-
Notifications
You must be signed in to change notification settings - Fork 0
/
API_test2_summary.py
177 lines (153 loc) · 6.96 KB
/
API_test2_summary.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
172
173
174
175
176
177
import os
import json
import base64
from openai import OpenAI
import requests
# # 设置代理环境变量
# os.environ["http_proxy"] = "http://10.105.20.64:7890"
# os.environ["https_proxy"] = "http://10.105.20.64:7890"
def closeAI(model_name, image_path, question):
from openai import OpenAI
api_key = 'sk-Ee0f5EaLhGBg9SjeKmTBGG0J0Y6w2zBVmiK2Sqx2QdY0XFvZ'
base64_image = encode_image(image_path)
if model_name.startswith('gpt'):
client = OpenAI(
base_url='https://api.openai-proxy.org/v1',
api_key=api_key,
)
chat_completion = client.chat.completions.create(
messages=[
{
"role": "user",
"content": [
{
"type": "text",
"text": question
},
{
"type": "image_url",
"image_url": {
"url": f"data:image/jpeg;base64,{base64_image}"
}
}
]
}
],
model=model_name,
)
# print(chat_completion)
generated_content = chat_completion.choices[0].message.content
else:
url = "https://api.openai-proxy.org/anthropic/v1/messages"
headers = {
"x-api-key": api_key,
"anthropic-version": "2023-06-01",
"content-type": "application/json"
}
data = {
"model": model_name,
"max_tokens": 1024,
"messages": [
{
"role": "user",
"content": [
{
"type": "image",
"source": {
"type": "base64",
"media_type": "image/jpeg",
"data": base64_image,
},
},
{
"type": "text",
"text": question
}
],
}
],
}
response = requests.post(url, headers=headers, json=data)
generated_content = response.json()['content'][0]['text']
# print(response.json())
# 提取生成的内容
return generated_content
def encode_image(image_path):
with open(image_path, "rb") as image_file:
return base64.b64encode(image_file.read()).decode('utf-8')
def process_templates(image_path, templates, model_name, caption, ref, title):
results = {}
for key, template in templates.items():
question = template.format(caption=caption, ref=ref, title=title)
results[key] = closeAI(model_name, image_path, question)
return results
def update_json_file(json_file, base_image_path, templates, model_name):
output_data = []
with open(json_file, 'r', encoding='utf-8') as f:
for line in f:
try:
data = json.loads(line.strip())
image_path = os.path.join(base_image_path, data['filename'])
caption = data.get('caption', '')
ref = data.get('ref', '')
title = data.get('title', '')
# 检查 ref 是否为空，如果为空则跳过
if not ref:
print(f"Skipping file due to empty 'ref': {data['filename']}")
continue
generated_contents = process_templates(image_path, templates, model_name, caption, ref, title)
data['generated_contents'] = generated_contents
output_data.append(data)
except json.JSONDecodeError as e:
print(f"Error decoding JSON on line: {line.strip()}")
print(f"Error message: {e}")
except Exception as e:
print(f"Error processing data: {e}")
output_file = json_file.replace('.json', '_updated_multi_reasoning.json')
# output_file = json_file.replace('.json', '_updated_gpt4_single_2.json')
with open(output_file, 'w', encoding='utf-8') as f:
for entry in output_data:
f.write(json.dumps(entry, ensure_ascii=False) + '\n')
# 定义 JSON 文件路径和基础图片路径
# base_image_path = './benchmark_single/single'
json_file = './benchmark_multi/multi_lu.jsonl'
base_image_path = './benchmark_multi/multi'
#Chart_Summarization
# templates = {
# 'Chart_Summarization': """Role: You are an AI visual assistant who can analyze a scientific paper graph.
# Task: You are provided with an image, its caption, its description, and the paper's title. Generate a chart summarization that aims at summarizing the trend-like or high-level characteristics from the plot.
# Requirements: The answer should be within 100 words, should be as related to the image as possible, and should clearly differentiate from a simple chart description by providing a more macro-level summary. If applicable, compare and summarize across multiple images, highlighting overall patterns and differences. The output should be in paragraph format, not a list.
# Caption: {caption}
# Des: {ref}
# Title: {title}
# END_OF_INSTRUCTIONS"""
# }
#background_reasoning
templates = {
'reasoning_questions': """Role: You are an AI visual assistant who can analyze a scientific paper graph.
Task: You are provided with an image, its caption and its description. Generate one Reasoning question based solely on the provided image and information. For each question, provide the following:
Question: The question itself.
Answer: The final, concise answer.
Requirements:
These questions require a certain level of astronomical background and needs to involve some analysis.
Only ask questions that you can clearly answer yourself. Answers should be within 50 words.
Caption: {caption}
Des: {ref}
END_OF_INSTRUCTIONS"""
}
# templates_multi = {
# 'reasoning_questions': """Role: You are an AI visual assistant who can analyze a scientific paper graph.
# Task: You are provided with an image and its caption. Generate four Reasoning questions based solely on the provided image and information. For each question, provide the following:
# Question: The question itself.
# Answer: The final, concise answer.
# Requirements:
# These question require numerical calculations, comparisons, and other information that cannot be directly obtained from the image.
# Tow questions should involve comparisons between multiple images. Only ask questions that you can clearly answer yourself. Answers should be within 20 words.
# Caption: {caption}
# Des: {ref}
# END_OF_INSTRUCTIONS"""
# }
# 更新 JSON 文件
update_json_file(json_file, base_image_path, templates, "claude-3-5-sonnet-20240620")
# update_json_file(json_file, base_image_path, templates, "gpt-4o")
# update_json_file(json_file, base_image_path, templates, "gpt-4o-mini")