NVIDIA’s new AI chatbot runs locally on your PC

NVIDIA just released a free demo version of a chatbot that runs locally on your PC. This is pretty neat, as it gives the chatbot access to your files and documents. You can feed Chat with RTX a selection of personal data and have it create summaries based on that information. You can also ask it questions, just like any chatbot, and it’ll dive into your data for answers.

The company says it allows users to “quickly, easily connect local files on a PC as a dataset to an open-source large language model like Mistral or Llama 2.” NVIDIA gives an example of a user asking the chatbot about a restaurant their partner recommended while in Las Vegas. The software scans local files to find the answer. It supports a whole bunch of file formats, including .txt, .pdf, .doc/.docx and .xml. The company says it’ll load relevant files into its dataset “in seconds.”

Chat with RTX also integrates YouTube videos and playlists. You can add a video URL into the dataset and it’ll integrate the knowledge contained in the clip for contextual queries. NVIDIA says this will be useful when asking for travel recommendations “based on content from favorite influencer videos” or when looking for tutorials and summaries derived from educational resources.

The Verge had some hands-on time with the chatbot and came away impressed, writing that they could see it as “a valuable part of data research for journalists or anyone who needs to analyze a collection of documents.”

This sounds like a big step toward something resembling an actual digital assistant that works within the contextual framework of your personal data. With most chatbots, the data is sent off to the cloud, but Chat with RTX “lets users process sensitive data on a local PC without the need to share it with a third party or have an internet connection.” So it’s safer and more contextually aware.

There are some limitations. This is a demo product so you should expect plenty of bugs, though NVIDIA should start squashing them once users begin issuing error reports and the like. There are also some strict hardware limitations here. Chat with RTX only works on Windows PCs with NVIDIA GeForce RTX 30 Series GPUs or higher and at least 8GB of VRAM.

NVIDIA has really been showing off its AI prowess lately, as the company just launched its next-generation of artificial intelligence supercomputer chips. It’s absolutely swimming in profits, due primarily to the company’s AI and data center segments.

This article originally appeared on Engadget at https://www.engadget.com/nvidias-new-ai-chatbot-runs-locally-on-your-pc-163406121.html?src=rss https://www.engadget.com/nvidias-new-ai-chatbot-runs-locally-on-your-pc-163406121.html?src=rss
Vytvorené 1y | 13. 2. 2024, 17:30:34


Ak chcete pridať komentár, prihláste sa

Ostatné príspevky v tejto skupine

Playdate Season 2 review: Taria & Como and Black Hole Havoc

We've officially made it to the end of Playdate Season Two, and what a season it's been. Despite having half the number of titles as Season One, this latest round of weekly game releases has made a

6. 7. 2025, 21:50:18 | Engadget
The Stop Killing Games initiative has hit a major milestone, but the fight's just begun

A petition to preserve video game access recently achieved an

6. 7. 2025, 19:40:16 | Engadget
Prime Day deals include $100 off the Apple Watch Series 10

Prime Day 2025 will be here in a few days, but you can already shop great tech deals on Amazon before the July 8-11

6. 7. 2025, 17:20:15 | Engadget
Elon Musk's proposed America Party is already attracting the attention of the ultra-rich

Just a day after former White House advisor Elon Musk claimed on X that he's creating a

6. 7. 2025, 17:20:14 | Engadget
The Amazon Smart Plug falls to only $13 for Prime Day

If you're looking for an ultra-affordable away to make a part of your home a little smarter, look no further than the

6. 7. 2025, 15:10:08 | Engadget
The latest Kindle Scribe drops to a record-low price for Prime Day

The latest Amazon Kindle Scribe ereader

6. 7. 2025, 15:10:07 | Engadget