1
Concordia: JIT-Compiled Persistent-Kernel Checkpointing for Fault-Tolerant LLM Inference
专为LLM推理设计的JIT编译持久内核检查点方案,以低开销实现故障恢复,保障长时任务稳定执行。
arXiv:2606.23521v1 Announce Type: cross Abstract: Long-running LLM agents keep valuable state resident on GPUs: KV caches, request schedulers, communi…