Update app.py

app.py

@@ -1,58 +1,21 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
-import os
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
 
-# Replace with your model's repo
-client = InferenceClient("yuvrajpant56/Mistral_Posttrain_SFT",token=os.environ["HF_TOKEN"])
+model_id = "yuvrajpant56/Mistral_Posttrain_SFT"
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    # Build simple prompt using system message and conversation history
-    prompt = f"{system_message.strip()}\n\n"
+# Load model & tokenizer
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16, device_map="auto")
 
-    for user_msg, bot_msg in history:
-        if user_msg:
-            prompt += f"User: {user_msg}\n"
-        if bot_msg:
-            prompt += f"Assistant: {bot_msg}\n"
-    prompt += f"User: {message}\nAssistant:"
+def generate_text(prompt):
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    outputs = model.generate(**inputs, max_new_tokens=100)
+    return tokenizer.decode(outputs[0], skip_special_tokens=True)
 
-    response = ""
-    for token in client.text_generation(
-        prompt,
-        max_new_tokens=max_tokens,
-        temperature=temperature,
-        top_p=top_p,
-        stream=True,
-    ):
-        response += token.token
-        yield response
-
-# Build Gradio ChatInterface
-demo = gr.ChatInterface(
-    fn=respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a helpful assistant.", 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"),
-    ],
-    title="🧠 Mistral SFT Chatbot",
-    description="Interact with your fine-tuned Mistral model using Hugging Face Inference API.",
-    theme="soft",
-    examples=[
-        ["Tell me a story about the Himalayas.", "You are a helpful assistant.", 512, 0.7, 0.95],
-        ["What is quantum computing?", "Explain like I’m 5.", 512, 0.8, 0.9]
-    ],
-    type="openai",  # also removes the warning about 'tuples'
-)
-
-
-if __name__ == "__main__":
-    demo.launch()
+gr.Interface(fn=generate_text,
+             inputs="text",
+             outputs="text",
+             title="Mistral SFT Text Generator",
+             description="Type a prompt and let the fine-tuned Mistral model generate the rest."
+).launch()