AI RESEARCH
ForesightKV: Optimizing KV Cache Eviction for Reasoning Models by Learning Long-Term Contribution
arXiv CS.LG
•
ArXi:2602.03203v2 Announce Type: replace-cross Recently, large language models (LLMs) have shown remarkable reasoning abilities by producing long reasoning traces. However, as the sequence length grows, the key-value (KV) cache expands linearly, incurring significant memory and computation costs. Existing KV cache eviction methods mitigate this issue by discarding less important KV pairs, but often fail to capture complex KV dependencies, resulting in performance degradation. To better balance efficiency and performance, we