Update app.py

app.py

@@ -1,6 +1,37 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
+from threading import Thread
 
+# 1. Configuration
+MODEL_ID = "ConceptModels/Concept-7b-V1-Full"
+
+# 2. Load Model and Tokenizer (Done once at startup)
+print(f"Loading {MODEL_ID}... this may take a while.")
+
+try:
+    tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+    
+    # Attempt to use GPU if available, otherwise CPU
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    print(f"Running on device: {device}")
+
+    model = AutoModelForCausalLM.from_pretrained(
+        MODEL_ID,
+        torch_dtype=torch.float16 if device == "cuda" else torch.float32,
+        device_map="auto" if device == "cuda" else None,
+        # Uncomment the line below to use 4-bit quantization (requires pip install bitsandbytes)
+        # load_in_4bit=True 
+    )
+    # If using CPU, move model explicitly
+    if device == "cpu":
+        model.to("cpu")
+        
+    print("Model loaded successfully.")
+
+except Exception as e:
+    print(f"Error loading model: {e}")
+    raise e
 
 def respond(
     message,
@@ -9,45 +40,49 @@ def respond(
     max_tokens,
     temperature,
     top_p,
-    hf_token: gr.OAuthToken,
+    hf_token=None, # Not strictly needed for local if logged in via CLI, but kept for signature compatibility
 ):
-    """
-    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(token=hf_token.token, model="ConceptModels/Concept-7b-V1-Full")
-
+    # 3. Format the conversation
+    # We construct the list of messages including system, history, and current input
     messages = [{"role": "system", "content": system_message}]
-
     messages.extend(history)
-
     messages.append({"role": "user", "content": message})
 
-    response = ""
+    # Apply the model's specific chat template
+    input_ids = tokenizer.apply_chat_template(
+        messages, 
+        return_tensors="pt", 
+        add_generation_prompt=True
+    ).to(model.device)
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
+    # 4. Setup Streaming
+    streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
+
+    generate_kwargs = dict(
+        input_ids=input_ids,
+        streamer=streamer,
+        max_new_tokens=max_tokens,
+        do_sample=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
+    # 5. Run generation in a separate thread so we can yield tokens
+    t = Thread(target=model.generate, kwargs=generate_kwargs)
+    t.start()
 
+    # 6. Yield output as it generates
+    partial_message = ""
+    for new_token in streamer:
+        partial_message += new_token
+        yield partial_message
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
+# 7. Gradio Interface
 chatbot = gr.ChatInterface(
     respond,
     type="messages",
     additional_inputs=[
-        gr.Textbox(value="You are an AI called Concept. Your made for programming in any type of code.", label="System message"),
+        gr.Textbox(value="You are an AI called Concept. You are made for programming in any type of code.", 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(
@@ -61,10 +96,9 @@ chatbot = gr.ChatInterface(
 )
 
 with gr.Blocks() as demo:
-    with gr.Sidebar():
-        gr.LoginButton()
+    # Removed LoginButton because local execution usually relies on environment login
+    # or public models.
     chatbot.render()
 
-
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()