Rickey Bowers Jr.

Assuming you have CUDA 11.6+ installed, and torch has been replaced by CPU only version. Activate python environment for this repo and install correct version of torch: ```bash .\venv\scripts\activate pip...

Use ```--iters```, the default is 10000 -- 100 batches of 100. It can continue if you stop and want to resume.

The model is trained to represent a scene - that training is a computationally intensive process. It will stop (default) when 100 batches have completed. It trains because the dependent...

I've gotten to installing the extensions. They don't want to build with MS Tools, and I haven't quite discovered why. Unfortunately, the default installation does not install the CUDA version...

I'm thinking you are using an old version of pip; because if mine errors it will show a list of allowed versions. What does ```pip list``` show? I can browse...

Obviously, your local setup is garbage. How about we ignore that and create a new virtual environment for this repo? ```bash rmdir /S venv python -m venv venv .\venv\scripts\activate ```...

Well, you don't need to use venv if you're using conda, but I never could build tiny-cuda-nn with conda. I abandoned that approach. You could try manually specifying with ```set...

pip only installs into the active environment: venv or conda or local install -- they are all insolated by design.

if you were using venv, I'd say to add it to ```venv\Scripts\activate.bat```, but I don't know for conda.

The two errors I'm stuck on are: ``` D:\_MP\stable-dreamfusion\venv\lib\site-packages\torch\include\pybind11\cast.h(1429): error: too few arguments for template template parameter "Tuple" detected during instantiation of class "pybind11::detail::tuple_caster [with Tuple=std::pair, Ts=]" D:\_MP\stable-dreamfusion\venv\lib\site-packages\torch\include\pybind11\cast.h(1503): error: too...