Which format to use for triple file when train?

[tonylmai]

The instruction for training ColBERT says that it requires a JSONL triples file with a [qid, pid+, pid-] list per line, for example triples="/path/to/MSMARCO/triples.train.small.tsv". But triples.train.small.tsv contains question, answer and passage.

Which format should I use?

[hhhhzs666]

What is the training data format? The introduction is not clear

[okhat]

Both formats work. I suggest sticking to a jsonl format