HF RL Explorer

Align KTO with DPO: Simplify max length init logic

Align KTO with DPO: Simplify max length init logic: a task in trl-r2e-v0-1 (Harbor dataset). Title: Align KTO with DPO: Simplify max length init logic

The task

**Title:** Align KTO with DPO: Simplify max_length init logic

Part of AdithyaSK/trl-r2e-v0-1.