Upload folder using huggingface_hub
Browse files- README.md +41 -0
- checkpoint-4950/config.json +53 -0
- checkpoint-4950/model.safetensors +3 -0
- checkpoint-4950/optimizer.pt +3 -0
- checkpoint-4950/rng_state.pth +3 -0
- checkpoint-4950/scheduler.pt +3 -0
- checkpoint-4950/trainer_state.json +1482 -0
- checkpoint-4950/training_args.bin +3 -0
- config.json +53 -0
- model.safetensors +3 -0
- preprocessor_config.json +22 -0
- runs/Oct28_19-20-06_r-compstak-document-ai-fine-tuning-z11p1xee-34b24-oo9j1/events.out.tfevents.1730143207.r-compstak-document-ai-fine-tuning-z11p1xee-34b24-oo9j1.1353.0 +2 -2
- runs/Oct28_19-20-06_r-compstak-document-ai-fine-tuning-z11p1xee-34b24-oo9j1/events.out.tfevents.1730144209.r-compstak-document-ai-fine-tuning-z11p1xee-34b24-oo9j1.1353.1 +3 -0
- training_args.bin +3 -0
- training_params.json +29 -0
README.md
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
---
|
| 3 |
+
tags:
|
| 4 |
+
- autotrain
|
| 5 |
+
- image-classification
|
| 6 |
+
base_model: google/vit-large-patch16-224
|
| 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: 0.3821594715118408
|
| 22 |
+
|
| 23 |
+
f1_macro: 0.8644630522360383
|
| 24 |
+
|
| 25 |
+
f1_micro: 0.886969696969697
|
| 26 |
+
|
| 27 |
+
f1_weighted: 0.8837489529217776
|
| 28 |
+
|
| 29 |
+
precision_macro: 0.8700338902181693
|
| 30 |
+
|
| 31 |
+
precision_micro: 0.886969696969697
|
| 32 |
+
|
| 33 |
+
precision_weighted: 0.8838390180385471
|
| 34 |
+
|
| 35 |
+
recall_macro: 0.8628333333333335
|
| 36 |
+
|
| 37 |
+
recall_micro: 0.886969696969697
|
| 38 |
+
|
| 39 |
+
recall_weighted: 0.886969696969697
|
| 40 |
+
|
| 41 |
+
accuracy: 0.886969696969697
|
checkpoint-4950/config.json
ADDED
|
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-large-patch16-224",
|
| 3 |
+
"_num_labels": 12,
|
| 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": 1024,
|
| 12 |
+
"id2label": {
|
| 13 |
+
"0": "Appraisal Report",
|
| 14 |
+
"1": "Block",
|
| 15 |
+
"2": "Columns",
|
| 16 |
+
"3": "Commision Form",
|
| 17 |
+
"4": "Don't Know",
|
| 18 |
+
"5": "Free Text",
|
| 19 |
+
"6": "Lease Agreement",
|
| 20 |
+
"7": "MLS",
|
| 21 |
+
"8": "Offering Memo",
|
| 22 |
+
"9": "Profile",
|
| 23 |
+
"10": "Rent Roll",
|
| 24 |
+
"11": "Table"
|
| 25 |
+
},
|
| 26 |
+
"image_size": 224,
|
| 27 |
+
"initializer_range": 0.02,
|
| 28 |
+
"intermediate_size": 4096,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"Appraisal Report": 0,
|
| 31 |
+
"Block": 1,
|
| 32 |
+
"Columns": 2,
|
| 33 |
+
"Commision Form": 3,
|
| 34 |
+
"Don't Know": 4,
|
| 35 |
+
"Free Text": 5,
|
| 36 |
+
"Lease Agreement": 6,
|
| 37 |
+
"MLS": 7,
|
| 38 |
+
"Offering Memo": 8,
|
| 39 |
+
"Profile": 9,
|
| 40 |
+
"Rent Roll": 10,
|
| 41 |
+
"Table": 11
|
| 42 |
+
},
|
| 43 |
+
"layer_norm_eps": 1e-12,
|
| 44 |
+
"model_type": "vit",
|
| 45 |
+
"num_attention_heads": 16,
|
| 46 |
+
"num_channels": 3,
|
| 47 |
+
"num_hidden_layers": 24,
|
| 48 |
+
"patch_size": 16,
|
| 49 |
+
"problem_type": "single_label_classification",
|
| 50 |
+
"qkv_bias": true,
|
| 51 |
+
"torch_dtype": "float32",
|
| 52 |
+
"transformers_version": "4.45.0"
|
| 53 |
+
}
|
checkpoint-4950/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:158a10c1203884182679a6e220f39cf8280d7ec317bd6af4ebdeb5ddd7c512ec
|
| 3 |
+
size 1213302272
|
checkpoint-4950/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5f6805ec665ba846643b8b6096f03b69d2d29e13c6c255b0c2d6ac0b15a28b84
|
| 3 |
+
size 2426839466
|
checkpoint-4950/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:31f71d682aa0c2e4ea00eba2b6d4f1a8b122f32ea22190485863ebcf93527382
|
| 3 |
+
size 14244
|
checkpoint-4950/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a6c883fc03f4c5e1952b5ce7222f8096e08b0228d866f9f131599b16338ef0c0
|
| 3 |
+
size 1064
|
checkpoint-4950/trainer_state.json
ADDED
|
@@ -0,0 +1,1482 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.3821594715118408,
|
| 3 |
+
"best_model_checkpoint": "classify-google-augment-3/checkpoint-4950",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4950,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.015151515151515152,
|
| 13 |
+
"grad_norm": 12.085535049438477,
|
| 14 |
+
"learning_rate": 2.3232323232323234e-06,
|
| 15 |
+
"loss": 2.4513,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.030303030303030304,
|
| 20 |
+
"grad_norm": 10.15959358215332,
|
| 21 |
+
"learning_rate": 4.848484848484849e-06,
|
| 22 |
+
"loss": 2.234,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.045454545454545456,
|
| 27 |
+
"grad_norm": 12.21349048614502,
|
| 28 |
+
"learning_rate": 7.3737373737373745e-06,
|
| 29 |
+
"loss": 1.9342,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.06060606060606061,
|
| 34 |
+
"grad_norm": 13.986493110656738,
|
| 35 |
+
"learning_rate": 9.898989898989899e-06,
|
| 36 |
+
"loss": 1.7956,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.07575757575757576,
|
| 41 |
+
"grad_norm": 13.179245948791504,
|
| 42 |
+
"learning_rate": 1.2424242424242424e-05,
|
| 43 |
+
"loss": 1.7165,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.09090909090909091,
|
| 48 |
+
"grad_norm": 15.372575759887695,
|
| 49 |
+
"learning_rate": 1.494949494949495e-05,
|
| 50 |
+
"loss": 1.519,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.10606060606060606,
|
| 55 |
+
"grad_norm": 9.555817604064941,
|
| 56 |
+
"learning_rate": 1.7474747474747475e-05,
|
| 57 |
+
"loss": 1.4716,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.12121212121212122,
|
| 62 |
+
"grad_norm": 14.15468692779541,
|
| 63 |
+
"learning_rate": 2e-05,
|
| 64 |
+
"loss": 1.5611,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.13636363636363635,
|
| 69 |
+
"grad_norm": 9.72482681274414,
|
| 70 |
+
"learning_rate": 2.2525252525252528e-05,
|
| 71 |
+
"loss": 1.3692,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.15151515151515152,
|
| 76 |
+
"grad_norm": 11.219098091125488,
|
| 77 |
+
"learning_rate": 2.505050505050505e-05,
|
| 78 |
+
"loss": 1.3761,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.16666666666666666,
|
| 83 |
+
"grad_norm": 14.293442726135254,
|
| 84 |
+
"learning_rate": 2.7575757575757578e-05,
|
| 85 |
+
"loss": 1.5677,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.18181818181818182,
|
| 90 |
+
"grad_norm": 12.68210506439209,
|
| 91 |
+
"learning_rate": 3.01010101010101e-05,
|
| 92 |
+
"loss": 1.4212,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.19696969696969696,
|
| 97 |
+
"grad_norm": 10.856423377990723,
|
| 98 |
+
"learning_rate": 3.2626262626262624e-05,
|
| 99 |
+
"loss": 1.3204,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.21212121212121213,
|
| 104 |
+
"grad_norm": 11.644044876098633,
|
| 105 |
+
"learning_rate": 3.515151515151515e-05,
|
| 106 |
+
"loss": 1.3228,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.22727272727272727,
|
| 111 |
+
"grad_norm": 7.3117356300354,
|
| 112 |
+
"learning_rate": 3.767676767676768e-05,
|
| 113 |
+
"loss": 1.4005,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.24242424242424243,
|
| 118 |
+
"grad_norm": 12.309542655944824,
|
| 119 |
+
"learning_rate": 4.0202020202020204e-05,
|
| 120 |
+
"loss": 1.2739,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.25757575757575757,
|
| 125 |
+
"grad_norm": 11.981978416442871,
|
| 126 |
+
"learning_rate": 4.2727272727272724e-05,
|
| 127 |
+
"loss": 1.3082,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.2727272727272727,
|
| 132 |
+
"grad_norm": 8.469183921813965,
|
| 133 |
+
"learning_rate": 4.525252525252526e-05,
|
| 134 |
+
"loss": 1.1932,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.2878787878787879,
|
| 139 |
+
"grad_norm": 13.018028259277344,
|
| 140 |
+
"learning_rate": 4.7777777777777784e-05,
|
| 141 |
+
"loss": 1.2612,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.30303030303030304,
|
| 146 |
+
"grad_norm": 11.944835662841797,
|
| 147 |
+
"learning_rate": 4.9966329966329964e-05,
|
| 148 |
+
"loss": 1.3266,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.3181818181818182,
|
| 153 |
+
"grad_norm": 11.489705085754395,
|
| 154 |
+
"learning_rate": 4.968574635241302e-05,
|
| 155 |
+
"loss": 1.3537,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.3333333333333333,
|
| 160 |
+
"grad_norm": 14.265945434570312,
|
| 161 |
+
"learning_rate": 4.940516273849607e-05,
|
| 162 |
+
"loss": 1.4162,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.3484848484848485,
|
| 167 |
+
"grad_norm": 10.913326263427734,
|
| 168 |
+
"learning_rate": 4.912457912457913e-05,
|
| 169 |
+
"loss": 1.0754,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.36363636363636365,
|
| 174 |
+
"grad_norm": 7.5248236656188965,
|
| 175 |
+
"learning_rate": 4.884399551066218e-05,
|
| 176 |
+
"loss": 1.3724,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.3787878787878788,
|
| 181 |
+
"grad_norm": 9.98138427734375,
|
| 182 |
+
"learning_rate": 4.856341189674523e-05,
|
| 183 |
+
"loss": 1.184,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.3939393939393939,
|
| 188 |
+
"grad_norm": 9.07705307006836,
|
| 189 |
+
"learning_rate": 4.828282828282829e-05,
|
| 190 |
+
"loss": 1.0418,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.4090909090909091,
|
| 195 |
+
"grad_norm": 8.512435913085938,
|
| 196 |
+
"learning_rate": 4.800224466891134e-05,
|
| 197 |
+
"loss": 1.1425,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.42424242424242425,
|
| 202 |
+
"grad_norm": 13.559457778930664,
|
| 203 |
+
"learning_rate": 4.7721661054994394e-05,
|
| 204 |
+
"loss": 1.4285,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.4393939393939394,
|
| 209 |
+
"grad_norm": 7.820692539215088,
|
| 210 |
+
"learning_rate": 4.7441077441077445e-05,
|
| 211 |
+
"loss": 1.1221,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.45454545454545453,
|
| 216 |
+
"grad_norm": 6.074573040008545,
|
| 217 |
+
"learning_rate": 4.7160493827160495e-05,
|
| 218 |
+
"loss": 0.9862,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.4696969696969697,
|
| 223 |
+
"grad_norm": 6.816392421722412,
|
| 224 |
+
"learning_rate": 4.687991021324355e-05,
|
| 225 |
+
"loss": 1.1894,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.48484848484848486,
|
| 230 |
+
"grad_norm": 9.136533737182617,
|
| 231 |
+
"learning_rate": 4.65993265993266e-05,
|
| 232 |
+
"loss": 1.0674,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5,
|
| 237 |
+
"grad_norm": 8.841058731079102,
|
| 238 |
+
"learning_rate": 4.631874298540965e-05,
|
| 239 |
+
"loss": 1.0798,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5151515151515151,
|
| 244 |
+
"grad_norm": 10.076774597167969,
|
| 245 |
+
"learning_rate": 4.60381593714927e-05,
|
| 246 |
+
"loss": 1.0903,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5303030303030303,
|
| 251 |
+
"grad_norm": 13.101428985595703,
|
| 252 |
+
"learning_rate": 4.575757575757576e-05,
|
| 253 |
+
"loss": 1.0672,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5454545454545454,
|
| 258 |
+
"grad_norm": 10.879769325256348,
|
| 259 |
+
"learning_rate": 4.547699214365882e-05,
|
| 260 |
+
"loss": 1.005,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.5606060606060606,
|
| 265 |
+
"grad_norm": 5.972232341766357,
|
| 266 |
+
"learning_rate": 4.519640852974186e-05,
|
| 267 |
+
"loss": 1.026,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.5757575757575758,
|
| 272 |
+
"grad_norm": 12.992452621459961,
|
| 273 |
+
"learning_rate": 4.491582491582492e-05,
|
| 274 |
+
"loss": 1.0436,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.5909090909090909,
|
| 279 |
+
"grad_norm": 11.431863784790039,
|
| 280 |
+
"learning_rate": 4.463524130190797e-05,
|
| 281 |
+
"loss": 1.006,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.6060606060606061,
|
| 286 |
+
"grad_norm": 13.05385971069336,
|
| 287 |
+
"learning_rate": 4.4354657687991025e-05,
|
| 288 |
+
"loss": 1.1255,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.6212121212121212,
|
| 293 |
+
"grad_norm": 9.93174934387207,
|
| 294 |
+
"learning_rate": 4.4074074074074076e-05,
|
| 295 |
+
"loss": 1.0457,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.6363636363636364,
|
| 300 |
+
"grad_norm": 14.331412315368652,
|
| 301 |
+
"learning_rate": 4.3793490460157126e-05,
|
| 302 |
+
"loss": 0.997,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.6515151515151515,
|
| 307 |
+
"grad_norm": 11.081893920898438,
|
| 308 |
+
"learning_rate": 4.351290684624018e-05,
|
| 309 |
+
"loss": 0.999,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.6666666666666666,
|
| 314 |
+
"grad_norm": 8.453628540039062,
|
| 315 |
+
"learning_rate": 4.3232323232323234e-05,
|
| 316 |
+
"loss": 1.0695,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.6818181818181818,
|
| 321 |
+
"grad_norm": 8.907005310058594,
|
| 322 |
+
"learning_rate": 4.295173961840629e-05,
|
| 323 |
+
"loss": 0.967,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.696969696969697,
|
| 328 |
+
"grad_norm": 10.75546646118164,
|
| 329 |
+
"learning_rate": 4.267115600448934e-05,
|
| 330 |
+
"loss": 1.0517,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.7121212121212122,
|
| 335 |
+
"grad_norm": 9.127320289611816,
|
| 336 |
+
"learning_rate": 4.239057239057239e-05,
|
| 337 |
+
"loss": 1.0049,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.7272727272727273,
|
| 342 |
+
"grad_norm": 10.124237060546875,
|
| 343 |
+
"learning_rate": 4.210998877665545e-05,
|
| 344 |
+
"loss": 0.969,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.7424242424242424,
|
| 349 |
+
"grad_norm": 5.617876052856445,
|
| 350 |
+
"learning_rate": 4.18294051627385e-05,
|
| 351 |
+
"loss": 1.0031,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.7575757575757576,
|
| 356 |
+
"grad_norm": 11.184277534484863,
|
| 357 |
+
"learning_rate": 4.154882154882155e-05,
|
| 358 |
+
"loss": 1.0209,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.7727272727272727,
|
| 363 |
+
"grad_norm": 6.545035362243652,
|
| 364 |
+
"learning_rate": 4.12682379349046e-05,
|
| 365 |
+
"loss": 0.8416,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.7878787878787878,
|
| 370 |
+
"grad_norm": 4.944152355194092,
|
| 371 |
+
"learning_rate": 4.0987654320987657e-05,
|
| 372 |
+
"loss": 1.0029,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.803030303030303,
|
| 377 |
+
"grad_norm": 8.916969299316406,
|
| 378 |
+
"learning_rate": 4.070707070707071e-05,
|
| 379 |
+
"loss": 0.8808,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.8181818181818182,
|
| 384 |
+
"grad_norm": 4.9258713722229,
|
| 385 |
+
"learning_rate": 4.0426487093153764e-05,
|
| 386 |
+
"loss": 0.9536,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.8333333333333334,
|
| 391 |
+
"grad_norm": 11.556329727172852,
|
| 392 |
+
"learning_rate": 4.0145903479236814e-05,
|
| 393 |
+
"loss": 0.937,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.8484848484848485,
|
| 398 |
+
"grad_norm": 10.06991958618164,
|
| 399 |
+
"learning_rate": 3.9865319865319865e-05,
|
| 400 |
+
"loss": 0.8424,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.8636363636363636,
|
| 405 |
+
"grad_norm": 8.390497207641602,
|
| 406 |
+
"learning_rate": 3.958473625140292e-05,
|
| 407 |
+
"loss": 0.9453,
|
| 408 |
+
"step": 1425
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.8787878787878788,
|
| 412 |
+
"grad_norm": 8.02001667022705,
|
| 413 |
+
"learning_rate": 3.930415263748597e-05,
|
| 414 |
+
"loss": 0.8522,
|
| 415 |
+
"step": 1450
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.8939393939393939,
|
| 419 |
+
"grad_norm": 8.541698455810547,
|
| 420 |
+
"learning_rate": 3.902356902356902e-05,
|
| 421 |
+
"loss": 0.9519,
|
| 422 |
+
"step": 1475
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.9090909090909091,
|
| 426 |
+
"grad_norm": 7.404212951660156,
|
| 427 |
+
"learning_rate": 3.874298540965208e-05,
|
| 428 |
+
"loss": 0.9692,
|
| 429 |
+
"step": 1500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.9242424242424242,
|
| 433 |
+
"grad_norm": 10.684733390808105,
|
| 434 |
+
"learning_rate": 3.846240179573513e-05,
|
| 435 |
+
"loss": 1.0417,
|
| 436 |
+
"step": 1525
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.9393939393939394,
|
| 440 |
+
"grad_norm": 6.182531833648682,
|
| 441 |
+
"learning_rate": 3.818181818181819e-05,
|
| 442 |
+
"loss": 0.8632,
|
| 443 |
+
"step": 1550
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.9545454545454546,
|
| 447 |
+
"grad_norm": 8.63337516784668,
|
| 448 |
+
"learning_rate": 3.790123456790123e-05,
|
| 449 |
+
"loss": 0.9318,
|
| 450 |
+
"step": 1575
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.9696969696969697,
|
| 454 |
+
"grad_norm": 11.64743709564209,
|
| 455 |
+
"learning_rate": 3.762065095398429e-05,
|
| 456 |
+
"loss": 0.8535,
|
| 457 |
+
"step": 1600
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.9848484848484849,
|
| 461 |
+
"grad_norm": 11.15438461303711,
|
| 462 |
+
"learning_rate": 3.7340067340067345e-05,
|
| 463 |
+
"loss": 0.9558,
|
| 464 |
+
"step": 1625
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.0,
|
| 468 |
+
"grad_norm": 7.817443370819092,
|
| 469 |
+
"learning_rate": 3.7059483726150395e-05,
|
| 470 |
+
"loss": 0.9244,
|
| 471 |
+
"step": 1650
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 1.0,
|
| 475 |
+
"eval_accuracy": 0.7645454545454545,
|
| 476 |
+
"eval_f1_macro": 0.6989492792384068,
|
| 477 |
+
"eval_f1_micro": 0.7645454545454545,
|
| 478 |
+
"eval_f1_weighted": 0.7496986753676582,
|
| 479 |
+
"eval_loss": 0.752944827079773,
|
| 480 |
+
"eval_precision_macro": 0.8012284685903541,
|
| 481 |
+
"eval_precision_micro": 0.7645454545454545,
|
| 482 |
+
"eval_precision_weighted": 0.817408067065806,
|
| 483 |
+
"eval_recall_macro": 0.7035,
|
| 484 |
+
"eval_recall_micro": 0.7645454545454545,
|
| 485 |
+
"eval_recall_weighted": 0.7645454545454545,
|
| 486 |
+
"eval_runtime": 18.9801,
|
| 487 |
+
"eval_samples_per_second": 173.866,
|
| 488 |
+
"eval_steps_per_second": 10.906,
|
| 489 |
+
"step": 1650
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.0151515151515151,
|
| 493 |
+
"grad_norm": 10.560444831848145,
|
| 494 |
+
"learning_rate": 3.677890011223345e-05,
|
| 495 |
+
"loss": 0.7833,
|
| 496 |
+
"step": 1675
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 1.0303030303030303,
|
| 500 |
+
"grad_norm": 4.251949310302734,
|
| 501 |
+
"learning_rate": 3.6498316498316496e-05,
|
| 502 |
+
"loss": 0.7379,
|
| 503 |
+
"step": 1700
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.0454545454545454,
|
| 507 |
+
"grad_norm": 10.605439186096191,
|
| 508 |
+
"learning_rate": 3.621773288439955e-05,
|
| 509 |
+
"loss": 0.7779,
|
| 510 |
+
"step": 1725
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 1.0606060606060606,
|
| 514 |
+
"grad_norm": 7.418851375579834,
|
| 515 |
+
"learning_rate": 3.59371492704826e-05,
|
| 516 |
+
"loss": 0.7347,
|
| 517 |
+
"step": 1750
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 1.0757575757575757,
|
| 521 |
+
"grad_norm": 10.209125518798828,
|
| 522 |
+
"learning_rate": 3.565656565656566e-05,
|
| 523 |
+
"loss": 0.8636,
|
| 524 |
+
"step": 1775
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 1.0909090909090908,
|
| 528 |
+
"grad_norm": 5.001626491546631,
|
| 529 |
+
"learning_rate": 3.537598204264871e-05,
|
| 530 |
+
"loss": 0.8716,
|
| 531 |
+
"step": 1800
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.106060606060606,
|
| 535 |
+
"grad_norm": 5.035640239715576,
|
| 536 |
+
"learning_rate": 3.509539842873176e-05,
|
| 537 |
+
"loss": 0.7529,
|
| 538 |
+
"step": 1825
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.121212121212121,
|
| 542 |
+
"grad_norm": 7.653929710388184,
|
| 543 |
+
"learning_rate": 3.481481481481482e-05,
|
| 544 |
+
"loss": 0.8541,
|
| 545 |
+
"step": 1850
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.1363636363636362,
|
| 549 |
+
"grad_norm": 11.623481750488281,
|
| 550 |
+
"learning_rate": 3.453423120089787e-05,
|
| 551 |
+
"loss": 0.8159,
|
| 552 |
+
"step": 1875
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.1515151515151516,
|
| 556 |
+
"grad_norm": 9.897392272949219,
|
| 557 |
+
"learning_rate": 3.425364758698092e-05,
|
| 558 |
+
"loss": 0.5642,
|
| 559 |
+
"step": 1900
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 1.1666666666666667,
|
| 563 |
+
"grad_norm": 8.24569320678711,
|
| 564 |
+
"learning_rate": 3.3973063973063976e-05,
|
| 565 |
+
"loss": 1.0238,
|
| 566 |
+
"step": 1925
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 1.1818181818181819,
|
| 570 |
+
"grad_norm": 6.044914245605469,
|
| 571 |
+
"learning_rate": 3.3692480359147026e-05,
|
| 572 |
+
"loss": 0.8917,
|
| 573 |
+
"step": 1950
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 1.196969696969697,
|
| 577 |
+
"grad_norm": 5.3622002601623535,
|
| 578 |
+
"learning_rate": 3.3411896745230084e-05,
|
| 579 |
+
"loss": 0.7951,
|
| 580 |
+
"step": 1975
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 1.2121212121212122,
|
| 584 |
+
"grad_norm": 8.479634284973145,
|
| 585 |
+
"learning_rate": 3.3131313131313134e-05,
|
| 586 |
+
"loss": 0.9219,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.2272727272727273,
|
| 591 |
+
"grad_norm": 10.730965614318848,
|
| 592 |
+
"learning_rate": 3.2850729517396184e-05,
|
| 593 |
+
"loss": 0.6391,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.2424242424242424,
|
| 598 |
+
"grad_norm": 2.901815891265869,
|
| 599 |
+
"learning_rate": 3.2570145903479235e-05,
|
| 600 |
+
"loss": 0.7946,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 1.2575757575757576,
|
| 605 |
+
"grad_norm": 7.174593448638916,
|
| 606 |
+
"learning_rate": 3.228956228956229e-05,
|
| 607 |
+
"loss": 0.7458,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.2727272727272727,
|
| 612 |
+
"grad_norm": 5.743284702301025,
|
| 613 |
+
"learning_rate": 3.200897867564535e-05,
|
| 614 |
+
"loss": 0.8514,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.2878787878787878,
|
| 619 |
+
"grad_norm": 5.6034464836120605,
|
| 620 |
+
"learning_rate": 3.172839506172839e-05,
|
| 621 |
+
"loss": 0.5711,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.303030303030303,
|
| 626 |
+
"grad_norm": 6.2251996994018555,
|
| 627 |
+
"learning_rate": 3.144781144781145e-05,
|
| 628 |
+
"loss": 0.8115,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.3181818181818181,
|
| 633 |
+
"grad_norm": 6.7410078048706055,
|
| 634 |
+
"learning_rate": 3.11672278338945e-05,
|
| 635 |
+
"loss": 0.7262,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.3333333333333333,
|
| 640 |
+
"grad_norm": 5.232451438903809,
|
| 641 |
+
"learning_rate": 3.088664421997756e-05,
|
| 642 |
+
"loss": 0.7573,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.3484848484848486,
|
| 647 |
+
"grad_norm": 4.0043439865112305,
|
| 648 |
+
"learning_rate": 3.060606060606061e-05,
|
| 649 |
+
"loss": 0.9982,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.3636363636363638,
|
| 654 |
+
"grad_norm": 8.409883499145508,
|
| 655 |
+
"learning_rate": 3.0325476992143658e-05,
|
| 656 |
+
"loss": 0.7902,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.378787878787879,
|
| 661 |
+
"grad_norm": 8.475225448608398,
|
| 662 |
+
"learning_rate": 3.004489337822671e-05,
|
| 663 |
+
"loss": 0.8343,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.393939393939394,
|
| 668 |
+
"grad_norm": 4.989223480224609,
|
| 669 |
+
"learning_rate": 2.976430976430977e-05,
|
| 670 |
+
"loss": 0.8126,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.4090909090909092,
|
| 675 |
+
"grad_norm": 5.526180267333984,
|
| 676 |
+
"learning_rate": 2.9483726150392822e-05,
|
| 677 |
+
"loss": 0.6589,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.4242424242424243,
|
| 682 |
+
"grad_norm": 7.382469177246094,
|
| 683 |
+
"learning_rate": 2.920314253647587e-05,
|
| 684 |
+
"loss": 0.7384,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.4393939393939394,
|
| 689 |
+
"grad_norm": 5.158092498779297,
|
| 690 |
+
"learning_rate": 2.8922558922558923e-05,
|
| 691 |
+
"loss": 0.6578,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.4545454545454546,
|
| 696 |
+
"grad_norm": 4.976596832275391,
|
| 697 |
+
"learning_rate": 2.8641975308641977e-05,
|
| 698 |
+
"loss": 0.8378,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.4696969696969697,
|
| 703 |
+
"grad_norm": 9.176383018493652,
|
| 704 |
+
"learning_rate": 2.836139169472503e-05,
|
| 705 |
+
"loss": 0.7935,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.4848484848484849,
|
| 710 |
+
"grad_norm": 10.877906799316406,
|
| 711 |
+
"learning_rate": 2.808080808080808e-05,
|
| 712 |
+
"loss": 0.8241,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.5,
|
| 717 |
+
"grad_norm": 12.3035249710083,
|
| 718 |
+
"learning_rate": 2.7800224466891134e-05,
|
| 719 |
+
"loss": 0.8298,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.5151515151515151,
|
| 724 |
+
"grad_norm": 7.093743324279785,
|
| 725 |
+
"learning_rate": 2.7519640852974188e-05,
|
| 726 |
+
"loss": 0.7239,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.5303030303030303,
|
| 731 |
+
"grad_norm": 14.64828872680664,
|
| 732 |
+
"learning_rate": 2.7239057239057242e-05,
|
| 733 |
+
"loss": 0.7086,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.5454545454545454,
|
| 738 |
+
"grad_norm": 5.189187526702881,
|
| 739 |
+
"learning_rate": 2.6958473625140296e-05,
|
| 740 |
+
"loss": 0.7458,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.5606060606060606,
|
| 745 |
+
"grad_norm": 7.915388107299805,
|
| 746 |
+
"learning_rate": 2.6677890011223346e-05,
|
| 747 |
+
"loss": 0.765,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.5757575757575757,
|
| 752 |
+
"grad_norm": 5.491411209106445,
|
| 753 |
+
"learning_rate": 2.63973063973064e-05,
|
| 754 |
+
"loss": 0.8382,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.5909090909090908,
|
| 759 |
+
"grad_norm": 10.231388092041016,
|
| 760 |
+
"learning_rate": 2.6116722783389453e-05,
|
| 761 |
+
"loss": 0.6905,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.606060606060606,
|
| 766 |
+
"grad_norm": 4.599724769592285,
|
| 767 |
+
"learning_rate": 2.5836139169472507e-05,
|
| 768 |
+
"loss": 0.7544,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.621212121212121,
|
| 773 |
+
"grad_norm": 9.038063049316406,
|
| 774 |
+
"learning_rate": 2.5555555555555554e-05,
|
| 775 |
+
"loss": 0.5843,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.6363636363636362,
|
| 780 |
+
"grad_norm": 7.875892639160156,
|
| 781 |
+
"learning_rate": 2.5274971941638608e-05,
|
| 782 |
+
"loss": 0.7318,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.6515151515151514,
|
| 787 |
+
"grad_norm": 7.997163772583008,
|
| 788 |
+
"learning_rate": 2.499438832772166e-05,
|
| 789 |
+
"loss": 0.6789,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.6666666666666665,
|
| 794 |
+
"grad_norm": 12.051643371582031,
|
| 795 |
+
"learning_rate": 2.4725028058361395e-05,
|
| 796 |
+
"loss": 0.8427,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.6818181818181817,
|
| 801 |
+
"grad_norm": 18.78903579711914,
|
| 802 |
+
"learning_rate": 2.4444444444444445e-05,
|
| 803 |
+
"loss": 0.7387,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.696969696969697,
|
| 808 |
+
"grad_norm": 5.628903865814209,
|
| 809 |
+
"learning_rate": 2.41638608305275e-05,
|
| 810 |
+
"loss": 0.6316,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.7121212121212122,
|
| 815 |
+
"grad_norm": 6.309241771697998,
|
| 816 |
+
"learning_rate": 2.388327721661055e-05,
|
| 817 |
+
"loss": 0.6701,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.7272727272727273,
|
| 822 |
+
"grad_norm": 6.3270392417907715,
|
| 823 |
+
"learning_rate": 2.3602693602693603e-05,
|
| 824 |
+
"loss": 0.5524,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.7424242424242424,
|
| 829 |
+
"grad_norm": 3.400372266769409,
|
| 830 |
+
"learning_rate": 2.3322109988776656e-05,
|
| 831 |
+
"loss": 0.6322,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.7575757575757576,
|
| 836 |
+
"grad_norm": 10.937955856323242,
|
| 837 |
+
"learning_rate": 2.304152637485971e-05,
|
| 838 |
+
"loss": 0.5497,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.7727272727272727,
|
| 843 |
+
"grad_norm": 9.74146842956543,
|
| 844 |
+
"learning_rate": 2.2760942760942764e-05,
|
| 845 |
+
"loss": 0.7272,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.7878787878787878,
|
| 850 |
+
"grad_norm": 24.714502334594727,
|
| 851 |
+
"learning_rate": 2.2480359147025814e-05,
|
| 852 |
+
"loss": 0.7961,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.803030303030303,
|
| 857 |
+
"grad_norm": 3.8761136531829834,
|
| 858 |
+
"learning_rate": 2.2199775533108868e-05,
|
| 859 |
+
"loss": 0.6801,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.8181818181818183,
|
| 864 |
+
"grad_norm": 5.883018493652344,
|
| 865 |
+
"learning_rate": 2.191919191919192e-05,
|
| 866 |
+
"loss": 0.7082,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.8333333333333335,
|
| 871 |
+
"grad_norm": 7.082830905914307,
|
| 872 |
+
"learning_rate": 2.1638608305274975e-05,
|
| 873 |
+
"loss": 0.5892,
|
| 874 |
+
"step": 3025
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.8484848484848486,
|
| 878 |
+
"grad_norm": 10.057491302490234,
|
| 879 |
+
"learning_rate": 2.1358024691358026e-05,
|
| 880 |
+
"loss": 0.7226,
|
| 881 |
+
"step": 3050
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 1.8636363636363638,
|
| 885 |
+
"grad_norm": 3.2809817790985107,
|
| 886 |
+
"learning_rate": 2.107744107744108e-05,
|
| 887 |
+
"loss": 0.546,
|
| 888 |
+
"step": 3075
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 1.878787878787879,
|
| 892 |
+
"grad_norm": 8.243097305297852,
|
| 893 |
+
"learning_rate": 2.079685746352413e-05,
|
| 894 |
+
"loss": 0.6747,
|
| 895 |
+
"step": 3100
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 1.893939393939394,
|
| 899 |
+
"grad_norm": 7.2149457931518555,
|
| 900 |
+
"learning_rate": 2.0516273849607184e-05,
|
| 901 |
+
"loss": 0.6502,
|
| 902 |
+
"step": 3125
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 1.9090909090909092,
|
| 906 |
+
"grad_norm": 5.218500137329102,
|
| 907 |
+
"learning_rate": 2.0235690235690234e-05,
|
| 908 |
+
"loss": 0.5866,
|
| 909 |
+
"step": 3150
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1.9242424242424243,
|
| 913 |
+
"grad_norm": 8.202606201171875,
|
| 914 |
+
"learning_rate": 1.995510662177329e-05,
|
| 915 |
+
"loss": 0.7029,
|
| 916 |
+
"step": 3175
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 1.9393939393939394,
|
| 920 |
+
"grad_norm": 10.254151344299316,
|
| 921 |
+
"learning_rate": 1.967452300785634e-05,
|
| 922 |
+
"loss": 0.6972,
|
| 923 |
+
"step": 3200
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 1.9545454545454546,
|
| 927 |
+
"grad_norm": 3.413534641265869,
|
| 928 |
+
"learning_rate": 1.9393939393939395e-05,
|
| 929 |
+
"loss": 0.6599,
|
| 930 |
+
"step": 3225
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.9696969696969697,
|
| 934 |
+
"grad_norm": 5.466010570526123,
|
| 935 |
+
"learning_rate": 1.911335578002245e-05,
|
| 936 |
+
"loss": 0.6938,
|
| 937 |
+
"step": 3250
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 1.9848484848484849,
|
| 941 |
+
"grad_norm": 4.317994594573975,
|
| 942 |
+
"learning_rate": 1.88327721661055e-05,
|
| 943 |
+
"loss": 0.7309,
|
| 944 |
+
"step": 3275
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 2.0,
|
| 948 |
+
"grad_norm": 7.49802303314209,
|
| 949 |
+
"learning_rate": 1.8552188552188553e-05,
|
| 950 |
+
"loss": 0.5147,
|
| 951 |
+
"step": 3300
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 2.0,
|
| 955 |
+
"eval_accuracy": 0.8475757575757575,
|
| 956 |
+
"eval_f1_macro": 0.8117394542384239,
|
| 957 |
+
"eval_f1_micro": 0.8475757575757575,
|
| 958 |
+
"eval_f1_weighted": 0.840866356559439,
|
| 959 |
+
"eval_loss": 0.5004541873931885,
|
| 960 |
+
"eval_precision_macro": 0.8356258037544531,
|
| 961 |
+
"eval_precision_micro": 0.8475757575757575,
|
| 962 |
+
"eval_precision_weighted": 0.8505208734646063,
|
| 963 |
+
"eval_recall_macro": 0.8084166666666667,
|
| 964 |
+
"eval_recall_micro": 0.8475757575757575,
|
| 965 |
+
"eval_recall_weighted": 0.8475757575757575,
|
| 966 |
+
"eval_runtime": 18.9039,
|
| 967 |
+
"eval_samples_per_second": 174.567,
|
| 968 |
+
"eval_steps_per_second": 10.95,
|
| 969 |
+
"step": 3300
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 2.015151515151515,
|
| 973 |
+
"grad_norm": 4.281924247741699,
|
| 974 |
+
"learning_rate": 1.8271604938271607e-05,
|
| 975 |
+
"loss": 0.4413,
|
| 976 |
+
"step": 3325
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 2.0303030303030303,
|
| 980 |
+
"grad_norm": 11.078526496887207,
|
| 981 |
+
"learning_rate": 1.799102132435466e-05,
|
| 982 |
+
"loss": 0.516,
|
| 983 |
+
"step": 3350
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 2.0454545454545454,
|
| 987 |
+
"grad_norm": 12.010851860046387,
|
| 988 |
+
"learning_rate": 1.771043771043771e-05,
|
| 989 |
+
"loss": 0.6546,
|
| 990 |
+
"step": 3375
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 2.0606060606060606,
|
| 994 |
+
"grad_norm": 6.438499450683594,
|
| 995 |
+
"learning_rate": 1.7429854096520764e-05,
|
| 996 |
+
"loss": 0.6509,
|
| 997 |
+
"step": 3400
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 2.0757575757575757,
|
| 1001 |
+
"grad_norm": 2.6832661628723145,
|
| 1002 |
+
"learning_rate": 1.7149270482603815e-05,
|
| 1003 |
+
"loss": 0.6495,
|
| 1004 |
+
"step": 3425
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 2.090909090909091,
|
| 1008 |
+
"grad_norm": 6.37797737121582,
|
| 1009 |
+
"learning_rate": 1.686868686868687e-05,
|
| 1010 |
+
"loss": 0.6738,
|
| 1011 |
+
"step": 3450
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 2.106060606060606,
|
| 1015 |
+
"grad_norm": 5.069665908813477,
|
| 1016 |
+
"learning_rate": 1.6588103254769922e-05,
|
| 1017 |
+
"loss": 0.5145,
|
| 1018 |
+
"step": 3475
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 2.121212121212121,
|
| 1022 |
+
"grad_norm": 12.002437591552734,
|
| 1023 |
+
"learning_rate": 1.6307519640852976e-05,
|
| 1024 |
+
"loss": 0.7411,
|
| 1025 |
+
"step": 3500
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 2.1363636363636362,
|
| 1029 |
+
"grad_norm": 4.099394798278809,
|
| 1030 |
+
"learning_rate": 1.6026936026936026e-05,
|
| 1031 |
+
"loss": 0.6736,
|
| 1032 |
+
"step": 3525
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 2.1515151515151514,
|
| 1036 |
+
"grad_norm": 11.855905532836914,
|
| 1037 |
+
"learning_rate": 1.574635241301908e-05,
|
| 1038 |
+
"loss": 0.4738,
|
| 1039 |
+
"step": 3550
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 2.1666666666666665,
|
| 1043 |
+
"grad_norm": 9.715377807617188,
|
| 1044 |
+
"learning_rate": 1.5465768799102134e-05,
|
| 1045 |
+
"loss": 0.5321,
|
| 1046 |
+
"step": 3575
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.1818181818181817,
|
| 1050 |
+
"grad_norm": 10.646961212158203,
|
| 1051 |
+
"learning_rate": 1.5185185185185186e-05,
|
| 1052 |
+
"loss": 0.5856,
|
| 1053 |
+
"step": 3600
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.196969696969697,
|
| 1057 |
+
"grad_norm": 9.710843086242676,
|
| 1058 |
+
"learning_rate": 1.490460157126824e-05,
|
| 1059 |
+
"loss": 0.6343,
|
| 1060 |
+
"step": 3625
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.212121212121212,
|
| 1064 |
+
"grad_norm": 5.35206937789917,
|
| 1065 |
+
"learning_rate": 1.4624017957351292e-05,
|
| 1066 |
+
"loss": 0.7256,
|
| 1067 |
+
"step": 3650
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.227272727272727,
|
| 1071 |
+
"grad_norm": 6.582408428192139,
|
| 1072 |
+
"learning_rate": 1.4343434343434345e-05,
|
| 1073 |
+
"loss": 0.5843,
|
| 1074 |
+
"step": 3675
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.242424242424242,
|
| 1078 |
+
"grad_norm": 11.637269973754883,
|
| 1079 |
+
"learning_rate": 1.4062850729517396e-05,
|
| 1080 |
+
"loss": 0.5725,
|
| 1081 |
+
"step": 3700
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.257575757575758,
|
| 1085 |
+
"grad_norm": 11.256560325622559,
|
| 1086 |
+
"learning_rate": 1.378226711560045e-05,
|
| 1087 |
+
"loss": 0.5129,
|
| 1088 |
+
"step": 3725
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.2727272727272725,
|
| 1092 |
+
"grad_norm": 4.118545055389404,
|
| 1093 |
+
"learning_rate": 1.3501683501683501e-05,
|
| 1094 |
+
"loss": 0.6303,
|
| 1095 |
+
"step": 3750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.287878787878788,
|
| 1099 |
+
"grad_norm": 8.819905281066895,
|
| 1100 |
+
"learning_rate": 1.3221099887766555e-05,
|
| 1101 |
+
"loss": 0.4861,
|
| 1102 |
+
"step": 3775
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.303030303030303,
|
| 1106 |
+
"grad_norm": 11.89466667175293,
|
| 1107 |
+
"learning_rate": 1.2940516273849607e-05,
|
| 1108 |
+
"loss": 0.6803,
|
| 1109 |
+
"step": 3800
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.3181818181818183,
|
| 1113 |
+
"grad_norm": 14.192046165466309,
|
| 1114 |
+
"learning_rate": 1.2659932659932661e-05,
|
| 1115 |
+
"loss": 0.6194,
|
| 1116 |
+
"step": 3825
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 2.3333333333333335,
|
| 1120 |
+
"grad_norm": 8.376368522644043,
|
| 1121 |
+
"learning_rate": 1.2379349046015713e-05,
|
| 1122 |
+
"loss": 0.5394,
|
| 1123 |
+
"step": 3850
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 2.3484848484848486,
|
| 1127 |
+
"grad_norm": 2.375666856765747,
|
| 1128 |
+
"learning_rate": 1.2098765432098767e-05,
|
| 1129 |
+
"loss": 0.5348,
|
| 1130 |
+
"step": 3875
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.3636363636363638,
|
| 1134 |
+
"grad_norm": 6.248928070068359,
|
| 1135 |
+
"learning_rate": 1.1818181818181819e-05,
|
| 1136 |
+
"loss": 0.6926,
|
| 1137 |
+
"step": 3900
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 2.378787878787879,
|
| 1141 |
+
"grad_norm": 3.3347110748291016,
|
| 1142 |
+
"learning_rate": 1.153759820426487e-05,
|
| 1143 |
+
"loss": 0.5196,
|
| 1144 |
+
"step": 3925
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 2.393939393939394,
|
| 1148 |
+
"grad_norm": 7.418989658355713,
|
| 1149 |
+
"learning_rate": 1.1257014590347924e-05,
|
| 1150 |
+
"loss": 0.5975,
|
| 1151 |
+
"step": 3950
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.409090909090909,
|
| 1155 |
+
"grad_norm": 17.091835021972656,
|
| 1156 |
+
"learning_rate": 1.0976430976430976e-05,
|
| 1157 |
+
"loss": 0.5818,
|
| 1158 |
+
"step": 3975
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 2.4242424242424243,
|
| 1162 |
+
"grad_norm": 4.748253345489502,
|
| 1163 |
+
"learning_rate": 1.0695847362514029e-05,
|
| 1164 |
+
"loss": 0.5008,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.4393939393939394,
|
| 1169 |
+
"grad_norm": 2.7156593799591064,
|
| 1170 |
+
"learning_rate": 1.0415263748597082e-05,
|
| 1171 |
+
"loss": 0.5977,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.4545454545454546,
|
| 1176 |
+
"grad_norm": 3.704885482788086,
|
| 1177 |
+
"learning_rate": 1.0134680134680136e-05,
|
| 1178 |
+
"loss": 0.428,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.4696969696969697,
|
| 1183 |
+
"grad_norm": 6.289621829986572,
|
| 1184 |
+
"learning_rate": 9.854096520763188e-06,
|
| 1185 |
+
"loss": 0.6108,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.484848484848485,
|
| 1190 |
+
"grad_norm": 11.862101554870605,
|
| 1191 |
+
"learning_rate": 9.573512906846242e-06,
|
| 1192 |
+
"loss": 0.6182,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.5,
|
| 1197 |
+
"grad_norm": 2.4363837242126465,
|
| 1198 |
+
"learning_rate": 9.292929292929294e-06,
|
| 1199 |
+
"loss": 0.6187,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.515151515151515,
|
| 1204 |
+
"grad_norm": 4.281114101409912,
|
| 1205 |
+
"learning_rate": 9.012345679012346e-06,
|
| 1206 |
+
"loss": 0.5247,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.5303030303030303,
|
| 1211 |
+
"grad_norm": 7.008238315582275,
|
| 1212 |
+
"learning_rate": 8.7317620650954e-06,
|
| 1213 |
+
"loss": 0.4063,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.5454545454545454,
|
| 1218 |
+
"grad_norm": 10.460261344909668,
|
| 1219 |
+
"learning_rate": 8.451178451178452e-06,
|
| 1220 |
+
"loss": 0.4699,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.5606060606060606,
|
| 1225 |
+
"grad_norm": 6.779576301574707,
|
| 1226 |
+
"learning_rate": 8.170594837261504e-06,
|
| 1227 |
+
"loss": 0.632,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.5757575757575757,
|
| 1232 |
+
"grad_norm": 12.081551551818848,
|
| 1233 |
+
"learning_rate": 7.890011223344557e-06,
|
| 1234 |
+
"loss": 0.5803,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.590909090909091,
|
| 1239 |
+
"grad_norm": 14.813763618469238,
|
| 1240 |
+
"learning_rate": 7.609427609427609e-06,
|
| 1241 |
+
"loss": 0.5825,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.606060606060606,
|
| 1246 |
+
"grad_norm": 3.7761809825897217,
|
| 1247 |
+
"learning_rate": 7.328843995510662e-06,
|
| 1248 |
+
"loss": 0.4587,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.621212121212121,
|
| 1253 |
+
"grad_norm": 2.71801495552063,
|
| 1254 |
+
"learning_rate": 7.048260381593716e-06,
|
| 1255 |
+
"loss": 0.5465,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.6363636363636362,
|
| 1260 |
+
"grad_norm": 5.783569812774658,
|
| 1261 |
+
"learning_rate": 6.767676767676769e-06,
|
| 1262 |
+
"loss": 0.5003,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.6515151515151514,
|
| 1267 |
+
"grad_norm": 12.182541847229004,
|
| 1268 |
+
"learning_rate": 6.487093153759821e-06,
|
| 1269 |
+
"loss": 0.5783,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 2.6666666666666665,
|
| 1274 |
+
"grad_norm": 12.15306282043457,
|
| 1275 |
+
"learning_rate": 6.206509539842874e-06,
|
| 1276 |
+
"loss": 0.5159,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.6818181818181817,
|
| 1281 |
+
"grad_norm": 8.337636947631836,
|
| 1282 |
+
"learning_rate": 5.925925925925927e-06,
|
| 1283 |
+
"loss": 0.4119,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 2.6969696969696972,
|
| 1288 |
+
"grad_norm": 3.530006170272827,
|
| 1289 |
+
"learning_rate": 5.645342312008979e-06,
|
| 1290 |
+
"loss": 0.4734,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.712121212121212,
|
| 1295 |
+
"grad_norm": 7.290157794952393,
|
| 1296 |
+
"learning_rate": 5.364758698092032e-06,
|
| 1297 |
+
"loss": 0.5549,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.7272727272727275,
|
| 1302 |
+
"grad_norm": 6.58420467376709,
|
| 1303 |
+
"learning_rate": 5.0841750841750845e-06,
|
| 1304 |
+
"loss": 0.4441,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.742424242424242,
|
| 1309 |
+
"grad_norm": 4.345494747161865,
|
| 1310 |
+
"learning_rate": 4.803591470258137e-06,
|
| 1311 |
+
"loss": 0.48,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.757575757575758,
|
| 1316 |
+
"grad_norm": 10.148066520690918,
|
| 1317 |
+
"learning_rate": 4.523007856341189e-06,
|
| 1318 |
+
"loss": 0.4423,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.7727272727272725,
|
| 1323 |
+
"grad_norm": 6.80890417098999,
|
| 1324 |
+
"learning_rate": 4.242424242424243e-06,
|
| 1325 |
+
"loss": 0.4497,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.787878787878788,
|
| 1330 |
+
"grad_norm": 9.686090469360352,
|
| 1331 |
+
"learning_rate": 3.961840628507295e-06,
|
| 1332 |
+
"loss": 0.3885,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.8030303030303028,
|
| 1337 |
+
"grad_norm": 6.605715274810791,
|
| 1338 |
+
"learning_rate": 3.681257014590348e-06,
|
| 1339 |
+
"loss": 0.4989,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 2.8181818181818183,
|
| 1344 |
+
"grad_norm": 5.217257499694824,
|
| 1345 |
+
"learning_rate": 3.400673400673401e-06,
|
| 1346 |
+
"loss": 0.5379,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 2.8333333333333335,
|
| 1351 |
+
"grad_norm": 8.507195472717285,
|
| 1352 |
+
"learning_rate": 3.1200897867564538e-06,
|
| 1353 |
+
"loss": 0.6228,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 2.8484848484848486,
|
| 1358 |
+
"grad_norm": 5.713125228881836,
|
| 1359 |
+
"learning_rate": 2.8395061728395062e-06,
|
| 1360 |
+
"loss": 0.4659,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 2.8636363636363638,
|
| 1365 |
+
"grad_norm": 13.924797058105469,
|
| 1366 |
+
"learning_rate": 2.558922558922559e-06,
|
| 1367 |
+
"loss": 0.5267,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 2.878787878787879,
|
| 1372 |
+
"grad_norm": 9.560832023620605,
|
| 1373 |
+
"learning_rate": 2.2783389450056116e-06,
|
| 1374 |
+
"loss": 0.5839,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 2.893939393939394,
|
| 1379 |
+
"grad_norm": 4.680112361907959,
|
| 1380 |
+
"learning_rate": 1.9977553310886645e-06,
|
| 1381 |
+
"loss": 0.3282,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 2.909090909090909,
|
| 1386 |
+
"grad_norm": 3.320807695388794,
|
| 1387 |
+
"learning_rate": 1.7171717171717171e-06,
|
| 1388 |
+
"loss": 0.4741,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 2.9242424242424243,
|
| 1393 |
+
"grad_norm": 6.167539119720459,
|
| 1394 |
+
"learning_rate": 1.43658810325477e-06,
|
| 1395 |
+
"loss": 0.5005,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 2.9393939393939394,
|
| 1400 |
+
"grad_norm": 6.241283893585205,
|
| 1401 |
+
"learning_rate": 1.156004489337823e-06,
|
| 1402 |
+
"loss": 0.4634,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 2.9545454545454546,
|
| 1407 |
+
"grad_norm": 2.0378332138061523,
|
| 1408 |
+
"learning_rate": 8.754208754208755e-07,
|
| 1409 |
+
"loss": 0.4892,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 2.9696969696969697,
|
| 1414 |
+
"grad_norm": 8.896339416503906,
|
| 1415 |
+
"learning_rate": 5.948372615039282e-07,
|
| 1416 |
+
"loss": 0.5611,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 2.984848484848485,
|
| 1421 |
+
"grad_norm": 2.1297545433044434,
|
| 1422 |
+
"learning_rate": 3.1425364758698096e-07,
|
| 1423 |
+
"loss": 0.516,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.0,
|
| 1428 |
+
"grad_norm": 10.862909317016602,
|
| 1429 |
+
"learning_rate": 3.367003367003367e-08,
|
| 1430 |
+
"loss": 0.4453,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.0,
|
| 1435 |
+
"eval_accuracy": 0.886969696969697,
|
| 1436 |
+
"eval_f1_macro": 0.8644630522360383,
|
| 1437 |
+
"eval_f1_micro": 0.886969696969697,
|
| 1438 |
+
"eval_f1_weighted": 0.8837489529217776,
|
| 1439 |
+
"eval_loss": 0.3821594715118408,
|
| 1440 |
+
"eval_precision_macro": 0.8700338902181693,
|
| 1441 |
+
"eval_precision_micro": 0.886969696969697,
|
| 1442 |
+
"eval_precision_weighted": 0.8838390180385471,
|
| 1443 |
+
"eval_recall_macro": 0.8628333333333335,
|
| 1444 |
+
"eval_recall_micro": 0.886969696969697,
|
| 1445 |
+
"eval_recall_weighted": 0.886969696969697,
|
| 1446 |
+
"eval_runtime": 18.9146,
|
| 1447 |
+
"eval_samples_per_second": 174.469,
|
| 1448 |
+
"eval_steps_per_second": 10.944,
|
| 1449 |
+
"step": 4950
|
| 1450 |
+
}
|
| 1451 |
+
],
|
| 1452 |
+
"logging_steps": 25,
|
| 1453 |
+
"max_steps": 4950,
|
| 1454 |
+
"num_input_tokens_seen": 0,
|
| 1455 |
+
"num_train_epochs": 3,
|
| 1456 |
+
"save_steps": 500,
|
| 1457 |
+
"stateful_callbacks": {
|
| 1458 |
+
"EarlyStoppingCallback": {
|
| 1459 |
+
"args": {
|
| 1460 |
+
"early_stopping_patience": 5,
|
| 1461 |
+
"early_stopping_threshold": 0.01
|
| 1462 |
+
},
|
| 1463 |
+
"attributes": {
|
| 1464 |
+
"early_stopping_patience_counter": 0
|
| 1465 |
+
}
|
| 1466 |
+
},
|
| 1467 |
+
"TrainerControl": {
|
| 1468 |
+
"args": {
|
| 1469 |
+
"should_epoch_stop": false,
|
| 1470 |
+
"should_evaluate": false,
|
| 1471 |
+
"should_log": false,
|
| 1472 |
+
"should_save": true,
|
| 1473 |
+
"should_training_stop": true
|
| 1474 |
+
},
|
| 1475 |
+
"attributes": {}
|
| 1476 |
+
}
|
| 1477 |
+
},
|
| 1478 |
+
"total_flos": 1.084816011852841e+19,
|
| 1479 |
+
"train_batch_size": 8,
|
| 1480 |
+
"trial_name": null,
|
| 1481 |
+
"trial_params": null
|
| 1482 |
+
}
|
checkpoint-4950/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3399f9c0daeb86c574359321a66cf504cae7b543b5213638c5189770dc12e9c8
|
| 3 |
+
size 5304
|
config.json
ADDED
|
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-large-patch16-224",
|
| 3 |
+
"_num_labels": 12,
|
| 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": 1024,
|
| 12 |
+
"id2label": {
|
| 13 |
+
"0": "Appraisal Report",
|
| 14 |
+
"1": "Block",
|
| 15 |
+
"2": "Columns",
|
| 16 |
+
"3": "Commision Form",
|
| 17 |
+
"4": "Don't Know",
|
| 18 |
+
"5": "Free Text",
|
| 19 |
+
"6": "Lease Agreement",
|
| 20 |
+
"7": "MLS",
|
| 21 |
+
"8": "Offering Memo",
|
| 22 |
+
"9": "Profile",
|
| 23 |
+
"10": "Rent Roll",
|
| 24 |
+
"11": "Table"
|
| 25 |
+
},
|
| 26 |
+
"image_size": 224,
|
| 27 |
+
"initializer_range": 0.02,
|
| 28 |
+
"intermediate_size": 4096,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"Appraisal Report": 0,
|
| 31 |
+
"Block": 1,
|
| 32 |
+
"Columns": 2,
|
| 33 |
+
"Commision Form": 3,
|
| 34 |
+
"Don't Know": 4,
|
| 35 |
+
"Free Text": 5,
|
| 36 |
+
"Lease Agreement": 6,
|
| 37 |
+
"MLS": 7,
|
| 38 |
+
"Offering Memo": 8,
|
| 39 |
+
"Profile": 9,
|
| 40 |
+
"Rent Roll": 10,
|
| 41 |
+
"Table": 11
|
| 42 |
+
},
|
| 43 |
+
"layer_norm_eps": 1e-12,
|
| 44 |
+
"model_type": "vit",
|
| 45 |
+
"num_attention_heads": 16,
|
| 46 |
+
"num_channels": 3,
|
| 47 |
+
"num_hidden_layers": 24,
|
| 48 |
+
"patch_size": 16,
|
| 49 |
+
"problem_type": "single_label_classification",
|
| 50 |
+
"qkv_bias": true,
|
| 51 |
+
"torch_dtype": "float32",
|
| 52 |
+
"transformers_version": "4.45.0"
|
| 53 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:158a10c1203884182679a6e220f39cf8280d7ec317bd6af4ebdeb5ddd7c512ec
|
| 3 |
+
size 1213302272
|
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": 224,
|
| 20 |
+
"width": 224
|
| 21 |
+
}
|
| 22 |
+
}
|
runs/Oct28_19-20-06_r-compstak-document-ai-fine-tuning-z11p1xee-34b24-oo9j1/events.out.tfevents.1730143207.r-compstak-document-ai-fine-tuning-z11p1xee-34b24-oo9j1.1353.0
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 sha256:214a730391b7f7e14207df3e87a31be13f9d54e5b08bcbb6a19fb0f49988783c
|
| 3 |
+
size 50078
|
runs/Oct28_19-20-06_r-compstak-document-ai-fine-tuning-z11p1xee-34b24-oo9j1/events.out.tfevents.1730144209.r-compstak-document-ai-fine-tuning-z11p1xee-34b24-oo9j1.1353.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4395521c9714f7782e45e46faa7855e9d802343623bc3968867bb0bcf7014fbe
|
| 3 |
+
size 921
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3399f9c0daeb86c574359321a66cf504cae7b543b5213638c5189770dc12e9c8
|
| 3 |
+
size 5304
|
training_params.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "classify-google-augment-3/autotrain-data",
|
| 3 |
+
"model": "google/vit-large-patch16-224",
|
| 4 |
+
"username": "compstak",
|
| 5 |
+
"lr": 5e-05,
|
| 6 |
+
"epochs": 3,
|
| 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": "classify-google-augment-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 |
+
}
|