3 ms·
Also, similar to Orca-Math but without a teacher model. They also followed an iterative DPO/KTO scheme, but with no length normalized NLL loss term.
by popinman322 2y ago
Also, similar to Orca-Math but without a teacher model. They also followed an iterative DPO/KTO scheme, but with no length normalized NLL loss term.
- algo_trader 2y agoIf we had a magical (fast) oracle for grading responses, have people done search/expert iteration for LLMs? Specifically for codegen, i am playing with an iterative interpreter that can quickly (re)evaluate a tree of similar responses