Story

arxiv_cs_lg ยท Nov 17, 2026 ยท paper

Source brief

Communicate the Change: Efficient Expert Parallelism via Activation Compression

arxiv.orgNov 17, 2026
original source linked

In brief

Scaling has driven the success of large language models (LLM), with sparse mixture-of-experts (MoE) architectures enabling greater parameter capacity without a proportional increase in computation. At large scales, Mo...

Continues in

Efficient Learning โ€” open the evidence trace โ†’

Continue reading

Read the original at arxiv.org โ†’Open in live feedRead that dayโ€™s brief

Earlier in this thread 4 items