Implement coding assistant chatbot using Qwen3 model with PEFT adapter and Gradio interface

app.py

@@ -1,5 +1,33 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import PeftModel
+
+
+# Load model và tokenizer
+print("Đang tải model...")
+base_model_name = "Qwen/Qwen3-0.6B"
+adapter_repo = "cochi1706/coding-assistant"
+
+# Load base model
+base_model = AutoModelForCausalLM.from_pretrained(
+    base_model_name,
+    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
+    device_map="auto" if torch.cuda.is_available() else None,
+)
+
+# Load PEFT adapter
+model = PeftModel.from_pretrained(base_model, adapter_repo)
+
+# Load tokenizer
+tokenizer = AutoTokenizer.from_pretrained(adapter_repo)
+
+# Set padding token nếu chưa có
+if tokenizer.pad_token is None:
+    tokenizer.pad_token = tokenizer.eos_token
+
+model.eval()
+print("Model đã sẵn sàng!")
 
 
 def respond(
@@ -9,47 +37,99 @@ def respond(
     max_tokens,
     temperature,
     top_p,
-    hf_token: gr.OAuthToken,
 ):
     """
-    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
+    Tạo phản hồi từ model coding assistant
     """
-    client = InferenceClient(token=hf_token.token, model="openai/gpt-oss-20b")
-
+    # Chuẩn bị prompt với chat template
     messages = [{"role": "system", "content": system_message}]
-
     messages.extend(history)
-
     messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
+    
+    # Format messages với chat template
+    prompt = tokenizer.apply_chat_template(
         messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        choices = message.choices
-        token = ""
-        if len(choices) and choices[0].delta.content:
-            token = choices[0].delta.content
-
-        response += token
-        yield response
+        tokenize=False,
+        add_generation_prompt=True
+    )
+    
+    # Tokenize
+    inputs = tokenizer(prompt, return_tensors="pt")
+    if torch.cuda.is_available():
+        inputs = {k: v.to(model.device) for k, v in inputs.items()}
+    
+    # Generate với streaming token-by-token
+    input_length = inputs["input_ids"].shape[1]
+    response = ""
+    
+    with torch.no_grad():
+        # Khởi tạo với input_ids
+        generated_ids = inputs["input_ids"].clone()
+        
+        for _ in range(max_tokens):
+            # Forward pass
+            outputs = model(generated_ids)
+            logits = outputs.logits[:, -1, :]
+            
+            # Apply temperature và top_p
+            if temperature != 1.0:
+                logits = logits / temperature
+            
+            # Top-p sampling
+            if top_p < 1.0:
+                sorted_logits, sorted_indices = torch.sort(logits, descending=True)
+                cumulative_probs = torch.cumsum(torch.softmax(sorted_logits, dim=-1), dim=-1)
+                sorted_indices_to_remove = cumulative_probs > top_p
+                sorted_indices_to_remove[..., 1:] = sorted_indices_to_remove[..., :-1].clone()
+                sorted_indices_to_remove[..., 0] = 0
+                indices_to_remove = sorted_indices_to_remove.scatter(1, sorted_indices, sorted_indices_to_remove)
+                logits[indices_to_remove] = float('-inf')
+            
+            # Sample next token
+            probs = torch.softmax(logits, dim=-1)
+            next_token = torch.multinomial(probs, num_samples=1)
+            
+            # Kiểm tra EOS token
+            if next_token.item() == tokenizer.eos_token_id:
+                break
+            
+            # Thêm token vào generated_ids
+            generated_ids = torch.cat([generated_ids, next_token], dim=1)
+            
+            # Decode token mới và stream
+            new_text = tokenizer.decode([next_token.item()], skip_special_tokens=True)
+            response += new_text
+            yield response
 
 
 """
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
+Chatbot hỗ trợ lập trình sử dụng Qwen3 fine-tuned model
 """
 chatbot = gr.ChatInterface(
     respond,
+    title="🤖 Coding Assistant",
+    description="Chatbot hỗ trợ lập trình",
     type="messages",
     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.Textbox(
+            value="You are a helpful coding assistant. Provide clear, concise, and accurate code solutions and explanations.",
+            label="System message",
+            lines=3,
+        ),
+        gr.Slider(
+            minimum=1,
+            maximum=2048,
+            value=512,
+            step=1,
+            label="Max new tokens",
+        ),
+        gr.Slider(
+            minimum=0.1,
+            maximum=2.0,
+            value=0.7,
+            step=0.1,
+            label="Temperature",
+        ),
         gr.Slider(
             minimum=0.1,
             maximum=1.0,
@@ -60,10 +140,7 @@ chatbot = gr.ChatInterface(
     ],
 )
 
-with gr.Blocks() as demo:
-    with gr.Sidebar():
-        gr.LoginButton()
-    chatbot.render()
+demo = chatbot
 
 
 if __name__ == "__main__":