ghostplant

@Lurnco We merged a PR that work for Windows environment. Can you let us know whether it works for you?

Have you try using `python3.exe -m pip install --user --upgrade git+https://github.com/microsoft/tutel@main`? It should work well on Windows. If there is ERROR, can you share the detail of log file?

@Lurnco CUDA environment is not needed, but it will only install CPU version. The error you show in the first case is not related to CUDA. It is likely your...

> I also haven't CUDA environment set a CUDA_HOME:torch/utils/cpp_extention.py _find_cuda_home() cuda_home="C:" error: RuntimeError: Error compiling objects for extension If you don't have CUDA environment, can you install Tutel successfully and...

Seems like it is an issue of Visual Studio for conda, you may refer this: https://github.com/pytorch/pytorch/issues/30379

It is expected to set an upper bound of sample size when creating `tutel.moe_layer(..)`. If not set by default, the upper bound will be automatically set to be the batch...

In v0.1.5, `expected_sample_size` will be automatic decided by MoE Layer, so that you no longer need to manually set `self._moe_layer.expected_sample_size = 1600`.

Do you mean something like this? ```py self._layer = tutel.moe.moe_layer(gate1_type={..}, gate2_type={..}, ..) output1 = self._layer(data, use_gate1) output2 = self._layer(data, use_gate2) ```

We are going to merge this: https://github.com/microsoft/tutel/pull/71/files You can create new moe layers by specifying a list of original gating types. And when forwarding the moe layer, you can use...

It's done. Feel free to share any feedbacks. Thanks!