zhenyih

Results 4 comments of zhenyih

UsUsing MegatronCommOverlapCallback(tp_comm_overlap=True) causes segfault.

ai generated, please verify The segmentation fault occurs because of a conflict in MPI initialization when using MegatronCommOverlapCallback with tp_comm_overlap=True. The error happens during tensor parallel communication setup when the...

Support Canary-Qwen-2.5b inference in vLLM and TensorRT

ai generated, please verify Based on my analysis of the NeMo codebase, there is currently no explicit support for Canary-Qwen-2.5b model inference in either vLLM or TensorRT. While NeMo supports...

compute_rnnt_timestamps fails with empty char offsets

AI-generated solution, please verify The error you're seeing occurs because the parakeet-tdt-0.6b-v2 model has difficulty generating character offsets for very short audio files (less than 1 second). When processing your...

NCCL TImeout

ai generated, please verify To resolve NCCL timeout errors occurring after extended training runs, try these solutions: 1. Increase NCCL timeout values by setting these environment variables before launching your...