jmgmcm comments

Repositories
Issues
Comments

Results 2 comments of


                                            jmgmcm

Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

Same here. My last training around one week ago was fine with similar parameters. Now the CPU memory (64 GB) is not enough with 8 bit during Transformer 2. I...

Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

Current version (26f4f02) works fine again (64 GB RAM RTX5090). Thank you to the developers.