Update app.py

app.py

@@ -2,6 +2,7 @@ import gradio as gr
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 import spaces
 from threading import Thread
+from typing import Iterator
 
 # Load model and tokenizer
 model_name = "Magpie-Align/MagpieLM-4B-Chat-v0.1"
@@ -16,27 +17,25 @@ model.to(device)
 
 MAX_INPUT_TOKEN_LENGTH = 4096  # You may need to adjust this value
 
-@spaces.GPU(enable_queue=True)
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens=2048,
-    temperature=0.6,
-    top_p=0.9,
-    repetition_penalty=1.0,
-):
-    messages = [{"role": "system", "content": system_message}]
+@spaces.GPU
+def generate(
+    message: str,
+    chat_history: list[tuple[str, str]],
+    system_prompt: str,
+    max_new_tokens: int = 1024,
+    temperature: float = 0.6,
+    top_p: float = 0.9,
+    top_k: int = 50,
+    repetition_penalty: float = 1.2,
+) -> Iterator[str]:
+    conversation = []
+    if system_prompt:
+        conversation.append({"role": "system", "content": system_prompt})
+    for user, assistant in chat_history:
+        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
+    conversation.append({"role": "user", "content": 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})
- 
-    input_ids = tokenizer.apply_chat_template(messages, return_tensors="pt")
+    input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt")
     if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH:
         input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
         gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
@@ -46,36 +45,31 @@ def respond(
     generate_kwargs = dict(
         input_ids=input_ids,
         streamer=streamer,
-        max_new_tokens=max_tokens,
+        max_new_tokens=max_new_tokens,
         do_sample=True,
         top_p=top_p,
+        top_k=top_k,
         temperature=temperature,
+        num_beams=1,
         repetition_penalty=repetition_penalty,
     )
-    
-    thread = Thread(target=model.generate, kwargs=generate_kwargs)
-    thread.start()
-
-    def stream():
-        for text in streamer:
-            yield text
+    t = Thread(target=model.generate, kwargs=generate_kwargs)
+    t.start()
 
-    return stream()
+    outputs = []
+    for text in streamer:
+        outputs.append(text)
+        yield "".join(outputs)
 
 demo = gr.ChatInterface(
-    respond,
+    generate,
     additional_inputs=[
-        gr.Textbox(value="You are Magpie, a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
+        gr.Textbox(value="You are Magpie, a friendly Chatbot.", label="System prompt"),
+        gr.Slider(minimum=1, maximum=2048, value=1024, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.6, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.9,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-        gr.Slider(minimum=0.5, maximum=1.5, value=1.0, step=0.1, label="Repetition Penalty"),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.9, step=0.05, label="Top-p"),
+        gr.Slider(minimum=1, maximum=100, value=50, step=1, label="Top-k"),
+        gr.Slider(minimum=0.5, maximum=2.0, value=1.2, step=0.1, label="Repetition Penalty"),
     ],
 )