[Feature Request] Support Epsilon Sampling
radinplaid opened this issue · comments
Epsilon sampling is a compelling alternative/complement to top_p and top_k sampling and would make a good addition to CTranslate2: https://arxiv.org/abs/2305.09860
Fast inference engine for Transformer models
radinplaid opened this issue · comments
Epsilon sampling is a compelling alternative/complement to top_p and top_k sampling and would make a good addition to CTranslate2: https://arxiv.org/abs/2305.09860