mirror of
https://github.com/open-compass/opencompass.git
synced 2025-05-30 16:03:24 +08:00
52 lines
1.6 KiB
Python
52 lines
1.6 KiB
Python
from opencompass.openicl.icl_prompt_template import PromptTemplate
|
|
from opencompass.openicl.icl_retriever import FixKRetriever
|
|
from opencompass.openicl.icl_inferencer import PPLInferencer
|
|
from opencompass.openicl.icl_evaluator import AccEvaluator
|
|
from opencompass.datasets import HFDataset
|
|
|
|
|
|
_hint = "The following are semantic matching questions. \n" \
|
|
"Please determine whether the following two sentences are semantically duplicate: " \
|
|
"0 means not duplicate, 1 means duplicate.\n"
|
|
QQP_infer_cfg = dict(
|
|
ice_template=dict(
|
|
type=PromptTemplate,
|
|
template="Sentence one: {question1}\nSentence two: {question2}\nResult: {label}",
|
|
),
|
|
prompt_template=dict(
|
|
type=PromptTemplate,
|
|
template={
|
|
answer:
|
|
f"{_hint}</E>Sentence one: {{question1}}\nSentence two: {{question2}}\nResult: {answer}"
|
|
for answer in [0, 1]
|
|
},
|
|
ice_token='</E>',
|
|
),
|
|
retriever=dict(type=FixKRetriever, fix_id_list=[0, 1, 2, 3, 4]),
|
|
inferencer=dict(type=PPLInferencer))
|
|
|
|
QQP_eval_cfg = dict(evaluator=dict(type=AccEvaluator), )
|
|
|
|
|
|
QQP_datasets = []
|
|
for _split in ["validation", "test"]:
|
|
|
|
QQP_reader_cfg = dict(
|
|
input_columns=['question1', 'question2'],
|
|
output_column='label',
|
|
train_split="train",
|
|
test_split=_split
|
|
)
|
|
|
|
QQP_datasets.append(
|
|
dict(
|
|
abbr=f'QQP-{_split}',
|
|
type=HFDataset,
|
|
path='glue',
|
|
name='qqp',
|
|
reader_cfg=QQP_reader_cfg,
|
|
infer_cfg=QQP_infer_cfg,
|
|
eval_cfg=QQP_eval_cfg
|
|
)
|
|
)
|