Commit
•
e2a098e
1
Parent(s):
0cd4d76
Upload folder using huggingface_hub
Browse files- README.md +42 -0
- checkpoint-264/config.json +84 -0
- checkpoint-264/model.safetensors +3 -0
- checkpoint-264/optimizer.pt +3 -0
- checkpoint-264/rng_state.pth +3 -0
- checkpoint-264/scheduler.pt +3 -0
- checkpoint-264/trainer_state.json +1923 -0
- checkpoint-264/training_args.bin +3 -0
- config.json +84 -0
- model.safetensors +3 -0
- preprocessor_config.json +46 -0
- training_args.bin +3 -0
- training_params.json +29 -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 |
+
- as-cle-bert/banana-disease-classification
|
| 15 |
+
---
|
| 16 |
+
|
| 17 |
+
# Model Trained Using AutoTrain
|
| 18 |
+
|
| 19 |
+
- Problem type: Image Classification
|
| 20 |
+
|
| 21 |
+
## Validation Metrics
|
| 22 |
+
loss: 0.011418873444199562
|
| 23 |
+
|
| 24 |
+
f1_macro: 1.0
|
| 25 |
+
|
| 26 |
+
f1_micro: 1.0
|
| 27 |
+
|
| 28 |
+
f1_weighted: 1.0
|
| 29 |
+
|
| 30 |
+
precision_macro: 1.0
|
| 31 |
+
|
| 32 |
+
precision_micro: 1.0
|
| 33 |
+
|
| 34 |
+
precision_weighted: 1.0
|
| 35 |
+
|
| 36 |
+
recall_macro: 1.0
|
| 37 |
+
|
| 38 |
+
recall_micro: 1.0
|
| 39 |
+
|
| 40 |
+
recall_weighted: 1.0
|
| 41 |
+
|
| 42 |
+
accuracy: 1.0
|
checkpoint-264/config.json
ADDED
|
@@ -0,0 +1,84 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "microsoft/beit-base-patch16-224-pt22k-ft22k",
|
| 3 |
+
"_num_labels": 7,
|
| 4 |
+
"add_fpn": false,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"BeitForImageClassification"
|
| 7 |
+
],
|
| 8 |
+
"attention_probs_dropout_prob": 0.0,
|
| 9 |
+
"auxiliary_channels": 256,
|
| 10 |
+
"auxiliary_concat_input": false,
|
| 11 |
+
"auxiliary_loss_weight": 0.4,
|
| 12 |
+
"auxiliary_num_convs": 1,
|
| 13 |
+
"drop_path_rate": 0.1,
|
| 14 |
+
"hidden_act": "gelu",
|
| 15 |
+
"hidden_dropout_prob": 0.0,
|
| 16 |
+
"hidden_size": 768,
|
| 17 |
+
"id2label": {
|
| 18 |
+
"0": "banana_healthy_leaf",
|
| 19 |
+
"1": "black_sigatoka",
|
| 20 |
+
"2": "bract_mosaic_virus",
|
| 21 |
+
"3": "insect_pest",
|
| 22 |
+
"4": "moko_disease",
|
| 23 |
+
"5": "panama_disease",
|
| 24 |
+
"6": "yellow_sigatoka"
|
| 25 |
+
},
|
| 26 |
+
"image_size": 224,
|
| 27 |
+
"initializer_range": 0.02,
|
| 28 |
+
"intermediate_size": 3072,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"banana_healthy_leaf": 0,
|
| 31 |
+
"black_sigatoka": 1,
|
| 32 |
+
"bract_mosaic_virus": 2,
|
| 33 |
+
"insect_pest": 3,
|
| 34 |
+
"moko_disease": 4,
|
| 35 |
+
"panama_disease": 5,
|
| 36 |
+
"yellow_sigatoka": 6
|
| 37 |
+
},
|
| 38 |
+
"layer_norm_eps": 1e-12,
|
| 39 |
+
"layer_scale_init_value": 0.1,
|
| 40 |
+
"model_type": "beit",
|
| 41 |
+
"num_attention_heads": 12,
|
| 42 |
+
"num_channels": 3,
|
| 43 |
+
"num_hidden_layers": 12,
|
| 44 |
+
"out_features": [
|
| 45 |
+
"stage12"
|
| 46 |
+
],
|
| 47 |
+
"out_indices": [
|
| 48 |
+
12
|
| 49 |
+
],
|
| 50 |
+
"patch_size": 16,
|
| 51 |
+
"pool_scales": [
|
| 52 |
+
1,
|
| 53 |
+
2,
|
| 54 |
+
3,
|
| 55 |
+
6
|
| 56 |
+
],
|
| 57 |
+
"problem_type": "single_label_classification",
|
| 58 |
+
"reshape_hidden_states": true,
|
| 59 |
+
"semantic_loss_ignore_index": 255,
|
| 60 |
+
"stage_names": [
|
| 61 |
+
"stem",
|
| 62 |
+
"stage1",
|
| 63 |
+
"stage2",
|
| 64 |
+
"stage3",
|
| 65 |
+
"stage4",
|
| 66 |
+
"stage5",
|
| 67 |
+
"stage6",
|
| 68 |
+
"stage7",
|
| 69 |
+
"stage8",
|
| 70 |
+
"stage9",
|
| 71 |
+
"stage10",
|
| 72 |
+
"stage11",
|
| 73 |
+
"stage12"
|
| 74 |
+
],
|
| 75 |
+
"torch_dtype": "float32",
|
| 76 |
+
"transformers_version": "4.39.0",
|
| 77 |
+
"use_absolute_position_embeddings": false,
|
| 78 |
+
"use_auxiliary_head": true,
|
| 79 |
+
"use_mask_token": false,
|
| 80 |
+
"use_mean_pooling": true,
|
| 81 |
+
"use_relative_position_bias": true,
|
| 82 |
+
"use_shared_relative_position_bias": false,
|
| 83 |
+
"vocab_size": 8192
|
| 84 |
+
}
|
checkpoint-264/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0b35f130963ec8b66044115b2000e40353aef8d421ed5242fb32a6954dc33fec
|
| 3 |
+
size 343095708
|
checkpoint-264/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d37957dfd87d12aa0ccbb9685d326b392bac90c204a0e0cb24cb73d7ea9d822
|
| 3 |
+
size 686319098
|
checkpoint-264/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:254600c20a459c792a733eb3210f08d6765e54864ed0b8bec5bc3c5c16ea31b8
|
| 3 |
+
size 14054
|
checkpoint-264/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8ee2802e88455fa306fc6f58b015507c9e650e0d133b89385cd179192619e4ed
|
| 3 |
+
size 1064
|
checkpoint-264/trainer_state.json
ADDED
|
@@ -0,0 +1,1923 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.011418873444199562,
|
| 3 |
+
"best_model_checkpoint": "beit-banana-diseases/checkpoint-264",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 264,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01,
|
| 13 |
+
"grad_norm": 21.393463134765625,
|
| 14 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 15 |
+
"loss": 1.9713,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.02,
|
| 20 |
+
"grad_norm": 22.38655662536621,
|
| 21 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 22 |
+
"loss": 1.9772,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.03,
|
| 27 |
+
"grad_norm": 21.854021072387695,
|
| 28 |
+
"learning_rate": 5.555555555555556e-06,
|
| 29 |
+
"loss": 1.7078,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.05,
|
| 34 |
+
"grad_norm": 24.193124771118164,
|
| 35 |
+
"learning_rate": 7.4074074074074075e-06,
|
| 36 |
+
"loss": 2.1819,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.06,
|
| 41 |
+
"grad_norm": 20.156389236450195,
|
| 42 |
+
"learning_rate": 9.259259259259259e-06,
|
| 43 |
+
"loss": 1.7843,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.07,
|
| 48 |
+
"grad_norm": 21.425861358642578,
|
| 49 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 50 |
+
"loss": 1.5904,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.08,
|
| 55 |
+
"grad_norm": 23.77149200439453,
|
| 56 |
+
"learning_rate": 1.2962962962962962e-05,
|
| 57 |
+
"loss": 1.8766,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.09,
|
| 62 |
+
"grad_norm": 21.238325119018555,
|
| 63 |
+
"learning_rate": 1.4814814814814815e-05,
|
| 64 |
+
"loss": 1.6849,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.1,
|
| 69 |
+
"grad_norm": 21.050870895385742,
|
| 70 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 71 |
+
"loss": 1.9811,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.11,
|
| 76 |
+
"grad_norm": 21.381715774536133,
|
| 77 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 78 |
+
"loss": 1.7029,
|
| 79 |
+
"step": 10
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.12,
|
| 83 |
+
"grad_norm": 23.12903594970703,
|
| 84 |
+
"learning_rate": 2.037037037037037e-05,
|
| 85 |
+
"loss": 1.5096,
|
| 86 |
+
"step": 11
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.14,
|
| 90 |
+
"grad_norm": 21.264657974243164,
|
| 91 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 92 |
+
"loss": 1.748,
|
| 93 |
+
"step": 12
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.15,
|
| 97 |
+
"grad_norm": 22.664464950561523,
|
| 98 |
+
"learning_rate": 2.4074074074074074e-05,
|
| 99 |
+
"loss": 1.5399,
|
| 100 |
+
"step": 13
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.16,
|
| 104 |
+
"grad_norm": 23.9726619720459,
|
| 105 |
+
"learning_rate": 2.5925925925925925e-05,
|
| 106 |
+
"loss": 1.6543,
|
| 107 |
+
"step": 14
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.17,
|
| 111 |
+
"grad_norm": 25.409692764282227,
|
| 112 |
+
"learning_rate": 2.777777777777778e-05,
|
| 113 |
+
"loss": 1.4427,
|
| 114 |
+
"step": 15
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.18,
|
| 118 |
+
"grad_norm": 19.801799774169922,
|
| 119 |
+
"learning_rate": 2.962962962962963e-05,
|
| 120 |
+
"loss": 1.415,
|
| 121 |
+
"step": 16
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.19,
|
| 125 |
+
"grad_norm": 18.942399978637695,
|
| 126 |
+
"learning_rate": 3.148148148148148e-05,
|
| 127 |
+
"loss": 1.3031,
|
| 128 |
+
"step": 17
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.2,
|
| 132 |
+
"grad_norm": 17.01711654663086,
|
| 133 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 134 |
+
"loss": 1.382,
|
| 135 |
+
"step": 18
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.22,
|
| 139 |
+
"grad_norm": 25.648284912109375,
|
| 140 |
+
"learning_rate": 3.518518518518519e-05,
|
| 141 |
+
"loss": 1.5428,
|
| 142 |
+
"step": 19
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.23,
|
| 146 |
+
"grad_norm": 17.678979873657227,
|
| 147 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 148 |
+
"loss": 1.1447,
|
| 149 |
+
"step": 20
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.24,
|
| 153 |
+
"grad_norm": 20.149110794067383,
|
| 154 |
+
"learning_rate": 3.888888888888889e-05,
|
| 155 |
+
"loss": 1.0376,
|
| 156 |
+
"step": 21
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.25,
|
| 160 |
+
"grad_norm": 22.123849868774414,
|
| 161 |
+
"learning_rate": 4.074074074074074e-05,
|
| 162 |
+
"loss": 1.037,
|
| 163 |
+
"step": 22
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.26,
|
| 167 |
+
"grad_norm": 16.500776290893555,
|
| 168 |
+
"learning_rate": 4.259259259259259e-05,
|
| 169 |
+
"loss": 0.7835,
|
| 170 |
+
"step": 23
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.27,
|
| 174 |
+
"grad_norm": 24.026493072509766,
|
| 175 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 176 |
+
"loss": 1.1297,
|
| 177 |
+
"step": 24
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.28,
|
| 181 |
+
"grad_norm": 15.726734161376953,
|
| 182 |
+
"learning_rate": 4.62962962962963e-05,
|
| 183 |
+
"loss": 0.872,
|
| 184 |
+
"step": 25
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.3,
|
| 188 |
+
"grad_norm": 27.0452823638916,
|
| 189 |
+
"learning_rate": 4.814814814814815e-05,
|
| 190 |
+
"loss": 1.4377,
|
| 191 |
+
"step": 26
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.31,
|
| 195 |
+
"grad_norm": 19.147762298583984,
|
| 196 |
+
"learning_rate": 5e-05,
|
| 197 |
+
"loss": 0.8577,
|
| 198 |
+
"step": 27
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.32,
|
| 202 |
+
"grad_norm": 18.310501098632812,
|
| 203 |
+
"learning_rate": 4.9789029535864986e-05,
|
| 204 |
+
"loss": 0.7565,
|
| 205 |
+
"step": 28
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.33,
|
| 209 |
+
"grad_norm": 39.3529052734375,
|
| 210 |
+
"learning_rate": 4.957805907172996e-05,
|
| 211 |
+
"loss": 1.1609,
|
| 212 |
+
"step": 29
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.34,
|
| 216 |
+
"grad_norm": 18.195653915405273,
|
| 217 |
+
"learning_rate": 4.936708860759494e-05,
|
| 218 |
+
"loss": 0.7501,
|
| 219 |
+
"step": 30
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.35,
|
| 223 |
+
"grad_norm": 18.973045349121094,
|
| 224 |
+
"learning_rate": 4.9156118143459915e-05,
|
| 225 |
+
"loss": 0.5452,
|
| 226 |
+
"step": 31
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.36,
|
| 230 |
+
"grad_norm": 34.03499221801758,
|
| 231 |
+
"learning_rate": 4.89451476793249e-05,
|
| 232 |
+
"loss": 0.7635,
|
| 233 |
+
"step": 32
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.38,
|
| 237 |
+
"grad_norm": 20.804288864135742,
|
| 238 |
+
"learning_rate": 4.8734177215189874e-05,
|
| 239 |
+
"loss": 0.5612,
|
| 240 |
+
"step": 33
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.39,
|
| 244 |
+
"grad_norm": 17.79737663269043,
|
| 245 |
+
"learning_rate": 4.852320675105486e-05,
|
| 246 |
+
"loss": 0.6733,
|
| 247 |
+
"step": 34
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.4,
|
| 251 |
+
"grad_norm": 16.90382194519043,
|
| 252 |
+
"learning_rate": 4.8312236286919834e-05,
|
| 253 |
+
"loss": 0.6375,
|
| 254 |
+
"step": 35
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.41,
|
| 258 |
+
"grad_norm": 10.320564270019531,
|
| 259 |
+
"learning_rate": 4.810126582278481e-05,
|
| 260 |
+
"loss": 0.2782,
|
| 261 |
+
"step": 36
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.42,
|
| 265 |
+
"grad_norm": 23.366121292114258,
|
| 266 |
+
"learning_rate": 4.789029535864979e-05,
|
| 267 |
+
"loss": 0.5793,
|
| 268 |
+
"step": 37
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.43,
|
| 272 |
+
"grad_norm": 10.37669563293457,
|
| 273 |
+
"learning_rate": 4.767932489451477e-05,
|
| 274 |
+
"loss": 0.2511,
|
| 275 |
+
"step": 38
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.44,
|
| 279 |
+
"grad_norm": 12.599074363708496,
|
| 280 |
+
"learning_rate": 4.7468354430379746e-05,
|
| 281 |
+
"loss": 0.3908,
|
| 282 |
+
"step": 39
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.45,
|
| 286 |
+
"grad_norm": 12.652225494384766,
|
| 287 |
+
"learning_rate": 4.725738396624473e-05,
|
| 288 |
+
"loss": 0.382,
|
| 289 |
+
"step": 40
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.47,
|
| 293 |
+
"grad_norm": 12.52493953704834,
|
| 294 |
+
"learning_rate": 4.704641350210971e-05,
|
| 295 |
+
"loss": 0.5196,
|
| 296 |
+
"step": 41
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.48,
|
| 300 |
+
"grad_norm": 24.851703643798828,
|
| 301 |
+
"learning_rate": 4.683544303797468e-05,
|
| 302 |
+
"loss": 0.759,
|
| 303 |
+
"step": 42
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.49,
|
| 307 |
+
"grad_norm": 20.065210342407227,
|
| 308 |
+
"learning_rate": 4.6624472573839666e-05,
|
| 309 |
+
"loss": 0.4902,
|
| 310 |
+
"step": 43
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.5,
|
| 314 |
+
"grad_norm": 6.980092525482178,
|
| 315 |
+
"learning_rate": 4.641350210970464e-05,
|
| 316 |
+
"loss": 0.1772,
|
| 317 |
+
"step": 44
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.51,
|
| 321 |
+
"grad_norm": 10.875155448913574,
|
| 322 |
+
"learning_rate": 4.6202531645569625e-05,
|
| 323 |
+
"loss": 0.2554,
|
| 324 |
+
"step": 45
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.52,
|
| 328 |
+
"grad_norm": 8.317543029785156,
|
| 329 |
+
"learning_rate": 4.59915611814346e-05,
|
| 330 |
+
"loss": 0.1368,
|
| 331 |
+
"step": 46
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.53,
|
| 335 |
+
"grad_norm": 15.295212745666504,
|
| 336 |
+
"learning_rate": 4.5780590717299585e-05,
|
| 337 |
+
"loss": 0.539,
|
| 338 |
+
"step": 47
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.55,
|
| 342 |
+
"grad_norm": 9.764492988586426,
|
| 343 |
+
"learning_rate": 4.556962025316456e-05,
|
| 344 |
+
"loss": 0.3554,
|
| 345 |
+
"step": 48
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.56,
|
| 349 |
+
"grad_norm": 18.125457763671875,
|
| 350 |
+
"learning_rate": 4.535864978902954e-05,
|
| 351 |
+
"loss": 0.5302,
|
| 352 |
+
"step": 49
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.57,
|
| 356 |
+
"grad_norm": 7.70007848739624,
|
| 357 |
+
"learning_rate": 4.5147679324894514e-05,
|
| 358 |
+
"loss": 0.2058,
|
| 359 |
+
"step": 50
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.58,
|
| 363 |
+
"grad_norm": 15.114970207214355,
|
| 364 |
+
"learning_rate": 4.49367088607595e-05,
|
| 365 |
+
"loss": 0.4506,
|
| 366 |
+
"step": 51
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.59,
|
| 370 |
+
"grad_norm": 3.854482889175415,
|
| 371 |
+
"learning_rate": 4.4725738396624474e-05,
|
| 372 |
+
"loss": 0.0844,
|
| 373 |
+
"step": 52
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.6,
|
| 377 |
+
"grad_norm": 15.301838874816895,
|
| 378 |
+
"learning_rate": 4.451476793248946e-05,
|
| 379 |
+
"loss": 0.443,
|
| 380 |
+
"step": 53
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.61,
|
| 384 |
+
"grad_norm": 12.818358421325684,
|
| 385 |
+
"learning_rate": 4.430379746835443e-05,
|
| 386 |
+
"loss": 0.4188,
|
| 387 |
+
"step": 54
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.62,
|
| 391 |
+
"grad_norm": 20.202260971069336,
|
| 392 |
+
"learning_rate": 4.409282700421941e-05,
|
| 393 |
+
"loss": 0.518,
|
| 394 |
+
"step": 55
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.64,
|
| 398 |
+
"grad_norm": 10.593685150146484,
|
| 399 |
+
"learning_rate": 4.388185654008439e-05,
|
| 400 |
+
"loss": 0.2015,
|
| 401 |
+
"step": 56
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.65,
|
| 405 |
+
"grad_norm": 6.216127872467041,
|
| 406 |
+
"learning_rate": 4.367088607594937e-05,
|
| 407 |
+
"loss": 0.1223,
|
| 408 |
+
"step": 57
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.66,
|
| 412 |
+
"grad_norm": 8.717305183410645,
|
| 413 |
+
"learning_rate": 4.345991561181435e-05,
|
| 414 |
+
"loss": 0.1161,
|
| 415 |
+
"step": 58
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.67,
|
| 419 |
+
"grad_norm": 5.678544998168945,
|
| 420 |
+
"learning_rate": 4.324894514767933e-05,
|
| 421 |
+
"loss": 0.1481,
|
| 422 |
+
"step": 59
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.68,
|
| 426 |
+
"grad_norm": 12.164253234863281,
|
| 427 |
+
"learning_rate": 4.3037974683544305e-05,
|
| 428 |
+
"loss": 0.3593,
|
| 429 |
+
"step": 60
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.69,
|
| 433 |
+
"grad_norm": 9.91303825378418,
|
| 434 |
+
"learning_rate": 4.282700421940928e-05,
|
| 435 |
+
"loss": 0.2332,
|
| 436 |
+
"step": 61
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.7,
|
| 440 |
+
"grad_norm": 11.6513090133667,
|
| 441 |
+
"learning_rate": 4.2616033755274265e-05,
|
| 442 |
+
"loss": 0.2646,
|
| 443 |
+
"step": 62
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.72,
|
| 447 |
+
"grad_norm": 10.128318786621094,
|
| 448 |
+
"learning_rate": 4.240506329113924e-05,
|
| 449 |
+
"loss": 0.166,
|
| 450 |
+
"step": 63
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.73,
|
| 454 |
+
"grad_norm": 7.344198226928711,
|
| 455 |
+
"learning_rate": 4.2194092827004224e-05,
|
| 456 |
+
"loss": 0.1824,
|
| 457 |
+
"step": 64
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.74,
|
| 461 |
+
"grad_norm": 14.342442512512207,
|
| 462 |
+
"learning_rate": 4.19831223628692e-05,
|
| 463 |
+
"loss": 0.2132,
|
| 464 |
+
"step": 65
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.75,
|
| 468 |
+
"grad_norm": 9.733878135681152,
|
| 469 |
+
"learning_rate": 4.177215189873418e-05,
|
| 470 |
+
"loss": 0.1968,
|
| 471 |
+
"step": 66
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.76,
|
| 475 |
+
"grad_norm": 6.932378768920898,
|
| 476 |
+
"learning_rate": 4.1561181434599153e-05,
|
| 477 |
+
"loss": 0.1487,
|
| 478 |
+
"step": 67
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.77,
|
| 482 |
+
"grad_norm": 1.614574909210205,
|
| 483 |
+
"learning_rate": 4.135021097046414e-05,
|
| 484 |
+
"loss": 0.0385,
|
| 485 |
+
"step": 68
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.78,
|
| 489 |
+
"grad_norm": 12.623132705688477,
|
| 490 |
+
"learning_rate": 4.113924050632912e-05,
|
| 491 |
+
"loss": 0.1805,
|
| 492 |
+
"step": 69
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.8,
|
| 496 |
+
"grad_norm": 8.182653427124023,
|
| 497 |
+
"learning_rate": 4.0928270042194096e-05,
|
| 498 |
+
"loss": 0.1422,
|
| 499 |
+
"step": 70
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.81,
|
| 503 |
+
"grad_norm": 2.2609386444091797,
|
| 504 |
+
"learning_rate": 4.071729957805907e-05,
|
| 505 |
+
"loss": 0.0578,
|
| 506 |
+
"step": 71
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.82,
|
| 510 |
+
"grad_norm": 1.1661690473556519,
|
| 511 |
+
"learning_rate": 4.050632911392405e-05,
|
| 512 |
+
"loss": 0.0243,
|
| 513 |
+
"step": 72
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.83,
|
| 517 |
+
"grad_norm": 18.96497917175293,
|
| 518 |
+
"learning_rate": 4.029535864978903e-05,
|
| 519 |
+
"loss": 0.3047,
|
| 520 |
+
"step": 73
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.84,
|
| 524 |
+
"grad_norm": 16.800785064697266,
|
| 525 |
+
"learning_rate": 4.008438818565401e-05,
|
| 526 |
+
"loss": 0.4561,
|
| 527 |
+
"step": 74
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.85,
|
| 531 |
+
"grad_norm": 25.772275924682617,
|
| 532 |
+
"learning_rate": 3.987341772151899e-05,
|
| 533 |
+
"loss": 0.5251,
|
| 534 |
+
"step": 75
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.86,
|
| 538 |
+
"grad_norm": 10.185676574707031,
|
| 539 |
+
"learning_rate": 3.966244725738397e-05,
|
| 540 |
+
"loss": 0.2006,
|
| 541 |
+
"step": 76
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.88,
|
| 545 |
+
"grad_norm": 6.298079013824463,
|
| 546 |
+
"learning_rate": 3.945147679324895e-05,
|
| 547 |
+
"loss": 0.0788,
|
| 548 |
+
"step": 77
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.89,
|
| 552 |
+
"grad_norm": 13.123534202575684,
|
| 553 |
+
"learning_rate": 3.924050632911392e-05,
|
| 554 |
+
"loss": 0.2158,
|
| 555 |
+
"step": 78
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.9,
|
| 559 |
+
"grad_norm": 4.510250568389893,
|
| 560 |
+
"learning_rate": 3.9029535864978904e-05,
|
| 561 |
+
"loss": 0.0525,
|
| 562 |
+
"step": 79
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.91,
|
| 566 |
+
"grad_norm": 8.665478706359863,
|
| 567 |
+
"learning_rate": 3.881856540084388e-05,
|
| 568 |
+
"loss": 0.1001,
|
| 569 |
+
"step": 80
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.92,
|
| 573 |
+
"grad_norm": 5.559294700622559,
|
| 574 |
+
"learning_rate": 3.8607594936708864e-05,
|
| 575 |
+
"loss": 0.0602,
|
| 576 |
+
"step": 81
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.93,
|
| 580 |
+
"grad_norm": 5.936877250671387,
|
| 581 |
+
"learning_rate": 3.839662447257384e-05,
|
| 582 |
+
"loss": 0.0586,
|
| 583 |
+
"step": 82
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.94,
|
| 587 |
+
"grad_norm": 4.139342784881592,
|
| 588 |
+
"learning_rate": 3.8185654008438823e-05,
|
| 589 |
+
"loss": 0.0709,
|
| 590 |
+
"step": 83
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.95,
|
| 594 |
+
"grad_norm": 1.722638487815857,
|
| 595 |
+
"learning_rate": 3.79746835443038e-05,
|
| 596 |
+
"loss": 0.0264,
|
| 597 |
+
"step": 84
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.97,
|
| 601 |
+
"grad_norm": 7.886575222015381,
|
| 602 |
+
"learning_rate": 3.7763713080168776e-05,
|
| 603 |
+
"loss": 0.1474,
|
| 604 |
+
"step": 85
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.98,
|
| 608 |
+
"grad_norm": 3.067516803741455,
|
| 609 |
+
"learning_rate": 3.755274261603376e-05,
|
| 610 |
+
"loss": 0.0431,
|
| 611 |
+
"step": 86
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.99,
|
| 615 |
+
"grad_norm": 7.004971027374268,
|
| 616 |
+
"learning_rate": 3.7341772151898736e-05,
|
| 617 |
+
"loss": 0.0518,
|
| 618 |
+
"step": 87
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.0,
|
| 622 |
+
"grad_norm": 42.861572265625,
|
| 623 |
+
"learning_rate": 3.713080168776372e-05,
|
| 624 |
+
"loss": 0.686,
|
| 625 |
+
"step": 88
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 1.0,
|
| 629 |
+
"eval_accuracy": 0.935064935064935,
|
| 630 |
+
"eval_f1_macro": 0.9315281668222845,
|
| 631 |
+
"eval_f1_micro": 0.935064935064935,
|
| 632 |
+
"eval_f1_weighted": 0.9315281668222845,
|
| 633 |
+
"eval_loss": 0.2894458770751953,
|
| 634 |
+
"eval_precision_macro": 0.9553571428571429,
|
| 635 |
+
"eval_precision_micro": 0.935064935064935,
|
| 636 |
+
"eval_precision_weighted": 0.9553571428571429,
|
| 637 |
+
"eval_recall_macro": 0.935064935064935,
|
| 638 |
+
"eval_recall_micro": 0.935064935064935,
|
| 639 |
+
"eval_recall_weighted": 0.935064935064935,
|
| 640 |
+
"eval_runtime": 23.6193,
|
| 641 |
+
"eval_samples_per_second": 3.26,
|
| 642 |
+
"eval_steps_per_second": 0.212,
|
| 643 |
+
"step": 88
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.01,
|
| 647 |
+
"grad_norm": 1.910532832145691,
|
| 648 |
+
"learning_rate": 3.6919831223628695e-05,
|
| 649 |
+
"loss": 0.0282,
|
| 650 |
+
"step": 89
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.02,
|
| 654 |
+
"grad_norm": 2.4399540424346924,
|
| 655 |
+
"learning_rate": 3.670886075949367e-05,
|
| 656 |
+
"loss": 0.0293,
|
| 657 |
+
"step": 90
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.03,
|
| 661 |
+
"grad_norm": 2.8237359523773193,
|
| 662 |
+
"learning_rate": 3.649789029535865e-05,
|
| 663 |
+
"loss": 0.0184,
|
| 664 |
+
"step": 91
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.05,
|
| 668 |
+
"grad_norm": 5.545835971832275,
|
| 669 |
+
"learning_rate": 3.628691983122363e-05,
|
| 670 |
+
"loss": 0.0542,
|
| 671 |
+
"step": 92
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.06,
|
| 675 |
+
"grad_norm": 23.268957138061523,
|
| 676 |
+
"learning_rate": 3.607594936708861e-05,
|
| 677 |
+
"loss": 0.3274,
|
| 678 |
+
"step": 93
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.07,
|
| 682 |
+
"grad_norm": 38.25579833984375,
|
| 683 |
+
"learning_rate": 3.586497890295359e-05,
|
| 684 |
+
"loss": 0.8174,
|
| 685 |
+
"step": 94
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.08,
|
| 689 |
+
"grad_norm": 29.19462776184082,
|
| 690 |
+
"learning_rate": 3.565400843881857e-05,
|
| 691 |
+
"loss": 0.4695,
|
| 692 |
+
"step": 95
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.09,
|
| 696 |
+
"grad_norm": 15.189168930053711,
|
| 697 |
+
"learning_rate": 3.5443037974683544e-05,
|
| 698 |
+
"loss": 0.5005,
|
| 699 |
+
"step": 96
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.1,
|
| 703 |
+
"grad_norm": 13.557092666625977,
|
| 704 |
+
"learning_rate": 3.523206751054853e-05,
|
| 705 |
+
"loss": 0.1581,
|
| 706 |
+
"step": 97
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.11,
|
| 710 |
+
"grad_norm": 24.019737243652344,
|
| 711 |
+
"learning_rate": 3.50210970464135e-05,
|
| 712 |
+
"loss": 0.1921,
|
| 713 |
+
"step": 98
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.12,
|
| 717 |
+
"grad_norm": 8.620636940002441,
|
| 718 |
+
"learning_rate": 3.4810126582278487e-05,
|
| 719 |
+
"loss": 0.1287,
|
| 720 |
+
"step": 99
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.14,
|
| 724 |
+
"grad_norm": 27.298948287963867,
|
| 725 |
+
"learning_rate": 3.459915611814346e-05,
|
| 726 |
+
"loss": 0.1861,
|
| 727 |
+
"step": 100
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.15,
|
| 731 |
+
"grad_norm": 2.0457346439361572,
|
| 732 |
+
"learning_rate": 3.438818565400844e-05,
|
| 733 |
+
"loss": 0.0309,
|
| 734 |
+
"step": 101
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.16,
|
| 738 |
+
"grad_norm": 0.20146967470645905,
|
| 739 |
+
"learning_rate": 3.4177215189873416e-05,
|
| 740 |
+
"loss": 0.0051,
|
| 741 |
+
"step": 102
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.17,
|
| 745 |
+
"grad_norm": 5.008052825927734,
|
| 746 |
+
"learning_rate": 3.39662447257384e-05,
|
| 747 |
+
"loss": 0.0413,
|
| 748 |
+
"step": 103
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.18,
|
| 752 |
+
"grad_norm": 7.44640588760376,
|
| 753 |
+
"learning_rate": 3.3755274261603375e-05,
|
| 754 |
+
"loss": 0.0625,
|
| 755 |
+
"step": 104
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.19,
|
| 759 |
+
"grad_norm": 1.3147083520889282,
|
| 760 |
+
"learning_rate": 3.354430379746836e-05,
|
| 761 |
+
"loss": 0.0202,
|
| 762 |
+
"step": 105
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.2,
|
| 766 |
+
"grad_norm": 30.254919052124023,
|
| 767 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 768 |
+
"loss": 0.5385,
|
| 769 |
+
"step": 106
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.22,
|
| 773 |
+
"grad_norm": 8.866931915283203,
|
| 774 |
+
"learning_rate": 3.312236286919831e-05,
|
| 775 |
+
"loss": 0.0891,
|
| 776 |
+
"step": 107
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.23,
|
| 780 |
+
"grad_norm": 6.436508655548096,
|
| 781 |
+
"learning_rate": 3.291139240506329e-05,
|
| 782 |
+
"loss": 0.0887,
|
| 783 |
+
"step": 108
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.24,
|
| 787 |
+
"grad_norm": 25.527908325195312,
|
| 788 |
+
"learning_rate": 3.270042194092827e-05,
|
| 789 |
+
"loss": 0.4717,
|
| 790 |
+
"step": 109
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.25,
|
| 794 |
+
"grad_norm": 20.949968338012695,
|
| 795 |
+
"learning_rate": 3.248945147679325e-05,
|
| 796 |
+
"loss": 0.2791,
|
| 797 |
+
"step": 110
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.26,
|
| 801 |
+
"grad_norm": 1.1503649950027466,
|
| 802 |
+
"learning_rate": 3.227848101265823e-05,
|
| 803 |
+
"loss": 0.0226,
|
| 804 |
+
"step": 111
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.27,
|
| 808 |
+
"grad_norm": 11.437671661376953,
|
| 809 |
+
"learning_rate": 3.2067510548523214e-05,
|
| 810 |
+
"loss": 0.1807,
|
| 811 |
+
"step": 112
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.28,
|
| 815 |
+
"grad_norm": 0.25332069396972656,
|
| 816 |
+
"learning_rate": 3.185654008438819e-05,
|
| 817 |
+
"loss": 0.0062,
|
| 818 |
+
"step": 113
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.3,
|
| 822 |
+
"grad_norm": 1.8780378103256226,
|
| 823 |
+
"learning_rate": 3.1645569620253167e-05,
|
| 824 |
+
"loss": 0.0284,
|
| 825 |
+
"step": 114
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.31,
|
| 829 |
+
"grad_norm": 6.392487049102783,
|
| 830 |
+
"learning_rate": 3.143459915611814e-05,
|
| 831 |
+
"loss": 0.06,
|
| 832 |
+
"step": 115
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.32,
|
| 836 |
+
"grad_norm": 23.06869888305664,
|
| 837 |
+
"learning_rate": 3.1223628691983126e-05,
|
| 838 |
+
"loss": 0.3186,
|
| 839 |
+
"step": 116
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.33,
|
| 843 |
+
"grad_norm": 9.919283866882324,
|
| 844 |
+
"learning_rate": 3.10126582278481e-05,
|
| 845 |
+
"loss": 0.0903,
|
| 846 |
+
"step": 117
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.34,
|
| 850 |
+
"grad_norm": 13.233536720275879,
|
| 851 |
+
"learning_rate": 3.0801687763713086e-05,
|
| 852 |
+
"loss": 0.1328,
|
| 853 |
+
"step": 118
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.35,
|
| 857 |
+
"grad_norm": 18.54045867919922,
|
| 858 |
+
"learning_rate": 3.059071729957806e-05,
|
| 859 |
+
"loss": 0.2366,
|
| 860 |
+
"step": 119
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.36,
|
| 864 |
+
"grad_norm": 1.4564893245697021,
|
| 865 |
+
"learning_rate": 3.0379746835443042e-05,
|
| 866 |
+
"loss": 0.0253,
|
| 867 |
+
"step": 120
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.38,
|
| 871 |
+
"grad_norm": 0.2900470495223999,
|
| 872 |
+
"learning_rate": 3.0168776371308015e-05,
|
| 873 |
+
"loss": 0.0065,
|
| 874 |
+
"step": 121
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.39,
|
| 878 |
+
"grad_norm": 0.4513213038444519,
|
| 879 |
+
"learning_rate": 2.9957805907172998e-05,
|
| 880 |
+
"loss": 0.0087,
|
| 881 |
+
"step": 122
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 1.4,
|
| 885 |
+
"grad_norm": 4.35510778427124,
|
| 886 |
+
"learning_rate": 2.9746835443037974e-05,
|
| 887 |
+
"loss": 0.0477,
|
| 888 |
+
"step": 123
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 1.41,
|
| 892 |
+
"grad_norm": 2.07529354095459,
|
| 893 |
+
"learning_rate": 2.9535864978902954e-05,
|
| 894 |
+
"loss": 0.0166,
|
| 895 |
+
"step": 124
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 1.42,
|
| 899 |
+
"grad_norm": 0.21565364301204681,
|
| 900 |
+
"learning_rate": 2.9324894514767937e-05,
|
| 901 |
+
"loss": 0.0047,
|
| 902 |
+
"step": 125
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 1.43,
|
| 906 |
+
"grad_norm": 1.532738208770752,
|
| 907 |
+
"learning_rate": 2.9113924050632914e-05,
|
| 908 |
+
"loss": 0.0174,
|
| 909 |
+
"step": 126
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1.44,
|
| 913 |
+
"grad_norm": 7.4955949783325195,
|
| 914 |
+
"learning_rate": 2.8902953586497894e-05,
|
| 915 |
+
"loss": 0.1436,
|
| 916 |
+
"step": 127
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 1.45,
|
| 920 |
+
"grad_norm": 27.709796905517578,
|
| 921 |
+
"learning_rate": 2.869198312236287e-05,
|
| 922 |
+
"loss": 0.784,
|
| 923 |
+
"step": 128
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 1.47,
|
| 927 |
+
"grad_norm": 0.33867835998535156,
|
| 928 |
+
"learning_rate": 2.848101265822785e-05,
|
| 929 |
+
"loss": 0.0058,
|
| 930 |
+
"step": 129
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.48,
|
| 934 |
+
"grad_norm": 0.512848436832428,
|
| 935 |
+
"learning_rate": 2.8270042194092826e-05,
|
| 936 |
+
"loss": 0.0065,
|
| 937 |
+
"step": 130
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 1.49,
|
| 941 |
+
"grad_norm": 18.563438415527344,
|
| 942 |
+
"learning_rate": 2.805907172995781e-05,
|
| 943 |
+
"loss": 0.1981,
|
| 944 |
+
"step": 131
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 1.5,
|
| 948 |
+
"grad_norm": 18.82804298400879,
|
| 949 |
+
"learning_rate": 2.7848101265822786e-05,
|
| 950 |
+
"loss": 0.1722,
|
| 951 |
+
"step": 132
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 1.51,
|
| 955 |
+
"grad_norm": 4.428062438964844,
|
| 956 |
+
"learning_rate": 2.7637130801687766e-05,
|
| 957 |
+
"loss": 0.038,
|
| 958 |
+
"step": 133
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 1.52,
|
| 962 |
+
"grad_norm": 3.624861478805542,
|
| 963 |
+
"learning_rate": 2.7426160337552742e-05,
|
| 964 |
+
"loss": 0.0346,
|
| 965 |
+
"step": 134
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 1.53,
|
| 969 |
+
"grad_norm": 6.063285827636719,
|
| 970 |
+
"learning_rate": 2.7215189873417722e-05,
|
| 971 |
+
"loss": 0.0494,
|
| 972 |
+
"step": 135
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 1.55,
|
| 976 |
+
"grad_norm": 12.928357124328613,
|
| 977 |
+
"learning_rate": 2.7004219409282698e-05,
|
| 978 |
+
"loss": 0.3157,
|
| 979 |
+
"step": 136
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 1.56,
|
| 983 |
+
"grad_norm": 1.6940982341766357,
|
| 984 |
+
"learning_rate": 2.679324894514768e-05,
|
| 985 |
+
"loss": 0.0192,
|
| 986 |
+
"step": 137
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.57,
|
| 990 |
+
"grad_norm": 0.3945218622684479,
|
| 991 |
+
"learning_rate": 2.6582278481012658e-05,
|
| 992 |
+
"loss": 0.0058,
|
| 993 |
+
"step": 138
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 1.58,
|
| 997 |
+
"grad_norm": 0.15542316436767578,
|
| 998 |
+
"learning_rate": 2.6371308016877638e-05,
|
| 999 |
+
"loss": 0.0032,
|
| 1000 |
+
"step": 139
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 1.59,
|
| 1004 |
+
"grad_norm": 0.11174587905406952,
|
| 1005 |
+
"learning_rate": 2.616033755274262e-05,
|
| 1006 |
+
"loss": 0.0025,
|
| 1007 |
+
"step": 140
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 1.6,
|
| 1011 |
+
"grad_norm": 23.488004684448242,
|
| 1012 |
+
"learning_rate": 2.5949367088607597e-05,
|
| 1013 |
+
"loss": 0.8641,
|
| 1014 |
+
"step": 141
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.61,
|
| 1018 |
+
"grad_norm": 0.5066362023353577,
|
| 1019 |
+
"learning_rate": 2.5738396624472577e-05,
|
| 1020 |
+
"loss": 0.0083,
|
| 1021 |
+
"step": 142
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 1.62,
|
| 1025 |
+
"grad_norm": 32.906803131103516,
|
| 1026 |
+
"learning_rate": 2.5527426160337553e-05,
|
| 1027 |
+
"loss": 0.6371,
|
| 1028 |
+
"step": 143
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 1.64,
|
| 1032 |
+
"grad_norm": 11.186979293823242,
|
| 1033 |
+
"learning_rate": 2.5316455696202533e-05,
|
| 1034 |
+
"loss": 0.1184,
|
| 1035 |
+
"step": 144
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 1.65,
|
| 1039 |
+
"grad_norm": 26.368669509887695,
|
| 1040 |
+
"learning_rate": 2.510548523206751e-05,
|
| 1041 |
+
"loss": 0.2113,
|
| 1042 |
+
"step": 145
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 1.66,
|
| 1046 |
+
"grad_norm": 2.7630584239959717,
|
| 1047 |
+
"learning_rate": 2.4894514767932493e-05,
|
| 1048 |
+
"loss": 0.0236,
|
| 1049 |
+
"step": 146
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 1.67,
|
| 1053 |
+
"grad_norm": 26.372888565063477,
|
| 1054 |
+
"learning_rate": 2.468354430379747e-05,
|
| 1055 |
+
"loss": 0.339,
|
| 1056 |
+
"step": 147
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 1.68,
|
| 1060 |
+
"grad_norm": 7.727011203765869,
|
| 1061 |
+
"learning_rate": 2.447257383966245e-05,
|
| 1062 |
+
"loss": 0.091,
|
| 1063 |
+
"step": 148
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 1.69,
|
| 1067 |
+
"grad_norm": 7.276378631591797,
|
| 1068 |
+
"learning_rate": 2.426160337552743e-05,
|
| 1069 |
+
"loss": 0.0345,
|
| 1070 |
+
"step": 149
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 1.7,
|
| 1074 |
+
"grad_norm": 1.4434276819229126,
|
| 1075 |
+
"learning_rate": 2.4050632911392405e-05,
|
| 1076 |
+
"loss": 0.0191,
|
| 1077 |
+
"step": 150
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 1.72,
|
| 1081 |
+
"grad_norm": 1.5443305969238281,
|
| 1082 |
+
"learning_rate": 2.3839662447257385e-05,
|
| 1083 |
+
"loss": 0.0139,
|
| 1084 |
+
"step": 151
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 1.73,
|
| 1088 |
+
"grad_norm": 7.824341297149658,
|
| 1089 |
+
"learning_rate": 2.3628691983122365e-05,
|
| 1090 |
+
"loss": 0.1015,
|
| 1091 |
+
"step": 152
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 1.74,
|
| 1095 |
+
"grad_norm": 0.20900146663188934,
|
| 1096 |
+
"learning_rate": 2.341772151898734e-05,
|
| 1097 |
+
"loss": 0.0033,
|
| 1098 |
+
"step": 153
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 1.75,
|
| 1102 |
+
"grad_norm": 0.32141363620758057,
|
| 1103 |
+
"learning_rate": 2.320675105485232e-05,
|
| 1104 |
+
"loss": 0.0059,
|
| 1105 |
+
"step": 154
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 1.76,
|
| 1109 |
+
"grad_norm": 21.58867073059082,
|
| 1110 |
+
"learning_rate": 2.29957805907173e-05,
|
| 1111 |
+
"loss": 0.5102,
|
| 1112 |
+
"step": 155
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.77,
|
| 1116 |
+
"grad_norm": 1.1995337009429932,
|
| 1117 |
+
"learning_rate": 2.278481012658228e-05,
|
| 1118 |
+
"loss": 0.0088,
|
| 1119 |
+
"step": 156
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 1.78,
|
| 1123 |
+
"grad_norm": 8.670762062072754,
|
| 1124 |
+
"learning_rate": 2.2573839662447257e-05,
|
| 1125 |
+
"loss": 0.0449,
|
| 1126 |
+
"step": 157
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 1.8,
|
| 1130 |
+
"grad_norm": 43.51858139038086,
|
| 1131 |
+
"learning_rate": 2.2362869198312237e-05,
|
| 1132 |
+
"loss": 0.7478,
|
| 1133 |
+
"step": 158
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 1.81,
|
| 1137 |
+
"grad_norm": 21.07337760925293,
|
| 1138 |
+
"learning_rate": 2.2151898734177217e-05,
|
| 1139 |
+
"loss": 0.4485,
|
| 1140 |
+
"step": 159
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 1.82,
|
| 1144 |
+
"grad_norm": 0.5225343704223633,
|
| 1145 |
+
"learning_rate": 2.1940928270042196e-05,
|
| 1146 |
+
"loss": 0.0063,
|
| 1147 |
+
"step": 160
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 1.83,
|
| 1151 |
+
"grad_norm": 0.6471118330955505,
|
| 1152 |
+
"learning_rate": 2.1729957805907176e-05,
|
| 1153 |
+
"loss": 0.0052,
|
| 1154 |
+
"step": 161
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 1.84,
|
| 1158 |
+
"grad_norm": 5.603309154510498,
|
| 1159 |
+
"learning_rate": 2.1518987341772153e-05,
|
| 1160 |
+
"loss": 0.0676,
|
| 1161 |
+
"step": 162
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 1.85,
|
| 1165 |
+
"grad_norm": 18.191370010375977,
|
| 1166 |
+
"learning_rate": 2.1308016877637132e-05,
|
| 1167 |
+
"loss": 0.1428,
|
| 1168 |
+
"step": 163
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 1.86,
|
| 1172 |
+
"grad_norm": 15.37728214263916,
|
| 1173 |
+
"learning_rate": 2.1097046413502112e-05,
|
| 1174 |
+
"loss": 0.4738,
|
| 1175 |
+
"step": 164
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 1.88,
|
| 1179 |
+
"grad_norm": 0.964952290058136,
|
| 1180 |
+
"learning_rate": 2.088607594936709e-05,
|
| 1181 |
+
"loss": 0.0135,
|
| 1182 |
+
"step": 165
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 1.89,
|
| 1186 |
+
"grad_norm": 20.822654724121094,
|
| 1187 |
+
"learning_rate": 2.067510548523207e-05,
|
| 1188 |
+
"loss": 0.359,
|
| 1189 |
+
"step": 166
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 1.9,
|
| 1193 |
+
"grad_norm": 15.292733192443848,
|
| 1194 |
+
"learning_rate": 2.0464135021097048e-05,
|
| 1195 |
+
"loss": 0.3184,
|
| 1196 |
+
"step": 167
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 1.91,
|
| 1200 |
+
"grad_norm": 1.5272413492202759,
|
| 1201 |
+
"learning_rate": 2.0253164556962025e-05,
|
| 1202 |
+
"loss": 0.0186,
|
| 1203 |
+
"step": 168
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 1.92,
|
| 1207 |
+
"grad_norm": 17.016498565673828,
|
| 1208 |
+
"learning_rate": 2.0042194092827004e-05,
|
| 1209 |
+
"loss": 0.2154,
|
| 1210 |
+
"step": 169
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 1.93,
|
| 1214 |
+
"grad_norm": 1.4779866933822632,
|
| 1215 |
+
"learning_rate": 1.9831223628691984e-05,
|
| 1216 |
+
"loss": 0.0165,
|
| 1217 |
+
"step": 170
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 1.94,
|
| 1221 |
+
"grad_norm": 0.8727405071258545,
|
| 1222 |
+
"learning_rate": 1.962025316455696e-05,
|
| 1223 |
+
"loss": 0.0102,
|
| 1224 |
+
"step": 171
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 1.95,
|
| 1228 |
+
"grad_norm": 13.600626945495605,
|
| 1229 |
+
"learning_rate": 1.940928270042194e-05,
|
| 1230 |
+
"loss": 0.1604,
|
| 1231 |
+
"step": 172
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 1.97,
|
| 1235 |
+
"grad_norm": 0.77298504114151,
|
| 1236 |
+
"learning_rate": 1.919831223628692e-05,
|
| 1237 |
+
"loss": 0.0094,
|
| 1238 |
+
"step": 173
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 1.98,
|
| 1242 |
+
"grad_norm": 8.220309257507324,
|
| 1243 |
+
"learning_rate": 1.89873417721519e-05,
|
| 1244 |
+
"loss": 0.0887,
|
| 1245 |
+
"step": 174
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 1.99,
|
| 1249 |
+
"grad_norm": 15.369383811950684,
|
| 1250 |
+
"learning_rate": 1.877637130801688e-05,
|
| 1251 |
+
"loss": 0.3962,
|
| 1252 |
+
"step": 175
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 2.0,
|
| 1256 |
+
"grad_norm": 7.943267822265625,
|
| 1257 |
+
"learning_rate": 1.856540084388186e-05,
|
| 1258 |
+
"loss": 0.0342,
|
| 1259 |
+
"step": 176
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 2.0,
|
| 1263 |
+
"eval_accuracy": 0.961038961038961,
|
| 1264 |
+
"eval_f1_macro": 0.9603007518796992,
|
| 1265 |
+
"eval_f1_micro": 0.961038961038961,
|
| 1266 |
+
"eval_f1_weighted": 0.9603007518796992,
|
| 1267 |
+
"eval_loss": 0.05716231092810631,
|
| 1268 |
+
"eval_precision_macro": 0.9693877551020408,
|
| 1269 |
+
"eval_precision_micro": 0.961038961038961,
|
| 1270 |
+
"eval_precision_weighted": 0.9693877551020408,
|
| 1271 |
+
"eval_recall_macro": 0.961038961038961,
|
| 1272 |
+
"eval_recall_micro": 0.961038961038961,
|
| 1273 |
+
"eval_recall_weighted": 0.961038961038961,
|
| 1274 |
+
"eval_runtime": 23.4765,
|
| 1275 |
+
"eval_samples_per_second": 3.28,
|
| 1276 |
+
"eval_steps_per_second": 0.213,
|
| 1277 |
+
"step": 176
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.01,
|
| 1281 |
+
"grad_norm": 0.05514498054981232,
|
| 1282 |
+
"learning_rate": 1.8354430379746836e-05,
|
| 1283 |
+
"loss": 0.0014,
|
| 1284 |
+
"step": 177
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 2.02,
|
| 1288 |
+
"grad_norm": 2.1552014350891113,
|
| 1289 |
+
"learning_rate": 1.8143459915611816e-05,
|
| 1290 |
+
"loss": 0.0202,
|
| 1291 |
+
"step": 178
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.03,
|
| 1295 |
+
"grad_norm": 0.025016020983457565,
|
| 1296 |
+
"learning_rate": 1.7932489451476795e-05,
|
| 1297 |
+
"loss": 0.0007,
|
| 1298 |
+
"step": 179
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.05,
|
| 1302 |
+
"grad_norm": 15.627620697021484,
|
| 1303 |
+
"learning_rate": 1.7721518987341772e-05,
|
| 1304 |
+
"loss": 0.1231,
|
| 1305 |
+
"step": 180
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.06,
|
| 1309 |
+
"grad_norm": 0.46118250489234924,
|
| 1310 |
+
"learning_rate": 1.751054852320675e-05,
|
| 1311 |
+
"loss": 0.0063,
|
| 1312 |
+
"step": 181
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.07,
|
| 1316 |
+
"grad_norm": 15.023947715759277,
|
| 1317 |
+
"learning_rate": 1.729957805907173e-05,
|
| 1318 |
+
"loss": 0.1147,
|
| 1319 |
+
"step": 182
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.08,
|
| 1323 |
+
"grad_norm": 11.93247127532959,
|
| 1324 |
+
"learning_rate": 1.7088607594936708e-05,
|
| 1325 |
+
"loss": 0.0811,
|
| 1326 |
+
"step": 183
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.09,
|
| 1330 |
+
"grad_norm": 2.5019609928131104,
|
| 1331 |
+
"learning_rate": 1.6877637130801688e-05,
|
| 1332 |
+
"loss": 0.0119,
|
| 1333 |
+
"step": 184
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.1,
|
| 1337 |
+
"grad_norm": 0.09159235656261444,
|
| 1338 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 1339 |
+
"loss": 0.0014,
|
| 1340 |
+
"step": 185
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 2.11,
|
| 1344 |
+
"grad_norm": 26.242231369018555,
|
| 1345 |
+
"learning_rate": 1.6455696202531644e-05,
|
| 1346 |
+
"loss": 0.236,
|
| 1347 |
+
"step": 186
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 2.12,
|
| 1351 |
+
"grad_norm": 2.3096747398376465,
|
| 1352 |
+
"learning_rate": 1.6244725738396624e-05,
|
| 1353 |
+
"loss": 0.0093,
|
| 1354 |
+
"step": 187
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 2.14,
|
| 1358 |
+
"grad_norm": 0.5707495212554932,
|
| 1359 |
+
"learning_rate": 1.6033755274261607e-05,
|
| 1360 |
+
"loss": 0.0059,
|
| 1361 |
+
"step": 188
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 2.15,
|
| 1365 |
+
"grad_norm": 6.667356014251709,
|
| 1366 |
+
"learning_rate": 1.5822784810126583e-05,
|
| 1367 |
+
"loss": 0.0563,
|
| 1368 |
+
"step": 189
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 2.16,
|
| 1372 |
+
"grad_norm": 0.10668758302927017,
|
| 1373 |
+
"learning_rate": 1.5611814345991563e-05,
|
| 1374 |
+
"loss": 0.0023,
|
| 1375 |
+
"step": 190
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 2.17,
|
| 1379 |
+
"grad_norm": 4.116157531738281,
|
| 1380 |
+
"learning_rate": 1.5400843881856543e-05,
|
| 1381 |
+
"loss": 0.0335,
|
| 1382 |
+
"step": 191
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 2.18,
|
| 1386 |
+
"grad_norm": 26.253101348876953,
|
| 1387 |
+
"learning_rate": 1.5189873417721521e-05,
|
| 1388 |
+
"loss": 0.1706,
|
| 1389 |
+
"step": 192
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 2.19,
|
| 1393 |
+
"grad_norm": 1.0219850540161133,
|
| 1394 |
+
"learning_rate": 1.4978902953586499e-05,
|
| 1395 |
+
"loss": 0.0066,
|
| 1396 |
+
"step": 193
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 2.2,
|
| 1400 |
+
"grad_norm": 41.47359848022461,
|
| 1401 |
+
"learning_rate": 1.4767932489451477e-05,
|
| 1402 |
+
"loss": 0.6927,
|
| 1403 |
+
"step": 194
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 2.22,
|
| 1407 |
+
"grad_norm": 9.33047866821289,
|
| 1408 |
+
"learning_rate": 1.4556962025316457e-05,
|
| 1409 |
+
"loss": 0.0786,
|
| 1410 |
+
"step": 195
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 2.23,
|
| 1414 |
+
"grad_norm": 0.5574650168418884,
|
| 1415 |
+
"learning_rate": 1.4345991561181435e-05,
|
| 1416 |
+
"loss": 0.0057,
|
| 1417 |
+
"step": 196
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 2.24,
|
| 1421 |
+
"grad_norm": 0.4510033428668976,
|
| 1422 |
+
"learning_rate": 1.4135021097046413e-05,
|
| 1423 |
+
"loss": 0.0043,
|
| 1424 |
+
"step": 197
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 2.25,
|
| 1428 |
+
"grad_norm": 0.07699360698461533,
|
| 1429 |
+
"learning_rate": 1.3924050632911393e-05,
|
| 1430 |
+
"loss": 0.0015,
|
| 1431 |
+
"step": 198
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 2.26,
|
| 1435 |
+
"grad_norm": 12.339035034179688,
|
| 1436 |
+
"learning_rate": 1.3713080168776371e-05,
|
| 1437 |
+
"loss": 0.1513,
|
| 1438 |
+
"step": 199
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 2.27,
|
| 1442 |
+
"grad_norm": 29.081424713134766,
|
| 1443 |
+
"learning_rate": 1.3502109704641349e-05,
|
| 1444 |
+
"loss": 0.2272,
|
| 1445 |
+
"step": 200
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 2.28,
|
| 1449 |
+
"grad_norm": 1.5039241313934326,
|
| 1450 |
+
"learning_rate": 1.3291139240506329e-05,
|
| 1451 |
+
"loss": 0.0113,
|
| 1452 |
+
"step": 201
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 2.3,
|
| 1456 |
+
"grad_norm": 1.7569365501403809,
|
| 1457 |
+
"learning_rate": 1.308016877637131e-05,
|
| 1458 |
+
"loss": 0.0143,
|
| 1459 |
+
"step": 202
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 2.31,
|
| 1463 |
+
"grad_norm": 4.633851051330566,
|
| 1464 |
+
"learning_rate": 1.2869198312236289e-05,
|
| 1465 |
+
"loss": 0.0273,
|
| 1466 |
+
"step": 203
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 2.32,
|
| 1470 |
+
"grad_norm": 9.019872665405273,
|
| 1471 |
+
"learning_rate": 1.2658227848101267e-05,
|
| 1472 |
+
"loss": 0.1055,
|
| 1473 |
+
"step": 204
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 2.33,
|
| 1477 |
+
"grad_norm": 0.14426438510417938,
|
| 1478 |
+
"learning_rate": 1.2447257383966246e-05,
|
| 1479 |
+
"loss": 0.0023,
|
| 1480 |
+
"step": 205
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 2.34,
|
| 1484 |
+
"grad_norm": 0.1260414719581604,
|
| 1485 |
+
"learning_rate": 1.2236286919831224e-05,
|
| 1486 |
+
"loss": 0.0023,
|
| 1487 |
+
"step": 206
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 2.35,
|
| 1491 |
+
"grad_norm": 0.14926879107952118,
|
| 1492 |
+
"learning_rate": 1.2025316455696203e-05,
|
| 1493 |
+
"loss": 0.0018,
|
| 1494 |
+
"step": 207
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 2.36,
|
| 1498 |
+
"grad_norm": 0.3730136752128601,
|
| 1499 |
+
"learning_rate": 1.1814345991561182e-05,
|
| 1500 |
+
"loss": 0.0058,
|
| 1501 |
+
"step": 208
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 2.38,
|
| 1505 |
+
"grad_norm": 4.101214408874512,
|
| 1506 |
+
"learning_rate": 1.160337552742616e-05,
|
| 1507 |
+
"loss": 0.0279,
|
| 1508 |
+
"step": 209
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 2.39,
|
| 1512 |
+
"grad_norm": 2.671245574951172,
|
| 1513 |
+
"learning_rate": 1.139240506329114e-05,
|
| 1514 |
+
"loss": 0.023,
|
| 1515 |
+
"step": 210
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 2.4,
|
| 1519 |
+
"grad_norm": 6.172547817230225,
|
| 1520 |
+
"learning_rate": 1.1181434599156118e-05,
|
| 1521 |
+
"loss": 0.0333,
|
| 1522 |
+
"step": 211
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 2.41,
|
| 1526 |
+
"grad_norm": 18.09273910522461,
|
| 1527 |
+
"learning_rate": 1.0970464135021098e-05,
|
| 1528 |
+
"loss": 0.3601,
|
| 1529 |
+
"step": 212
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 2.42,
|
| 1533 |
+
"grad_norm": 9.900489807128906,
|
| 1534 |
+
"learning_rate": 1.0759493670886076e-05,
|
| 1535 |
+
"loss": 0.5438,
|
| 1536 |
+
"step": 213
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 2.43,
|
| 1540 |
+
"grad_norm": 27.930665969848633,
|
| 1541 |
+
"learning_rate": 1.0548523206751056e-05,
|
| 1542 |
+
"loss": 0.2069,
|
| 1543 |
+
"step": 214
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 2.44,
|
| 1547 |
+
"grad_norm": 0.493826687335968,
|
| 1548 |
+
"learning_rate": 1.0337552742616034e-05,
|
| 1549 |
+
"loss": 0.0039,
|
| 1550 |
+
"step": 215
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 2.45,
|
| 1554 |
+
"grad_norm": 0.08800631016492844,
|
| 1555 |
+
"learning_rate": 1.0126582278481012e-05,
|
| 1556 |
+
"loss": 0.0015,
|
| 1557 |
+
"step": 216
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 2.47,
|
| 1561 |
+
"grad_norm": 1.6704317331314087,
|
| 1562 |
+
"learning_rate": 9.915611814345992e-06,
|
| 1563 |
+
"loss": 0.0121,
|
| 1564 |
+
"step": 217
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 2.48,
|
| 1568 |
+
"grad_norm": 0.060208290815353394,
|
| 1569 |
+
"learning_rate": 9.70464135021097e-06,
|
| 1570 |
+
"loss": 0.001,
|
| 1571 |
+
"step": 218
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 2.49,
|
| 1575 |
+
"grad_norm": 0.42390766739845276,
|
| 1576 |
+
"learning_rate": 9.49367088607595e-06,
|
| 1577 |
+
"loss": 0.0043,
|
| 1578 |
+
"step": 219
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 2.5,
|
| 1582 |
+
"grad_norm": 5.121891021728516,
|
| 1583 |
+
"learning_rate": 9.28270042194093e-06,
|
| 1584 |
+
"loss": 0.0373,
|
| 1585 |
+
"step": 220
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 2.51,
|
| 1589 |
+
"grad_norm": 0.2980020344257355,
|
| 1590 |
+
"learning_rate": 9.071729957805908e-06,
|
| 1591 |
+
"loss": 0.0019,
|
| 1592 |
+
"step": 221
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 2.52,
|
| 1596 |
+
"grad_norm": 1.821443796157837,
|
| 1597 |
+
"learning_rate": 8.860759493670886e-06,
|
| 1598 |
+
"loss": 0.0096,
|
| 1599 |
+
"step": 222
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 2.53,
|
| 1603 |
+
"grad_norm": 0.2338918149471283,
|
| 1604 |
+
"learning_rate": 8.649789029535866e-06,
|
| 1605 |
+
"loss": 0.0033,
|
| 1606 |
+
"step": 223
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 2.55,
|
| 1610 |
+
"grad_norm": 0.03316354379057884,
|
| 1611 |
+
"learning_rate": 8.438818565400844e-06,
|
| 1612 |
+
"loss": 0.001,
|
| 1613 |
+
"step": 224
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 2.56,
|
| 1617 |
+
"grad_norm": 44.416473388671875,
|
| 1618 |
+
"learning_rate": 8.227848101265822e-06,
|
| 1619 |
+
"loss": 0.2124,
|
| 1620 |
+
"step": 225
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 2.57,
|
| 1624 |
+
"grad_norm": 0.2957440912723541,
|
| 1625 |
+
"learning_rate": 8.016877637130803e-06,
|
| 1626 |
+
"loss": 0.0041,
|
| 1627 |
+
"step": 226
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 2.58,
|
| 1631 |
+
"grad_norm": 18.5585994720459,
|
| 1632 |
+
"learning_rate": 7.805907172995782e-06,
|
| 1633 |
+
"loss": 0.0596,
|
| 1634 |
+
"step": 227
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 2.59,
|
| 1638 |
+
"grad_norm": 5.535640716552734,
|
| 1639 |
+
"learning_rate": 7.5949367088607605e-06,
|
| 1640 |
+
"loss": 0.0476,
|
| 1641 |
+
"step": 228
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 2.6,
|
| 1645 |
+
"grad_norm": 38.173789978027344,
|
| 1646 |
+
"learning_rate": 7.3839662447257386e-06,
|
| 1647 |
+
"loss": 0.6888,
|
| 1648 |
+
"step": 229
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 2.61,
|
| 1652 |
+
"grad_norm": 18.576967239379883,
|
| 1653 |
+
"learning_rate": 7.1729957805907175e-06,
|
| 1654 |
+
"loss": 0.3636,
|
| 1655 |
+
"step": 230
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 2.62,
|
| 1659 |
+
"grad_norm": 3.596296548843384,
|
| 1660 |
+
"learning_rate": 6.9620253164556965e-06,
|
| 1661 |
+
"loss": 0.0224,
|
| 1662 |
+
"step": 231
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 2.64,
|
| 1666 |
+
"grad_norm": 0.09587596356868744,
|
| 1667 |
+
"learning_rate": 6.7510548523206746e-06,
|
| 1668 |
+
"loss": 0.0022,
|
| 1669 |
+
"step": 232
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 2.65,
|
| 1673 |
+
"grad_norm": 0.15251006186008453,
|
| 1674 |
+
"learning_rate": 6.540084388185655e-06,
|
| 1675 |
+
"loss": 0.0026,
|
| 1676 |
+
"step": 233
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 2.66,
|
| 1680 |
+
"grad_norm": 0.9377198219299316,
|
| 1681 |
+
"learning_rate": 6.329113924050633e-06,
|
| 1682 |
+
"loss": 0.0081,
|
| 1683 |
+
"step": 234
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 2.67,
|
| 1687 |
+
"grad_norm": 11.142963409423828,
|
| 1688 |
+
"learning_rate": 6.118143459915612e-06,
|
| 1689 |
+
"loss": 0.1642,
|
| 1690 |
+
"step": 235
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 2.68,
|
| 1694 |
+
"grad_norm": 31.128833770751953,
|
| 1695 |
+
"learning_rate": 5.907172995780591e-06,
|
| 1696 |
+
"loss": 0.3947,
|
| 1697 |
+
"step": 236
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 2.69,
|
| 1701 |
+
"grad_norm": 0.6137996315956116,
|
| 1702 |
+
"learning_rate": 5.69620253164557e-06,
|
| 1703 |
+
"loss": 0.0033,
|
| 1704 |
+
"step": 237
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 2.7,
|
| 1708 |
+
"grad_norm": 0.1146240383386612,
|
| 1709 |
+
"learning_rate": 5.485232067510549e-06,
|
| 1710 |
+
"loss": 0.0018,
|
| 1711 |
+
"step": 238
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 2.72,
|
| 1715 |
+
"grad_norm": 14.48742389678955,
|
| 1716 |
+
"learning_rate": 5.274261603375528e-06,
|
| 1717 |
+
"loss": 0.2166,
|
| 1718 |
+
"step": 239
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 2.73,
|
| 1722 |
+
"grad_norm": 1.7509044408798218,
|
| 1723 |
+
"learning_rate": 5.063291139240506e-06,
|
| 1724 |
+
"loss": 0.0174,
|
| 1725 |
+
"step": 240
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 2.74,
|
| 1729 |
+
"grad_norm": 0.0605483315885067,
|
| 1730 |
+
"learning_rate": 4.852320675105485e-06,
|
| 1731 |
+
"loss": 0.001,
|
| 1732 |
+
"step": 241
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 2.75,
|
| 1736 |
+
"grad_norm": 0.1251741498708725,
|
| 1737 |
+
"learning_rate": 4.641350210970465e-06,
|
| 1738 |
+
"loss": 0.0018,
|
| 1739 |
+
"step": 242
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 2.76,
|
| 1743 |
+
"grad_norm": 0.15386824309825897,
|
| 1744 |
+
"learning_rate": 4.430379746835443e-06,
|
| 1745 |
+
"loss": 0.0025,
|
| 1746 |
+
"step": 243
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 2.77,
|
| 1750 |
+
"grad_norm": 6.340598106384277,
|
| 1751 |
+
"learning_rate": 4.219409282700422e-06,
|
| 1752 |
+
"loss": 0.0342,
|
| 1753 |
+
"step": 244
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 2.78,
|
| 1757 |
+
"grad_norm": 0.18014036118984222,
|
| 1758 |
+
"learning_rate": 4.008438818565402e-06,
|
| 1759 |
+
"loss": 0.0022,
|
| 1760 |
+
"step": 245
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 2.8,
|
| 1764 |
+
"grad_norm": 17.62770652770996,
|
| 1765 |
+
"learning_rate": 3.7974683544303802e-06,
|
| 1766 |
+
"loss": 0.1143,
|
| 1767 |
+
"step": 246
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 2.81,
|
| 1771 |
+
"grad_norm": 0.17265360057353973,
|
| 1772 |
+
"learning_rate": 3.5864978902953588e-06,
|
| 1773 |
+
"loss": 0.0026,
|
| 1774 |
+
"step": 247
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 2.82,
|
| 1778 |
+
"grad_norm": 6.071741580963135,
|
| 1779 |
+
"learning_rate": 3.3755274261603373e-06,
|
| 1780 |
+
"loss": 0.0373,
|
| 1781 |
+
"step": 248
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 2.83,
|
| 1785 |
+
"grad_norm": 1.1394612789154053,
|
| 1786 |
+
"learning_rate": 3.1645569620253167e-06,
|
| 1787 |
+
"loss": 0.0098,
|
| 1788 |
+
"step": 249
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 2.84,
|
| 1792 |
+
"grad_norm": 0.09031672030687332,
|
| 1793 |
+
"learning_rate": 2.9535864978902956e-06,
|
| 1794 |
+
"loss": 0.0017,
|
| 1795 |
+
"step": 250
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 2.85,
|
| 1799 |
+
"grad_norm": 23.36667251586914,
|
| 1800 |
+
"learning_rate": 2.7426160337552745e-06,
|
| 1801 |
+
"loss": 0.7473,
|
| 1802 |
+
"step": 251
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 2.86,
|
| 1806 |
+
"grad_norm": 0.17166578769683838,
|
| 1807 |
+
"learning_rate": 2.531645569620253e-06,
|
| 1808 |
+
"loss": 0.0015,
|
| 1809 |
+
"step": 252
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 2.88,
|
| 1813 |
+
"grad_norm": 0.7906249165534973,
|
| 1814 |
+
"learning_rate": 2.3206751054852324e-06,
|
| 1815 |
+
"loss": 0.0071,
|
| 1816 |
+
"step": 253
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 2.89,
|
| 1820 |
+
"grad_norm": 13.26647663116455,
|
| 1821 |
+
"learning_rate": 2.109704641350211e-06,
|
| 1822 |
+
"loss": 0.1543,
|
| 1823 |
+
"step": 254
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 2.9,
|
| 1827 |
+
"grad_norm": 0.22246617078781128,
|
| 1828 |
+
"learning_rate": 1.8987341772151901e-06,
|
| 1829 |
+
"loss": 0.002,
|
| 1830 |
+
"step": 255
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 2.91,
|
| 1834 |
+
"grad_norm": 5.0251784324646,
|
| 1835 |
+
"learning_rate": 1.6877637130801686e-06,
|
| 1836 |
+
"loss": 0.0597,
|
| 1837 |
+
"step": 256
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 2.92,
|
| 1841 |
+
"grad_norm": 0.6754881143569946,
|
| 1842 |
+
"learning_rate": 1.4767932489451478e-06,
|
| 1843 |
+
"loss": 0.0062,
|
| 1844 |
+
"step": 257
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 2.93,
|
| 1848 |
+
"grad_norm": 0.03976951539516449,
|
| 1849 |
+
"learning_rate": 1.2658227848101265e-06,
|
| 1850 |
+
"loss": 0.0009,
|
| 1851 |
+
"step": 258
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 2.94,
|
| 1855 |
+
"grad_norm": 0.01055199932307005,
|
| 1856 |
+
"learning_rate": 1.0548523206751055e-06,
|
| 1857 |
+
"loss": 0.0003,
|
| 1858 |
+
"step": 259
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 2.95,
|
| 1862 |
+
"grad_norm": 20.83010482788086,
|
| 1863 |
+
"learning_rate": 8.438818565400843e-07,
|
| 1864 |
+
"loss": 0.371,
|
| 1865 |
+
"step": 260
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 2.97,
|
| 1869 |
+
"grad_norm": 0.4786561131477356,
|
| 1870 |
+
"learning_rate": 6.329113924050633e-07,
|
| 1871 |
+
"loss": 0.0055,
|
| 1872 |
+
"step": 261
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 2.98,
|
| 1876 |
+
"grad_norm": 11.4153413772583,
|
| 1877 |
+
"learning_rate": 4.2194092827004216e-07,
|
| 1878 |
+
"loss": 0.0949,
|
| 1879 |
+
"step": 262
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 2.99,
|
| 1883 |
+
"grad_norm": 0.1931201070547104,
|
| 1884 |
+
"learning_rate": 2.1097046413502108e-07,
|
| 1885 |
+
"loss": 0.0028,
|
| 1886 |
+
"step": 263
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 3.0,
|
| 1890 |
+
"grad_norm": 6.672695636749268,
|
| 1891 |
+
"learning_rate": 0.0,
|
| 1892 |
+
"loss": 0.0624,
|
| 1893 |
+
"step": 264
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 3.0,
|
| 1897 |
+
"eval_accuracy": 1.0,
|
| 1898 |
+
"eval_f1_macro": 1.0,
|
| 1899 |
+
"eval_f1_micro": 1.0,
|
| 1900 |
+
"eval_f1_weighted": 1.0,
|
| 1901 |
+
"eval_loss": 0.011418873444199562,
|
| 1902 |
+
"eval_precision_macro": 1.0,
|
| 1903 |
+
"eval_precision_micro": 1.0,
|
| 1904 |
+
"eval_precision_weighted": 1.0,
|
| 1905 |
+
"eval_recall_macro": 1.0,
|
| 1906 |
+
"eval_recall_micro": 1.0,
|
| 1907 |
+
"eval_recall_weighted": 1.0,
|
| 1908 |
+
"eval_runtime": 23.0544,
|
| 1909 |
+
"eval_samples_per_second": 3.34,
|
| 1910 |
+
"eval_steps_per_second": 0.217,
|
| 1911 |
+
"step": 264
|
| 1912 |
+
}
|
| 1913 |
+
],
|
| 1914 |
+
"logging_steps": 1,
|
| 1915 |
+
"max_steps": 264,
|
| 1916 |
+
"num_input_tokens_seen": 0,
|
| 1917 |
+
"num_train_epochs": 3,
|
| 1918 |
+
"save_steps": 500,
|
| 1919 |
+
"total_flos": 1.626709167691776e+17,
|
| 1920 |
+
"train_batch_size": 8,
|
| 1921 |
+
"trial_name": null,
|
| 1922 |
+
"trial_params": null
|
| 1923 |
+
}
|
checkpoint-264/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:90ab5e32042ddb9b9e8492598d406c290602192b20e03add7b080f031e40d9e2
|
| 3 |
+
size 4920
|
config.json
ADDED
|
@@ -0,0 +1,84 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "microsoft/beit-base-patch16-224-pt22k-ft22k",
|
| 3 |
+
"_num_labels": 7,
|
| 4 |
+
"add_fpn": false,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"BeitForImageClassification"
|
| 7 |
+
],
|
| 8 |
+
"attention_probs_dropout_prob": 0.0,
|
| 9 |
+
"auxiliary_channels": 256,
|
| 10 |
+
"auxiliary_concat_input": false,
|
| 11 |
+
"auxiliary_loss_weight": 0.4,
|
| 12 |
+
"auxiliary_num_convs": 1,
|
| 13 |
+
"drop_path_rate": 0.1,
|
| 14 |
+
"hidden_act": "gelu",
|
| 15 |
+
"hidden_dropout_prob": 0.0,
|
| 16 |
+
"hidden_size": 768,
|
| 17 |
+
"id2label": {
|
| 18 |
+
"0": "banana_healthy_leaf",
|
| 19 |
+
"1": "black_sigatoka",
|
| 20 |
+
"2": "bract_mosaic_virus",
|
| 21 |
+
"3": "insect_pest",
|
| 22 |
+
"4": "moko_disease",
|
| 23 |
+
"5": "panama_disease",
|
| 24 |
+
"6": "yellow_sigatoka"
|
| 25 |
+
},
|
| 26 |
+
"image_size": 224,
|
| 27 |
+
"initializer_range": 0.02,
|
| 28 |
+
"intermediate_size": 3072,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"banana_healthy_leaf": 0,
|
| 31 |
+
"black_sigatoka": 1,
|
| 32 |
+
"bract_mosaic_virus": 2,
|
| 33 |
+
"insect_pest": 3,
|
| 34 |
+
"moko_disease": 4,
|
| 35 |
+
"panama_disease": 5,
|
| 36 |
+
"yellow_sigatoka": 6
|
| 37 |
+
},
|
| 38 |
+
"layer_norm_eps": 1e-12,
|
| 39 |
+
"layer_scale_init_value": 0.1,
|
| 40 |
+
"model_type": "beit",
|
| 41 |
+
"num_attention_heads": 12,
|
| 42 |
+
"num_channels": 3,
|
| 43 |
+
"num_hidden_layers": 12,
|
| 44 |
+
"out_features": [
|
| 45 |
+
"stage12"
|
| 46 |
+
],
|
| 47 |
+
"out_indices": [
|
| 48 |
+
12
|
| 49 |
+
],
|
| 50 |
+
"patch_size": 16,
|
| 51 |
+
"pool_scales": [
|
| 52 |
+
1,
|
| 53 |
+
2,
|
| 54 |
+
3,
|
| 55 |
+
6
|
| 56 |
+
],
|
| 57 |
+
"problem_type": "single_label_classification",
|
| 58 |
+
"reshape_hidden_states": true,
|
| 59 |
+
"semantic_loss_ignore_index": 255,
|
| 60 |
+
"stage_names": [
|
| 61 |
+
"stem",
|
| 62 |
+
"stage1",
|
| 63 |
+
"stage2",
|
| 64 |
+
"stage3",
|
| 65 |
+
"stage4",
|
| 66 |
+
"stage5",
|
| 67 |
+
"stage6",
|
| 68 |
+
"stage7",
|
| 69 |
+
"stage8",
|
| 70 |
+
"stage9",
|
| 71 |
+
"stage10",
|
| 72 |
+
"stage11",
|
| 73 |
+
"stage12"
|
| 74 |
+
],
|
| 75 |
+
"torch_dtype": "float32",
|
| 76 |
+
"transformers_version": "4.39.0",
|
| 77 |
+
"use_absolute_position_embeddings": false,
|
| 78 |
+
"use_auxiliary_head": true,
|
| 79 |
+
"use_mask_token": false,
|
| 80 |
+
"use_mean_pooling": true,
|
| 81 |
+
"use_relative_position_bias": true,
|
| 82 |
+
"use_shared_relative_position_bias": false,
|
| 83 |
+
"vocab_size": 8192
|
| 84 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0b35f130963ec8b66044115b2000e40353aef8d421ed5242fb32a6954dc33fec
|
| 3 |
+
size 343095708
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_valid_processor_keys": [
|
| 3 |
+
"images",
|
| 4 |
+
"segmentation_maps",
|
| 5 |
+
"do_resize",
|
| 6 |
+
"size",
|
| 7 |
+
"resample",
|
| 8 |
+
"do_center_crop",
|
| 9 |
+
"crop_size",
|
| 10 |
+
"do_rescale",
|
| 11 |
+
"rescale_factor",
|
| 12 |
+
"do_normalize",
|
| 13 |
+
"image_mean",
|
| 14 |
+
"image_std",
|
| 15 |
+
"do_reduce_labels",
|
| 16 |
+
"return_tensors",
|
| 17 |
+
"data_format",
|
| 18 |
+
"input_data_format"
|
| 19 |
+
],
|
| 20 |
+
"crop_size": {
|
| 21 |
+
"height": 224,
|
| 22 |
+
"width": 224
|
| 23 |
+
},
|
| 24 |
+
"do_center_crop": false,
|
| 25 |
+
"do_normalize": true,
|
| 26 |
+
"do_reduce_labels": false,
|
| 27 |
+
"do_rescale": true,
|
| 28 |
+
"do_resize": true,
|
| 29 |
+
"image_mean": [
|
| 30 |
+
0.5,
|
| 31 |
+
0.5,
|
| 32 |
+
0.5
|
| 33 |
+
],
|
| 34 |
+
"image_processor_type": "BeitImageProcessor",
|
| 35 |
+
"image_std": [
|
| 36 |
+
0.5,
|
| 37 |
+
0.5,
|
| 38 |
+
0.5
|
| 39 |
+
],
|
| 40 |
+
"resample": 2,
|
| 41 |
+
"rescale_factor": 0.00392156862745098,
|
| 42 |
+
"size": {
|
| 43 |
+
"height": 224,
|
| 44 |
+
"width": 224
|
| 45 |
+
}
|
| 46 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:90ab5e32042ddb9b9e8492598d406c290602192b20e03add7b080f031e40d9e2
|
| 3 |
+
size 4920
|
training_params.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "as-cle-bert/banana-disease-classification",
|
| 3 |
+
"model": "microsoft/beit-base-patch16-224-pt22k-ft22k",
|
| 4 |
+
"username": "as-cle-bert",
|
| 5 |
+
"lr": 5e-05,
|
| 6 |
+
"epochs": 3,
|
| 7 |
+
"batch_size": 8,
|
| 8 |
+
"warmup_ratio": 0.1,
|
| 9 |
+
"gradient_accumulation": 1,
|
| 10 |
+
"optimizer": "adamw_torch",
|
| 11 |
+
"scheduler": "linear",
|
| 12 |
+
"weight_decay": 0.0,
|
| 13 |
+
"max_grad_norm": 1.0,
|
| 14 |
+
"seed": 42,
|
| 15 |
+
"train_split": "train",
|
| 16 |
+
"valid_split": "test",
|
| 17 |
+
"logging_steps": -1,
|
| 18 |
+
"project_name": "beit-banana-diseases",
|
| 19 |
+
"auto_find_batch_size": false,
|
| 20 |
+
"mixed_precision": "no",
|
| 21 |
+
"save_total_limit": 1,
|
| 22 |
+
"save_strategy": "epoch",
|
| 23 |
+
"push_to_hub": true,
|
| 24 |
+
"repo_id": "as-cle-bert/beit-banana-diseases",
|
| 25 |
+
"evaluation_strategy": "epoch",
|
| 26 |
+
"image_column": "image",
|
| 27 |
+
"target_column": "label",
|
| 28 |
+
"log": "none"
|
| 29 |
+
}
|