Wenlei Xie
Wenlei Xie
`astype` is now renamed as `cast` in https://github.com/facebookresearch/torcharrow/pull/43
In PyTorch, there is also `Tensor.type_as` : https://pytorch.org/docs/stable/generated/torch.Tensor.type_as.html#torch.Tensor.type_as
Yeah, today it's adhoc supported based on arity: https://github.com/facebookresearch/torcharrow/pull/296 I think we need a variadic version for generic UDF call. In your case, it has 12 parameters?
The conversion between Arrow and Velox is also zero-copy. Velox is also collaborating with Arrow Community and Voltron Data, as announced in https://www.globenewswire.com/news-release/2022/06/23/2468271/0/en/Voltron-Data-Announces-Commitment-to-Improve-Interoperability-Between-Apache-Arrow-and-the-Velox-Open-Source-Project-Created-by-Meta.html Quoted > ... this collaboration will enable...
cc @RenfeiChen-FB
That makes sense. Thanks for the suggestion. The reason why I listed all the libraries explicitly is because when I setup my Mac, some of the package installments failed due...
@wangkuiyi Sounds great! Shall we still make `scripts/deps_mac.bash` callable and can install individual packages? So in general user will just call `scripts/deps.bash`, but if something goes wrong , they can...
@erip : This is due to previously Velox/TrochArrow is built with `arch=native` that causes GCC very aggressively use AVX512. We have fixed it in https://github.com/facebookresearch/torcharrow/pull/315 (in fact the recent portability...
Yeah. TorchArrow's expected arithmetic behavior is PyTorch, both type promotion and semantic (e.g. underflow/overflow/division by 0 behavior).