Log inSign up
Log inSign up
Hamish Ivison
607 posts
Hamish Ivison profile banner
@hamishivi

Hamish Ivison

@hamishivi
Antipodean Abroad. I (try to) do NLP research. PhD student @uwcse, prev @allen_ai @google @Sydney_Uni 🇦🇺🇨🇦🇬🇧
Earth
hamishivi.github.io
Joined March 2016
775 Following
2,667 Followers
RepliesRepliesRepostsRepostsMediaMedia

Hamish Ivison’s posts

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·US TIDA·Ads Info·© 2026 X Corp.
  • @hamishivi
    Hamish Ivison
    @hamishivi
    Nov 16
    Check out our long horizon infra eval!
    @TengX6
    Teng Xiao
    @TengX6
    Nov 16
    [1/4] 🚀 Introducing Infrathon (infrathonx.github.io) Can LLM agents autonomously improve real-world AI infrastructure, make new discoveries, and outperform human-designed baselines on long-horizon, open-ended optimization tasks? AI infra optimization requires repeated
    00:00
  • @hamishivi
    Hamish Ivison
    @hamishivi
    Nov 3
    All u need is good data and good infra -> many points improvement with tmax!
    @oscar_yinn
    Oscar Yinn
    @oscar_yinn
    Nov 3
    Terminalmaxxing cannot be stopped 🤫 Introducing TMax v1.1! We improved our data and infrastructure, and managed to improve our models by up to 6 points on TB2. TMax v1.1 dominates the Pareto curve for open terminal agents under 30B. As usual, we release everything: our
    00:00
    2
  • @hamishivi
    Hamish Ivison
    @hamishivi
    Nov 2
    Stop by to listen to me yap abt tmax and rl envs!
    @real_jjmachan
    Jithin James
    @real_jjmachan
    Nov 2
    This weeks paper club we have @hamishivi, first author of TMax speaking to us about the paper and building data recipes for terminal agents in general. RL envs are something that excites you do stop by for our paper-club! We only ask you to bring good questions 😉
  • @hamishivi
    Hamish Ivison
    @hamishivi
    Oct 4
    Excited for COLM next week! (and my time properly visiting SF! 🌉) Come learn about Meta-RL for Search at our poster: colm.cc/virtual/2026/p… Otherwise, I'll be around! Please reach out if you would like to chat abt whatever, esp credit assignment, RL, terminal agents!
    4
  • @hamishivi
    Hamish Ivison
    @hamishivi
    Sep 30
    Helped out a little on this! This setup nicely generalises to thing like subagents/swams/etc - just use multiple context files with differing permissions! Lots of details packed into this paper so please give a read!
    @RulinShao
    Rulin Shao
    @RulinShao
    Sep 30
    ‼️The Bitter Lesson for context management: Giving LMs unrestricted control over their context beats human-designed SOTA! Introducing 🩵Context Language Models (CLMs)🩵 - Natively manage their own context - Treat context as a file - Learn policies in CLM weights, no harness