Skip to content

check CUDA_DEVICE_MAX_CONNECTIONS and TORCH_NCCL_AVOID_RECORD_STREAMS #1242

check CUDA_DEVICE_MAX_CONNECTIONS and TORCH_NCCL_AVOID_RECORD_STREAMS

check CUDA_DEVICE_MAX_CONNECTIONS and TORCH_NCCL_AVOID_RECORD_STREAMS #1242

Annotations

2 warnings

training_16GPU_4DP2TP2PP_MSP (t_cluster)

succeeded Dec 18, 2024 in 1m 36s