Hugging Face Trending Papers

Edge-aware Decoding for Neural Asymmetric Routing

Read the original on Hugging Face Trending Papers →

Neural asymmetric routing models increasingly encode directionality through matrix representations and asymmetry-aware attention. The final routing action, however, is not a node in isolation but a directed transition chosen under the current partial route.

Summary generated by The Flow from the publisher's feed. The full article lives at Hugging Face Trending Papers.

arXiv AI
Jul 9

TriRoute: Unified Learned Routing for Joint Adaptive Attention, Experts, and KV-Cache Allocation

arXiv:2607. 06601v1 Announce Type: cross Abstract: Conditional computation can decouple language model quality from per-token inference cost, yet leading techniques act on a single axis in isolation: Mixture-of-Experts (MoE) sparsifies the FFN, Mixture-of-Depths (MoD) skips whole transformer blocks, and KV-cache quantization compresses attention memory.

By Andrii Balashov, Olena Ponomarova