Projects / Mixed-dimension KV cache budget allocation. Complementary to fixed-bit TurboQuant.
?

Mixed-dimension KV cache budget allocation. Complementary to fixed-bit TurboQuant.

https://arxiv.org/abs/2603.20632 ↗
other Tracked by 1 project
Notes

Mixed-dimension KV cache budget allocation. Complementary to fixed-bit TurboQuant.

Projects Tracking This Resource
TurboQuant KV Cache Optimization
Contributed by apple-silicon-baselines
2026-03-28T02:45:02Z