app.py

from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer, Qwen2VLForConditionalGeneration, AutoProcessor
import spaces
import gradio as gr
import os
from threading import Thread
from PIL import Image
from qwen_vl_utils import process_vision_info

# model_name = "Qwen/Qwen2.5-7B-Instruct"
model_name = "Qwen/Qwen2.5-7B-Instruct"

# model = AutoModelForCausalLM.from_pretrained(
#     model_name, 
#     torch_dtype="auto", 
#     device_map='auto'
# )
# model = Qwen2VLForConditionalGeneration.from_pretrained(model_name, trust_remote_code=True, torch_dtype="auto").cuda().eval()
# model = Qwen2VLForConditionalGeneration.from_pretrained(model_name, trust_remote_code=True, torch_dtype=torch.float16).to("cuda").eval()

# tokenizer = AutoTokenizer.from_pretrained(model_name)

model = Qwen2VLForConditionalGeneration.from_pretrained(
    "Qwen/Qwen2-VL-72B-Instruct-AWQ", torch_dtype="auto", device_map="auto"
)
processor = AutoProcessor.from_pretrained(model_name, trust_remote_code=True)

@spaces.GPU
def stream_chat(message: str, history: list, temperature: float, max_new_tokens: int, top_p: float, top_k: int, penalty: float):
    messages = [
        {
            "role": "user",
            "content": [
                {
                    "type": "image",
                    "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg",
                },
                {"type": "text", "text": "Describe this image."},
            ],
        }
    ]
    # Preparation for inference
    text = processor.apply_chat_template(
        messages, tokenize=False, add_generation_prompt=True
    )
    image_inputs, video_inputs = process_vision_info(messages)
    inputs = processor(
        text=[text],
        images=image_inputs,
        videos=video_inputs,
        padding=True,
        return_tensors="pt",
    )
    inputs = inputs.to("cuda")
    generated_ids = model.generate(**inputs, max_new_tokens=128)
    generated_ids_trimmed = [
        out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
    ]
    output_text = processor.batch_decode(
        generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
    )
    print(output_text)
    return output_text



    # 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": [
    #     {
    #         "type": "text",
    #         "text": "这是什么"
    #     },
    #     {
    #         "type": "image_url",
    #         "image_url": {
    #             "url": "https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg"
    #         }
    #     }
    # ]})

    # print(f"Conversation is -\n{conversation}")
    
    # input_ids = tokenizer.apply_chat_template(conversation, tokenize=False, add_generation_prompt=True)
    # inputs = tokenizer(input_ids, return_tensors="pt").to(0)
    
    # 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,
    #     repetition_penalty=penalty,
    #     max_new_tokens=max_new_tokens, 
    #     do_sample=True, 
    #     temperature=temperature,
    #     eos_token_id = [151645, 151643],
    # )
    
    # 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() as demo:
    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()