nolangclement's profile picture. Stripe, MIT CS

nolan

@nolangclement

Stripe, MIT CS

nolan reposted

Jamba model launch takeaways - Potentially a new leader for ultra-long prompt use-cases (RAG) ‣ First open-source model of this size to combine MAMBA state-space model architecture, Mixture-Of-Experts (MOE) and the transformer ‣ 256k context window, more than 2X the size of…

ArtificialAnlys's tweet image. Jamba model launch takeaways - Potentially a new leader for ultra-long prompt use-cases (RAG)

‣ First open-source model of this size to combine MAMBA state-space model architecture, Mixture-Of-Experts (MOE) and the transformer
‣ 256k context window, more than 2X the size of…
ArtificialAnlys's tweet image. Jamba model launch takeaways - Potentially a new leader for ultra-long prompt use-cases (RAG)

‣ First open-source model of this size to combine MAMBA state-space model architecture, Mixture-Of-Experts (MOE) and the transformer
‣ 256k context window, more than 2X the size of…

nolan reposted

the mission continues

We remain committed to our partnership with OpenAI and have confidence in our product roadmap, our ability to continue to innovate with everything we announced at Microsoft Ignite, and in continuing to support our customers and partners. We look forward to getting to know Emmett…



nolan reposted

Generated some characters on @AutoCubeAI and turned them into VRM avatars. Environment is generated too! Process: AutoCube -> Blender -> Mixamo -> UniVRM


nolan reposted

I'm prototyping an AI-powered pet NFT, drop your address / let me know if you want to be whitelisted for a free mint! It's in VRM format so you can use it as your avatar as well. The design is simple but it's something I want personally (originally for a game a year ago but…


nolan reposted

Just generated "a futuristic vehicle that combines elements of a motorcycle and a hovercraft" on @AutoCubeAI (AI randomly came up with the prompt too) and am now driving it in @hyperfy_io with my @vipeio avatar! 🏍️🌬️ Link to world: hyperfy.io/autocube


nolan reposted

Launching @OpenAI Shap-e support on @OpenRouterAI and @windowai! Generate high quality 3D assets in < 30 seconds, and build sustainable apps where users can pay for their own usage.


nolan reposted

Letting users switch between LLMs on a public @streamlit app. Aimed for a simple, zero-cost dev setup: 1. No LLM API keys needed. 2. Uses only standard dependencies like openai and @LangChainAI, even for Claude/PaLM. Playground URL: openrouter.streamlit.app


nolan reposted

A decentralized approach I’m excited about by @xanderatallah is windowai.io


nolan reposted

Windowai.io is now live on the Chrome store ✨ It's a new way to build AI apps: users choose the model, and developers don't have to pay. Check out apps launched in the last two weeks below. chrome.google.com/webstore/detai…


nolan reposted

It's a very early ecosystem, but you can discover other apps on the "Product Hunt for Window apps," Skylight: skylightai.io Made by @YKouloumbis and @nolangclement


Download the extension and check out some of the apps people have built on skylightai.io ! Hop in the discord via windowai.io to learn more about the API and get started building your own creations 🚀

And we're finally live on the store! Get the extension for Chrome here. More browsers coming soon.



nolan reposted

Pitting GPT3.5 vs GPT4 vs GPT-NeoXT in chess 👀 GPT4 is slow but provides great game analysis...

Can you beat an LLM at chess? Try playing Chess GPT! Powered by @windowai_io (bring your own AI model). Link: play-chess-gpt.vercel.app



nolan reposted

Window AI now has a Twitter and public roadmap, with several starter tasks if you're looking to contribute! 🐦 @windowai_io 🐙 Roadmap: github.com/alexanderatall… 🐣Starter tasks: github.com/alexanderatall…


nolan reposted

if you're a recruiter i have the receipts i wasn't lying about being the 100x dev

Bro hasn’t seen our shipping timelines yet 💀 3.5 hours

YKouloumbis's tweet image. Bro hasn’t seen our shipping timelines yet 💀 3.5 hours


Loading...

Something went wrong.


Something went wrong.