Update app.py

app.py

@@ -1,64 +1,68 @@
+import os
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer, BitsAndBytesConfig
+import torch
 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")
+# Retrieve the API key from environment variables
+hf_api_key = os.getenv('HF_API_KEY')
 
+# Check if the API key is available
+if hf_api_key is None:
+    raise ValueError("API key not found. Please set the HF_API_KEY environment variable.")
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+# Model configuration
+bnb_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_use_double_quant=True,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_compute_dtype=torch.bfloat16
+)
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+MODEL_NAME = 'meta-llama/Llama-3.1-8B-Instruct'
 
-    messages.append({"role": "user", "content": message})
+# Load tokenizer and model with the API key
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, use_auth_token=hf_api_key)
+model = AutoModelForCausalLM.from_pretrained(MODEL_NAME, device_map='cuda:0', quantization_config=bnb_config, use_auth_token=hf_api_key)
 
-    response = ""
+# Function to generate responses
+def generate_response(message, history):
+    try:
+        # Create a properly formatted conversation history
+        full_conversation = ""
+        for user_msg, assistant_msg in history:
+            full_conversation += f"User: {user_msg}\nAssistant: {assistant_msg}\n"
+        full_conversation += f"User: {message}\n"
 
-    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
+        # Tokenize the full conversation
+        inputs = tokenizer(full_conversation, return_tensors="pt", truncation=True, max_length=2048).to(model.device)
 
-        response += token
-        yield response
+        # Generate response
+        outputs = model.generate(
+            **inputs,
+            max_new_tokens=150,
+            do_sample=True,
+            temperature=0.7,
+            top_p=0.9,
+            repetition_penalty=1.2
+        )
 
+        # Decode the generated text
+        response = tokenizer.decode(outputs[0], skip_special_tokens=True).strip()
+        response = response.split("User:")[-1].split("Assistant:")[-1].strip()
 
-"""
-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)",
-        ),
-    ],
-)
+        return response
 
+    except Exception as e:
+        print("Error during response generation:", str(e))
+        return f"An error occurred during response generation: {str(e)}"
+
+# Create the Gradio ChatInterface
+chat_interface = gr.ChatInterface(
+    fn=generate_response,
+    title="Llama 3 Chatbot",
+    description="Interact with the Llama 3 model. Type a message and receive a response.",
+    examples=["Hello, how are you?", "What's the weather like?"]
+)
 
-if __name__ == "__main__":
-    demo.launch()
+# Launch the interface
+chat_interface.launch(debug=True, share=True)