-
Notifications
You must be signed in to change notification settings - Fork 0
/
pope_llava.py
172 lines (150 loc) · 6.59 KB
/
pope_llava.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
168
169
170
171
import argparse
import torch
import os
import json
from tqdm import tqdm
import requests
from io import BytesIO
import shortuuid
import sys
import os
sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__)))))
sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
# print(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__)))))
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, process_images
from PIL import Image
import math
import re
# import kornia
from transformers import set_seed
# from vcd_utils.vcd_add_noise import add_diffusion_noise
# from vcd_utils.vcd_sample import evolve_vcd_sampling
# evolve_vcd_sampling()
def recorder(out):
word_list = re.split(r'[^\w]+', out.lower())
if "yes" in word_list:
return "Yes"
else:
return "No"
# def image_parser(args):
# out = args.image_file.split(args.sep)
# return out
def load_image(image_file):
# 预处理
if image_file.startswith("http") or image_file.startswith("https"):
response = requests.get(image_file)
image = Image.open(BytesIO(response.content)).convert("RGB")
else:
image = Image.open(image_file).convert("RGB")
return image
def load_images(image_files):
# 图片列表
out = []
for image_file in image_files:
image = load_image(image_file)
out.append(image)
return out
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,device="cuda:5")
questions = [json.loads(q) for q in open(os.path.expanduser(args.question_file), "r")]
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 line in tqdm(questions):
idx = line["question_id"]
image_file = line["image"]
qs = line["text"]
cur_prompt = qs
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
if "llama-2" in model_name.lower():
conv_mode = "llava_llama_2"
elif "v1" in model_name.lower():
conv_mode = "llava_v1"
elif "mpt" in model_name.lower():
conv_mode = "mpt"
else:
conv_mode = "llava_v0"
conv = conv_templates[conv_mode].copy()
conv.append_message(conv.roles[0], qs)# + " Please answer this question with one word.")
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(5)
image_files = [image_file]
images = load_images(image_files)
images_tensor = process_images(
images,
image_processor, # clip
model.config
).to(model.device, dtype=torch.float16)
stop_str = conv.sep if conv.sep_style != SeparatorStyle.TWO else conv.sep2
keywords = [stop_str]
stopping_criteria = KeywordsStoppingCriteria(keywords, tokenizer, input_ids)
with torch.inference_mode():
output_dict = model.generate(
input_ids,
images=images_tensor,
do_sample=True if args.temperature > 0 else False,
temperature=args.temperature,
top_p=args.top_p,
top_k=args.top_k,
max_new_tokens=5,
use_deco=True,
alpha = args.aplha,
threshold_top_p = args.threshold_top_p,
threshold_top_k = args.threshold_top_k,
early_exit_layers=[i for i in range(args.start_layer, args.end_layer)],
output_hidden_states=True,
return_dict_in_generate=True,
stopping_criteria=[stopping_criteria])
output_ids = output_dict.sequences
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()
print(idx)
print(outputs)
ans_file.write(json.dumps({"question_id": idx,
"prompt": cur_prompt,
"text": recorder(outputs),
"model_id": model_name,
"image": image_file,
"metadata": {}}) + "\n")
ans_file.flush()
ans_file.close()
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--model-path", type=str, default=".../llava-v1.5-7b")
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="")
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=-1)
parser.add_argument("--top_p", type=float, default=None)
parser.add_argument("--top_k", type=int, default=None)
parser.add_argument("--alpha", type=float, default=0.6)
parser.add_argument("--threshold_top_p", type=float, default=0.9)
parser.add_argument("--threshold_top_k", type=int, default=20)
parser.add_argument("--start_layer", type=int, default=20)
parser.add_argument("--end_layer", type=int, default=29)
parser.add_argument("--seed", type=int, default=42)
args = parser.parse_args()
set_seed(args.seed)
eval_model(args)