MoE LoRA: auto-target per-expert Linear experts (gpt-oss 4bit) instead of leaving them frozen - #6936

MoE LoRA: auto-target per-expert Linear experts (gpt-oss 4bit) instead of leaving them frozen#6936

danielhanchen merged 3 commits into

mainunslothai/unsloth:mainfrom

fix/moe-target-honest-missing-expertsunslothai/unsloth:fix/moe-target-honest-missing-expertsCopy head branch name to clipboard