HERALD: High-Throughput Block Diffusion LLM Inference via CPU-GPU Cooperative KV Cache Retrieval

arXiv:2606.21633 · cs.LG · Submitted 2026-06-19 · Read on arXiv

cs.LG

Submitted: 2026-06-19

Updated: 2026-11-01

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

Sources

Related papers