metadata
license: bsd-3-clause
BanBTP2.0-v10: A 7,112,960 (~7.1M) Million parameter model designed to run on any hardware. 100 times smaller than BanBTPV9!!! It uses a custom new banbtp arch instead of gpt2.
license: bsd-3-clause
BanBTP2.0-v10: A 7,112,960 (~7.1M) Million parameter model designed to run on any hardware. 100 times smaller than BanBTPV9!!! It uses a custom new banbtp arch instead of gpt2.