Santosh Bhavani

I'm also having the same problem. The MS COCO format was not working so I converted to CSV and now I'm getting the same error.

You're right it's a bit outdated. We can switch to latest, I was just copying from a previous docker file -- CEO & Founder @ SemanticMD +1 281 816 5734...

I'm trying to combine this with a docker compose file that also runs couchdb but running into some issues. Let me know if you were able to successfully launch both...

`services: couchdb: image: couchdb ports: - "5984:5984" dicomweb: build: . depends_on: - couchdb ports: - "5985:5985" command: npm start` This was my simple docker compose file but the issue I...

[JAX Toolbox](https://github.com/NVIDIA/JAX-Toolbox) includes the nightly container referenced above along with open [Dockerfiles](https://github.com/NVIDIA/JAX-Toolbox/tree/main/.github/container).

I'd recommend using a JAX base container from [JAX Toolbox](https://github.com/NVIDIA/JAX-Toolbox/pkgs/container/jax) which is validated with a nightly CI on NVIDIA GPUs. The [Dockerfiles](https://github.com/NVIDIA/JAX-Toolbox/tree/main/.github/container) are open to modify as well.

I'd recommend using a JAX base container from [JAX Toolbox](https://github.com/NVIDIA/JAX-Toolbox/pkgs/container/jax) which is validated with a nightly CI on NVIDIA GPUs. The [Dockerfiles](https://github.com/NVIDIA/JAX-Toolbox/tree/main/.github/container) are open to modify as well.

You can also pull this container to run grok: `ghcr.io/nvidia/jax:grok` from [JAX Toolbox](https://github.com/NVIDIA/JAX-Toolbox)

> It should work even faster with the 40 series as they support FP8, so no quantization needed. Just cast to FP8 - when PyTorch adds support for it, of...

[TransformerEngine v0.7 added FP8 support on Ada](https://docs.nvidia.com/deeplearning/transformer-engine/release-notes/index.html)