-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathbenchmark_test.py
166 lines (137 loc) · 7.07 KB
/
benchmark_test.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
import os
from tqdm import tqdm
from rich.console import Console
from ..LLM.LLM_state import get_api_client
from ..utils.file import import_file
from ..Vic.Vic_main import Vic, Vic_async
from ..utils.file import to_base64_image
console = Console()
class Benchmarktest():
def __init__(self,**kwargs) -> None:
self.benchmark = kwargs.get("benchmark")
self.indicator = kwargs.get("indicator")
self.model = get_api_client()
self.dirname = self.benchmark.dirname
self.model_name = self.model.llm.model_name
self.benchmark_name = self.benchmark.benchmark_name
self.exchange_info_df = kwargs.get("exchange_info_df",None)
@property
def result_name(self):
return os.path.join(self.dirname, f"{self.model_name}_{self.benchmark_name}_{self.indicator}_result.tsv")
def factory(self):
match self.indicator:
case "vic":
self.vic()
case "vic_ready":
self.vic_ready()
case "vic_m":
self.vic_m()
case "switch_info":
self.info_concate()
self.switch_info()
case "original":
self.original()
case 'cot':
self.cot()
case _:
raise ValueError("Indicator not supported")
def checking_exiting_file(self):
if os.path.exists(self.result_name):
df = import_file(self.result_name)
console.print(f"Loaded {len(df)} results from {self.result_name}",style="bold green")
return len(df)
else:
console.print(f"Created {self.result_name}",style="bold green")
return 0
def original(self):
existing_len = self.checking_exiting_file()
benchmark = self.benchmark[existing_len:]
for row in tqdm(benchmark.iterrows(),total=len(benchmark)):
row_info = row[0]
row_df = row.left
prompt = row_info['prompt']
image = [to_base64_image(row_info['image'])]
input_data = {'query':prompt,'image':image}
response = self.model.request(input_data)
row_df.loc[:,f'{self.model_name}_response'] = response
self.store(row_df)
def cot(self):
existing_len = self.checking_exiting_file()
benchmark = self.benchmark[existing_len:]
for row in tqdm(benchmark.iterrows(),total=len(benchmark)):
row_info = row[0]
row_df = row.left
cot_prompt = 'Answer the following question step by step based on image.' #You should also focus on the information from image. Explain each step of your reasoning process clearly and thoroughly before arriving at the final answer. Even if the problem seems simple, break it down into small, logical steps to ensure clarity and correctness.'
prompt = cot_prompt+row_info['prompt']
image = [to_base64_image(row_info['image'])]
input_data = {'query':prompt,'image':image}
response = self.model.request(input_data)
row_df.loc[:,f'{self.model_name}_response'] = response
self.store(row_df)
def vic(self):
existing_len = self.checking_exiting_file()
benchmark = self.benchmark[existing_len:]
for row in tqdm(benchmark.iterrows(),total=len(benchmark)):
row_info = row[0]
row_df = row.left
prompt = row_info['prompt']
image = [to_base64_image(row_info['image'])]
response = Vic(query=prompt,image=image)
row_df.loc[:,f'visual_chain'] = str(response['visual_inference_chain'])
row_df.loc[:,f'{self.model_name}_intermediate_response'] = response['intermediate_response']
row_df.loc[:,f'{self.model_name}_response'] = response['response']
self.store(row_df)
def vic_ready(self):
existing_len = self.checking_exiting_file()
benchmark = self.benchmark[existing_len:]
for row in tqdm(benchmark.iterrows(),total=len(benchmark)):
row_info = row[0]
row_df = row.left
prompt = row_info['prompt']
image = [to_base64_image(row_info['image'])]
visual_chain = row.visual_chain.iloc[0]
response = Vic(query=prompt,image=image,vic=visual_chain)
row_df.loc[:,f'visual_chain'] = str(response['visual_inference_chain'])
row_df.loc[:,f'{self.model_name}_intermediate_response'] = response['intermediate_response']
row_df.loc[:,f'{self.model_name}_response'] = response['response']
self.store(row_df)
def vic_m(self):
existing_len = self.checking_exiting_file()
benchmark = self.benchmark[existing_len:]
for row in tqdm(benchmark.iterrows(),total=len(benchmark)):
row_info = row[0]
row_df = row.left
prompt = row_info['prompt']
image = [to_base64_image(row_info['image'])]
visual_chain = row.visual_chain.iloc[0]
response = Vic(query=prompt,image=image,vic=visual_chain,vic_m=True)
row_df.loc[:,f'visual_chain'] = str(response['visual_inference_chain'])
row_df.loc[:,f'{self.model_name}_intermediate_response'] = response['intermediate_response']
row_df.loc[:,f'{self.model_name}_response'] = response['response']
self.store(row_df)
def info_concate(self):
if self.exchange_info_df is None:
raise ValueError("exchange_info_df is required")
exchange_info_column = [col for col in self.exchange_info_df.columns if 'intermediate_response' in col]
self.benchmark.df['exchange_info'] = self.exchange_info_df[exchange_info_column].values.tolist()
def switch_info(self):
existing_len = self.checking_exiting_file()
benchmark = self.benchmark[existing_len:]
for row in tqdm(benchmark.iterrows(),total=len(benchmark)):
row_info = row[0]
row_df = row.left
prompt = row_info['prompt']
image = [to_base64_image(row_info['image'])]
visual_chain = row.visual_chain.iloc[0]
exchange_info = row.df.exchange_info.iloc[0]
response = Vic(query=prompt,image=image,vic=visual_chain,extract_info=exchange_info)
row_df.loc[:,f'visual_chain'] = str(response['visual_inference_chain'])
row_df.loc[:,f'{self.model_name}_intermediate_response'] = response['intermediate_response']
row_df.loc[:,f'{self.model_name}_response'] = response['response']
self.store(row_df)
def store(self,row_df):
if not os.path.exists(self.result_name):
row_df.to_csv(self.result_name,sep='\t',index=False)
else:
row_df.to_csv(self.result_name,sep='\t',mode='a',index=False,header=False)
console.print(f"Stored 1 result in {self.result_name}",style="bold green")