Hugging Face's logo

Spaces:
MahatirTusher
/
PneumoInsight
Sleeping

App Files Community
mahatirahmedtusher commited on
Commit
db577ac
·
1 Parent(s): f7c34e2

Initial commit

Browse files
Files changed (13) hide show
  1. Examples/Bacterial.jpeg +0 -0
  2. Examples/Normal.jpeg +0 -0
  3. Examples/Viral.jpeg +0 -0
  4. app.py +155 -0
  5. models/config.json +34 -0
  6. models/model.safetensors +3 -0
  7. models/optimizer.pt +3 -0
  8. models/preprocessor_config.json +22 -0
  9. models/rng_state.pth +3 -0
  10. models/scheduler.pt +3 -0
  11. models/trainer_state.json +1551 -0
  12. models/training_args.bin +3 -0
  13. requirements.txt +7 -0
Examples/Bacterial.jpeg ADDED
Examples/Normal.jpeg ADDED
Examples/Viral.jpeg ADDED
app.py ADDED
@@ -0,0 +1,155 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import gradio as gr
3
+ from PIL import Image
4
+ import torch
5
+ from transformers import ViTForImageClassification, ViTImageProcessor
6
+ from datasets import load_dataset
7
+ import matplotlib.pyplot as plt
8
+ import numpy as np
9
+ import cv2
10
+
11
+ # Model and processor configuration
12
+ model_name_or_path = "google/vit-base-patch16-224-in21k"
13
+ processor = ViTImageProcessor.from_pretrained(model_name_or_path)
14
+
15
+ # Load dataset (adjust dataset_path accordingly)
16
+ dataset_path = "pawlo2013/chest_xray"
17
+ train_dataset = load_dataset(dataset_path, split="train")
18
+ class_names = train_dataset.features["label"].names
19
+
20
+ # Load ViT model
21
+ model = ViTForImageClassification.from_pretrained(
22
+ "./models",
23
+ num_labels=len(class_names),
24
+ id2label={str(i): label for i, label in enumerate(class_names)},
25
+ label2id={label: i for i, label in enumerate(class_names)},
26
+ )
27
+
28
+ # Set model to evaluation mode
29
+ model.eval()
30
+
31
+
32
+ # Define the classification function
33
+ def classify_and_visualize(img, device="cpu", discard_ratio=0.9, head_fusion="mean"):
34
+ img = img.convert("RGB")
35
+ processed_input = processor(images=img, return_tensors="pt").to(device)
36
+
37
+ processed_input = processed_input["pixel_values"].to(device)
38
+
39
+ with torch.no_grad():
40
+ outputs = model(processed_input, output_attentions=True)
41
+ logits = outputs.logits
42
+ probabilities = torch.softmax(logits, dim=1)[0].tolist()
43
+ prediction = torch.argmax(logits, dim=-1).item()
44
+ predicted_class = class_names[prediction]
45
+
46
+ result = {class_name: prob for class_name, prob in zip(class_names, probabilities)}
47
+
48
+ # Generate attention heatmap
49
+ heatmap_img = show_final_layer_attention_maps(
50
+ outputs, processed_input, device, discard_ratio, head_fusion
51
+ )
52
+
53
+ return {"probabilities": result, "heatmap": heatmap_img}
54
+
55
+
56
+ def format_output(output):
57
+ return (output["probabilities"], output["heatmap"])
58
+
59
+
60
+ # Function to load examples from a folder
61
+ def load_examples_from_folder(folder_path):
62
+ examples = []
63
+ for file in os.listdir(folder_path):
64
+ if file.endswith((".png", ".jpg", ".jpeg")):
65
+ examples.append(Image.open(os.path.join(folder_path, file)))
66
+ return examples
67
+
68
+
69
+ # Function to show final layer attention maps
70
+ def show_final_layer_attention_maps(
71
+ outputs,
72
+ processed_input,
73
+ device,
74
+ discard_ratio=0.6,
75
+ head_fusion="max",
76
+ only_last_layer=False,
77
+ ):
78
+
79
+ with torch.no_grad():
80
+
81
+ image = processed_input.squeeze(0)
82
+
83
+ image = image - image.min()
84
+ image = image / image.max()
85
+
86
+ result = torch.eye(outputs.attentions[0].size(-1)).to(device)
87
+ if only_last_layer:
88
+ attention_list = outputs.attentions[-1].unsqueeze(0).to(device)
89
+ else:
90
+ attention_list = outputs.attentions
91
+
92
+ for attention in attention_list:
93
+ if head_fusion == "mean":
94
+ attention_heads_fused = attention.mean(axis=1)
95
+ elif head_fusion == "max":
96
+ attention_heads_fused = attention.max(axis=1)[0]
97
+ elif head_fusion == "min":
98
+ attention_heads_fused = attention.min(axis=1)[0]
99
+
100
+ flat = attention_heads_fused.view(attention_heads_fused.size(0), -1)
101
+ _, indices = flat.topk(int(flat.size(-1) * discard_ratio), -1, False)
102
+ indices = indices[indices != 0]
103
+ flat[0, indices] = 0
104
+
105
+ I = torch.eye(attention_heads_fused.size(-1)).to(device)
106
+ a = (attention_heads_fused + 1.0 * I) / 2
107
+ a = a / a.sum(dim=-1)
108
+
109
+ result = torch.matmul(a, result)
110
+
111
+ mask = result[0, 0, 1:]
112
+ width = int(mask.size(-1) ** 0.5)
113
+ mask = mask.reshape(width, width).cpu().numpy()
114
+ mask = mask / np.max(mask)
115
+
116
+ mask = cv2.resize(mask, (224, 224))
117
+
118
+ mask = (mask - np.min(mask)) / (np.max(mask) - np.min(mask))
119
+ heatmap = plt.cm.jet(mask)[:, :, :3]
120
+
121
+ showed_img = image.permute(1, 2, 0).detach().cpu().numpy()
122
+ showed_img = (showed_img - np.min(showed_img)) / (
123
+ np.max(showed_img) - np.min(showed_img)
124
+ )
125
+ superimposed_img = heatmap * 0.4 + showed_img * 0.6
126
+
127
+ superimposed_img_pil = Image.fromarray(
128
+ (superimposed_img * 255).astype(np.uint8)
129
+ )
130
+
131
+ return superimposed_img_pil
132
+
133
+
134
+ # Define the path to the examples folder
135
+ examples_folder = "./examples"
136
+ examples = load_examples_from_folder(examples_folder)
137
+
138
+ # Create the Gradio interface
139
+ iface = gr.Interface(
140
+ fn=lambda img: format_output(classify_and_visualize(img)),
141
+ inputs=gr.Image(type="pil", label="Upload X-Ray Image"),
142
+ outputs=[
143
+ gr.Label(),
144
+ gr.Image(label="Attention Heatmap"),
145
+ ],
146
+ examples=examples,
147
+ cache_examples=False,
148
+ allow_flagging=False,
149
+ concurrency_limit=1,
150
+ title="Pneumonia X-Ray 3-Class Classification with Vision Transformer (ViT) using data augmentation",
151
+ description="Upload an X-ray image to classify it as normal, viral or bacterial pneumonia. Checkout the model in more details [here](https://huggingface.co/pawlo2013/vit-pneumonia-x-ray_3_class). The examples presented are taken from the test set of [Kermany et al. (2018) dataset.](https://data.mendeley.com/datasets/rscbjbr9sj/2.) The attention heatmap over all layers of the transfomer done by the attention rollout techinique by the implementation of [jacobgil](https://github.com/jacobgil/vit-explain).",
152
+ )
153
+ # Launch the app
154
+ if __name__ == "__main__":
155
+ iface.launch(debug=True)
models/config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "google/vit-base-patch16-224-in21k",
3
+ "architectures": [
4
+ "ViTForImageClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.0,
7
+ "encoder_stride": 16,
8
+ "hidden_act": "gelu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 768,
11
+ "id2label": {
12
+ "0": "LABEL_0",
13
+ "1": "LABEL_1",
14
+ "2": "LABEL_2"
15
+ },
16
+ "image_size": 224,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 3072,
19
+ "label2id": {
20
+ "LABEL_0": 0,
21
+ "LABEL_1": 1,
22
+ "LABEL_2": 2
23
+ },
24
+ "layer_norm_eps": 1e-12,
25
+ "model_type": "vit",
26
+ "num_attention_heads": 12,
27
+ "num_channels": 3,
28
+ "num_hidden_layers": 12,
29
+ "patch_size": 16,
30
+ "problem_type": "single_label_classification",
31
+ "qkv_bias": true,
32
+ "torch_dtype": "float32",
33
+ "transformers_version": "4.38.1"
34
+ }
models/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7259957cb4dd213d39e5d536abde8da37aa08cf56c729541f5e87a5fa6faa33f
3
+ size 343227052
models/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e273e892c8cb94bf00a5d414bc46df66f9f53bd7359d3572b918f22b98ce4b93
3
+ size 56750370
models/preprocessor_config.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "do_rescale": true,
4
+ "do_resize": true,
5
+ "image_mean": [
6
+ 0.5,
7
+ 0.5,
8
+ 0.5
9
+ ],
10
+ "image_processor_type": "ViTImageProcessor",
11
+ "image_std": [
12
+ 0.5,
13
+ 0.5,
14
+ 0.5
15
+ ],
16
+ "resample": 2,
17
+ "rescale_factor": 0.00392156862745098,
18
+ "size": {
19
+ "height": 224,
20
+ "width": 224
21
+ }
22
+ }
models/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e61ed5a12974ddc84aba2cf0e3ebe942a9f431b3349c3ae1f29a37ab202683c4
3
+ size 13990
models/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:665e0ada3e15db3564e43a6ff9d215f31f6a3ac1540edfe099d0df6966ec7aa8
3
+ size 1064
models/trainer_state.json ADDED
@@ -0,0 +1,1551 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.7440524101257324,
3
+ "best_model_checkpoint": "./vit-pneumonia-x-ray_data_augmentation_frozen_model/checkpoint-1800",
4
+ "epoch": 12.32876712328767,
5
+ "eval_steps": 100,
6
+ "global_step": 1800,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.07,
13
+ "grad_norm": 1.2999861240386963,
14
+ "learning_rate": 0.0009986301369863013,
15
+ "loss": 0.9052,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.14,
20
+ "grad_norm": 1.1751593351364136,
21
+ "learning_rate": 0.0009972602739726027,
22
+ "loss": 0.7342,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.21,
27
+ "grad_norm": 1.3162366151809692,
28
+ "learning_rate": 0.000995890410958904,
29
+ "loss": 0.6777,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.27,
34
+ "grad_norm": 1.1881186962127686,
35
+ "learning_rate": 0.0009945205479452055,
36
+ "loss": 0.6455,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.34,
41
+ "grad_norm": 4.0501508712768555,
42
+ "learning_rate": 0.0009931506849315068,
43
+ "loss": 0.6433,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.41,
48
+ "grad_norm": 2.4241373538970947,
49
+ "learning_rate": 0.0009917808219178082,
50
+ "loss": 0.6633,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.48,
55
+ "grad_norm": 1.7584506273269653,
56
+ "learning_rate": 0.0009904109589041095,
57
+ "loss": 0.5253,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.55,
62
+ "grad_norm": 1.115493893623352,
63
+ "learning_rate": 0.000989041095890411,
64
+ "loss": 0.5291,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.62,
69
+ "grad_norm": 1.0183758735656738,
70
+ "learning_rate": 0.0009876712328767123,
71
+ "loss": 0.5742,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.68,
76
+ "grad_norm": 1.7618106603622437,
77
+ "learning_rate": 0.0009863013698630137,
78
+ "loss": 0.5098,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.68,
83
+ "eval_accuracy": 0.7198443579766537,
84
+ "eval_accuracy_class_Bacterial": 0.84375,
85
+ "eval_accuracy_class_Normal": 0.7967479674796748,
86
+ "eval_accuracy_class_Viral": 0.4148148148148148,
87
+ "eval_f1": 0.7061270215472498,
88
+ "eval_loss": 0.8028925657272339,
89
+ "eval_precision": 0.727092782714888,
90
+ "eval_recall": 0.7198443579766537,
91
+ "eval_runtime": 42.6101,
92
+ "eval_samples_per_second": 12.063,
93
+ "eval_steps_per_second": 0.399,
94
+ "step": 100
95
+ },
96
+ {
97
+ "epoch": 0.75,
98
+ "grad_norm": 0.9144254326820374,
99
+ "learning_rate": 0.000984931506849315,
100
+ "loss": 0.5965,
101
+ "step": 110
102
+ },
103
+ {
104
+ "epoch": 0.82,
105
+ "grad_norm": 0.9542173743247986,
106
+ "learning_rate": 0.0009835616438356163,
107
+ "loss": 0.5742,
108
+ "step": 120
109
+ },
110
+ {
111
+ "epoch": 0.89,
112
+ "grad_norm": 1.90875244140625,
113
+ "learning_rate": 0.0009821917808219179,
114
+ "loss": 0.5374,
115
+ "step": 130
116
+ },
117
+ {
118
+ "epoch": 0.96,
119
+ "grad_norm": 0.8156168460845947,
120
+ "learning_rate": 0.0009808219178082192,
121
+ "loss": 0.5118,
122
+ "step": 140
123
+ },
124
+ {
125
+ "epoch": 1.03,
126
+ "grad_norm": 1.6298660039901733,
127
+ "learning_rate": 0.0009794520547945205,
128
+ "loss": 0.5168,
129
+ "step": 150
130
+ },
131
+ {
132
+ "epoch": 1.1,
133
+ "grad_norm": 0.5898642539978027,
134
+ "learning_rate": 0.0009780821917808218,
135
+ "loss": 0.5665,
136
+ "step": 160
137
+ },
138
+ {
139
+ "epoch": 1.16,
140
+ "grad_norm": 1.1714237928390503,
141
+ "learning_rate": 0.0009767123287671234,
142
+ "loss": 0.5061,
143
+ "step": 170
144
+ },
145
+ {
146
+ "epoch": 1.23,
147
+ "grad_norm": 1.8745484352111816,
148
+ "learning_rate": 0.0009753424657534247,
149
+ "loss": 0.4868,
150
+ "step": 180
151
+ },
152
+ {
153
+ "epoch": 1.3,
154
+ "grad_norm": 0.7712134718894958,
155
+ "learning_rate": 0.0009739726027397261,
156
+ "loss": 0.5125,
157
+ "step": 190
158
+ },
159
+ {
160
+ "epoch": 1.37,
161
+ "grad_norm": 0.753567099571228,
162
+ "learning_rate": 0.0009726027397260274,
163
+ "loss": 0.4609,
164
+ "step": 200
165
+ },
166
+ {
167
+ "epoch": 1.37,
168
+ "eval_accuracy": 0.7509727626459144,
169
+ "eval_accuracy_class_Bacterial": 0.8125,
170
+ "eval_accuracy_class_Normal": 0.7804878048780488,
171
+ "eval_accuracy_class_Viral": 0.6074074074074074,
172
+ "eval_f1": 0.7503972489190844,
173
+ "eval_loss": 0.759770393371582,
174
+ "eval_precision": 0.753673142519536,
175
+ "eval_recall": 0.7509727626459144,
176
+ "eval_runtime": 38.2089,
177
+ "eval_samples_per_second": 13.452,
178
+ "eval_steps_per_second": 0.445,
179
+ "step": 200
180
+ },
181
+ {
182
+ "epoch": 1.44,
183
+ "grad_norm": 0.7043547034263611,
184
+ "learning_rate": 0.0009712328767123288,
185
+ "loss": 0.4749,
186
+ "step": 210
187
+ },
188
+ {
189
+ "epoch": 1.51,
190
+ "grad_norm": 0.5875887274742126,
191
+ "learning_rate": 0.0009698630136986302,
192
+ "loss": 0.4831,
193
+ "step": 220
194
+ },
195
+ {
196
+ "epoch": 1.58,
197
+ "grad_norm": 0.5488404035568237,
198
+ "learning_rate": 0.0009684931506849315,
199
+ "loss": 0.5314,
200
+ "step": 230
201
+ },
202
+ {
203
+ "epoch": 1.64,
204
+ "grad_norm": 1.2711101770401,
205
+ "learning_rate": 0.0009671232876712329,
206
+ "loss": 0.5463,
207
+ "step": 240
208
+ },
209
+ {
210
+ "epoch": 1.71,
211
+ "grad_norm": 0.31795018911361694,
212
+ "learning_rate": 0.0009657534246575343,
213
+ "loss": 0.5109,
214
+ "step": 250
215
+ },
216
+ {
217
+ "epoch": 1.78,
218
+ "grad_norm": 1.2750917673110962,
219
+ "learning_rate": 0.0009643835616438357,
220
+ "loss": 0.5053,
221
+ "step": 260
222
+ },
223
+ {
224
+ "epoch": 1.85,
225
+ "grad_norm": 0.9569109678268433,
226
+ "learning_rate": 0.000963013698630137,
227
+ "loss": 0.4965,
228
+ "step": 270
229
+ },
230
+ {
231
+ "epoch": 1.92,
232
+ "grad_norm": 0.9257662892341614,
233
+ "learning_rate": 0.0009616438356164384,
234
+ "loss": 0.5175,
235
+ "step": 280
236
+ },
237
+ {
238
+ "epoch": 1.99,
239
+ "grad_norm": 0.9531387686729431,
240
+ "learning_rate": 0.0009602739726027398,
241
+ "loss": 0.4751,
242
+ "step": 290
243
+ },
244
+ {
245
+ "epoch": 2.05,
246
+ "grad_norm": 1.1072851419448853,
247
+ "learning_rate": 0.0009589041095890411,
248
+ "loss": 0.482,
249
+ "step": 300
250
+ },
251
+ {
252
+ "epoch": 2.05,
253
+ "eval_accuracy": 0.7392996108949417,
254
+ "eval_accuracy_class_Bacterial": 0.7265625,
255
+ "eval_accuracy_class_Normal": 0.7804878048780488,
256
+ "eval_accuracy_class_Viral": 0.725925925925926,
257
+ "eval_f1": 0.7441369344372653,
258
+ "eval_loss": 0.7680220603942871,
259
+ "eval_precision": 0.7562073779920531,
260
+ "eval_recall": 0.7392996108949417,
261
+ "eval_runtime": 39.4785,
262
+ "eval_samples_per_second": 13.02,
263
+ "eval_steps_per_second": 0.431,
264
+ "step": 300
265
+ },
266
+ {
267
+ "epoch": 2.12,
268
+ "grad_norm": 0.49826326966285706,
269
+ "learning_rate": 0.0009575342465753425,
270
+ "loss": 0.5238,
271
+ "step": 310
272
+ },
273
+ {
274
+ "epoch": 2.19,
275
+ "grad_norm": 2.0074150562286377,
276
+ "learning_rate": 0.0009561643835616438,
277
+ "loss": 0.5213,
278
+ "step": 320
279
+ },
280
+ {
281
+ "epoch": 2.26,
282
+ "grad_norm": 0.6424590945243835,
283
+ "learning_rate": 0.0009547945205479453,
284
+ "loss": 0.4442,
285
+ "step": 330
286
+ },
287
+ {
288
+ "epoch": 2.33,
289
+ "grad_norm": 0.5362870097160339,
290
+ "learning_rate": 0.0009534246575342466,
291
+ "loss": 0.4302,
292
+ "step": 340
293
+ },
294
+ {
295
+ "epoch": 2.4,
296
+ "grad_norm": 0.8034541010856628,
297
+ "learning_rate": 0.000952054794520548,
298
+ "loss": 0.4882,
299
+ "step": 350
300
+ },
301
+ {
302
+ "epoch": 2.47,
303
+ "grad_norm": 0.7750194072723389,
304
+ "learning_rate": 0.0009506849315068493,
305
+ "loss": 0.4668,
306
+ "step": 360
307
+ },
308
+ {
309
+ "epoch": 2.53,
310
+ "grad_norm": 0.5913398265838623,
311
+ "learning_rate": 0.0009493150684931508,
312
+ "loss": 0.4412,
313
+ "step": 370
314
+ },
315
+ {
316
+ "epoch": 2.6,
317
+ "grad_norm": 2.182452917098999,
318
+ "learning_rate": 0.0009479452054794521,
319
+ "loss": 0.492,
320
+ "step": 380
321
+ },
322
+ {
323
+ "epoch": 2.67,
324
+ "grad_norm": 0.4750209450721741,
325
+ "learning_rate": 0.0009465753424657535,
326
+ "loss": 0.4839,
327
+ "step": 390
328
+ },
329
+ {
330
+ "epoch": 2.74,
331
+ "grad_norm": 0.8099997639656067,
332
+ "learning_rate": 0.0009452054794520548,
333
+ "loss": 0.4513,
334
+ "step": 400
335
+ },
336
+ {
337
+ "epoch": 2.74,
338
+ "eval_accuracy": 0.7587548638132295,
339
+ "eval_accuracy_class_Bacterial": 0.8203125,
340
+ "eval_accuracy_class_Normal": 0.8048780487804879,
341
+ "eval_accuracy_class_Viral": 0.6,
342
+ "eval_f1": 0.7561734408999475,
343
+ "eval_loss": 0.7605456709861755,
344
+ "eval_precision": 0.7582411293175139,
345
+ "eval_recall": 0.7587548638132295,
346
+ "eval_runtime": 38.7682,
347
+ "eval_samples_per_second": 13.258,
348
+ "eval_steps_per_second": 0.439,
349
+ "step": 400
350
+ },
351
+ {
352
+ "epoch": 2.81,
353
+ "grad_norm": 0.31849750876426697,
354
+ "learning_rate": 0.0009438356164383562,
355
+ "loss": 0.4884,
356
+ "step": 410
357
+ },
358
+ {
359
+ "epoch": 2.88,
360
+ "grad_norm": 0.7159171104431152,
361
+ "learning_rate": 0.0009424657534246576,
362
+ "loss": 0.492,
363
+ "step": 420
364
+ },
365
+ {
366
+ "epoch": 2.95,
367
+ "grad_norm": 0.706955075263977,
368
+ "learning_rate": 0.0009410958904109589,
369
+ "loss": 0.4354,
370
+ "step": 430
371
+ },
372
+ {
373
+ "epoch": 3.01,
374
+ "grad_norm": 0.8797878623008728,
375
+ "learning_rate": 0.0009397260273972603,
376
+ "loss": 0.4362,
377
+ "step": 440
378
+ },
379
+ {
380
+ "epoch": 3.08,
381
+ "grad_norm": 1.2229899168014526,
382
+ "learning_rate": 0.0009383561643835617,
383
+ "loss": 0.5546,
384
+ "step": 450
385
+ },
386
+ {
387
+ "epoch": 3.15,
388
+ "grad_norm": 0.4625228941440582,
389
+ "learning_rate": 0.0009369863013698631,
390
+ "loss": 0.4412,
391
+ "step": 460
392
+ },
393
+ {
394
+ "epoch": 3.22,
395
+ "grad_norm": 0.7467411160469055,
396
+ "learning_rate": 0.0009356164383561644,
397
+ "loss": 0.4433,
398
+ "step": 470
399
+ },
400
+ {
401
+ "epoch": 3.29,
402
+ "grad_norm": 0.7149803042411804,
403
+ "learning_rate": 0.0009342465753424658,
404
+ "loss": 0.3969,
405
+ "step": 480
406
+ },
407
+ {
408
+ "epoch": 3.36,
409
+ "grad_norm": 1.1275142431259155,
410
+ "learning_rate": 0.0009328767123287672,
411
+ "loss": 0.4433,
412
+ "step": 490
413
+ },
414
+ {
415
+ "epoch": 3.42,
416
+ "grad_norm": 0.5531851649284363,
417
+ "learning_rate": 0.0009315068493150685,
418
+ "loss": 0.4367,
419
+ "step": 500
420
+ },
421
+ {
422
+ "epoch": 3.42,
423
+ "eval_accuracy": 0.7607003891050583,
424
+ "eval_accuracy_class_Bacterial": 0.78125,
425
+ "eval_accuracy_class_Normal": 0.7967479674796748,
426
+ "eval_accuracy_class_Viral": 0.6888888888888889,
427
+ "eval_f1": 0.7612120105726318,
428
+ "eval_loss": 0.7907313108444214,
429
+ "eval_precision": 0.7620375478182957,
430
+ "eval_recall": 0.7607003891050583,
431
+ "eval_runtime": 39.4844,
432
+ "eval_samples_per_second": 13.018,
433
+ "eval_steps_per_second": 0.431,
434
+ "step": 500
435
+ },
436
+ {
437
+ "epoch": 3.49,
438
+ "grad_norm": 1.5219345092773438,
439
+ "learning_rate": 0.0009301369863013699,
440
+ "loss": 0.4563,
441
+ "step": 510
442
+ },
443
+ {
444
+ "epoch": 3.56,
445
+ "grad_norm": 0.6871808171272278,
446
+ "learning_rate": 0.0009287671232876712,
447
+ "loss": 0.4929,
448
+ "step": 520
449
+ },
450
+ {
451
+ "epoch": 3.63,
452
+ "grad_norm": 0.7525418996810913,
453
+ "learning_rate": 0.0009273972602739727,
454
+ "loss": 0.4474,
455
+ "step": 530
456
+ },
457
+ {
458
+ "epoch": 3.7,
459
+ "grad_norm": 0.3728916049003601,
460
+ "learning_rate": 0.000926027397260274,
461
+ "loss": 0.4444,
462
+ "step": 540
463
+ },
464
+ {
465
+ "epoch": 3.77,
466
+ "grad_norm": 0.6982372999191284,
467
+ "learning_rate": 0.0009246575342465754,
468
+ "loss": 0.394,
469
+ "step": 550
470
+ },
471
+ {
472
+ "epoch": 3.84,
473
+ "grad_norm": 0.6600722670555115,
474
+ "learning_rate": 0.0009232876712328767,
475
+ "loss": 0.5548,
476
+ "step": 560
477
+ },
478
+ {
479
+ "epoch": 3.9,
480
+ "grad_norm": 0.5913540720939636,
481
+ "learning_rate": 0.0009219178082191782,
482
+ "loss": 0.4652,
483
+ "step": 570
484
+ },
485
+ {
486
+ "epoch": 3.97,
487
+ "grad_norm": 0.2932947874069214,
488
+ "learning_rate": 0.0009205479452054795,
489
+ "loss": 0.3682,
490
+ "step": 580
491
+ },
492
+ {
493
+ "epoch": 4.04,
494
+ "grad_norm": 0.6950840950012207,
495
+ "learning_rate": 0.0009191780821917809,
496
+ "loss": 0.4862,
497
+ "step": 590
498
+ },
499
+ {
500
+ "epoch": 4.11,
501
+ "grad_norm": 0.7572476863861084,
502
+ "learning_rate": 0.0009178082191780823,
503
+ "loss": 0.4134,
504
+ "step": 600
505
+ },
506
+ {
507
+ "epoch": 4.11,
508
+ "eval_accuracy": 0.7587548638132295,
509
+ "eval_accuracy_class_Bacterial": 0.7734375,
510
+ "eval_accuracy_class_Normal": 0.7560975609756098,
511
+ "eval_accuracy_class_Viral": 0.7333333333333333,
512
+ "eval_f1": 0.762153811730061,
513
+ "eval_loss": 0.7795141935348511,
514
+ "eval_precision": 0.7718812461468243,
515
+ "eval_recall": 0.7587548638132295,
516
+ "eval_runtime": 39.1561,
517
+ "eval_samples_per_second": 13.127,
518
+ "eval_steps_per_second": 0.434,
519
+ "step": 600
520
+ },
521
+ {
522
+ "epoch": 4.18,
523
+ "grad_norm": 1.036367416381836,
524
+ "learning_rate": 0.0009164383561643836,
525
+ "loss": 0.3799,
526
+ "step": 610
527
+ },
528
+ {
529
+ "epoch": 4.25,
530
+ "grad_norm": 1.7970045804977417,
531
+ "learning_rate": 0.000915068493150685,
532
+ "loss": 0.4556,
533
+ "step": 620
534
+ },
535
+ {
536
+ "epoch": 4.32,
537
+ "grad_norm": 0.6464496850967407,
538
+ "learning_rate": 0.0009136986301369863,
539
+ "loss": 0.3978,
540
+ "step": 630
541
+ },
542
+ {
543
+ "epoch": 4.38,
544
+ "grad_norm": 1.043805480003357,
545
+ "learning_rate": 0.0009123287671232878,
546
+ "loss": 0.3975,
547
+ "step": 640
548
+ },
549
+ {
550
+ "epoch": 4.45,
551
+ "grad_norm": 0.627406120300293,
552
+ "learning_rate": 0.0009109589041095891,
553
+ "loss": 0.4197,
554
+ "step": 650
555
+ },
556
+ {
557
+ "epoch": 4.52,
558
+ "grad_norm": 0.669355034828186,
559
+ "learning_rate": 0.0009095890410958905,
560
+ "loss": 0.5468,
561
+ "step": 660
562
+ },
563
+ {
564
+ "epoch": 4.59,
565
+ "grad_norm": 0.7727690935134888,
566
+ "learning_rate": 0.0009082191780821918,
567
+ "loss": 0.506,
568
+ "step": 670
569
+ },
570
+ {
571
+ "epoch": 4.66,
572
+ "grad_norm": 0.4808361828327179,
573
+ "learning_rate": 0.0009068493150684933,
574
+ "loss": 0.4329,
575
+ "step": 680
576
+ },
577
+ {
578
+ "epoch": 4.73,
579
+ "grad_norm": 0.9221294522285461,
580
+ "learning_rate": 0.0009054794520547946,
581
+ "loss": 0.3699,
582
+ "step": 690
583
+ },
584
+ {
585
+ "epoch": 4.79,
586
+ "grad_norm": 0.6050639152526855,
587
+ "learning_rate": 0.0009041095890410959,
588
+ "loss": 0.5776,
589
+ "step": 700
590
+ },
591
+ {
592
+ "epoch": 4.79,
593
+ "eval_accuracy": 0.7723735408560312,
594
+ "eval_accuracy_class_Bacterial": 0.85546875,
595
+ "eval_accuracy_class_Normal": 0.7967479674796748,
596
+ "eval_accuracy_class_Viral": 0.5925925925925926,
597
+ "eval_f1": 0.7697496825583363,
598
+ "eval_loss": 0.7780735492706299,
599
+ "eval_precision": 0.7771621725411038,
600
+ "eval_recall": 0.7723735408560312,
601
+ "eval_runtime": 43.4411,
602
+ "eval_samples_per_second": 11.832,
603
+ "eval_steps_per_second": 0.391,
604
+ "step": 700
605
+ },
606
+ {
607
+ "epoch": 4.86,
608
+ "grad_norm": 1.3572362661361694,
609
+ "learning_rate": 0.0009027397260273973,
610
+ "loss": 0.4805,
611
+ "step": 710
612
+ },
613
+ {
614
+ "epoch": 4.93,
615
+ "grad_norm": 0.6859280467033386,
616
+ "learning_rate": 0.0009013698630136987,
617
+ "loss": 0.4412,
618
+ "step": 720
619
+ },
620
+ {
621
+ "epoch": 5.0,
622
+ "grad_norm": 0.9770642518997192,
623
+ "learning_rate": 0.0009000000000000001,
624
+ "loss": 0.3328,
625
+ "step": 730
626
+ },
627
+ {
628
+ "epoch": 5.07,
629
+ "grad_norm": 0.7935605645179749,
630
+ "learning_rate": 0.0008986301369863014,
631
+ "loss": 0.4048,
632
+ "step": 740
633
+ },
634
+ {
635
+ "epoch": 5.14,
636
+ "grad_norm": 0.7372242212295532,
637
+ "learning_rate": 0.0008972602739726028,
638
+ "loss": 0.4263,
639
+ "step": 750
640
+ },
641
+ {
642
+ "epoch": 5.21,
643
+ "grad_norm": 0.8754793405532837,
644
+ "learning_rate": 0.0008958904109589042,
645
+ "loss": 0.3936,
646
+ "step": 760
647
+ },
648
+ {
649
+ "epoch": 5.27,
650
+ "grad_norm": 1.1648764610290527,
651
+ "learning_rate": 0.0008945205479452056,
652
+ "loss": 0.3619,
653
+ "step": 770
654
+ },
655
+ {
656
+ "epoch": 5.34,
657
+ "grad_norm": 0.8821219801902771,
658
+ "learning_rate": 0.0008931506849315069,
659
+ "loss": 0.3882,
660
+ "step": 780
661
+ },
662
+ {
663
+ "epoch": 5.41,
664
+ "grad_norm": 0.9254215359687805,
665
+ "learning_rate": 0.0008917808219178082,
666
+ "loss": 0.3538,
667
+ "step": 790
668
+ },
669
+ {
670
+ "epoch": 5.48,
671
+ "grad_norm": 0.3486805856227875,
672
+ "learning_rate": 0.0008904109589041097,
673
+ "loss": 0.4369,
674
+ "step": 800
675
+ },
676
+ {
677
+ "epoch": 5.48,
678
+ "eval_accuracy": 0.7801556420233463,
679
+ "eval_accuracy_class_Bacterial": 0.80859375,
680
+ "eval_accuracy_class_Normal": 0.7886178861788617,
681
+ "eval_accuracy_class_Viral": 0.7185185185185186,
682
+ "eval_f1": 0.7813822202992926,
683
+ "eval_loss": 0.8660529851913452,
684
+ "eval_precision": 0.7850433799518873,
685
+ "eval_recall": 0.7801556420233463,
686
+ "eval_runtime": 39.0462,
687
+ "eval_samples_per_second": 13.164,
688
+ "eval_steps_per_second": 0.435,
689
+ "step": 800
690
+ },
691
+ {
692
+ "epoch": 5.55,
693
+ "grad_norm": 1.4223288297653198,
694
+ "learning_rate": 0.000889041095890411,
695
+ "loss": 0.4764,
696
+ "step": 810
697
+ },
698
+ {
699
+ "epoch": 5.62,
700
+ "grad_norm": 1.5775070190429688,
701
+ "learning_rate": 0.0008876712328767124,
702
+ "loss": 0.4166,
703
+ "step": 820
704
+ },
705
+ {
706
+ "epoch": 5.68,
707
+ "grad_norm": 0.44408535957336426,
708
+ "learning_rate": 0.0008863013698630137,
709
+ "loss": 0.4337,
710
+ "step": 830
711
+ },
712
+ {
713
+ "epoch": 5.75,
714
+ "grad_norm": 0.7388616800308228,
715
+ "learning_rate": 0.0008849315068493152,
716
+ "loss": 0.4474,
717
+ "step": 840
718
+ },
719
+ {
720
+ "epoch": 5.82,
721
+ "grad_norm": 0.6610634922981262,
722
+ "learning_rate": 0.0008835616438356165,
723
+ "loss": 0.3813,
724
+ "step": 850
725
+ },
726
+ {
727
+ "epoch": 5.89,
728
+ "grad_norm": 0.9140748977661133,
729
+ "learning_rate": 0.0008821917808219179,
730
+ "loss": 0.4658,
731
+ "step": 860
732
+ },
733
+ {
734
+ "epoch": 5.96,
735
+ "grad_norm": 0.9210271239280701,
736
+ "learning_rate": 0.0008808219178082192,
737
+ "loss": 0.3908,
738
+ "step": 870
739
+ },
740
+ {
741
+ "epoch": 6.03,
742
+ "grad_norm": 0.4311777651309967,
743
+ "learning_rate": 0.0008794520547945207,
744
+ "loss": 0.4913,
745
+ "step": 880
746
+ },
747
+ {
748
+ "epoch": 6.1,
749
+ "grad_norm": 0.6495469212532043,
750
+ "learning_rate": 0.000878082191780822,
751
+ "loss": 0.3922,
752
+ "step": 890
753
+ },
754
+ {
755
+ "epoch": 6.16,
756
+ "grad_norm": 0.5745411515235901,
757
+ "learning_rate": 0.0008767123287671232,
758
+ "loss": 0.3959,
759
+ "step": 900
760
+ },
761
+ {
762
+ "epoch": 6.16,
763
+ "eval_accuracy": 0.7645914396887159,
764
+ "eval_accuracy_class_Bacterial": 0.78125,
765
+ "eval_accuracy_class_Normal": 0.7642276422764228,
766
+ "eval_accuracy_class_Viral": 0.7333333333333333,
767
+ "eval_f1": 0.767354221797679,
768
+ "eval_loss": 0.8347097635269165,
769
+ "eval_precision": 0.7748247691701339,
770
+ "eval_recall": 0.7645914396887159,
771
+ "eval_runtime": 40.6733,
772
+ "eval_samples_per_second": 12.637,
773
+ "eval_steps_per_second": 0.418,
774
+ "step": 900
775
+ },
776
+ {
777
+ "epoch": 6.23,
778
+ "grad_norm": 0.3549049496650696,
779
+ "learning_rate": 0.0008753424657534247,
780
+ "loss": 0.402,
781
+ "step": 910
782
+ },
783
+ {
784
+ "epoch": 6.3,
785
+ "grad_norm": 1.2331146001815796,
786
+ "learning_rate": 0.000873972602739726,
787
+ "loss": 0.4013,
788
+ "step": 920
789
+ },
790
+ {
791
+ "epoch": 6.37,
792
+ "grad_norm": 0.4287501275539398,
793
+ "learning_rate": 0.0008726027397260274,
794
+ "loss": 0.451,
795
+ "step": 930
796
+ },
797
+ {
798
+ "epoch": 6.44,
799
+ "grad_norm": 0.30888569355010986,
800
+ "learning_rate": 0.0008712328767123287,
801
+ "loss": 0.3853,
802
+ "step": 940
803
+ },
804
+ {
805
+ "epoch": 6.51,
806
+ "grad_norm": 0.3006976544857025,
807
+ "learning_rate": 0.0008698630136986301,
808
+ "loss": 0.3756,
809
+ "step": 950
810
+ },
811
+ {
812
+ "epoch": 6.58,
813
+ "grad_norm": 0.7747860550880432,
814
+ "learning_rate": 0.0008684931506849315,
815
+ "loss": 0.472,
816
+ "step": 960
817
+ },
818
+ {
819
+ "epoch": 6.64,
820
+ "grad_norm": 0.8170182108879089,
821
+ "learning_rate": 0.0008671232876712329,
822
+ "loss": 0.3817,
823
+ "step": 970
824
+ },
825
+ {
826
+ "epoch": 6.71,
827
+ "grad_norm": 0.6512866616249084,
828
+ "learning_rate": 0.0008657534246575342,
829
+ "loss": 0.4345,
830
+ "step": 980
831
+ },
832
+ {
833
+ "epoch": 6.78,
834
+ "grad_norm": 1.3372316360473633,
835
+ "learning_rate": 0.0008643835616438355,
836
+ "loss": 0.4163,
837
+ "step": 990
838
+ },
839
+ {
840
+ "epoch": 6.85,
841
+ "grad_norm": 0.5636022686958313,
842
+ "learning_rate": 0.000863013698630137,
843
+ "loss": 0.3524,
844
+ "step": 1000
845
+ },
846
+ {
847
+ "epoch": 6.85,
848
+ "eval_accuracy": 0.7782101167315175,
849
+ "eval_accuracy_class_Bacterial": 0.83203125,
850
+ "eval_accuracy_class_Normal": 0.8048780487804879,
851
+ "eval_accuracy_class_Viral": 0.6518518518518519,
852
+ "eval_f1": 0.7774953947468894,
853
+ "eval_loss": 0.8164880871772766,
854
+ "eval_precision": 0.7802158087533432,
855
+ "eval_recall": 0.7782101167315175,
856
+ "eval_runtime": 38.1288,
857
+ "eval_samples_per_second": 13.481,
858
+ "eval_steps_per_second": 0.446,
859
+ "step": 1000
860
+ },
861
+ {
862
+ "epoch": 6.92,
863
+ "grad_norm": 0.6771811842918396,
864
+ "learning_rate": 0.0008616438356164383,
865
+ "loss": 0.3926,
866
+ "step": 1010
867
+ },
868
+ {
869
+ "epoch": 6.99,
870
+ "grad_norm": 0.5836020112037659,
871
+ "learning_rate": 0.0008602739726027397,
872
+ "loss": 0.4507,
873
+ "step": 1020
874
+ },
875
+ {
876
+ "epoch": 7.05,
877
+ "grad_norm": 0.9095780849456787,
878
+ "learning_rate": 0.000858904109589041,
879
+ "loss": 0.4257,
880
+ "step": 1030
881
+ },
882
+ {
883
+ "epoch": 7.12,
884
+ "grad_norm": 0.735991358757019,
885
+ "learning_rate": 0.0008575342465753425,
886
+ "loss": 0.4297,
887
+ "step": 1040
888
+ },
889
+ {
890
+ "epoch": 7.19,
891
+ "grad_norm": 0.21994538605213165,
892
+ "learning_rate": 0.0008561643835616438,
893
+ "loss": 0.4011,
894
+ "step": 1050
895
+ },
896
+ {
897
+ "epoch": 7.26,
898
+ "grad_norm": 0.3590526878833771,
899
+ "learning_rate": 0.0008547945205479452,
900
+ "loss": 0.3651,
901
+ "step": 1060
902
+ },
903
+ {
904
+ "epoch": 7.33,
905
+ "grad_norm": 1.078801155090332,
906
+ "learning_rate": 0.0008534246575342465,
907
+ "loss": 0.4736,
908
+ "step": 1070
909
+ },
910
+ {
911
+ "epoch": 7.4,
912
+ "grad_norm": 0.7849373817443848,
913
+ "learning_rate": 0.000852054794520548,
914
+ "loss": 0.4423,
915
+ "step": 1080
916
+ },
917
+ {
918
+ "epoch": 7.47,
919
+ "grad_norm": 0.43372392654418945,
920
+ "learning_rate": 0.0008506849315068493,
921
+ "loss": 0.398,
922
+ "step": 1090
923
+ },
924
+ {
925
+ "epoch": 7.53,
926
+ "grad_norm": 0.8436893224716187,
927
+ "learning_rate": 0.0008493150684931506,
928
+ "loss": 0.422,
929
+ "step": 1100
930
+ },
931
+ {
932
+ "epoch": 7.53,
933
+ "eval_accuracy": 0.7762645914396887,
934
+ "eval_accuracy_class_Bacterial": 0.86328125,
935
+ "eval_accuracy_class_Normal": 0.7804878048780488,
936
+ "eval_accuracy_class_Viral": 0.6074074074074074,
937
+ "eval_f1": 0.7741563369372261,
938
+ "eval_loss": 0.7580455541610718,
939
+ "eval_precision": 0.7825488253813928,
940
+ "eval_recall": 0.7762645914396887,
941
+ "eval_runtime": 39.4721,
942
+ "eval_samples_per_second": 13.022,
943
+ "eval_steps_per_second": 0.431,
944
+ "step": 1100
945
+ },
946
+ {
947
+ "epoch": 7.6,
948
+ "grad_norm": 1.0213713645935059,
949
+ "learning_rate": 0.000847945205479452,
950
+ "loss": 0.3556,
951
+ "step": 1110
952
+ },
953
+ {
954
+ "epoch": 7.67,
955
+ "grad_norm": 0.45328274369239807,
956
+ "learning_rate": 0.0008465753424657534,
957
+ "loss": 0.4054,
958
+ "step": 1120
959
+ },
960
+ {
961
+ "epoch": 7.74,
962
+ "grad_norm": 0.28321486711502075,
963
+ "learning_rate": 0.0008452054794520548,
964
+ "loss": 0.3806,
965
+ "step": 1130
966
+ },
967
+ {
968
+ "epoch": 7.81,
969
+ "grad_norm": 0.9029455184936523,
970
+ "learning_rate": 0.0008438356164383561,
971
+ "loss": 0.404,
972
+ "step": 1140
973
+ },
974
+ {
975
+ "epoch": 7.88,
976
+ "grad_norm": 0.3433306813240051,
977
+ "learning_rate": 0.0008424657534246575,
978
+ "loss": 0.3265,
979
+ "step": 1150
980
+ },
981
+ {
982
+ "epoch": 7.95,
983
+ "grad_norm": 0.9373074769973755,
984
+ "learning_rate": 0.0008410958904109589,
985
+ "loss": 0.4464,
986
+ "step": 1160
987
+ },
988
+ {
989
+ "epoch": 8.01,
990
+ "grad_norm": 0.5745645761489868,
991
+ "learning_rate": 0.0008397260273972603,
992
+ "loss": 0.4057,
993
+ "step": 1170
994
+ },
995
+ {
996
+ "epoch": 8.08,
997
+ "grad_norm": 0.38959360122680664,
998
+ "learning_rate": 0.0008383561643835616,
999
+ "loss": 0.3368,
1000
+ "step": 1180
1001
+ },
1002
+ {
1003
+ "epoch": 8.15,
1004
+ "grad_norm": 0.3645295202732086,
1005
+ "learning_rate": 0.0008369863013698629,
1006
+ "loss": 0.4245,
1007
+ "step": 1190
1008
+ },
1009
+ {
1010
+ "epoch": 8.22,
1011
+ "grad_norm": 0.8835442066192627,
1012
+ "learning_rate": 0.0008356164383561644,
1013
+ "loss": 0.4398,
1014
+ "step": 1200
1015
+ },
1016
+ {
1017
+ "epoch": 8.22,
1018
+ "eval_accuracy": 0.7276264591439688,
1019
+ "eval_accuracy_class_Bacterial": 0.69140625,
1020
+ "eval_accuracy_class_Normal": 0.8048780487804879,
1021
+ "eval_accuracy_class_Viral": 0.725925925925926,
1022
+ "eval_f1": 0.7303442580310747,
1023
+ "eval_loss": 0.8645263910293579,
1024
+ "eval_precision": 0.7390963461212989,
1025
+ "eval_recall": 0.7276264591439688,
1026
+ "eval_runtime": 88.6089,
1027
+ "eval_samples_per_second": 5.801,
1028
+ "eval_steps_per_second": 0.192,
1029
+ "step": 1200
1030
+ },
1031
+ {
1032
+ "epoch": 8.29,
1033
+ "grad_norm": 0.235728457570076,
1034
+ "learning_rate": 0.0008342465753424657,
1035
+ "loss": 0.4229,
1036
+ "step": 1210
1037
+ },
1038
+ {
1039
+ "epoch": 8.36,
1040
+ "grad_norm": 0.9462645053863525,
1041
+ "learning_rate": 0.0008328767123287671,
1042
+ "loss": 0.3902,
1043
+ "step": 1220
1044
+ },
1045
+ {
1046
+ "epoch": 8.42,
1047
+ "grad_norm": 0.3837108910083771,
1048
+ "learning_rate": 0.0008315068493150684,
1049
+ "loss": 0.4508,
1050
+ "step": 1230
1051
+ },
1052
+ {
1053
+ "epoch": 8.49,
1054
+ "grad_norm": 0.5294187068939209,
1055
+ "learning_rate": 0.0008301369863013699,
1056
+ "loss": 0.3643,
1057
+ "step": 1240
1058
+ },
1059
+ {
1060
+ "epoch": 8.56,
1061
+ "grad_norm": 0.4534919261932373,
1062
+ "learning_rate": 0.0008287671232876712,
1063
+ "loss": 0.4308,
1064
+ "step": 1250
1065
+ },
1066
+ {
1067
+ "epoch": 8.63,
1068
+ "grad_norm": 0.5512118935585022,
1069
+ "learning_rate": 0.0008273972602739726,
1070
+ "loss": 0.3734,
1071
+ "step": 1260
1072
+ },
1073
+ {
1074
+ "epoch": 8.7,
1075
+ "grad_norm": 0.446801096200943,
1076
+ "learning_rate": 0.000826027397260274,
1077
+ "loss": 0.3612,
1078
+ "step": 1270
1079
+ },
1080
+ {
1081
+ "epoch": 8.77,
1082
+ "grad_norm": 1.0712846517562866,
1083
+ "learning_rate": 0.0008246575342465754,
1084
+ "loss": 0.361,
1085
+ "step": 1280
1086
+ },
1087
+ {
1088
+ "epoch": 8.84,
1089
+ "grad_norm": 1.1732456684112549,
1090
+ "learning_rate": 0.0008232876712328767,
1091
+ "loss": 0.3792,
1092
+ "step": 1290
1093
+ },
1094
+ {
1095
+ "epoch": 8.9,
1096
+ "grad_norm": 0.5159270167350769,
1097
+ "learning_rate": 0.000821917808219178,
1098
+ "loss": 0.3962,
1099
+ "step": 1300
1100
+ },
1101
+ {
1102
+ "epoch": 8.9,
1103
+ "eval_accuracy": 0.754863813229572,
1104
+ "eval_accuracy_class_Bacterial": 0.74609375,
1105
+ "eval_accuracy_class_Normal": 0.8048780487804879,
1106
+ "eval_accuracy_class_Viral": 0.725925925925926,
1107
+ "eval_f1": 0.7560885847797696,
1108
+ "eval_loss": 0.8278017640113831,
1109
+ "eval_precision": 0.7588608378059307,
1110
+ "eval_recall": 0.754863813229572,
1111
+ "eval_runtime": 476.7256,
1112
+ "eval_samples_per_second": 1.078,
1113
+ "eval_steps_per_second": 0.036,
1114
+ "step": 1300
1115
+ },
1116
+ {
1117
+ "epoch": 8.97,
1118
+ "grad_norm": 0.48536455631256104,
1119
+ "learning_rate": 0.0008205479452054795,
1120
+ "loss": 0.4093,
1121
+ "step": 1310
1122
+ },
1123
+ {
1124
+ "epoch": 9.04,
1125
+ "grad_norm": 0.9068573117256165,
1126
+ "learning_rate": 0.0008191780821917808,
1127
+ "loss": 0.47,
1128
+ "step": 1320
1129
+ },
1130
+ {
1131
+ "epoch": 9.11,
1132
+ "grad_norm": 0.3649793863296509,
1133
+ "learning_rate": 0.0008178082191780822,
1134
+ "loss": 0.3298,
1135
+ "step": 1330
1136
+ },
1137
+ {
1138
+ "epoch": 9.18,
1139
+ "grad_norm": 0.5880826711654663,
1140
+ "learning_rate": 0.0008164383561643835,
1141
+ "loss": 0.3088,
1142
+ "step": 1340
1143
+ },
1144
+ {
1145
+ "epoch": 9.25,
1146
+ "grad_norm": 0.9857625365257263,
1147
+ "learning_rate": 0.000815068493150685,
1148
+ "loss": 0.3509,
1149
+ "step": 1350
1150
+ },
1151
+ {
1152
+ "epoch": 9.32,
1153
+ "grad_norm": 1.4002394676208496,
1154
+ "learning_rate": 0.0008136986301369863,
1155
+ "loss": 0.4272,
1156
+ "step": 1360
1157
+ },
1158
+ {
1159
+ "epoch": 9.38,
1160
+ "grad_norm": 0.28744077682495117,
1161
+ "learning_rate": 0.0008123287671232877,
1162
+ "loss": 0.4068,
1163
+ "step": 1370
1164
+ },
1165
+ {
1166
+ "epoch": 9.45,
1167
+ "grad_norm": 0.3996258080005646,
1168
+ "learning_rate": 0.000810958904109589,
1169
+ "loss": 0.339,
1170
+ "step": 1380
1171
+ },
1172
+ {
1173
+ "epoch": 9.52,
1174
+ "grad_norm": 0.9040880799293518,
1175
+ "learning_rate": 0.0008095890410958904,
1176
+ "loss": 0.3977,
1177
+ "step": 1390
1178
+ },
1179
+ {
1180
+ "epoch": 9.59,
1181
+ "grad_norm": 0.965135931968689,
1182
+ "learning_rate": 0.0008082191780821918,
1183
+ "loss": 0.3871,
1184
+ "step": 1400
1185
+ },
1186
+ {
1187
+ "epoch": 9.59,
1188
+ "eval_accuracy": 0.7607003891050583,
1189
+ "eval_accuracy_class_Bacterial": 0.73828125,
1190
+ "eval_accuracy_class_Normal": 0.7723577235772358,
1191
+ "eval_accuracy_class_Viral": 0.7925925925925926,
1192
+ "eval_f1": 0.7650308027920366,
1193
+ "eval_loss": 0.8378809094429016,
1194
+ "eval_precision": 0.7797474302208566,
1195
+ "eval_recall": 0.7607003891050583,
1196
+ "eval_runtime": 24.5865,
1197
+ "eval_samples_per_second": 20.906,
1198
+ "eval_steps_per_second": 0.691,
1199
+ "step": 1400
1200
+ },
1201
+ {
1202
+ "epoch": 9.66,
1203
+ "grad_norm": 0.6330709457397461,
1204
+ "learning_rate": 0.0008068493150684931,
1205
+ "loss": 0.3691,
1206
+ "step": 1410
1207
+ },
1208
+ {
1209
+ "epoch": 9.73,
1210
+ "grad_norm": 0.1988159418106079,
1211
+ "learning_rate": 0.0008054794520547945,
1212
+ "loss": 0.3915,
1213
+ "step": 1420
1214
+ },
1215
+ {
1216
+ "epoch": 9.79,
1217
+ "grad_norm": 0.35764390230178833,
1218
+ "learning_rate": 0.0008041095890410959,
1219
+ "loss": 0.4255,
1220
+ "step": 1430
1221
+ },
1222
+ {
1223
+ "epoch": 9.86,
1224
+ "grad_norm": 0.5993324518203735,
1225
+ "learning_rate": 0.0008027397260273973,
1226
+ "loss": 0.3154,
1227
+ "step": 1440
1228
+ },
1229
+ {
1230
+ "epoch": 9.93,
1231
+ "grad_norm": 0.6987205147743225,
1232
+ "learning_rate": 0.0008013698630136986,
1233
+ "loss": 0.4347,
1234
+ "step": 1450
1235
+ },
1236
+ {
1237
+ "epoch": 10.0,
1238
+ "grad_norm": 1.6289706230163574,
1239
+ "learning_rate": 0.0008,
1240
+ "loss": 0.5012,
1241
+ "step": 1460
1242
+ },
1243
+ {
1244
+ "epoch": 10.07,
1245
+ "grad_norm": 0.42524194717407227,
1246
+ "learning_rate": 0.0007986301369863014,
1247
+ "loss": 0.3247,
1248
+ "step": 1470
1249
+ },
1250
+ {
1251
+ "epoch": 10.14,
1252
+ "grad_norm": 0.45430487394332886,
1253
+ "learning_rate": 0.0007972602739726027,
1254
+ "loss": 0.3774,
1255
+ "step": 1480
1256
+ },
1257
+ {
1258
+ "epoch": 10.21,
1259
+ "grad_norm": 0.4011427164077759,
1260
+ "learning_rate": 0.0007958904109589041,
1261
+ "loss": 0.3483,
1262
+ "step": 1490
1263
+ },
1264
+ {
1265
+ "epoch": 10.27,
1266
+ "grad_norm": 1.9553821086883545,
1267
+ "learning_rate": 0.0007945205479452054,
1268
+ "loss": 0.3949,
1269
+ "step": 1500
1270
+ },
1271
+ {
1272
+ "epoch": 10.27,
1273
+ "eval_accuracy": 0.7704280155642024,
1274
+ "eval_accuracy_class_Bacterial": 0.76171875,
1275
+ "eval_accuracy_class_Normal": 0.7967479674796748,
1276
+ "eval_accuracy_class_Viral": 0.762962962962963,
1277
+ "eval_f1": 0.7727560365613958,
1278
+ "eval_loss": 0.8412158489227295,
1279
+ "eval_precision": 0.7787042029583747,
1280
+ "eval_recall": 0.7704280155642024,
1281
+ "eval_runtime": 26.6938,
1282
+ "eval_samples_per_second": 19.255,
1283
+ "eval_steps_per_second": 0.637,
1284
+ "step": 1500
1285
+ },
1286
+ {
1287
+ "epoch": 10.34,
1288
+ "grad_norm": 0.5882077217102051,
1289
+ "learning_rate": 0.0007931506849315069,
1290
+ "loss": 0.4126,
1291
+ "step": 1510
1292
+ },
1293
+ {
1294
+ "epoch": 10.41,
1295
+ "grad_norm": 1.9042034149169922,
1296
+ "learning_rate": 0.0007917808219178082,
1297
+ "loss": 0.4275,
1298
+ "step": 1520
1299
+ },
1300
+ {
1301
+ "epoch": 10.48,
1302
+ "grad_norm": 0.38827410340309143,
1303
+ "learning_rate": 0.0007904109589041096,
1304
+ "loss": 0.4241,
1305
+ "step": 1530
1306
+ },
1307
+ {
1308
+ "epoch": 10.55,
1309
+ "grad_norm": 1.0685482025146484,
1310
+ "learning_rate": 0.0007890410958904109,
1311
+ "loss": 0.3723,
1312
+ "step": 1540
1313
+ },
1314
+ {
1315
+ "epoch": 10.62,
1316
+ "grad_norm": 0.4978479743003845,
1317
+ "learning_rate": 0.0007876712328767124,
1318
+ "loss": 0.4213,
1319
+ "step": 1550
1320
+ },
1321
+ {
1322
+ "epoch": 10.68,
1323
+ "grad_norm": 0.5836212635040283,
1324
+ "learning_rate": 0.0007863013698630137,
1325
+ "loss": 0.3702,
1326
+ "step": 1560
1327
+ },
1328
+ {
1329
+ "epoch": 10.75,
1330
+ "grad_norm": 0.7332190871238708,
1331
+ "learning_rate": 0.0007849315068493151,
1332
+ "loss": 0.3701,
1333
+ "step": 1570
1334
+ },
1335
+ {
1336
+ "epoch": 10.82,
1337
+ "grad_norm": 0.6048933267593384,
1338
+ "learning_rate": 0.0007835616438356164,
1339
+ "loss": 0.4229,
1340
+ "step": 1580
1341
+ },
1342
+ {
1343
+ "epoch": 10.89,
1344
+ "grad_norm": 0.8896064162254333,
1345
+ "learning_rate": 0.0007821917808219178,
1346
+ "loss": 0.3429,
1347
+ "step": 1590
1348
+ },
1349
+ {
1350
+ "epoch": 10.96,
1351
+ "grad_norm": 0.3327595591545105,
1352
+ "learning_rate": 0.0007808219178082192,
1353
+ "loss": 0.3547,
1354
+ "step": 1600
1355
+ },
1356
+ {
1357
+ "epoch": 10.96,
1358
+ "eval_accuracy": 0.7645914396887159,
1359
+ "eval_accuracy_class_Bacterial": 0.7265625,
1360
+ "eval_accuracy_class_Normal": 0.7967479674796748,
1361
+ "eval_accuracy_class_Viral": 0.8074074074074075,
1362
+ "eval_f1": 0.7688402503119753,
1363
+ "eval_loss": 0.8228224515914917,
1364
+ "eval_precision": 0.7840447085340081,
1365
+ "eval_recall": 0.7645914396887159,
1366
+ "eval_runtime": 25.0146,
1367
+ "eval_samples_per_second": 20.548,
1368
+ "eval_steps_per_second": 0.68,
1369
+ "step": 1600
1370
+ },
1371
+ {
1372
+ "epoch": 11.03,
1373
+ "grad_norm": 1.2333165407180786,
1374
+ "learning_rate": 0.0007794520547945205,
1375
+ "loss": 0.2956,
1376
+ "step": 1610
1377
+ },
1378
+ {
1379
+ "epoch": 11.1,
1380
+ "grad_norm": 0.970270037651062,
1381
+ "learning_rate": 0.0007780821917808219,
1382
+ "loss": 0.4526,
1383
+ "step": 1620
1384
+ },
1385
+ {
1386
+ "epoch": 11.16,
1387
+ "grad_norm": 0.8335412740707397,
1388
+ "learning_rate": 0.0007767123287671233,
1389
+ "loss": 0.4005,
1390
+ "step": 1630
1391
+ },
1392
+ {
1393
+ "epoch": 11.23,
1394
+ "grad_norm": 0.5635302662849426,
1395
+ "learning_rate": 0.0007753424657534247,
1396
+ "loss": 0.3713,
1397
+ "step": 1640
1398
+ },
1399
+ {
1400
+ "epoch": 11.3,
1401
+ "grad_norm": 0.555288553237915,
1402
+ "learning_rate": 0.000773972602739726,
1403
+ "loss": 0.4158,
1404
+ "step": 1650
1405
+ },
1406
+ {
1407
+ "epoch": 11.37,
1408
+ "grad_norm": 0.8734630346298218,
1409
+ "learning_rate": 0.0007726027397260274,
1410
+ "loss": 0.3844,
1411
+ "step": 1660
1412
+ },
1413
+ {
1414
+ "epoch": 11.44,
1415
+ "grad_norm": 1.3194448947906494,
1416
+ "learning_rate": 0.0007712328767123288,
1417
+ "loss": 0.3602,
1418
+ "step": 1670
1419
+ },
1420
+ {
1421
+ "epoch": 11.51,
1422
+ "grad_norm": 0.9547590613365173,
1423
+ "learning_rate": 0.0007698630136986301,
1424
+ "loss": 0.3285,
1425
+ "step": 1680
1426
+ },
1427
+ {
1428
+ "epoch": 11.58,
1429
+ "grad_norm": 0.9226530194282532,
1430
+ "learning_rate": 0.0007684931506849315,
1431
+ "loss": 0.3075,
1432
+ "step": 1690
1433
+ },
1434
+ {
1435
+ "epoch": 11.64,
1436
+ "grad_norm": 0.3840011656284332,
1437
+ "learning_rate": 0.0007671232876712328,
1438
+ "loss": 0.3994,
1439
+ "step": 1700
1440
+ },
1441
+ {
1442
+ "epoch": 11.64,
1443
+ "eval_accuracy": 0.77431906614786,
1444
+ "eval_accuracy_class_Bacterial": 0.7890625,
1445
+ "eval_accuracy_class_Normal": 0.7886178861788617,
1446
+ "eval_accuracy_class_Viral": 0.7333333333333333,
1447
+ "eval_f1": 0.7774772964245277,
1448
+ "eval_loss": 0.8398252129554749,
1449
+ "eval_precision": 0.7855846838648818,
1450
+ "eval_recall": 0.77431906614786,
1451
+ "eval_runtime": 26.6906,
1452
+ "eval_samples_per_second": 19.258,
1453
+ "eval_steps_per_second": 0.637,
1454
+ "step": 1700
1455
+ },
1456
+ {
1457
+ "epoch": 11.71,
1458
+ "grad_norm": 0.4347558617591858,
1459
+ "learning_rate": 0.0007657534246575343,
1460
+ "loss": 0.3146,
1461
+ "step": 1710
1462
+ },
1463
+ {
1464
+ "epoch": 11.78,
1465
+ "grad_norm": 0.5778964757919312,
1466
+ "learning_rate": 0.0007643835616438356,
1467
+ "loss": 0.3872,
1468
+ "step": 1720
1469
+ },
1470
+ {
1471
+ "epoch": 11.85,
1472
+ "grad_norm": 0.7429023385047913,
1473
+ "learning_rate": 0.000763013698630137,
1474
+ "loss": 0.3561,
1475
+ "step": 1730
1476
+ },
1477
+ {
1478
+ "epoch": 11.92,
1479
+ "grad_norm": 0.938795268535614,
1480
+ "learning_rate": 0.0007616438356164383,
1481
+ "loss": 0.4158,
1482
+ "step": 1740
1483
+ },
1484
+ {
1485
+ "epoch": 11.99,
1486
+ "grad_norm": 0.4394015967845917,
1487
+ "learning_rate": 0.0007602739726027398,
1488
+ "loss": 0.3729,
1489
+ "step": 1750
1490
+ },
1491
+ {
1492
+ "epoch": 12.05,
1493
+ "grad_norm": 0.6519060134887695,
1494
+ "learning_rate": 0.0007589041095890411,
1495
+ "loss": 0.3371,
1496
+ "step": 1760
1497
+ },
1498
+ {
1499
+ "epoch": 12.12,
1500
+ "grad_norm": 0.3657008707523346,
1501
+ "learning_rate": 0.0007575342465753425,
1502
+ "loss": 0.3517,
1503
+ "step": 1770
1504
+ },
1505
+ {
1506
+ "epoch": 12.19,
1507
+ "grad_norm": 0.4903722107410431,
1508
+ "learning_rate": 0.0007561643835616439,
1509
+ "loss": 0.3818,
1510
+ "step": 1780
1511
+ },
1512
+ {
1513
+ "epoch": 12.26,
1514
+ "grad_norm": 0.41789180040359497,
1515
+ "learning_rate": 0.0007547945205479452,
1516
+ "loss": 0.3878,
1517
+ "step": 1790
1518
+ },
1519
+ {
1520
+ "epoch": 12.33,
1521
+ "grad_norm": 0.6251237392425537,
1522
+ "learning_rate": 0.0007534246575342466,
1523
+ "loss": 0.3717,
1524
+ "step": 1800
1525
+ },
1526
+ {
1527
+ "epoch": 12.33,
1528
+ "eval_accuracy": 0.7704280155642024,
1529
+ "eval_accuracy_class_Bacterial": 0.78515625,
1530
+ "eval_accuracy_class_Normal": 0.8048780487804879,
1531
+ "eval_accuracy_class_Viral": 0.7111111111111111,
1532
+ "eval_f1": 0.7718021949322796,
1533
+ "eval_loss": 0.7440524101257324,
1534
+ "eval_precision": 0.774278133812427,
1535
+ "eval_recall": 0.7704280155642024,
1536
+ "eval_runtime": 26.5014,
1537
+ "eval_samples_per_second": 19.395,
1538
+ "eval_steps_per_second": 0.641,
1539
+ "step": 1800
1540
+ }
1541
+ ],
1542
+ "logging_steps": 10,
1543
+ "max_steps": 7300,
1544
+ "num_input_tokens_seen": 0,
1545
+ "num_train_epochs": 50,
1546
+ "save_steps": 100,
1547
+ "total_flos": 4.439400889630114e+18,
1548
+ "train_batch_size": 32,
1549
+ "trial_name": null,
1550
+ "trial_params": null
1551
+ }
models/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d6c11078065b9bfd3a041f71cbf2e838578aa9ab78caac1121bbc0b2e39bf16
3
+ size 4920
requirements.txt ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ torch
2
+ transformers == 4.38.1
3
+ datasets
4
+ numpy
5
+ opencv-python
6
+ Pillow
7
+ matplotlib