Update app.py

app.py

@@ -1,64 +1,86 @@
+import os
+import logging
+import random
+import pickle
+import torch
+from transformers import BertForSequenceClassification, BertTokenizerFast
+from huggingface_hub import hf_hub_download
 import gradio as gr
-from huggingface_hub import InferenceClient
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+# Configuration
+MODEL_NAME = "vinapatri/intent-classification-jkn-kis"
+CONFIDENCE_THRESHOLD = 0.6  # Adjust this based on your model's performance
 
+# --- Model Functions ---
+def load_model_components():
+    model = BertForSequenceClassification.from_pretrained(MODEL_NAME)
+    tokenizer = BertTokenizerFast.from_pretrained(MODEL_NAME)
+    
+    label_encoder_path = hf_hub_download(repo_id=MODEL_NAME, filename="label_encoder.pkl")
+    responses_path = hf_hub_download(repo_id=MODEL_NAME, filename="tag_to_responses.pkl")
+    
+    with open(label_encoder_path, "rb") as f:
+        le = pickle.load(f)
+    
+    with open(responses_path, "rb") as f:
+        tag_to_responses = pickle.load(f)
+    
+    # Ensure we have an 'unknown' response category
+    if 'unknown' not in tag_to_responses:
+        tag_to_responses['unknown'] = [
+            "Maaf, saya tidak mengerti pertanyaan Anda",
+            "Saya belum bisa menjawab pertanyaan tersebut"
+        ]
+    
+    return model, tokenizer, le, tag_to_responses
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+model, tokenizer, le, tag_to_responses = load_model_components()
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+def predict_intent(text):
+    inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)
+    
+    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    model.to(device)
+    inputs = {k: v.to(device) for k, v in inputs.items()}
+    
+    with torch.no_grad():
+        outputs = model(**inputs)
+    
+    # Get confidence scores
+    probabilities = torch.softmax(outputs.logits, dim=1)[0]
+    confidence, predicted_class_id = torch.max(probabilities, 0)
+    confidence = confidence.item()
+    
+    # Return 'unknown' if confidence is below threshold
+    if confidence < CONFIDENCE_THRESHOLD:
+        return "unknown"
+    
+    tag = le.inverse_transform([predicted_class_id.item()])[0]
+    return tag
 
-    messages.append({"role": "user", "content": message})
+def get_response(user_input):
+    tag = predict_intent(user_input)
+    responses = tag_to_responses.get(tag, tag_to_responses['unknown'])
+    return random.choice(responses)
 
-    response = ""
+# --- Gradio Interface ---
+def chat_interface(message, history):
+    return get_response(message)
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
+gradio_app = gr.ChatInterface(
+    fn=chat_interface,
+    title="JKN-KIS Intent Classification",
+    description="Bot untuk klasifikasi intent terkait JKN-KIS",
+    examples=[
+        "Bagaimana cara daftar Mobile JKN?",
+        "Gimana cara buat rujukan online?",
+        "Kenapa OTP Mobile JKN tidak masuk?"
+    ]
 )
 
-
 if __name__ == "__main__":
-    demo.launch()
+    logging.basicConfig(
+        format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
+        level=logging.INFO
+    )
+    gradio_app.launch(server_name="0.0.0.0", server_port=7860)