generate.py

#
# Copyright 2016 The BigDL Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

import torch
import time
import argparse
import numpy as np

from ipex_llm.transformers import Qwen2VLForConditionalGeneration
from qwen_vl_utils import process_vision_info


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='Predict Tokens using generate() API for Qwen2-VL model')
    parser.add_argument('--repo-id-or-model-path', type=str, default="Qwen/Qwen2-VL-7B-Instruct",
                        help='The huggingface repo id for the Qwen2-VL model to be downloaded'
                             ', or the path to the huggingface checkpoint folder')
    parser.add_argument('--prompt', type=str, default="图片里有什么？",
                        help='Prompt to infer') 
    parser.add_argument('--image-url-or-path', type=str,
                        default='http://farm6.staticflickr.com/5268/5602445367_3504763978_z.jpg' ,
                        help='The URL or path to the image to infer')
    
    parser.add_argument('--n-predict', type=int, default=32,
                        help='Max tokens to predict')
    parser.add_argument('--modelscope', action="store_true", default=False, 
                        help="Use models from modelscope")

    args = parser.parse_args()
    if args.modelscope:
        from modelscope import AutoProcessor
        model_hub = 'modelscope'
    else:
        from transformers import AutoProcessor
        model_hub = 'huggingface'
        
    model_path = args.repo_id_or_model_path

    model = Qwen2VLForConditionalGeneration.from_pretrained(model_path,
                                                            load_in_4bit=True,
                                                            optimize_model=True,
                                                            trust_remote_code=True,
                                                            modules_to_not_convert=["vision"],
                                                            use_cache=True,
                                                            model_hub=model_hub)

    # Use .float() for better output, and use .half() for better speed
    model = model.half().to("xpu")

    # The following code for generation is adapted from https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct#quickstart

    # The default range for the number of visual tokens per image in the model is 4-16384.
    # You can set min_pixels and max_pixels according to your needs, such as a token count range of 256-1280,
    # to balance speed and memory usage.
    min_pixels = 256*28*28
    max_pixels = 1280*28*28
    processor = AutoProcessor.from_pretrained(model_path, min_pixels=min_pixels, max_pixels=max_pixels)

    prompt = args.prompt
    image_path = args.image_url_or_path

    with torch.inference_mode():
        messages = [
            {
                "role": "user",
                "content": [
                    {
                        "type": "image",
                        "image": image_path,
                    },
                    {"type": "text", "text": prompt},
                ],
            }
        ]
        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('xpu')

        # ipex_llm model needs a warmup, then inference time can be accurate
        generated_ids = model.generate(
            **inputs,
            max_new_tokens=args.n_predict
        )

        st = time.time()
        generated_ids = model.generate(
            **inputs,
            max_new_tokens=args.n_predict
        )
        torch.xpu.synchronize()
        end = time.time()
        generated_ids = generated_ids.cpu()
        generated_ids = [
            output_ids[len(input_ids):] for input_ids, output_ids in zip(inputs.input_ids, generated_ids)
        ]

        response = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
        print(f'Inference time: {end-st} s')
        print('-'*20, 'Input Image', '-'*20)
        print(image_path)
        print('-'*20, 'Prompt', '-'*20)
        print(prompt)
        print('-'*20, 'Output', '-'*20)
        print(response)