import argparse import torch import os import json from tqdm import tqdm import shortuuid import numpy as np from longva.constants import IMAGE_TOKEN_INDEX from longva.longva.conversation import conv_templates from longva.model.builder import load_pretrained_model from longva.mm_utils import tokenizer_image_token, process_images,transform_input_id from torch.utils.data import Dataset, DataLoader from PIL import Image import math from longva.model.builder import load_pretrained_model from longva.mm_utils import tokenizer_image_token, process_images from longva.constants import IMAGE_TOKEN_INDEX def split_list(lst, n): """Split a list into n (roughly) equal-sized chunks""" chunk_size = math.ceil(len(lst) / n) # integer division return [lst[i:i + chunk_size] for i in range(0, len(lst), chunk_size)] def get_chunk(lst, n, k): chunks = split_list(lst, n) return chunks[k] # Custom dataset class class CustomDataset(Dataset): def __init__(self, questions, image_folder, tokenizer, image_processor, model_config): self.questions = questions self.image_folder = image_folder self.tokenizer = tokenizer self.image_processor = image_processor self.model_config = model_config def __getitem__(self, index): line = self.questions[index] image_file = line["image"] qs = line["text"] # qs = "" + '\n' + qs # conv = conv_templates[args.conv_mode].copy() # conv.append_message(conv.roles[0], qs) # conv.append_message(conv.roles[1], None) # prompt = conv.get_prompt() # prompt = "<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\n\nDescribe the image in details.<|im_end|>\n<|im_start|>assistant\n" prompt = f"<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\n\n{qs}<|im_end|>\n<|im_start|>assistant\n" if ".mp4" in image_file: new_path=os.path.join(self.image_folder,image_file.replace(".mp4","")) num_images =len(os.listdir(new_path)) frames = [] for n in range(1, num_images + 1): # 假设 num_images 是图片数量 image_path = os.path.join(new_path, f"{n}.png") # 图片名称为1.png, 2.png, ... with Image.open(image_path) as frame: frame = np.array(frame) frames.append(frame) image_tensor = self.image_processor.preprocess(frames, return_tensors="pt")["pixel_values"] size=[0] flag=["video"] else: image = Image.open(os.path.join(self.image_folder, image_file)).convert('RGB') image_tensor = process_images([image], self.image_processor, self.model_config) size=[image.size] flag=["image"] input_ids = tokenizer_image_token(prompt, self.tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt') return input_ids, image_tensor, size, flag def __len__(self): return len(self.questions) # DataLoader def create_data_loader(questions, image_folder, tokenizer, image_processor, model_config, batch_size=1, num_workers=4): assert batch_size == 1, "batch_size must be 1" dataset = CustomDataset(questions, image_folder, tokenizer, image_processor, model_config) data_loader = DataLoader(dataset, batch_size=batch_size, num_workers=num_workers, shuffle=False) return data_loader def eval_model(args): # Model tokenizer, model, image_processor, _ = load_pretrained_model(args.model_path, None, "llava_qwen", device_map="cuda:0") questions = [json.loads(q) for q in open(os.path.expanduser(args.question_file), "r")] questions = get_chunk(questions, args.num_chunks, args.chunk_idx) answers_file = os.path.expanduser(args.answers_file) os.makedirs(os.path.dirname(answers_file), exist_ok=True) ans_file = open(answers_file, "w") data_loader = create_data_loader(questions, args.image_folder, tokenizer, image_processor, model.config) gen_kwargs = {"do_sample": False, "temperature": 0, "top_p": None, "num_beams": 1, "use_cache": True, "max_new_tokens": 128} for (input_ids, image_tensor, size,flag), line in tqdm(zip(data_loader, questions), total=len(questions)): model.memory.reset() idx = line["question_id"] cur_prompt = line["text"] image_tensor=image_tensor.squeeze(0).to('cuda', dtype=torch.float16) input_ids = input_ids.to(device='cuda', non_blocking=True) if flag[0][0]=="image": num_tokens=(image_tensor.shape[1]-1) *144 with torch.inference_mode(): output_ids = model.generate(input_ids, images=image_tensor, image_sizes=size, modalities=["image"],**gen_kwargs) elif flag[0][0]=="video": num_tokens=(image_tensor.shape[0]) *144 with torch.inference_mode(): output_ids = model.generate(input_ids, images=[image_tensor], modalities=["video"],**gen_kwargs) transform_input_ids=transform_input_id(input_ids,num_tokens,model.config.vocab_size-1) output_ids=output_ids[:,transform_input_ids.shape[1]:] outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip() ans_id = shortuuid.uuid() ans_file.write(json.dumps({"question_id": idx, "prompt": cur_prompt, "text": outputs, "answer_id": ans_id, "model_id": "long_qwen", "metadata": {}}) + "\n") # ans_file.flush() ans_file.close() if __name__ == "__main__": parser = argparse.ArgumentParser() parser.add_argument("--model-path", type=str, default=None) parser.add_argument("--model-base", type=str, default=None) parser.add_argument("--model-type", type=str, default=None) parser.add_argument("--image-folder", type=str, default=None) parser.add_argument("--question-file", type=str, default=None) parser.add_argument("--answers-file", type=str, default=None) parser.add_argument("--conv-mode", type=str, default=None) parser.add_argument("--num-chunks", type=int, default=1) parser.add_argument("--chunk-idx", type=int, default=0) parser.add_argument("--temperature", type=float, default=0.2) parser.add_argument("--top_p", type=float, default=None) parser.add_argument("--num_beams", type=int, default=1) parser.add_argument("--max_new_tokens", type=int, default=128) args = parser.parse_args() eval_model(args)