Nebius AI monthly digest, July 2024
Nebius AI monthly digest, July 2024
In July, we introduced our LLM R&D team and shared more of the team’s experience. We also launched a GPU auction and published diverse materials, including guest post about the London Institute’s research based on our infra.
In-house LLM R&D: Nebius AI’s secret ingredient for truly AI‑centric cloud
To build a full-fledged ML platform, we realized it’s necessary to perform large-scale distributed training in-house. That’s why we formed the LLM R&D team, leveraging our compute capacities to let us deeply specialize Nebius AI, while also advancing our own AI efforts. Read the full story here.
For some time now, though, the team has been sharing its experience in managing heavy tasks around building an LLM. One of the ML engineers on the team, Maksim Nekrashevich, has been among the speakers sharing such insights. Check out his talk for the LLMOps Space community featuring Philip Tannor, CEO and Co-Founder at Deepchecks. Maxim discussed building the fine-tuning pipeline for LLM alignment:
JupyterHub with PyTorch and CUDA®
JupyterHub, provided along with PyTorch and CUDA on our Marketplace, is a multi-user server for notebooks. Here’s why you should consider using this version.
The latest on our docs and blog
Guides to auctions and other billing concepts
Auctions with descending asking price might be new to you, but don’t worry: the detailed guide in our documentation will help you figure them out. Meet Alice, Bob, Cyril and Daniel there — they’re our model bidders! Also, you can now learn more about our billing concepts: threshold and payment methods.
Nebius AI at the frontiers of physics and maths
The guest post on our blog is written by Ananyo Bhattacharya, Chief Science Writer at the London Institute for Mathematical Sciences. We are honored to host multiple projects of LIMS on our infra.
Krisp’s advancements in real‑time voice AI
Our client Krisp works in the field of Accent Localization, an AI-powered real-time voice conversion technology that removes the accent from call center agent speech, resulting in a US-native speech. The goal is to improve the clarity and comprehension in human-to-human communication, when the agent — located, say, in India — talks to a US-based customer.
Mixtures of Experts and scaling laws
The latest article in our AI research series explores the steps scientists have taken on the road toward the perfect Mixture of Experts.
Explore Nebius AI
See also
Our June’s highlights included offering up to 8 GPUs via self-service, decoding GPU pricing and adding L40S to our portfolio. We also detailed our Terraform provider, discussed LLM dataprep, announced DVC.ai partnership and introduced feature requests.
In May, we published a guide on using DVC in Nebius, discussed Slurm vs K8s and fundamentals of LoRA, picked the best GPUs for CV, hosted a webinar on applying RAG — and welcomed our tech partner Weights & Biases with its Launch agent in Marketplace.
Our main news of the past month is that Nebius AI has become available to everyone! We also participated in the MLOps podcast and published several videos about setting up training, as well as stories about how Nebius AI clients are building their models.