Update app.py

app.py

@@ -2,27 +2,33 @@ import os
 import gradio as gr
 import spaces
 import torch
+from threading import Thread
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
+# Constants for model behavior
 MAX_MAX_NEW_TOKENS = 2048
 DEFAULT_MAX_NEW_TOKENS = 1024
 MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
 
+# Models selection
+MODELS = {
+    "Fast-Model": "Artples/L-MChat-Small",
+    "Quality-Model": "Artples/L-MChat-7b"
+}
+
+# Description for the application
 DESCRIPTION = """\
 # L-MChat
 This Space demonstrates [L-MChat](https://huggingface.co/collections/Artples/l-mchat-663265a8351231c428318a8f) by L-AI.
 """
 
+# Check for GPU availability
 if not torch.cuda.is_available():
     DESCRIPTION += "\n<p>Running on CPU! This demo does not work on CPU.</p>"
 
-model_details = {
-    "Fast-Model": "Artples/L-MChat-Small",
-    "Quality-Model": "Artples/L-MChat-7b"
-}
-
-models = {name: AutoModelForCausalLM.from_pretrained(model_id, device_map="auto") for name, model_id in model_details.items()}
-tokenizers = {name: AutoTokenizer.from_pretrained(model_id) for name, model_id in model_details.items()}
+# Load models and tokenizers
+models = {name: AutoModelForCausalLM.from_pretrained(model_id, device_map="auto") for name, model_id in MODELS.items()}
+tokenizers = {name: AutoTokenizer.from_pretrained(model_id) for name, model_id in MODELS.items()}
 
 @spaces.GPU(enable_queue=True, duration=90)
 def generate(
@@ -39,35 +45,52 @@ def generate(
     model = models[model_choice]
     tokenizer = tokenizers[model_choice]
 
-    conversation = [{"role": "system", "content": system_prompt}] if system_prompt else []
-    conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}] for user, assistant in chat_history)
+    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})
 
     input_ids = tokenizer(conversation, return_tensors="pt", truncation=True, max_length=MAX_INPUT_TOKEN_LENGTH).input_ids
     input_ids = input_ids.to(model.device)
 
-    output_ids = model.generate(input_ids, max_length=MAX_INPUT_TOKEN_LENGTH + max_new_tokens, temperature=temperature, top_p=top_p, top_k=top_k, repetition_penalty=repetition_penalty)
-    output_text = tokenizer.decode(output_ids[0], skip_special_tokens=True)
-
-    return output_text
+    output_ids = model.generate(
+        input_ids,
+        max_length=input_ids.shape[1] + max_new_tokens,
+        temperature=temperature,
+        top_p=top_p,
+        top_k=top_k,
+        repetition_penalty=repetition_penalty,
+        num_return_sequences=1,
+    )
 
+    return tokenizer.decode(output_ids[0], skip_special_tokens=True)
 
-chat_interface = gr.ChatInterface(
-    theme='ehristoforu/RE_Theme',
+# Gradio Interface
+chat_interface = gr.Interface(
     fn=generate,
-    additional_inputs=[gr.Textbox(label="System prompt", lines=6), gr.Dropdown(label="Model Choice", choices=list(model_details.keys()), value="Quality-Model")],
-    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?"],
-        ["Write a 100-word article on 'Benefits of Open-Source in AI research'"],
+    inputs=[
+        gr.Dropdown(label="Choose Model", choices=list(MODELS.keys()), default="Quality-Model"),
+        gr.ChatBox(),
+        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=0.1, maximum=4.0, step=0.1, value=0.6),
+        gr.Slider(label="Top-p (nucleus sampling)", minimum=0.05, maximum=1.0, step=0.05, value=0.9),
+        gr.Slider(label="Top-k", minimum=1, maximum=1000, step=1, value=50),
+        gr.Slider(label="Repetition penalty", minimum=1.0, maximum=2.0, step=0.05, value=1.2),
     ],
+    outputs="text",
+    theme='ehristoforu/RE_Theme',
+    examples=[
+        ["Quality-Model", "Hello there! How are you doing?", [], "Let's start the conversation.", 1024, 0.6, 0.9, 50, 1.2]
+    ]
 )
 
+# Main execution block
 with gr.Blocks(css="style.css") as demo:
     gr.Markdown(DESCRIPTION)
     chat_interface.render()
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.queue(max_size=20).launch()