Upload folder using huggingface_hub
Browse files- README.md +31 -0
- checkpoint-270/config.json +34 -0
- checkpoint-270/model.safetensors +3 -0
- checkpoint-270/optimizer.pt +3 -0
- checkpoint-270/rng_state.pth +3 -0
- checkpoint-270/scheduler.pt +3 -0
- checkpoint-270/trainer_state.json +1052 -0
- checkpoint-270/training_args.bin +3 -0
- config.json +34 -0
- model.safetensors +3 -0
- preprocessor_config.json +26 -0
- runs/Jul31_13-24-57_r-chispidev-ia-banana-fwzg7678-62e49-8u7fh/events.out.tfevents.1722432298.r-chispidev-ia-banana-fwzg7678-62e49-8u7fh.147.0 +2 -2
- runs/Jul31_13-24-57_r-chispidev-ia-banana-fwzg7678-62e49-8u7fh/events.out.tfevents.1722432616.r-chispidev-ia-banana-fwzg7678-62e49-8u7fh.147.1 +3 -0
- training_args.bin +3 -0
- training_params.json +29 -0
README.md
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
---
|
| 3 |
+
tags:
|
| 4 |
+
- autotrain
|
| 5 |
+
- image-classification
|
| 6 |
+
base_model: google/mobilenet_v2_1.0_224
|
| 7 |
+
widget:
|
| 8 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/tiger.jpg
|
| 9 |
+
example_title: Tiger
|
| 10 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/teapot.jpg
|
| 11 |
+
example_title: Teapot
|
| 12 |
+
- src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/palace.jpg
|
| 13 |
+
example_title: Palace
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
# Model Trained Using AutoTrain
|
| 17 |
+
|
| 18 |
+
- Problem type: Image Classification
|
| 19 |
+
|
| 20 |
+
## Validation Metrics
|
| 21 |
+
loss: 0.42044156789779663
|
| 22 |
+
|
| 23 |
+
f1: 0.9361702127659575
|
| 24 |
+
|
| 25 |
+
precision: 0.88
|
| 26 |
+
|
| 27 |
+
recall: 1.0
|
| 28 |
+
|
| 29 |
+
auc: 0.8340909090909091
|
| 30 |
+
|
| 31 |
+
accuracy: 0.8888888888888888
|
checkpoint-270/config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/mobilenet_v2_1.0_224",
|
| 3 |
+
"_num_labels": 2,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"MobileNetV2ForImageClassification"
|
| 6 |
+
],
|
| 7 |
+
"classifier_dropout_prob": 0.2,
|
| 8 |
+
"depth_divisible_by": 8,
|
| 9 |
+
"depth_multiplier": 1.0,
|
| 10 |
+
"expand_ratio": 6,
|
| 11 |
+
"finegrained_output": true,
|
| 12 |
+
"first_layer_is_expansion": true,
|
| 13 |
+
"hidden_act": "relu6",
|
| 14 |
+
"id2label": {
|
| 15 |
+
"0": "rotten banana",
|
| 16 |
+
"1": "yellow banana"
|
| 17 |
+
},
|
| 18 |
+
"image_size": 224,
|
| 19 |
+
"initializer_range": 0.02,
|
| 20 |
+
"label2id": {
|
| 21 |
+
"rotten banana": 0,
|
| 22 |
+
"yellow banana": 1
|
| 23 |
+
},
|
| 24 |
+
"layer_norm_eps": 0.001,
|
| 25 |
+
"min_depth": 8,
|
| 26 |
+
"model_type": "mobilenet_v2",
|
| 27 |
+
"num_channels": 3,
|
| 28 |
+
"output_stride": 32,
|
| 29 |
+
"problem_type": "single_label_classification",
|
| 30 |
+
"semantic_loss_ignore_index": 255,
|
| 31 |
+
"tf_padding": true,
|
| 32 |
+
"torch_dtype": "float32",
|
| 33 |
+
"transformers_version": "4.43.1"
|
| 34 |
+
}
|
checkpoint-270/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4f8aa363d171b4e520dac9d00149698714b8f5c90b1a6327e4b9f4d35fd97233
|
| 3 |
+
size 9080216
|
checkpoint-270/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6495cd5a1a68af10dd3adf057e0464a349556dfa1f73ea2cab5850a18bc1dd6d
|
| 3 |
+
size 17939322
|
checkpoint-270/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:31f98014c52404d9bd685338d13c07e11f3be1c58c712812368c6d4ec102c8a8
|
| 3 |
+
size 13990
|
checkpoint-270/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:259ffb58c0acaf40e622dd34936c6507d4f7114a47fe31c6997724ffab4fdd83
|
| 3 |
+
size 1064
|
checkpoint-270/trainer_state.json
ADDED
|
@@ -0,0 +1,1052 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.42044156789779663,
|
| 3 |
+
"best_model_checkpoint": "autotrain-pky99-ias73/checkpoint-270",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 270,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.037037037037037035,
|
| 13 |
+
"grad_norm": 18.27926254272461,
|
| 14 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 15 |
+
"loss": 0.7067,
|
| 16 |
+
"step": 2
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.07407407407407407,
|
| 20 |
+
"grad_norm": 15.84852123260498,
|
| 21 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 22 |
+
"loss": 0.6894,
|
| 23 |
+
"step": 4
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.1111111111111111,
|
| 27 |
+
"grad_norm": 15.550982475280762,
|
| 28 |
+
"learning_rate": 5.555555555555556e-06,
|
| 29 |
+
"loss": 0.7159,
|
| 30 |
+
"step": 6
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.14814814814814814,
|
| 34 |
+
"grad_norm": 18.58649253845215,
|
| 35 |
+
"learning_rate": 7.4074074074074075e-06,
|
| 36 |
+
"loss": 0.8481,
|
| 37 |
+
"step": 8
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.18518518518518517,
|
| 41 |
+
"grad_norm": 13.416651725769043,
|
| 42 |
+
"learning_rate": 9.259259259259259e-06,
|
| 43 |
+
"loss": 0.6869,
|
| 44 |
+
"step": 10
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.2222222222222222,
|
| 48 |
+
"grad_norm": 17.145116806030273,
|
| 49 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 50 |
+
"loss": 0.7312,
|
| 51 |
+
"step": 12
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.25925925925925924,
|
| 55 |
+
"grad_norm": 17.046096801757812,
|
| 56 |
+
"learning_rate": 1.2962962962962962e-05,
|
| 57 |
+
"loss": 0.7309,
|
| 58 |
+
"step": 14
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.2962962962962963,
|
| 62 |
+
"grad_norm": 13.83819580078125,
|
| 63 |
+
"learning_rate": 1.4814814814814815e-05,
|
| 64 |
+
"loss": 0.6518,
|
| 65 |
+
"step": 16
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.3333333333333333,
|
| 69 |
+
"grad_norm": 19.314144134521484,
|
| 70 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 71 |
+
"loss": 0.8117,
|
| 72 |
+
"step": 18
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.37037037037037035,
|
| 76 |
+
"grad_norm": 15.31916332244873,
|
| 77 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 78 |
+
"loss": 0.6198,
|
| 79 |
+
"step": 20
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.4074074074074074,
|
| 83 |
+
"grad_norm": 16.181686401367188,
|
| 84 |
+
"learning_rate": 2.037037037037037e-05,
|
| 85 |
+
"loss": 0.6906,
|
| 86 |
+
"step": 22
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.4444444444444444,
|
| 90 |
+
"grad_norm": 14.573526382446289,
|
| 91 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 92 |
+
"loss": 0.6868,
|
| 93 |
+
"step": 24
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.48148148148148145,
|
| 97 |
+
"grad_norm": 16.38991928100586,
|
| 98 |
+
"learning_rate": 2.4074074074074074e-05,
|
| 99 |
+
"loss": 0.6247,
|
| 100 |
+
"step": 26
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.5185185185185185,
|
| 104 |
+
"grad_norm": 13.956877708435059,
|
| 105 |
+
"learning_rate": 2.5925925925925925e-05,
|
| 106 |
+
"loss": 0.6263,
|
| 107 |
+
"step": 28
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.5555555555555556,
|
| 111 |
+
"grad_norm": 18.015186309814453,
|
| 112 |
+
"learning_rate": 2.777777777777778e-05,
|
| 113 |
+
"loss": 0.8402,
|
| 114 |
+
"step": 30
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.5925925925925926,
|
| 118 |
+
"grad_norm": 10.778483390808105,
|
| 119 |
+
"learning_rate": 2.962962962962963e-05,
|
| 120 |
+
"loss": 0.4364,
|
| 121 |
+
"step": 32
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.6296296296296297,
|
| 125 |
+
"grad_norm": 13.538158416748047,
|
| 126 |
+
"learning_rate": 3.148148148148148e-05,
|
| 127 |
+
"loss": 0.5065,
|
| 128 |
+
"step": 34
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.6666666666666666,
|
| 132 |
+
"grad_norm": 10.736392974853516,
|
| 133 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 134 |
+
"loss": 0.5143,
|
| 135 |
+
"step": 36
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.7037037037037037,
|
| 139 |
+
"grad_norm": 18.392257690429688,
|
| 140 |
+
"learning_rate": 3.518518518518519e-05,
|
| 141 |
+
"loss": 0.7119,
|
| 142 |
+
"step": 38
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.7407407407407407,
|
| 146 |
+
"grad_norm": 12.574856758117676,
|
| 147 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 148 |
+
"loss": 0.4684,
|
| 149 |
+
"step": 40
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.7777777777777778,
|
| 153 |
+
"grad_norm": 10.016814231872559,
|
| 154 |
+
"learning_rate": 3.888888888888889e-05,
|
| 155 |
+
"loss": 0.4852,
|
| 156 |
+
"step": 42
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.8148148148148148,
|
| 160 |
+
"grad_norm": 10.29903507232666,
|
| 161 |
+
"learning_rate": 4.074074074074074e-05,
|
| 162 |
+
"loss": 0.4805,
|
| 163 |
+
"step": 44
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.8518518518518519,
|
| 167 |
+
"grad_norm": 8.819818496704102,
|
| 168 |
+
"learning_rate": 4.259259259259259e-05,
|
| 169 |
+
"loss": 0.3533,
|
| 170 |
+
"step": 46
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.8888888888888888,
|
| 174 |
+
"grad_norm": 9.893930435180664,
|
| 175 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 176 |
+
"loss": 0.4406,
|
| 177 |
+
"step": 48
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.9259259259259259,
|
| 181 |
+
"grad_norm": 9.269518852233887,
|
| 182 |
+
"learning_rate": 4.62962962962963e-05,
|
| 183 |
+
"loss": 0.3606,
|
| 184 |
+
"step": 50
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.9629629629629629,
|
| 188 |
+
"grad_norm": 9.66145133972168,
|
| 189 |
+
"learning_rate": 4.814814814814815e-05,
|
| 190 |
+
"loss": 0.7552,
|
| 191 |
+
"step": 52
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.0,
|
| 195 |
+
"grad_norm": 15.34220027923584,
|
| 196 |
+
"learning_rate": 5e-05,
|
| 197 |
+
"loss": 0.421,
|
| 198 |
+
"step": 54
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 1.0,
|
| 202 |
+
"eval_accuracy": 0.7592592592592593,
|
| 203 |
+
"eval_auc": 0.5136363636363637,
|
| 204 |
+
"eval_f1": 0.8631578947368421,
|
| 205 |
+
"eval_loss": 0.5203155279159546,
|
| 206 |
+
"eval_precision": 0.803921568627451,
|
| 207 |
+
"eval_recall": 0.9318181818181818,
|
| 208 |
+
"eval_runtime": 2.0799,
|
| 209 |
+
"eval_samples_per_second": 25.963,
|
| 210 |
+
"eval_steps_per_second": 3.366,
|
| 211 |
+
"step": 54
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 1.037037037037037,
|
| 215 |
+
"grad_norm": 17.741943359375,
|
| 216 |
+
"learning_rate": 4.9794238683127575e-05,
|
| 217 |
+
"loss": 0.6145,
|
| 218 |
+
"step": 56
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.074074074074074,
|
| 222 |
+
"grad_norm": 7.202783107757568,
|
| 223 |
+
"learning_rate": 4.958847736625515e-05,
|
| 224 |
+
"loss": 0.3385,
|
| 225 |
+
"step": 58
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 1.1111111111111112,
|
| 229 |
+
"grad_norm": 6.882026672363281,
|
| 230 |
+
"learning_rate": 4.938271604938271e-05,
|
| 231 |
+
"loss": 0.4704,
|
| 232 |
+
"step": 60
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 1.1481481481481481,
|
| 236 |
+
"grad_norm": 9.586283683776855,
|
| 237 |
+
"learning_rate": 4.9176954732510286e-05,
|
| 238 |
+
"loss": 0.2562,
|
| 239 |
+
"step": 62
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.1851851851851851,
|
| 243 |
+
"grad_norm": 11.778535842895508,
|
| 244 |
+
"learning_rate": 4.8971193415637865e-05,
|
| 245 |
+
"loss": 0.2755,
|
| 246 |
+
"step": 64
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 1.2222222222222223,
|
| 250 |
+
"grad_norm": 5.810285568237305,
|
| 251 |
+
"learning_rate": 4.876543209876544e-05,
|
| 252 |
+
"loss": 0.2591,
|
| 253 |
+
"step": 66
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 1.2592592592592593,
|
| 257 |
+
"grad_norm": 22.250574111938477,
|
| 258 |
+
"learning_rate": 4.855967078189301e-05,
|
| 259 |
+
"loss": 0.5711,
|
| 260 |
+
"step": 68
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 1.2962962962962963,
|
| 264 |
+
"grad_norm": 4.9341816902160645,
|
| 265 |
+
"learning_rate": 4.835390946502058e-05,
|
| 266 |
+
"loss": 0.1883,
|
| 267 |
+
"step": 70
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 1.3333333333333333,
|
| 271 |
+
"grad_norm": 3.9070465564727783,
|
| 272 |
+
"learning_rate": 4.814814814814815e-05,
|
| 273 |
+
"loss": 0.2058,
|
| 274 |
+
"step": 72
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 1.3703703703703702,
|
| 278 |
+
"grad_norm": 10.579364776611328,
|
| 279 |
+
"learning_rate": 4.794238683127572e-05,
|
| 280 |
+
"loss": 0.4203,
|
| 281 |
+
"step": 74
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.4074074074074074,
|
| 285 |
+
"grad_norm": 11.414931297302246,
|
| 286 |
+
"learning_rate": 4.773662551440329e-05,
|
| 287 |
+
"loss": 0.2615,
|
| 288 |
+
"step": 76
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 1.4444444444444444,
|
| 292 |
+
"grad_norm": 15.772526741027832,
|
| 293 |
+
"learning_rate": 4.7530864197530866e-05,
|
| 294 |
+
"loss": 0.7251,
|
| 295 |
+
"step": 78
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.4814814814814814,
|
| 299 |
+
"grad_norm": 13.619314193725586,
|
| 300 |
+
"learning_rate": 4.732510288065844e-05,
|
| 301 |
+
"loss": 0.3141,
|
| 302 |
+
"step": 80
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.5185185185185186,
|
| 306 |
+
"grad_norm": 20.24803924560547,
|
| 307 |
+
"learning_rate": 4.711934156378601e-05,
|
| 308 |
+
"loss": 0.5813,
|
| 309 |
+
"step": 82
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 1.5555555555555556,
|
| 313 |
+
"grad_norm": 11.862203598022461,
|
| 314 |
+
"learning_rate": 4.691358024691358e-05,
|
| 315 |
+
"loss": 0.3863,
|
| 316 |
+
"step": 84
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 1.5925925925925926,
|
| 320 |
+
"grad_norm": 12.273361206054688,
|
| 321 |
+
"learning_rate": 4.6707818930041156e-05,
|
| 322 |
+
"loss": 0.224,
|
| 323 |
+
"step": 86
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.6296296296296298,
|
| 327 |
+
"grad_norm": 3.214340925216675,
|
| 328 |
+
"learning_rate": 4.650205761316873e-05,
|
| 329 |
+
"loss": 0.1151,
|
| 330 |
+
"step": 88
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 1.6666666666666665,
|
| 334 |
+
"grad_norm": 12.43146800994873,
|
| 335 |
+
"learning_rate": 4.62962962962963e-05,
|
| 336 |
+
"loss": 0.3468,
|
| 337 |
+
"step": 90
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 1.7037037037037037,
|
| 341 |
+
"grad_norm": 3.3505775928497314,
|
| 342 |
+
"learning_rate": 4.609053497942387e-05,
|
| 343 |
+
"loss": 0.567,
|
| 344 |
+
"step": 92
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 1.7407407407407407,
|
| 348 |
+
"grad_norm": 7.76352596282959,
|
| 349 |
+
"learning_rate": 4.5884773662551446e-05,
|
| 350 |
+
"loss": 0.3589,
|
| 351 |
+
"step": 94
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 1.7777777777777777,
|
| 355 |
+
"grad_norm": 11.790644645690918,
|
| 356 |
+
"learning_rate": 4.567901234567901e-05,
|
| 357 |
+
"loss": 0.4001,
|
| 358 |
+
"step": 96
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 1.8148148148148149,
|
| 362 |
+
"grad_norm": 15.57990550994873,
|
| 363 |
+
"learning_rate": 4.5473251028806584e-05,
|
| 364 |
+
"loss": 0.9304,
|
| 365 |
+
"step": 98
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 1.8518518518518519,
|
| 369 |
+
"grad_norm": 13.815410614013672,
|
| 370 |
+
"learning_rate": 4.5267489711934157e-05,
|
| 371 |
+
"loss": 0.2938,
|
| 372 |
+
"step": 100
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 1.8888888888888888,
|
| 376 |
+
"grad_norm": 6.8774094581604,
|
| 377 |
+
"learning_rate": 4.506172839506173e-05,
|
| 378 |
+
"loss": 0.1514,
|
| 379 |
+
"step": 102
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 1.925925925925926,
|
| 383 |
+
"grad_norm": 9.06995964050293,
|
| 384 |
+
"learning_rate": 4.48559670781893e-05,
|
| 385 |
+
"loss": 0.1889,
|
| 386 |
+
"step": 104
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 1.9629629629629628,
|
| 390 |
+
"grad_norm": 9.242486953735352,
|
| 391 |
+
"learning_rate": 4.4650205761316874e-05,
|
| 392 |
+
"loss": 0.1605,
|
| 393 |
+
"step": 106
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 2.0,
|
| 397 |
+
"grad_norm": 16.733461380004883,
|
| 398 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 399 |
+
"loss": 0.263,
|
| 400 |
+
"step": 108
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 2.0,
|
| 404 |
+
"eval_accuracy": 0.7777777777777778,
|
| 405 |
+
"eval_auc": 0.5818181818181818,
|
| 406 |
+
"eval_f1": 0.8723404255319149,
|
| 407 |
+
"eval_loss": 0.5064903497695923,
|
| 408 |
+
"eval_precision": 0.82,
|
| 409 |
+
"eval_recall": 0.9318181818181818,
|
| 410 |
+
"eval_runtime": 2.3112,
|
| 411 |
+
"eval_samples_per_second": 23.365,
|
| 412 |
+
"eval_steps_per_second": 3.029,
|
| 413 |
+
"step": 108
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 2.037037037037037,
|
| 417 |
+
"grad_norm": 10.532149314880371,
|
| 418 |
+
"learning_rate": 4.423868312757202e-05,
|
| 419 |
+
"loss": 0.252,
|
| 420 |
+
"step": 110
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 2.074074074074074,
|
| 424 |
+
"grad_norm": 22.404003143310547,
|
| 425 |
+
"learning_rate": 4.403292181069959e-05,
|
| 426 |
+
"loss": 0.8314,
|
| 427 |
+
"step": 112
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 2.111111111111111,
|
| 431 |
+
"grad_norm": 5.451999187469482,
|
| 432 |
+
"learning_rate": 4.3827160493827164e-05,
|
| 433 |
+
"loss": 0.0909,
|
| 434 |
+
"step": 114
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.148148148148148,
|
| 438 |
+
"grad_norm": 5.845940113067627,
|
| 439 |
+
"learning_rate": 4.3621399176954737e-05,
|
| 440 |
+
"loss": 0.1798,
|
| 441 |
+
"step": 116
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 2.185185185185185,
|
| 445 |
+
"grad_norm": 3.3942580223083496,
|
| 446 |
+
"learning_rate": 4.341563786008231e-05,
|
| 447 |
+
"loss": 0.1201,
|
| 448 |
+
"step": 118
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 2.2222222222222223,
|
| 452 |
+
"grad_norm": 4.614056587219238,
|
| 453 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 454 |
+
"loss": 0.098,
|
| 455 |
+
"step": 120
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 2.259259259259259,
|
| 459 |
+
"grad_norm": 30.02410888671875,
|
| 460 |
+
"learning_rate": 4.300411522633745e-05,
|
| 461 |
+
"loss": 0.6446,
|
| 462 |
+
"step": 122
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 2.2962962962962963,
|
| 466 |
+
"grad_norm": 14.608016967773438,
|
| 467 |
+
"learning_rate": 4.279835390946502e-05,
|
| 468 |
+
"loss": 0.3173,
|
| 469 |
+
"step": 124
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 2.3333333333333335,
|
| 473 |
+
"grad_norm": 17.3658390045166,
|
| 474 |
+
"learning_rate": 4.259259259259259e-05,
|
| 475 |
+
"loss": 0.3116,
|
| 476 |
+
"step": 126
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.3703703703703702,
|
| 480 |
+
"grad_norm": 2.008209228515625,
|
| 481 |
+
"learning_rate": 4.2386831275720165e-05,
|
| 482 |
+
"loss": 0.0642,
|
| 483 |
+
"step": 128
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 2.4074074074074074,
|
| 487 |
+
"grad_norm": 2.245250940322876,
|
| 488 |
+
"learning_rate": 4.2181069958847744e-05,
|
| 489 |
+
"loss": 0.0924,
|
| 490 |
+
"step": 130
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 2.4444444444444446,
|
| 494 |
+
"grad_norm": 15.064545631408691,
|
| 495 |
+
"learning_rate": 4.197530864197531e-05,
|
| 496 |
+
"loss": 0.2853,
|
| 497 |
+
"step": 132
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 2.4814814814814814,
|
| 501 |
+
"grad_norm": 3.873878002166748,
|
| 502 |
+
"learning_rate": 4.176954732510288e-05,
|
| 503 |
+
"loss": 0.0772,
|
| 504 |
+
"step": 134
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 2.5185185185185186,
|
| 508 |
+
"grad_norm": 4.94734525680542,
|
| 509 |
+
"learning_rate": 4.1563786008230455e-05,
|
| 510 |
+
"loss": 0.1424,
|
| 511 |
+
"step": 136
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 2.5555555555555554,
|
| 515 |
+
"grad_norm": 3.7207255363464355,
|
| 516 |
+
"learning_rate": 4.135802469135803e-05,
|
| 517 |
+
"loss": 0.3227,
|
| 518 |
+
"step": 138
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 2.5925925925925926,
|
| 522 |
+
"grad_norm": 4.773040294647217,
|
| 523 |
+
"learning_rate": 4.11522633744856e-05,
|
| 524 |
+
"loss": 0.0844,
|
| 525 |
+
"step": 140
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 2.6296296296296298,
|
| 529 |
+
"grad_norm": 7.692733287811279,
|
| 530 |
+
"learning_rate": 4.094650205761317e-05,
|
| 531 |
+
"loss": 0.0881,
|
| 532 |
+
"step": 142
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 2.6666666666666665,
|
| 536 |
+
"grad_norm": 3.4888436794281006,
|
| 537 |
+
"learning_rate": 4.074074074074074e-05,
|
| 538 |
+
"loss": 0.0933,
|
| 539 |
+
"step": 144
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 2.7037037037037037,
|
| 543 |
+
"grad_norm": 23.996261596679688,
|
| 544 |
+
"learning_rate": 4.053497942386831e-05,
|
| 545 |
+
"loss": 1.0403,
|
| 546 |
+
"step": 146
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 2.7407407407407405,
|
| 550 |
+
"grad_norm": 5.3896050453186035,
|
| 551 |
+
"learning_rate": 4.032921810699588e-05,
|
| 552 |
+
"loss": 0.074,
|
| 553 |
+
"step": 148
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 2.7777777777777777,
|
| 557 |
+
"grad_norm": 18.857404708862305,
|
| 558 |
+
"learning_rate": 4.012345679012346e-05,
|
| 559 |
+
"loss": 0.3853,
|
| 560 |
+
"step": 150
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 2.814814814814815,
|
| 564 |
+
"grad_norm": 2.667506456375122,
|
| 565 |
+
"learning_rate": 3.9917695473251035e-05,
|
| 566 |
+
"loss": 0.056,
|
| 567 |
+
"step": 152
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 2.851851851851852,
|
| 571 |
+
"grad_norm": 24.5335693359375,
|
| 572 |
+
"learning_rate": 3.971193415637861e-05,
|
| 573 |
+
"loss": 0.8115,
|
| 574 |
+
"step": 154
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 2.888888888888889,
|
| 578 |
+
"grad_norm": 5.093173027038574,
|
| 579 |
+
"learning_rate": 3.950617283950617e-05,
|
| 580 |
+
"loss": 0.0812,
|
| 581 |
+
"step": 156
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 2.925925925925926,
|
| 585 |
+
"grad_norm": 2.674487829208374,
|
| 586 |
+
"learning_rate": 3.9300411522633746e-05,
|
| 587 |
+
"loss": 0.0673,
|
| 588 |
+
"step": 158
|
| 589 |
+
},
|
| 590 |
+
{
|
| 591 |
+
"epoch": 2.962962962962963,
|
| 592 |
+
"grad_norm": 12.101727485656738,
|
| 593 |
+
"learning_rate": 3.909465020576132e-05,
|
| 594 |
+
"loss": 0.4882,
|
| 595 |
+
"step": 160
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 3.0,
|
| 599 |
+
"grad_norm": 1.7577975988388062,
|
| 600 |
+
"learning_rate": 3.888888888888889e-05,
|
| 601 |
+
"loss": 0.7461,
|
| 602 |
+
"step": 162
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 3.0,
|
| 606 |
+
"eval_accuracy": 0.7777777777777778,
|
| 607 |
+
"eval_auc": 0.765909090909091,
|
| 608 |
+
"eval_f1": 0.85,
|
| 609 |
+
"eval_loss": 0.634387731552124,
|
| 610 |
+
"eval_precision": 0.9444444444444444,
|
| 611 |
+
"eval_recall": 0.7727272727272727,
|
| 612 |
+
"eval_runtime": 2.2065,
|
| 613 |
+
"eval_samples_per_second": 24.473,
|
| 614 |
+
"eval_steps_per_second": 3.172,
|
| 615 |
+
"step": 162
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 3.037037037037037,
|
| 619 |
+
"grad_norm": 2.8391764163970947,
|
| 620 |
+
"learning_rate": 3.868312757201646e-05,
|
| 621 |
+
"loss": 0.9041,
|
| 622 |
+
"step": 164
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 3.074074074074074,
|
| 626 |
+
"grad_norm": 1.376434564590454,
|
| 627 |
+
"learning_rate": 3.8477366255144036e-05,
|
| 628 |
+
"loss": 0.8279,
|
| 629 |
+
"step": 166
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 3.111111111111111,
|
| 633 |
+
"grad_norm": 1.2460780143737793,
|
| 634 |
+
"learning_rate": 3.82716049382716e-05,
|
| 635 |
+
"loss": 0.0379,
|
| 636 |
+
"step": 168
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 3.148148148148148,
|
| 640 |
+
"grad_norm": 3.860896110534668,
|
| 641 |
+
"learning_rate": 3.806584362139918e-05,
|
| 642 |
+
"loss": 0.0716,
|
| 643 |
+
"step": 170
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 3.185185185185185,
|
| 647 |
+
"grad_norm": 18.640295028686523,
|
| 648 |
+
"learning_rate": 3.786008230452675e-05,
|
| 649 |
+
"loss": 0.3966,
|
| 650 |
+
"step": 172
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 3.2222222222222223,
|
| 654 |
+
"grad_norm": 10.263703346252441,
|
| 655 |
+
"learning_rate": 3.7654320987654326e-05,
|
| 656 |
+
"loss": 0.1088,
|
| 657 |
+
"step": 174
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 3.259259259259259,
|
| 661 |
+
"grad_norm": 26.68462562561035,
|
| 662 |
+
"learning_rate": 3.74485596707819e-05,
|
| 663 |
+
"loss": 0.9791,
|
| 664 |
+
"step": 176
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 3.2962962962962963,
|
| 668 |
+
"grad_norm": 4.9114274978637695,
|
| 669 |
+
"learning_rate": 3.724279835390947e-05,
|
| 670 |
+
"loss": 0.1261,
|
| 671 |
+
"step": 178
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 3.3333333333333335,
|
| 675 |
+
"grad_norm": 3.88952898979187,
|
| 676 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 677 |
+
"loss": 0.0454,
|
| 678 |
+
"step": 180
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 3.3703703703703702,
|
| 682 |
+
"grad_norm": 3.1947505474090576,
|
| 683 |
+
"learning_rate": 3.683127572016461e-05,
|
| 684 |
+
"loss": 0.0586,
|
| 685 |
+
"step": 182
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 3.4074074074074074,
|
| 689 |
+
"grad_norm": 0.9005553126335144,
|
| 690 |
+
"learning_rate": 3.662551440329218e-05,
|
| 691 |
+
"loss": 0.0294,
|
| 692 |
+
"step": 184
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 3.4444444444444446,
|
| 696 |
+
"grad_norm": 1.0244694948196411,
|
| 697 |
+
"learning_rate": 3.6419753086419754e-05,
|
| 698 |
+
"loss": 0.0483,
|
| 699 |
+
"step": 186
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 3.4814814814814814,
|
| 703 |
+
"grad_norm": 1.7269971370697021,
|
| 704 |
+
"learning_rate": 3.6213991769547327e-05,
|
| 705 |
+
"loss": 0.1445,
|
| 706 |
+
"step": 188
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 3.5185185185185186,
|
| 710 |
+
"grad_norm": 24.60822868347168,
|
| 711 |
+
"learning_rate": 3.60082304526749e-05,
|
| 712 |
+
"loss": 0.5923,
|
| 713 |
+
"step": 190
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 3.5555555555555554,
|
| 717 |
+
"grad_norm": 18.20575523376465,
|
| 718 |
+
"learning_rate": 3.580246913580247e-05,
|
| 719 |
+
"loss": 0.4842,
|
| 720 |
+
"step": 192
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 3.5925925925925926,
|
| 724 |
+
"grad_norm": 18.561235427856445,
|
| 725 |
+
"learning_rate": 3.5596707818930044e-05,
|
| 726 |
+
"loss": 0.3398,
|
| 727 |
+
"step": 194
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 3.6296296296296298,
|
| 731 |
+
"grad_norm": 16.065948486328125,
|
| 732 |
+
"learning_rate": 3.539094650205762e-05,
|
| 733 |
+
"loss": 0.2199,
|
| 734 |
+
"step": 196
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 3.6666666666666665,
|
| 738 |
+
"grad_norm": 0.938701868057251,
|
| 739 |
+
"learning_rate": 3.518518518518519e-05,
|
| 740 |
+
"loss": 0.0764,
|
| 741 |
+
"step": 198
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 3.7037037037037037,
|
| 745 |
+
"grad_norm": 23.948251724243164,
|
| 746 |
+
"learning_rate": 3.497942386831276e-05,
|
| 747 |
+
"loss": 0.7831,
|
| 748 |
+
"step": 200
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 3.7407407407407405,
|
| 752 |
+
"grad_norm": 7.113195896148682,
|
| 753 |
+
"learning_rate": 3.4773662551440334e-05,
|
| 754 |
+
"loss": 0.0832,
|
| 755 |
+
"step": 202
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 3.7777777777777777,
|
| 759 |
+
"grad_norm": 4.066124439239502,
|
| 760 |
+
"learning_rate": 3.45679012345679e-05,
|
| 761 |
+
"loss": 0.0639,
|
| 762 |
+
"step": 204
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 3.814814814814815,
|
| 766 |
+
"grad_norm": 5.707618236541748,
|
| 767 |
+
"learning_rate": 3.436213991769547e-05,
|
| 768 |
+
"loss": 0.0549,
|
| 769 |
+
"step": 206
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 3.851851851851852,
|
| 773 |
+
"grad_norm": 0.38951224088668823,
|
| 774 |
+
"learning_rate": 3.4156378600823045e-05,
|
| 775 |
+
"loss": 0.0265,
|
| 776 |
+
"step": 208
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 3.888888888888889,
|
| 780 |
+
"grad_norm": 28.686899185180664,
|
| 781 |
+
"learning_rate": 3.395061728395062e-05,
|
| 782 |
+
"loss": 0.7846,
|
| 783 |
+
"step": 210
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 3.925925925925926,
|
| 787 |
+
"grad_norm": 1.2721365690231323,
|
| 788 |
+
"learning_rate": 3.374485596707819e-05,
|
| 789 |
+
"loss": 0.0313,
|
| 790 |
+
"step": 212
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 3.962962962962963,
|
| 794 |
+
"grad_norm": 4.036009311676025,
|
| 795 |
+
"learning_rate": 3.353909465020576e-05,
|
| 796 |
+
"loss": 0.1673,
|
| 797 |
+
"step": 214
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 4.0,
|
| 801 |
+
"grad_norm": 48.41086196899414,
|
| 802 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 803 |
+
"loss": 1.8,
|
| 804 |
+
"step": 216
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 4.0,
|
| 808 |
+
"eval_accuracy": 0.8148148148148148,
|
| 809 |
+
"eval_auc": 0.7136363636363636,
|
| 810 |
+
"eval_f1": 0.8837209302325582,
|
| 811 |
+
"eval_loss": 0.44954243302345276,
|
| 812 |
+
"eval_precision": 0.9047619047619048,
|
| 813 |
+
"eval_recall": 0.8636363636363636,
|
| 814 |
+
"eval_runtime": 2.0755,
|
| 815 |
+
"eval_samples_per_second": 26.018,
|
| 816 |
+
"eval_steps_per_second": 3.373,
|
| 817 |
+
"step": 216
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 4.037037037037037,
|
| 821 |
+
"grad_norm": 19.771888732910156,
|
| 822 |
+
"learning_rate": 3.312757201646091e-05,
|
| 823 |
+
"loss": 0.2898,
|
| 824 |
+
"step": 218
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 4.074074074074074,
|
| 828 |
+
"grad_norm": 1.0788342952728271,
|
| 829 |
+
"learning_rate": 3.292181069958848e-05,
|
| 830 |
+
"loss": 0.0248,
|
| 831 |
+
"step": 220
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 4.111111111111111,
|
| 835 |
+
"grad_norm": 17.729440689086914,
|
| 836 |
+
"learning_rate": 3.271604938271605e-05,
|
| 837 |
+
"loss": 0.1917,
|
| 838 |
+
"step": 222
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 4.148148148148148,
|
| 842 |
+
"grad_norm": 27.10383415222168,
|
| 843 |
+
"learning_rate": 3.2510288065843625e-05,
|
| 844 |
+
"loss": 0.7218,
|
| 845 |
+
"step": 224
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 4.185185185185185,
|
| 849 |
+
"grad_norm": 0.5653066635131836,
|
| 850 |
+
"learning_rate": 3.230452674897119e-05,
|
| 851 |
+
"loss": 0.1118,
|
| 852 |
+
"step": 226
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 4.222222222222222,
|
| 856 |
+
"grad_norm": 0.976607620716095,
|
| 857 |
+
"learning_rate": 3.209876543209876e-05,
|
| 858 |
+
"loss": 0.0155,
|
| 859 |
+
"step": 228
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 4.2592592592592595,
|
| 863 |
+
"grad_norm": 1.2977440357208252,
|
| 864 |
+
"learning_rate": 3.1893004115226336e-05,
|
| 865 |
+
"loss": 0.0213,
|
| 866 |
+
"step": 230
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 4.296296296296296,
|
| 870 |
+
"grad_norm": 2.7349634170532227,
|
| 871 |
+
"learning_rate": 3.168724279835391e-05,
|
| 872 |
+
"loss": 0.028,
|
| 873 |
+
"step": 232
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 4.333333333333333,
|
| 877 |
+
"grad_norm": 20.332021713256836,
|
| 878 |
+
"learning_rate": 3.148148148148148e-05,
|
| 879 |
+
"loss": 0.2017,
|
| 880 |
+
"step": 234
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 4.37037037037037,
|
| 884 |
+
"grad_norm": 2.7718288898468018,
|
| 885 |
+
"learning_rate": 3.127572016460906e-05,
|
| 886 |
+
"loss": 0.202,
|
| 887 |
+
"step": 236
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 4.407407407407407,
|
| 891 |
+
"grad_norm": 12.309640884399414,
|
| 892 |
+
"learning_rate": 3.1069958847736626e-05,
|
| 893 |
+
"loss": 0.1372,
|
| 894 |
+
"step": 238
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 4.444444444444445,
|
| 898 |
+
"grad_norm": 24.895954132080078,
|
| 899 |
+
"learning_rate": 3.08641975308642e-05,
|
| 900 |
+
"loss": 0.6885,
|
| 901 |
+
"step": 240
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 4.481481481481482,
|
| 905 |
+
"grad_norm": 26.356069564819336,
|
| 906 |
+
"learning_rate": 3.065843621399177e-05,
|
| 907 |
+
"loss": 0.4002,
|
| 908 |
+
"step": 242
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 4.518518518518518,
|
| 912 |
+
"grad_norm": 4.215313911437988,
|
| 913 |
+
"learning_rate": 3.0452674897119343e-05,
|
| 914 |
+
"loss": 0.4445,
|
| 915 |
+
"step": 244
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 4.555555555555555,
|
| 919 |
+
"grad_norm": 1.8379865884780884,
|
| 920 |
+
"learning_rate": 3.0246913580246916e-05,
|
| 921 |
+
"loss": 0.0547,
|
| 922 |
+
"step": 246
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"epoch": 4.592592592592593,
|
| 926 |
+
"grad_norm": 17.198705673217773,
|
| 927 |
+
"learning_rate": 3.0041152263374488e-05,
|
| 928 |
+
"loss": 0.2228,
|
| 929 |
+
"step": 248
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 4.62962962962963,
|
| 933 |
+
"grad_norm": 4.950334548950195,
|
| 934 |
+
"learning_rate": 2.9835390946502057e-05,
|
| 935 |
+
"loss": 0.4299,
|
| 936 |
+
"step": 250
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 4.666666666666667,
|
| 940 |
+
"grad_norm": 0.7647513747215271,
|
| 941 |
+
"learning_rate": 2.962962962962963e-05,
|
| 942 |
+
"loss": 0.0204,
|
| 943 |
+
"step": 252
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 4.703703703703704,
|
| 947 |
+
"grad_norm": 7.0774993896484375,
|
| 948 |
+
"learning_rate": 2.9423868312757202e-05,
|
| 949 |
+
"loss": 0.0711,
|
| 950 |
+
"step": 254
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 4.7407407407407405,
|
| 954 |
+
"grad_norm": 3.5618622303009033,
|
| 955 |
+
"learning_rate": 2.9218106995884775e-05,
|
| 956 |
+
"loss": 1.0106,
|
| 957 |
+
"step": 256
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 4.777777777777778,
|
| 961 |
+
"grad_norm": 19.40982437133789,
|
| 962 |
+
"learning_rate": 2.9012345679012347e-05,
|
| 963 |
+
"loss": 0.1852,
|
| 964 |
+
"step": 258
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 4.814814814814815,
|
| 968 |
+
"grad_norm": 1.7325761318206787,
|
| 969 |
+
"learning_rate": 2.880658436213992e-05,
|
| 970 |
+
"loss": 0.0338,
|
| 971 |
+
"step": 260
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 4.851851851851852,
|
| 975 |
+
"grad_norm": 0.40004846453666687,
|
| 976 |
+
"learning_rate": 2.860082304526749e-05,
|
| 977 |
+
"loss": 0.5516,
|
| 978 |
+
"step": 262
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 4.888888888888889,
|
| 982 |
+
"grad_norm": 24.856149673461914,
|
| 983 |
+
"learning_rate": 2.839506172839506e-05,
|
| 984 |
+
"loss": 0.3494,
|
| 985 |
+
"step": 264
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 4.925925925925926,
|
| 989 |
+
"grad_norm": 5.626762390136719,
|
| 990 |
+
"learning_rate": 2.8189300411522634e-05,
|
| 991 |
+
"loss": 0.1587,
|
| 992 |
+
"step": 266
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 4.962962962962963,
|
| 996 |
+
"grad_norm": 0.4980678856372833,
|
| 997 |
+
"learning_rate": 2.7983539094650207e-05,
|
| 998 |
+
"loss": 0.0123,
|
| 999 |
+
"step": 268
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 5.0,
|
| 1003 |
+
"grad_norm": 50.487037658691406,
|
| 1004 |
+
"learning_rate": 2.777777777777778e-05,
|
| 1005 |
+
"loss": 2.1749,
|
| 1006 |
+
"step": 270
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 5.0,
|
| 1010 |
+
"eval_accuracy": 0.8888888888888888,
|
| 1011 |
+
"eval_auc": 0.8340909090909091,
|
| 1012 |
+
"eval_f1": 0.9361702127659575,
|
| 1013 |
+
"eval_loss": 0.42044156789779663,
|
| 1014 |
+
"eval_precision": 0.88,
|
| 1015 |
+
"eval_recall": 1.0,
|
| 1016 |
+
"eval_runtime": 2.1391,
|
| 1017 |
+
"eval_samples_per_second": 25.244,
|
| 1018 |
+
"eval_steps_per_second": 3.272,
|
| 1019 |
+
"step": 270
|
| 1020 |
+
}
|
| 1021 |
+
],
|
| 1022 |
+
"logging_steps": 2,
|
| 1023 |
+
"max_steps": 540,
|
| 1024 |
+
"num_input_tokens_seen": 0,
|
| 1025 |
+
"num_train_epochs": 10,
|
| 1026 |
+
"save_steps": 500,
|
| 1027 |
+
"stateful_callbacks": {
|
| 1028 |
+
"EarlyStoppingCallback": {
|
| 1029 |
+
"args": {
|
| 1030 |
+
"early_stopping_patience": 5,
|
| 1031 |
+
"early_stopping_threshold": 0.01
|
| 1032 |
+
},
|
| 1033 |
+
"attributes": {
|
| 1034 |
+
"early_stopping_patience_counter": 0
|
| 1035 |
+
}
|
| 1036 |
+
},
|
| 1037 |
+
"TrainerControl": {
|
| 1038 |
+
"args": {
|
| 1039 |
+
"should_epoch_stop": false,
|
| 1040 |
+
"should_evaluate": false,
|
| 1041 |
+
"should_log": false,
|
| 1042 |
+
"should_save": true,
|
| 1043 |
+
"should_training_stop": false
|
| 1044 |
+
},
|
| 1045 |
+
"attributes": {}
|
| 1046 |
+
}
|
| 1047 |
+
},
|
| 1048 |
+
"total_flos": 2797124962222080.0,
|
| 1049 |
+
"train_batch_size": 4,
|
| 1050 |
+
"trial_name": null,
|
| 1051 |
+
"trial_params": null
|
| 1052 |
+
}
|
checkpoint-270/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6f45ce932e3634c82baf6a39ef3f7e175f87db12adc545872f811263f92c8c5b
|
| 3 |
+
size 5240
|
config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/mobilenet_v2_1.0_224",
|
| 3 |
+
"_num_labels": 2,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"MobileNetV2ForImageClassification"
|
| 6 |
+
],
|
| 7 |
+
"classifier_dropout_prob": 0.2,
|
| 8 |
+
"depth_divisible_by": 8,
|
| 9 |
+
"depth_multiplier": 1.0,
|
| 10 |
+
"expand_ratio": 6,
|
| 11 |
+
"finegrained_output": true,
|
| 12 |
+
"first_layer_is_expansion": true,
|
| 13 |
+
"hidden_act": "relu6",
|
| 14 |
+
"id2label": {
|
| 15 |
+
"0": "rotten banana",
|
| 16 |
+
"1": "yellow banana"
|
| 17 |
+
},
|
| 18 |
+
"image_size": 224,
|
| 19 |
+
"initializer_range": 0.02,
|
| 20 |
+
"label2id": {
|
| 21 |
+
"rotten banana": 0,
|
| 22 |
+
"yellow banana": 1
|
| 23 |
+
},
|
| 24 |
+
"layer_norm_eps": 0.001,
|
| 25 |
+
"min_depth": 8,
|
| 26 |
+
"model_type": "mobilenet_v2",
|
| 27 |
+
"num_channels": 3,
|
| 28 |
+
"output_stride": 32,
|
| 29 |
+
"problem_type": "single_label_classification",
|
| 30 |
+
"semantic_loss_ignore_index": 255,
|
| 31 |
+
"tf_padding": true,
|
| 32 |
+
"torch_dtype": "float32",
|
| 33 |
+
"transformers_version": "4.43.1"
|
| 34 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4f8aa363d171b4e520dac9d00149698714b8f5c90b1a6327e4b9f4d35fd97233
|
| 3 |
+
size 9080216
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"crop_size": {
|
| 3 |
+
"height": 224,
|
| 4 |
+
"width": 224
|
| 5 |
+
},
|
| 6 |
+
"do_center_crop": true,
|
| 7 |
+
"do_normalize": true,
|
| 8 |
+
"do_rescale": true,
|
| 9 |
+
"do_resize": true,
|
| 10 |
+
"image_mean": [
|
| 11 |
+
0.5,
|
| 12 |
+
0.5,
|
| 13 |
+
0.5
|
| 14 |
+
],
|
| 15 |
+
"image_processor_type": "MobileNetV2ImageProcessor",
|
| 16 |
+
"image_std": [
|
| 17 |
+
0.5,
|
| 18 |
+
0.5,
|
| 19 |
+
0.5
|
| 20 |
+
],
|
| 21 |
+
"resample": 2,
|
| 22 |
+
"rescale_factor": 0.00392156862745098,
|
| 23 |
+
"size": {
|
| 24 |
+
"shortest_edge": 256
|
| 25 |
+
}
|
| 26 |
+
}
|
runs/Jul31_13-24-57_r-chispidev-ia-banana-fwzg7678-62e49-8u7fh/events.out.tfevents.1722432298.r-chispidev-ia-banana-fwzg7678-62e49-8u7fh.147.0
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 sha256:9e89c9077970f28c17bf61127fcc6681a8df519665c3beb0e84aebd9e3eb4adb
|
| 3 |
+
size 67314
|
runs/Jul31_13-24-57_r-chispidev-ia-banana-fwzg7678-62e49-8u7fh/events.out.tfevents.1722432616.r-chispidev-ia-banana-fwzg7678-62e49-8u7fh.147.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4ec4b4dce9cc0921e923ab38fc55e0a1eb6a48ee0aeb146013f0a953071f1750
|
| 3 |
+
size 607
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6f45ce932e3634c82baf6a39ef3f7e175f87db12adc545872f811263f92c8c5b
|
| 3 |
+
size 5240
|
training_params.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"data_path": "autotrain-pky99-ias73/autotrain-data",
|
| 3 |
+
"model": "google/mobilenet_v2_1.0_224",
|
| 4 |
+
"username": "ChispiDEV",
|
| 5 |
+
"lr": 5e-05,
|
| 6 |
+
"epochs": 10,
|
| 7 |
+
"batch_size": 4,
|
| 8 |
+
"warmup_ratio": 0.1,
|
| 9 |
+
"gradient_accumulation": 1,
|
| 10 |
+
"optimizer": "adamw_torch",
|
| 11 |
+
"scheduler": "linear",
|
| 12 |
+
"weight_decay": 0.0,
|
| 13 |
+
"max_grad_norm": 1.0,
|
| 14 |
+
"seed": 42,
|
| 15 |
+
"train_split": "train",
|
| 16 |
+
"valid_split": "validation",
|
| 17 |
+
"logging_steps": -1,
|
| 18 |
+
"project_name": "autotrain-pky99-ias73",
|
| 19 |
+
"auto_find_batch_size": false,
|
| 20 |
+
"mixed_precision": "fp16",
|
| 21 |
+
"save_total_limit": 1,
|
| 22 |
+
"push_to_hub": true,
|
| 23 |
+
"eval_strategy": "epoch",
|
| 24 |
+
"image_column": "autotrain_image",
|
| 25 |
+
"target_column": "autotrain_label",
|
| 26 |
+
"log": "tensorboard",
|
| 27 |
+
"early_stopping_patience": 5,
|
| 28 |
+
"early_stopping_threshold": 0.01
|
| 29 |
+
}
|