Liquid Clustering is now GA on Databricks -- a new layout strategy that outperforms partitioning and does not require tuning. This went from an idea in our engineering team to supporting 1000+ customers and 100+ petabytes of tables in just one year! databricks.com/blog/announcin…

Congrats to our former WeWork neighbors PatronusAI on raising their Series A!!!

We are excited to announce our $17M Series A led by Glenn Solomon at Notable Capital with participation from Nnamdi Iregbulem and lsvp, Datadog, Inc., @matthartman, Jonathan Frankle, Jason Warner, Tristan Handy, Barr Moses, Aparna Sinha, Nadim Hossain and a number of tech industry executives

We are announcing new grants for research into systemic AI safety.

Initially backed by up to £8.5 million, this program will fund researchers to advance the science underpinning AI safety.

Read more: gov.uk/government/new…

Sounds like continuing to live as a hero to me.

It is always such a joy being able to give back to the research community. Especially in an age where so many insights are happening in close labs and only whispered a coffee shops and conference after parties.

This is super exciting! We’ve been working on diffusion models for some time, and releasing CommonCanvas! Much more to come on this front soon :)

LoRA Learns Less and Forgets Less

Low-Rank Adaptation (LoRA) is a widely-used parameter-efficient finetuning method for large language models. LoRA saves memory by training only low rank perturbations to selected weight matrices. In this work, we compare the performance of

Quite excited that CommonCanvas is JUST out! 🖼️

• First open source text-to-image models trained fully on openly licensed images (SD2 and SDXL architectures)
• The dataset, with ~70M openly licensed creative commons images + synthetic captions, is also out (largest dataset on

Dataset drop!! Excited to announce the CommonCanvas dataset, a large-scale Creative-Commons dataset is now open-source on Huggingface (1/n)

This was a monumental effort from so many people. Aaron Gokaslan at Cornell, A. Feder Cooper at Cornell/Yale soon, jasmine collins Landan Seguin @Austinjjac Mihir Patel and Cory Stephenson at Databricks, and Volodymyr Kuleshov 🇺🇦 at Cornell.

Amazing work by Aaron Gokaslan! This is a great diffusion model that was trained on open CC data! Great contribution, and amazing that he open sourced it!

CommonCanvas is here! A new dataset for training text and image models, and a family of models to go with it. All Creative Commons licensed. Amazing work by Aaron Gokaslan!!!

At Databricks Mosaic Research, we (and our customers! we actually make money) train custom LLMs for specific use cases. Today, we're sharing a large part of our infra stack and many of the tricks we use to get reliable performance at scale. (1/N)

Another Dolphin-2.9.1 🐬 coming right at you! Thanks Databricks Mosaic Research for the excellent base model, Crusoe Energy for sponsoring the compute! Thanks to my crew Lucas Atkins and Fernando Fernandes Neto for all the support! Uncensored models will hurt your feelings. 😱 You are

Good take!