Closed wenbingl closed 8 months ago
The sentencepiece based tokenizer token-id is int32. To be consistent, add an option to cast the output token-id.
The sentencepiece based tokenizer token-id is int32. To be consistent, add an option to cast the output token-id.