import sys
import argparse
from dpr_qa_test import DPRTest
def main(args):
parser = argparse.ArgumentParser(description='To get arguments')
parser.add_argument('--max_seq_lens',
nargs='*',
type=int,
default=[256],
help='Max sequence length of one input text for the model',
required=False)
parser.add_argument('--max_seq_len_passages',
nargs='*',
type=int,
default=[256],
help='Longest length of each passage/context sequence. Maximum number of tokens for the passage text. Longer ones will be cut down.',
required=False)
parser.add_argument('--max_seq_len_queries',
nargs='*',
type=int,
default=[64],
help='Longest length of each query sequence. Maximum number of tokens for the query text. Longer ones will be cut down.',
required=False)
parser.add_argument('--embed_titles',
nargs='*',
type=bool,
default=[True, False],
help='Whether to concatenate title and passage to a text pair that is then used to create the embedding.',
required=False)
parser.add_argument('--context_window_sizes',
nargs='*',
type=int,
default=[150, 175],
help='The size, in characters, of the window around the answer span that is used when displaying the context around the answer.',
required=False)
parser.add_argument('--doc_strides',
nargs='*',
type=int,
default=[100, 128],
help='Length of striding window for splitting long texts (used if len(text) > max_seq_len)',
required=False)
parser.add_argument('--retriever_top_ks',
nargs='*',
type=int,
default=[3, 5, 7],
help='How many documents to return per query.',
required=False)
parser.add_argument('--reader_top_ks',
nargs='*',
type=int,
default=[3, 5, 7],
help='The maximum number of answers to return',
required=False)
parser.add_argument('--reader_models',
nargs='*',
type=str,
default=['ktrapeznikov/albert-xlarge-v2-squad-v2',
'deepset/roberta-base-squad2',
'deepset/minilm-uncased-squad2',
'ahotrod/albert_xxlargev1_squad2_512'],
help='The maximum number of answers to return',
required=False)
parser.add_argument('--text_datasets',
nargs='*',
type=str,
default=['titleText-threeSentences.csv',
'titleText-paragraphs.csv'],
help='The maximum number of answers to return',
required=False)
parser.add_argument('--qa_datasets',
nargs='*',
type=str,
default=['qa-SQUAD.json'],
help='The maximum number of answers to return',
required=False)
parser.add_argument('--report_out_dir',
type=str,
default='dpr-qa-report.csv',
help='The maximum number of answers to return',
required=False)
parser.add_argument('--sample_out_dir',
type=str,
default='dpr-qa-sample.json',
help='The maximum number of answers to return',
required=False)
args = parser.parse_args()
report_generator = DPRTest.report_generator(max_seq_lens=args.max_seq_lens,
max_seq_len_passages=args.max_seq_len_passages,
max_seq_len_queries=args.max_seq_len_queries,
embed_titles=args.embed_titles,
context_window_sizes=args.context_window_sizes,
doc_strides=args.doc_strides,
retriever_top_ks=args.retriever_top_ks,
reader_top_ks=args.reader_top_ks,
reader_models=args.reader_models,
text_datasets=args.text_datasets,
qa_datasets=args.qa_datasets,
report_out_dir=args.qa_datasets,
sample_out_dir=args.sample_out_dir,
)
_, _ = report_generator.get_report()
if __name__ == "__main__":
main(sys.argv[1:])
Datasets
Models
