Update app.py

app.py

@@ -1,64 +1,84 @@
 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("GoofyLM/gonzalez-v1")
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
 
+# Load model and tokenizer locally
+model_name = "GoofyLM/gonzalez-v1"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    torch_dtype=torch.float16,  # Use float16 for efficiency
+    device_map="auto"  # Automatically distribute across available GPUs/devices
+)
 
 def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
+    message, 
+    history: list[tuple[str, str]], 
+    system_message, 
+    max_tokens, 
+    temperature, 
     top_p,
 ):
+    # Format messages for the model
     messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
+    for user_msg, assistant_msg in history:
+        if user_msg:
+            messages.append({"role": "user", "content": user_msg})
+        if assistant_msg:
+            messages.append({"role": "assistant", "content": assistant_msg})
     messages.append({"role": "user", "content": message})
 
+    # Convert messages to model input format
+    chat_template = tokenizer.apply_chat_template(
+        messages, 
+        tokenize=False, 
+        add_generation_prompt=True
+    )
+    
+    # Tokenize the input
+    inputs = tokenizer(chat_template, return_tensors="pt").to(model.device)
+    
+    # Generate response with streaming
+    input_length = inputs.input_ids.shape[1]
+    generated_tokens = []
+    
+    # Set up generation parameters
+    gen_kwargs = {
+        "max_new_tokens": max_tokens,
+        "temperature": temperature,
+        "top_p": top_p,
+        "do_sample": temperature > 0,
+        "pad_token_id": tokenizer.eos_token_id,
+    }
+    
+    # Stream the generation
     response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
+    for output in model.generate(
+        **inputs,
+        **gen_kwargs,
+        streamer=transformers.TextStreamer(tokenizer, skip_prompt=True),
     ):
-        token = message.choices[0].delta.content
-
-        response += token
+        # Skip input tokens
+        if len(output) <= input_length:
+            continue
+        
+        # Get new tokens
+        new_tokens = output[input_length:]
+        decoded = tokenizer.decode(new_tokens, skip_special_tokens=True)
+        response = decoded
         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 Gonzalez.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=72, step=1, label="Max new tokens"),
+        gr.Textbox(value="You are a Gonzalez-v1.", 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)",
+            minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"
         ),
     ],
 )
 
-
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()