Update app.py

app.py

@@ -1,85 +1,148 @@
+import torch
+from PIL import Image
 import gradio as gr
+import spaces
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
+import os
+from threading import Thread
 
-import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer
-from transformers import TextStreamer
 
-import spaces
+HF_TOKEN = os.environ.get("HF_TOKEN", None)
+MODELS = "Rorical/0-roleplay"
+MODEL_NAME = MODELS.split("/")[-1]
+
+TITLE = "<h1><center>Rorical/0-roleplay</center></h1>"
 
-# Load model and tokenizer
-model = AutoModelForCausalLM.from_pretrained("Rorical/0-roleplay", return_dict=True, trust_remote_code=True, device_map="auto")
-tokenizer = AutoTokenizer.from_pretrained("Rorical/0-roleplay", trust_remote_code=True)
+DESCRIPTION = f"""
+<h3>MODEL: <a href="https://hf.co/{MODELS}">{MODEL_NAME}</a></h3>
+<center>
+<p>Qwen is the large language model built by Alibaba Cloud.
+<br>
+Feel free to test without log.
+</p>
+</center>
+"""
+
+CSS = """
+.duplicate-button {
+margin: auto !important;
+color: white !important;
+background: black !important;
+border-radius: 100vh !important;
+}
+h3 {
+text-align: center;
+}
+"""
+
+model = AutoModelForCausalLM.from_pretrained(
+          MODELS,
+          torch_dtype=torch.float16,
+          device_map="auto",
+        )
+tokenizer = AutoTokenizer.from_pretrained(MODELS)
 tokenizer.chat_template = "{% for message in messages %}{{'' + ((message['role'] + '\n') if message['role'] != '' else '') + message['content'] + '' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '星野\n' }}{% endif %}"
 
-# Define the response function
 @spaces.GPU
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    # Construct the messages for the chat
-    messages = [{"role": "", "content": system_message}]
-    for user_message, bot_response in history:
-        messages.append({"role": "老师", "content": user_message})  # Assuming the user is "老师"
-        messages.append({"role": "星野", "content": bot_response})  # Assuming the bot is "星野"
-    messages.append({"role": "老师", "content": message})  # Append the latest user message
+def stream_chat(message: str, history: list, temperature: float, max_new_tokens: int, top_p: float, top_k: int, penalty: float):
+    print(f'message is - {message}')
+    print(f'history is - {history}')
+    conversation = []
+    for prompt, answer in history:
+        conversation.extend([{"role": "user", "content": prompt}, {"role": "assistant", "content": answer}])
+    conversation.append({"role": "user", "content": message})
+
+    print(f"Conversation is -\n{conversation}")
     
-    # Tokenize and prepare inputs
-    inputs = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt")
-    inputs = inputs.to("cuda")
+    input_ids = tokenizer.apply_chat_template(conversation, tokenize=False, add_generation_prompt=True)
+    inputs = tokenizer(input_ids, return_tensors="pt").to(0)
     
-    # Generate response
-    streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
-    generate_ids = model.generate(
-        inputs,
-        max_length=max_tokens,
-        temperature=temperature,
+    streamer = TextIteratorStreamer(tokenizer, timeout=10., skip_prompt=True, skip_special_tokens=True)
+
+    generate_kwargs = dict(
+        inputs, 
+        streamer=streamer,
+        top_k=top_k,
         top_p=top_p,
-        streamer=streamer
+        repetition_penalty=penalty,
+        max_new_tokens=max_new_tokens, 
+        do_sample=True, 
+        temperature=temperature,
+        eos_token_id = [151645, 151643],
     )
     
-    # Decode the generated response
-    response = tokenizer.decode(generate_ids[0], skip_special_tokens=True)
-    response = response.replace("星野\n", "", 1)  # Remove the "星野\n" prefix if it exists
-    response = response.split("老师\n")[0]  # Split on the next possible user input
-    
-    return response, history + [(message, response)]
-
-# Default prompt for the chatbot
-prompt = """以下是小鸟游星野的介绍
-星野是阿拜多斯高中对策委员会的委员长，同时也是学生会副主席。语气懒散，经常自称为大叔，实际上是自己默默承担一切的女生。
-比起工作，她更喜欢玩。 正因为如此，她经常被委员会的其他人骂。 但是，一旦任务开始，她就会在前线勇敢地战斗以保护她的战友。
-她在阿拜多斯上高中。与星野一起在对策委员会的成员有白子，茜香，野乃美，和绫音。
-星野的年龄是17岁，生日为1月2日。
-星野有一头粉红色的头发，头巾一直长到她的腿上。 
-星野有蓝色和橙色眼睛的异色症。
-星野其实更符合认真而默默努力的类型。她实际上不相信其它的学校和大人，是对策委员会中最谨慎保守的人。当然，这并不妨碍老师和星野增进关系，成为她唯一信任的大人。
-是萝莉、有呆毛、天然萌���早熟、学生会副会长、异色瞳、慵懒。
-星野对海洋动物很感兴趣，对鱼类的知识了解得不少。她在拿到附录中包含2000多种热带鱼图鉴的书后，迫不及待地找了家店坐下来阅读。
-在众多海洋动物中，星野最喜欢的当属鲸鱼，情人节时星野还在海洋馆买了鲸鱼的巧克力作为纪念。
-星野还对寻宝有着十分浓厚的兴趣，曾和老师探索了阿拜多斯多个角落。
-星野给人一种白天睡不醒的瞌睡虫形象。"""
-
-# Create the Gradio interface
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value=prompt, label="System message", lines=5),
-        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)",
-        ),
-    ],
-)
+    thread = Thread(target=model.generate, kwargs=generate_kwargs)
+    thread.start()
+
+    buffer = ""
+    for new_text in streamer:
+        buffer += new_text
+        yield buffer
+
+
+
+chatbot = gr.Chatbot(height=450)
+
+with gr.Blocks(css=CSS) as demo:
+    gr.HTML(TITLE)
+    gr.HTML(DESCRIPTION)
+    gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
+    gr.ChatInterface(
+        fn=stream_chat,
+        chatbot=chatbot,
+        fill_height=True,
+        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
+        additional_inputs=[
+            gr.Slider(
+                minimum=0,
+                maximum=1,
+                step=0.1,
+                value=0.8,
+                label="Temperature",
+                render=False,
+            ),
+            gr.Slider(
+                minimum=128,
+                maximum=4096,
+                step=1,
+                value=1024,
+                label="Max new tokens",
+                render=False,
+            ),
+            gr.Slider(
+                minimum=0.0,
+                maximum=1.0,
+                step=0.1,
+                value=0.8,
+                label="top_p",
+                render=False,
+            ),
+            gr.Slider(
+                minimum=1,
+                maximum=20,
+                step=1,
+                value=20,
+                label="top_k",
+                render=False,
+            ),
+            gr.Slider(
+                minimum=0.0,
+                maximum=2.0,
+                step=0.1,
+                value=1.0,
+                label="Repetition penalty",
+                render=False,
+            ),
+        ],
+        examples=[
+            ["Help me study vocabulary: write a sentence for me to fill in the blank, and I'll try to pick the correct option."],
+            ["What are 5 creative things I could do with my kids' art? I don't want to throw them away, but it's also so much clutter."],
+            ["Tell me a random fun fact about the Roman Empire."],
+            ["Show me a code snippet of a website's sticky header in CSS and JavaScript."],
+        ],
+        cache_examples=False,
+    )
+
 
 if __name__ == "__main__":
     demo.launch()