shawntan/scattermoe Issues
Can't use torch.compile
UpdatedQuestion: Multi-node training
Updated 3No module named 'torch'
Updated 4ParallelLinear with bias
Closed 2Megablocks example
UpdatedExperts with different capacity
Closed 4Accuracy Issues
Closed 11Segfault CUDA 12.2
Updated 1pytest fail
Closed 6Mixtral inference example
Closed 5Tensor Parallelism
Updated 3