Free Compute @ ICLR

Prime Intellect democratizes AI development at scale. Our platform makes it easy to find global compute resources and train state-of-the-art models through distributed training across clusters. Collectively own the resulting open AI innovations, from language models to scientific breakthroughs.

$50 Free Compute for ICLR Attendees

We democratize powerful compute access to advance AI research by ag. Train bigger models, run more experiments, accelerate discovery.

Prime Intellect supports ICLR's groundbreaking AI researchers with $50 free compute grants for all attendees.

  • Accelerate research on our aggregated compute network
  • Access on-demand Nvidia H100s/A100s for scalable training

Redeem by filling out this form

*Limited to 1 per attendee, expires at 12/31/24, non-transferable

<div data-tf-live="01HX71224YAKD4H9CC279CZ07Q"></div><script src="//embed.typeform.com/next/embed.js"></script>

Fast Compute Grants

We're launching Fast Compute Grants to support ambitious research into the novel models, architectures, techniques, and paradigms needed to make open and decentralized AGI a reality.

We're on a mission to accelerate open and decentralized AI progress by enabling anyone to contribute compute, capital or code to collaboratively train powerful open models. Our ultimate goal? Openly accessible AGI that benefits everyone. But we can't do it alone and we want to do this together with you.

That's why we're launching Fast Compute Grants to support ambitious research into novel models, architectures, techniques, and paradigms needed to make open and decentralized AGI a reality.

A few ideas of what we’d be excited about

  • Decentralized Llama-3 MoE Sparse upcycling via DiPaCo
  • Scientific foundation models with new architectures ala HyenaDNA
  • scGPT as a Llama-3 fine-tune (S/O to Felix)
  • Distributed training across a heterogenous swarm of consumer devices (S/O to Omkaar)
  • 13B parameter BitNet + infini-Attention + DenseFormer + MoD + In Context-Pretraining + 2 stage pretraining (S/O to Wing)
  • Upcycle w c-BTX to an 8 expert sparse MoE + MoA (S/O to Wing)
  • Coding agent models
  • Novel applied super-alignment research
  • Efficient long-context context window extension
  • Exploring new Transformer and alternative architectures

If you're working on something in this vein that could use a boost from free GPUs, we want to hear from you. We'll provide:

  • $500-$100k worth of Prime Intellect compute credits.
  • Exposure to our ecosystem of AI hackers and distributed computing experts
  • Promotion of your work to our community and partners

You can apply via this form, and we’ll get back to you in 5-10 days.

  1. What you're working on and why it's important
  2. How much compute you need and what you'd do with it
  3. Who you are and any past work we should know about
  4. Any code/papers/demos/other material that'll get us excited

The bar for quality is high but there are no other hoops. Anyone from anywhere can apply. Just email your pitch to contact@primeintellect.ai. We'll get back to you within 2 weeks if it seems like a good fit.

Our goal is to get a critical mass of brilliant people pointed at the hardest problems in open and decentralized AI and equip them to make rapid progress.

The future won't build itself - let's get to work.

‍

Apply Below

<div data-tf-live="01HWZ9Q3VNFQD4J12DDTB8931G"></div><script src="//embed.typeform.com/next/embed.js"></script>