Geoffrey Angus

From the visualization standpoint, you can take inspiration from some of our existing visualization functionality: https://github.com/ludwig-ai/ludwig/blob/49b4c79cc89c5369a37d3cb4922893898004ce13/ludwig/visualize.py#L2388 In order to change the threshold in `predict`, you can build off of functionality...

Hi @fire, sorry for the late response here. To your points: 1. NVIDIA Triton does support C++ backends: https://github.com/triton-inference-server/pytorch_backend 2. I do believe that it needs a CUDA runtime, but...

Primarily reviewed the TorchScript-relevant parts of the PR– it's starting to look good. A couple of comments then I think those parts look good to me.

Superseded by https://github.com/ludwig-ai/ludwig/pull/2733.

Hi @aarnphm, can you clarify this point? > you can provide the adapter_name via the swagger UI, under the same level as prompt key. I've deployed an OPT model using...

Got it– I'm also trying now by using the vanilla `openllm` CLI tool. Here's how I deployed the model: ``` openllm start falcon --model-id tiiuae/falcon-7b-instruct ``` Here's what I submitted:...

Hi @philippe-solodov-wd– can you describe your use case in more detail? Are you trying to train on a GPU-enabled machine, without using the GPU?

Hi @vijayi1– taking a look!

Hello @chayanray– Yes, it looks like you may have to download the `llm` subpackage for Ludwig. I see that you are on `0.8.dev`. If you therefore installed by git cloning...

Hi @GlacierPurpleBison, taking a look. Do you know if this bug is a SageMaker-specific bug, or if this occurs when initializing a vanilla LoRAX docker container as well?