From the 1 of 2 linked papers with an AI index.
2 papers
cs.NI2026
Incast-Free MoE Rate-Based Scheduling
Evyatar Cohen, Jose Yallouz, Alexander Shpiner +3
The paper shows that round-robin scheduling in Mixture of Experts (MoE) models creates an exponential incast problem, and introduces a proactive fair rate‑based scheduling framewor…
cs.NI2026
Avoiding Cross-Datacenter Collective Congestion via Disaggregated Buffering
Mariano Scazzariello, Noga H. Rotman, Dima Gavrilenko +6
LLM training at the scale of tens of thousands of GPUs now spans multiple datacenters (DC), making cross-DC collectives over long-haul links unavoidable. A critical and overlooked…