Free(): Learning to Forget in Malloc-Only Reasoning Models
IntermediateYilun Zheng, Dongyang Ma et al.Feb 8arXiv
LLMs can think for many steps, but when they keep every step forever, the extra tokens turn into noise and make answers worse, not better.
#Free()LM#self-forgetting#context pruning