_rate": 3.7950690518120305e-05,
|
| 1085 |
+
"loss": 0.0,
|
| 1086 |
+
"step": 3800
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 1.2762762762762763,
|
| 1090 |
+
"grad_norm": 0.00021806442236993462,
|
| 1091 |
+
"learning_rate": 3.7834831773102235e-05,
|
| 1092 |
+
"loss": 0.0001,
|
| 1093 |
+
"step": 3825
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 1.284617951284618,
|
| 1097 |
+
"grad_norm": 0.0004119027580600232,
|
| 1098 |
+
"learning_rate": 3.771897302808416e-05,
|
| 1099 |
+
"loss": 0.0,
|
| 1100 |
+
"step": 3850
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 1.2929596262929597,
|
| 1104 |
+
"grad_norm": 0.0005425213603302836,
|
| 1105 |
+
"learning_rate": 3.760311428306609e-05,
|
| 1106 |
+
"loss": 0.0256,
|
| 1107 |
+
"step": 3875
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 1.3013013013013013,
|
| 1111 |
+
"grad_norm": 0.0020909614395350218,
|
| 1112 |
+
"learning_rate": 3.748725553804801e-05,
|
| 1113 |
+
"loss": 0.0001,
|
| 1114 |
+
"step": 3900
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 1.309642976309643,
|
| 1118 |
+
"grad_norm": 6.709009647369385,
|
| 1119 |
+
"learning_rate": 3.7371396793029937e-05,
|
| 1120 |
+
"loss": 0.0233,
|
| 1121 |
+
"step": 3925
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 1.3179846513179847,
|
| 1125 |
+
"grad_norm": 0.00022848871594760567,
|
| 1126 |
+
"learning_rate": 3.725553804801187e-05,
|
| 1127 |
+
"loss": 0.0341,
|
| 1128 |
+
"step": 3950
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 1.3263263263263263,
|
| 1132 |
+
"grad_norm": 0.00047850931878201663,
|
| 1133 |
+
"learning_rate": 3.713967930299379e-05,
|
| 1134 |
+
"loss": 0.0311,
|
| 1135 |
+
"step": 3975
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 1.3346680013346681,
|
| 1139 |
+
"grad_norm": 0.0011013607727363706,
|
| 1140 |
+
"learning_rate": 3.702382055797572e-05,
|
| 1141 |
+
"loss": 0.0001,
|
| 1142 |
+
"step": 4000
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.3430096763430097,
|
| 1146 |
+
"grad_norm": 0.03528917208313942,
|
| 1147 |
+
"learning_rate": 3.6907961812957645e-05,
|
| 1148 |
+
"loss": 0.0428,
|
| 1149 |
+
"step": 4025
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.3513513513513513,
|
| 1153 |
+
"grad_norm": 0.0020049402955919504,
|
| 1154 |
+
"learning_rate": 3.679210306793957e-05,
|
| 1155 |
+
"loss": 0.001,
|
| 1156 |
+
"step": 4050
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 1.3596930263596931,
|
| 1160 |
+
"grad_norm": 6.149662017822266,
|
| 1161 |
+
"learning_rate": 3.667624432292149e-05,
|
| 1162 |
+
"loss": 0.0331,
|
| 1163 |
+
"step": 4075
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 1.3680347013680347,
|
| 1167 |
+
"grad_norm": 0.0008450220921076834,
|
| 1168 |
+
"learning_rate": 3.656038557790342e-05,
|
| 1169 |
+
"loss": 0.0138,
|
| 1170 |
+
"step": 4100
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 1.3763763763763763,
|
| 1174 |
+
"grad_norm": 0.0007816566503606737,
|
| 1175 |
+
"learning_rate": 3.644452683288535e-05,
|
| 1176 |
+
"loss": 0.0239,
|
| 1177 |
+
"step": 4125
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 1.3847180513847182,
|
| 1181 |
+
"grad_norm": 0.004917916841804981,
|
| 1182 |
+
"learning_rate": 3.632866808786728e-05,
|
| 1183 |
+
"loss": 0.0056,
|
| 1184 |
+
"step": 4150
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 1.3930597263930597,
|
| 1188 |
+
"grad_norm": 0.00037450282252393663,
|
| 1189 |
+
"learning_rate": 3.62128093428492e-05,
|
| 1190 |
+
"loss": 0.0014,
|
| 1191 |
+
"step": 4175
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 1.4014014014014013,
|
| 1195 |
+
"grad_norm": 0.0002487332094460726,
|
| 1196 |
+
"learning_rate": 3.6096950597831124e-05,
|
| 1197 |
+
"loss": 0.0081,
|
| 1198 |
+
"step": 4200
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 1.4097430764097432,
|
| 1202 |
+
"grad_norm": 0.0002034240314969793,
|
| 1203 |
+
"learning_rate": 3.5981091852813055e-05,
|
| 1204 |
+
"loss": 0.0,
|
| 1205 |
+
"step": 4225
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 1.4180847514180848,
|
| 1209 |
+
"grad_norm": 0.0002251726109534502,
|
| 1210 |
+
"learning_rate": 3.586523310779498e-05,
|
| 1211 |
+
"loss": 0.0376,
|
| 1212 |
+
"step": 4250
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 1.4264264264264264,
|
| 1216 |
+
"grad_norm": 0.0068957749754190445,
|
| 1217 |
+
"learning_rate": 3.57493743627769e-05,
|
| 1218 |
+
"loss": 0.0547,
|
| 1219 |
+
"step": 4275
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 1.4347681014347682,
|
| 1223 |
+
"grad_norm": 0.14848369359970093,
|
| 1224 |
+
"learning_rate": 3.5633515617758826e-05,
|
| 1225 |
+
"loss": 0.0398,
|
| 1226 |
+
"step": 4300
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 1.4431097764431098,
|
| 1230 |
+
"grad_norm": 0.006616557948291302,
|
| 1231 |
+
"learning_rate": 3.5517656872740756e-05,
|
| 1232 |
+
"loss": 0.0009,
|
| 1233 |
+
"step": 4325
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 1.4514514514514514,
|
| 1237 |
+
"grad_norm": 0.0264744870364666,
|
| 1238 |
+
"learning_rate": 3.540179812772268e-05,
|
| 1239 |
+
"loss": 0.0219,
|
| 1240 |
+
"step": 4350
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 1.4597931264597932,
|
| 1244 |
+
"grad_norm": 0.011937894858419895,
|
| 1245 |
+
"learning_rate": 3.528593938270461e-05,
|
| 1246 |
+
"loss": 0.0284,
|
| 1247 |
+
"step": 4375
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 1.4681348014681348,
|
| 1251 |
+
"grad_norm": 0.00034804438473656774,
|
| 1252 |
+
"learning_rate": 3.5170080637686534e-05,
|
| 1253 |
+
"loss": 0.0027,
|
| 1254 |
+
"step": 4400
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.4764764764764764,
|
| 1258 |
+
"grad_norm": 0.00043386907782405615,
|
| 1259 |
+
"learning_rate": 3.505422189266846e-05,
|
| 1260 |
+
"loss": 0.0,
|
| 1261 |
+
"step": 4425
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 1.4848181514848182,
|
| 1265 |
+
"grad_norm": 0.0003097046574112028,
|
| 1266 |
+
"learning_rate": 3.493836314765039e-05,
|
| 1267 |
+
"loss": 0.0005,
|
| 1268 |
+
"step": 4450
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 1.4931598264931598,
|
| 1272 |
+
"grad_norm": 0.0003939344023820013,
|
| 1273 |
+
"learning_rate": 3.482250440263231e-05,
|
| 1274 |
+
"loss": 0.0002,
|
| 1275 |
+
"step": 4475
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.5015015015015014,
|
| 1279 |
+
"grad_norm": 0.00022050646657589823,
|
| 1280 |
+
"learning_rate": 3.470664565761424e-05,
|
| 1281 |
+
"loss": 0.0001,
|
| 1282 |
+
"step": 4500
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.5098431765098432,
|
| 1286 |
+
"grad_norm": 0.0003953372943215072,
|
| 1287 |
+
"learning_rate": 3.459078691259616e-05,
|
| 1288 |
+
"loss": 0.0001,
|
| 1289 |
+
"step": 4525
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 1.5181848515181848,
|
| 1293 |
+
"grad_norm": 0.00029923071269877255,
|
| 1294 |
+
"learning_rate": 3.447492816757809e-05,
|
| 1295 |
+
"loss": 0.0408,
|
| 1296 |
+
"step": 4550
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 1.5265265265265264,
|
| 1300 |
+
"grad_norm": 0.0014319141628220677,
|
| 1301 |
+
"learning_rate": 3.4359069422560014e-05,
|
| 1302 |
+
"loss": 0.0001,
|
| 1303 |
+
"step": 4575
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 1.5348682015348682,
|
| 1307 |
+
"grad_norm": 0.0010346460621804,
|
| 1308 |
+
"learning_rate": 3.4243210677541944e-05,
|
| 1309 |
+
"loss": 0.016,
|
| 1310 |
+
"step": 4600
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 1.5432098765432098,
|
| 1314 |
+
"grad_norm": 0.0012767055304720998,
|
| 1315 |
+
"learning_rate": 3.412735193252387e-05,
|
| 1316 |
+
"loss": 0.0001,
|
| 1317 |
+
"step": 4625
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.5515515515515514,
|
| 1321 |
+
"grad_norm": 0.0006332869525067508,
|
| 1322 |
+
"learning_rate": 3.401149318750579e-05,
|
| 1323 |
+
"loss": 0.0001,
|
| 1324 |
+
"step": 4650
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 1.5598932265598933,
|
| 1328 |
+
"grad_norm": 0.013449088670313358,
|
| 1329 |
+
"learning_rate": 3.389563444248772e-05,
|
| 1330 |
+
"loss": 0.0124,
|
| 1331 |
+
"step": 4675
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 1.5682349015682349,
|
| 1335 |
+
"grad_norm": 0.0019397336291149259,
|
| 1336 |
+
"learning_rate": 3.3779775697469646e-05,
|
| 1337 |
+
"loss": 0.0488,
|
| 1338 |
+
"step": 4700
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 1.5765765765765765,
|
| 1342 |
+
"grad_norm": 0.001804558327421546,
|
| 1343 |
+
"learning_rate": 3.3663916952451576e-05,
|
| 1344 |
+
"loss": 0.0001,
|
| 1345 |
+
"step": 4725
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.5849182515849183,
|
| 1349 |
+
"grad_norm": 0.0007520993240177631,
|
| 1350 |
+
"learning_rate": 3.35480582074335e-05,
|
| 1351 |
+
"loss": 0.0005,
|
| 1352 |
+
"step": 4750
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 1.5932599265932599,
|
| 1356 |
+
"grad_norm": 0.0034453764092177153,
|
| 1357 |
+
"learning_rate": 3.3432199462415424e-05,
|
| 1358 |
+
"loss": 0.0,
|
| 1359 |
+
"step": 4775
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.6016016016016015,
|
| 1363 |
+
"grad_norm": 0.0007546308333985507,
|
| 1364 |
+
"learning_rate": 3.331634071739735e-05,
|
| 1365 |
+
"loss": 0.0284,
|
| 1366 |
+
"step": 4800
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 1.6099432766099433,
|
| 1370 |
+
"grad_norm": 0.018241139128804207,
|
| 1371 |
+
"learning_rate": 3.320048197237928e-05,
|
| 1372 |
+
"loss": 0.0,
|
| 1373 |
+
"step": 4825
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 1.6182849516182851,
|
| 1377 |
+
"grad_norm": 0.0001632855273783207,
|
| 1378 |
+
"learning_rate": 3.308462322736121e-05,
|
| 1379 |
+
"loss": 0.0007,
|
| 1380 |
+
"step": 4850
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 1.6266266266266265,
|
| 1384 |
+
"grad_norm": 0.0043875593692064285,
|
| 1385 |
+
"learning_rate": 3.2968764482343125e-05,
|
| 1386 |
+
"loss": 0.0,
|
| 1387 |
+
"step": 4875
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 1.6349683016349683,
|
| 1391 |
+
"grad_norm": 0.0005138323176652193,
|
| 1392 |
+
"learning_rate": 3.2852905737325056e-05,
|
| 1393 |
+
"loss": 0.0322,
|
| 1394 |
+
"step": 4900
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.6433099766433101,
|
| 1398 |
+
"grad_norm": 0.0033361820969730616,
|
| 1399 |
+
"learning_rate": 3.273704699230698e-05,
|
| 1400 |
+
"loss": 0.0285,
|
| 1401 |
+
"step": 4925
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.6516516516516515,
|
| 1405 |
+
"grad_norm": 0.0008180899312719703,
|
| 1406 |
+
"learning_rate": 3.262118824728891e-05,
|
| 1407 |
+
"loss": 0.0001,
|
| 1408 |
+
"step": 4950
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 1.6599933266599933,
|
| 1412 |
+
"grad_norm": 0.0002278868923895061,
|
| 1413 |
+
"learning_rate": 3.2505329502270834e-05,
|
| 1414 |
+
"loss": 0.0008,
|
| 1415 |
+
"step": 4975
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 1.6683350016683351,
|
| 1419 |
+
"grad_norm": 0.0010708553018048406,
|
| 1420 |
+
"learning_rate": 3.238947075725276e-05,
|
| 1421 |
+
"loss": 0.0001,
|
| 1422 |
+
"step": 5000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 1.6766766766766765,
|
| 1426 |
+
"grad_norm": 0.0006777086528018117,
|
| 1427 |
+
"learning_rate": 3.227361201223468e-05,
|
| 1428 |
+
"loss": 0.0,
|
| 1429 |
+
"step": 5025
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 1.6850183516850183,
|
| 1433 |
+
"grad_norm": 0.0007990729645825922,
|
| 1434 |
+
"learning_rate": 3.215775326721661e-05,
|
| 1435 |
+
"loss": 0.0588,
|
| 1436 |
+
"step": 5050
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 1.6933600266933602,
|
| 1440 |
+
"grad_norm": 0.00486621493473649,
|
| 1441 |
+
"learning_rate": 3.2041894522198535e-05,
|
| 1442 |
+
"loss": 0.0109,
|
| 1443 |
+
"step": 5075
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 1.7017017017017015,
|
| 1447 |
+
"grad_norm": 0.020536229014396667,
|
| 1448 |
+
"learning_rate": 3.1926035777180466e-05,
|
| 1449 |
+
"loss": 0.0241,
|
| 1450 |
+
"step": 5100
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.7100433767100434,
|
| 1454 |
+
"grad_norm": 0.003294334514066577,
|
| 1455 |
+
"learning_rate": 3.181017703216239e-05,
|
| 1456 |
+
"loss": 0.023,
|
| 1457 |
+
"step": 5125
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.7183850517183852,
|
| 1461 |
+
"grad_norm": 0.0024423557333648205,
|
| 1462 |
+
"learning_rate": 3.169431828714431e-05,
|
| 1463 |
+
"loss": 0.0002,
|
| 1464 |
+
"step": 5150
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.7267267267267268,
|
| 1468 |
+
"grad_norm": 0.010359351523220539,
|
| 1469 |
+
"learning_rate": 3.1578459542126244e-05,
|
| 1470 |
+
"loss": 0.0208,
|
| 1471 |
+
"step": 5175
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 1.7350684017350684,
|
| 1475 |
+
"grad_norm": 0.008612546138465405,
|
| 1476 |
+
"learning_rate": 3.146260079710817e-05,
|
| 1477 |
+
"loss": 0.0002,
|
| 1478 |
+
"step": 5200
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 1.7434100767434102,
|
| 1482 |
+
"grad_norm": 0.0005397327477112412,
|
| 1483 |
+
"learning_rate": 3.13467420520901e-05,
|
| 1484 |
+
"loss": 0.0078,
|
| 1485 |
+
"step": 5225
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 1.7517517517517518,
|
| 1489 |
+
"grad_norm": 0.00102256890386343,
|
| 1490 |
+
"learning_rate": 3.1230883307072015e-05,
|
| 1491 |
+
"loss": 0.0,
|
| 1492 |
+
"step": 5250
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 1.7600934267600934,
|
| 1496 |
+
"grad_norm": 0.004368093330413103,
|
| 1497 |
+
"learning_rate": 3.1115024562053945e-05,
|
| 1498 |
+
"loss": 0.0485,
|
| 1499 |
+
"step": 5275
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 1.7684351017684352,
|
| 1503 |
+
"grad_norm": 0.0030503266025334597,
|
| 1504 |
+
"learning_rate": 3.099916581703587e-05,
|
| 1505 |
+
"loss": 0.0002,
|
| 1506 |
+
"step": 5300
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.7767767767767768,
|
| 1510 |
+
"grad_norm": 0.005283331032842398,
|
| 1511 |
+
"learning_rate": 3.08833070720178e-05,
|
| 1512 |
+
"loss": 0.0003,
|
| 1513 |
+
"step": 5325
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 1.7851184517851184,
|
| 1517 |
+
"grad_norm": 0.004458490759134293,
|
| 1518 |
+
"learning_rate": 3.077208267680045e-05,
|
| 1519 |
+
"loss": 0.0769,
|
| 1520 |
+
"step": 5350
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 1.7934601267934602,
|
| 1524 |
+
"grad_norm": 0.015544002875685692,
|
| 1525 |
+
"learning_rate": 3.065622393178238e-05,
|
| 1526 |
+
"loss": 0.0005,
|
| 1527 |
+
"step": 5375
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.8018018018018018,
|
| 1531 |
+
"grad_norm": 0.01679283380508423,
|
| 1532 |
+
"learning_rate": 3.0540365186764294e-05,
|
| 1533 |
+
"loss": 0.0006,
|
| 1534 |
+
"step": 5400
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 1.8101434768101434,
|
| 1538 |
+
"grad_norm": 0.0037997430190443993,
|
| 1539 |
+
"learning_rate": 3.0424506441746225e-05,
|
| 1540 |
+
"loss": 0.0217,
|
| 1541 |
+
"step": 5425
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 1.8184851518184852,
|
| 1545 |
+
"grad_norm": 0.003071027109399438,
|
| 1546 |
+
"learning_rate": 3.030864769672815e-05,
|
| 1547 |
+
"loss": 0.0001,
|
| 1548 |
+
"step": 5450
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 1.8268268268268268,
|
| 1552 |
+
"grad_norm": 0.0016744754975661635,
|
| 1553 |
+
"learning_rate": 3.0192788951710076e-05,
|
| 1554 |
+
"loss": 0.0001,
|
| 1555 |
+
"step": 5475
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.8351685018351684,
|
| 1559 |
+
"grad_norm": 0.019519317895174026,
|
| 1560 |
+
"learning_rate": 3.0076930206692e-05,
|
| 1561 |
+
"loss": 0.0474,
|
| 1562 |
+
"step": 5500
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.8435101768435103,
|
| 1566 |
+
"grad_norm": 0.016507979482412338,
|
| 1567 |
+
"learning_rate": 2.996107146167393e-05,
|
| 1568 |
+
"loss": 0.0004,
|
| 1569 |
+
"step": 5525
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.8518518518518519,
|
| 1573 |
+
"grad_norm": 0.002733981003984809,
|
| 1574 |
+
"learning_rate": 2.9845212716655857e-05,
|
| 1575 |
+
"loss": 0.0049,
|
| 1576 |
+
"step": 5550
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 1.8601935268601935,
|
| 1580 |
+
"grad_norm": 0.002405300736427307,
|
| 1581 |
+
"learning_rate": 2.972935397163778e-05,
|
| 1582 |
+
"loss": 0.0012,
|
| 1583 |
+
"step": 5575
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.8685352018685353,
|
| 1587 |
+
"grad_norm": 0.0028873877599835396,
|
| 1588 |
+
"learning_rate": 2.9613495226619708e-05,
|
| 1589 |
+
"loss": 0.0001,
|
| 1590 |
+
"step": 5600
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 1.8768768768768769,
|
| 1594 |
+
"grad_norm": 0.0029900181107223034,
|
| 1595 |
+
"learning_rate": 2.949763648160163e-05,
|
| 1596 |
+
"loss": 0.0204,
|
| 1597 |
+
"step": 5625
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.8852185518852185,
|
| 1601 |
+
"grad_norm": 0.0025838094297796488,
|
| 1602 |
+
"learning_rate": 2.938177773658356e-05,
|
| 1603 |
+
"loss": 0.0001,
|
| 1604 |
+
"step": 5650
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 1.8935602268935603,
|
| 1608 |
+
"grad_norm": 0.0009950456442311406,
|
| 1609 |
+
"learning_rate": 2.9265918991565482e-05,
|
| 1610 |
+
"loss": 0.0,
|
| 1611 |
+
"step": 5675
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.901901901901902,
|
| 1615 |
+
"grad_norm": 0.0015871904324740171,
|
| 1616 |
+
"learning_rate": 2.9150060246547413e-05,
|
| 1617 |
+
"loss": 0.0002,
|
| 1618 |
+
"step": 5700
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.9102435769102435,
|
| 1622 |
+
"grad_norm": 0.0018013713415712118,
|
| 1623 |
+
"learning_rate": 2.9034201501529333e-05,
|
| 1624 |
+
"loss": 0.0676,
|
| 1625 |
+
"step": 5725
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.9185852519185853,
|
| 1629 |
+
"grad_norm": 0.006069496739655733,
|
| 1630 |
+
"learning_rate": 2.8918342756511263e-05,
|
| 1631 |
+
"loss": 0.0378,
|
| 1632 |
+
"step": 5750
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 1.926926926926927,
|
| 1636 |
+
"grad_norm": 0.0046617318876087666,
|
| 1637 |
+
"learning_rate": 2.880248401149319e-05,
|
| 1638 |
+
"loss": 0.0037,
|
| 1639 |
+
"step": 5775
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 1.9352686019352685,
|
| 1643 |
+
"grad_norm": 0.0014451640890911222,
|
| 1644 |
+
"learning_rate": 2.8686625266475114e-05,
|
| 1645 |
+
"loss": 0.0003,
|
| 1646 |
+
"step": 5800
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 1.9436102769436103,
|
| 1650 |
+
"grad_norm": 0.0014788233675062656,
|
| 1651 |
+
"learning_rate": 2.8570766521457045e-05,
|
| 1652 |
+
"loss": 0.0266,
|
| 1653 |
+
"step": 5825
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.951951951951952,
|
| 1657 |
+
"grad_norm": 0.0011505110887810588,
|
| 1658 |
+
"learning_rate": 2.8454907776438965e-05,
|
| 1659 |
+
"loss": 0.0,
|
| 1660 |
+
"step": 5850
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.9602936269602935,
|
| 1664 |
+
"grad_norm": 0.004947907291352749,
|
| 1665 |
+
"learning_rate": 2.8339049031420895e-05,
|
| 1666 |
+
"loss": 0.0098,
|
| 1667 |
+
"step": 5875
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 1.9686353019686353,
|
| 1671 |
+
"grad_norm": 0.0015439976705238223,
|
| 1672 |
+
"learning_rate": 2.8223190286402816e-05,
|
| 1673 |
+
"loss": 0.0001,
|
| 1674 |
+
"step": 5900
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 1.976976976976977,
|
| 1678 |
+
"grad_norm": 0.003472746815532446,
|
| 1679 |
+
"learning_rate": 2.8107331541384746e-05,
|
| 1680 |
+
"loss": 0.0001,
|
| 1681 |
+
"step": 5925
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 1.9853186519853185,
|
| 1685 |
+
"grad_norm": 0.0020330217666924,
|
| 1686 |
+
"learning_rate": 2.799147279636667e-05,
|
| 1687 |
+
"loss": 0.0001,
|
| 1688 |
+
"step": 5950
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 1.9936603269936604,
|
| 1692 |
+
"grad_norm": 0.000714894209522754,
|
| 1693 |
+
"learning_rate": 2.7875614051348597e-05,
|
| 1694 |
+
"loss": 0.0005,
|
| 1695 |
+
"step": 5975
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 2.0,
|
| 1699 |
+
"eval_accuracy": 0.997997997997998,
|
| 1700 |
+
"eval_auc": 0.9992763878743608,
|
| 1701 |
+
"eval_f1": 0.9142857142857143,
|
| 1702 |
+
"eval_loss": 0.009895329363644123,
|
| 1703 |
+
"eval_precision": 0.9696969696969697,
|
| 1704 |
+
"eval_recall": 0.8648648648648649,
|
| 1705 |
+
"eval_runtime": 78.0253,
|
| 1706 |
+
"eval_samples_per_second": 76.821,
|
| 1707 |
+
"eval_steps_per_second": 4.806,
|
| 1708 |
+
"step": 5994
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 2.002002002002002,
|
| 1712 |
+
"grad_norm": 0.0006246891571208835,
|
| 1713 |
+
"learning_rate": 2.775975530633052e-05,
|
| 1714 |
+
"loss": 0.0,
|
| 1715 |
+
"step": 6000
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 2.0103436770103436,
|
| 1719 |
+
"grad_norm": 0.009155196137726307,
|
| 1720 |
+
"learning_rate": 2.7643896561312448e-05,
|
| 1721 |
+
"loss": 0.0,
|
| 1722 |
+
"step": 6025
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 2.0186853520186854,
|
| 1726 |
+
"grad_norm": 0.0010195561917498708,
|
| 1727 |
+
"learning_rate": 2.752803781629438e-05,
|
| 1728 |
+
"loss": 0.0005,
|
| 1729 |
+
"step": 6050
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 2.027027027027027,
|
| 1733 |
+
"grad_norm": 0.00037840736331418157,
|
| 1734 |
+
"learning_rate": 2.7412179071276302e-05,
|
| 1735 |
+
"loss": 0.0,
|
| 1736 |
+
"step": 6075
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 2.0353687020353686,
|
| 1740 |
+
"grad_norm": 2.285806655883789,
|
| 1741 |
+
"learning_rate": 2.729632032625823e-05,
|
| 1742 |
+
"loss": 0.0008,
|
| 1743 |
+
"step": 6100
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 2.0437103770437104,
|
| 1747 |
+
"grad_norm": 0.000318177102599293,
|
| 1748 |
+
"learning_rate": 2.7180461581240153e-05,
|
| 1749 |
+
"loss": 0.0,
|
| 1750 |
+
"step": 6125
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 2.052052052052052,
|
| 1754 |
+
"grad_norm": 0.00042503979057073593,
|
| 1755 |
+
"learning_rate": 2.706460283622208e-05,
|
| 1756 |
+
"loss": 0.0001,
|
| 1757 |
+
"step": 6150
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 2.0603937270603936,
|
| 1761 |
+
"grad_norm": 0.0004734281974378973,
|
| 1762 |
+
"learning_rate": 2.6948744091204004e-05,
|
| 1763 |
+
"loss": 0.0,
|
| 1764 |
+
"step": 6175
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 2.0687354020687354,
|
| 1768 |
+
"grad_norm": 0.00044175496441312134,
|
| 1769 |
+
"learning_rate": 2.683288534618593e-05,
|
| 1770 |
+
"loss": 0.0001,
|
| 1771 |
+
"step": 6200
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 2.0770770770770772,
|
| 1775 |
+
"grad_norm": 0.0005970322526991367,
|
| 1776 |
+
"learning_rate": 2.6717026601167854e-05,
|
| 1777 |
+
"loss": 0.0,
|
| 1778 |
+
"step": 6225
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 2.0854187520854186,
|
| 1782 |
+
"grad_norm": 0.00034779220004566014,
|
| 1783 |
+
"learning_rate": 2.6601167856149785e-05,
|
| 1784 |
+
"loss": 0.0001,
|
| 1785 |
+
"step": 6250
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 2.0937604270937604,
|
| 1789 |
+
"grad_norm": 0.0002642270992510021,
|
| 1790 |
+
"learning_rate": 2.6485309111131712e-05,
|
| 1791 |
+
"loss": 0.0046,
|
| 1792 |
+
"step": 6275
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"epoch": 2.1021021021021022,
|
| 1796 |
+
"grad_norm": 0.0005227423971518874,
|
| 1797 |
+
"learning_rate": 2.6369450366113636e-05,
|
| 1798 |
+
"loss": 0.004,
|
| 1799 |
+
"step": 6300
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 2.1104437771104436,
|
| 1803 |
+
"grad_norm": 0.0027825473807752132,
|
| 1804 |
+
"learning_rate": 2.6253591621095563e-05,
|
| 1805 |
+
"loss": 0.0441,
|
| 1806 |
+
"step": 6325
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 2.1187854521187854,
|
| 1810 |
+
"grad_norm": 0.0015170240076258779,
|
| 1811 |
+
"learning_rate": 2.6137732876077486e-05,
|
| 1812 |
+
"loss": 0.0004,
|
| 1813 |
+
"step": 6350
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 2.1271271271271273,
|
| 1817 |
+
"grad_norm": 0.0009673828026279807,
|
| 1818 |
+
"learning_rate": 2.6021874131059414e-05,
|
| 1819 |
+
"loss": 0.0001,
|
| 1820 |
+
"step": 6375
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 2.1354688021354686,
|
| 1824 |
+
"grad_norm": 0.0016775678377598524,
|
| 1825 |
+
"learning_rate": 2.5906015386041337e-05,
|
| 1826 |
+
"loss": 0.0033,
|
| 1827 |
+
"step": 6400
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 2.1438104771438105,
|
| 1831 |
+
"grad_norm": 0.4511735737323761,
|
| 1832 |
+
"learning_rate": 2.5790156641023268e-05,
|
| 1833 |
+
"loss": 0.0226,
|
| 1834 |
+
"step": 6425
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 2.1521521521521523,
|
| 1838 |
+
"grad_norm": 0.00039455911610275507,
|
| 1839 |
+
"learning_rate": 2.5674297896005188e-05,
|
| 1840 |
+
"loss": 0.0005,
|
| 1841 |
+
"step": 6450
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 2.1604938271604937,
|
| 1845 |
+
"grad_norm": 0.0009889901848509908,
|
| 1846 |
+
"learning_rate": 2.555843915098712e-05,
|
| 1847 |
+
"loss": 0.0298,
|
| 1848 |
+
"step": 6475
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 2.1688355021688355,
|
| 1852 |
+
"grad_norm": 0.0007957897614687681,
|
| 1853 |
+
"learning_rate": 2.5442580405969046e-05,
|
| 1854 |
+
"loss": 0.0242,
|
| 1855 |
+
"step": 6500
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 2.1771771771771773,
|
| 1859 |
+
"grad_norm": 0.00022962476941756904,
|
| 1860 |
+
"learning_rate": 2.532672166095097e-05,
|
| 1861 |
+
"loss": 0.0,
|
| 1862 |
+
"step": 6525
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 2.1855188521855187,
|
| 1866 |
+
"grad_norm": 0.0003863972960971296,
|
| 1867 |
+
"learning_rate": 2.5210862915932896e-05,
|
| 1868 |
+
"loss": 0.0,
|
| 1869 |
+
"step": 6550
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 2.1938605271938605,
|
| 1873 |
+
"grad_norm": 0.0011119615519419312,
|
| 1874 |
+
"learning_rate": 2.509500417091482e-05,
|
| 1875 |
+
"loss": 0.0001,
|
| 1876 |
+
"step": 6575
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 2.2022022022022023,
|
| 1880 |
+
"grad_norm": 0.0002698484167922288,
|
| 1881 |
+
"learning_rate": 2.4979145425896747e-05,
|
| 1882 |
+
"loss": 0.0001,
|
| 1883 |
+
"step": 6600
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 2.2105438772105437,
|
| 1887 |
+
"grad_norm": 0.0003634936292655766,
|
| 1888 |
+
"learning_rate": 2.4863286680878674e-05,
|
| 1889 |
+
"loss": 0.0,
|
| 1890 |
+
"step": 6625
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 2.2188855522188855,
|
| 1894 |
+
"grad_norm": 0.10544054955244064,
|
| 1895 |
+
"learning_rate": 2.47474279358606e-05,
|
| 1896 |
+
"loss": 0.0729,
|
| 1897 |
+
"step": 6650
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 2.2272272272272273,
|
| 1901 |
+
"grad_norm": 0.0028164817485958338,
|
| 1902 |
+
"learning_rate": 2.4631569190842525e-05,
|
| 1903 |
+
"loss": 0.0194,
|
| 1904 |
+
"step": 6675
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 2.2355689022355687,
|
| 1908 |
+
"grad_norm": 0.0028918420430272818,
|
| 1909 |
+
"learning_rate": 2.4515710445824452e-05,
|
| 1910 |
+
"loss": 0.0004,
|
| 1911 |
+
"step": 6700
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 2.2439105772439105,
|
| 1915 |
+
"grad_norm": 0.003687906078994274,
|
| 1916 |
+
"learning_rate": 2.439985170080638e-05,
|
| 1917 |
+
"loss": 0.0001,
|
| 1918 |
+
"step": 6725
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 2.2522522522522523,
|
| 1922 |
+
"grad_norm": 0.0016472985735163093,
|
| 1923 |
+
"learning_rate": 2.4283992955788303e-05,
|
| 1924 |
+
"loss": 0.0016,
|
| 1925 |
+
"step": 6750
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 2.2605939272605937,
|
| 1929 |
+
"grad_norm": 0.0029458694625645876,
|
| 1930 |
+
"learning_rate": 2.416813421077023e-05,
|
| 1931 |
+
"loss": 0.0001,
|
| 1932 |
+
"step": 6775
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 2.2689356022689355,
|
| 1936 |
+
"grad_norm": 0.0009019913850352168,
|
| 1937 |
+
"learning_rate": 2.4052275465752154e-05,
|
| 1938 |
+
"loss": 0.0001,
|
| 1939 |
+
"step": 6800
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 2.2772772772772774,
|
| 1943 |
+
"grad_norm": 0.0009054698166437447,
|
| 1944 |
+
"learning_rate": 2.393641672073408e-05,
|
| 1945 |
+
"loss": 0.0001,
|
| 1946 |
+
"step": 6825
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 2.2856189522856187,
|
| 1950 |
+
"grad_norm": 0.001989714102819562,
|
| 1951 |
+
"learning_rate": 2.3820557975716008e-05,
|
| 1952 |
+
"loss": 0.0001,
|
| 1953 |
+
"step": 6850
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 2.2939606272939606,
|
| 1957 |
+
"grad_norm": 0.0007018431788310409,
|
| 1958 |
+
"learning_rate": 2.3704699230697935e-05,
|
| 1959 |
+
"loss": 0.0001,
|
| 1960 |
+
"step": 6875
|
| 1961 |
+
},
|
| 1962 |
+
{
|
| 1963 |
+
"epoch": 2.3023023023023024,
|
| 1964 |
+
"grad_norm": 0.0006220221403054893,
|
| 1965 |
+
"learning_rate": 2.3588840485679862e-05,
|
| 1966 |
+
"loss": 0.0001,
|
| 1967 |
+
"step": 6900
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 2.3106439773106437,
|
| 1971 |
+
"grad_norm": 0.001839311677031219,
|
| 1972 |
+
"learning_rate": 2.3472981740661786e-05,
|
| 1973 |
+
"loss": 0.0,
|
| 1974 |
+
"step": 6925
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 2.3189856523189856,
|
| 1978 |
+
"grad_norm": 0.0005744717200286686,
|
| 1979 |
+
"learning_rate": 2.3357122995643713e-05,
|
| 1980 |
+
"loss": 0.0,
|
| 1981 |
+
"step": 6950
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 2.3273273273273274,
|
| 1985 |
+
"grad_norm": 0.0004123161779716611,
|
| 1986 |
+
"learning_rate": 2.324126425062564e-05,
|
| 1987 |
+
"loss": 0.0001,
|
| 1988 |
+
"step": 6975
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 2.3356690023356688,
|
| 1992 |
+
"grad_norm": 0.0013543826062232256,
|
| 1993 |
+
"learning_rate": 2.3125405505607564e-05,
|
| 1994 |
+
"loss": 0.0,
|
| 1995 |
+
"step": 7000
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 2.3440106773440106,
|
| 1999 |
+
"grad_norm": 0.001142465160228312,
|
| 2000 |
+
"learning_rate": 2.300954676058949e-05,
|
| 2001 |
+
"loss": 0.0198,
|
| 2002 |
+
"step": 7025
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 2.3523523523523524,
|
| 2006 |
+
"grad_norm": 0.0012254125904291868,
|
| 2007 |
+
"learning_rate": 2.2893688015571414e-05,
|
| 2008 |
+
"loss": 0.0008,
|
| 2009 |
+
"step": 7050
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 2.360694027360694,
|
| 2013 |
+
"grad_norm": 0.0004832675331272185,
|
| 2014 |
+
"learning_rate": 2.277782927055334e-05,
|
| 2015 |
+
"loss": 0.0154,
|
| 2016 |
+
"step": 7075
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 2.3690357023690356,
|
| 2020 |
+
"grad_norm": 0.00016073780716396868,
|
| 2021 |
+
"learning_rate": 2.266197052553527e-05,
|
| 2022 |
+
"loss": 0.0,
|
| 2023 |
+
"step": 7100
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 2.3773773773773774,
|
| 2027 |
+
"grad_norm": 0.002141432836651802,
|
| 2028 |
+
"learning_rate": 2.2546111780517196e-05,
|
| 2029 |
+
"loss": 0.0001,
|
| 2030 |
+
"step": 7125
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 2.3857190523857192,
|
| 2034 |
+
"grad_norm": 0.0022791263181716204,
|
| 2035 |
+
"learning_rate": 2.2430253035499123e-05,
|
| 2036 |
+
"loss": 0.0442,
|
| 2037 |
+
"step": 7150
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 2.3940607273940606,
|
| 2041 |
+
"grad_norm": 0.0010367584181949496,
|
| 2042 |
+
"learning_rate": 2.2314394290481047e-05,
|
| 2043 |
+
"loss": 0.0004,
|
| 2044 |
+
"step": 7175
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 2.4024024024024024,
|
| 2048 |
+
"grad_norm": 0.014609686098992825,
|
| 2049 |
+
"learning_rate": 2.2198535545462974e-05,
|
| 2050 |
+
"loss": 0.0001,
|
| 2051 |
+
"step": 7200
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 2.4107440774107443,
|
| 2055 |
+
"grad_norm": 0.0009007493499666452,
|
| 2056 |
+
"learning_rate": 2.2082676800444897e-05,
|
| 2057 |
+
"loss": 0.0001,
|
| 2058 |
+
"step": 7225
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 2.4190857524190856,
|
| 2062 |
+
"grad_norm": 0.005582588724792004,
|
| 2063 |
+
"learning_rate": 2.1966818055426824e-05,
|
| 2064 |
+
"loss": 0.0416,
|
| 2065 |
+
"step": 7250
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 2.4274274274274275,
|
| 2069 |
+
"grad_norm": 0.030880291014909744,
|
| 2070 |
+
"learning_rate": 2.185095931040875e-05,
|
| 2071 |
+
"loss": 0.0002,
|
| 2072 |
+
"step": 7275
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 2.4357691024357693,
|
| 2076 |
+
"grad_norm": 0.007195931393653154,
|
| 2077 |
+
"learning_rate": 2.1735100565390675e-05,
|
| 2078 |
+
"loss": 0.0019,
|
| 2079 |
+
"step": 7300
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 2.4441107774441106,
|
| 2083 |
+
"grad_norm": 0.0043434868566691875,
|
| 2084 |
+
"learning_rate": 2.1619241820372602e-05,
|
| 2085 |
+
"loss": 0.0003,
|
| 2086 |
+
"step": 7325
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 2.4524524524524525,
|
| 2090 |
+
"grad_norm": 0.00130070885643363,
|
| 2091 |
+
"learning_rate": 2.150338307535453e-05,
|
| 2092 |
+
"loss": 0.0001,
|
| 2093 |
+
"step": 7350
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 2.4607941274607943,
|
| 2097 |
+
"grad_norm": 0.011537245474755764,
|
| 2098 |
+
"learning_rate": 2.1387524330336456e-05,
|
| 2099 |
+
"loss": 0.0001,
|
| 2100 |
+
"step": 7375
|
| 2101 |
+
},
|
| 2102 |
+
{
|
| 2103 |
+
"epoch": 2.4691358024691357,
|
| 2104 |
+
"grad_norm": 0.0012115085264667869,
|
| 2105 |
+
"learning_rate": 2.127166558531838e-05,
|
| 2106 |
+
"loss": 0.0001,
|
| 2107 |
+
"step": 7400
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 2.4774774774774775,
|
| 2111 |
+
"grad_norm": 0.0006190771819092333,
|
| 2112 |
+
"learning_rate": 2.1155806840300307e-05,
|
| 2113 |
+
"loss": 0.0001,
|
| 2114 |
+
"step": 7425
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 2.4858191524858193,
|
| 2118 |
+
"grad_norm": 0.0007142933318391442,
|
| 2119 |
+
"learning_rate": 2.1039948095282234e-05,
|
| 2120 |
+
"loss": 0.0,
|
| 2121 |
+
"step": 7450
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 2.4941608274941607,
|
| 2125 |
+
"grad_norm": 0.0024094288237392902,
|
| 2126 |
+
"learning_rate": 2.0924089350264158e-05,
|
| 2127 |
+
"loss": 0.0349,
|
| 2128 |
+
"step": 7475
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 2.5025025025025025,
|
| 2132 |
+
"grad_norm": 0.0028155215550214052,
|
| 2133 |
+
"learning_rate": 2.0808230605246085e-05,
|
| 2134 |
+
"loss": 0.0349,
|
| 2135 |
+
"step": 7500
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 2.5108441775108443,
|
| 2139 |
+
"grad_norm": 0.00238401023671031,
|
| 2140 |
+
"learning_rate": 2.069237186022801e-05,
|
| 2141 |
+
"loss": 0.0087,
|
| 2142 |
+
"step": 7525
|
| 2143 |
+
},
|
| 2144 |
+
{
|
| 2145 |
+
"epoch": 2.5191858525191857,
|
| 2146 |
+
"grad_norm": 0.0014537558890879154,
|
| 2147 |
+
"learning_rate": 2.0576513115209936e-05,
|
| 2148 |
+
"loss": 0.0033,
|
| 2149 |
+
"step": 7550
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"epoch": 2.5275275275275275,
|
| 2153 |
+
"grad_norm": 0.0015429011546075344,
|
| 2154 |
+
"learning_rate": 2.0460654370191863e-05,
|
| 2155 |
+
"loss": 0.0143,
|
| 2156 |
+
"step": 7575
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 2.5358692025358693,
|
| 2160 |
+
"grad_norm": 0.0013287997571751475,
|
| 2161 |
+
"learning_rate": 2.034479562517379e-05,
|
| 2162 |
+
"loss": 0.0014,
|
| 2163 |
+
"step": 7600
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 2.5442108775442107,
|
| 2167 |
+
"grad_norm": 0.0010292051592841744,
|
| 2168 |
+
"learning_rate": 2.0228936880155717e-05,
|
| 2169 |
+
"loss": 0.0049,
|
| 2170 |
+
"step": 7625
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 2.5525525525525525,
|
| 2174 |
+
"grad_norm": 0.007215111516416073,
|
| 2175 |
+
"learning_rate": 2.011307813513764e-05,
|
| 2176 |
+
"loss": 0.005,
|
| 2177 |
+
"step": 7650
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 2.5608942275608944,
|
| 2181 |
+
"grad_norm": 0.007656760048121214,
|
| 2182 |
+
"learning_rate": 1.9997219390119568e-05,
|
| 2183 |
+
"loss": 0.0355,
|
| 2184 |
+
"step": 7675
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 2.569235902569236,
|
| 2188 |
+
"grad_norm": 0.001918874098919332,
|
| 2189 |
+
"learning_rate": 1.9881360645101492e-05,
|
| 2190 |
+
"loss": 0.0045,
|
| 2191 |
+
"step": 7700
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 2.5775775775775776,
|
| 2195 |
+
"grad_norm": 0.0005019979435019195,
|
| 2196 |
+
"learning_rate": 1.976550190008342e-05,
|
| 2197 |
+
"loss": 0.0001,
|
| 2198 |
+
"step": 7725
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 2.5859192525859194,
|
| 2202 |
+
"grad_norm": 0.00030773007893003523,
|
| 2203 |
+
"learning_rate": 1.9649643155065346e-05,
|
| 2204 |
+
"loss": 0.0001,
|
| 2205 |
+
"step": 7750
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 2.594260927594261,
|
| 2209 |
+
"grad_norm": 0.0003611762367654592,
|
| 2210 |
+
"learning_rate": 1.953378441004727e-05,
|
| 2211 |
+
"loss": 0.0001,
|
| 2212 |
+
"step": 7775
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 2.6026026026026026,
|
| 2216 |
+
"grad_norm": 0.0019309103954583406,
|
| 2217 |
+
"learning_rate": 1.9417925665029197e-05,
|
| 2218 |
+
"loss": 0.0,
|
| 2219 |
+
"step": 7800
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 2.6109442776109444,
|
| 2223 |
+
"grad_norm": 0.004138192627578974,
|
| 2224 |
+
"learning_rate": 1.930206692001112e-05,
|
| 2225 |
+
"loss": 0.0533,
|
| 2226 |
+
"step": 7825
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 2.619285952619286,
|
| 2230 |
+
"grad_norm": 0.042996328324079514,
|
| 2231 |
+
"learning_rate": 1.918620817499305e-05,
|
| 2232 |
+
"loss": 0.0001,
|
| 2233 |
+
"step": 7850
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 2.6276276276276276,
|
| 2237 |
+
"grad_norm": 0.0017391805304214358,
|
| 2238 |
+
"learning_rate": 1.9070349429974978e-05,
|
| 2239 |
+
"loss": 0.0001,
|
| 2240 |
+
"step": 7875
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 2.6359693026359694,
|
| 2244 |
+
"grad_norm": 0.002806081436574459,
|
| 2245 |
+
"learning_rate": 1.89544906849569e-05,
|
| 2246 |
+
"loss": 0.0002,
|
| 2247 |
+
"step": 7900
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 2.6443109776443112,
|
| 2251 |
+
"grad_norm": 0.0011362434597685933,
|
| 2252 |
+
"learning_rate": 1.883863193993883e-05,
|
| 2253 |
+
"loss": 0.0001,
|
| 2254 |
+
"step": 7925
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"epoch": 2.6526526526526526,
|
| 2258 |
+
"grad_norm": 0.0013199028326198459,
|
| 2259 |
+
"learning_rate": 1.8722773194920752e-05,
|
| 2260 |
+
"loss": 0.0001,
|
| 2261 |
+
"step": 7950
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 2.6609943276609944,
|
| 2265 |
+
"grad_norm": 0.0009320782264694571,
|
| 2266 |
+
"learning_rate": 1.860691444990268e-05,
|
| 2267 |
+
"loss": 0.0001,
|
| 2268 |
+
"step": 7975
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 2.6693360026693362,
|
| 2272 |
+
"grad_norm": 0.0012447205372154713,
|
| 2273 |
+
"learning_rate": 1.8491055704884607e-05,
|
| 2274 |
+
"loss": 0.0001,
|
| 2275 |
+
"step": 8000
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 2.6776776776776776,
|
| 2279 |
+
"grad_norm": 0.0005381664377637208,
|
| 2280 |
+
"learning_rate": 1.837519695986653e-05,
|
| 2281 |
+
"loss": 0.0,
|
| 2282 |
+
"step": 8025
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 2.6860193526860194,
|
| 2286 |
+
"grad_norm": 0.0014626341871917248,
|
| 2287 |
+
"learning_rate": 1.8259338214848457e-05,
|
| 2288 |
+
"loss": 0.0002,
|
| 2289 |
+
"step": 8050
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 2.6943610276943613,
|
| 2293 |
+
"grad_norm": 0.0006017221603542566,
|
| 2294 |
+
"learning_rate": 1.8143479469830384e-05,
|
| 2295 |
+
"loss": 0.0001,
|
| 2296 |
+
"step": 8075
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 2.7027027027027026,
|
| 2300 |
+
"grad_norm": 0.0003165393427480012,
|
| 2301 |
+
"learning_rate": 1.802762072481231e-05,
|
| 2302 |
+
"loss": 0.0023,
|
| 2303 |
+
"step": 8100
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 2.7110443777110445,
|
| 2307 |
+
"grad_norm": 0.0005794435273855925,
|
| 2308 |
+
"learning_rate": 1.7911761979794235e-05,
|
| 2309 |
+
"loss": 0.0001,
|
| 2310 |
+
"step": 8125
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 2.7193860527193863,
|
| 2314 |
+
"grad_norm": 0.0025553391315042973,
|
| 2315 |
+
"learning_rate": 1.7795903234776162e-05,
|
| 2316 |
+
"loss": 0.0003,
|
| 2317 |
+
"step": 8150
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 2.7277277277277276,
|
| 2321 |
+
"grad_norm": 0.0002689988468773663,
|
| 2322 |
+
"learning_rate": 1.768004448975809e-05,
|
| 2323 |
+
"loss": 0.0,
|
| 2324 |
+
"step": 8175
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 2.7360694027360695,
|
| 2328 |
+
"grad_norm": 0.0001785249332897365,
|
| 2329 |
+
"learning_rate": 1.7564185744740013e-05,
|
| 2330 |
+
"loss": 0.0,
|
| 2331 |
+
"step": 8200
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 2.7444110777444113,
|
| 2335 |
+
"grad_norm": 0.000306436704704538,
|
| 2336 |
+
"learning_rate": 1.744832699972194e-05,
|
| 2337 |
+
"loss": 0.0,
|
| 2338 |
+
"step": 8225
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 2.7527527527527527,
|
| 2342 |
+
"grad_norm": 0.00017187983030453324,
|
| 2343 |
+
"learning_rate": 1.7332468254703864e-05,
|
| 2344 |
+
"loss": 0.0007,
|
| 2345 |
+
"step": 8250
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 2.7610944277610945,
|
| 2349 |
+
"grad_norm": 0.00022748754417989403,
|
| 2350 |
+
"learning_rate": 1.721660950968579e-05,
|
| 2351 |
+
"loss": 0.0097,
|
| 2352 |
+
"step": 8275
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 2.7694361027694363,
|
| 2356 |
+
"grad_norm": 0.0016567155253142118,
|
| 2357 |
+
"learning_rate": 1.7100750764667718e-05,
|
| 2358 |
+
"loss": 0.0018,
|
| 2359 |
+
"step": 8300
|
| 2360 |
+
},
|
| 2361 |
+
{
|
| 2362 |
+
"epoch": 2.7777777777777777,
|
| 2363 |
+
"grad_norm": 0.0004330672090873122,
|
| 2364 |
+
"learning_rate": 1.6984892019649645e-05,
|
| 2365 |
+
"loss": 0.0001,
|
| 2366 |
+
"step": 8325
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 2.7861194527861195,
|
| 2370 |
+
"grad_norm": 0.006991783156991005,
|
| 2371 |
+
"learning_rate": 1.6869033274631572e-05,
|
| 2372 |
+
"loss": 0.0174,
|
| 2373 |
+
"step": 8350
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 2.7944611277944613,
|
| 2377 |
+
"grad_norm": 0.0004689243796747178,
|
| 2378 |
+
"learning_rate": 1.6753174529613496e-05,
|
| 2379 |
+
"loss": 0.0015,
|
| 2380 |
+
"step": 8375
|
| 2381 |
+
},
|
| 2382 |
+
{
|
| 2383 |
+
"epoch": 2.8028028028028027,
|
| 2384 |
+
"grad_norm": 0.0006888713105581701,
|
| 2385 |
+
"learning_rate": 1.6637315784595423e-05,
|
| 2386 |
+
"loss": 0.0001,
|
| 2387 |
+
"step": 8400
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 2.8111444778111445,
|
| 2391 |
+
"grad_norm": 0.0015181319322437048,
|
| 2392 |
+
"learning_rate": 1.6521457039577347e-05,
|
| 2393 |
+
"loss": 0.0013,
|
| 2394 |
+
"step": 8425
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 2.8194861528194863,
|
| 2398 |
+
"grad_norm": 0.00026851208531297743,
|
| 2399 |
+
"learning_rate": 1.6405598294559274e-05,
|
| 2400 |
+
"loss": 0.0067,
|
| 2401 |
+
"step": 8450
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 2.8278278278278277,
|
| 2405 |
+
"grad_norm": 0.0003072840045206249,
|
| 2406 |
+
"learning_rate": 1.62897395495412e-05,
|
| 2407 |
+
"loss": 0.0,
|
| 2408 |
+
"step": 8475
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 2.8361695028361695,
|
| 2412 |
+
"grad_norm": 0.000419327465351671,
|
| 2413 |
+
"learning_rate": 1.6173880804523125e-05,
|
| 2414 |
+
"loss": 0.0,
|
| 2415 |
+
"step": 8500
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 2.8445111778445114,
|
| 2419 |
+
"grad_norm": 0.0008704416686668992,
|
| 2420 |
+
"learning_rate": 1.6058022059505052e-05,
|
| 2421 |
+
"loss": 0.0,
|
| 2422 |
+
"step": 8525
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 2.8528528528528527,
|
| 2426 |
+
"grad_norm": 0.00026631655055098236,
|
| 2427 |
+
"learning_rate": 1.5942163314486975e-05,
|
| 2428 |
+
"loss": 0.0,
|
| 2429 |
+
"step": 8550
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 2.8611945278611945,
|
| 2433 |
+
"grad_norm": 0.00015377814997918904,
|
| 2434 |
+
"learning_rate": 1.5826304569468906e-05,
|
| 2435 |
+
"loss": 0.0,
|
| 2436 |
+
"step": 8575
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 2.8695362028695364,
|
| 2440 |
+
"grad_norm": 0.0001188891037600115,
|
| 2441 |
+
"learning_rate": 1.571044582445083e-05,
|
| 2442 |
+
"loss": 0.0,
|
| 2443 |
+
"step": 8600
|
| 2444 |
+
},
|
| 2445 |
+
{
|
| 2446 |
+
"epoch": 2.8778778778778777,
|
| 2447 |
+
"grad_norm": 0.00026092492043972015,
|
| 2448 |
+
"learning_rate": 1.5594587079432757e-05,
|
| 2449 |
+
"loss": 0.0,
|
| 2450 |
+
"step": 8625
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 2.8862195528862196,
|
| 2454 |
+
"grad_norm": 0.0021843460854142904,
|
| 2455 |
+
"learning_rate": 1.5478728334414684e-05,
|
| 2456 |
+
"loss": 0.0025,
|
| 2457 |
+
"step": 8650
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 2.8945612278945614,
|
| 2461 |
+
"grad_norm": 0.00022988433192949742,
|
| 2462 |
+
"learning_rate": 1.5362869589396608e-05,
|
| 2463 |
+
"loss": 0.0094,
|
| 2464 |
+
"step": 8675
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"epoch": 2.9029029029029028,
|
| 2468 |
+
"grad_norm": 0.00016332197992596775,
|
| 2469 |
+
"learning_rate": 1.5247010844378535e-05,
|
| 2470 |
+
"loss": 0.006,
|
| 2471 |
+
"step": 8700
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 2.9112445779112446,
|
| 2475 |
+
"grad_norm": 0.00020633054373320192,
|
| 2476 |
+
"learning_rate": 1.513115209936046e-05,
|
| 2477 |
+
"loss": 0.0002,
|
| 2478 |
+
"step": 8725
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 2.9195862529195864,
|
| 2482 |
+
"grad_norm": 0.00025880796601995826,
|
| 2483 |
+
"learning_rate": 1.5015293354342385e-05,
|
| 2484 |
+
"loss": 0.0,
|
| 2485 |
+
"step": 8750
|
| 2486 |
+
},
|
| 2487 |
+
{
|
| 2488 |
+
"epoch": 2.9279279279279278,
|
| 2489 |
+
"grad_norm": 0.0003759913379326463,
|
| 2490 |
+
"learning_rate": 1.489943460932431e-05,
|
| 2491 |
+
"loss": 0.0003,
|
| 2492 |
+
"step": 8775
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 2.9362696029362696,
|
| 2496 |
+
"grad_norm": 0.00019277114188298583,
|
| 2497 |
+
"learning_rate": 1.4783575864306238e-05,
|
| 2498 |
+
"loss": 0.0044,
|
| 2499 |
+
"step": 8800
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 2.9446112779446114,
|
| 2503 |
+
"grad_norm": 0.0007216227240860462,
|
| 2504 |
+
"learning_rate": 1.4667717119288165e-05,
|
| 2505 |
+
"loss": 0.0,
|
| 2506 |
+
"step": 8825
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"epoch": 2.952952952952953,
|
| 2510 |
+
"grad_norm": 0.0001798528101062402,
|
| 2511 |
+
"learning_rate": 1.4551858374270092e-05,
|
| 2512 |
+
"loss": 0.0304,
|
| 2513 |
+
"step": 8850
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 2.9612946279612946,
|
| 2517 |
+
"grad_norm": 0.00019520522619131953,
|
| 2518 |
+
"learning_rate": 1.4435999629252017e-05,
|
| 2519 |
+
"loss": 0.0001,
|
| 2520 |
+
"step": 8875
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"epoch": 2.9696363029696364,
|
| 2524 |
+
"grad_norm": 0.0001343146723229438,
|
| 2525 |
+
"learning_rate": 1.4320140884233943e-05,
|
| 2526 |
+
"loss": 0.0489,
|
| 2527 |
+
"step": 8900
|
| 2528 |
+
},
|
| 2529 |
+
{
|
| 2530 |
+
"epoch": 2.977977977977978,
|
| 2531 |
+
"grad_norm": 12.330880165100098,
|
| 2532 |
+
"learning_rate": 1.4204282139215868e-05,
|
| 2533 |
+
"loss": 0.0347,
|
| 2534 |
+
"step": 8925
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 2.9863196529863196,
|
| 2538 |
+
"grad_norm": 0.00025471311528235674,
|
| 2539 |
+
"learning_rate": 1.4088423394197794e-05,
|
| 2540 |
+
"loss": 0.0245,
|
| 2541 |
+
"step": 8950
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 2.9946613279946614,
|
| 2545 |
+
"grad_norm": 0.0009992901468649507,
|
| 2546 |
+
"learning_rate": 1.397256464917972e-05,
|
| 2547 |
+
"loss": 0.0337,
|
| 2548 |
+
"step": 8975
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 3.0,
|
| 2552 |
+
"eval_accuracy": 0.9983316649983317,
|
| 2553 |
+
"eval_auc": 0.9996484660336011,
|
| 2554 |
+
"eval_f1": 0.935064935064935,
|
| 2555 |
+
"eval_loss": 0.007099705282598734,
|
| 2556 |
+
"eval_precision": 0.9,
|
| 2557 |
+
"eval_recall": 0.972972972972973,
|
| 2558 |
+
"eval_runtime": 77.3547,
|
| 2559 |
+
"eval_samples_per_second": 77.487,
|
| 2560 |
+
"eval_steps_per_second": 4.848,
|
| 2561 |
+
"step": 8991
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 3.003003003003003,
|
| 2565 |
+
"grad_norm": 0.0015147333033382893,
|
| 2566 |
+
"learning_rate": 1.3856705904161646e-05,
|
| 2567 |
+
"loss": 0.0003,
|
| 2568 |
+
"step": 9000
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 3.0113446780113446,
|
| 2572 |
+
"grad_norm": 0.002197373891249299,
|
| 2573 |
+
"learning_rate": 1.3740847159143572e-05,
|
| 2574 |
+
"loss": 0.0002,
|
| 2575 |
+
"step": 9025
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 3.0196863530196865,
|
| 2579 |
+
"grad_norm": 0.0003656800545286387,
|
| 2580 |
+
"learning_rate": 1.36249884141255e-05,
|
| 2581 |
+
"loss": 0.001,
|
| 2582 |
+
"step": 9050
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 3.028028028028028,
|
| 2586 |
+
"grad_norm": 0.00043537470628507435,
|
| 2587 |
+
"learning_rate": 1.3509129669107426e-05,
|
| 2588 |
+
"loss": 0.0,
|
| 2589 |
+
"step": 9075
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 3.0363697030363697,
|
| 2593 |
+
"grad_norm": 0.0017388892592862248,
|
| 2594 |
+
"learning_rate": 1.3393270924089351e-05,
|
| 2595 |
+
"loss": 0.0,
|
| 2596 |
+
"step": 9100
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 3.0447113780447115,
|
| 2600 |
+
"grad_norm": 0.00047031775466166437,
|
| 2601 |
+
"learning_rate": 1.3277412179071278e-05,
|
| 2602 |
+
"loss": 0.0,
|
| 2603 |
+
"step": 9125
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 3.053053053053053,
|
| 2607 |
+
"grad_norm": 0.0003807338362094015,
|
| 2608 |
+
"learning_rate": 1.3161553434053204e-05,
|
| 2609 |
+
"loss": 0.0,
|
| 2610 |
+
"step": 9150
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 3.0613947280613947,
|
| 2614 |
+
"grad_norm": 0.00046666429261676967,
|
| 2615 |
+
"learning_rate": 1.3045694689035129e-05,
|
| 2616 |
+
"loss": 0.0,
|
| 2617 |
+
"step": 9175
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 3.0697364030697365,
|
| 2621 |
+
"grad_norm": 0.000990239786915481,
|
| 2622 |
+
"learning_rate": 1.2929835944017054e-05,
|
| 2623 |
+
"loss": 0.0001,
|
| 2624 |
+
"step": 9200
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 3.078078078078078,
|
| 2628 |
+
"grad_norm": 0.0009268262074328959,
|
| 2629 |
+
"learning_rate": 1.281397719899898e-05,
|
| 2630 |
+
"loss": 0.0,
|
| 2631 |
+
"step": 9225
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 3.0864197530864197,
|
| 2635 |
+
"grad_norm": 0.0001752979151206091,
|
| 2636 |
+
"learning_rate": 1.2698118453980907e-05,
|
| 2637 |
+
"loss": 0.0,
|
| 2638 |
+
"step": 9250
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 3.0947614280947615,
|
| 2642 |
+
"grad_norm": 0.00017092602502088994,
|
| 2643 |
+
"learning_rate": 1.2582259708962832e-05,
|
| 2644 |
+
"loss": 0.0001,
|
| 2645 |
+
"step": 9275
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 3.103103103103103,
|
| 2649 |
+
"grad_norm": 0.0003801763814408332,
|
| 2650 |
+
"learning_rate": 1.246640096394476e-05,
|
| 2651 |
+
"loss": 0.0,
|
| 2652 |
+
"step": 9300
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 3.1114447781114447,
|
| 2656 |
+
"grad_norm": 0.0007715587271377444,
|
| 2657 |
+
"learning_rate": 1.2350542218926685e-05,
|
| 2658 |
+
"loss": 0.0,
|
| 2659 |
+
"step": 9325
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 3.1197864531197865,
|
| 2663 |
+
"grad_norm": 0.0001741141895763576,
|
| 2664 |
+
"learning_rate": 1.223468347390861e-05,
|
| 2665 |
+
"loss": 0.0,
|
| 2666 |
+
"step": 9350
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 3.128128128128128,
|
| 2670 |
+
"grad_norm": 0.00024119947920553386,
|
| 2671 |
+
"learning_rate": 1.2118824728890537e-05,
|
| 2672 |
+
"loss": 0.0,
|
| 2673 |
+
"step": 9375
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 3.1364698031364697,
|
| 2677 |
+
"grad_norm": 0.0003123109054286033,
|
| 2678 |
+
"learning_rate": 1.2002965983872463e-05,
|
| 2679 |
+
"loss": 0.0027,
|
| 2680 |
+
"step": 9400
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 3.1448114781448115,
|
| 2684 |
+
"grad_norm": 0.0002255546278320253,
|
| 2685 |
+
"learning_rate": 1.188710723885439e-05,
|
| 2686 |
+
"loss": 0.0001,
|
| 2687 |
+
"step": 9425
|
| 2688 |
+
},
|
| 2689 |
+
{
|
| 2690 |
+
"epoch": 3.153153153153153,
|
| 2691 |
+
"grad_norm": 0.00022687349701300263,
|
| 2692 |
+
"learning_rate": 1.1771248493836315e-05,
|
| 2693 |
+
"loss": 0.0,
|
| 2694 |
+
"step": 9450
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 3.1614948281614947,
|
| 2698 |
+
"grad_norm": 0.0009094116394408047,
|
| 2699 |
+
"learning_rate": 1.1655389748818242e-05,
|
| 2700 |
+
"loss": 0.0001,
|
| 2701 |
+
"step": 9475
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 3.1698365031698366,
|
| 2705 |
+
"grad_norm": 0.00019586944836191833,
|
| 2706 |
+
"learning_rate": 1.1539531003800168e-05,
|
| 2707 |
+
"loss": 0.0001,
|
| 2708 |
+
"step": 9500
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 3.1781781781781784,
|
| 2712 |
+
"grad_norm": 0.0013932987349107862,
|
| 2713 |
+
"learning_rate": 1.1423672258782093e-05,
|
| 2714 |
+
"loss": 0.0,
|
| 2715 |
+
"step": 9525
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 3.1865198531865198,
|
| 2719 |
+
"grad_norm": 0.00014476373326033354,
|
| 2720 |
+
"learning_rate": 1.130781351376402e-05,
|
| 2721 |
+
"loss": 0.0,
|
| 2722 |
+
"step": 9550
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 3.1948615281948616,
|
| 2726 |
+
"grad_norm": 0.00020363023213576525,
|
| 2727 |
+
"learning_rate": 1.1191954768745945e-05,
|
| 2728 |
+
"loss": 0.0,
|
| 2729 |
+
"step": 9575
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 3.2032032032032034,
|
| 2733 |
+
"grad_norm": 0.0005529238260351121,
|
| 2734 |
+
"learning_rate": 1.1076096023727873e-05,
|
| 2735 |
+
"loss": 0.0005,
|
| 2736 |
+
"step": 9600
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 3.2115448782115448,
|
| 2740 |
+
"grad_norm": 0.00039333957829512656,
|
| 2741 |
+
"learning_rate": 1.0960237278709798e-05,
|
| 2742 |
+
"loss": 0.0001,
|
| 2743 |
+
"step": 9625
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 3.2198865532198866,
|
| 2747 |
+
"grad_norm": 0.0006521272589452565,
|
| 2748 |
+
"learning_rate": 1.0844378533691723e-05,
|
| 2749 |
+
"loss": 0.0,
|
| 2750 |
+
"step": 9650
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 3.2282282282282284,
|
| 2754 |
+
"grad_norm": Infinity,
|
| 2755 |
+
"learning_rate": 1.0733154138474373e-05,
|
| 2756 |
+
"loss": 0.0231,
|
| 2757 |
+
"step": 9675
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 3.23656990323657,
|
| 2761 |
+
"grad_norm": 0.0003595724992919713,
|
| 2762 |
+
"learning_rate": 1.0617295393456298e-05,
|
| 2763 |
+
"loss": 0.0,
|
| 2764 |
+
"step": 9700
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 3.2449115782449116,
|
| 2768 |
+
"grad_norm": 0.00021334755001589656,
|
| 2769 |
+
"learning_rate": 1.0501436648438225e-05,
|
| 2770 |
+
"loss": 0.0003,
|
| 2771 |
+
"step": 9725
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 3.2532532532532534,
|
| 2775 |
+
"grad_norm": 0.00028163444949314,
|
| 2776 |
+
"learning_rate": 1.038557790342015e-05,
|
| 2777 |
+
"loss": 0.0,
|
| 2778 |
+
"step": 9750
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 3.261594928261595,
|
| 2782 |
+
"grad_norm": 0.00030702631920576096,
|
| 2783 |
+
"learning_rate": 1.0269719158402076e-05,
|
| 2784 |
+
"loss": 0.0,
|
| 2785 |
+
"step": 9775
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 3.2699366032699366,
|
| 2789 |
+
"grad_norm": 0.0002560736029408872,
|
| 2790 |
+
"learning_rate": 1.0153860413384003e-05,
|
| 2791 |
+
"loss": 0.0,
|
| 2792 |
+
"step": 9800
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 3.2782782782782784,
|
| 2796 |
+
"grad_norm": 7.156394713092595e-05,
|
| 2797 |
+
"learning_rate": 1.0038001668365928e-05,
|
| 2798 |
+
"loss": 0.0004,
|
| 2799 |
+
"step": 9825
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 3.28661995328662,
|
| 2803 |
+
"grad_norm": 0.00014015445776749402,
|
| 2804 |
+
"learning_rate": 9.922142923347855e-06,
|
| 2805 |
+
"loss": 0.0211,
|
| 2806 |
+
"step": 9850
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 3.2949616282949616,
|
| 2810 |
+
"grad_norm": 8.722938946448267e-05,
|
| 2811 |
+
"learning_rate": 9.806284178329781e-06,
|
| 2812 |
+
"loss": 0.0355,
|
| 2813 |
+
"step": 9875
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 3.3033033033033035,
|
| 2817 |
+
"grad_norm": 0.00010665127047104761,
|
| 2818 |
+
"learning_rate": 9.690425433311706e-06,
|
| 2819 |
+
"loss": 0.0,
|
| 2820 |
+
"step": 9900
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 3.311644978311645,
|
| 2824 |
+
"grad_norm": 0.0003643149684648961,
|
| 2825 |
+
"learning_rate": 9.574566688293633e-06,
|
| 2826 |
+
"loss": 0.0159,
|
| 2827 |
+
"step": 9925
|
| 2828 |
+
},
|
| 2829 |
+
{
|
| 2830 |
+
"epoch": 3.3199866533199867,
|
| 2831 |
+
"grad_norm": 0.000472669635200873,
|
| 2832 |
+
"learning_rate": 9.458707943275559e-06,
|
| 2833 |
+
"loss": 0.0001,
|
| 2834 |
+
"step": 9950
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 3.3283283283283285,
|
| 2838 |
+
"grad_norm": 0.0002747498219832778,
|
| 2839 |
+
"learning_rate": 9.342849198257484e-06,
|
| 2840 |
+
"loss": 0.0002,
|
| 2841 |
+
"step": 9975
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 3.33667000333667,
|
| 2845 |
+
"grad_norm": 0.000370238529285416,
|
| 2846 |
+
"learning_rate": 9.226990453239411e-06,
|
| 2847 |
+
"loss": 0.0,
|
| 2848 |
+
"step": 10000
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"epoch": 3.3450116783450117,
|
| 2852 |
+
"grad_norm": 0.0001508592686150223,
|
| 2853 |
+
"learning_rate": 9.111131708221338e-06,
|
| 2854 |
+
"loss": 0.0153,
|
| 2855 |
+
"step": 10025
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 3.3533533533533535,
|
| 2859 |
+
"grad_norm": 0.00012634581071324646,
|
| 2860 |
+
"learning_rate": 8.995272963203264e-06,
|
| 2861 |
+
"loss": 0.0,
|
| 2862 |
+
"step": 10050
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 3.361695028361695,
|
| 2866 |
+
"grad_norm": 7.435754378093407e-05,
|
| 2867 |
+
"learning_rate": 8.879414218185189e-06,
|
| 2868 |
+
"loss": 0.0,
|
| 2869 |
+
"step": 10075
|
| 2870 |
+
},
|
| 2871 |
+
{
|
| 2872 |
+
"epoch": 3.3700367033700367,
|
| 2873 |
+
"grad_norm": 0.00024650897830724716,
|
| 2874 |
+
"learning_rate": 8.763555473167115e-06,
|
| 2875 |
+
"loss": 0.0,
|
| 2876 |
+
"step": 10100
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 3.3783783783783785,
|
| 2880 |
+
"grad_norm": 0.00031506569939665496,
|
| 2881 |
+
"learning_rate": 8.64769672814904e-06,
|
| 2882 |
+
"loss": 0.0,
|
| 2883 |
+
"step": 10125
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 3.38672005338672,
|
| 2887 |
+
"grad_norm": 0.00038327754009515047,
|
| 2888 |
+
"learning_rate": 8.531837983130967e-06,
|
| 2889 |
+
"loss": 0.0,
|
| 2890 |
+
"step": 10150
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 3.3950617283950617,
|
| 2894 |
+
"grad_norm": 0.00010452201968291774,
|
| 2895 |
+
"learning_rate": 8.415979238112894e-06,
|
| 2896 |
+
"loss": 0.0,
|
| 2897 |
+
"step": 10175
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 3.4034034034034035,
|
| 2901 |
+
"grad_norm": 0.003248438471928239,
|
| 2902 |
+
"learning_rate": 8.30012049309482e-06,
|
| 2903 |
+
"loss": 0.0,
|
| 2904 |
+
"step": 10200
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 3.411745078411745,
|
| 2908 |
+
"grad_norm": 0.0001290349755436182,
|
| 2909 |
+
"learning_rate": 8.184261748076745e-06,
|
| 2910 |
+
"loss": 0.0,
|
| 2911 |
+
"step": 10225
|
| 2912 |
+
},
|
| 2913 |
+
{
|
| 2914 |
+
"epoch": 3.4200867534200867,
|
| 2915 |
+
"grad_norm": 0.0004630775947589427,
|
| 2916 |
+
"learning_rate": 8.06840300305867e-06,
|
| 2917 |
+
"loss": 0.0,
|
| 2918 |
+
"step": 10250
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 3.4284284284284285,
|
| 2922 |
+
"grad_norm": 0.00028067638049833477,
|
| 2923 |
+
"learning_rate": 7.952544258040597e-06,
|
| 2924 |
+
"loss": 0.0,
|
| 2925 |
+
"step": 10275
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 3.43677010343677,
|
| 2929 |
+
"grad_norm": 0.00016888586105778813,
|
| 2930 |
+
"learning_rate": 7.836685513022524e-06,
|
| 2931 |
+
"loss": 0.0,
|
| 2932 |
+
"step": 10300
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"epoch": 3.4451117784451117,
|
| 2936 |
+
"grad_norm": 0.00010634879436111078,
|
| 2937 |
+
"learning_rate": 7.72082676800445e-06,
|
| 2938 |
+
"loss": 0.0,
|
| 2939 |
+
"step": 10325
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 3.4534534534534536,
|
| 2943 |
+
"grad_norm": 0.004633384756743908,
|
| 2944 |
+
"learning_rate": 7.604968022986375e-06,
|
| 2945 |
+
"loss": 0.0,
|
| 2946 |
+
"step": 10350
|
| 2947 |
+
},
|
| 2948 |
+
{
|
| 2949 |
+
"epoch": 3.461795128461795,
|
| 2950 |
+
"grad_norm": 0.00013694832159671932,
|
| 2951 |
+
"learning_rate": 7.489109277968301e-06,
|
| 2952 |
+
"loss": 0.0,
|
| 2953 |
+
"step": 10375
|
| 2954 |
+
},
|
| 2955 |
+
{
|
| 2956 |
+
"epoch": 3.4701368034701368,
|
| 2957 |
+
"grad_norm": 0.0003641396469902247,
|
| 2958 |
+
"learning_rate": 7.373250532950228e-06,
|
| 2959 |
+
"loss": 0.0337,
|
| 2960 |
+
"step": 10400
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 3.4784784784784786,
|
| 2964 |
+
"grad_norm": 0.000520822883117944,
|
| 2965 |
+
"learning_rate": 7.257391787932154e-06,
|
| 2966 |
+
"loss": 0.0,
|
| 2967 |
+
"step": 10425
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 3.4868201534868204,
|
| 2971 |
+
"grad_norm": 0.000284605281194672,
|
| 2972 |
+
"learning_rate": 7.141533042914079e-06,
|
| 2973 |
+
"loss": 0.0,
|
| 2974 |
+
"step": 10450
|
| 2975 |
+
},
|
| 2976 |
+
{
|
| 2977 |
+
"epoch": 3.4951618284951618,
|
| 2978 |
+
"grad_norm": 0.0006933792028576136,
|
| 2979 |
+
"learning_rate": 7.025674297896006e-06,
|
| 2980 |
+
"loss": 0.0002,
|
| 2981 |
+
"step": 10475
|
| 2982 |
+
},
|
| 2983 |
+
{
|
| 2984 |
+
"epoch": 3.5035035035035036,
|
| 2985 |
+
"grad_norm": 0.0002910966577474028,
|
| 2986 |
+
"learning_rate": 6.914449902678655e-06,
|
| 2987 |
+
"loss": 0.0221,
|
| 2988 |
+
"step": 10500
|
| 2989 |
+
},
|
| 2990 |
+
{
|
| 2991 |
+
"epoch": 3.5118451785118454,
|
| 2992 |
+
"grad_norm": 0.0009816793026402593,
|
| 2993 |
+
"learning_rate": 6.79859115766058e-06,
|
| 2994 |
+
"loss": 0.0,
|
| 2995 |
+
"step": 10525
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 3.520186853520187,
|
| 2999 |
+
"grad_norm": 0.00022771142539568245,
|
| 3000 |
+
"learning_rate": 6.6827324126425065e-06,
|
| 3001 |
+
"loss": 0.0,
|
| 3002 |
+
"step": 10550
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 3.5285285285285286,
|
| 3006 |
+
"grad_norm": 0.00018489710055291653,
|
| 3007 |
+
"learning_rate": 6.566873667624434e-06,
|
| 3008 |
+
"loss": 0.0,
|
| 3009 |
+
"step": 10575
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 3.5368702035368704,
|
| 3013 |
+
"grad_norm": 0.0005421001696959138,
|
| 3014 |
+
"learning_rate": 6.451014922606359e-06,
|
| 3015 |
+
"loss": 0.0001,
|
| 3016 |
+
"step": 10600
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 3.545211878545212,
|
| 3020 |
+
"grad_norm": 0.00024389017198700458,
|
| 3021 |
+
"learning_rate": 6.335156177588284e-06,
|
| 3022 |
+
"loss": 0.0,
|
| 3023 |
+
"step": 10625
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 3.5535535535535536,
|
| 3027 |
+
"grad_norm": 0.0002914558572228998,
|
| 3028 |
+
"learning_rate": 6.219297432570211e-06,
|
| 3029 |
+
"loss": 0.0,
|
| 3030 |
+
"step": 10650
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 3.5618952285618954,
|
| 3034 |
+
"grad_norm": 0.0002373899769736454,
|
| 3035 |
+
"learning_rate": 6.103438687552137e-06,
|
| 3036 |
+
"loss": 0.0,
|
| 3037 |
+
"step": 10675
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 3.570236903570237,
|
| 3041 |
+
"grad_norm": 0.006195495370775461,
|
| 3042 |
+
"learning_rate": 5.987579942534063e-06,
|
| 3043 |
+
"loss": 0.0,
|
| 3044 |
+
"step": 10700
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 3.5785785785785786,
|
| 3048 |
+
"grad_norm": 0.00012051707017235458,
|
| 3049 |
+
"learning_rate": 5.8717211975159885e-06,
|
| 3050 |
+
"loss": 0.0,
|
| 3051 |
+
"step": 10725
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 3.5869202535869205,
|
| 3055 |
+
"grad_norm": 0.002486045239493251,
|
| 3056 |
+
"learning_rate": 5.755862452497915e-06,
|
| 3057 |
+
"loss": 0.0001,
|
| 3058 |
+
"step": 10750
|
| 3059 |
+
},
|
| 3060 |
+
{
|
| 3061 |
+
"epoch": 3.595261928595262,
|
| 3062 |
+
"grad_norm": 0.0001380012690788135,
|
| 3063 |
+
"learning_rate": 5.640003707479841e-06,
|
| 3064 |
+
"loss": 0.0,
|
| 3065 |
+
"step": 10775
|
| 3066 |
+
},
|
| 3067 |
+
{
|
| 3068 |
+
"epoch": 3.6036036036036037,
|
| 3069 |
+
"grad_norm": 0.00035765368374995887,
|
| 3070 |
+
"learning_rate": 5.524144962461767e-06,
|
| 3071 |
+
"loss": 0.0,
|
| 3072 |
+
"step": 10800
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 3.6119452786119455,
|
| 3076 |
+
"grad_norm": 0.00022478456958197057,
|
| 3077 |
+
"learning_rate": 5.408286217443693e-06,
|
| 3078 |
+
"loss": 0.0,
|
| 3079 |
+
"step": 10825
|
| 3080 |
+
},
|
| 3081 |
+
{
|
| 3082 |
+
"epoch": 3.620286953620287,
|
| 3083 |
+
"grad_norm": 0.00024634675355628133,
|
| 3084 |
+
"learning_rate": 5.292427472425619e-06,
|
| 3085 |
+
"loss": 0.0,
|
| 3086 |
+
"step": 10850
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 3.6286286286286287,
|
| 3090 |
+
"grad_norm": 9.515749115962535e-05,
|
| 3091 |
+
"learning_rate": 5.176568727407545e-06,
|
| 3092 |
+
"loss": 0.0,
|
| 3093 |
+
"step": 10875
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 3.6369703036369705,
|
| 3097 |
+
"grad_norm": 0.0015800537075847387,
|
| 3098 |
+
"learning_rate": 5.0607099823894705e-06,
|
| 3099 |
+
"loss": 0.0,
|
| 3100 |
+
"step": 10900
|
| 3101 |
+
},
|
| 3102 |
+
{
|
| 3103 |
+
"epoch": 3.645311978645312,
|
| 3104 |
+
"grad_norm": 0.00018024566816166043,
|
| 3105 |
+
"learning_rate": 4.944851237371398e-06,
|
| 3106 |
+
"loss": 0.0,
|
| 3107 |
+
"step": 10925
|
| 3108 |
+
},
|
| 3109 |
+
{
|
| 3110 |
+
"epoch": 3.6536536536536537,
|
| 3111 |
+
"grad_norm": 0.00037315645022317767,
|
| 3112 |
+
"learning_rate": 4.828992492353323e-06,
|
| 3113 |
+
"loss": 0.0,
|
| 3114 |
+
"step": 10950
|
| 3115 |
+
},
|
| 3116 |
+
{
|
| 3117 |
+
"epoch": 3.6619953286619955,
|
| 3118 |
+
"grad_norm": 0.00012279977090656757,
|
| 3119 |
+
"learning_rate": 4.713133747335249e-06,
|
| 3120 |
+
"loss": 0.0098,
|
| 3121 |
+
"step": 10975
|
| 3122 |
+
},
|
| 3123 |
+
{
|
| 3124 |
+
"epoch": 3.670337003670337,
|
| 3125 |
+
"grad_norm": 0.00033114643883891404,
|
| 3126 |
+
"learning_rate": 4.5972750023171755e-06,
|
| 3127 |
+
"loss": 0.0,
|
| 3128 |
+
"step": 11000
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 3.6786786786786787,
|
| 3132 |
+
"grad_norm": 0.00015320570673793554,
|
| 3133 |
+
"learning_rate": 4.481416257299101e-06,
|
| 3134 |
+
"loss": 0.0,
|
| 3135 |
+
"step": 11025
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 3.6870203536870205,
|
| 3139 |
+
"grad_norm": 0.00033146454370580614,
|
| 3140 |
+
"learning_rate": 4.365557512281027e-06,
|
| 3141 |
+
"loss": 0.0131,
|
| 3142 |
+
"step": 11050
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 3.695362028695362,
|
| 3146 |
+
"grad_norm": 0.0005683086346834898,
|
| 3147 |
+
"learning_rate": 4.249698767262953e-06,
|
| 3148 |
+
"loss": 0.0,
|
| 3149 |
+
"step": 11075
|
| 3150 |
+
},
|
| 3151 |
+
{
|
| 3152 |
+
"epoch": 3.7037037037037037,
|
| 3153 |
+
"grad_norm": 0.0013795130653306842,
|
| 3154 |
+
"learning_rate": 4.13384002224488e-06,
|
| 3155 |
+
"loss": 0.0001,
|
| 3156 |
+
"step": 11100
|
| 3157 |
+
},
|
| 3158 |
+
{
|
| 3159 |
+
"epoch": 3.7120453787120455,
|
| 3160 |
+
"grad_norm": 0.0038050978910177946,
|
| 3161 |
+
"learning_rate": 4.017981277226805e-06,
|
| 3162 |
+
"loss": 0.0004,
|
| 3163 |
+
"step": 11125
|
| 3164 |
+
},
|
| 3165 |
+
{
|
| 3166 |
+
"epoch": 3.720387053720387,
|
| 3167 |
+
"grad_norm": 0.00022509918198920786,
|
| 3168 |
+
"learning_rate": 3.902122532208731e-06,
|
| 3169 |
+
"loss": 0.0001,
|
| 3170 |
+
"step": 11150
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 3.7287287287287287,
|
| 3174 |
+
"grad_norm": 0.0005176962004043162,
|
| 3175 |
+
"learning_rate": 3.7862637871906575e-06,
|
| 3176 |
+
"loss": 0.0,
|
| 3177 |
+
"step": 11175
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 3.7370704037370706,
|
| 3181 |
+
"grad_norm": 0.0005020920070819557,
|
| 3182 |
+
"learning_rate": 3.6704050421725833e-06,
|
| 3183 |
+
"loss": 0.0598,
|
| 3184 |
+
"step": 11200
|
| 3185 |
+
},
|
| 3186 |
+
{
|
| 3187 |
+
"epoch": 3.745412078745412,
|
| 3188 |
+
"grad_norm": 0.0011666618520393968,
|
| 3189 |
+
"learning_rate": 3.5545462971545095e-06,
|
| 3190 |
+
"loss": 0.0001,
|
| 3191 |
+
"step": 11225
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 3.7537537537537538,
|
| 3195 |
+
"grad_norm": 0.009826689027249813,
|
| 3196 |
+
"learning_rate": 3.4386875521364354e-06,
|
| 3197 |
+
"loss": 0.0005,
|
| 3198 |
+
"step": 11250
|
| 3199 |
+
},
|
| 3200 |
+
{
|
| 3201 |
+
"epoch": 3.7620954287620956,
|
| 3202 |
+
"grad_norm": 0.011009103618562222,
|
| 3203 |
+
"learning_rate": 3.3228288071183616e-06,
|
| 3204 |
+
"loss": 0.0371,
|
| 3205 |
+
"step": 11275
|
| 3206 |
+
},
|
| 3207 |
+
{
|
| 3208 |
+
"epoch": 3.770437103770437,
|
| 3209 |
+
"grad_norm": 0.01023051142692566,
|
| 3210 |
+
"learning_rate": 3.2069700621002874e-06,
|
| 3211 |
+
"loss": 0.0001,
|
| 3212 |
+
"step": 11300
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 3.7787787787787788,
|
| 3216 |
+
"grad_norm": 0.0009649486164562404,
|
| 3217 |
+
"learning_rate": 3.0911113170822137e-06,
|
| 3218 |
+
"loss": 0.0001,
|
| 3219 |
+
"step": 11325
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 3.7871204537871206,
|
| 3223 |
+
"grad_norm": 0.003300480078905821,
|
| 3224 |
+
"learning_rate": 2.97525257206414e-06,
|
| 3225 |
+
"loss": 0.0073,
|
| 3226 |
+
"step": 11350
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 3.795462128795462,
|
| 3230 |
+
"grad_norm": 0.004097791388630867,
|
| 3231 |
+
"learning_rate": 2.8593938270460653e-06,
|
| 3232 |
+
"loss": 0.0016,
|
| 3233 |
+
"step": 11375
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 3.803803803803804,
|
| 3237 |
+
"grad_norm": 0.0011136537650600076,
|
| 3238 |
+
"learning_rate": 2.7435350820279915e-06,
|
| 3239 |
+
"loss": 0.0001,
|
| 3240 |
+
"step": 11400
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 3.8121454788121456,
|
| 3244 |
+
"grad_norm": 0.001229249406605959,
|
| 3245 |
+
"learning_rate": 2.6276763370099178e-06,
|
| 3246 |
+
"loss": 0.0002,
|
| 3247 |
+
"step": 11425
|
| 3248 |
+
},
|
| 3249 |
+
{
|
| 3250 |
+
"epoch": 3.820487153820487,
|
| 3251 |
+
"grad_norm": 0.001040545990690589,
|
| 3252 |
+
"learning_rate": 2.5118175919918436e-06,
|
| 3253 |
+
"loss": 0.0104,
|
| 3254 |
+
"step": 11450
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 3.828828828828829,
|
| 3258 |
+
"grad_norm": 0.0006933126132935286,
|
| 3259 |
+
"learning_rate": 2.39595884697377e-06,
|
| 3260 |
+
"loss": 0.0006,
|
| 3261 |
+
"step": 11475
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 3.8371705038371706,
|
| 3265 |
+
"grad_norm": 0.0001256452960660681,
|
| 3266 |
+
"learning_rate": 2.2801001019556957e-06,
|
| 3267 |
+
"loss": 0.0109,
|
| 3268 |
+
"step": 11500
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 3.845512178845512,
|
| 3272 |
+
"grad_norm": 0.00040541382622905076,
|
| 3273 |
+
"learning_rate": 2.1642413569376215e-06,
|
| 3274 |
+
"loss": 0.0,
|
| 3275 |
+
"step": 11525
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 3.853853853853854,
|
| 3279 |
+
"grad_norm": 0.007394419051706791,
|
| 3280 |
+
"learning_rate": 2.0483826119195477e-06,
|
| 3281 |
+
"loss": 0.0,
|
| 3282 |
+
"step": 11550
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 3.8621955288621956,
|
| 3286 |
+
"grad_norm": 0.00017597030091565102,
|
| 3287 |
+
"learning_rate": 1.932523866901474e-06,
|
| 3288 |
+
"loss": 0.0,
|
| 3289 |
+
"step": 11575
|
| 3290 |
+
},
|
| 3291 |
+
{
|
| 3292 |
+
"epoch": 3.870537203870537,
|
| 3293 |
+
"grad_norm": 0.00032824205118231475,
|
| 3294 |
+
"learning_rate": 1.8166651218834e-06,
|
| 3295 |
+
"loss": 0.0,
|
| 3296 |
+
"step": 11600
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 3.878878878878879,
|
| 3300 |
+
"grad_norm": 0.00014257676957640797,
|
| 3301 |
+
"learning_rate": 1.700806376865326e-06,
|
| 3302 |
+
"loss": 0.0,
|
| 3303 |
+
"step": 11625
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 3.8872205538872207,
|
| 3307 |
+
"grad_norm": 0.003545379266142845,
|
| 3308 |
+
"learning_rate": 1.5849476318472518e-06,
|
| 3309 |
+
"loss": 0.0,
|
| 3310 |
+
"step": 11650
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 3.895562228895562,
|
| 3314 |
+
"grad_norm": 0.00017835904145613313,
|
| 3315 |
+
"learning_rate": 1.4690888868291779e-06,
|
| 3316 |
+
"loss": 0.0,
|
| 3317 |
+
"step": 11675
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 3.903903903903904,
|
| 3321 |
+
"grad_norm": 0.000521348207257688,
|
| 3322 |
+
"learning_rate": 1.353230141811104e-06,
|
| 3323 |
+
"loss": 0.0,
|
| 3324 |
+
"step": 11700
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 3.9122455789122457,
|
| 3328 |
+
"grad_norm": 0.0002032867632806301,
|
| 3329 |
+
"learning_rate": 1.23737139679303e-06,
|
| 3330 |
+
"loss": 0.0,
|
| 3331 |
+
"step": 11725
|
| 3332 |
+
},
|
| 3333 |
+
{
|
| 3334 |
+
"epoch": 3.920587253920587,
|
| 3335 |
+
"grad_norm": 0.0018738268408924341,
|
| 3336 |
+
"learning_rate": 1.121512651774956e-06,
|
| 3337 |
+
"loss": 0.0009,
|
| 3338 |
+
"step": 11750
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 3.928928928928929,
|
| 3342 |
+
"grad_norm": 0.0014717292506247759,
|
| 3343 |
+
"learning_rate": 1.005653906756882e-06,
|
| 3344 |
+
"loss": 0.0,
|
| 3345 |
+
"step": 11775
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 3.9372706039372707,
|
| 3349 |
+
"grad_norm": 0.00021975289564579725,
|
| 3350 |
+
"learning_rate": 8.897951617388081e-07,
|
| 3351 |
+
"loss": 0.0,
|
| 3352 |
+
"step": 11800
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 3.945612278945612,
|
| 3356 |
+
"grad_norm": 0.0005329960258677602,
|
| 3357 |
+
"learning_rate": 7.739364167207342e-07,
|
| 3358 |
+
"loss": 0.0001,
|
| 3359 |
+
"step": 11825
|
| 3360 |
+
},
|
| 3361 |
+
{
|
| 3362 |
+
"epoch": 3.953953953953954,
|
| 3363 |
+
"grad_norm": 0.0001847147213993594,
|
| 3364 |
+
"learning_rate": 6.580776717026602e-07,
|
| 3365 |
+
"loss": 0.0,
|
| 3366 |
+
"step": 11850
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 3.9622956289622957,
|
| 3370 |
+
"grad_norm": 0.00019793544197455049,
|
| 3371 |
+
"learning_rate": 5.422189266845861e-07,
|
| 3372 |
+
"loss": 0.0,
|
| 3373 |
+
"step": 11875
|
| 3374 |
+
},
|
| 3375 |
+
{
|
| 3376 |
+
"epoch": 3.970637303970637,
|
| 3377 |
+
"grad_norm": 0.0004685576423071325,
|
| 3378 |
+
"learning_rate": 4.2636018166651216e-07,
|
| 3379 |
+
"loss": 0.0,
|
| 3380 |
+
"step": 11900
|
| 3381 |
+
},
|
| 3382 |
+
{
|
| 3383 |
+
"epoch": 3.978978978978979,
|
| 3384 |
+
"grad_norm": 0.00016259195399470627,
|
| 3385 |
+
"learning_rate": 3.1050143664843824e-07,
|
| 3386 |
+
"loss": 0.0,
|
| 3387 |
+
"step": 11925
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 3.9873206539873207,
|
| 3391 |
+
"grad_norm": 0.0003845692554023117,
|
| 3392 |
+
"learning_rate": 1.9464269163036427e-07,
|
| 3393 |
+
"loss": 0.0,
|
| 3394 |
+
"step": 11950
|
| 3395 |
+
},
|
| 3396 |
+
{
|
| 3397 |
+
"epoch": 3.995662328995662,
|
| 3398 |
+
"grad_norm": 0.0001322088937740773,
|
| 3399 |
+
"learning_rate": 7.87839466122903e-08,
|
| 3400 |
+
"loss": 0.0225,
|
| 3401 |
+
"step": 11975
|
| 3402 |
+
},
|
| 3403 |
+
{
|
| 3404 |
+
"epoch": 4.0,
|
| 3405 |
+
"eval_accuracy": 0.9991658324991658,
|
| 3406 |
+
"eval_auc": 0.9999155405405404,
|
| 3407 |
+
"eval_f1": 0.9659863945578231,
|
| 3408 |
+
"eval_loss": 0.002970785601064563,
|
| 3409 |
+
"eval_precision": 0.9726027397260274,
|
| 3410 |
+
"eval_recall": 0.9594594594594594,
|
| 3411 |
+
"eval_runtime": 77.7268,
|
| 3412 |
+
"eval_samples_per_second": 77.116,
|
| 3413 |
+
"eval_steps_per_second": 4.825,
|
| 3414 |
+
"step": 11988
|
| 3415 |
+
}
|
| 3416 |
+
],
|
| 3417 |
+
"logging_steps": 25,
|
| 3418 |
+
"max_steps": 11988,
|
| 3419 |
+
"num_input_tokens_seen": 0,
|
| 3420 |
+
"num_train_epochs": 4,
|
| 3421 |
+
"save_steps": 500,
|
| 3422 |
+
"stateful_callbacks": {
|
| 3423 |
+
"EarlyStoppingCallback": {
|
| 3424 |
+
"args": {
|
| 3425 |
+
"early_stopping_patience": 5,
|
| 3426 |
+
"early_stopping_threshold": 0.01
|
| 3427 |
+
},
|
| 3428 |
+
"attributes": {
|
| 3429 |
+
"early_stopping_patience_counter": 3
|
| 3430 |
+
}
|
| 3431 |
+
},
|
| 3432 |
+
"TrainerControl": {
|
| 3433 |
+
"args": {
|
| 3434 |
+
"should_epoch_stop": false,
|
| 3435 |
+
"should_evaluate": false,
|
| 3436 |
+
"should_log": false,
|
| 3437 |
+
"should_save": true,
|
| 3438 |
+
"should_training_stop": true
|
| 3439 |
+
},
|
| 3440 |
+
"attributes": {}
|
| 3441 |
+
}
|
| 3442 |
+
},
|
| 3443 |
+
"total_flos": 2.1910999185868456e+19,
|
| 3444 |
+
"train_batch_size": 8,
|
| 3445 |
+
"trial_name": null,
|
| 3446 |
+
"trial_params": null
|
| 3447 |
+
}
|
checkpoint-11988/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1e1f3384e989a1539b5906b36be83b4eac30c9e6deb0978c70bb1e0130837932
|
| 3 |
+
size 5240
|
config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-base-patch16-384",
|
| 3 |
+
"_num_labels": 2,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"ViTForImageClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_probs_dropout_prob": 0.0,
|
| 8 |
+
"encoder_stride": 16,
|
| 9 |
+
"hidden_act": "gelu",
|
| 10 |
+
"hidden_dropout_prob": 0.0,
|
| 11 |
+
"hidden_size": 768,
|
| 12 |
+
"id2label": {
|
| 13 |
+
"0": "not_soyjak",
|
| 14 |
+
"1": "soyjak"
|
| 15 |
+
},
|
| 16 |
+
"image_size": 384,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 3072,
|
| 19 |
+
"label2id": {
|
| 20 |
+
"not_soyjak": 0,
|
| 21 |
+
"soyjak": 1
|
| 22 |
+
},
|
| 23 |
+
"layer_norm_eps": 1e-12,
|
| 24 |
+
"model_type": "vit",
|
| 25 |
+
"num_attention_heads": 12,
|
| 26 |
+
"num_channels": 3,
|
| 27 |
+
"num_hidden_layers": 12,
|
| 28 |
+
"patch_size": 16,
|
| 29 |
+
"problem_type": "single_label_classification",
|
| 30 |
+
"qkv_bias": true,
|
| 31 |
+
"torch_dtype": "float32",
|
| 32 |
+
"transformers_version": "4.45.0"
|
| 33 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:79c52f74f6defcd2becb5f8c8d19d415c07a5ce8587e929a351d1a886356b70f
|
| 3 |
+
size 344391328
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_normalize": true,
|
| 3 |
+
"do_rescale": true,
|
| 4 |
+
"do_resize": true,
|
| 5 |
+
"image_mean": [
|
| 6 |
+
0.5,
|
| 7 |
+
0.5,
|
| 8 |
+
0.5
|
| 9 |
+
],
|
| 10 |
+
"image_processor_type": "ViTImageProcessor",
|
| 11 |
+
"image_std": [
|
| 12 |
+
0.5,
|
| 13 |
+
0.5,
|
| 14 |
+
0.5
|
| 15 |
+
],
|
| 16 |
+
"resample": 2,
|
| 17 |
+
"rescale_factor": 0.00392156862745098,
|
| 18 |
+
"size": {
|
| 19 |
+
"height": 384,
|
| 20 |
+
"width": 384
|
| 21 |
+
}
|
| 22 |
+
}
|
runs/Oct21_02-57-43_r-haiefff-sjstuff-xmuxso1w-882ba-zoike/events.out.tfevents.1729479465.r-haiefff-sjstuff-xmuxso1w-882ba-zoike.172.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a7d6050b550d161da793e6a838f5d542515bfbffd48dea5d9ab95b65f72d9f7
|
| 3 |
+
size 108437
|
runs/Oct21_02-57-43_r-haiefff-sjstuff-xmuxso1w-882ba-zoike/events.out.tfevents.1729483109.r-haiefff-sjstuff-xmuxso1w-882ba-zoike.172.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d4fba17fc52bfc3f79b37e671a4269994a8362f1e30d2c52cab261b385e1295b
|
| 3 |
+
size 607
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1e1f3384e989a1539b5906b36be83b4eac30c9e6deb0978c70bb1e0130837932
|
| 3 |
+
size 5240
|
training_params.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "sj-det-3/autotrain-data",
|
| 3 |
+
"model": "google/vit-base-patch16-384",
|
| 4 |
+
"username": "haiefff",
|
| 5 |
+
"lr": 5e-05,
|
| 6 |
+
"epochs": 4,
|
| 7 |
+
"batch_size": 8,
|
| 8 |
+
"warmup_ratio": 0.1,
|
| 9 |
+
"gradient_accumulation": 1,
|
| 10 |
+
"optimizer": "adamw_torch",
|
| 11 |
+
"scheduler": "linear",
|
| 12 |
+
"weight_decay": 0.0,
|
| 13 |
+
"max_grad_norm": 1.0,
|
| 14 |
+
"seed": 42,
|
| 15 |
+
"train_split": "train",
|
| 16 |
+
"valid_split": "validation",
|
| 17 |
+
"logging_steps": -1,
|
| 18 |
+
"project_name": "sj-det-3",
|
| 19 |
+
"auto_find_batch_size": false,
|
| 20 |
+
"mixed_precision": "fp16",
|
| 21 |
+
"save_total_limit": 1,
|
| 22 |
+
"push_to_hub": true,
|
| 23 |
+
"eval_strategy": "epoch",
|
| 24 |
+
"image_column": "autotrain_image",
|
| 25 |
+
"target_column": "autotrain_label",
|
| 26 |
+
"log": "tensorboard",
|
| 27 |
+
"early_stopping_patience": 5,
|
| 28 |
+
"early_stopping_threshold": 0.01
|
| 29 |
+
}
|