File size: 6,235 Bytes
4963c36 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 |
import argparse
import torch
import os
import json
from tqdm import tqdm
import shortuuid
from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN
from llava.conversation import conv_templates, SeparatorStyle
from llava.model.builder import load_pretrained_model
from llava.utils import disable_torch_init
from llava.mm_utils import tokenizer_image_token, get_model_name_from_path, KeywordsStoppingCriteria
from PIL import Image
import math
import re
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]
def eval_model(args):
# Model
disable_torch_init()
model_path = os.path.expanduser(args.model_path)
model_name = get_model_name_from_path(model_path)
# import pudb; pudb.set_trace()
tokenizer, model, image_processor, context_len = load_pretrained_model(args.model_path, args.model_base, model_name)
questions=[]
# with open(os.path.expanduser(args.question_file), "r") as f:
# questions = json.load(f)
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")
for i in tqdm(range(0,len(questions),args.batch_size)):
input_batch=[]
input_image_batch=[]
count=i
image_folder=[]
batch_end = min(i + args.batch_size, len(questions))
for j in range(i,batch_end):
image_file=questions[j]['image_id']
qs = questions[j]['question']
# import pudb; pudb.set_trace()
qs="[refer] output the bounding box of the <ref>" + qs + "</ref> in the image."
if model.config.mm_use_im_start_end:
qs = DEFAULT_IM_START_TOKEN + DEFAULT_IMAGE_TOKEN + DEFAULT_IM_END_TOKEN + '\n' + qs
else:
qs = DEFAULT_IMAGE_TOKEN + '\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()
input_ids = tokenizer_image_token(prompt, tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).cuda()
input_batch.append(input_ids)
image = Image.open(os.path.join(args.image_folder, image_file))
image = image.resize((336, 336))
image_folder.append(image)
stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2
keywords = [stop_str]
stopping_criteria = KeywordsStoppingCriteria(keywords, tokenizer, input_ids)
max_length = max(tensor.size(1) for tensor in input_batch)
final_input_list = [torch.cat((torch.zeros((1,max_length - tensor.size(1)), dtype=tensor.dtype,device=tensor.get_device()), tensor),dim=1) for tensor in input_batch]
final_input_tensors=torch.cat(final_input_list,dim=0)
image_tensor_batch = image_processor.preprocess(image_folder,crop_size ={'height': 336, 'width': 336},size = {'shortest_edge': 336}, return_tensors='pt')['pixel_values']
with torch.inference_mode():
output_ids = model.generate(final_input_tensors, images=image_tensor_batch.half().cuda(), do_sample=False , temperature=args.temperature, top_p=args.top_p, num_beams=1, max_new_tokens=256,length_penalty=2.0, use_cache=True)
# input_token_len = final_input_tensors.shape[1]
# n_diff_input_output = (final_input_tensors != output_ids[:, :input_token_len]).sum().item()
# if n_diff_input_output > 0:
# print(f'[Warning] {n_diff_input_output} output_ids are not the same as the input_ids')
# outputs = tokenizer.batch_decode(output_ids[:, input_token_len:], skip_special_tokens=True)
outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)
# import pudb; pudb.set_trace()
for k in range(0,len(final_input_list)):
output = outputs[k].strip()
if output.endswith(stop_str):
output = output[:-len(stop_str)]
output = output.strip()
ans_id = shortuuid.uuid()
ans_file.write(json.dumps({
"question_id": questions[count]["question_id"],
"image_id": questions[count]["image_id"],
"answer": output,
"bbox": questions[count]['bbox'],
"poly": questions[count]['poly'],
"question":questions[count]['question'],
"dataset": questions[count]['dataset']
}) + "\n")
count=count+1
ans_file.flush()
ans_file.close()
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--model-path", type=str, default="facebook/opt-350m")
parser.add_argument("--model-base", type=str, default=None)
parser.add_argument("--image-folder", type=str, default="")
parser.add_argument("--question-file", type=str, default="tables/question.jsonl")
parser.add_argument("--answers-file", type=str, default="answer.jsonl")
parser.add_argument("--conv-mode", type=str, default="llava_v1")
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("--batch_size",type=int, default=1)
args = parser.parse_args()
eval_model(args)
|