-
Notifications
You must be signed in to change notification settings - Fork 15
/
model_vqa_mmbench.py
167 lines (138 loc) · 6.71 KB
/
model_vqa_mmbench.py
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
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
import argparse
import torch
import os
import json
import pandas as pd
from tqdm import tqdm
import shortuuid
from vcoder_llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN
from vcoder_llava.vcoder_conversation import conv_templates, SeparatorStyle
from vcoder_llava.model.builder import load_pretrained_model
from vcoder_llava.utils import disable_torch_init
from vcoder_llava.mm_utils import tokenizer_image_token, process_images, load_image_from_base64, get_model_name_from_path
from PIL import Image
import math
all_options = ['A', 'B', 'C', 'D']
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 is_none(value):
if value is None:
return True
if type(value) is float and math.isnan(value):
return True
if type(value) is str and value.lower() == 'nan':
return True
if type(value) is str and value.lower() == 'none':
return True
return False
def get_options(row, options):
parsed_options = []
for option in options:
option_value = row[option]
if is_none(option_value):
break
parsed_options.append(option_value)
return parsed_options
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)
tokenizer, model, image_processor, _, _, context_len = load_pretrained_model(model_path, args.model_base, model_name)
questions = pd.read_table(os.path.expanduser(args.question_file))
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")
if 'plain' in model_name and 'finetune' not in model_name.lower() and 'mmtag' not in args.conv_mode:
args.conv_mode = args.conv_mode + '_mmtag'
print(f'It seems that this is a plain model, but it is not using a mmtag prompt, auto switching to {args.conv_mode}.')
for index, row in tqdm(questions.iterrows(), total=len(questions)):
options = get_options(row, all_options)
cur_option_char = all_options[:len(options)]
if args.all_rounds:
num_rounds = len(options)
else:
num_rounds = 1
for round_idx in range(num_rounds):
idx = row['index']
question = row['question']
hint = row['hint']
image = load_image_from_base64(row['image'])
if not is_none(hint):
question = hint + '\n' + question
for option_char, option in zip(all_options[:len(options)], options):
question = question + '\n' + option_char + '. ' + option
qs = cur_prompt = question
qs = DEFAULT_IMAGE_TOKEN + '\n' + qs
if args.single_pred_prompt:
if args.lang == 'cn':
qs = qs + '\n' + "请直接回答选项字母。"
else:
qs = qs + '\n' + "Answer with the option's letter from the given choices directly."
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()
image_tensor = process_images([image], image_processor, model.config)[0]
# image_tensor = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0]
stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2
with torch.inference_mode():
output_ids = model.generate(
input_ids,
images=image_tensor.unsqueeze(0).half().cuda(),
do_sample=True if args.temperature > 0 else False,
temperature=args.temperature,
top_p=args.top_p,
num_beams=args.num_beams,
# no_repeat_ngram_size=3,
max_new_tokens=1024,
use_cache=True)
input_token_len = input_ids.shape[1]
n_diff_input_output = (input_ids != 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)[0]
outputs = outputs.strip()
if outputs.endswith(stop_str):
outputs = outputs[:-len(stop_str)]
outputs = outputs.strip()
ans_id = shortuuid.uuid()
ans_file.write(json.dumps({"question_id": idx,
"round_id": round_idx,
"prompt": cur_prompt,
"text": outputs,
"options": options,
"option_char": cur_option_char,
"answer_id": ans_id,
"model_id": model_name,
"metadata": {}}) + "\n")
ans_file.flush()
# rotate options
options = options[1:] + options[:1]
cur_option_char = cur_option_char[1:] + cur_option_char[:1]
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("--all-rounds", action="store_true")
parser.add_argument("--single-pred-prompt", action="store_true")
parser.add_argument("--lang", type=str, default="en")
args = parser.parse_args()
eval_model(args)