Ani Rahul

DeepGrove @deepgrove_ai · 4d

Today, in the first of a series of blog posts we’re calling “Behind the Grove”, we’re offering a look into some of the technical work we do at DeepGrove.

Our first post is about our pretraining infrastructure, Grove. Grove is built around 2 backends: Grove-FSDP and DDP. We detail how we optimized both backends, culminating in an over 2x speedup over public baselines.

A production-scale pretraining run of Maple 20B-A1B (deepgrove.ai/maple-preview) on 4T tokens takes a mere 2 weeks on 64 H100s.

We view infrastructure as the backbone of our work, and as one of the many pieces that help us build toward the frontier of efficiency and performance.

To read the full post, visit deepgrove.ai/infra.

View quoted post on X

Hanchen Li Verified account @lihanc02 · 5d

Bro...@AfterQuery

View quoted post on X

Cody Blakeney Verified account @code_star · Nov 2

The reason it’s ONLY 85% margin is because of open source.

There should be no doubt in your mind if there was not open models capable of doing much of the same tasks being served at only 25-40% margin Anthropic and other closed labs would be happy to take 200%+ margin.

It’s competitive pressure and it’s good for everyone.

View quoted post on X

General Compute Verified account @general_compute · Oct 27

🚨 Big news today: We’ve raised $100M for our Series A! 🎉 🚀

We believe demand for fast inference is many times larger than demand for slow inference, and it’s about to go exponential.

Thanks to our co-leads @BlackbirdVC and @Abstractvc. Two visionary partners join our journey 🎉 🚀

View quoted post on X