Similar Items: ZipCCL: Efficient Lossless Data Compression of Communication Collectives for Accelerating LLM Training
- SplitZip: Ultra Fast Lossless KV Compression for Disaggregated LLM Serving
- CCL-Bench 1.0: A Trace-Based Benchmark for LLM Infrastructure
- SparseRL-Sync: Lossless Weight Synchronization with ~100x Less Communication
- CCL-D: A High-Precision Diagnostic System for Slow and Hang Anomalies in Large-Scale Model Training
- RcLLM: Accelerating Generative Recommendation via Beyond-Prefix KV Caching
- ResiHP: Taming LLM Training Failures with Dynamic Hybrid