forked from openai/weak-to-strong
-
Notifications
You must be signed in to change notification settings - Fork 0
/
incontext.py
60 lines (50 loc) · 2.07 KB
/
incontext.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
import pandas as pd
import random
import re
import argparse
from weak_to_strong.model import TransformerWithHead
from safetensors.torch import load_model
from datasets import load_dataset
from weak_to_strong.train import ModelConfig
from train_simple import MODEL_CONFIGS, MODELS_DICT
from weak_to_strong.common import get_tokenizer
def load_transformer_model(path, model):
load_model(model, path)
def load_custom_dataset(task):
if task == 'equivalence':
path = 'dangnguyen0420/equivalence_relation'
else:
path = 'dangnguyen0420/hierarchical_equivalence'
dataset = load_dataset(path)
train_df = pd.DataFrame(dataset['train'])
test_df = pd.DataFrame(dataset['test'])
return train_df, test_df
def zero_shot(df, model, tokenizer, max_tokens=5):
for ind in df.index:
input = df['input'][ind]
model_inputs = tokenizer([input], return_tensors="pt").to('cuda')
output = model.lm.generate(**model_inputs, max_new_tokens=max_tokens, pad_token_id=tokenizer.eos_token_id)
output = tokenizer.batch_decode(output, skip_special_tokens=True)[0]
output = output[len(input):]
print("Prompt:", input)
print("Response:", output)
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument('--model_name', type=str, choices=MODELS_DICT.keys(), help='Model name')
parser.add_argument('--model_path', type=str, help='Path to model')
parser.add_argument('--task', type=str, choices=['equivalence', 'hierarchical'], help='Task name')
args = parser.parse_args()
return args
def main():
args = parse_args()
model_config = MODELS_DICT[args.model_name]
custom_kwargs = model_config.custom_kwargs or {}
tokenizer = get_tokenizer(args.model_name)
model = TransformerWithHead.from_pretrained(
model_config.name, num_labels=2, linear_probe=False, **custom_kwargs
).to("cuda")
load_transformer_model(args.model_path, model)
_, test_df = load_custom_dataset(args.task)
zero_shot(test_df, model, tokenizer)
if __name__ == "__main__":
main()