✨Paper out in final form: exciting results from our semi-supervised pose estimation package, Lightning Pose, which is now adopted by a number of great neuroscience labs. Please give it a whirl: github.com/danbider/light…

If you want to learn more about how the Llama3 team used annealing to assess data quality check out our paper! At ICML? go chat with Mansheej Paul about it!

Awesome to see so much open science shared in the Llama 3.1 paper, including a shoutout to Cody Blakeney and Mansheej Paul's work. There are also great details on RLHF and other aspects of Llama 3.1.

Pretraining data ablations are expensive: how can we measure data quality fast and cheap? If you're at ICML, come find out at the ES-FoMo poster session today in Lehar 2 at 1 pm: icml.cc/virtual/2024/w…

LLM as a judge works well by burning extra Inference compute on chain of thought and self critiques. Reward models work well due to Bradley Terry style objectives being a good fit for most current preference datasets Now you can have the best of both worlds!

Code and models for our latest work Critique-out-Loud (CLoud) Reward models is now released! Check out our paper (arxiv.org/abs/2408.11791) for more details on using reward models to reason before predicting a reward score.

Agreed ;) But in all seriousness, its cool to see everyone converging on reward models that perform explicit reasoning by critiquing out loud. Super excited to see how people build on top of these works.

If you want to read more about the curriculum training used in OLMo 2 checkout our (Mansheej Paul Brett Larsen Sean Owen) paper! Congrats on the release to everyone at AI2! (but especially Luca Soldaini 🎀 and Kyle Lo <3 data ) arxiv.org/abs/2406.03476

Critique out loud reward models made it into the Kimi k1.5 technical report! Super cool to see someone scale it up to 800k inputs and to see how much better reward modeling it led to!

💥New Paper! Algorithmic Phases of In-Context Learning: We show that transformers learn a superposition of different algorithmic solutions depending on the data diversity, training time and context length! 1/n

How can we use small LLMs to shift more AI workloads onto our laptops and phones? In our paper and open-source code, we pair on-device LLMs (ollama) with frontier LLMs in the cloud (@openai, @together), to solve token-intensive workloads on your 💻 at 17.5% of the cloud cost

Deep learning training is a mathematical dumpster fire. But it turns out that if you *fix* the math, everything kinda just works…fp8 training, hyperparameter transfer, training stability, and more. [1/n]

Imagine if boats had twitter. They’d be like “@dock is this true?”

Imagine if threads had twitter. They’d be like “@lock can I do?”

Engineers spend 70% of their time understanding code, not writing it. That’s why we built Asimov at Reflection AI. The best-in-class code research agent, built for teams and organizations.

Imagine if memory pointers had twitter. They’d be like “Malloc is this true?”

It was kinda a movie