Currently the code only support GPT-2 but this limits performance. I'll add support for LLama-7b as well to improve our absolute performance.
Currently the code only support GPT-2 but this limits performance. I'll add support for LLama-7b as well to improve our absolute performance.