Commit
•
7222601
1
Parent(s):
7558cfe
Upload folder using huggingface_hub
Browse files- README.md +42 -0
- checkpoint-10234/config.json +63 -0
- checkpoint-10234/model.safetensors +3 -0
- checkpoint-10234/optimizer.pt +3 -0
- checkpoint-10234/rng_state.pth +3 -0
- checkpoint-10234/scheduler.pt +3 -0
- checkpoint-10234/trainer_state.json +3010 -0
- checkpoint-10234/training_args.bin +3 -0
- config.json +63 -0
- model.safetensors +3 -0
- preprocessor_config.json +36 -0
- runs/May03_11-45-08_60f4804cf903/events.out.tfevents.1714736709.60f4804cf903.13020.0 +2 -2
- runs/May03_11-45-08_60f4804cf903/events.out.tfevents.1714740351.60f4804cf903.13020.1 +3 -0
- training_args.bin +3 -0
- training_params.json +27 -0
README.md
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
---
|
| 3 |
+
tags:
|
| 4 |
+
- autotrain
|
| 5 |
+
- image-classification
|
| 6 |
+
widget:
|
| 7 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/tiger.jpg
|
| 8 |
+
example_title: Tiger
|
| 9 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/teapot.jpg
|
| 10 |
+
example_title: Teapot
|
| 11 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/palace.jpg
|
| 12 |
+
example_title: Palace
|
| 13 |
+
datasets:
|
| 14 |
+
- autotrain-vit-large-patch16-224/autotrain-data
|
| 15 |
+
---
|
| 16 |
+
|
| 17 |
+
# Model Trained Using AutoTrain
|
| 18 |
+
|
| 19 |
+
- Problem type: Image Classification
|
| 20 |
+
|
| 21 |
+
## Validation Metrics
|
| 22 |
+
loss: 0.23535355925559998
|
| 23 |
+
|
| 24 |
+
f1_macro: 0.833050479179196
|
| 25 |
+
|
| 26 |
+
f1_micro: 0.9214749061113008
|
| 27 |
+
|
| 28 |
+
f1_weighted: 0.9214057890123231
|
| 29 |
+
|
| 30 |
+
precision_macro: 0.8898753259090769
|
| 31 |
+
|
| 32 |
+
precision_micro: 0.9214749061113008
|
| 33 |
+
|
| 34 |
+
precision_weighted: 0.9262844759199463
|
| 35 |
+
|
| 36 |
+
recall_macro: 0.7987766214501736
|
| 37 |
+
|
| 38 |
+
recall_micro: 0.9214749061113008
|
| 39 |
+
|
| 40 |
+
recall_weighted: 0.9214749061113008
|
| 41 |
+
|
| 42 |
+
accuracy: 0.9214749061113008
|
checkpoint-10234/config.json
ADDED
|
@@ -0,0 +1,63 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-large-patch16-224",
|
| 3 |
+
"_num_labels": 17,
|
| 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": "AadhaarBack",
|
| 14 |
+
"1": "AadhaarBackMasked",
|
| 15 |
+
"2": "AadhaarFront",
|
| 16 |
+
"3": "AadhaarFrontMasked",
|
| 17 |
+
"4": "AadhaarRegular",
|
| 18 |
+
"5": "AadhaarRegularMasked",
|
| 19 |
+
"6": "AadhaarSelfGenerated",
|
| 20 |
+
"7": "DrivingLicenseNew",
|
| 21 |
+
"8": "DrivingLicenseOld",
|
| 22 |
+
"9": "MultipleOVDs",
|
| 23 |
+
"10": "PanCard",
|
| 24 |
+
"11": "PassportFirst",
|
| 25 |
+
"12": "PassportLast",
|
| 26 |
+
"13": "PassportRegular",
|
| 27 |
+
"14": "VoterCardBack",
|
| 28 |
+
"15": "VoterCardFront",
|
| 29 |
+
"16": "VoterCardRegular"
|
| 30 |
+
},
|
| 31 |
+
"image_size": 224,
|
| 32 |
+
"initializer_range": 0.02,
|
| 33 |
+
"intermediate_size": 4096,
|
| 34 |
+
"label2id": {
|
| 35 |
+
"AadhaarBack": 0,
|
| 36 |
+
"AadhaarBackMasked": 1,
|
| 37 |
+
"AadhaarFront": 2,
|
| 38 |
+
"AadhaarFrontMasked": 3,
|
| 39 |
+
"AadhaarRegular": 4,
|
| 40 |
+
"AadhaarRegularMasked": 5,
|
| 41 |
+
"AadhaarSelfGenerated": 6,
|
| 42 |
+
"DrivingLicenseNew": 7,
|
| 43 |
+
"DrivingLicenseOld": 8,
|
| 44 |
+
"MultipleOVDs": 9,
|
| 45 |
+
"PanCard": 10,
|
| 46 |
+
"PassportFirst": 11,
|
| 47 |
+
"PassportLast": 12,
|
| 48 |
+
"PassportRegular": 13,
|
| 49 |
+
"VoterCardBack": 14,
|
| 50 |
+
"VoterCardFront": 15,
|
| 51 |
+
"VoterCardRegular": 16
|
| 52 |
+
},
|
| 53 |
+
"layer_norm_eps": 1e-12,
|
| 54 |
+
"model_type": "vit",
|
| 55 |
+
"num_attention_heads": 16,
|
| 56 |
+
"num_channels": 3,
|
| 57 |
+
"num_hidden_layers": 24,
|
| 58 |
+
"patch_size": 16,
|
| 59 |
+
"problem_type": "single_label_classification",
|
| 60 |
+
"qkv_bias": true,
|
| 61 |
+
"torch_dtype": "float32",
|
| 62 |
+
"transformers_version": "4.40.1"
|
| 63 |
+
}
|
checkpoint-10234/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b84b79d2c3d3dffecdc42ac6506d7890bf3056236ac00e06bac2769c5ad6701e
|
| 3 |
+
size 1213322772
|
checkpoint-10234/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c20bd5f4df1d9da70f94fa82d93310df3c94af012afb082a51dc190efb5ae5d7
|
| 3 |
+
size 2426880554
|
checkpoint-10234/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6ac0dd2a1da46f73db9d63afe770e294b0981b00947eb7449cb44c38cb3632ce
|
| 3 |
+
size 14180
|
checkpoint-10234/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:788ca17bcea37ff2f0a361aa95b563072d37969bb81e423f05e49f475379ac9b
|
| 3 |
+
size 1064
|
checkpoint-10234/trainer_state.json
ADDED
|
@@ -0,0 +1,3010 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.23535355925559998,
|
| 3 |
+
"best_model_checkpoint": "autotrain-vit-large-patch16-224/checkpoint-10234",
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 10234,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01709986320109439,
|
| 13 |
+
"grad_norm": 16.03294563293457,
|
| 14 |
+
"learning_rate": 8.207934336525308e-07,
|
| 15 |
+
"loss": 3.1283,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.03419972640218878,
|
| 20 |
+
"grad_norm": 15.965583801269531,
|
| 21 |
+
"learning_rate": 1.6073871409028727e-06,
|
| 22 |
+
"loss": 2.7271,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.05129958960328317,
|
| 27 |
+
"grad_norm": 10.72908878326416,
|
| 28 |
+
"learning_rate": 2.4623803009575924e-06,
|
| 29 |
+
"loss": 2.2885,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.06839945280437756,
|
| 34 |
+
"grad_norm": 12.793851852416992,
|
| 35 |
+
"learning_rate": 3.3173734610123124e-06,
|
| 36 |
+
"loss": 1.9045,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.08549931600547196,
|
| 41 |
+
"grad_norm": 11.025067329406738,
|
| 42 |
+
"learning_rate": 4.172366621067032e-06,
|
| 43 |
+
"loss": 1.4383,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.10259917920656635,
|
| 48 |
+
"grad_norm": 13.093396186828613,
|
| 49 |
+
"learning_rate": 5.027359781121752e-06,
|
| 50 |
+
"loss": 1.3066,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.11969904240766074,
|
| 55 |
+
"grad_norm": 8.261626243591309,
|
| 56 |
+
"learning_rate": 5.882352941176471e-06,
|
| 57 |
+
"loss": 1.1223,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.13679890560875513,
|
| 62 |
+
"grad_norm": 9.678178787231445,
|
| 63 |
+
"learning_rate": 6.73734610123119e-06,
|
| 64 |
+
"loss": 1.1502,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.1538987688098495,
|
| 69 |
+
"grad_norm": 13.990431785583496,
|
| 70 |
+
"learning_rate": 7.592339261285911e-06,
|
| 71 |
+
"loss": 0.9833,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.17099863201094392,
|
| 76 |
+
"grad_norm": 11.807621002197266,
|
| 77 |
+
"learning_rate": 8.44733242134063e-06,
|
| 78 |
+
"loss": 0.8433,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.1880984952120383,
|
| 83 |
+
"grad_norm": 21.150636672973633,
|
| 84 |
+
"learning_rate": 9.302325581395349e-06,
|
| 85 |
+
"loss": 0.8553,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.2051983584131327,
|
| 90 |
+
"grad_norm": 16.56648826599121,
|
| 91 |
+
"learning_rate": 1.0157318741450068e-05,
|
| 92 |
+
"loss": 1.0502,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.22229822161422708,
|
| 97 |
+
"grad_norm": 8.26397705078125,
|
| 98 |
+
"learning_rate": 1.1012311901504789e-05,
|
| 99 |
+
"loss": 0.8686,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2393980848153215,
|
| 104 |
+
"grad_norm": 7.628605842590332,
|
| 105 |
+
"learning_rate": 1.1867305061559508e-05,
|
| 106 |
+
"loss": 0.7838,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.25649794801641584,
|
| 111 |
+
"grad_norm": 6.690649509429932,
|
| 112 |
+
"learning_rate": 1.2722298221614229e-05,
|
| 113 |
+
"loss": 0.7284,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.27359781121751026,
|
| 118 |
+
"grad_norm": 9.397786140441895,
|
| 119 |
+
"learning_rate": 1.3577291381668946e-05,
|
| 120 |
+
"loss": 0.6408,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.29069767441860467,
|
| 125 |
+
"grad_norm": 1.714457392692566,
|
| 126 |
+
"learning_rate": 1.4432284541723667e-05,
|
| 127 |
+
"loss": 0.7046,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.307797537619699,
|
| 132 |
+
"grad_norm": 16.014860153198242,
|
| 133 |
+
"learning_rate": 1.5287277701778386e-05,
|
| 134 |
+
"loss": 0.7838,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.32489740082079344,
|
| 139 |
+
"grad_norm": 10.126370429992676,
|
| 140 |
+
"learning_rate": 1.6142270861833107e-05,
|
| 141 |
+
"loss": 0.735,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.34199726402188785,
|
| 146 |
+
"grad_norm": 10.995970726013184,
|
| 147 |
+
"learning_rate": 1.6997264021887825e-05,
|
| 148 |
+
"loss": 0.7523,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.3590971272229822,
|
| 153 |
+
"grad_norm": 7.909482955932617,
|
| 154 |
+
"learning_rate": 1.7852257181942546e-05,
|
| 155 |
+
"loss": 0.7157,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.3761969904240766,
|
| 160 |
+
"grad_norm": 10.751182556152344,
|
| 161 |
+
"learning_rate": 1.8707250341997263e-05,
|
| 162 |
+
"loss": 0.7966,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.393296853625171,
|
| 167 |
+
"grad_norm": 8.872684478759766,
|
| 168 |
+
"learning_rate": 1.9562243502051984e-05,
|
| 169 |
+
"loss": 0.6923,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.4103967168262654,
|
| 174 |
+
"grad_norm": 12.542756080627441,
|
| 175 |
+
"learning_rate": 2.0417236662106705e-05,
|
| 176 |
+
"loss": 0.8022,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.4274965800273598,
|
| 181 |
+
"grad_norm": 4.314883708953857,
|
| 182 |
+
"learning_rate": 2.1272229822161423e-05,
|
| 183 |
+
"loss": 0.6959,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.44459644322845415,
|
| 188 |
+
"grad_norm": 10.603111267089844,
|
| 189 |
+
"learning_rate": 2.2127222982216144e-05,
|
| 190 |
+
"loss": 0.7282,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.46169630642954856,
|
| 195 |
+
"grad_norm": 12.411425590515137,
|
| 196 |
+
"learning_rate": 2.298221614227086e-05,
|
| 197 |
+
"loss": 0.8191,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.478796169630643,
|
| 202 |
+
"grad_norm": 6.108178615570068,
|
| 203 |
+
"learning_rate": 2.3837209302325582e-05,
|
| 204 |
+
"loss": 0.507,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.49589603283173733,
|
| 209 |
+
"grad_norm": 11.323492050170898,
|
| 210 |
+
"learning_rate": 2.46922024623803e-05,
|
| 211 |
+
"loss": 0.5333,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.5129958960328317,
|
| 216 |
+
"grad_norm": 9.001029014587402,
|
| 217 |
+
"learning_rate": 2.554719562243502e-05,
|
| 218 |
+
"loss": 0.6816,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.5300957592339262,
|
| 223 |
+
"grad_norm": 1.6873475313186646,
|
| 224 |
+
"learning_rate": 2.6402188782489738e-05,
|
| 225 |
+
"loss": 0.6621,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5471956224350205,
|
| 230 |
+
"grad_norm": 9.71667194366455,
|
| 231 |
+
"learning_rate": 2.7257181942544462e-05,
|
| 232 |
+
"loss": 0.5873,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5642954856361149,
|
| 237 |
+
"grad_norm": 5.7018866539001465,
|
| 238 |
+
"learning_rate": 2.811217510259918e-05,
|
| 239 |
+
"loss": 0.6119,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5813953488372093,
|
| 244 |
+
"grad_norm": 6.869658946990967,
|
| 245 |
+
"learning_rate": 2.8967168262653897e-05,
|
| 246 |
+
"loss": 0.8036,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5984952120383037,
|
| 251 |
+
"grad_norm": 10.419452667236328,
|
| 252 |
+
"learning_rate": 2.982216142270862e-05,
|
| 253 |
+
"loss": 0.6884,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.615595075239398,
|
| 258 |
+
"grad_norm": 6.166721343994141,
|
| 259 |
+
"learning_rate": 3.067715458276334e-05,
|
| 260 |
+
"loss": 0.652,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.6326949384404925,
|
| 265 |
+
"grad_norm": 10.317444801330566,
|
| 266 |
+
"learning_rate": 3.1532147742818057e-05,
|
| 267 |
+
"loss": 0.6249,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.6497948016415869,
|
| 272 |
+
"grad_norm": 11.344230651855469,
|
| 273 |
+
"learning_rate": 3.2387140902872784e-05,
|
| 274 |
+
"loss": 0.6797,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6668946648426812,
|
| 279 |
+
"grad_norm": 8.010811805725098,
|
| 280 |
+
"learning_rate": 3.32421340629275e-05,
|
| 281 |
+
"loss": 0.8409,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.6839945280437757,
|
| 286 |
+
"grad_norm": 8.49295425415039,
|
| 287 |
+
"learning_rate": 3.409712722298222e-05,
|
| 288 |
+
"loss": 0.5985,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.70109439124487,
|
| 293 |
+
"grad_norm": 14.201081275939941,
|
| 294 |
+
"learning_rate": 3.4952120383036933e-05,
|
| 295 |
+
"loss": 0.7377,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.7181942544459644,
|
| 300 |
+
"grad_norm": 12.270833015441895,
|
| 301 |
+
"learning_rate": 3.580711354309166e-05,
|
| 302 |
+
"loss": 0.5195,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.7352941176470589,
|
| 307 |
+
"grad_norm": 2.879152297973633,
|
| 308 |
+
"learning_rate": 3.6662106703146375e-05,
|
| 309 |
+
"loss": 0.6526,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.7523939808481532,
|
| 314 |
+
"grad_norm": 8.10263442993164,
|
| 315 |
+
"learning_rate": 3.7517099863201096e-05,
|
| 316 |
+
"loss": 0.7119,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.7694938440492476,
|
| 321 |
+
"grad_norm": 4.584366321563721,
|
| 322 |
+
"learning_rate": 3.837209302325582e-05,
|
| 323 |
+
"loss": 0.6515,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.786593707250342,
|
| 328 |
+
"grad_norm": 6.819866180419922,
|
| 329 |
+
"learning_rate": 3.922708618331054e-05,
|
| 330 |
+
"loss": 0.5304,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.8036935704514364,
|
| 335 |
+
"grad_norm": 9.715897560119629,
|
| 336 |
+
"learning_rate": 4.008207934336525e-05,
|
| 337 |
+
"loss": 0.5718,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.8207934336525308,
|
| 342 |
+
"grad_norm": 6.830766677856445,
|
| 343 |
+
"learning_rate": 4.093707250341998e-05,
|
| 344 |
+
"loss": 0.6228,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.8378932968536251,
|
| 349 |
+
"grad_norm": 8.660406112670898,
|
| 350 |
+
"learning_rate": 4.1792065663474694e-05,
|
| 351 |
+
"loss": 0.7842,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.8549931600547196,
|
| 356 |
+
"grad_norm": 8.981478691101074,
|
| 357 |
+
"learning_rate": 4.2647058823529415e-05,
|
| 358 |
+
"loss": 0.7702,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.872093023255814,
|
| 363 |
+
"grad_norm": 5.491425037384033,
|
| 364 |
+
"learning_rate": 4.3502051983584136e-05,
|
| 365 |
+
"loss": 0.6483,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.8891928864569083,
|
| 370 |
+
"grad_norm": 8.423704147338867,
|
| 371 |
+
"learning_rate": 4.435704514363886e-05,
|
| 372 |
+
"loss": 0.6889,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.9062927496580028,
|
| 377 |
+
"grad_norm": 11.879225730895996,
|
| 378 |
+
"learning_rate": 4.521203830369357e-05,
|
| 379 |
+
"loss": 0.7592,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.9233926128590971,
|
| 384 |
+
"grad_norm": 10.297099113464355,
|
| 385 |
+
"learning_rate": 4.606703146374829e-05,
|
| 386 |
+
"loss": 0.6994,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.9404924760601915,
|
| 391 |
+
"grad_norm": 34.574771881103516,
|
| 392 |
+
"learning_rate": 4.692202462380301e-05,
|
| 393 |
+
"loss": 0.6254,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.957592339261286,
|
| 398 |
+
"grad_norm": 7.829855442047119,
|
| 399 |
+
"learning_rate": 4.7777017783857733e-05,
|
| 400 |
+
"loss": 0.62,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.9746922024623803,
|
| 405 |
+
"grad_norm": 10.330737113952637,
|
| 406 |
+
"learning_rate": 4.863201094391245e-05,
|
| 407 |
+
"loss": 0.5799,
|
| 408 |
+
"step": 1425
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.9917920656634747,
|
| 412 |
+
"grad_norm": 7.56599235534668,
|
| 413 |
+
"learning_rate": 4.948700410396717e-05,
|
| 414 |
+
"loss": 0.6286,
|
| 415 |
+
"step": 1450
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 1.0,
|
| 419 |
+
"eval_accuracy": 0.8296346876066917,
|
| 420 |
+
"eval_f1_macro": 0.5237337030166633,
|
| 421 |
+
"eval_f1_micro": 0.8296346876066917,
|
| 422 |
+
"eval_f1_weighted": 0.818531665770605,
|
| 423 |
+
"eval_loss": 0.5858680605888367,
|
| 424 |
+
"eval_precision_macro": 0.6349808627313414,
|
| 425 |
+
"eval_precision_micro": 0.8296346876066917,
|
| 426 |
+
"eval_precision_weighted": 0.8428198229026794,
|
| 427 |
+
"eval_recall_macro": 0.5158734772675513,
|
| 428 |
+
"eval_recall_micro": 0.8296346876066917,
|
| 429 |
+
"eval_recall_weighted": 0.8296346876066917,
|
| 430 |
+
"eval_runtime": 29.6775,
|
| 431 |
+
"eval_samples_per_second": 98.694,
|
| 432 |
+
"eval_steps_per_second": 6.2,
|
| 433 |
+
"step": 1462
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 1.008891928864569,
|
| 437 |
+
"grad_norm": 7.980313301086426,
|
| 438 |
+
"learning_rate": 4.996200030399757e-05,
|
| 439 |
+
"loss": 0.6037,
|
| 440 |
+
"step": 1475
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.0259917920656634,
|
| 444 |
+
"grad_norm": 6.049495220184326,
|
| 445 |
+
"learning_rate": 4.9867001063991494e-05,
|
| 446 |
+
"loss": 0.6332,
|
| 447 |
+
"step": 1500
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 1.043091655266758,
|
| 451 |
+
"grad_norm": 11.885618209838867,
|
| 452 |
+
"learning_rate": 4.977200182398541e-05,
|
| 453 |
+
"loss": 0.761,
|
| 454 |
+
"step": 1525
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 1.0601915184678523,
|
| 458 |
+
"grad_norm": 14.27984619140625,
|
| 459 |
+
"learning_rate": 4.9677002583979335e-05,
|
| 460 |
+
"loss": 0.4568,
|
| 461 |
+
"step": 1550
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.0772913816689467,
|
| 465 |
+
"grad_norm": 6.687506198883057,
|
| 466 |
+
"learning_rate": 4.958200334397325e-05,
|
| 467 |
+
"loss": 0.6308,
|
| 468 |
+
"step": 1575
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.094391244870041,
|
| 472 |
+
"grad_norm": 6.140110492706299,
|
| 473 |
+
"learning_rate": 4.948700410396717e-05,
|
| 474 |
+
"loss": 0.5896,
|
| 475 |
+
"step": 1600
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 1.1114911080711354,
|
| 479 |
+
"grad_norm": 8.485834121704102,
|
| 480 |
+
"learning_rate": 4.939200486396109e-05,
|
| 481 |
+
"loss": 0.7167,
|
| 482 |
+
"step": 1625
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.1285909712722297,
|
| 486 |
+
"grad_norm": 8.836180686950684,
|
| 487 |
+
"learning_rate": 4.929700562395501e-05,
|
| 488 |
+
"loss": 0.5643,
|
| 489 |
+
"step": 1650
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.1456908344733243,
|
| 493 |
+
"grad_norm": 5.937542915344238,
|
| 494 |
+
"learning_rate": 4.9202006383948926e-05,
|
| 495 |
+
"loss": 0.5568,
|
| 496 |
+
"step": 1675
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 1.1627906976744187,
|
| 500 |
+
"grad_norm": 4.0352582931518555,
|
| 501 |
+
"learning_rate": 4.910700714394285e-05,
|
| 502 |
+
"loss": 0.7228,
|
| 503 |
+
"step": 1700
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.179890560875513,
|
| 507 |
+
"grad_norm": 6.907437801361084,
|
| 508 |
+
"learning_rate": 4.9012007903936766e-05,
|
| 509 |
+
"loss": 0.5816,
|
| 510 |
+
"step": 1725
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 1.1969904240766074,
|
| 514 |
+
"grad_norm": 6.376392841339111,
|
| 515 |
+
"learning_rate": 4.891700866393069e-05,
|
| 516 |
+
"loss": 0.5199,
|
| 517 |
+
"step": 1750
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 1.2140902872777017,
|
| 521 |
+
"grad_norm": 9.630695343017578,
|
| 522 |
+
"learning_rate": 4.882200942392461e-05,
|
| 523 |
+
"loss": 0.6383,
|
| 524 |
+
"step": 1775
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 1.231190150478796,
|
| 528 |
+
"grad_norm": 6.556370258331299,
|
| 529 |
+
"learning_rate": 4.872701018391853e-05,
|
| 530 |
+
"loss": 0.5977,
|
| 531 |
+
"step": 1800
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.2482900136798905,
|
| 535 |
+
"grad_norm": 8.746458053588867,
|
| 536 |
+
"learning_rate": 4.863201094391245e-05,
|
| 537 |
+
"loss": 0.5699,
|
| 538 |
+
"step": 1825
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.265389876880985,
|
| 542 |
+
"grad_norm": 6.0708537101745605,
|
| 543 |
+
"learning_rate": 4.853701170390637e-05,
|
| 544 |
+
"loss": 0.6479,
|
| 545 |
+
"step": 1850
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.2824897400820794,
|
| 549 |
+
"grad_norm": 7.421648979187012,
|
| 550 |
+
"learning_rate": 4.8442012463900295e-05,
|
| 551 |
+
"loss": 0.7258,
|
| 552 |
+
"step": 1875
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.2995896032831737,
|
| 556 |
+
"grad_norm": 11.180326461791992,
|
| 557 |
+
"learning_rate": 4.834701322389421e-05,
|
| 558 |
+
"loss": 0.5853,
|
| 559 |
+
"step": 1900
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 1.316689466484268,
|
| 563 |
+
"grad_norm": 12.808277130126953,
|
| 564 |
+
"learning_rate": 4.8252013983888135e-05,
|
| 565 |
+
"loss": 0.625,
|
| 566 |
+
"step": 1925
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 1.3337893296853625,
|
| 570 |
+
"grad_norm": 5.873989105224609,
|
| 571 |
+
"learning_rate": 4.815701474388205e-05,
|
| 572 |
+
"loss": 0.6044,
|
| 573 |
+
"step": 1950
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 1.350889192886457,
|
| 577 |
+
"grad_norm": 6.302676200866699,
|
| 578 |
+
"learning_rate": 4.8062015503875976e-05,
|
| 579 |
+
"loss": 0.7381,
|
| 580 |
+
"step": 1975
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 1.3679890560875512,
|
| 584 |
+
"grad_norm": 10.768912315368652,
|
| 585 |
+
"learning_rate": 4.796701626386989e-05,
|
| 586 |
+
"loss": 0.7416,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.3850889192886457,
|
| 591 |
+
"grad_norm": 3.4966821670532227,
|
| 592 |
+
"learning_rate": 4.7872017023863817e-05,
|
| 593 |
+
"loss": 0.5816,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.40218878248974,
|
| 598 |
+
"grad_norm": 20.856693267822266,
|
| 599 |
+
"learning_rate": 4.7777017783857733e-05,
|
| 600 |
+
"loss": 0.674,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 1.4192886456908345,
|
| 605 |
+
"grad_norm": 3.353882074356079,
|
| 606 |
+
"learning_rate": 4.768201854385165e-05,
|
| 607 |
+
"loss": 0.6317,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.4363885088919288,
|
| 612 |
+
"grad_norm": 4.5955939292907715,
|
| 613 |
+
"learning_rate": 4.758701930384557e-05,
|
| 614 |
+
"loss": 0.5058,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.4534883720930232,
|
| 619 |
+
"grad_norm": 3.8296780586242676,
|
| 620 |
+
"learning_rate": 4.749202006383949e-05,
|
| 621 |
+
"loss": 0.55,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4705882352941178,
|
| 626 |
+
"grad_norm": 8.511550903320312,
|
| 627 |
+
"learning_rate": 4.739702082383341e-05,
|
| 628 |
+
"loss": 0.7234,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.487688098495212,
|
| 633 |
+
"grad_norm": 7.452313423156738,
|
| 634 |
+
"learning_rate": 4.730202158382733e-05,
|
| 635 |
+
"loss": 0.4657,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.5047879616963065,
|
| 640 |
+
"grad_norm": 5.583883762359619,
|
| 641 |
+
"learning_rate": 4.720702234382125e-05,
|
| 642 |
+
"loss": 0.5554,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.5218878248974008,
|
| 647 |
+
"grad_norm": 3.6462392807006836,
|
| 648 |
+
"learning_rate": 4.711202310381517e-05,
|
| 649 |
+
"loss": 0.6182,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.5389876880984952,
|
| 654 |
+
"grad_norm": 5.7832207679748535,
|
| 655 |
+
"learning_rate": 4.701702386380909e-05,
|
| 656 |
+
"loss": 0.4694,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.5560875512995898,
|
| 661 |
+
"grad_norm": 3.335848331451416,
|
| 662 |
+
"learning_rate": 4.692202462380301e-05,
|
| 663 |
+
"loss": 0.6423,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.573187414500684,
|
| 668 |
+
"grad_norm": 15.582513809204102,
|
| 669 |
+
"learning_rate": 4.682702538379693e-05,
|
| 670 |
+
"loss": 0.6912,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.5902872777017785,
|
| 675 |
+
"grad_norm": 8.966438293457031,
|
| 676 |
+
"learning_rate": 4.673202614379085e-05,
|
| 677 |
+
"loss": 0.4998,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.6073871409028728,
|
| 682 |
+
"grad_norm": 6.383381366729736,
|
| 683 |
+
"learning_rate": 4.663702690378477e-05,
|
| 684 |
+
"loss": 0.5879,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.6244870041039672,
|
| 689 |
+
"grad_norm": 3.314211845397949,
|
| 690 |
+
"learning_rate": 4.6542027663778694e-05,
|
| 691 |
+
"loss": 0.5865,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.6415868673050615,
|
| 696 |
+
"grad_norm": 7.680863380432129,
|
| 697 |
+
"learning_rate": 4.644702842377261e-05,
|
| 698 |
+
"loss": 0.5731,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.658686730506156,
|
| 703 |
+
"grad_norm": 15.084454536437988,
|
| 704 |
+
"learning_rate": 4.6352029183766534e-05,
|
| 705 |
+
"loss": 0.6018,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.6757865937072505,
|
| 710 |
+
"grad_norm": 4.016242027282715,
|
| 711 |
+
"learning_rate": 4.625702994376046e-05,
|
| 712 |
+
"loss": 0.4291,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.6928864569083446,
|
| 717 |
+
"grad_norm": 6.622522354125977,
|
| 718 |
+
"learning_rate": 4.6162030703754375e-05,
|
| 719 |
+
"loss": 0.5729,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.7099863201094392,
|
| 724 |
+
"grad_norm": 7.149597644805908,
|
| 725 |
+
"learning_rate": 4.606703146374829e-05,
|
| 726 |
+
"loss": 0.6435,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.7270861833105335,
|
| 731 |
+
"grad_norm": 4.473049163818359,
|
| 732 |
+
"learning_rate": 4.597203222374221e-05,
|
| 733 |
+
"loss": 0.5068,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.744186046511628,
|
| 738 |
+
"grad_norm": 8.462128639221191,
|
| 739 |
+
"learning_rate": 4.587703298373613e-05,
|
| 740 |
+
"loss": 0.4735,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.7612859097127223,
|
| 745 |
+
"grad_norm": 3.4091100692749023,
|
| 746 |
+
"learning_rate": 4.578203374373005e-05,
|
| 747 |
+
"loss": 0.591,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.7783857729138166,
|
| 752 |
+
"grad_norm": 4.581203937530518,
|
| 753 |
+
"learning_rate": 4.568703450372397e-05,
|
| 754 |
+
"loss": 0.4642,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.7954856361149112,
|
| 759 |
+
"grad_norm": 7.462313175201416,
|
| 760 |
+
"learning_rate": 4.559203526371789e-05,
|
| 761 |
+
"loss": 0.5036,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.8125854993160053,
|
| 766 |
+
"grad_norm": 7.391847610473633,
|
| 767 |
+
"learning_rate": 4.549703602371181e-05,
|
| 768 |
+
"loss": 0.5039,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.8296853625171,
|
| 773 |
+
"grad_norm": 16.374174118041992,
|
| 774 |
+
"learning_rate": 4.540203678370573e-05,
|
| 775 |
+
"loss": 0.6136,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.8467852257181943,
|
| 780 |
+
"grad_norm": 7.047588348388672,
|
| 781 |
+
"learning_rate": 4.5307037543699654e-05,
|
| 782 |
+
"loss": 0.4051,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.8638850889192886,
|
| 787 |
+
"grad_norm": 4.181390285491943,
|
| 788 |
+
"learning_rate": 4.521203830369357e-05,
|
| 789 |
+
"loss": 0.7065,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.8809849521203832,
|
| 794 |
+
"grad_norm": 2.601778984069824,
|
| 795 |
+
"learning_rate": 4.5117039063687494e-05,
|
| 796 |
+
"loss": 0.5517,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.8980848153214773,
|
| 801 |
+
"grad_norm": 2.9935567378997803,
|
| 802 |
+
"learning_rate": 4.502203982368141e-05,
|
| 803 |
+
"loss": 0.3861,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.915184678522572,
|
| 808 |
+
"grad_norm": 7.431847095489502,
|
| 809 |
+
"learning_rate": 4.4927040583675335e-05,
|
| 810 |
+
"loss": 0.5537,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.9322845417236663,
|
| 815 |
+
"grad_norm": 7.30590295791626,
|
| 816 |
+
"learning_rate": 4.483204134366925e-05,
|
| 817 |
+
"loss": 0.5082,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.9493844049247606,
|
| 822 |
+
"grad_norm": 6.986799240112305,
|
| 823 |
+
"learning_rate": 4.4737042103663176e-05,
|
| 824 |
+
"loss": 0.5727,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.966484268125855,
|
| 829 |
+
"grad_norm": 4.823264122009277,
|
| 830 |
+
"learning_rate": 4.464204286365709e-05,
|
| 831 |
+
"loss": 0.5773,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.9835841313269493,
|
| 836 |
+
"grad_norm": 0.19885371625423431,
|
| 837 |
+
"learning_rate": 4.4547043623651016e-05,
|
| 838 |
+
"loss": 0.5056,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 2.0,
|
| 843 |
+
"eval_accuracy": 0.9095254353021509,
|
| 844 |
+
"eval_f1_macro": 0.7148556374573719,
|
| 845 |
+
"eval_f1_micro": 0.9095254353021509,
|
| 846 |
+
"eval_f1_weighted": 0.906235846022486,
|
| 847 |
+
"eval_loss": 0.2884175181388855,
|
| 848 |
+
"eval_precision_macro": 0.779691628927798,
|
| 849 |
+
"eval_precision_micro": 0.9095254353021509,
|
| 850 |
+
"eval_precision_weighted": 0.9118121732094211,
|
| 851 |
+
"eval_recall_macro": 0.6904984139815875,
|
| 852 |
+
"eval_recall_micro": 0.9095254353021509,
|
| 853 |
+
"eval_recall_weighted": 0.9095254353021509,
|
| 854 |
+
"eval_runtime": 29.8929,
|
| 855 |
+
"eval_samples_per_second": 97.983,
|
| 856 |
+
"eval_steps_per_second": 6.155,
|
| 857 |
+
"step": 2924
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 2.000683994528044,
|
| 861 |
+
"grad_norm": 8.045063972473145,
|
| 862 |
+
"learning_rate": 4.445204438364493e-05,
|
| 863 |
+
"loss": 0.5348,
|
| 864 |
+
"step": 2925
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 2.017783857729138,
|
| 868 |
+
"grad_norm": 7.1167521476745605,
|
| 869 |
+
"learning_rate": 4.435704514363886e-05,
|
| 870 |
+
"loss": 0.4205,
|
| 871 |
+
"step": 2950
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 2.0348837209302326,
|
| 875 |
+
"grad_norm": 7.508376121520996,
|
| 876 |
+
"learning_rate": 4.4262045903632774e-05,
|
| 877 |
+
"loss": 0.5367,
|
| 878 |
+
"step": 2975
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 2.0519835841313268,
|
| 882 |
+
"grad_norm": 10.526673316955566,
|
| 883 |
+
"learning_rate": 4.416704666362669e-05,
|
| 884 |
+
"loss": 0.4326,
|
| 885 |
+
"step": 3000
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 2.0690834473324213,
|
| 889 |
+
"grad_norm": 7.46666145324707,
|
| 890 |
+
"learning_rate": 4.407204742362061e-05,
|
| 891 |
+
"loss": 0.6013,
|
| 892 |
+
"step": 3025
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 2.086183310533516,
|
| 896 |
+
"grad_norm": 3.7393107414245605,
|
| 897 |
+
"learning_rate": 4.397704818361453e-05,
|
| 898 |
+
"loss": 0.5544,
|
| 899 |
+
"step": 3050
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 2.10328317373461,
|
| 903 |
+
"grad_norm": 6.285150051116943,
|
| 904 |
+
"learning_rate": 4.3882048943608455e-05,
|
| 905 |
+
"loss": 0.5717,
|
| 906 |
+
"step": 3075
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 2.1203830369357046,
|
| 910 |
+
"grad_norm": 8.156254768371582,
|
| 911 |
+
"learning_rate": 4.378704970360237e-05,
|
| 912 |
+
"loss": 0.5008,
|
| 913 |
+
"step": 3100
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 2.1374829001367988,
|
| 917 |
+
"grad_norm": 10.280779838562012,
|
| 918 |
+
"learning_rate": 4.3692050463596295e-05,
|
| 919 |
+
"loss": 0.3871,
|
| 920 |
+
"step": 3125
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.1545827633378933,
|
| 924 |
+
"grad_norm": 11.541882514953613,
|
| 925 |
+
"learning_rate": 4.359705122359021e-05,
|
| 926 |
+
"loss": 0.4684,
|
| 927 |
+
"step": 3150
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.1716826265389875,
|
| 931 |
+
"grad_norm": 12.31938648223877,
|
| 932 |
+
"learning_rate": 4.3502051983584136e-05,
|
| 933 |
+
"loss": 0.5055,
|
| 934 |
+
"step": 3175
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.188782489740082,
|
| 938 |
+
"grad_norm": 6.534464359283447,
|
| 939 |
+
"learning_rate": 4.340705274357805e-05,
|
| 940 |
+
"loss": 0.682,
|
| 941 |
+
"step": 3200
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.2058823529411766,
|
| 945 |
+
"grad_norm": 6.284212112426758,
|
| 946 |
+
"learning_rate": 4.3312053503571976e-05,
|
| 947 |
+
"loss": 0.5189,
|
| 948 |
+
"step": 3225
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.2229822161422708,
|
| 952 |
+
"grad_norm": 4.796578884124756,
|
| 953 |
+
"learning_rate": 4.321705426356589e-05,
|
| 954 |
+
"loss": 0.5516,
|
| 955 |
+
"step": 3250
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.2400820793433653,
|
| 959 |
+
"grad_norm": 0.08828147500753403,
|
| 960 |
+
"learning_rate": 4.312205502355982e-05,
|
| 961 |
+
"loss": 0.4325,
|
| 962 |
+
"step": 3275
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.2571819425444595,
|
| 966 |
+
"grad_norm": 3.709010362625122,
|
| 967 |
+
"learning_rate": 4.3027055783553734e-05,
|
| 968 |
+
"loss": 0.524,
|
| 969 |
+
"step": 3300
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 2.274281805745554,
|
| 973 |
+
"grad_norm": 6.48209810256958,
|
| 974 |
+
"learning_rate": 4.293205654354766e-05,
|
| 975 |
+
"loss": 0.4304,
|
| 976 |
+
"step": 3325
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 2.2913816689466486,
|
| 980 |
+
"grad_norm": 7.461411952972412,
|
| 981 |
+
"learning_rate": 4.2837057303541574e-05,
|
| 982 |
+
"loss": 0.5894,
|
| 983 |
+
"step": 3350
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 2.3084815321477428,
|
| 987 |
+
"grad_norm": 6.687474727630615,
|
| 988 |
+
"learning_rate": 4.27420580635355e-05,
|
| 989 |
+
"loss": 0.4504,
|
| 990 |
+
"step": 3375
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 2.3255813953488373,
|
| 994 |
+
"grad_norm": 4.4408745765686035,
|
| 995 |
+
"learning_rate": 4.2647058823529415e-05,
|
| 996 |
+
"loss": 0.5091,
|
| 997 |
+
"step": 3400
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 2.3426812585499315,
|
| 1001 |
+
"grad_norm": 7.8160552978515625,
|
| 1002 |
+
"learning_rate": 4.255205958352333e-05,
|
| 1003 |
+
"loss": 0.3612,
|
| 1004 |
+
"step": 3425
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 2.359781121751026,
|
| 1008 |
+
"grad_norm": 4.452754497528076,
|
| 1009 |
+
"learning_rate": 4.2457060343517255e-05,
|
| 1010 |
+
"loss": 0.6184,
|
| 1011 |
+
"step": 3450
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 2.37688098495212,
|
| 1015 |
+
"grad_norm": 12.292354583740234,
|
| 1016 |
+
"learning_rate": 4.236206110351117e-05,
|
| 1017 |
+
"loss": 0.5868,
|
| 1018 |
+
"step": 3475
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 2.3939808481532148,
|
| 1022 |
+
"grad_norm": 3.688055992126465,
|
| 1023 |
+
"learning_rate": 4.226706186350509e-05,
|
| 1024 |
+
"loss": 0.5325,
|
| 1025 |
+
"step": 3500
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 2.4110807113543093,
|
| 1029 |
+
"grad_norm": 5.77033805847168,
|
| 1030 |
+
"learning_rate": 4.217206262349901e-05,
|
| 1031 |
+
"loss": 0.3577,
|
| 1032 |
+
"step": 3525
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 2.4281805745554035,
|
| 1036 |
+
"grad_norm": 4.816909313201904,
|
| 1037 |
+
"learning_rate": 4.207706338349293e-05,
|
| 1038 |
+
"loss": 0.4681,
|
| 1039 |
+
"step": 3550
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 2.445280437756498,
|
| 1043 |
+
"grad_norm": 8.695076942443848,
|
| 1044 |
+
"learning_rate": 4.198206414348685e-05,
|
| 1045 |
+
"loss": 0.4151,
|
| 1046 |
+
"step": 3575
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.462380300957592,
|
| 1050 |
+
"grad_norm": 10.278604507446289,
|
| 1051 |
+
"learning_rate": 4.188706490348077e-05,
|
| 1052 |
+
"loss": 0.4165,
|
| 1053 |
+
"step": 3600
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.4794801641586868,
|
| 1057 |
+
"grad_norm": 17.01686668395996,
|
| 1058 |
+
"learning_rate": 4.1792065663474694e-05,
|
| 1059 |
+
"loss": 0.5273,
|
| 1060 |
+
"step": 3625
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.496580027359781,
|
| 1064 |
+
"grad_norm": 7.072406768798828,
|
| 1065 |
+
"learning_rate": 4.170086639306886e-05,
|
| 1066 |
+
"loss": 0.4383,
|
| 1067 |
+
"step": 3650
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.5136798905608755,
|
| 1071 |
+
"grad_norm": 5.283012866973877,
|
| 1072 |
+
"learning_rate": 4.1605867153062774e-05,
|
| 1073 |
+
"loss": 0.578,
|
| 1074 |
+
"step": 3675
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.53077975376197,
|
| 1078 |
+
"grad_norm": 3.6656084060668945,
|
| 1079 |
+
"learning_rate": 4.15108679130567e-05,
|
| 1080 |
+
"loss": 0.491,
|
| 1081 |
+
"step": 3700
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.547879616963064,
|
| 1085 |
+
"grad_norm": 9.410102844238281,
|
| 1086 |
+
"learning_rate": 4.1415868673050615e-05,
|
| 1087 |
+
"loss": 0.4382,
|
| 1088 |
+
"step": 3725
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.5649794801641588,
|
| 1092 |
+
"grad_norm": 8.41618824005127,
|
| 1093 |
+
"learning_rate": 4.132086943304454e-05,
|
| 1094 |
+
"loss": 0.5658,
|
| 1095 |
+
"step": 3750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.582079343365253,
|
| 1099 |
+
"grad_norm": 1.3863807916641235,
|
| 1100 |
+
"learning_rate": 4.1225870193038455e-05,
|
| 1101 |
+
"loss": 0.5228,
|
| 1102 |
+
"step": 3775
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.5991792065663475,
|
| 1106 |
+
"grad_norm": 3.929469585418701,
|
| 1107 |
+
"learning_rate": 4.113087095303238e-05,
|
| 1108 |
+
"loss": 0.4993,
|
| 1109 |
+
"step": 3800
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.616279069767442,
|
| 1113 |
+
"grad_norm": 4.706065654754639,
|
| 1114 |
+
"learning_rate": 4.1035871713026296e-05,
|
| 1115 |
+
"loss": 0.4045,
|
| 1116 |
+
"step": 3825
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 2.633378932968536,
|
| 1120 |
+
"grad_norm": 2.805471658706665,
|
| 1121 |
+
"learning_rate": 4.094087247302022e-05,
|
| 1122 |
+
"loss": 0.4283,
|
| 1123 |
+
"step": 3850
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 2.650478796169631,
|
| 1127 |
+
"grad_norm": 8.297080993652344,
|
| 1128 |
+
"learning_rate": 4.0845873233014136e-05,
|
| 1129 |
+
"loss": 0.4499,
|
| 1130 |
+
"step": 3875
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.667578659370725,
|
| 1134 |
+
"grad_norm": 2.637622594833374,
|
| 1135 |
+
"learning_rate": 4.075087399300806e-05,
|
| 1136 |
+
"loss": 0.5087,
|
| 1137 |
+
"step": 3900
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 2.6846785225718195,
|
| 1141 |
+
"grad_norm": 2.461876392364502,
|
| 1142 |
+
"learning_rate": 4.065587475300198e-05,
|
| 1143 |
+
"loss": 0.5104,
|
| 1144 |
+
"step": 3925
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 2.701778385772914,
|
| 1148 |
+
"grad_norm": 6.3255486488342285,
|
| 1149 |
+
"learning_rate": 4.05608755129959e-05,
|
| 1150 |
+
"loss": 0.5861,
|
| 1151 |
+
"step": 3950
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.718878248974008,
|
| 1155 |
+
"grad_norm": 11.981715202331543,
|
| 1156 |
+
"learning_rate": 4.046587627298982e-05,
|
| 1157 |
+
"loss": 0.5593,
|
| 1158 |
+
"step": 3975
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 2.7359781121751023,
|
| 1162 |
+
"grad_norm": 7.206056594848633,
|
| 1163 |
+
"learning_rate": 4.037087703298374e-05,
|
| 1164 |
+
"loss": 0.4665,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.753077975376197,
|
| 1169 |
+
"grad_norm": 1.9749879837036133,
|
| 1170 |
+
"learning_rate": 4.027587779297766e-05,
|
| 1171 |
+
"loss": 0.4984,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.7701778385772915,
|
| 1176 |
+
"grad_norm": 4.363369464874268,
|
| 1177 |
+
"learning_rate": 4.018087855297158e-05,
|
| 1178 |
+
"loss": 0.5785,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.7872777017783856,
|
| 1183 |
+
"grad_norm": 6.443023681640625,
|
| 1184 |
+
"learning_rate": 4.00858793129655e-05,
|
| 1185 |
+
"loss": 0.5287,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.80437756497948,
|
| 1190 |
+
"grad_norm": 6.106603622436523,
|
| 1191 |
+
"learning_rate": 3.999088007295942e-05,
|
| 1192 |
+
"loss": 0.6016,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.8214774281805743,
|
| 1197 |
+
"grad_norm": 9.416887283325195,
|
| 1198 |
+
"learning_rate": 3.989588083295334e-05,
|
| 1199 |
+
"loss": 0.5532,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.838577291381669,
|
| 1204 |
+
"grad_norm": 2.5579991340637207,
|
| 1205 |
+
"learning_rate": 3.9800881592947256e-05,
|
| 1206 |
+
"loss": 0.4579,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.8556771545827635,
|
| 1211 |
+
"grad_norm": 10.23747444152832,
|
| 1212 |
+
"learning_rate": 3.970588235294117e-05,
|
| 1213 |
+
"loss": 0.5402,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.8727770177838576,
|
| 1218 |
+
"grad_norm": 9.435909271240234,
|
| 1219 |
+
"learning_rate": 3.96108831129351e-05,
|
| 1220 |
+
"loss": 0.4845,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.889876880984952,
|
| 1225 |
+
"grad_norm": 3.5227017402648926,
|
| 1226 |
+
"learning_rate": 3.9515883872929014e-05,
|
| 1227 |
+
"loss": 0.3969,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.9069767441860463,
|
| 1232 |
+
"grad_norm": 2.499969482421875,
|
| 1233 |
+
"learning_rate": 3.942088463292294e-05,
|
| 1234 |
+
"loss": 0.4392,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.924076607387141,
|
| 1239 |
+
"grad_norm": 8.345962524414062,
|
| 1240 |
+
"learning_rate": 3.9325885392916854e-05,
|
| 1241 |
+
"loss": 0.4584,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.9411764705882355,
|
| 1246 |
+
"grad_norm": 3.583355188369751,
|
| 1247 |
+
"learning_rate": 3.923088615291078e-05,
|
| 1248 |
+
"loss": 0.3816,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.9582763337893296,
|
| 1253 |
+
"grad_norm": 1.7532494068145752,
|
| 1254 |
+
"learning_rate": 3.91358869129047e-05,
|
| 1255 |
+
"loss": 0.3497,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.975376196990424,
|
| 1260 |
+
"grad_norm": 9.220611572265625,
|
| 1261 |
+
"learning_rate": 3.904088767289862e-05,
|
| 1262 |
+
"loss": 0.5747,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.9924760601915183,
|
| 1267 |
+
"grad_norm": 1.715482234954834,
|
| 1268 |
+
"learning_rate": 3.894588843289254e-05,
|
| 1269 |
+
"loss": 0.4701,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 3.0,
|
| 1274 |
+
"eval_accuracy": 0.9071355411403209,
|
| 1275 |
+
"eval_f1_macro": 0.7615092412094518,
|
| 1276 |
+
"eval_f1_micro": 0.9071355411403209,
|
| 1277 |
+
"eval_f1_weighted": 0.9053250303728207,
|
| 1278 |
+
"eval_loss": 0.2766139805316925,
|
| 1279 |
+
"eval_precision_macro": 0.8223788212097681,
|
| 1280 |
+
"eval_precision_micro": 0.9071355411403209,
|
| 1281 |
+
"eval_precision_weighted": 0.9095949572268466,
|
| 1282 |
+
"eval_recall_macro": 0.7322063659665187,
|
| 1283 |
+
"eval_recall_micro": 0.9071355411403209,
|
| 1284 |
+
"eval_recall_weighted": 0.9071355411403209,
|
| 1285 |
+
"eval_runtime": 29.8386,
|
| 1286 |
+
"eval_samples_per_second": 98.161,
|
| 1287 |
+
"eval_steps_per_second": 6.167,
|
| 1288 |
+
"step": 4386
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 3.009575923392613,
|
| 1292 |
+
"grad_norm": 4.551536560058594,
|
| 1293 |
+
"learning_rate": 3.885088919288646e-05,
|
| 1294 |
+
"loss": 0.4776,
|
| 1295 |
+
"step": 4400
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 3.026675786593707,
|
| 1299 |
+
"grad_norm": 10.111888885498047,
|
| 1300 |
+
"learning_rate": 3.875588995288038e-05,
|
| 1301 |
+
"loss": 0.5966,
|
| 1302 |
+
"step": 4425
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 3.0437756497948016,
|
| 1306 |
+
"grad_norm": 13.998083114624023,
|
| 1307 |
+
"learning_rate": 3.86608907128743e-05,
|
| 1308 |
+
"loss": 0.4699,
|
| 1309 |
+
"step": 4450
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 3.060875512995896,
|
| 1313 |
+
"grad_norm": 10.541669845581055,
|
| 1314 |
+
"learning_rate": 3.856589147286822e-05,
|
| 1315 |
+
"loss": 0.5154,
|
| 1316 |
+
"step": 4475
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 3.0779753761969904,
|
| 1320 |
+
"grad_norm": 3.1978766918182373,
|
| 1321 |
+
"learning_rate": 3.847089223286214e-05,
|
| 1322 |
+
"loss": 0.4304,
|
| 1323 |
+
"step": 4500
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 3.095075239398085,
|
| 1327 |
+
"grad_norm": 5.206043720245361,
|
| 1328 |
+
"learning_rate": 3.8375892992856064e-05,
|
| 1329 |
+
"loss": 0.3223,
|
| 1330 |
+
"step": 4525
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 3.112175102599179,
|
| 1334 |
+
"grad_norm": 9.216475486755371,
|
| 1335 |
+
"learning_rate": 3.828089375284998e-05,
|
| 1336 |
+
"loss": 0.6146,
|
| 1337 |
+
"step": 4550
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 3.1292749658002736,
|
| 1341 |
+
"grad_norm": 4.771245002746582,
|
| 1342 |
+
"learning_rate": 3.81858945128439e-05,
|
| 1343 |
+
"loss": 0.3407,
|
| 1344 |
+
"step": 4575
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 3.146374829001368,
|
| 1348 |
+
"grad_norm": 8.766341209411621,
|
| 1349 |
+
"learning_rate": 3.8090895272837814e-05,
|
| 1350 |
+
"loss": 0.4866,
|
| 1351 |
+
"step": 4600
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 3.1634746922024624,
|
| 1355 |
+
"grad_norm": 6.949918270111084,
|
| 1356 |
+
"learning_rate": 3.799589603283174e-05,
|
| 1357 |
+
"loss": 0.4899,
|
| 1358 |
+
"step": 4625
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 3.180574555403557,
|
| 1362 |
+
"grad_norm": 1.0002543926239014,
|
| 1363 |
+
"learning_rate": 3.7900896792825655e-05,
|
| 1364 |
+
"loss": 0.481,
|
| 1365 |
+
"step": 4650
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 3.197674418604651,
|
| 1369 |
+
"grad_norm": 9.768813133239746,
|
| 1370 |
+
"learning_rate": 3.780589755281958e-05,
|
| 1371 |
+
"loss": 0.4338,
|
| 1372 |
+
"step": 4675
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 3.2147742818057456,
|
| 1376 |
+
"grad_norm": 4.305062770843506,
|
| 1377 |
+
"learning_rate": 3.7710898312813495e-05,
|
| 1378 |
+
"loss": 0.3783,
|
| 1379 |
+
"step": 4700
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 3.23187414500684,
|
| 1383 |
+
"grad_norm": 0.7493396997451782,
|
| 1384 |
+
"learning_rate": 3.761589907280742e-05,
|
| 1385 |
+
"loss": 0.5001,
|
| 1386 |
+
"step": 4725
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 3.2489740082079344,
|
| 1390 |
+
"grad_norm": 5.481766700744629,
|
| 1391 |
+
"learning_rate": 3.7520899832801336e-05,
|
| 1392 |
+
"loss": 0.4894,
|
| 1393 |
+
"step": 4750
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 3.266073871409029,
|
| 1397 |
+
"grad_norm": 3.8942747116088867,
|
| 1398 |
+
"learning_rate": 3.742590059279526e-05,
|
| 1399 |
+
"loss": 0.4761,
|
| 1400 |
+
"step": 4775
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 3.283173734610123,
|
| 1404 |
+
"grad_norm": 8.91220760345459,
|
| 1405 |
+
"learning_rate": 3.7330901352789176e-05,
|
| 1406 |
+
"loss": 0.5526,
|
| 1407 |
+
"step": 4800
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 3.3002735978112177,
|
| 1411 |
+
"grad_norm": 1.765817642211914,
|
| 1412 |
+
"learning_rate": 3.72359021127831e-05,
|
| 1413 |
+
"loss": 0.5665,
|
| 1414 |
+
"step": 4825
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 3.317373461012312,
|
| 1418 |
+
"grad_norm": 6.553985118865967,
|
| 1419 |
+
"learning_rate": 3.714090287277702e-05,
|
| 1420 |
+
"loss": 0.287,
|
| 1421 |
+
"step": 4850
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 3.3344733242134064,
|
| 1425 |
+
"grad_norm": 5.839998245239258,
|
| 1426 |
+
"learning_rate": 3.704590363277094e-05,
|
| 1427 |
+
"loss": 0.4927,
|
| 1428 |
+
"step": 4875
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 3.3515731874145005,
|
| 1432 |
+
"grad_norm": 8.003843307495117,
|
| 1433 |
+
"learning_rate": 3.6950904392764864e-05,
|
| 1434 |
+
"loss": 0.552,
|
| 1435 |
+
"step": 4900
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 3.368673050615595,
|
| 1439 |
+
"grad_norm": 3.3592958450317383,
|
| 1440 |
+
"learning_rate": 3.685590515275878e-05,
|
| 1441 |
+
"loss": 0.5703,
|
| 1442 |
+
"step": 4925
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 3.3857729138166897,
|
| 1446 |
+
"grad_norm": 4.747976779937744,
|
| 1447 |
+
"learning_rate": 3.6760905912752705e-05,
|
| 1448 |
+
"loss": 0.5218,
|
| 1449 |
+
"step": 4950
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 3.402872777017784,
|
| 1453 |
+
"grad_norm": 4.351706504821777,
|
| 1454 |
+
"learning_rate": 3.666590667274662e-05,
|
| 1455 |
+
"loss": 0.4671,
|
| 1456 |
+
"step": 4975
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 3.4199726402188784,
|
| 1460 |
+
"grad_norm": 5.302130222320557,
|
| 1461 |
+
"learning_rate": 3.6570907432740545e-05,
|
| 1462 |
+
"loss": 0.3913,
|
| 1463 |
+
"step": 5000
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 3.4370725034199725,
|
| 1467 |
+
"grad_norm": 1.5275038480758667,
|
| 1468 |
+
"learning_rate": 3.647590819273446e-05,
|
| 1469 |
+
"loss": 0.3149,
|
| 1470 |
+
"step": 5025
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 3.454172366621067,
|
| 1474 |
+
"grad_norm": 0.8008119463920593,
|
| 1475 |
+
"learning_rate": 3.638090895272838e-05,
|
| 1476 |
+
"loss": 0.5232,
|
| 1477 |
+
"step": 5050
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 3.471272229822161,
|
| 1481 |
+
"grad_norm": 6.361542701721191,
|
| 1482 |
+
"learning_rate": 3.6285909712722296e-05,
|
| 1483 |
+
"loss": 0.4149,
|
| 1484 |
+
"step": 5075
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 3.488372093023256,
|
| 1488 |
+
"grad_norm": 0.8227640390396118,
|
| 1489 |
+
"learning_rate": 3.619091047271622e-05,
|
| 1490 |
+
"loss": 0.5698,
|
| 1491 |
+
"step": 5100
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 3.5054719562243504,
|
| 1495 |
+
"grad_norm": 4.734671115875244,
|
| 1496 |
+
"learning_rate": 3.609591123271014e-05,
|
| 1497 |
+
"loss": 0.4835,
|
| 1498 |
+
"step": 5125
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 3.5225718194254445,
|
| 1502 |
+
"grad_norm": 6.9522576332092285,
|
| 1503 |
+
"learning_rate": 3.600091199270406e-05,
|
| 1504 |
+
"loss": 0.5387,
|
| 1505 |
+
"step": 5150
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 3.539671682626539,
|
| 1509 |
+
"grad_norm": 3.4246246814727783,
|
| 1510 |
+
"learning_rate": 3.590591275269798e-05,
|
| 1511 |
+
"loss": 0.4044,
|
| 1512 |
+
"step": 5175
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 3.556771545827633,
|
| 1516 |
+
"grad_norm": 6.183599472045898,
|
| 1517 |
+
"learning_rate": 3.58109135126919e-05,
|
| 1518 |
+
"loss": 0.4785,
|
| 1519 |
+
"step": 5200
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 3.573871409028728,
|
| 1523 |
+
"grad_norm": 3.738954782485962,
|
| 1524 |
+
"learning_rate": 3.571591427268582e-05,
|
| 1525 |
+
"loss": 0.5516,
|
| 1526 |
+
"step": 5225
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 3.5909712722298224,
|
| 1530 |
+
"grad_norm": 7.439993381500244,
|
| 1531 |
+
"learning_rate": 3.562091503267974e-05,
|
| 1532 |
+
"loss": 0.523,
|
| 1533 |
+
"step": 5250
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 3.6080711354309165,
|
| 1537 |
+
"grad_norm": 4.419105052947998,
|
| 1538 |
+
"learning_rate": 3.552591579267366e-05,
|
| 1539 |
+
"loss": 0.4118,
|
| 1540 |
+
"step": 5275
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 3.625170998632011,
|
| 1544 |
+
"grad_norm": 16.105833053588867,
|
| 1545 |
+
"learning_rate": 3.543091655266758e-05,
|
| 1546 |
+
"loss": 0.5083,
|
| 1547 |
+
"step": 5300
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 3.6422708618331052,
|
| 1551 |
+
"grad_norm": 5.370678424835205,
|
| 1552 |
+
"learning_rate": 3.53359173126615e-05,
|
| 1553 |
+
"loss": 0.4519,
|
| 1554 |
+
"step": 5325
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 3.6593707250342,
|
| 1558 |
+
"grad_norm": 11.25657844543457,
|
| 1559 |
+
"learning_rate": 3.524091807265542e-05,
|
| 1560 |
+
"loss": 0.3801,
|
| 1561 |
+
"step": 5350
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 3.6764705882352944,
|
| 1565 |
+
"grad_norm": 3.6341259479522705,
|
| 1566 |
+
"learning_rate": 3.514591883264934e-05,
|
| 1567 |
+
"loss": 0.3978,
|
| 1568 |
+
"step": 5375
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 3.6935704514363885,
|
| 1572 |
+
"grad_norm": 0.18588456511497498,
|
| 1573 |
+
"learning_rate": 3.505091959264326e-05,
|
| 1574 |
+
"loss": 0.4341,
|
| 1575 |
+
"step": 5400
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 3.7106703146374826,
|
| 1579 |
+
"grad_norm": 4.884586811065674,
|
| 1580 |
+
"learning_rate": 3.495592035263718e-05,
|
| 1581 |
+
"loss": 0.5095,
|
| 1582 |
+
"step": 5425
|
| 1583 |
+
},
|
| 1584 |
+
{
|
| 1585 |
+
"epoch": 3.7277701778385772,
|
| 1586 |
+
"grad_norm": 5.724035739898682,
|
| 1587 |
+
"learning_rate": 3.4860921112631104e-05,
|
| 1588 |
+
"loss": 0.4625,
|
| 1589 |
+
"step": 5450
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 3.744870041039672,
|
| 1593 |
+
"grad_norm": 5.873854637145996,
|
| 1594 |
+
"learning_rate": 3.476592187262502e-05,
|
| 1595 |
+
"loss": 0.4019,
|
| 1596 |
+
"step": 5475
|
| 1597 |
+
},
|
| 1598 |
+
{
|
| 1599 |
+
"epoch": 3.761969904240766,
|
| 1600 |
+
"grad_norm": 3.843010902404785,
|
| 1601 |
+
"learning_rate": 3.467092263261894e-05,
|
| 1602 |
+
"loss": 0.5835,
|
| 1603 |
+
"step": 5500
|
| 1604 |
+
},
|
| 1605 |
+
{
|
| 1606 |
+
"epoch": 3.7790697674418605,
|
| 1607 |
+
"grad_norm": 5.200079441070557,
|
| 1608 |
+
"learning_rate": 3.457592339261286e-05,
|
| 1609 |
+
"loss": 0.4441,
|
| 1610 |
+
"step": 5525
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 3.7961696306429547,
|
| 1614 |
+
"grad_norm": 7.437492847442627,
|
| 1615 |
+
"learning_rate": 3.448092415260678e-05,
|
| 1616 |
+
"loss": 0.3881,
|
| 1617 |
+
"step": 5550
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 3.8132694938440492,
|
| 1621 |
+
"grad_norm": 5.660449504852295,
|
| 1622 |
+
"learning_rate": 3.43859249126007e-05,
|
| 1623 |
+
"loss": 0.446,
|
| 1624 |
+
"step": 5575
|
| 1625 |
+
},
|
| 1626 |
+
{
|
| 1627 |
+
"epoch": 3.830369357045144,
|
| 1628 |
+
"grad_norm": 2.2128264904022217,
|
| 1629 |
+
"learning_rate": 3.429092567259462e-05,
|
| 1630 |
+
"loss": 0.4257,
|
| 1631 |
+
"step": 5600
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 3.847469220246238,
|
| 1635 |
+
"grad_norm": 5.218535423278809,
|
| 1636 |
+
"learning_rate": 3.419592643258854e-05,
|
| 1637 |
+
"loss": 0.4946,
|
| 1638 |
+
"step": 5625
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 3.8645690834473325,
|
| 1642 |
+
"grad_norm": 1.4989601373672485,
|
| 1643 |
+
"learning_rate": 3.410092719258246e-05,
|
| 1644 |
+
"loss": 0.3866,
|
| 1645 |
+
"step": 5650
|
| 1646 |
+
},
|
| 1647 |
+
{
|
| 1648 |
+
"epoch": 3.8816689466484267,
|
| 1649 |
+
"grad_norm": 11.225085258483887,
|
| 1650 |
+
"learning_rate": 3.400592795257638e-05,
|
| 1651 |
+
"loss": 0.4764,
|
| 1652 |
+
"step": 5675
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 3.8987688098495212,
|
| 1656 |
+
"grad_norm": 5.653949737548828,
|
| 1657 |
+
"learning_rate": 3.39109287125703e-05,
|
| 1658 |
+
"loss": 0.4362,
|
| 1659 |
+
"step": 5700
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 3.915868673050616,
|
| 1663 |
+
"grad_norm": 10.87972354888916,
|
| 1664 |
+
"learning_rate": 3.381592947256422e-05,
|
| 1665 |
+
"loss": 0.4708,
|
| 1666 |
+
"step": 5725
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 3.93296853625171,
|
| 1670 |
+
"grad_norm": 5.110008239746094,
|
| 1671 |
+
"learning_rate": 3.372093023255814e-05,
|
| 1672 |
+
"loss": 0.4373,
|
| 1673 |
+
"step": 5750
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 3.9500683994528045,
|
| 1677 |
+
"grad_norm": 4.611896991729736,
|
| 1678 |
+
"learning_rate": 3.3625930992552064e-05,
|
| 1679 |
+
"loss": 0.4808,
|
| 1680 |
+
"step": 5775
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 3.9671682626538987,
|
| 1684 |
+
"grad_norm": 3.967752695083618,
|
| 1685 |
+
"learning_rate": 3.353093175254598e-05,
|
| 1686 |
+
"loss": 0.3827,
|
| 1687 |
+
"step": 5800
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 3.9842681258549932,
|
| 1691 |
+
"grad_norm": 7.738473892211914,
|
| 1692 |
+
"learning_rate": 3.3435932512539904e-05,
|
| 1693 |
+
"loss": 0.4564,
|
| 1694 |
+
"step": 5825
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 4.0,
|
| 1698 |
+
"eval_accuracy": 0.8968931375896211,
|
| 1699 |
+
"eval_f1_macro": 0.6842607301838518,
|
| 1700 |
+
"eval_f1_micro": 0.8968931375896211,
|
| 1701 |
+
"eval_f1_weighted": 0.8922998232908407,
|
| 1702 |
+
"eval_loss": 0.3079198896884918,
|
| 1703 |
+
"eval_precision_macro": 0.7711077815931238,
|
| 1704 |
+
"eval_precision_micro": 0.8968931375896211,
|
| 1705 |
+
"eval_precision_weighted": 0.9026293299812516,
|
| 1706 |
+
"eval_recall_macro": 0.6595431638738259,
|
| 1707 |
+
"eval_recall_micro": 0.8968931375896211,
|
| 1708 |
+
"eval_recall_weighted": 0.8968931375896211,
|
| 1709 |
+
"eval_runtime": 29.6904,
|
| 1710 |
+
"eval_samples_per_second": 98.651,
|
| 1711 |
+
"eval_steps_per_second": 6.197,
|
| 1712 |
+
"step": 5848
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 4.001367989056088,
|
| 1716 |
+
"grad_norm": 12.561144828796387,
|
| 1717 |
+
"learning_rate": 3.334093327253382e-05,
|
| 1718 |
+
"loss": 0.285,
|
| 1719 |
+
"step": 5850
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 4.0184678522571815,
|
| 1723 |
+
"grad_norm": 4.646189212799072,
|
| 1724 |
+
"learning_rate": 3.3245934032527745e-05,
|
| 1725 |
+
"loss": 0.4118,
|
| 1726 |
+
"step": 5875
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 4.035567715458276,
|
| 1730 |
+
"grad_norm": 3.1821000576019287,
|
| 1731 |
+
"learning_rate": 3.315093479252166e-05,
|
| 1732 |
+
"loss": 0.5941,
|
| 1733 |
+
"step": 5900
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 4.052667578659371,
|
| 1737 |
+
"grad_norm": 6.912432670593262,
|
| 1738 |
+
"learning_rate": 3.3055935552515586e-05,
|
| 1739 |
+
"loss": 0.3519,
|
| 1740 |
+
"step": 5925
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 4.069767441860465,
|
| 1744 |
+
"grad_norm": 2.416905641555786,
|
| 1745 |
+
"learning_rate": 3.29609363125095e-05,
|
| 1746 |
+
"loss": 0.4982,
|
| 1747 |
+
"step": 5950
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 4.08686730506156,
|
| 1751 |
+
"grad_norm": 8.833620071411133,
|
| 1752 |
+
"learning_rate": 3.286593707250342e-05,
|
| 1753 |
+
"loss": 0.3608,
|
| 1754 |
+
"step": 5975
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 4.1039671682626535,
|
| 1758 |
+
"grad_norm": 0.06249883025884628,
|
| 1759 |
+
"learning_rate": 3.2770937832497336e-05,
|
| 1760 |
+
"loss": 0.4883,
|
| 1761 |
+
"step": 6000
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 4.121067031463748,
|
| 1765 |
+
"grad_norm": 2.5257112979888916,
|
| 1766 |
+
"learning_rate": 3.267593859249126e-05,
|
| 1767 |
+
"loss": 0.3371,
|
| 1768 |
+
"step": 6025
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 4.138166894664843,
|
| 1772 |
+
"grad_norm": 5.706219673156738,
|
| 1773 |
+
"learning_rate": 3.258093935248518e-05,
|
| 1774 |
+
"loss": 0.3998,
|
| 1775 |
+
"step": 6050
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 4.155266757865937,
|
| 1779 |
+
"grad_norm": 9.878058433532715,
|
| 1780 |
+
"learning_rate": 3.24859401124791e-05,
|
| 1781 |
+
"loss": 0.5082,
|
| 1782 |
+
"step": 6075
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 4.172366621067032,
|
| 1786 |
+
"grad_norm": 0.6408124566078186,
|
| 1787 |
+
"learning_rate": 3.239094087247302e-05,
|
| 1788 |
+
"loss": 0.2915,
|
| 1789 |
+
"step": 6100
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 4.1894664842681255,
|
| 1793 |
+
"grad_norm": 4.155546188354492,
|
| 1794 |
+
"learning_rate": 3.229594163246694e-05,
|
| 1795 |
+
"loss": 0.421,
|
| 1796 |
+
"step": 6125
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 4.20656634746922,
|
| 1800 |
+
"grad_norm": 2.4020016193389893,
|
| 1801 |
+
"learning_rate": 3.2200942392460865e-05,
|
| 1802 |
+
"loss": 0.3445,
|
| 1803 |
+
"step": 6150
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 4.223666210670315,
|
| 1807 |
+
"grad_norm": 9.44356632232666,
|
| 1808 |
+
"learning_rate": 3.210594315245478e-05,
|
| 1809 |
+
"loss": 0.478,
|
| 1810 |
+
"step": 6175
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 4.240766073871409,
|
| 1814 |
+
"grad_norm": 6.012132167816162,
|
| 1815 |
+
"learning_rate": 3.2010943912448705e-05,
|
| 1816 |
+
"loss": 0.4779,
|
| 1817 |
+
"step": 6200
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 4.257865937072504,
|
| 1821 |
+
"grad_norm": 5.681918621063232,
|
| 1822 |
+
"learning_rate": 3.191594467244262e-05,
|
| 1823 |
+
"loss": 0.5116,
|
| 1824 |
+
"step": 6225
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 4.2749658002735975,
|
| 1828 |
+
"grad_norm": 7.404222011566162,
|
| 1829 |
+
"learning_rate": 3.1820945432436546e-05,
|
| 1830 |
+
"loss": 0.3494,
|
| 1831 |
+
"step": 6250
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 4.292065663474692,
|
| 1835 |
+
"grad_norm": 14.846484184265137,
|
| 1836 |
+
"learning_rate": 3.172594619243046e-05,
|
| 1837 |
+
"loss": 0.5149,
|
| 1838 |
+
"step": 6275
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 4.309165526675787,
|
| 1842 |
+
"grad_norm": 1.3198221921920776,
|
| 1843 |
+
"learning_rate": 3.1630946952424386e-05,
|
| 1844 |
+
"loss": 0.4053,
|
| 1845 |
+
"step": 6300
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 4.326265389876881,
|
| 1849 |
+
"grad_norm": 3.9458107948303223,
|
| 1850 |
+
"learning_rate": 3.15359477124183e-05,
|
| 1851 |
+
"loss": 0.4499,
|
| 1852 |
+
"step": 6325
|
| 1853 |
+
},
|
| 1854 |
+
{
|
| 1855 |
+
"epoch": 4.343365253077975,
|
| 1856 |
+
"grad_norm": 4.272012233734131,
|
| 1857 |
+
"learning_rate": 3.144094847241223e-05,
|
| 1858 |
+
"loss": 0.4298,
|
| 1859 |
+
"step": 6350
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 4.3604651162790695,
|
| 1863 |
+
"grad_norm": 8.574115753173828,
|
| 1864 |
+
"learning_rate": 3.1345949232406144e-05,
|
| 1865 |
+
"loss": 0.3282,
|
| 1866 |
+
"step": 6375
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 4.377564979480164,
|
| 1870 |
+
"grad_norm": 10.030281066894531,
|
| 1871 |
+
"learning_rate": 3.125094999240006e-05,
|
| 1872 |
+
"loss": 0.486,
|
| 1873 |
+
"step": 6400
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 4.394664842681259,
|
| 1877 |
+
"grad_norm": 8.065329551696777,
|
| 1878 |
+
"learning_rate": 3.115595075239398e-05,
|
| 1879 |
+
"loss": 0.332,
|
| 1880 |
+
"step": 6425
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 4.411764705882353,
|
| 1884 |
+
"grad_norm": 10.601835250854492,
|
| 1885 |
+
"learning_rate": 3.10609515123879e-05,
|
| 1886 |
+
"loss": 0.4674,
|
| 1887 |
+
"step": 6450
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 4.428864569083447,
|
| 1891 |
+
"grad_norm": 4.035892486572266,
|
| 1892 |
+
"learning_rate": 3.096595227238182e-05,
|
| 1893 |
+
"loss": 0.5078,
|
| 1894 |
+
"step": 6475
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"epoch": 4.4459644322845415,
|
| 1898 |
+
"grad_norm": 6.8672685623168945,
|
| 1899 |
+
"learning_rate": 3.087095303237574e-05,
|
| 1900 |
+
"loss": 0.3255,
|
| 1901 |
+
"step": 6500
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 4.463064295485636,
|
| 1905 |
+
"grad_norm": 4.926382541656494,
|
| 1906 |
+
"learning_rate": 3.077595379236966e-05,
|
| 1907 |
+
"loss": 0.4204,
|
| 1908 |
+
"step": 6525
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 4.480164158686731,
|
| 1912 |
+
"grad_norm": 4.504785060882568,
|
| 1913 |
+
"learning_rate": 3.068095455236358e-05,
|
| 1914 |
+
"loss": 0.4257,
|
| 1915 |
+
"step": 6550
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 4.497264021887825,
|
| 1919 |
+
"grad_norm": 10.662145614624023,
|
| 1920 |
+
"learning_rate": 3.05859553123575e-05,
|
| 1921 |
+
"loss": 0.4991,
|
| 1922 |
+
"step": 6575
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 4.514363885088919,
|
| 1926 |
+
"grad_norm": 4.892814636230469,
|
| 1927 |
+
"learning_rate": 3.0490956072351423e-05,
|
| 1928 |
+
"loss": 0.3588,
|
| 1929 |
+
"step": 6600
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 4.5314637482900135,
|
| 1933 |
+
"grad_norm": 2.461637258529663,
|
| 1934 |
+
"learning_rate": 3.039595683234534e-05,
|
| 1935 |
+
"loss": 0.3509,
|
| 1936 |
+
"step": 6625
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 4.548563611491108,
|
| 1940 |
+
"grad_norm": 0.9665390253067017,
|
| 1941 |
+
"learning_rate": 3.0300957592339263e-05,
|
| 1942 |
+
"loss": 0.4587,
|
| 1943 |
+
"step": 6650
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 4.565663474692203,
|
| 1947 |
+
"grad_norm": 5.2920002937316895,
|
| 1948 |
+
"learning_rate": 3.020595835233318e-05,
|
| 1949 |
+
"loss": 0.3853,
|
| 1950 |
+
"step": 6675
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 4.582763337893297,
|
| 1954 |
+
"grad_norm": 6.542552471160889,
|
| 1955 |
+
"learning_rate": 3.0110959112327104e-05,
|
| 1956 |
+
"loss": 0.475,
|
| 1957 |
+
"step": 6700
|
| 1958 |
+
},
|
| 1959 |
+
{
|
| 1960 |
+
"epoch": 4.599863201094391,
|
| 1961 |
+
"grad_norm": 3.7444801330566406,
|
| 1962 |
+
"learning_rate": 3.0015959872321024e-05,
|
| 1963 |
+
"loss": 0.3638,
|
| 1964 |
+
"step": 6725
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 4.6169630642954855,
|
| 1968 |
+
"grad_norm": 6.542828559875488,
|
| 1969 |
+
"learning_rate": 2.992096063231494e-05,
|
| 1970 |
+
"loss": 0.4892,
|
| 1971 |
+
"step": 6750
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 4.63406292749658,
|
| 1975 |
+
"grad_norm": 1.4131304025650024,
|
| 1976 |
+
"learning_rate": 2.9825961392308865e-05,
|
| 1977 |
+
"loss": 0.4617,
|
| 1978 |
+
"step": 6775
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 4.651162790697675,
|
| 1982 |
+
"grad_norm": 2.6974217891693115,
|
| 1983 |
+
"learning_rate": 2.973096215230278e-05,
|
| 1984 |
+
"loss": 0.3848,
|
| 1985 |
+
"step": 6800
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 4.668262653898768,
|
| 1989 |
+
"grad_norm": 5.639090061187744,
|
| 1990 |
+
"learning_rate": 2.9635962912296705e-05,
|
| 1991 |
+
"loss": 0.3584,
|
| 1992 |
+
"step": 6825
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 4.685362517099863,
|
| 1996 |
+
"grad_norm": 11.76826000213623,
|
| 1997 |
+
"learning_rate": 2.9540963672290622e-05,
|
| 1998 |
+
"loss": 0.4498,
|
| 1999 |
+
"step": 6850
|
| 2000 |
+
},
|
| 2001 |
+
{
|
| 2002 |
+
"epoch": 4.7024623803009575,
|
| 2003 |
+
"grad_norm": 3.208690643310547,
|
| 2004 |
+
"learning_rate": 2.9445964432284546e-05,
|
| 2005 |
+
"loss": 0.3597,
|
| 2006 |
+
"step": 6875
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 4.719562243502052,
|
| 2010 |
+
"grad_norm": 5.359434604644775,
|
| 2011 |
+
"learning_rate": 2.9350965192278463e-05,
|
| 2012 |
+
"loss": 0.3621,
|
| 2013 |
+
"step": 6900
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 4.736662106703147,
|
| 2017 |
+
"grad_norm": 6.385953426361084,
|
| 2018 |
+
"learning_rate": 2.9255965952272386e-05,
|
| 2019 |
+
"loss": 0.4097,
|
| 2020 |
+
"step": 6925
|
| 2021 |
+
},
|
| 2022 |
+
{
|
| 2023 |
+
"epoch": 4.75376196990424,
|
| 2024 |
+
"grad_norm": 10.398621559143066,
|
| 2025 |
+
"learning_rate": 2.9160966712266303e-05,
|
| 2026 |
+
"loss": 0.3647,
|
| 2027 |
+
"step": 6950
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 4.770861833105335,
|
| 2031 |
+
"grad_norm": 2.8369972705841064,
|
| 2032 |
+
"learning_rate": 2.9065967472260224e-05,
|
| 2033 |
+
"loss": 0.3396,
|
| 2034 |
+
"step": 6975
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 4.7879616963064295,
|
| 2038 |
+
"grad_norm": 6.925849437713623,
|
| 2039 |
+
"learning_rate": 2.897096823225414e-05,
|
| 2040 |
+
"loss": 0.3689,
|
| 2041 |
+
"step": 7000
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 4.805061559507524,
|
| 2045 |
+
"grad_norm": 3.137173652648926,
|
| 2046 |
+
"learning_rate": 2.8875968992248064e-05,
|
| 2047 |
+
"loss": 0.5454,
|
| 2048 |
+
"step": 7025
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 4.822161422708619,
|
| 2052 |
+
"grad_norm": 12.329541206359863,
|
| 2053 |
+
"learning_rate": 2.878096975224198e-05,
|
| 2054 |
+
"loss": 0.417,
|
| 2055 |
+
"step": 7050
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 4.839261285909712,
|
| 2059 |
+
"grad_norm": 1.9504649639129639,
|
| 2060 |
+
"learning_rate": 2.8685970512235905e-05,
|
| 2061 |
+
"loss": 0.3552,
|
| 2062 |
+
"step": 7075
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 4.856361149110807,
|
| 2066 |
+
"grad_norm": 6.352153778076172,
|
| 2067 |
+
"learning_rate": 2.859097127222982e-05,
|
| 2068 |
+
"loss": 0.4427,
|
| 2069 |
+
"step": 7100
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 4.8734610123119015,
|
| 2073 |
+
"grad_norm": 5.652337074279785,
|
| 2074 |
+
"learning_rate": 2.8495972032223745e-05,
|
| 2075 |
+
"loss": 0.4454,
|
| 2076 |
+
"step": 7125
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 4.890560875512996,
|
| 2080 |
+
"grad_norm": 4.495980262756348,
|
| 2081 |
+
"learning_rate": 2.8400972792217662e-05,
|
| 2082 |
+
"loss": 0.4396,
|
| 2083 |
+
"step": 7150
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 4.907660738714091,
|
| 2087 |
+
"grad_norm": 9.805567741394043,
|
| 2088 |
+
"learning_rate": 2.8305973552211586e-05,
|
| 2089 |
+
"loss": 0.4436,
|
| 2090 |
+
"step": 7175
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 4.924760601915184,
|
| 2094 |
+
"grad_norm": 2.151742696762085,
|
| 2095 |
+
"learning_rate": 2.8210974312205503e-05,
|
| 2096 |
+
"loss": 0.4238,
|
| 2097 |
+
"step": 7200
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 4.941860465116279,
|
| 2101 |
+
"grad_norm": 4.97649621963501,
|
| 2102 |
+
"learning_rate": 2.8115975072199423e-05,
|
| 2103 |
+
"loss": 0.4693,
|
| 2104 |
+
"step": 7225
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 4.9589603283173735,
|
| 2108 |
+
"grad_norm": 2.5066914558410645,
|
| 2109 |
+
"learning_rate": 2.802097583219334e-05,
|
| 2110 |
+
"loss": 0.4568,
|
| 2111 |
+
"step": 7250
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 4.976060191518468,
|
| 2115 |
+
"grad_norm": 13.676875114440918,
|
| 2116 |
+
"learning_rate": 2.7925976592187263e-05,
|
| 2117 |
+
"loss": 0.3527,
|
| 2118 |
+
"step": 7275
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 4.993160054719562,
|
| 2122 |
+
"grad_norm": 3.240598678588867,
|
| 2123 |
+
"learning_rate": 2.783097735218118e-05,
|
| 2124 |
+
"loss": 0.4515,
|
| 2125 |
+
"step": 7300
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 5.0,
|
| 2129 |
+
"eval_accuracy": 0.9166951177876408,
|
| 2130 |
+
"eval_f1_macro": 0.7653415866627981,
|
| 2131 |
+
"eval_f1_micro": 0.9166951177876408,
|
| 2132 |
+
"eval_f1_weighted": 0.9141778991500009,
|
| 2133 |
+
"eval_loss": 0.24330385029315948,
|
| 2134 |
+
"eval_precision_macro": 0.794330339771912,
|
| 2135 |
+
"eval_precision_micro": 0.9166951177876408,
|
| 2136 |
+
"eval_precision_weighted": 0.9162095802352392,
|
| 2137 |
+
"eval_recall_macro": 0.7496536933673316,
|
| 2138 |
+
"eval_recall_micro": 0.9166951177876408,
|
| 2139 |
+
"eval_recall_weighted": 0.9166951177876408,
|
| 2140 |
+
"eval_runtime": 29.505,
|
| 2141 |
+
"eval_samples_per_second": 99.271,
|
| 2142 |
+
"eval_steps_per_second": 6.236,
|
| 2143 |
+
"step": 7310
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 5.010259917920656,
|
| 2147 |
+
"grad_norm": 3.2240583896636963,
|
| 2148 |
+
"learning_rate": 2.7735978112175104e-05,
|
| 2149 |
+
"loss": 0.406,
|
| 2150 |
+
"step": 7325
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 5.027359781121751,
|
| 2154 |
+
"grad_norm": 1.1182893514633179,
|
| 2155 |
+
"learning_rate": 2.7640978872169028e-05,
|
| 2156 |
+
"loss": 0.4151,
|
| 2157 |
+
"step": 7350
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 5.0444596443228455,
|
| 2161 |
+
"grad_norm": 3.5205605030059814,
|
| 2162 |
+
"learning_rate": 2.7545979632162945e-05,
|
| 2163 |
+
"loss": 0.4349,
|
| 2164 |
+
"step": 7375
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 5.06155950752394,
|
| 2168 |
+
"grad_norm": 5.898115634918213,
|
| 2169 |
+
"learning_rate": 2.7450980392156865e-05,
|
| 2170 |
+
"loss": 0.4127,
|
| 2171 |
+
"step": 7400
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 5.078659370725034,
|
| 2175 |
+
"grad_norm": 2.958627700805664,
|
| 2176 |
+
"learning_rate": 2.7355981152150785e-05,
|
| 2177 |
+
"loss": 0.4479,
|
| 2178 |
+
"step": 7425
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 5.095759233926128,
|
| 2182 |
+
"grad_norm": 4.672809600830078,
|
| 2183 |
+
"learning_rate": 2.7260981912144705e-05,
|
| 2184 |
+
"loss": 0.4976,
|
| 2185 |
+
"step": 7450
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 5.112859097127223,
|
| 2189 |
+
"grad_norm": 8.304091453552246,
|
| 2190 |
+
"learning_rate": 2.7165982672138622e-05,
|
| 2191 |
+
"loss": 0.4065,
|
| 2192 |
+
"step": 7475
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 5.1299589603283176,
|
| 2196 |
+
"grad_norm": 2.276463270187378,
|
| 2197 |
+
"learning_rate": 2.7070983432132546e-05,
|
| 2198 |
+
"loss": 0.3508,
|
| 2199 |
+
"step": 7500
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 5.147058823529412,
|
| 2203 |
+
"grad_norm": 0.10079041868448257,
|
| 2204 |
+
"learning_rate": 2.6975984192126463e-05,
|
| 2205 |
+
"loss": 0.4274,
|
| 2206 |
+
"step": 7525
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 5.164158686730506,
|
| 2210 |
+
"grad_norm": 5.3365936279296875,
|
| 2211 |
+
"learning_rate": 2.6880984952120387e-05,
|
| 2212 |
+
"loss": 0.2555,
|
| 2213 |
+
"step": 7550
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 5.1812585499316,
|
| 2217 |
+
"grad_norm": 0.33459940552711487,
|
| 2218 |
+
"learning_rate": 2.6785985712114303e-05,
|
| 2219 |
+
"loss": 0.4309,
|
| 2220 |
+
"step": 7575
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 5.198358413132695,
|
| 2224 |
+
"grad_norm": 6.256367206573486,
|
| 2225 |
+
"learning_rate": 2.6690986472108227e-05,
|
| 2226 |
+
"loss": 0.3263,
|
| 2227 |
+
"step": 7600
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 5.2154582763337896,
|
| 2231 |
+
"grad_norm": 2.592665672302246,
|
| 2232 |
+
"learning_rate": 2.6595987232102144e-05,
|
| 2233 |
+
"loss": 0.4089,
|
| 2234 |
+
"step": 7625
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 5.232558139534884,
|
| 2238 |
+
"grad_norm": 5.69423246383667,
|
| 2239 |
+
"learning_rate": 2.6500987992096064e-05,
|
| 2240 |
+
"loss": 0.4288,
|
| 2241 |
+
"step": 7650
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 5.249658002735978,
|
| 2245 |
+
"grad_norm": 7.463606834411621,
|
| 2246 |
+
"learning_rate": 2.6405988752089985e-05,
|
| 2247 |
+
"loss": 0.35,
|
| 2248 |
+
"step": 7675
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 5.266757865937072,
|
| 2252 |
+
"grad_norm": 4.466919898986816,
|
| 2253 |
+
"learning_rate": 2.6310989512083905e-05,
|
| 2254 |
+
"loss": 0.4364,
|
| 2255 |
+
"step": 7700
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 5.283857729138167,
|
| 2259 |
+
"grad_norm": 5.60076904296875,
|
| 2260 |
+
"learning_rate": 2.621599027207782e-05,
|
| 2261 |
+
"loss": 0.2817,
|
| 2262 |
+
"step": 7725
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 5.300957592339262,
|
| 2266 |
+
"grad_norm": 1.8708001375198364,
|
| 2267 |
+
"learning_rate": 2.6120991032071745e-05,
|
| 2268 |
+
"loss": 0.3567,
|
| 2269 |
+
"step": 7750
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 5.318057455540355,
|
| 2273 |
+
"grad_norm": 4.970040798187256,
|
| 2274 |
+
"learning_rate": 2.6025991792065662e-05,
|
| 2275 |
+
"loss": 0.2724,
|
| 2276 |
+
"step": 7775
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 5.33515731874145,
|
| 2280 |
+
"grad_norm": 1.4727894067764282,
|
| 2281 |
+
"learning_rate": 2.5930992552059586e-05,
|
| 2282 |
+
"loss": 0.3978,
|
| 2283 |
+
"step": 7800
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 5.352257181942544,
|
| 2287 |
+
"grad_norm": 4.697504043579102,
|
| 2288 |
+
"learning_rate": 2.5835993312053503e-05,
|
| 2289 |
+
"loss": 0.447,
|
| 2290 |
+
"step": 7825
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 5.369357045143639,
|
| 2294 |
+
"grad_norm": 2.3965351581573486,
|
| 2295 |
+
"learning_rate": 2.5740994072047426e-05,
|
| 2296 |
+
"loss": 0.3596,
|
| 2297 |
+
"step": 7850
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 5.386456908344734,
|
| 2301 |
+
"grad_norm": 8.727534294128418,
|
| 2302 |
+
"learning_rate": 2.5645994832041343e-05,
|
| 2303 |
+
"loss": 0.3787,
|
| 2304 |
+
"step": 7875
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 5.403556771545827,
|
| 2308 |
+
"grad_norm": 5.903003692626953,
|
| 2309 |
+
"learning_rate": 2.5550995592035264e-05,
|
| 2310 |
+
"loss": 0.3899,
|
| 2311 |
+
"step": 7900
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 5.420656634746922,
|
| 2315 |
+
"grad_norm": 4.570006847381592,
|
| 2316 |
+
"learning_rate": 2.5455996352029187e-05,
|
| 2317 |
+
"loss": 0.4558,
|
| 2318 |
+
"step": 7925
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 5.437756497948016,
|
| 2322 |
+
"grad_norm": 0.9773057699203491,
|
| 2323 |
+
"learning_rate": 2.5360997112023104e-05,
|
| 2324 |
+
"loss": 0.3345,
|
| 2325 |
+
"step": 7950
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 5.454856361149111,
|
| 2329 |
+
"grad_norm": 3.641374349594116,
|
| 2330 |
+
"learning_rate": 2.5265997872017028e-05,
|
| 2331 |
+
"loss": 0.4155,
|
| 2332 |
+
"step": 7975
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 5.471956224350206,
|
| 2336 |
+
"grad_norm": 5.382412433624268,
|
| 2337 |
+
"learning_rate": 2.5170998632010945e-05,
|
| 2338 |
+
"loss": 0.4111,
|
| 2339 |
+
"step": 8000
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 5.489056087551299,
|
| 2343 |
+
"grad_norm": 2.235764980316162,
|
| 2344 |
+
"learning_rate": 2.507599939200487e-05,
|
| 2345 |
+
"loss": 0.3047,
|
| 2346 |
+
"step": 8025
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 5.506155950752394,
|
| 2350 |
+
"grad_norm": 3.8063552379608154,
|
| 2351 |
+
"learning_rate": 2.4981000151998785e-05,
|
| 2352 |
+
"loss": 0.4702,
|
| 2353 |
+
"step": 8050
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 5.523255813953488,
|
| 2357 |
+
"grad_norm": 3.573058843612671,
|
| 2358 |
+
"learning_rate": 2.4886000911992706e-05,
|
| 2359 |
+
"loss": 0.4619,
|
| 2360 |
+
"step": 8075
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 5.540355677154583,
|
| 2364 |
+
"grad_norm": 7.515729904174805,
|
| 2365 |
+
"learning_rate": 2.4791001671986626e-05,
|
| 2366 |
+
"loss": 0.3179,
|
| 2367 |
+
"step": 8100
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 5.557455540355678,
|
| 2371 |
+
"grad_norm": 1.7556642293930054,
|
| 2372 |
+
"learning_rate": 2.4696002431980546e-05,
|
| 2373 |
+
"loss": 0.3382,
|
| 2374 |
+
"step": 8125
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 5.574555403556771,
|
| 2378 |
+
"grad_norm": 2.0541181564331055,
|
| 2379 |
+
"learning_rate": 2.4601003191974463e-05,
|
| 2380 |
+
"loss": 0.4592,
|
| 2381 |
+
"step": 8150
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 5.591655266757866,
|
| 2385 |
+
"grad_norm": 2.7318685054779053,
|
| 2386 |
+
"learning_rate": 2.4506003951968383e-05,
|
| 2387 |
+
"loss": 0.4169,
|
| 2388 |
+
"step": 8175
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 5.60875512995896,
|
| 2392 |
+
"grad_norm": 8.095924377441406,
|
| 2393 |
+
"learning_rate": 2.4411004711962304e-05,
|
| 2394 |
+
"loss": 0.4333,
|
| 2395 |
+
"step": 8200
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 5.625854993160055,
|
| 2399 |
+
"grad_norm": 2.2528128623962402,
|
| 2400 |
+
"learning_rate": 2.4316005471956224e-05,
|
| 2401 |
+
"loss": 0.402,
|
| 2402 |
+
"step": 8225
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 5.642954856361149,
|
| 2406 |
+
"grad_norm": 10.526260375976562,
|
| 2407 |
+
"learning_rate": 2.4221006231950147e-05,
|
| 2408 |
+
"loss": 0.3477,
|
| 2409 |
+
"step": 8250
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 5.660054719562243,
|
| 2413 |
+
"grad_norm": 5.1542487144470215,
|
| 2414 |
+
"learning_rate": 2.4126006991944068e-05,
|
| 2415 |
+
"loss": 0.3982,
|
| 2416 |
+
"step": 8275
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 5.677154582763338,
|
| 2420 |
+
"grad_norm": 5.238847255706787,
|
| 2421 |
+
"learning_rate": 2.4031007751937988e-05,
|
| 2422 |
+
"loss": 0.416,
|
| 2423 |
+
"step": 8300
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 5.694254445964432,
|
| 2427 |
+
"grad_norm": 8.261098861694336,
|
| 2428 |
+
"learning_rate": 2.3936008511931908e-05,
|
| 2429 |
+
"loss": 0.2776,
|
| 2430 |
+
"step": 8325
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 5.711354309165527,
|
| 2434 |
+
"grad_norm": 6.928776741027832,
|
| 2435 |
+
"learning_rate": 2.3841009271925825e-05,
|
| 2436 |
+
"loss": 0.3845,
|
| 2437 |
+
"step": 8350
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 5.728454172366621,
|
| 2441 |
+
"grad_norm": 9.97043514251709,
|
| 2442 |
+
"learning_rate": 2.3746010031919745e-05,
|
| 2443 |
+
"loss": 0.4559,
|
| 2444 |
+
"step": 8375
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 5.745554035567715,
|
| 2448 |
+
"grad_norm": 1.063759446144104,
|
| 2449 |
+
"learning_rate": 2.3651010791913666e-05,
|
| 2450 |
+
"loss": 0.2716,
|
| 2451 |
+
"step": 8400
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 5.76265389876881,
|
| 2455 |
+
"grad_norm": 2.098356246948242,
|
| 2456 |
+
"learning_rate": 2.3556011551907586e-05,
|
| 2457 |
+
"loss": 0.3414,
|
| 2458 |
+
"step": 8425
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 5.779753761969904,
|
| 2462 |
+
"grad_norm": 5.218485355377197,
|
| 2463 |
+
"learning_rate": 2.3461012311901506e-05,
|
| 2464 |
+
"loss": 0.3264,
|
| 2465 |
+
"step": 8450
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 5.796853625170999,
|
| 2469 |
+
"grad_norm": 5.245093822479248,
|
| 2470 |
+
"learning_rate": 2.3366013071895427e-05,
|
| 2471 |
+
"loss": 0.3575,
|
| 2472 |
+
"step": 8475
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 5.813953488372093,
|
| 2476 |
+
"grad_norm": 2.249113082885742,
|
| 2477 |
+
"learning_rate": 2.3271013831889347e-05,
|
| 2478 |
+
"loss": 0.3042,
|
| 2479 |
+
"step": 8500
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 5.831053351573187,
|
| 2483 |
+
"grad_norm": 8.331680297851562,
|
| 2484 |
+
"learning_rate": 2.3176014591883267e-05,
|
| 2485 |
+
"loss": 0.349,
|
| 2486 |
+
"step": 8525
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 5.848153214774282,
|
| 2490 |
+
"grad_norm": 5.0042805671691895,
|
| 2491 |
+
"learning_rate": 2.3081015351877187e-05,
|
| 2492 |
+
"loss": 0.4322,
|
| 2493 |
+
"step": 8550
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 5.865253077975376,
|
| 2497 |
+
"grad_norm": 15.829472541809082,
|
| 2498 |
+
"learning_rate": 2.2986016111871104e-05,
|
| 2499 |
+
"loss": 0.5301,
|
| 2500 |
+
"step": 8575
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 5.882352941176471,
|
| 2504 |
+
"grad_norm": 5.099222660064697,
|
| 2505 |
+
"learning_rate": 2.2891016871865025e-05,
|
| 2506 |
+
"loss": 0.4318,
|
| 2507 |
+
"step": 8600
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 5.899452804377565,
|
| 2511 |
+
"grad_norm": 3.355231285095215,
|
| 2512 |
+
"learning_rate": 2.2796017631858945e-05,
|
| 2513 |
+
"loss": 0.3114,
|
| 2514 |
+
"step": 8625
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 5.916552667578659,
|
| 2518 |
+
"grad_norm": 5.962894916534424,
|
| 2519 |
+
"learning_rate": 2.2701018391852865e-05,
|
| 2520 |
+
"loss": 0.389,
|
| 2521 |
+
"step": 8650
|
| 2522 |
+
},
|
| 2523 |
+
{
|
| 2524 |
+
"epoch": 5.933652530779754,
|
| 2525 |
+
"grad_norm": 0.5913941860198975,
|
| 2526 |
+
"learning_rate": 2.2606019151846785e-05,
|
| 2527 |
+
"loss": 0.3845,
|
| 2528 |
+
"step": 8675
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 5.950752393980848,
|
| 2532 |
+
"grad_norm": 2.7873880863189697,
|
| 2533 |
+
"learning_rate": 2.2511019911840706e-05,
|
| 2534 |
+
"loss": 0.3045,
|
| 2535 |
+
"step": 8700
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 5.967852257181942,
|
| 2539 |
+
"grad_norm": 6.107980728149414,
|
| 2540 |
+
"learning_rate": 2.2416020671834626e-05,
|
| 2541 |
+
"loss": 0.4016,
|
| 2542 |
+
"step": 8725
|
| 2543 |
+
},
|
| 2544 |
+
{
|
| 2545 |
+
"epoch": 5.984952120383037,
|
| 2546 |
+
"grad_norm": 2.5966835021972656,
|
| 2547 |
+
"learning_rate": 2.2321021431828546e-05,
|
| 2548 |
+
"loss": 0.3413,
|
| 2549 |
+
"step": 8750
|
| 2550 |
+
},
|
| 2551 |
+
{
|
| 2552 |
+
"epoch": 6.0,
|
| 2553 |
+
"eval_accuracy": 0.9095254353021509,
|
| 2554 |
+
"eval_f1_macro": 0.787176394659271,
|
| 2555 |
+
"eval_f1_micro": 0.9095254353021509,
|
| 2556 |
+
"eval_f1_weighted": 0.9072901390204282,
|
| 2557 |
+
"eval_loss": 0.29091742634773254,
|
| 2558 |
+
"eval_precision_macro": 0.8953297645243259,
|
| 2559 |
+
"eval_precision_micro": 0.9095254353021509,
|
| 2560 |
+
"eval_precision_weighted": 0.9174594106183492,
|
| 2561 |
+
"eval_recall_macro": 0.7531104129954299,
|
| 2562 |
+
"eval_recall_micro": 0.9095254353021509,
|
| 2563 |
+
"eval_recall_weighted": 0.9095254353021509,
|
| 2564 |
+
"eval_runtime": 29.8878,
|
| 2565 |
+
"eval_samples_per_second": 98.0,
|
| 2566 |
+
"eval_steps_per_second": 6.156,
|
| 2567 |
+
"step": 8772
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 6.002051983584131,
|
| 2571 |
+
"grad_norm": 6.595887660980225,
|
| 2572 |
+
"learning_rate": 2.2226022191822466e-05,
|
| 2573 |
+
"loss": 0.3606,
|
| 2574 |
+
"step": 8775
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 6.019151846785226,
|
| 2578 |
+
"grad_norm": 6.390087127685547,
|
| 2579 |
+
"learning_rate": 2.2131022951816387e-05,
|
| 2580 |
+
"loss": 0.4026,
|
| 2581 |
+
"step": 8800
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 6.03625170998632,
|
| 2585 |
+
"grad_norm": 6.15175199508667,
|
| 2586 |
+
"learning_rate": 2.2036023711810304e-05,
|
| 2587 |
+
"loss": 0.3281,
|
| 2588 |
+
"step": 8825
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 6.053351573187414,
|
| 2592 |
+
"grad_norm": 18.778564453125,
|
| 2593 |
+
"learning_rate": 2.1941024471804227e-05,
|
| 2594 |
+
"loss": 0.4304,
|
| 2595 |
+
"step": 8850
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 6.070451436388509,
|
| 2599 |
+
"grad_norm": 9.964593887329102,
|
| 2600 |
+
"learning_rate": 2.1846025231798148e-05,
|
| 2601 |
+
"loss": 0.2677,
|
| 2602 |
+
"step": 8875
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 6.087551299589603,
|
| 2606 |
+
"grad_norm": 10.952362060546875,
|
| 2607 |
+
"learning_rate": 2.1751025991792068e-05,
|
| 2608 |
+
"loss": 0.5143,
|
| 2609 |
+
"step": 8900
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 6.104651162790698,
|
| 2613 |
+
"grad_norm": 2.978898048400879,
|
| 2614 |
+
"learning_rate": 2.1656026751785988e-05,
|
| 2615 |
+
"loss": 0.3024,
|
| 2616 |
+
"step": 8925
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 6.121751025991792,
|
| 2620 |
+
"grad_norm": 3.3607687950134277,
|
| 2621 |
+
"learning_rate": 2.156102751177991e-05,
|
| 2622 |
+
"loss": 0.2684,
|
| 2623 |
+
"step": 8950
|
| 2624 |
+
},
|
| 2625 |
+
{
|
| 2626 |
+
"epoch": 6.138850889192886,
|
| 2627 |
+
"grad_norm": 1.9624247550964355,
|
| 2628 |
+
"learning_rate": 2.146602827177383e-05,
|
| 2629 |
+
"loss": 0.3062,
|
| 2630 |
+
"step": 8975
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 6.155950752393981,
|
| 2634 |
+
"grad_norm": 1.7253330945968628,
|
| 2635 |
+
"learning_rate": 2.137102903176775e-05,
|
| 2636 |
+
"loss": 0.3242,
|
| 2637 |
+
"step": 9000
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 6.173050615595075,
|
| 2641 |
+
"grad_norm": 2.6368486881256104,
|
| 2642 |
+
"learning_rate": 2.1276029791761666e-05,
|
| 2643 |
+
"loss": 0.3678,
|
| 2644 |
+
"step": 9025
|
| 2645 |
+
},
|
| 2646 |
+
{
|
| 2647 |
+
"epoch": 6.19015047879617,
|
| 2648 |
+
"grad_norm": 7.951999664306641,
|
| 2649 |
+
"learning_rate": 2.1181030551755586e-05,
|
| 2650 |
+
"loss": 0.3242,
|
| 2651 |
+
"step": 9050
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 6.207250341997264,
|
| 2655 |
+
"grad_norm": 13.932353019714355,
|
| 2656 |
+
"learning_rate": 2.1086031311749506e-05,
|
| 2657 |
+
"loss": 0.244,
|
| 2658 |
+
"step": 9075
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 6.224350205198358,
|
| 2662 |
+
"grad_norm": 0.4225591719150543,
|
| 2663 |
+
"learning_rate": 2.0991032071743427e-05,
|
| 2664 |
+
"loss": 0.503,
|
| 2665 |
+
"step": 9100
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 6.241450068399453,
|
| 2669 |
+
"grad_norm": 5.91692590713501,
|
| 2670 |
+
"learning_rate": 2.0896032831737347e-05,
|
| 2671 |
+
"loss": 0.3721,
|
| 2672 |
+
"step": 9125
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 6.258549931600547,
|
| 2676 |
+
"grad_norm": 7.2567009925842285,
|
| 2677 |
+
"learning_rate": 2.0801033591731267e-05,
|
| 2678 |
+
"loss": 0.3032,
|
| 2679 |
+
"step": 9150
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 6.275649794801642,
|
| 2683 |
+
"grad_norm": 12.250884056091309,
|
| 2684 |
+
"learning_rate": 2.0706034351725188e-05,
|
| 2685 |
+
"loss": 0.4468,
|
| 2686 |
+
"step": 9175
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 6.292749658002736,
|
| 2690 |
+
"grad_norm": 5.396862030029297,
|
| 2691 |
+
"learning_rate": 2.0611035111719108e-05,
|
| 2692 |
+
"loss": 0.357,
|
| 2693 |
+
"step": 9200
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 6.30984952120383,
|
| 2697 |
+
"grad_norm": 0.8270015120506287,
|
| 2698 |
+
"learning_rate": 2.0516035871713028e-05,
|
| 2699 |
+
"loss": 0.2127,
|
| 2700 |
+
"step": 9225
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 6.326949384404925,
|
| 2704 |
+
"grad_norm": 6.996820449829102,
|
| 2705 |
+
"learning_rate": 2.042103663170695e-05,
|
| 2706 |
+
"loss": 0.4843,
|
| 2707 |
+
"step": 9250
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 6.344049247606019,
|
| 2711 |
+
"grad_norm": 3.144995927810669,
|
| 2712 |
+
"learning_rate": 2.0326037391700865e-05,
|
| 2713 |
+
"loss": 0.4409,
|
| 2714 |
+
"step": 9275
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 6.361149110807114,
|
| 2718 |
+
"grad_norm": 10.23337459564209,
|
| 2719 |
+
"learning_rate": 2.0231038151694785e-05,
|
| 2720 |
+
"loss": 0.406,
|
| 2721 |
+
"step": 9300
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 6.378248974008208,
|
| 2725 |
+
"grad_norm": 3.1300835609436035,
|
| 2726 |
+
"learning_rate": 2.0136038911688706e-05,
|
| 2727 |
+
"loss": 0.2881,
|
| 2728 |
+
"step": 9325
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 6.395348837209302,
|
| 2732 |
+
"grad_norm": 3.322756290435791,
|
| 2733 |
+
"learning_rate": 2.0041039671682626e-05,
|
| 2734 |
+
"loss": 0.2679,
|
| 2735 |
+
"step": 9350
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 6.412448700410397,
|
| 2739 |
+
"grad_norm": 7.402932643890381,
|
| 2740 |
+
"learning_rate": 1.9946040431676546e-05,
|
| 2741 |
+
"loss": 0.3229,
|
| 2742 |
+
"step": 9375
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 6.429548563611491,
|
| 2746 |
+
"grad_norm": 2.389960289001465,
|
| 2747 |
+
"learning_rate": 1.9851041191670467e-05,
|
| 2748 |
+
"loss": 0.3176,
|
| 2749 |
+
"step": 9400
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 6.446648426812586,
|
| 2753 |
+
"grad_norm": 4.060113906860352,
|
| 2754 |
+
"learning_rate": 1.9756041951664387e-05,
|
| 2755 |
+
"loss": 0.4683,
|
| 2756 |
+
"step": 9425
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 6.46374829001368,
|
| 2760 |
+
"grad_norm": 3.9129064083099365,
|
| 2761 |
+
"learning_rate": 1.966104271165831e-05,
|
| 2762 |
+
"loss": 0.3247,
|
| 2763 |
+
"step": 9450
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 6.480848153214774,
|
| 2767 |
+
"grad_norm": 7.171699523925781,
|
| 2768 |
+
"learning_rate": 1.9566043471652227e-05,
|
| 2769 |
+
"loss": 0.2698,
|
| 2770 |
+
"step": 9475
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 6.497948016415869,
|
| 2774 |
+
"grad_norm": 5.123276233673096,
|
| 2775 |
+
"learning_rate": 1.9471044231646148e-05,
|
| 2776 |
+
"loss": 0.3883,
|
| 2777 |
+
"step": 9500
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 6.515047879616963,
|
| 2781 |
+
"grad_norm": 1.7513083219528198,
|
| 2782 |
+
"learning_rate": 1.9376044991640068e-05,
|
| 2783 |
+
"loss": 0.3488,
|
| 2784 |
+
"step": 9525
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 6.532147742818058,
|
| 2788 |
+
"grad_norm": 3.334690809249878,
|
| 2789 |
+
"learning_rate": 1.9281045751633988e-05,
|
| 2790 |
+
"loss": 0.4843,
|
| 2791 |
+
"step": 9550
|
| 2792 |
+
},
|
| 2793 |
+
{
|
| 2794 |
+
"epoch": 6.549247606019152,
|
| 2795 |
+
"grad_norm": 2.4918205738067627,
|
| 2796 |
+
"learning_rate": 1.918604651162791e-05,
|
| 2797 |
+
"loss": 0.4017,
|
| 2798 |
+
"step": 9575
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 6.566347469220246,
|
| 2802 |
+
"grad_norm": 3.277615785598755,
|
| 2803 |
+
"learning_rate": 1.909104727162183e-05,
|
| 2804 |
+
"loss": 0.3807,
|
| 2805 |
+
"step": 9600
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 6.583447332421341,
|
| 2809 |
+
"grad_norm": 0.09262839704751968,
|
| 2810 |
+
"learning_rate": 1.899604803161575e-05,
|
| 2811 |
+
"loss": 0.3623,
|
| 2812 |
+
"step": 9625
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 6.600547195622435,
|
| 2816 |
+
"grad_norm": 9.153847694396973,
|
| 2817 |
+
"learning_rate": 1.890104879160967e-05,
|
| 2818 |
+
"loss": 0.3903,
|
| 2819 |
+
"step": 9650
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 6.617647058823529,
|
| 2823 |
+
"grad_norm": 4.24110221862793,
|
| 2824 |
+
"learning_rate": 1.880604955160359e-05,
|
| 2825 |
+
"loss": 0.3078,
|
| 2826 |
+
"step": 9675
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 6.634746922024624,
|
| 2830 |
+
"grad_norm": 5.625593185424805,
|
| 2831 |
+
"learning_rate": 1.871105031159751e-05,
|
| 2832 |
+
"loss": 0.3915,
|
| 2833 |
+
"step": 9700
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 6.651846785225718,
|
| 2837 |
+
"grad_norm": 6.439937591552734,
|
| 2838 |
+
"learning_rate": 1.8616051071591427e-05,
|
| 2839 |
+
"loss": 0.4783,
|
| 2840 |
+
"step": 9725
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 6.668946648426813,
|
| 2844 |
+
"grad_norm": 6.7055559158325195,
|
| 2845 |
+
"learning_rate": 1.8521051831585347e-05,
|
| 2846 |
+
"loss": 0.3364,
|
| 2847 |
+
"step": 9750
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 6.686046511627907,
|
| 2851 |
+
"grad_norm": 2.9428536891937256,
|
| 2852 |
+
"learning_rate": 1.8426052591579267e-05,
|
| 2853 |
+
"loss": 0.3409,
|
| 2854 |
+
"step": 9775
|
| 2855 |
+
},
|
| 2856 |
+
{
|
| 2857 |
+
"epoch": 6.703146374829001,
|
| 2858 |
+
"grad_norm": 3.669163942337036,
|
| 2859 |
+
"learning_rate": 1.8331053351573188e-05,
|
| 2860 |
+
"loss": 0.2637,
|
| 2861 |
+
"step": 9800
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 6.720246238030096,
|
| 2865 |
+
"grad_norm": 2.797004461288452,
|
| 2866 |
+
"learning_rate": 1.8236054111567108e-05,
|
| 2867 |
+
"loss": 0.3721,
|
| 2868 |
+
"step": 9825
|
| 2869 |
+
},
|
| 2870 |
+
{
|
| 2871 |
+
"epoch": 6.73734610123119,
|
| 2872 |
+
"grad_norm": 1.3925397396087646,
|
| 2873 |
+
"learning_rate": 1.8141054871561028e-05,
|
| 2874 |
+
"loss": 0.3227,
|
| 2875 |
+
"step": 9850
|
| 2876 |
+
},
|
| 2877 |
+
{
|
| 2878 |
+
"epoch": 6.754445964432285,
|
| 2879 |
+
"grad_norm": 9.197000503540039,
|
| 2880 |
+
"learning_rate": 1.804605563155495e-05,
|
| 2881 |
+
"loss": 0.3789,
|
| 2882 |
+
"step": 9875
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 6.771545827633379,
|
| 2886 |
+
"grad_norm": 7.823335647583008,
|
| 2887 |
+
"learning_rate": 1.795105639154887e-05,
|
| 2888 |
+
"loss": 0.2718,
|
| 2889 |
+
"step": 9900
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 6.788645690834473,
|
| 2893 |
+
"grad_norm": 1.3368386030197144,
|
| 2894 |
+
"learning_rate": 1.785605715154279e-05,
|
| 2895 |
+
"loss": 0.3345,
|
| 2896 |
+
"step": 9925
|
| 2897 |
+
},
|
| 2898 |
+
{
|
| 2899 |
+
"epoch": 6.805745554035568,
|
| 2900 |
+
"grad_norm": 4.986422061920166,
|
| 2901 |
+
"learning_rate": 1.776105791153671e-05,
|
| 2902 |
+
"loss": 0.3514,
|
| 2903 |
+
"step": 9950
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 6.822845417236662,
|
| 2907 |
+
"grad_norm": 3.647663116455078,
|
| 2908 |
+
"learning_rate": 1.7666058671530626e-05,
|
| 2909 |
+
"loss": 0.3476,
|
| 2910 |
+
"step": 9975
|
| 2911 |
+
},
|
| 2912 |
+
{
|
| 2913 |
+
"epoch": 6.839945280437757,
|
| 2914 |
+
"grad_norm": 10.99355697631836,
|
| 2915 |
+
"learning_rate": 1.7571059431524546e-05,
|
| 2916 |
+
"loss": 0.4088,
|
| 2917 |
+
"step": 10000
|
| 2918 |
+
},
|
| 2919 |
+
{
|
| 2920 |
+
"epoch": 6.857045143638851,
|
| 2921 |
+
"grad_norm": 6.52988862991333,
|
| 2922 |
+
"learning_rate": 1.7476060191518467e-05,
|
| 2923 |
+
"loss": 0.3857,
|
| 2924 |
+
"step": 10025
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 6.874145006839945,
|
| 2928 |
+
"grad_norm": 4.713931560516357,
|
| 2929 |
+
"learning_rate": 1.738106095151239e-05,
|
| 2930 |
+
"loss": 0.5173,
|
| 2931 |
+
"step": 10050
|
| 2932 |
+
},
|
| 2933 |
+
{
|
| 2934 |
+
"epoch": 6.89124487004104,
|
| 2935 |
+
"grad_norm": 1.7083477973937988,
|
| 2936 |
+
"learning_rate": 1.728606171150631e-05,
|
| 2937 |
+
"loss": 0.2722,
|
| 2938 |
+
"step": 10075
|
| 2939 |
+
},
|
| 2940 |
+
{
|
| 2941 |
+
"epoch": 6.908344733242134,
|
| 2942 |
+
"grad_norm": 2.890033483505249,
|
| 2943 |
+
"learning_rate": 1.719106247150023e-05,
|
| 2944 |
+
"loss": 0.346,
|
| 2945 |
+
"step": 10100
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 6.925444596443229,
|
| 2949 |
+
"grad_norm": 2.318939208984375,
|
| 2950 |
+
"learning_rate": 1.709606323149415e-05,
|
| 2951 |
+
"loss": 0.3563,
|
| 2952 |
+
"step": 10125
|
| 2953 |
+
},
|
| 2954 |
+
{
|
| 2955 |
+
"epoch": 6.942544459644322,
|
| 2956 |
+
"grad_norm": 1.522857666015625,
|
| 2957 |
+
"learning_rate": 1.700106399148807e-05,
|
| 2958 |
+
"loss": 0.4441,
|
| 2959 |
+
"step": 10150
|
| 2960 |
+
},
|
| 2961 |
+
{
|
| 2962 |
+
"epoch": 6.959644322845417,
|
| 2963 |
+
"grad_norm": 4.418458938598633,
|
| 2964 |
+
"learning_rate": 1.690986472108223e-05,
|
| 2965 |
+
"loss": 0.253,
|
| 2966 |
+
"step": 10175
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 6.976744186046512,
|
| 2970 |
+
"grad_norm": 7.665693759918213,
|
| 2971 |
+
"learning_rate": 1.6814865481076152e-05,
|
| 2972 |
+
"loss": 0.329,
|
| 2973 |
+
"step": 10200
|
| 2974 |
+
},
|
| 2975 |
+
{
|
| 2976 |
+
"epoch": 6.993844049247606,
|
| 2977 |
+
"grad_norm": 2.388054609298706,
|
| 2978 |
+
"learning_rate": 1.6719866241070072e-05,
|
| 2979 |
+
"loss": 0.3777,
|
| 2980 |
+
"step": 10225
|
| 2981 |
+
},
|
| 2982 |
+
{
|
| 2983 |
+
"epoch": 7.0,
|
| 2984 |
+
"eval_accuracy": 0.9214749061113008,
|
| 2985 |
+
"eval_f1_macro": 0.833050479179196,
|
| 2986 |
+
"eval_f1_micro": 0.9214749061113008,
|
| 2987 |
+
"eval_f1_weighted": 0.9214057890123231,
|
| 2988 |
+
"eval_loss": 0.23535355925559998,
|
| 2989 |
+
"eval_precision_macro": 0.8898753259090769,
|
| 2990 |
+
"eval_precision_micro": 0.9214749061113008,
|
| 2991 |
+
"eval_precision_weighted": 0.9262844759199463,
|
| 2992 |
+
"eval_recall_macro": 0.7987766214501736,
|
| 2993 |
+
"eval_recall_micro": 0.9214749061113008,
|
| 2994 |
+
"eval_recall_weighted": 0.9214749061113008,
|
| 2995 |
+
"eval_runtime": 30.0564,
|
| 2996 |
+
"eval_samples_per_second": 97.45,
|
| 2997 |
+
"eval_steps_per_second": 6.122,
|
| 2998 |
+
"step": 10234
|
| 2999 |
+
}
|
| 3000 |
+
],
|
| 3001 |
+
"logging_steps": 25,
|
| 3002 |
+
"max_steps": 14620,
|
| 3003 |
+
"num_input_tokens_seen": 0,
|
| 3004 |
+
"num_train_epochs": 10,
|
| 3005 |
+
"save_steps": 500,
|
| 3006 |
+
"total_flos": 2.242292315610641e+19,
|
| 3007 |
+
"train_batch_size": 8,
|
| 3008 |
+
"trial_name": null,
|
| 3009 |
+
"trial_params": null
|
| 3010 |
+
}
|
checkpoint-10234/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29e3d7db214846049e195034680a123a2249f7237975ce3e50e4108119dd919c
|
| 3 |
+
size 5048
|
config.json
ADDED
|
@@ -0,0 +1,63 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/vit-large-patch16-224",
|
| 3 |
+
"_num_labels": 17,
|
| 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": "AadhaarBack",
|
| 14 |
+
"1": "AadhaarBackMasked",
|
| 15 |
+
"2": "AadhaarFront",
|
| 16 |
+
"3": "AadhaarFrontMasked",
|
| 17 |
+
"4": "AadhaarRegular",
|
| 18 |
+
"5": "AadhaarRegularMasked",
|
| 19 |
+
"6": "AadhaarSelfGenerated",
|
| 20 |
+
"7": "DrivingLicenseNew",
|
| 21 |
+
"8": "DrivingLicenseOld",
|
| 22 |
+
"9": "MultipleOVDs",
|
| 23 |
+
"10": "PanCard",
|
| 24 |
+
"11": "PassportFirst",
|
| 25 |
+
"12": "PassportLast",
|
| 26 |
+
"13": "PassportRegular",
|
| 27 |
+
"14": "VoterCardBack",
|
| 28 |
+
"15": "VoterCardFront",
|
| 29 |
+
"16": "VoterCardRegular"
|
| 30 |
+
},
|
| 31 |
+
"image_size": 224,
|
| 32 |
+
"initializer_range": 0.02,
|
| 33 |
+
"intermediate_size": 4096,
|
| 34 |
+
"label2id": {
|
| 35 |
+
"AadhaarBack": 0,
|
| 36 |
+
"AadhaarBackMasked": 1,
|
| 37 |
+
"AadhaarFront": 2,
|
| 38 |
+
"AadhaarFrontMasked": 3,
|
| 39 |
+
"AadhaarRegular": 4,
|
| 40 |
+
"AadhaarRegularMasked": 5,
|
| 41 |
+
"AadhaarSelfGenerated": 6,
|
| 42 |
+
"DrivingLicenseNew": 7,
|
| 43 |
+
"DrivingLicenseOld": 8,
|
| 44 |
+
"MultipleOVDs": 9,
|
| 45 |
+
"PanCard": 10,
|
| 46 |
+
"PassportFirst": 11,
|
| 47 |
+
"PassportLast": 12,
|
| 48 |
+
"PassportRegular": 13,
|
| 49 |
+
"VoterCardBack": 14,
|
| 50 |
+
"VoterCardFront": 15,
|
| 51 |
+
"VoterCardRegular": 16
|
| 52 |
+
},
|
| 53 |
+
"layer_norm_eps": 1e-12,
|
| 54 |
+
"model_type": "vit",
|
| 55 |
+
"num_attention_heads": 16,
|
| 56 |
+
"num_channels": 3,
|
| 57 |
+
"num_hidden_layers": 24,
|
| 58 |
+
"patch_size": 16,
|
| 59 |
+
"problem_type": "single_label_classification",
|
| 60 |
+
"qkv_bias": true,
|
| 61 |
+
"torch_dtype": "float32",
|
| 62 |
+
"transformers_version": "4.40.1"
|
| 63 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b84b79d2c3d3dffecdc42ac6506d7890bf3056236ac00e06bac2769c5ad6701e
|
| 3 |
+
size 1213322772
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_valid_processor_keys": [
|
| 3 |
+
"images",
|
| 4 |
+
"do_resize",
|
| 5 |
+
"size",
|
| 6 |
+
"resample",
|
| 7 |
+
"do_rescale",
|
| 8 |
+
"rescale_factor",
|
| 9 |
+
"do_normalize",
|
| 10 |
+
"image_mean",
|
| 11 |
+
"image_std",
|
| 12 |
+
"return_tensors",
|
| 13 |
+
"data_format",
|
| 14 |
+
"input_data_format"
|
| 15 |
+
],
|
| 16 |
+
"do_normalize": true,
|
| 17 |
+
"do_rescale": true,
|
| 18 |
+
"do_resize": true,
|
| 19 |
+
"image_mean": [
|
| 20 |
+
0.5,
|
| 21 |
+
0.5,
|
| 22 |
+
0.5
|
| 23 |
+
],
|
| 24 |
+
"image_processor_type": "ViTImageProcessor",
|
| 25 |
+
"image_std": [
|
| 26 |
+
0.5,
|
| 27 |
+
0.5,
|
| 28 |
+
0.5
|
| 29 |
+
],
|
| 30 |
+
"resample": 2,
|
| 31 |
+
"rescale_factor": 0.00392156862745098,
|
| 32 |
+
"size": {
|
| 33 |
+
"height": 224,
|
| 34 |
+
"width": 224
|
| 35 |
+
}
|
| 36 |
+
}
|
runs/May03_11-45-08_60f4804cf903/events.out.tfevents.1714736709.60f4804cf903.13020.0
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 sha256:542761337e7c49648831050c6890d0b60dfa560e23bcd70a10c6824f74dc9d60
|
| 3 |
+
size 111117
|
runs/May03_11-45-08_60f4804cf903/events.out.tfevents.1714740351.60f4804cf903.13020.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d00104cfe81bc35cf0eeb5c14768e5e656952a424b0dc1e0ebf4e88a06688ab1
|
| 3 |
+
size 921
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29e3d7db214846049e195034680a123a2249f7237975ce3e50e4108119dd919c
|
| 3 |
+
size 5048
|
training_params.json
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "autotrain-vit-large-patch16-224/autotrain-data",
|
| 3 |
+
"model": "google/vit-large-patch16-224",
|
| 4 |
+
"username": "Kushagra07",
|
| 5 |
+
"lr": 5e-05,
|
| 6 |
+
"epochs": 10,
|
| 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": "autotrain-vit-large-patch16-224",
|
| 19 |
+
"auto_find_batch_size": true,
|
| 20 |
+
"mixed_precision": "fp16",
|
| 21 |
+
"save_total_limit": 1,
|
| 22 |
+
"push_to_hub": true,
|
| 23 |
+
"evaluation_strategy": "epoch",
|
| 24 |
+
"image_column": "autotrain_image",
|
| 25 |
+
"target_column": "autotrain_label",
|
| 26 |
+
"log": "tensorboard"
|
| 27 |
+
}
|