mirror of
https://github.com/Tencent/WeKnora.git
synced 2026-06-04 13:30:32 +08:00
1. Added CUDA debugging support 2. Improved memory management during model inference. 3. Fixed logits output Test Device: Windows WSL Ubuntu 20.04 (python3.10 CUDA 5090 32G)
4.1 KiB
4.1 KiB