Log inSign up
user avatar
trieu
@thtrieu_
inventor of #alphageometry. thinking about thinking @ deepmind.
Mountain View
github.com/thtrieu
Joined April 2014
147
Following
2,652
Followers
  • Pinned
    user avatar
    trieu
    @thtrieu_
    Feb 2
    Mathematicians 🤝AI researchers arxiv.org/abs/2601.22401. Our take on AI solving Erdos problems: * Many "Open" problems are actually just obscure: many cases the AI didn't find something new, only rediscovered solutions buried in the literature. We present our systematic approach
    user avatar
    Thang Luong
    @lmthang
    Feb 2
    Replying to @lmthang
    Here's the paper link to our scaled effort for tackling Erdős problems. We started with 700 problems marked ‘Open’ in the database. Our agent #Aletheia identified potential solutions to 200 problems. Initial human grading revealed 63 correct answers, followed by deep expert
    Image
    arXiv logo
    arxiv.org
    Semi-Autonomous Mathematics Discovery with Gemini: A Case Study on...
    We present a case study in semi-autonomous mathematics discovery, using Gemini to systematically evaluate 700 conjectures labeled 'Open' in Bloom's Erdős Problems database. We employ a hybrid...
    30K
  • user avatar
    trieu
    @thtrieu_
    Jan 17, 2024
    Proud of this work. Here's my 22min video explanation of the paper:
    user avatar
    Google DeepMind
    @GoogleDeepMind
    Jan 17, 2024
    Introducing AlphaGeometry: an AI system that solves Olympiad geometry problems at a level approaching a human gold-medalist. 📐 It was trained solely on synthetic data and marks a breakthrough for AI in mathematical reasoning. 🧵 dpmd.ai/alphageometry
    Image
    GIF
    2.4M
  • user avatar
    trieu
    @thtrieu_
    Jun 11, 2018
    We show large language models trained on massive text corpora (LM1b, CommonCrawl, Gutenberg) can be used for commonsense reasoning and obtain SOTA on Winograd Schema Challenge. Paper at arxiv.org/abs/1806.02847, results reproducible at git.io/vhaaf
    Image
  • user avatar
    trieu
    @thtrieu_
    Feb 16, 2019
    As also observed by OpenAI's GPT-2, training data quality is important. We release the STORIES corpus introduced in our work arxiv.org/abs/1806.02847. The corpus is a high quality subset of CommonCrawl with a total of ~7B words (~32GB) can be found here: console.cloud.google.com/storage/browse…
    Image
    Image
  • user avatar
    trieu
    @thtrieu_
    Jan 26, 2019
    Wow! An old project of mine github.com/thtrieu/darkfl… is now the 7th most popular Machine Learning project across all Github in 2018, alongside with Tensorflow and Scikit-learn? I really need to spend some time polishing it now...
    user avatar
    GitHub
    @github
    Jan 24, 2019
    From the programming languages you used most to the most popular data science packages, we’re digging into the data on Machine Learning from 2018. Find out what we discovered github.blog/2019-01-24-the…
    Image
  • user avatar
    trieu
    @thtrieu_
    Aug 1, 2018
    Human reasoning is not manipulating symbolic expressions
    user avatar
    CIFAR
    @CIFAR_News
    Jul 31, 2018
    Replying to @CIFAR_News
    What is thought?: Big questions from CIFAR Distinguished Fellow Geoffrey Hinton #dlrl2018
    Image
  • user avatar
    trieu
    @thtrieu_
    Jan 29, 2020
    Had the chance to sit next to Daniel @xpearhead in the early days of the project and tried out the interactive Meena. It has always been *this* surprising and funny :) BIG Congrats to the team with this publication. The possibilities to build up from here is endless.
    user avatar
    Kyle McDonald
    @kcimc
    Jan 29, 2020
    um.. google's latest chatbot is 😳 arxiv.org/abs/2001.09977
    Image
  • user avatar
    trieu
    @thtrieu_
    May 12, 2018
    Our work on learning longer-term dependencies is accepted at @icmlconf #icml2018
    user avatar
    Thang Luong
    @lmthang
    Mar 2, 2018
    Excited to share a new work by #GoogleAI resident @thtrieu_ (with @andrewdai, me, & Quoc Le) on training very long RNNs (up to 16K long). See paper for extreme cases of zero or little backprop on RNNs ;) arxiv.org/pdf/1803.00144…
    Image
  • user avatar
    trieu
    @thtrieu_
    Apr 29, 2018
    I'll be presenting my work #ICLR2018 on Wednesday. Come and have a chat :)
    Image
  • user avatar
    trieu
    @thtrieu_
    Jun 14, 2022
    "We see contributions to traditional conferences and publications in journals as an important part of our work, but also support efforts that go “beyond the research paper"".
    user avatar
    Sara Hooker
    adaption
    @sarahookr
    Jun 14, 2022
    I'm excited to finally share what I have been working on. Today we are officially launching Cohere For AI @forai_ml a non-profit research lab that aims to reimagine how, where, and by whom research is done. cohere.for.ai
    Image
  • user avatar
    trieu
    @thtrieu_
    Oct 18, 2018
    A commonsense reasoning task is "solved" even before its official introduction.
    user avatar
    Sebastian Ruder
    @seb_ruder
    Oct 12, 2018
    It's amazing how fast #NLProc is moving these days. We have now reached super-human performance on SWAG, a commonsense task that will only be introduced at @emnlp2018 in November! We need even more challenging tasks! BERT: arxiv.org/abs/1810.04805 SWAG: arxiv.org/abs/1808.05326
    Image
    Image
  • user avatar
    trieu
    @thtrieu_
    Feb 16, 2024
    10M toks with almost perfect needle in haystack, amazing.
    user avatar
    Jeff Dean
    @JeffDean
    Feb 15, 2024
    Gemini 1.5 Pro - A highly capable multimodal model with a 10M token context length Today we are releasing the first demonstrations of the capabilities of the Gemini 1.5 series, with the Gemini 1.5 Pro model. One of the key differentiators of this model is its incredibly long
    Image
    1.8K
  • user avatar
    trieu
    @thtrieu_
    Jan 18, 2024
    Yes, the fact that we can do this means the geometry we considered is quite narrow.
    user avatar
    Thomas Ahle
    Normal Computing 🧠🌡️
    @thomasahle
    Jan 18, 2024
    Making a synthetic dataset of mathematical proofs is hard! It's easy to make a whole lot of "1+1+1+...=491" style theorems. I'm surprised this method of random construction and transformation finds so many classical geometric theorems. Maybe because the domain is somewhat
    1.2K
  • user avatar
    trieu
    @thtrieu_
    Feb 9, 2019
    Want to see space-time contract/dilation? This series on Special Relativity (SR) is beautiful. The author squashed space-time to 2D, explained the two postulates by geometric intuition and run a simulator on top of it. It is the 3blue1brown of SR!