ElasWave: An Elastic-Native System for Scalable Hybrid-Parallel Training

Documenti analoghi

• Fault-Tolerant Hybrid-Parallel Training at Scale with Reliable and Efficient In-memory Checkpointing
  di: Wang, Yuxin, et al.
  Pubblicazione: (2023)
• On the Spectral Flattening of Quantized Embeddings
  di: Huang, Junlin, et al.
  Pubblicazione: (2026)
• BurstGPT: A Real-world Workload Dataset to Optimize LLM Serving Systems
  di: Wang, Yuxin, et al.
  Pubblicazione: (2024)
• Efficient MoE Inference with Fine-Grained Scheduling of Disaggregated Expert Parallelism
  di: Pan, Xinglin, et al.
  Pubblicazione: (2025)
• ElasTST: Towards Robust Varied-Horizon Forecasting with Elastic Time-Series Transformer
  di: Zhang, Jiawen, et al.
  Pubblicazione: (2024)