Update app.py

app.py

@@ -1,4 +1,3 @@
-from huggingface_hub import InferenceClient
 from unsloth import FastLanguageModel
 import torch
 import gradio as gr
@@ -6,91 +5,81 @@ import gradio as gr
 """
 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")
 model,tokenizer = FastLanguageModel.from_pretrained('./unified_model_v3')
         
-def generate_response(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    top_k
-):
-    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]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    # inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
-    # with torch.no_grad():
-    #     outputs = model.generate(**inputs,
-    #                              early_stopping=True,
-    #                              min_length=50,
-    #                              length_penalty=2,
-    #                              do_sample=True,
-    #                              max_new_tokens=max_tokens,
-    #                              top_p=top_p,
-    #                              top_k=top_k,
-    #                              temperature=temperature,
-    #                              repetition_penalty=1.2,
-    #                              num_return_sequences=1
-    #                     )
-    # response = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    # yield response
-    # response = response.split("### Answer:")[-1]
-
-    
-    for message in client.chat_completion(
-        messages=messages,
-        model=model,
-        repetition_penalty=1.2,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    generate_response,
-    additional_inputs=[
-        gr.Textbox(value="""Instruction:
+def generate_response(instruction,chat_history):
+    """Generates a response using your fine-tuned model."""
+    # FastLanguageModel.for_inference(model) # Enable native 2x faster inference within the function
+    prompt = f"""### Instruction:
     Respond to the user's question. Don't include code in your response.
 
     ### Question:
     {instruction}
 
     Please provide a unique, concise, and non-repetitive answer
-    ### Answer:""", 
-                   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)",
-        ),
-    ],
-)
-
-
-# if __name__ == "__main__":
+    ### Answer:"""
+
+    inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
+    with torch.no_grad():
+        outputs = model.generate(**inputs,early_stopping=True,min_length=50,length_penalty=2,do_sample=True,max_new_tokens=300,
+    top_p=0.95,
+    top_k=50,
+    temperature=0.7,
+    repetition_penalty=1.2,
+    num_return_sequences=1
+    )
+
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    response = response.split("### Answer:")[-1]
+    return response
+
+def update_chat_history(chat_history, user_message, bot_message):
+    """Update chat history to maintain relevance and avoid excessive growth."""
+    chat_history['user'].append(user_message)
+    chat_history['bot'].append(bot_message)
+    # Keep only the last N interactions
+    if len(chat_history['user']) > 5:
+        chat_history['user'] = chat_history['user'][-5:]
+        chat_history['bot'] = chat_history['bot'][-5:]
+    return chat_history
+
+def chatbot(input_text,chat_history):
+  messages = {
+      "user": [],
+      "bot": [],
+  }
+
+  for user_msg, bot_msg in chat_history:
+    messages["user"].append(user_msg)
+    messages["bot"].append(bot_msg)
+
+  bot_response = generate_response(input_text,messages)
+  chat_history.append(("User: " + input_text, bot_response))
+  messages = update_chat_history(messages, input_text, bot_response)
+  return "", chat_history
+
+with gr.Blocks() as demo:
+  gr.Markdown('## AILA INTERFACE DEMO')
+
+  with gr.Row():
+
+    user_input = gr.Textbox(
+        placeholder = "Type your message here...",
+        label = "Your Message",
+        lines = 1
+    )
+
+  submit_button = gr.Button('Submit')
+
+  chat_history = gr.Chatbot()
+
+
+
+  submit_button.click(
+      chatbot,
+      inputs = [user_input,chat_history],
+      outputs = [user_input, chat_history]
+  )
+
+
 demo.launch()