Upload folder using huggingface_hub
Browse files- README.md +48 -0
- checkpoint-4950/config.json +46 -0
- checkpoint-4950/model.safetensors +3 -0
- checkpoint-4950/optimizer.pt +3 -0
- checkpoint-4950/preprocessor_config.json +25 -0
- checkpoint-4950/rng_state.pth +3 -0
- checkpoint-4950/scheduler.pt +3 -0
- checkpoint-4950/trainer_state.json +1558 -0
- checkpoint-4950/training_args.bin +3 -0
- config.json +46 -0
- model.safetensors +3 -0
- preprocessor_config.json +25 -0
- runs/Nov19_07-08-55_659d872c1556/events.out.tfevents.1732000135.659d872c1556.241.0 +2 -2
- runs/Nov19_07-08-55_659d872c1556/events.out.tfevents.1732008067.659d872c1556.241.1 +3 -0
- training_args.bin +3 -0
- training_params.json +30 -0
README.md
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
---
|
| 3 |
+
tags:
|
| 4 |
+
- autotrain
|
| 5 |
+
- object-detection
|
| 6 |
+
- vision
|
| 7 |
+
base_model: hustvl/yolos-small
|
| 8 |
+
widget:
|
| 9 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/tiger.jpg
|
| 10 |
+
example_title: Tiger
|
| 11 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/teapot.jpg
|
| 12 |
+
example_title: Teapot
|
| 13 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/palace.jpg
|
| 14 |
+
example_title: Palace
|
| 15 |
+
datasets:
|
| 16 |
+
- Francesco/brain-tumor-m2pbp
|
| 17 |
+
---
|
| 18 |
+
|
| 19 |
+
# Model Trained Using AutoTrain
|
| 20 |
+
|
| 21 |
+
- Problem type: Object Detection
|
| 22 |
+
|
| 23 |
+
## Validation Metrics
|
| 24 |
+
loss: 1.1927123069763184
|
| 25 |
+
|
| 26 |
+
map: 0.2537
|
| 27 |
+
|
| 28 |
+
map_50: 0.502
|
| 29 |
+
|
| 30 |
+
map_75: 0.2303
|
| 31 |
+
|
| 32 |
+
map_small: 0.0016
|
| 33 |
+
|
| 34 |
+
map_medium: 0.1325
|
| 35 |
+
|
| 36 |
+
map_large: 0.3796
|
| 37 |
+
|
| 38 |
+
mar_1: 0.3388
|
| 39 |
+
|
| 40 |
+
mar_10: 0.4646
|
| 41 |
+
|
| 42 |
+
mar_100: 0.5012
|
| 43 |
+
|
| 44 |
+
mar_small: 0.0456
|
| 45 |
+
|
| 46 |
+
mar_medium: 0.4027
|
| 47 |
+
|
| 48 |
+
mar_large: 0.678
|
checkpoint-4950/config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "hustvl/yolos-small",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"YolosForObjectDetection"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"auxiliary_loss": false,
|
| 8 |
+
"bbox_cost": 5,
|
| 9 |
+
"bbox_loss_coefficient": 5,
|
| 10 |
+
"class_cost": 1,
|
| 11 |
+
"eos_coefficient": 0.1,
|
| 12 |
+
"giou_cost": 2,
|
| 13 |
+
"giou_loss_coefficient": 2,
|
| 14 |
+
"hidden_act": "gelu",
|
| 15 |
+
"hidden_dropout_prob": 0.0,
|
| 16 |
+
"hidden_size": 384,
|
| 17 |
+
"id2label": {
|
| 18 |
+
"0": "brain-tumor",
|
| 19 |
+
"1": "label0",
|
| 20 |
+
"2": "label1",
|
| 21 |
+
"3": "label2"
|
| 22 |
+
},
|
| 23 |
+
"image_size": [
|
| 24 |
+
512,
|
| 25 |
+
864
|
| 26 |
+
],
|
| 27 |
+
"initializer_range": 0.02,
|
| 28 |
+
"intermediate_size": 1536,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"brain-tumor": 0,
|
| 31 |
+
"label0": 1,
|
| 32 |
+
"label1": 2,
|
| 33 |
+
"label2": 3
|
| 34 |
+
},
|
| 35 |
+
"layer_norm_eps": 1e-12,
|
| 36 |
+
"model_type": "yolos",
|
| 37 |
+
"num_attention_heads": 6,
|
| 38 |
+
"num_channels": 3,
|
| 39 |
+
"num_detection_tokens": 100,
|
| 40 |
+
"num_hidden_layers": 12,
|
| 41 |
+
"patch_size": 16,
|
| 42 |
+
"qkv_bias": true,
|
| 43 |
+
"torch_dtype": "float32",
|
| 44 |
+
"transformers_version": "4.46.2",
|
| 45 |
+
"use_mid_position_embeddings": true
|
| 46 |
+
}
|
checkpoint-4950/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:91f9a40cd583f7353f09a374ff308a89d686aef6c0682d2edb9a5613668f6ae4
|
| 3 |
+
size 122629292
|
checkpoint-4950/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:03deecaeaff4b76eee0106b2f956c81cbec2d418a4791edc620fe4e234838a8e
|
| 3 |
+
size 245387514
|
checkpoint-4950/preprocessor_config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_annotations": true,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_pad": false,
|
| 5 |
+
"do_rescale": true,
|
| 6 |
+
"do_resize": false,
|
| 7 |
+
"format": "coco_detection",
|
| 8 |
+
"image_mean": [
|
| 9 |
+
0.485,
|
| 10 |
+
0.456,
|
| 11 |
+
0.406
|
| 12 |
+
],
|
| 13 |
+
"image_processor_type": "YolosImageProcessor",
|
| 14 |
+
"image_std": [
|
| 15 |
+
0.229,
|
| 16 |
+
0.224,
|
| 17 |
+
0.225
|
| 18 |
+
],
|
| 19 |
+
"pad_size": null,
|
| 20 |
+
"resample": 2,
|
| 21 |
+
"rescale_factor": 0.00392156862745098,
|
| 22 |
+
"size": {
|
| 23 |
+
"longest_edge": 640
|
| 24 |
+
}
|
| 25 |
+
}
|
checkpoint-4950/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d245e05e72192c132e0f2edb6fdcae0c578c890f0fe912f17ec7b0bba2d38cc3
|
| 3 |
+
size 14244
|
checkpoint-4950/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:21f017fc19796a7674cd02712e672332ba3015a9a7f7596919ebd03e98ecf2d3
|
| 3 |
+
size 1064
|
checkpoint-4950/trainer_state.json
ADDED
|
@@ -0,0 +1,1558 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.1927123069763184,
|
| 3 |
+
"best_model_checkpoint": "tumor-detect-yolo-small/checkpoint-4950",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4950,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.025252525252525252,
|
| 13 |
+
"grad_norm": 215.87857055664062,
|
| 14 |
+
"learning_rate": 2.2222222222222225e-06,
|
| 15 |
+
"loss": 3.8856,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.050505050505050504,
|
| 20 |
+
"grad_norm": 184.14230346679688,
|
| 21 |
+
"learning_rate": 4.747474747474747e-06,
|
| 22 |
+
"loss": 3.4913,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.07575757575757576,
|
| 27 |
+
"grad_norm": 73.90482330322266,
|
| 28 |
+
"learning_rate": 7.272727272727272e-06,
|
| 29 |
+
"loss": 2.8902,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.10101010101010101,
|
| 34 |
+
"grad_norm": 40.390384674072266,
|
| 35 |
+
"learning_rate": 9.7979797979798e-06,
|
| 36 |
+
"loss": 2.4434,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.12626262626262627,
|
| 41 |
+
"grad_norm": 42.501609802246094,
|
| 42 |
+
"learning_rate": 1.2323232323232325e-05,
|
| 43 |
+
"loss": 2.1762,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.15151515151515152,
|
| 48 |
+
"grad_norm": 56.84285354614258,
|
| 49 |
+
"learning_rate": 1.484848484848485e-05,
|
| 50 |
+
"loss": 1.9959,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.17676767676767677,
|
| 55 |
+
"grad_norm": 24.612760543823242,
|
| 56 |
+
"learning_rate": 1.7373737373737375e-05,
|
| 57 |
+
"loss": 1.9997,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.20202020202020202,
|
| 62 |
+
"grad_norm": 28.1953067779541,
|
| 63 |
+
"learning_rate": 1.98989898989899e-05,
|
| 64 |
+
"loss": 1.9539,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.22727272727272727,
|
| 69 |
+
"grad_norm": 27.381372451782227,
|
| 70 |
+
"learning_rate": 2.2323232323232324e-05,
|
| 71 |
+
"loss": 1.9312,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.25252525252525254,
|
| 76 |
+
"grad_norm": 29.632625579833984,
|
| 77 |
+
"learning_rate": 2.4848484848484847e-05,
|
| 78 |
+
"loss": 1.89,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.2777777777777778,
|
| 83 |
+
"grad_norm": 39.85140609741211,
|
| 84 |
+
"learning_rate": 2.7373737373737374e-05,
|
| 85 |
+
"loss": 1.8177,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.30303030303030304,
|
| 90 |
+
"grad_norm": 30.56757354736328,
|
| 91 |
+
"learning_rate": 2.98989898989899e-05,
|
| 92 |
+
"loss": 1.789,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.3282828282828283,
|
| 97 |
+
"grad_norm": 46.40056610107422,
|
| 98 |
+
"learning_rate": 3.2424242424242423e-05,
|
| 99 |
+
"loss": 1.7855,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.35353535353535354,
|
| 104 |
+
"grad_norm": 27.508628845214844,
|
| 105 |
+
"learning_rate": 3.494949494949495e-05,
|
| 106 |
+
"loss": 1.6843,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.3787878787878788,
|
| 111 |
+
"grad_norm": 55.3870964050293,
|
| 112 |
+
"learning_rate": 3.747474747474748e-05,
|
| 113 |
+
"loss": 1.7921,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.40404040404040403,
|
| 118 |
+
"grad_norm": 24.95541000366211,
|
| 119 |
+
"learning_rate": 4e-05,
|
| 120 |
+
"loss": 1.78,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.4292929292929293,
|
| 125 |
+
"grad_norm": 28.33966636657715,
|
| 126 |
+
"learning_rate": 4.252525252525253e-05,
|
| 127 |
+
"loss": 1.6173,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.45454545454545453,
|
| 132 |
+
"grad_norm": 27.43466567993164,
|
| 133 |
+
"learning_rate": 4.5050505050505056e-05,
|
| 134 |
+
"loss": 1.7889,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.4797979797979798,
|
| 139 |
+
"grad_norm": 22.498807907104492,
|
| 140 |
+
"learning_rate": 4.7575757575757576e-05,
|
| 141 |
+
"loss": 1.8349,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.5050505050505051,
|
| 146 |
+
"grad_norm": 30.1269474029541,
|
| 147 |
+
"learning_rate": 4.998877665544332e-05,
|
| 148 |
+
"loss": 1.6779,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.5303030303030303,
|
| 153 |
+
"grad_norm": 30.818307876586914,
|
| 154 |
+
"learning_rate": 4.970819304152638e-05,
|
| 155 |
+
"loss": 1.7294,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.5555555555555556,
|
| 160 |
+
"grad_norm": 25.11827850341797,
|
| 161 |
+
"learning_rate": 4.942760942760943e-05,
|
| 162 |
+
"loss": 1.7098,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.5808080808080808,
|
| 167 |
+
"grad_norm": 38.30568313598633,
|
| 168 |
+
"learning_rate": 4.914702581369248e-05,
|
| 169 |
+
"loss": 1.5827,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.6060606060606061,
|
| 174 |
+
"grad_norm": 42.69370651245117,
|
| 175 |
+
"learning_rate": 4.886644219977554e-05,
|
| 176 |
+
"loss": 1.6928,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.6313131313131313,
|
| 181 |
+
"grad_norm": 12.195549011230469,
|
| 182 |
+
"learning_rate": 4.858585858585859e-05,
|
| 183 |
+
"loss": 1.7731,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.6565656565656566,
|
| 188 |
+
"grad_norm": 14.210679054260254,
|
| 189 |
+
"learning_rate": 4.830527497194164e-05,
|
| 190 |
+
"loss": 1.6602,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.6818181818181818,
|
| 195 |
+
"grad_norm": 37.4093132019043,
|
| 196 |
+
"learning_rate": 4.8024691358024696e-05,
|
| 197 |
+
"loss": 1.6286,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.7070707070707071,
|
| 202 |
+
"grad_norm": 21.867361068725586,
|
| 203 |
+
"learning_rate": 4.7744107744107746e-05,
|
| 204 |
+
"loss": 1.6603,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.7323232323232324,
|
| 209 |
+
"grad_norm": 25.663557052612305,
|
| 210 |
+
"learning_rate": 4.74635241301908e-05,
|
| 211 |
+
"loss": 1.6566,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.7575757575757576,
|
| 216 |
+
"grad_norm": 27.501630783081055,
|
| 217 |
+
"learning_rate": 4.718294051627385e-05,
|
| 218 |
+
"loss": 1.6502,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.7828282828282829,
|
| 223 |
+
"grad_norm": 35.044532775878906,
|
| 224 |
+
"learning_rate": 4.691358024691358e-05,
|
| 225 |
+
"loss": 1.5283,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.8080808080808081,
|
| 230 |
+
"grad_norm": 38.075992584228516,
|
| 231 |
+
"learning_rate": 4.6632996632996634e-05,
|
| 232 |
+
"loss": 1.5263,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.8333333333333334,
|
| 237 |
+
"grad_norm": 19.746585845947266,
|
| 238 |
+
"learning_rate": 4.635241301907969e-05,
|
| 239 |
+
"loss": 1.6239,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.8585858585858586,
|
| 244 |
+
"grad_norm": 33.95107650756836,
|
| 245 |
+
"learning_rate": 4.607182940516274e-05,
|
| 246 |
+
"loss": 1.6435,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.8838383838383839,
|
| 251 |
+
"grad_norm": 35.092159271240234,
|
| 252 |
+
"learning_rate": 4.57912457912458e-05,
|
| 253 |
+
"loss": 1.5883,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.9090909090909091,
|
| 258 |
+
"grad_norm": 22.830068588256836,
|
| 259 |
+
"learning_rate": 4.551066217732884e-05,
|
| 260 |
+
"loss": 1.5553,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.9343434343434344,
|
| 265 |
+
"grad_norm": 34.29344177246094,
|
| 266 |
+
"learning_rate": 4.52300785634119e-05,
|
| 267 |
+
"loss": 1.5197,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.9595959595959596,
|
| 272 |
+
"grad_norm": 59.63193893432617,
|
| 273 |
+
"learning_rate": 4.494949494949495e-05,
|
| 274 |
+
"loss": 1.5119,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.9848484848484849,
|
| 279 |
+
"grad_norm": 22.16474151611328,
|
| 280 |
+
"learning_rate": 4.4668911335578006e-05,
|
| 281 |
+
"loss": 1.6414,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.0,
|
| 286 |
+
"eval_loss": 1.5094339847564697,
|
| 287 |
+
"eval_map": 0.1139,
|
| 288 |
+
"eval_map_50": 0.2748,
|
| 289 |
+
"eval_map_75": 0.0747,
|
| 290 |
+
"eval_map_label0": 0.0376,
|
| 291 |
+
"eval_map_label1": 0.2643,
|
| 292 |
+
"eval_map_label2": 0.0398,
|
| 293 |
+
"eval_map_large": 0.1716,
|
| 294 |
+
"eval_map_medium": 0.0391,
|
| 295 |
+
"eval_map_small": 0.0013,
|
| 296 |
+
"eval_mar_1": 0.1849,
|
| 297 |
+
"eval_mar_10": 0.3107,
|
| 298 |
+
"eval_mar_100": 0.3439,
|
| 299 |
+
"eval_mar_100_label0": 0.366,
|
| 300 |
+
"eval_mar_100_label1": 0.4665,
|
| 301 |
+
"eval_mar_100_label2": 0.1993,
|
| 302 |
+
"eval_mar_large": 0.4817,
|
| 303 |
+
"eval_mar_medium": 0.249,
|
| 304 |
+
"eval_mar_small": 0.0139,
|
| 305 |
+
"eval_runtime": 179.4201,
|
| 306 |
+
"eval_samples_per_second": 11.036,
|
| 307 |
+
"eval_steps_per_second": 0.691,
|
| 308 |
+
"step": 990
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.0101010101010102,
|
| 312 |
+
"grad_norm": 15.366436958312988,
|
| 313 |
+
"learning_rate": 4.438832772166106e-05,
|
| 314 |
+
"loss": 1.5708,
|
| 315 |
+
"step": 1000
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 1.0353535353535352,
|
| 319 |
+
"grad_norm": 29.506799697875977,
|
| 320 |
+
"learning_rate": 4.410774410774411e-05,
|
| 321 |
+
"loss": 1.56,
|
| 322 |
+
"step": 1025
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 1.0606060606060606,
|
| 326 |
+
"grad_norm": 18.874906539916992,
|
| 327 |
+
"learning_rate": 4.3827160493827164e-05,
|
| 328 |
+
"loss": 1.5524,
|
| 329 |
+
"step": 1050
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.0858585858585859,
|
| 333 |
+
"grad_norm": 21.787818908691406,
|
| 334 |
+
"learning_rate": 4.3546576879910214e-05,
|
| 335 |
+
"loss": 1.5174,
|
| 336 |
+
"step": 1075
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.1111111111111112,
|
| 340 |
+
"grad_norm": 20.070924758911133,
|
| 341 |
+
"learning_rate": 4.3265993265993265e-05,
|
| 342 |
+
"loss": 1.536,
|
| 343 |
+
"step": 1100
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 1.1363636363636362,
|
| 347 |
+
"grad_norm": 39.67344284057617,
|
| 348 |
+
"learning_rate": 4.298540965207632e-05,
|
| 349 |
+
"loss": 1.6004,
|
| 350 |
+
"step": 1125
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.1616161616161615,
|
| 354 |
+
"grad_norm": 22.05341148376465,
|
| 355 |
+
"learning_rate": 4.270482603815937e-05,
|
| 356 |
+
"loss": 1.5341,
|
| 357 |
+
"step": 1150
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 1.1868686868686869,
|
| 361 |
+
"grad_norm": 18.423086166381836,
|
| 362 |
+
"learning_rate": 4.242424242424243e-05,
|
| 363 |
+
"loss": 1.5609,
|
| 364 |
+
"step": 1175
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 1.2121212121212122,
|
| 368 |
+
"grad_norm": 23.607410430908203,
|
| 369 |
+
"learning_rate": 4.214365881032548e-05,
|
| 370 |
+
"loss": 1.5443,
|
| 371 |
+
"step": 1200
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 1.2373737373737375,
|
| 375 |
+
"grad_norm": 26.15247917175293,
|
| 376 |
+
"learning_rate": 4.186307519640853e-05,
|
| 377 |
+
"loss": 1.4498,
|
| 378 |
+
"step": 1225
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 1.2626262626262625,
|
| 382 |
+
"grad_norm": 18.705833435058594,
|
| 383 |
+
"learning_rate": 4.158249158249159e-05,
|
| 384 |
+
"loss": 1.4653,
|
| 385 |
+
"step": 1250
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 1.2878787878787878,
|
| 389 |
+
"grad_norm": 23.05366325378418,
|
| 390 |
+
"learning_rate": 4.130190796857464e-05,
|
| 391 |
+
"loss": 1.4813,
|
| 392 |
+
"step": 1275
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 1.3131313131313131,
|
| 396 |
+
"grad_norm": 17.575767517089844,
|
| 397 |
+
"learning_rate": 4.1021324354657695e-05,
|
| 398 |
+
"loss": 1.4978,
|
| 399 |
+
"step": 1300
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.3383838383838385,
|
| 403 |
+
"grad_norm": 15.836087226867676,
|
| 404 |
+
"learning_rate": 4.074074074074074e-05,
|
| 405 |
+
"loss": 1.5237,
|
| 406 |
+
"step": 1325
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 1.3636363636363638,
|
| 410 |
+
"grad_norm": 18.104198455810547,
|
| 411 |
+
"learning_rate": 4.0460157126823795e-05,
|
| 412 |
+
"loss": 1.5597,
|
| 413 |
+
"step": 1350
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.3888888888888888,
|
| 417 |
+
"grad_norm": 27.618770599365234,
|
| 418 |
+
"learning_rate": 4.0179573512906846e-05,
|
| 419 |
+
"loss": 1.5139,
|
| 420 |
+
"step": 1375
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.4141414141414141,
|
| 424 |
+
"grad_norm": 12.446822166442871,
|
| 425 |
+
"learning_rate": 3.98989898989899e-05,
|
| 426 |
+
"loss": 1.557,
|
| 427 |
+
"step": 1400
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 1.4393939393939394,
|
| 431 |
+
"grad_norm": 32.567569732666016,
|
| 432 |
+
"learning_rate": 3.961840628507295e-05,
|
| 433 |
+
"loss": 1.4802,
|
| 434 |
+
"step": 1425
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.4646464646464645,
|
| 438 |
+
"grad_norm": 20.30423355102539,
|
| 439 |
+
"learning_rate": 3.9337822671156003e-05,
|
| 440 |
+
"loss": 1.4664,
|
| 441 |
+
"step": 1450
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.4898989898989898,
|
| 445 |
+
"grad_norm": 23.72423553466797,
|
| 446 |
+
"learning_rate": 3.905723905723906e-05,
|
| 447 |
+
"loss": 1.4951,
|
| 448 |
+
"step": 1475
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 1.5151515151515151,
|
| 452 |
+
"grad_norm": 16.151382446289062,
|
| 453 |
+
"learning_rate": 3.877665544332211e-05,
|
| 454 |
+
"loss": 1.4668,
|
| 455 |
+
"step": 1500
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 1.5404040404040404,
|
| 459 |
+
"grad_norm": 15.978365898132324,
|
| 460 |
+
"learning_rate": 3.849607182940517e-05,
|
| 461 |
+
"loss": 1.5443,
|
| 462 |
+
"step": 1525
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 1.5656565656565657,
|
| 466 |
+
"grad_norm": 29.589649200439453,
|
| 467 |
+
"learning_rate": 3.821548821548822e-05,
|
| 468 |
+
"loss": 1.4429,
|
| 469 |
+
"step": 1550
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 1.5909090909090908,
|
| 473 |
+
"grad_norm": 29.138826370239258,
|
| 474 |
+
"learning_rate": 3.793490460157127e-05,
|
| 475 |
+
"loss": 1.4719,
|
| 476 |
+
"step": 1575
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 1.6161616161616161,
|
| 480 |
+
"grad_norm": 14.22740364074707,
|
| 481 |
+
"learning_rate": 3.7654320987654326e-05,
|
| 482 |
+
"loss": 1.4467,
|
| 483 |
+
"step": 1600
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 1.6414141414141414,
|
| 487 |
+
"grad_norm": 13.720157623291016,
|
| 488 |
+
"learning_rate": 3.7373737373737376e-05,
|
| 489 |
+
"loss": 1.5102,
|
| 490 |
+
"step": 1625
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 1.6666666666666665,
|
| 494 |
+
"grad_norm": 17.502704620361328,
|
| 495 |
+
"learning_rate": 3.7093153759820426e-05,
|
| 496 |
+
"loss": 1.4633,
|
| 497 |
+
"step": 1650
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 1.691919191919192,
|
| 501 |
+
"grad_norm": 15.681164741516113,
|
| 502 |
+
"learning_rate": 3.681257014590348e-05,
|
| 503 |
+
"loss": 1.4519,
|
| 504 |
+
"step": 1675
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 1.7171717171717171,
|
| 508 |
+
"grad_norm": 44.3072624206543,
|
| 509 |
+
"learning_rate": 3.6531986531986534e-05,
|
| 510 |
+
"loss": 1.4892,
|
| 511 |
+
"step": 1700
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 1.7424242424242424,
|
| 515 |
+
"grad_norm": 22.672714233398438,
|
| 516 |
+
"learning_rate": 3.625140291806959e-05,
|
| 517 |
+
"loss": 1.4918,
|
| 518 |
+
"step": 1725
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 1.7676767676767677,
|
| 522 |
+
"grad_norm": 29.78325843811035,
|
| 523 |
+
"learning_rate": 3.5970819304152635e-05,
|
| 524 |
+
"loss": 1.4672,
|
| 525 |
+
"step": 1750
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 1.7929292929292928,
|
| 529 |
+
"grad_norm": 23.5821590423584,
|
| 530 |
+
"learning_rate": 3.569023569023569e-05,
|
| 531 |
+
"loss": 1.4292,
|
| 532 |
+
"step": 1775
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 1.8181818181818183,
|
| 536 |
+
"grad_norm": 28.285043716430664,
|
| 537 |
+
"learning_rate": 3.540965207631874e-05,
|
| 538 |
+
"loss": 1.4431,
|
| 539 |
+
"step": 1800
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 1.8434343434343434,
|
| 543 |
+
"grad_norm": 29.603483200073242,
|
| 544 |
+
"learning_rate": 3.51290684624018e-05,
|
| 545 |
+
"loss": 1.482,
|
| 546 |
+
"step": 1825
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 1.8686868686868687,
|
| 550 |
+
"grad_norm": 20.012603759765625,
|
| 551 |
+
"learning_rate": 3.484848484848485e-05,
|
| 552 |
+
"loss": 1.5469,
|
| 553 |
+
"step": 1850
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 1.893939393939394,
|
| 557 |
+
"grad_norm": 21.001644134521484,
|
| 558 |
+
"learning_rate": 3.45679012345679e-05,
|
| 559 |
+
"loss": 1.5622,
|
| 560 |
+
"step": 1875
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.9191919191919191,
|
| 564 |
+
"grad_norm": 21.10232925415039,
|
| 565 |
+
"learning_rate": 3.428731762065096e-05,
|
| 566 |
+
"loss": 1.3952,
|
| 567 |
+
"step": 1900
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.9444444444444444,
|
| 571 |
+
"grad_norm": 14.726667404174805,
|
| 572 |
+
"learning_rate": 3.400673400673401e-05,
|
| 573 |
+
"loss": 1.3896,
|
| 574 |
+
"step": 1925
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 1.9696969696969697,
|
| 578 |
+
"grad_norm": 25.8332576751709,
|
| 579 |
+
"learning_rate": 3.3726150392817064e-05,
|
| 580 |
+
"loss": 1.4457,
|
| 581 |
+
"step": 1950
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 1.9949494949494948,
|
| 585 |
+
"grad_norm": 13.405915260314941,
|
| 586 |
+
"learning_rate": 3.3445566778900115e-05,
|
| 587 |
+
"loss": 1.4202,
|
| 588 |
+
"step": 1975
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 2.0,
|
| 592 |
+
"eval_loss": 1.396142601966858,
|
| 593 |
+
"eval_map": 0.1585,
|
| 594 |
+
"eval_map_50": 0.3518,
|
| 595 |
+
"eval_map_75": 0.1208,
|
| 596 |
+
"eval_map_label0": 0.0454,
|
| 597 |
+
"eval_map_label1": 0.3192,
|
| 598 |
+
"eval_map_label2": 0.1108,
|
| 599 |
+
"eval_map_large": 0.2453,
|
| 600 |
+
"eval_map_medium": 0.0656,
|
| 601 |
+
"eval_map_small": 0.001,
|
| 602 |
+
"eval_mar_1": 0.2464,
|
| 603 |
+
"eval_mar_10": 0.3794,
|
| 604 |
+
"eval_mar_100": 0.4095,
|
| 605 |
+
"eval_mar_100_label0": 0.4123,
|
| 606 |
+
"eval_mar_100_label1": 0.4825,
|
| 607 |
+
"eval_mar_100_label2": 0.3339,
|
| 608 |
+
"eval_mar_large": 0.5798,
|
| 609 |
+
"eval_mar_medium": 0.3002,
|
| 610 |
+
"eval_mar_small": 0.0202,
|
| 611 |
+
"eval_runtime": 129.5946,
|
| 612 |
+
"eval_samples_per_second": 15.278,
|
| 613 |
+
"eval_steps_per_second": 0.957,
|
| 614 |
+
"step": 1980
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.0202020202020203,
|
| 618 |
+
"grad_norm": 17.959142684936523,
|
| 619 |
+
"learning_rate": 3.3164983164983165e-05,
|
| 620 |
+
"loss": 1.3717,
|
| 621 |
+
"step": 2000
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 2.0454545454545454,
|
| 625 |
+
"grad_norm": 16.8206729888916,
|
| 626 |
+
"learning_rate": 3.288439955106622e-05,
|
| 627 |
+
"loss": 1.4279,
|
| 628 |
+
"step": 2025
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 2.0707070707070705,
|
| 632 |
+
"grad_norm": 34.60778045654297,
|
| 633 |
+
"learning_rate": 3.260381593714927e-05,
|
| 634 |
+
"loss": 1.4862,
|
| 635 |
+
"step": 2050
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 2.095959595959596,
|
| 639 |
+
"grad_norm": 27.798601150512695,
|
| 640 |
+
"learning_rate": 3.232323232323233e-05,
|
| 641 |
+
"loss": 1.4113,
|
| 642 |
+
"step": 2075
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 2.121212121212121,
|
| 646 |
+
"grad_norm": 36.2835578918457,
|
| 647 |
+
"learning_rate": 3.204264870931537e-05,
|
| 648 |
+
"loss": 1.3866,
|
| 649 |
+
"step": 2100
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 2.1464646464646466,
|
| 653 |
+
"grad_norm": 21.473899841308594,
|
| 654 |
+
"learning_rate": 3.176206509539843e-05,
|
| 655 |
+
"loss": 1.4915,
|
| 656 |
+
"step": 2125
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 2.1717171717171717,
|
| 660 |
+
"grad_norm": 14.18557357788086,
|
| 661 |
+
"learning_rate": 3.148148148148148e-05,
|
| 662 |
+
"loss": 1.3965,
|
| 663 |
+
"step": 2150
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 2.196969696969697,
|
| 667 |
+
"grad_norm": 25.503524780273438,
|
| 668 |
+
"learning_rate": 3.120089786756454e-05,
|
| 669 |
+
"loss": 1.3686,
|
| 670 |
+
"step": 2175
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 2.2222222222222223,
|
| 674 |
+
"grad_norm": 43.49673080444336,
|
| 675 |
+
"learning_rate": 3.092031425364759e-05,
|
| 676 |
+
"loss": 1.4288,
|
| 677 |
+
"step": 2200
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 2.2474747474747474,
|
| 681 |
+
"grad_norm": 13.61823558807373,
|
| 682 |
+
"learning_rate": 3.063973063973064e-05,
|
| 683 |
+
"loss": 1.309,
|
| 684 |
+
"step": 2225
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 2.2727272727272725,
|
| 688 |
+
"grad_norm": 22.609790802001953,
|
| 689 |
+
"learning_rate": 3.0359147025813696e-05,
|
| 690 |
+
"loss": 1.3974,
|
| 691 |
+
"step": 2250
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 2.297979797979798,
|
| 695 |
+
"grad_norm": 15.05124568939209,
|
| 696 |
+
"learning_rate": 3.007856341189675e-05,
|
| 697 |
+
"loss": 1.4293,
|
| 698 |
+
"step": 2275
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.323232323232323,
|
| 702 |
+
"grad_norm": 24.605804443359375,
|
| 703 |
+
"learning_rate": 2.9797979797979796e-05,
|
| 704 |
+
"loss": 1.3323,
|
| 705 |
+
"step": 2300
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 2.3484848484848486,
|
| 709 |
+
"grad_norm": 33.4686164855957,
|
| 710 |
+
"learning_rate": 2.951739618406285e-05,
|
| 711 |
+
"loss": 1.3623,
|
| 712 |
+
"step": 2325
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 2.3737373737373737,
|
| 716 |
+
"grad_norm": 22.23702621459961,
|
| 717 |
+
"learning_rate": 2.9236812570145904e-05,
|
| 718 |
+
"loss": 1.3697,
|
| 719 |
+
"step": 2350
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 2.398989898989899,
|
| 723 |
+
"grad_norm": 29.579387664794922,
|
| 724 |
+
"learning_rate": 2.8956228956228958e-05,
|
| 725 |
+
"loss": 1.4029,
|
| 726 |
+
"step": 2375
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 2.4242424242424243,
|
| 730 |
+
"grad_norm": 32.36582565307617,
|
| 731 |
+
"learning_rate": 2.8675645342312015e-05,
|
| 732 |
+
"loss": 1.359,
|
| 733 |
+
"step": 2400
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 2.4494949494949494,
|
| 737 |
+
"grad_norm": 16.40806007385254,
|
| 738 |
+
"learning_rate": 2.839506172839506e-05,
|
| 739 |
+
"loss": 1.4343,
|
| 740 |
+
"step": 2425
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 2.474747474747475,
|
| 744 |
+
"grad_norm": 27.034290313720703,
|
| 745 |
+
"learning_rate": 2.8114478114478115e-05,
|
| 746 |
+
"loss": 1.4051,
|
| 747 |
+
"step": 2450
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 2.5,
|
| 751 |
+
"grad_norm": 19.469768524169922,
|
| 752 |
+
"learning_rate": 2.783389450056117e-05,
|
| 753 |
+
"loss": 1.3616,
|
| 754 |
+
"step": 2475
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 2.525252525252525,
|
| 758 |
+
"grad_norm": 23.027786254882812,
|
| 759 |
+
"learning_rate": 2.7553310886644223e-05,
|
| 760 |
+
"loss": 1.3019,
|
| 761 |
+
"step": 2500
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 2.5505050505050506,
|
| 765 |
+
"grad_norm": 23.63875389099121,
|
| 766 |
+
"learning_rate": 2.7272727272727273e-05,
|
| 767 |
+
"loss": 1.3673,
|
| 768 |
+
"step": 2525
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 2.5757575757575757,
|
| 772 |
+
"grad_norm": 18.26418113708496,
|
| 773 |
+
"learning_rate": 2.6992143658810327e-05,
|
| 774 |
+
"loss": 1.3827,
|
| 775 |
+
"step": 2550
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 2.601010101010101,
|
| 779 |
+
"grad_norm": 43.460304260253906,
|
| 780 |
+
"learning_rate": 2.671156004489338e-05,
|
| 781 |
+
"loss": 1.3362,
|
| 782 |
+
"step": 2575
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 2.6262626262626263,
|
| 786 |
+
"grad_norm": 12.931845664978027,
|
| 787 |
+
"learning_rate": 2.6430976430976434e-05,
|
| 788 |
+
"loss": 1.456,
|
| 789 |
+
"step": 2600
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 2.6515151515151514,
|
| 793 |
+
"grad_norm": 19.35201072692871,
|
| 794 |
+
"learning_rate": 2.615039281705948e-05,
|
| 795 |
+
"loss": 1.2981,
|
| 796 |
+
"step": 2625
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 2.676767676767677,
|
| 800 |
+
"grad_norm": 13.542413711547852,
|
| 801 |
+
"learning_rate": 2.5869809203142535e-05,
|
| 802 |
+
"loss": 1.4668,
|
| 803 |
+
"step": 2650
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 2.702020202020202,
|
| 807 |
+
"grad_norm": 16.890657424926758,
|
| 808 |
+
"learning_rate": 2.5589225589225592e-05,
|
| 809 |
+
"loss": 1.321,
|
| 810 |
+
"step": 2675
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 2.7272727272727275,
|
| 814 |
+
"grad_norm": 32.06181716918945,
|
| 815 |
+
"learning_rate": 2.5308641975308646e-05,
|
| 816 |
+
"loss": 1.3692,
|
| 817 |
+
"step": 2700
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 2.7525252525252526,
|
| 821 |
+
"grad_norm": 17.713001251220703,
|
| 822 |
+
"learning_rate": 2.50280583613917e-05,
|
| 823 |
+
"loss": 1.3754,
|
| 824 |
+
"step": 2725
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 2.7777777777777777,
|
| 828 |
+
"grad_norm": 18.294958114624023,
|
| 829 |
+
"learning_rate": 2.474747474747475e-05,
|
| 830 |
+
"loss": 1.323,
|
| 831 |
+
"step": 2750
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 2.8030303030303028,
|
| 835 |
+
"grad_norm": 19.94571876525879,
|
| 836 |
+
"learning_rate": 2.44668911335578e-05,
|
| 837 |
+
"loss": 1.3726,
|
| 838 |
+
"step": 2775
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 2.8282828282828283,
|
| 842 |
+
"grad_norm": 33.24243927001953,
|
| 843 |
+
"learning_rate": 2.4186307519640854e-05,
|
| 844 |
+
"loss": 1.3252,
|
| 845 |
+
"step": 2800
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 2.8535353535353534,
|
| 849 |
+
"grad_norm": 16.95751190185547,
|
| 850 |
+
"learning_rate": 2.3905723905723908e-05,
|
| 851 |
+
"loss": 1.2888,
|
| 852 |
+
"step": 2825
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 2.878787878787879,
|
| 856 |
+
"grad_norm": 16.71131134033203,
|
| 857 |
+
"learning_rate": 2.362514029180696e-05,
|
| 858 |
+
"loss": 1.3961,
|
| 859 |
+
"step": 2850
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 2.904040404040404,
|
| 863 |
+
"grad_norm": 24.505714416503906,
|
| 864 |
+
"learning_rate": 2.3344556677890012e-05,
|
| 865 |
+
"loss": 1.2817,
|
| 866 |
+
"step": 2875
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 2.929292929292929,
|
| 870 |
+
"grad_norm": 19.099855422973633,
|
| 871 |
+
"learning_rate": 2.3063973063973065e-05,
|
| 872 |
+
"loss": 1.2262,
|
| 873 |
+
"step": 2900
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 2.9545454545454546,
|
| 877 |
+
"grad_norm": 18.465789794921875,
|
| 878 |
+
"learning_rate": 2.2783389450056116e-05,
|
| 879 |
+
"loss": 1.3006,
|
| 880 |
+
"step": 2925
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 2.9797979797979797,
|
| 884 |
+
"grad_norm": 23.741085052490234,
|
| 885 |
+
"learning_rate": 2.250280583613917e-05,
|
| 886 |
+
"loss": 1.3211,
|
| 887 |
+
"step": 2950
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 3.0,
|
| 891 |
+
"eval_loss": 1.2902663946151733,
|
| 892 |
+
"eval_map": 0.2016,
|
| 893 |
+
"eval_map_50": 0.4285,
|
| 894 |
+
"eval_map_75": 0.168,
|
| 895 |
+
"eval_map_label0": 0.0804,
|
| 896 |
+
"eval_map_label1": 0.3685,
|
| 897 |
+
"eval_map_label2": 0.1561,
|
| 898 |
+
"eval_map_large": 0.3051,
|
| 899 |
+
"eval_map_medium": 0.0957,
|
| 900 |
+
"eval_map_small": 0.0024,
|
| 901 |
+
"eval_mar_1": 0.2898,
|
| 902 |
+
"eval_mar_10": 0.4088,
|
| 903 |
+
"eval_mar_100": 0.4394,
|
| 904 |
+
"eval_mar_100_label0": 0.4586,
|
| 905 |
+
"eval_mar_100_label1": 0.5428,
|
| 906 |
+
"eval_mar_100_label2": 0.3169,
|
| 907 |
+
"eval_mar_large": 0.6007,
|
| 908 |
+
"eval_mar_medium": 0.341,
|
| 909 |
+
"eval_mar_small": 0.0322,
|
| 910 |
+
"eval_runtime": 135.6725,
|
| 911 |
+
"eval_samples_per_second": 14.594,
|
| 912 |
+
"eval_steps_per_second": 0.914,
|
| 913 |
+
"step": 2970
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 3.005050505050505,
|
| 917 |
+
"grad_norm": 16.176536560058594,
|
| 918 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 919 |
+
"loss": 1.2763,
|
| 920 |
+
"step": 2975
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 3.0303030303030303,
|
| 924 |
+
"grad_norm": 22.124528884887695,
|
| 925 |
+
"learning_rate": 2.1941638608305277e-05,
|
| 926 |
+
"loss": 1.2491,
|
| 927 |
+
"step": 3000
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 3.0555555555555554,
|
| 931 |
+
"grad_norm": 20.725881576538086,
|
| 932 |
+
"learning_rate": 2.166105499438833e-05,
|
| 933 |
+
"loss": 1.3218,
|
| 934 |
+
"step": 3025
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 3.080808080808081,
|
| 938 |
+
"grad_norm": 21.34439468383789,
|
| 939 |
+
"learning_rate": 2.138047138047138e-05,
|
| 940 |
+
"loss": 1.3862,
|
| 941 |
+
"step": 3050
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 3.106060606060606,
|
| 945 |
+
"grad_norm": 27.87030029296875,
|
| 946 |
+
"learning_rate": 2.1099887766554435e-05,
|
| 947 |
+
"loss": 1.2895,
|
| 948 |
+
"step": 3075
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 3.1313131313131315,
|
| 952 |
+
"grad_norm": 37.32668685913086,
|
| 953 |
+
"learning_rate": 2.0819304152637485e-05,
|
| 954 |
+
"loss": 1.3541,
|
| 955 |
+
"step": 3100
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 3.1565656565656566,
|
| 959 |
+
"grad_norm": 17.16813850402832,
|
| 960 |
+
"learning_rate": 2.0538720538720542e-05,
|
| 961 |
+
"loss": 1.2823,
|
| 962 |
+
"step": 3125
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 3.1818181818181817,
|
| 966 |
+
"grad_norm": 22.47783851623535,
|
| 967 |
+
"learning_rate": 2.0258136924803593e-05,
|
| 968 |
+
"loss": 1.3354,
|
| 969 |
+
"step": 3150
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 3.207070707070707,
|
| 973 |
+
"grad_norm": 16.478052139282227,
|
| 974 |
+
"learning_rate": 1.9977553310886646e-05,
|
| 975 |
+
"loss": 1.3117,
|
| 976 |
+
"step": 3175
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 3.2323232323232323,
|
| 980 |
+
"grad_norm": 24.550979614257812,
|
| 981 |
+
"learning_rate": 1.9696969696969697e-05,
|
| 982 |
+
"loss": 1.3446,
|
| 983 |
+
"step": 3200
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 3.257575757575758,
|
| 987 |
+
"grad_norm": 19.664234161376953,
|
| 988 |
+
"learning_rate": 1.941638608305275e-05,
|
| 989 |
+
"loss": 1.322,
|
| 990 |
+
"step": 3225
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 3.282828282828283,
|
| 994 |
+
"grad_norm": 20.59177589416504,
|
| 995 |
+
"learning_rate": 1.91358024691358e-05,
|
| 996 |
+
"loss": 1.2819,
|
| 997 |
+
"step": 3250
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 3.308080808080808,
|
| 1001 |
+
"grad_norm": 24.09212303161621,
|
| 1002 |
+
"learning_rate": 1.8855218855218858e-05,
|
| 1003 |
+
"loss": 1.2937,
|
| 1004 |
+
"step": 3275
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 3.3333333333333335,
|
| 1008 |
+
"grad_norm": 18.40042495727539,
|
| 1009 |
+
"learning_rate": 1.8574635241301908e-05,
|
| 1010 |
+
"loss": 1.3408,
|
| 1011 |
+
"step": 3300
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 3.3585858585858586,
|
| 1015 |
+
"grad_norm": 44.310447692871094,
|
| 1016 |
+
"learning_rate": 1.8294051627384962e-05,
|
| 1017 |
+
"loss": 1.2765,
|
| 1018 |
+
"step": 3325
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 3.3838383838383836,
|
| 1022 |
+
"grad_norm": 23.463211059570312,
|
| 1023 |
+
"learning_rate": 1.8013468013468016e-05,
|
| 1024 |
+
"loss": 1.2986,
|
| 1025 |
+
"step": 3350
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 3.409090909090909,
|
| 1029 |
+
"grad_norm": 24.56027603149414,
|
| 1030 |
+
"learning_rate": 1.7732884399551066e-05,
|
| 1031 |
+
"loss": 1.2789,
|
| 1032 |
+
"step": 3375
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 3.4343434343434343,
|
| 1036 |
+
"grad_norm": 12.601003646850586,
|
| 1037 |
+
"learning_rate": 1.745230078563412e-05,
|
| 1038 |
+
"loss": 1.2991,
|
| 1039 |
+
"step": 3400
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 3.45959595959596,
|
| 1043 |
+
"grad_norm": 14.440610885620117,
|
| 1044 |
+
"learning_rate": 1.7171717171717173e-05,
|
| 1045 |
+
"loss": 1.3621,
|
| 1046 |
+
"step": 3425
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 3.484848484848485,
|
| 1050 |
+
"grad_norm": 18.45439910888672,
|
| 1051 |
+
"learning_rate": 1.6891133557800227e-05,
|
| 1052 |
+
"loss": 1.3083,
|
| 1053 |
+
"step": 3450
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 3.51010101010101,
|
| 1057 |
+
"grad_norm": 18.901931762695312,
|
| 1058 |
+
"learning_rate": 1.6610549943883278e-05,
|
| 1059 |
+
"loss": 1.2736,
|
| 1060 |
+
"step": 3475
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 3.5353535353535355,
|
| 1064 |
+
"grad_norm": 40.01392364501953,
|
| 1065 |
+
"learning_rate": 1.632996632996633e-05,
|
| 1066 |
+
"loss": 1.2889,
|
| 1067 |
+
"step": 3500
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 3.5606060606060606,
|
| 1071 |
+
"grad_norm": 33.847312927246094,
|
| 1072 |
+
"learning_rate": 1.604938271604938e-05,
|
| 1073 |
+
"loss": 1.3858,
|
| 1074 |
+
"step": 3525
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 3.5858585858585856,
|
| 1078 |
+
"grad_norm": 15.7413330078125,
|
| 1079 |
+
"learning_rate": 1.5768799102132435e-05,
|
| 1080 |
+
"loss": 1.2634,
|
| 1081 |
+
"step": 3550
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 3.611111111111111,
|
| 1085 |
+
"grad_norm": 11.64206600189209,
|
| 1086 |
+
"learning_rate": 1.548821548821549e-05,
|
| 1087 |
+
"loss": 1.2464,
|
| 1088 |
+
"step": 3575
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 3.6363636363636362,
|
| 1092 |
+
"grad_norm": 22.535266876220703,
|
| 1093 |
+
"learning_rate": 1.5207631874298541e-05,
|
| 1094 |
+
"loss": 1.283,
|
| 1095 |
+
"step": 3600
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 3.6616161616161618,
|
| 1099 |
+
"grad_norm": 24.02773666381836,
|
| 1100 |
+
"learning_rate": 1.4927048260381593e-05,
|
| 1101 |
+
"loss": 1.2939,
|
| 1102 |
+
"step": 3625
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 3.686868686868687,
|
| 1106 |
+
"grad_norm": 24.32711410522461,
|
| 1107 |
+
"learning_rate": 1.4646464646464647e-05,
|
| 1108 |
+
"loss": 1.2618,
|
| 1109 |
+
"step": 3650
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 3.712121212121212,
|
| 1113 |
+
"grad_norm": 31.185592651367188,
|
| 1114 |
+
"learning_rate": 1.43658810325477e-05,
|
| 1115 |
+
"loss": 1.2082,
|
| 1116 |
+
"step": 3675
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 3.7373737373737375,
|
| 1120 |
+
"grad_norm": 19.14853286743164,
|
| 1121 |
+
"learning_rate": 1.4085297418630753e-05,
|
| 1122 |
+
"loss": 1.2204,
|
| 1123 |
+
"step": 3700
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 3.7626262626262625,
|
| 1127 |
+
"grad_norm": 19.671226501464844,
|
| 1128 |
+
"learning_rate": 1.3804713804713806e-05,
|
| 1129 |
+
"loss": 1.2873,
|
| 1130 |
+
"step": 3725
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 3.787878787878788,
|
| 1134 |
+
"grad_norm": 37.9943733215332,
|
| 1135 |
+
"learning_rate": 1.3524130190796858e-05,
|
| 1136 |
+
"loss": 1.2267,
|
| 1137 |
+
"step": 3750
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 3.813131313131313,
|
| 1141 |
+
"grad_norm": 39.25022506713867,
|
| 1142 |
+
"learning_rate": 1.3243546576879912e-05,
|
| 1143 |
+
"loss": 1.3132,
|
| 1144 |
+
"step": 3775
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 3.8383838383838382,
|
| 1148 |
+
"grad_norm": 31.733964920043945,
|
| 1149 |
+
"learning_rate": 1.2962962962962962e-05,
|
| 1150 |
+
"loss": 1.2981,
|
| 1151 |
+
"step": 3800
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 3.8636363636363638,
|
| 1155 |
+
"grad_norm": 13.638587951660156,
|
| 1156 |
+
"learning_rate": 1.2682379349046016e-05,
|
| 1157 |
+
"loss": 1.2567,
|
| 1158 |
+
"step": 3825
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 3.888888888888889,
|
| 1162 |
+
"grad_norm": 27.432302474975586,
|
| 1163 |
+
"learning_rate": 1.240179573512907e-05,
|
| 1164 |
+
"loss": 1.1788,
|
| 1165 |
+
"step": 3850
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 3.9141414141414144,
|
| 1169 |
+
"grad_norm": 26.565326690673828,
|
| 1170 |
+
"learning_rate": 1.2121212121212122e-05,
|
| 1171 |
+
"loss": 1.2726,
|
| 1172 |
+
"step": 3875
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 3.9393939393939394,
|
| 1176 |
+
"grad_norm": 15.759629249572754,
|
| 1177 |
+
"learning_rate": 1.1840628507295174e-05,
|
| 1178 |
+
"loss": 1.3079,
|
| 1179 |
+
"step": 3900
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 3.9646464646464645,
|
| 1183 |
+
"grad_norm": 32.35866165161133,
|
| 1184 |
+
"learning_rate": 1.1560044893378228e-05,
|
| 1185 |
+
"loss": 1.3048,
|
| 1186 |
+
"step": 3925
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 3.98989898989899,
|
| 1190 |
+
"grad_norm": 13.627097129821777,
|
| 1191 |
+
"learning_rate": 1.127946127946128e-05,
|
| 1192 |
+
"loss": 1.293,
|
| 1193 |
+
"step": 3950
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 4.0,
|
| 1197 |
+
"eval_loss": 1.2365508079528809,
|
| 1198 |
+
"eval_map": 0.2352,
|
| 1199 |
+
"eval_map_50": 0.4789,
|
| 1200 |
+
"eval_map_75": 0.2113,
|
| 1201 |
+
"eval_map_label0": 0.0958,
|
| 1202 |
+
"eval_map_label1": 0.3932,
|
| 1203 |
+
"eval_map_label2": 0.2165,
|
| 1204 |
+
"eval_map_large": 0.3554,
|
| 1205 |
+
"eval_map_medium": 0.1165,
|
| 1206 |
+
"eval_map_small": 0.0025,
|
| 1207 |
+
"eval_mar_1": 0.3252,
|
| 1208 |
+
"eval_mar_10": 0.4433,
|
| 1209 |
+
"eval_mar_100": 0.479,
|
| 1210 |
+
"eval_mar_100_label0": 0.5128,
|
| 1211 |
+
"eval_mar_100_label1": 0.553,
|
| 1212 |
+
"eval_mar_100_label2": 0.3714,
|
| 1213 |
+
"eval_mar_large": 0.6639,
|
| 1214 |
+
"eval_mar_medium": 0.3693,
|
| 1215 |
+
"eval_mar_small": 0.0282,
|
| 1216 |
+
"eval_runtime": 189.7649,
|
| 1217 |
+
"eval_samples_per_second": 10.434,
|
| 1218 |
+
"eval_steps_per_second": 0.653,
|
| 1219 |
+
"step": 3960
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 4.015151515151516,
|
| 1223 |
+
"grad_norm": 16.43737030029297,
|
| 1224 |
+
"learning_rate": 1.0998877665544333e-05,
|
| 1225 |
+
"loss": 1.3154,
|
| 1226 |
+
"step": 3975
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 4.040404040404041,
|
| 1230 |
+
"grad_norm": 12.887079238891602,
|
| 1231 |
+
"learning_rate": 1.0718294051627386e-05,
|
| 1232 |
+
"loss": 1.2436,
|
| 1233 |
+
"step": 4000
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 4.065656565656566,
|
| 1237 |
+
"grad_norm": 27.211605072021484,
|
| 1238 |
+
"learning_rate": 1.0437710437710438e-05,
|
| 1239 |
+
"loss": 1.2369,
|
| 1240 |
+
"step": 4025
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 4.090909090909091,
|
| 1244 |
+
"grad_norm": 30.36830711364746,
|
| 1245 |
+
"learning_rate": 1.0157126823793491e-05,
|
| 1246 |
+
"loss": 1.2688,
|
| 1247 |
+
"step": 4050
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 4.116161616161616,
|
| 1251 |
+
"grad_norm": 16.97003936767578,
|
| 1252 |
+
"learning_rate": 9.876543209876543e-06,
|
| 1253 |
+
"loss": 1.1834,
|
| 1254 |
+
"step": 4075
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 4.141414141414141,
|
| 1258 |
+
"grad_norm": 19.722625732421875,
|
| 1259 |
+
"learning_rate": 9.595959595959595e-06,
|
| 1260 |
+
"loss": 1.2216,
|
| 1261 |
+
"step": 4100
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 4.166666666666667,
|
| 1265 |
+
"grad_norm": 24.40361213684082,
|
| 1266 |
+
"learning_rate": 9.315375982042649e-06,
|
| 1267 |
+
"loss": 1.1382,
|
| 1268 |
+
"step": 4125
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 4.191919191919192,
|
| 1272 |
+
"grad_norm": 28.218984603881836,
|
| 1273 |
+
"learning_rate": 9.034792368125703e-06,
|
| 1274 |
+
"loss": 1.1971,
|
| 1275 |
+
"step": 4150
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 4.217171717171717,
|
| 1279 |
+
"grad_norm": 23.390958786010742,
|
| 1280 |
+
"learning_rate": 8.754208754208755e-06,
|
| 1281 |
+
"loss": 1.3141,
|
| 1282 |
+
"step": 4175
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 4.242424242424242,
|
| 1286 |
+
"grad_norm": 22.968050003051758,
|
| 1287 |
+
"learning_rate": 8.473625140291807e-06,
|
| 1288 |
+
"loss": 1.2497,
|
| 1289 |
+
"step": 4200
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 4.267676767676767,
|
| 1293 |
+
"grad_norm": 21.7801570892334,
|
| 1294 |
+
"learning_rate": 8.19304152637486e-06,
|
| 1295 |
+
"loss": 1.2535,
|
| 1296 |
+
"step": 4225
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 4.292929292929293,
|
| 1300 |
+
"grad_norm": 32.12472915649414,
|
| 1301 |
+
"learning_rate": 7.912457912457913e-06,
|
| 1302 |
+
"loss": 1.2288,
|
| 1303 |
+
"step": 4250
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 4.318181818181818,
|
| 1307 |
+
"grad_norm": 20.222686767578125,
|
| 1308 |
+
"learning_rate": 7.631874298540966e-06,
|
| 1309 |
+
"loss": 1.2274,
|
| 1310 |
+
"step": 4275
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 4.343434343434343,
|
| 1314 |
+
"grad_norm": 26.127695083618164,
|
| 1315 |
+
"learning_rate": 7.351290684624018e-06,
|
| 1316 |
+
"loss": 1.1986,
|
| 1317 |
+
"step": 4300
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 4.3686868686868685,
|
| 1321 |
+
"grad_norm": 16.555965423583984,
|
| 1322 |
+
"learning_rate": 7.0707070707070704e-06,
|
| 1323 |
+
"loss": 1.2522,
|
| 1324 |
+
"step": 4325
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 4.393939393939394,
|
| 1328 |
+
"grad_norm": 16.466201782226562,
|
| 1329 |
+
"learning_rate": 6.790123456790123e-06,
|
| 1330 |
+
"loss": 1.2335,
|
| 1331 |
+
"step": 4350
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 4.41919191919192,
|
| 1335 |
+
"grad_norm": 18.98979949951172,
|
| 1336 |
+
"learning_rate": 6.509539842873176e-06,
|
| 1337 |
+
"loss": 1.3064,
|
| 1338 |
+
"step": 4375
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 4.444444444444445,
|
| 1342 |
+
"grad_norm": 13.089599609375,
|
| 1343 |
+
"learning_rate": 6.2401795735129076e-06,
|
| 1344 |
+
"loss": 1.1926,
|
| 1345 |
+
"step": 4400
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 4.46969696969697,
|
| 1349 |
+
"grad_norm": 51.39873504638672,
|
| 1350 |
+
"learning_rate": 5.9595959595959605e-06,
|
| 1351 |
+
"loss": 1.211,
|
| 1352 |
+
"step": 4425
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 4.494949494949495,
|
| 1356 |
+
"grad_norm": 16.774805068969727,
|
| 1357 |
+
"learning_rate": 5.6790123456790125e-06,
|
| 1358 |
+
"loss": 1.2553,
|
| 1359 |
+
"step": 4450
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 4.52020202020202,
|
| 1363 |
+
"grad_norm": 15.329991340637207,
|
| 1364 |
+
"learning_rate": 5.398428731762065e-06,
|
| 1365 |
+
"loss": 1.1924,
|
| 1366 |
+
"step": 4475
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 4.545454545454545,
|
| 1370 |
+
"grad_norm": 27.15462875366211,
|
| 1371 |
+
"learning_rate": 5.117845117845118e-06,
|
| 1372 |
+
"loss": 1.2463,
|
| 1373 |
+
"step": 4500
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 4.570707070707071,
|
| 1377 |
+
"grad_norm": 21.43892478942871,
|
| 1378 |
+
"learning_rate": 4.83726150392817e-06,
|
| 1379 |
+
"loss": 1.2831,
|
| 1380 |
+
"step": 4525
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 4.595959595959596,
|
| 1384 |
+
"grad_norm": 33.843936920166016,
|
| 1385 |
+
"learning_rate": 4.556677890011223e-06,
|
| 1386 |
+
"loss": 1.1781,
|
| 1387 |
+
"step": 4550
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 4.621212121212121,
|
| 1391 |
+
"grad_norm": 20.4368896484375,
|
| 1392 |
+
"learning_rate": 4.276094276094277e-06,
|
| 1393 |
+
"loss": 1.2344,
|
| 1394 |
+
"step": 4575
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 4.646464646464646,
|
| 1398 |
+
"grad_norm": 25.551361083984375,
|
| 1399 |
+
"learning_rate": 3.995510662177329e-06,
|
| 1400 |
+
"loss": 1.2114,
|
| 1401 |
+
"step": 4600
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 4.671717171717171,
|
| 1405 |
+
"grad_norm": 18.057945251464844,
|
| 1406 |
+
"learning_rate": 3.714927048260382e-06,
|
| 1407 |
+
"loss": 1.2301,
|
| 1408 |
+
"step": 4625
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 4.696969696969697,
|
| 1412 |
+
"grad_norm": 25.322641372680664,
|
| 1413 |
+
"learning_rate": 3.4343434343434343e-06,
|
| 1414 |
+
"loss": 1.1766,
|
| 1415 |
+
"step": 4650
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 4.722222222222222,
|
| 1419 |
+
"grad_norm": 21.608871459960938,
|
| 1420 |
+
"learning_rate": 3.153759820426487e-06,
|
| 1421 |
+
"loss": 1.1757,
|
| 1422 |
+
"step": 4675
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 4.747474747474747,
|
| 1426 |
+
"grad_norm": 12.744592666625977,
|
| 1427 |
+
"learning_rate": 2.87317620650954e-06,
|
| 1428 |
+
"loss": 1.1878,
|
| 1429 |
+
"step": 4700
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 4.7727272727272725,
|
| 1433 |
+
"grad_norm": 18.519357681274414,
|
| 1434 |
+
"learning_rate": 2.5925925925925925e-06,
|
| 1435 |
+
"loss": 1.185,
|
| 1436 |
+
"step": 4725
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 4.797979797979798,
|
| 1440 |
+
"grad_norm": 17.793031692504883,
|
| 1441 |
+
"learning_rate": 2.312008978675646e-06,
|
| 1442 |
+
"loss": 1.1464,
|
| 1443 |
+
"step": 4750
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 4.8232323232323235,
|
| 1447 |
+
"grad_norm": 16.70855712890625,
|
| 1448 |
+
"learning_rate": 2.0314253647586983e-06,
|
| 1449 |
+
"loss": 1.2202,
|
| 1450 |
+
"step": 4775
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 4.848484848484849,
|
| 1454 |
+
"grad_norm": 23.166088104248047,
|
| 1455 |
+
"learning_rate": 1.750841750841751e-06,
|
| 1456 |
+
"loss": 1.2478,
|
| 1457 |
+
"step": 4800
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 4.873737373737374,
|
| 1461 |
+
"grad_norm": 25.1901798248291,
|
| 1462 |
+
"learning_rate": 1.4702581369248036e-06,
|
| 1463 |
+
"loss": 1.2884,
|
| 1464 |
+
"step": 4825
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 4.898989898989899,
|
| 1468 |
+
"grad_norm": 18.20812225341797,
|
| 1469 |
+
"learning_rate": 1.1896745230078565e-06,
|
| 1470 |
+
"loss": 1.1887,
|
| 1471 |
+
"step": 4850
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 4.924242424242424,
|
| 1475 |
+
"grad_norm": 18.666954040527344,
|
| 1476 |
+
"learning_rate": 9.09090909090909e-07,
|
| 1477 |
+
"loss": 1.2136,
|
| 1478 |
+
"step": 4875
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 4.94949494949495,
|
| 1482 |
+
"grad_norm": 31.150278091430664,
|
| 1483 |
+
"learning_rate": 6.285072951739619e-07,
|
| 1484 |
+
"loss": 1.2078,
|
| 1485 |
+
"step": 4900
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 4.974747474747475,
|
| 1489 |
+
"grad_norm": 12.956293106079102,
|
| 1490 |
+
"learning_rate": 3.479236812570146e-07,
|
| 1491 |
+
"loss": 1.2789,
|
| 1492 |
+
"step": 4925
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 5.0,
|
| 1496 |
+
"grad_norm": 21.77115249633789,
|
| 1497 |
+
"learning_rate": 6.734006734006734e-08,
|
| 1498 |
+
"loss": 1.1564,
|
| 1499 |
+
"step": 4950
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 5.0,
|
| 1503 |
+
"eval_loss": 1.1927123069763184,
|
| 1504 |
+
"eval_map": 0.2537,
|
| 1505 |
+
"eval_map_50": 0.502,
|
| 1506 |
+
"eval_map_75": 0.2303,
|
| 1507 |
+
"eval_map_label0": 0.1064,
|
| 1508 |
+
"eval_map_label1": 0.4212,
|
| 1509 |
+
"eval_map_label2": 0.2333,
|
| 1510 |
+
"eval_map_large": 0.3796,
|
| 1511 |
+
"eval_map_medium": 0.1325,
|
| 1512 |
+
"eval_map_small": 0.0016,
|
| 1513 |
+
"eval_mar_1": 0.3388,
|
| 1514 |
+
"eval_mar_10": 0.4646,
|
| 1515 |
+
"eval_mar_100": 0.5012,
|
| 1516 |
+
"eval_mar_100_label0": 0.5197,
|
| 1517 |
+
"eval_mar_100_label1": 0.5865,
|
| 1518 |
+
"eval_mar_100_label2": 0.3974,
|
| 1519 |
+
"eval_mar_large": 0.678,
|
| 1520 |
+
"eval_mar_medium": 0.4027,
|
| 1521 |
+
"eval_mar_small": 0.0456,
|
| 1522 |
+
"eval_runtime": 190.0886,
|
| 1523 |
+
"eval_samples_per_second": 10.416,
|
| 1524 |
+
"eval_steps_per_second": 0.652,
|
| 1525 |
+
"step": 4950
|
| 1526 |
+
}
|
| 1527 |
+
],
|
| 1528 |
+
"logging_steps": 25,
|
| 1529 |
+
"max_steps": 4950,
|
| 1530 |
+
"num_input_tokens_seen": 0,
|
| 1531 |
+
"num_train_epochs": 5,
|
| 1532 |
+
"save_steps": 500,
|
| 1533 |
+
"stateful_callbacks": {
|
| 1534 |
+
"EarlyStoppingCallback": {
|
| 1535 |
+
"args": {
|
| 1536 |
+
"early_stopping_patience": 5,
|
| 1537 |
+
"early_stopping_threshold": 0.01
|
| 1538 |
+
},
|
| 1539 |
+
"attributes": {
|
| 1540 |
+
"early_stopping_patience_counter": 0
|
| 1541 |
+
}
|
| 1542 |
+
},
|
| 1543 |
+
"TrainerControl": {
|
| 1544 |
+
"args": {
|
| 1545 |
+
"should_epoch_stop": false,
|
| 1546 |
+
"should_evaluate": false,
|
| 1547 |
+
"should_log": false,
|
| 1548 |
+
"should_save": true,
|
| 1549 |
+
"should_training_stop": true
|
| 1550 |
+
},
|
| 1551 |
+
"attributes": {}
|
| 1552 |
+
}
|
| 1553 |
+
},
|
| 1554 |
+
"total_flos": 8.94903394074624e+18,
|
| 1555 |
+
"train_batch_size": 8,
|
| 1556 |
+
"trial_name": null,
|
| 1557 |
+
"trial_params": null
|
| 1558 |
+
}
|
checkpoint-4950/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:93a9f65387e850cdb532b8cc40a1f985541a476dd8baaabe7a5b1b280545a9e9
|
| 3 |
+
size 5304
|
config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "hustvl/yolos-small",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"YolosForObjectDetection"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.0,
|
| 7 |
+
"auxiliary_loss": false,
|
| 8 |
+
"bbox_cost": 5,
|
| 9 |
+
"bbox_loss_coefficient": 5,
|
| 10 |
+
"class_cost": 1,
|
| 11 |
+
"eos_coefficient": 0.1,
|
| 12 |
+
"giou_cost": 2,
|
| 13 |
+
"giou_loss_coefficient": 2,
|
| 14 |
+
"hidden_act": "gelu",
|
| 15 |
+
"hidden_dropout_prob": 0.0,
|
| 16 |
+
"hidden_size": 384,
|
| 17 |
+
"id2label": {
|
| 18 |
+
"0": "brain-tumor",
|
| 19 |
+
"1": "label0",
|
| 20 |
+
"2": "label1",
|
| 21 |
+
"3": "label2"
|
| 22 |
+
},
|
| 23 |
+
"image_size": [
|
| 24 |
+
512,
|
| 25 |
+
864
|
| 26 |
+
],
|
| 27 |
+
"initializer_range": 0.02,
|
| 28 |
+
"intermediate_size": 1536,
|
| 29 |
+
"label2id": {
|
| 30 |
+
"brain-tumor": 0,
|
| 31 |
+
"label0": 1,
|
| 32 |
+
"label1": 2,
|
| 33 |
+
"label2": 3
|
| 34 |
+
},
|
| 35 |
+
"layer_norm_eps": 1e-12,
|
| 36 |
+
"model_type": "yolos",
|
| 37 |
+
"num_attention_heads": 6,
|
| 38 |
+
"num_channels": 3,
|
| 39 |
+
"num_detection_tokens": 100,
|
| 40 |
+
"num_hidden_layers": 12,
|
| 41 |
+
"patch_size": 16,
|
| 42 |
+
"qkv_bias": true,
|
| 43 |
+
"torch_dtype": "float32",
|
| 44 |
+
"transformers_version": "4.46.2",
|
| 45 |
+
"use_mid_position_embeddings": true
|
| 46 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:91f9a40cd583f7353f09a374ff308a89d686aef6c0682d2edb9a5613668f6ae4
|
| 3 |
+
size 122629292
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_annotations": true,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_pad": false,
|
| 5 |
+
"do_rescale": true,
|
| 6 |
+
"do_resize": false,
|
| 7 |
+
"format": "coco_detection",
|
| 8 |
+
"image_mean": [
|
| 9 |
+
0.485,
|
| 10 |
+
0.456,
|
| 11 |
+
0.406
|
| 12 |
+
],
|
| 13 |
+
"image_processor_type": "YolosImageProcessor",
|
| 14 |
+
"image_std": [
|
| 15 |
+
0.229,
|
| 16 |
+
0.224,
|
| 17 |
+
0.225
|
| 18 |
+
],
|
| 19 |
+
"pad_size": null,
|
| 20 |
+
"resample": 2,
|
| 21 |
+
"rescale_factor": 0.00392156862745098,
|
| 22 |
+
"size": {
|
| 23 |
+
"longest_edge": 640
|
| 24 |
+
}
|
| 25 |
+
}
|
runs/Nov19_07-08-55_659d872c1556/events.out.tfevents.1732000135.659d872c1556.241.0
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 sha256:8f5061c54c6ec2418a0e7d548dffbaf679ea3103cecc0c54eb2688fba4e4ed50
|
| 3 |
+
size 53626
|
runs/Nov19_07-08-55_659d872c1556/events.out.tfevents.1732008067.659d872c1556.241.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7d477735ed8524a485d99e88524dbe6fce610299b2787cab2dc2e38bd11cc8cd
|
| 3 |
+
size 1312
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:93a9f65387e850cdb532b8cc40a1f985541a476dd8baaabe7a5b1b280545a9e9
|
| 3 |
+
size 5304
|
training_params.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "Francesco/brain-tumor-m2pbp",
|
| 3 |
+
"model": "hustvl/yolos-small",
|
| 4 |
+
"username": "Aryan-401",
|
| 5 |
+
"lr": 5e-05,
|
| 6 |
+
"epochs": 4,
|
| 7 |
+
"batch_size": 32,
|
| 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+validation",
|
| 16 |
+
"valid_split": "test",
|
| 17 |
+
"logging_steps": -1,
|
| 18 |
+
"project_name": "tumor-detect-yolo-small",
|
| 19 |
+
"auto_find_batch_size": false,
|
| 20 |
+
"mixed_precision": "fp16",
|
| 21 |
+
"save_total_limit": 1,
|
| 22 |
+
"push_to_hub": true,
|
| 23 |
+
"eval_strategy": "epoch",
|
| 24 |
+
"image_column": "image",
|
| 25 |
+
"objects_column": "objects",
|
| 26 |
+
"log": "tensorboard",
|
| 27 |
+
"image_square_size": 640,
|
| 28 |
+
"early_stopping_patience": 5,
|
| 29 |
+
"early_stopping_threshold": 0.01
|
| 30 |
+
}
|