Update app.py

app.py

@@ -2,8 +2,8 @@ import os
 from threading import Thread
 from typing import Iterator
 
-import spaces
 import gradio as gr
+import spaces
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 
@@ -16,7 +16,8 @@ if torch.cuda.is_available():
     model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16, device_map="auto")
     tokenizer = AutoTokenizer.from_pretrained(model_id)
 
-def generate_and_display(
+@spaces.GPU
+def generate(
     message: str,
     chat_history: list[tuple[str, str]],
     system_prompt: str,
@@ -24,7 +25,7 @@ def generate_and_display(
     temperature: float = 0.7,
     top_p: float = 1.0,
     repetition_penalty: float = 1.1,
-) -> str:
+) -> Iterator[str]:
     conversation = []
     if system_prompt:
         conversation.append({"role": "system", "content": system_prompt})
@@ -35,12 +36,12 @@ def generate_and_display(
     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.")
+        gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
     input_ids = input_ids.to(model.device)
 
     streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
     generate_kwargs = dict(
-        input_ids=input_ids,
+        {"input_ids": input_ids},
         streamer=streamer,
         max_new_tokens=max_new_tokens,
         do_sample=True,
@@ -49,71 +50,59 @@ def generate_and_display(
         num_beams=1,
         repetition_penalty=repetition_penalty,
     )
+    t = Thread(target=model.generate, kwargs=generate_kwargs)
+    t.start()
 
     outputs = []
-    model_outputs = model.generate(**generate_kwargs)
-    for text in streamer.generate_from_iterator(model_outputs):
+    for text in streamer:
         outputs.append(text)
-    return "".join(outputs)
-
-def generate_response():
-    outputs = generate_and_display(
-        input_textbox.value,
-        chat_history=[],
-        system_prompt=system_prompt_textbox.value,
-        max_new_tokens=max_new_tokens_slider.value,
-        temperature=temperature_slider.value,
-        top_p=top_p_slider.value,
-        repetition_penalty=repetition_penalty_slider.value,
-    )
-    chat_output_textbox.value = outputs
+        yield "".join(outputs)
 
-input_textbox = gr.Textbox(label="User Input", lines=5, placeholder="Enter your message...")
-system_prompt_textbox = gr.Textbox(label="System Prompt", lines=5, placeholder="Enter system prompt (optional)...")
-max_new_tokens_slider = gr.Slider(
-    label="Max New Tokens",
-    minimum=1,
-    maximum=MAX_MAX_NEW_TOKENS,
-    step=1,
-    value=DEFAULT_MAX_NEW_TOKENS,
-)
-temperature_slider = gr.Slider(
-    label="Temperature",
-    minimum=0.01,
-    maximum=1.0,
-    step=0.01,
-    value=0.7,
-)
-top_p_slider = gr.Slider(
-    label="Top-p (Nucleus Sampling)",
-    minimum=0.05,
-    maximum=1.0,
-    step=0.01,
-    value=1.0,
-)
-repetition_penalty_slider = gr.Slider(
-    label="Repetition Penalty",
-    minimum=1.0,
-    maximum=2.0,
-    step=0.05,
-    value=1.1,
-)
-generate_button = gr.Button(label="Generate Response", command=generate_response)
-chat_output_textbox = gr.Textbox(label="Chat Output", lines=10)
-
-gr.Interface(
-    generate_and_display,
-    inputs=[input_textbox, system_prompt_textbox, max_new_tokens_slider, temperature_slider, top_p_slider, repetition_penalty_slider],
-    outputs=chat_output_textbox,
-    title="🦣MAmmoTH2",
-    description="A simple web interactive chat demo based on gradio.",
+chat_interface = gr.ChatInterface(
+    fn=generate,
+    additional_inputs=[
+        gr.Textbox(label="System prompt", lines=6),
+        gr.Slider(
+            label="Max new tokens",
+            minimum=1,
+            maximum=MAX_MAX_NEW_TOKENS,
+            step=1,
+            value=DEFAULT_MAX_NEW_TOKENS,
+        ),
+        gr.Slider(
+            label="Temperature",
+            minimum=0.01,
+            maximum=1.0,
+            step=0.01,
+            value=0.7,
+        ),
+        gr.Slider(
+            label="Top-p (nucleus sampling)",
+            minimum=0.05,
+            maximum=1.0,
+            step=0.01,
+            value=1.0,
+        ),
+        gr.Slider(
+            label="Repetition penalty",
+            minimum=1.0,
+            maximum=2.0,
+            step=0.05,
+            value=1.1,
+        ),
+    ],
+    stop_btn=None,
     examples=[
         ["Hello there! How are you doing?"],
         ["Can you explain briefly to me what is the Python programming language?"],
         ["Explain the plot of Cinderella in a sentence."],
-        ["How many hours does it take a man to eat a helicopter?"],
+        ["How many hours does it take a man to eat a Helicopter?"],
         ["Write a 100-word article on 'Benefits of Open-Source in AI research'"],
     ],
-    theme="default",
-    live=True,
-).launch()
+)
+
+with gr.Blocks(css="style.css") as demo:
+    chat_interface.render()
+
+if __name__ == "__main__":
+    demo.queue(max_size=20).launch()