Per-layer mixture-of-experts routing selects heterogeneous eviction-quantization tuples for KV cache compression, matching uncompressed accuracy at 14x reduction on LongBench subsets where uniform baselines degrade.
GQA: Training generalized multi-query transformer models from multi-head check- points
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MoE-nD: Per-Layer Mixture-of-Experts Routing for Multi-Axis KV Cache Compression
Per-layer mixture-of-experts routing selects heterogeneous eviction-quantization tuples for KV cache compression, matching uncompressed accuracy at 14x reduction on LongBench subsets where uniform baselines degrade.