Create app.py

app.py

@@ -0,0 +1,92 @@
+import gradio as gr
+import torch
+from transformers import BertTokenizer, BertForSequenceClassification, XLNetTokenizer, XLNetForSequenceClassification
+
+# โหลดโมเดลของคุณ
+logbert_model = BertForSequenceClassification.from_pretrained("Sirapatsorn/Spark_Log_Analysis", use_auth_token=True)
+xlnet_model = XLNetForSequenceClassification.from_pretrained("Sirapatsorn/Spark_Log_Analysis", use_auth_token=True)
+logbert_tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
+xlnet_tokenizer = XLNetTokenizer.from_pretrained("xlnet-base-cased")
+
+# ตรวจสอบว่ามี GPU หรือไม่
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+logbert_model.to(device)
+xlnet_model.to(device)
+
+# ฟังก์ชันการพยากรณ์
+def predict_log(text):
+    logbert_inputs = logbert_tokenizer(text, return_tensors="pt", truncation=True, max_length=512).to(device)
+    with torch.no_grad():
+        logbert_outputs = logbert_model(**logbert_inputs)
+        log_level = torch.argmax(logbert_outputs.logits, dim=1).item()
+        log_level_confidence = torch.softmax(logbert_outputs.logits, dim=1)[0][log_level].item()
+
+    log_levels = ["INFO", "WARN", "ERROR"]
+    log_level_result = log_levels[log_level]
+
+    xlnet_inputs = xlnet_tokenizer(text, return_tensors="pt", truncation=True, max_length=512).to(device)
+    with torch.no_grad():
+        xlnet_outputs = xlnet_model(**xlnet_inputs)
+        performance_value = xlnet_outputs.logits.item()
+
+    if performance_value < 0:
+        performance_status = "Good Performance"
+    elif performance_value < 3.0:
+        performance_status = "Normal Performance"
+    else:
+        performance_status = "Poor Performance"
+
+    return {
+        "Log Level": log_level_result,
+        "Confidence": f"{log_level_confidence:.2f}",
+        "Performance Value": f"{performance_value:.2f}",
+        "Performance Status": performance_status
+    }
+
+# ฟังก์ชันสำหรับ Gradio: พยากรณ์จากข้อความโดยตรง
+def predict_from_text(text):
+    prediction = predict_log(text)
+    return (f"Log Level: {prediction['Log Level']} (Confidence: {prediction['Confidence']})\n"
+            f"Performance Value: {prediction['Performance Value']}\n"
+            f"Performance Status: {prediction['Performance Status']}")
+
+def predict_from_file(file):
+    results = []
+    with open(file.name, 'r') as f:
+        for line in f:
+            prediction = predict_log(line.strip())
+            result_text = (f"Log: {line.strip()}\n"
+                           f"Log Level: {prediction['Log Level']} (Confidence: {prediction['Confidence']})\n"
+                           f"Performance Value: {prediction['Performance Value']}\n"
+                           f"Performance Status: {prediction['Performance Status']}")
+            results.append(result_text)
+    return "\n\n".join(results)
+
+custom_css = """
+.gr-button {
+    background-color: #FFA500 !important; 
+    color: #FFFFFF !important;
+    border: none !important;
+}
+"""
+
+with gr.Blocks(css=custom_css) as demo:
+    with gr.Tabs():
+        with gr.TabItem("Upload File"):
+            file_upload = gr.File(label="Upload Log File")
+            file_btn = gr.Button("Predict")
+            file_output = gr.Textbox(label="Output")
+            file_btn.click(predict_from_file, inputs=file_upload, outputs=file_output)
+        with gr.TabItem("Text Input"):
+            text_input = gr.Textbox(label="Enter Log Message")
+            text_btn = gr.Button("Predict")
+            text_output = gr.Textbox(label="Output")
+            text_btn.click(predict_from_text, inputs=text_input, outputs=text_output)
+
+    demo.css += """
+    .gr-button.gr-button-lg.gr-button-secondary {
+        display: none !important;
+    }
+    """
+
+demo.launch()