Workaround for ChatRTX on Linux

[Hyphaed]

Hi NVidia staff

congrats for the great work you are doing

Since Tensorflow, TensorRT-LLM, LlamaIndex and FAISS are libraries that can be used without problems under Linux

is there any workaround to have ChatRTX on Linux?

even if is trough terminal, without UI

Would be a great if you guide us on how to use ChatRTX on Linux Terminal

[oscarbg]

+1

[SvenMeyer]

NVIDIA, a 2.83 trillion USD company, big "product placement" on their website* for ChatRTX "Download here !!" ... and then ONLY WINDOZE ??? ... How sad is that.

• https://www.nvidia.com/en-us/ai-on-rtx/chatrtx/?ncid=so-infl-738981

[NotYuSheng]

Found this:
https://chat-withrtx.com/linux/

Seems like the solution is to either run from a vm or to use Wine

[NotYuSheng]

Found this: https://chat-withrtx.com/linux/

Seems like the solution is to either run from a vm or to use Wine

It seems like this website is not an official NVIDIA site. I've tried running the official Windows version with wine but have encountered issues.

0050:err:winediag:is_broken_driver Broken NVIDIA RandR detected, falling back to RandR 1.0. Please consider using the Nouveau driver instead.
00d4:err:winediag:is_broken_driver Broken NVIDIA RandR detected, falling back to RandR 1.0. Please consider using the Nouveau driver instead.
0034:err:winediag:is_broken_driver Broken NVIDIA RandR detected, falling back to RandR 1.0. Please consider using the Nouveau driver instead.
0024:err:winediag:is_broken_driver Broken NVIDIA RandR detected, falling back to RandR 1.0. Please consider using the Nouveau driver instead.

Specs

• OS: Ubuntu 24.04
• GPU: NVIDIA RTX A6000

[Hyphaed]

for Linux we can use Alpaca (meanwhile waiting for an official NVidia solution)

https://github.com/Jeffser/Alpaca

you will have different models to choose like mistral nemo (trained by Nvidia) or Llama Chatqa (also trained by Nvidia)

I recommend you to use ollama local instead of the ollama bundled with Alpaca

1 :: you install ollama
https://ollama.com/download

2 :: you open a terminal and type
ollama serve

3 :: while ollama is open in a terminal (and without closing the terminal)

Open Alpaca / Preferences / on URL of Remote Instance set http://127.0.0.1:11434

for me works better that way

you can always use the bundled ollama in alpaca instead of running the local ollama instance