Physicist, Machine Learner
Mountain View, CA
Joined July 2008
- Tweets687
- Following253
- Followers868
- Likes148
Guy Gur-Ari retweeted
Claude Sonnet 4.5 is now the default model for Augment Code.
We’re rolling it out to all customers over the next 24 hours, where it will be available alongside Sonnet 4 (for a limited time) and GPT-5 in the model picker.
Here’s how Sonnet 4.5 compares to Sonnet 4:
Guy Gur-Ari retweeted
Introducing Remote Agent - cloud-based dev agents with full-codebase context, deep IDE integration, and full toolchain access.
Available in VS Code.
Guy Gur-Ari retweeted
Augment Agent is now live.
Designed for large-scale codebases — with 200K context tokens, persistent memory, and deep tool integrations, Augment Agent is available in VS Code and JetBrains.
Guy Gur-Ari retweeted
Just bumped into @jyangballin on Stanford campus and discussed future of coding agents and swebench/sweagent. Very happy such smart and kind folks are working on this important problem with an open-source POV. The future is bright ☀️
Strongly believe open source is critical puzzle piece in AI safety story over the coming decade. To play our part, my research colleagues at Augment and I will begin by posting a blog post soon detailing all our learnings and hard-learned discoveries building coding agents for swebench
Guy Gur-Ari retweeted
Working on open source is hard.
We just made it a little easier.
Augment is free for open source contributors and maintainers. 🎉
Augment’s deep codebase understanding tears down some of the biggest pains for working on open-source projects– the context you need to contribute and the help you need to make quality PRs.
Sign up at augmentcode.com/opensource
Guy Gur-Ari retweeted
AI coding tools are everywhere. Yet they all fall short when scaling to 100s of developers and complex codebases. We’re solving this problem @augmentcode.
Introducing Augment Code, the first developer AI purpose-built for teams.🧵
Guy Gur-Ari retweeted
Happy to release our work on Language Model Cascades. Read on to learn how we can unify existing methods for interacting models (scratchpad/chain of thought, verifiers, tool-use, …) in the language of probabilistic programming.
paper: arxiv.org/abs/2207.10342
Guy Gur-Ari retweeted
1/ Super excited to introduce #Minerva 🦉(goo.gle/3yGpTN7). Minerva was trained on math and science found on the web and can solve many multi-step quantitative reasoning problems.
Very excited to present Minerva🦉: a language model capable of solving mathematical questions using step-by-step natural language reasoning.
Combining scale, data and others dramatically improves performance on the STEM benchmarks MATH and MMLU-STEM. goo.gle/3yGpTN7
Guy Gur-Ari retweeted
Very excited to announce a significant milestone in expanding reasoning capabilities of language models! 🎉🎉
We introduce #Minerva🦉: a language model that can solve mathematical questions using step-by-step natural language reasoning:
bit.ly/3OBj2d5
🧵
1/
Very excited to present Minerva🦉: a language model capable of solving mathematical questions using step-by-step natural language reasoning.
Combining scale, data and others dramatically improves performance on the STEM benchmarks MATH and MMLU-STEM. goo.gle/3yGpTN7
Guy Gur-Ari retweeted
Thrilled to announce🦉Minerva: a large language model capable of solving mathematical problems using step-by-step reasoning in natural language.
See blog here: goo.gle/3yGpTN7 and samples here: minerva-demo.github.io (1/n)
Very excited to present Minerva🦉: a language model capable of solving mathematical questions using step-by-step natural language reasoning.
Combining scale, data and others dramatically improves performance on the STEM benchmarks MATH and MMLU-STEM. goo.gle/3yGpTN7
Guy Gur-Ari retweeted
Super excited to share Minerva!! – a language model that is capable of solving MATH with 50% success rate, which was predicted to happen in 2025 by Steinhardt et. al. (bounded-regret.ghost.io/ai-f…)!
#Minerva
1/
Very excited to present Minerva🦉: a language model capable of solving mathematical questions using step-by-step natural language reasoning.
Combining scale, data and others dramatically improves performance on the STEM benchmarks MATH and MMLU-STEM. goo.gle/3yGpTN7
Guy Gur-Ari retweeted
Very excited to present Minerva🦉: a language model capable of solving mathematical questions using step-by-step natural language reasoning.
Combining scale, data and others dramatically improves performance on the STEM benchmarks MATH and MMLU-STEM. goo.gle/3yGpTN7
Guy Gur-Ari retweeted
Language models have shown strong performance on a variety of #NLU tasks but are weaker at solving tasks that involve quantitative reasoning. Learn how #Minerva uses step-by-step reasoning to achieve a new state of the art on quantitative reasoning tasks→ goo.gle/3yGpTN7
Guy Gur-Ari retweeted
imo a major AI safety contribution, both in short-term (applications) and long-term (AGI) scope
Guy Gur-Ari retweeted
This is incredible work on LLMs! Reading through this paper, I'm not only amazed by the huge amount of work behind BIG-bench, but also fascinated by the observations of current LLMs' (BIG-G, GPT, PaLM) behaviors on those tasks. A few things that I felt deeply resonated:
After 2 years of work by 442 contributors across 132 institutions, I am thrilled to announce that the github.com/google/BIG-bench paper is now live: arxiv.org/abs/2206.04615. BIG-bench consists of 204 diverse tasks to measure and extrapolate the capabilities of large language models.
Guy Gur-Ari retweeted
Thanks @LiamFedus @AJAndreassen @jaschasd @ethansdyer @guygr and team for the incredible work on BigBench !
You can find it on the Hugging Face Hub here
👉 hf.co/datasets/bigbench
Guy Gur-Ari retweeted
Today we release our paper in response to @ericweinstein's "Theory of Everything" #geometricunity: timothynguyen.files.wordpres… Our objections:
Guy Gur-Ari retweeted
In the Les Houches School of Physics, established in the wake of WWII, a *in-person* workshop on Physics & Machine Learning starting today: leshouches2020.krzakala.org Feels eerie, but so desired. Never had I imagined to have to restrict participation to European residents ...
Guy Gur-Ari retweeted
This was put in place to prevent backdoor access to the OpenAI API. Additionally we have finetuned on a specific dataset and use parameters optimized for our use case making AI Dungeon not necessarily representative of GPT-3 in general.