Xin Yao comments

Results 123 comments of


                                            Xin Yao

[Feature] Gpu cache for node and edge data

@dgl-bot

[Feature] Gpu cache for node and edge data

@dgl-bot

[cugraph-ops] Add RGCN cugraph-op into DGL

Close by https://github.com/dmlc/dgl/pull/4965.

v2ray-plugin exits unexpectedly on macOS Big Sur Beta 10

@bkm-clem It still exits unexpectedly after I built it myself. Could you kindly share the binary file that works normally on macOS 11?

The provided qkv memory layout is not supported! When using RoPE

We're aware of this bug and will push a fix to MCore. For now, you can add the following code in https://github.com/NVIDIA/Megatron-LM/blob/89574689447d694bb19dd86fc8a6153b4467ba9d/megatron/core/transformer/custom_layers/transformer_engine.py#L464 ```python # In PyTorch, the following two tensors...

[Feature Request] Grouped GEMM kernel

A `GroupedLinear` layer has been added in TE v1.9, and it has FP8 support.

Fix typo in README.md

Thanks. I'll include your commit in my internal fix PR.

Fix typo in README.md

Your commit is squash merged in https://github.com/NVIDIA/Megatron-LM/commit/2488e20d488b35259f6eb53267e8a0525a373599 and your author information is preserved. Thanks for your contribution. You can close this PR now.

fix _te_version issue in transformer_engine.py get_cpu_offload_context()

Fixed in https://github.com/NVIDIA/Megatron-LM/commit/98b43c91d004dec254f1610d9cffae8aff8550f3.

[Bugfix] Fix typo in moe doc

Good catch. Fixed in https://github.com/NVIDIA/Megatron-LM/commit/e223b920724c523a098558b2c128b4d6eb7f8ff0.