Add Nerf Studio Example

[nbardy]

This adds a nerf studio example to modal.

Setting this up for myself and the build setup got a little complicated.

Type of Change

• New example
• Example updates (Bug fixes, new features, etc.)
• Other (changes to the codebase, but not to examples)

Checklist

• Example is testable in synthetic monitoring system, or lambda-test: false is added to example frontmatter (---)
  • Example is tested by executing with modal run or an alternative cmd is provided in the example frontmatter (e.g. cmd: ["modal", "deploy"])
  • Example is tested by running with no arguments or the args are provided in the example frontmatter (e.g. args: ["--prompt", "Formula for room temperature superconductor:"]
• Example is documented with comments throughout, in a Literate Programming style.
• Example does not require third-party dependencies to be installed locally
• Example pins its dependencies
  • Example pins container images to a stable tag, not a dynamic tag like latest
  • Example specifies a python_version for the base image, if it is used
  • Example pins all dependencies to at least minor version, ~=x.y.z or ==x.y
  • Example dependencies with version < 1 are pinned to patch version, ==0.y.z

Outside contributors

You're great! Thanks for your contribution.

[charlesfrye]

I'm kinda surprised this works? The host driver is 12.2.

Is a later step maybe installing the CUDA toolkit at <=12.2?

[charlesfrye]

Would love to get a NeRF example in!

The install does seem kinda messy though.

Looking at their docs, I see

pip install torch==2.1.2+cu118 torchvision==0.16.2+cu118 --extra-index-url https://download.pytorch.org/whl/cu118

Torch now bundles the CUDA toolkit, including nvcc, so the above should work to get you everything. Maybe it's just a matter of setting CUDA_HOME and symlinking?

I also don't think we to install the editable version of the library, since folks won't be editing the library on Modal.

So I think we can shrink this down to a few pip_installs -- one for main deps (torch, transformers, ninja), one for tiny-cuda-nn, one for nerfstudio, and maybe one more for kplanes-nerfstudio.

HMU in the Modal Slack if you want to pair on it for a bit.

[nbardy]

dromni/nerfstudio

I ended up finding a docker image. I think this is probably the better way to do it.

I also don't think we to install the editable version of the library, since folks won't be editing the library on Modal.

I do want to fork the repo and develop on it from this setup. Probably should move that out to an optional flag to use a local version.

@charlesfrye
Hoping to pick this up in the evenings after work this week. But not so much time this week.