Update app.py

app.py

@@ -1,81 +1,27 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
-import os
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
 
 # ============================================================
-# ১. মডেল আইডি (সিক্রেট বা ডিফল্ট)
+# ১. মডেল লোড (আপনার mx-llms/BLM)
 # ============================================================
-model_id = os.getenv("MODEL_ID", "mx-llms/BLM")
-
-def respond(
-    message,
-    history: list[dict[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    hf_token: gr.OAuthToken | None,
-):
-    """
-    Hugging Face Inference API - টোকেন অটো-ডিটেক্ট:
-    - ইউজার লগইন করলে OAuth Token ব্যবহার করবে
-    - না করলে Space Secret-এর HF_TOKEN ব্যবহার করবে
-    """
-    # টোকেন নির্বাচন
-    if hf_token is not None:
-        token = hf_token.token
-    else:
-        token = os.getenv("HF_TOKEN")
-        if token is None:
-            yield "⚠️ Please login or set HF_TOKEN in Space Secrets."
-            return
-
-    client = InferenceClient(token=token, model=model_id)
-
-    messages = [{"role": "system", "content": system_message}]
-    messages.extend(history)
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-    try:
-        for msg in client.chat_completion(
-            messages,
-            max_tokens=max_tokens,
-            stream=True,
-            temperature=temperature,
-            top_p=top_p,
-        ):
-            token_text = msg.choices[0].delta.content or ""
-            response += token_text
-            yield response
-    except Exception as e:
-        yield f"❌ Error: {str(e)}. Please check logs."
-
-# ============================================================
-# ২. ChatInterface UI
-# ============================================================
-chatbot = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(
-            value="You are BLM, a helpful AI assistant created by MD Mushfiqur Rahim.",
-            label="System message"
-        ),
-        gr.Slider(minimum=1, maximum=4096, 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"),
-    ],
-    title="BLM AI Assistant",
-    description="Created by MD Mushfiqur Rahim. Please login (top-left) for full access.",
+model = AutoModelForCausalLM.from_pretrained(
+    "mx-llms/BLM",
+    trust_remote_code=True,
+    device_map="auto",
+    torch_dtype=torch.float16,
 )
+tokenizer = AutoTokenizer.from_pretrained("mx-llms/BLM", trust_remote_code=True)
+
+def respond(message, history):
+    messages = [{"role": "user", "content": message}]
+    prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+    inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
+    with torch.no_grad():
+        outputs = model.generate(**inputs, max_new_tokens=100)
+    return tokenizer.decode(outputs[0], skip_special_tokens=True)
 
 # ============================================================
-# ৩. Login Button সহ ডেমো
+# ২. Gradio UI
 # ============================================================
-with gr.Blocks(title="BLM AI Assistant") as demo:
-    with gr.Sidebar():
-        gr.LoginButton()
-    chatbot.render()
-
-if __name__ == "__main__":
-    demo.launch()
+gr.ChatInterface(fn=respond, title="BLM AI Assistant").launch()