Kv Cache News
The latest Kv Cache coverage curated for game developers.
Quantized Keys Steal Attention: Bias Correction for KV-Cache Compression in Video Diffusion
A new paper says INT2 KV-cache quantization in chunked video diffusion can quietly bias attention, making cached keys “steal” mass from the current chunk. The authors correct that Jensen bias …
Research
arXiv cs.GR
·
3 months, 2 weeks ago