AI300
  1. Cameron Wolfe / Researcher at Netflix
    2311.
    Believe it or not, one of the first blog posts I ever wrote was about the OPT model / library.›
    LLM score 25 · 6 months ago
  2. Alex Zhang / MIT CSAIL PhD
    2312.
    New mini experiment + blogpost + trajectories! tldr; we boost performance of RLM(GPT-5.2) to double the best performing number (38.7% --> 65.6%) on LongCoT-mini without any training! An example of the mismanaged geniuses hypothesis (MGH) we (@zli11010, @lateinteraction) https://t.co/6xx2vfiTCL
    LLM score 92 · 6 months ago
  3. Cameron Wolfe / Researcher at Netflix
    2313.
    Most people are aware of KL regularization for RL, but initial work (e.g., PPO paper) also used an entropy bonus for regularization purposes.›
    LLM score 92 · 6 months ago
  4. Lewis Tunstall / Hugging Face Researcher
    2314.
    gpt-5.5 is now available in the ml-intern! this means it gets access to the whole @huggingface infra: buckets, jobs, repos etc for doing ai research at scale›
    LLM score 92 · 6 months ago
  5. Lewis Tunstall / Hugging Face Researcher
    2315.
    We’ve now got multiple ML interns collaborating together on the hub! ›
    LLM score 85 · 6 months ago
  6. Jeffrey Wang / OpenAI Researcher
    2316.
    a beautiful & simple & important GPT-5.5 proof :) https://t.co/VrDc4BHtPn
    LLM score 92 · 6 months ago
  7. Aidan McLaughlin / OpenAI Research Scientist
    2317.
    there had long been a vacuum of an "mkbhd for ai"— someone with the class to do serious llm reviews and rise above engagement farming.›
    LLM score 75 · 6 months ago
  8. Damek Davis / Assoc. Professor Wharton Stats
    2318.
    It's been fun looking through the stage 1 cheatsheets, while we finish up the evaluation.›
    LLM score 75 · 6 months ago
  9. Yoshua Bengio
    2319.
    AI is advancing faster than our ability to manage it.›
    LLM score 92 · 6 months ago
  10. Jerry Tworek / ex OpenAI VP of RL
    2320.
    If anyone ever asks what does "frontier" model means, show them this picture as a definition: https://t.co/BEfamwdffs
    LLM score 75 · 6 months ago
  11. Jason Weston / Meta Research Scientist
    2321.
    DeepSeek-V4 uses our Hash routing approach developed back in 2021 -- see screenshot of their tech report! (Looks like a great model, congrats!)›
    LLM score 92 · 6 months ago
  12. Alex Zhang / MIT CSAIL PhD
    2322.
    Forgot to mention this but the wonderful folks at @AntimLabs are hosting and have eval'd the latest VLMs on VideoGameBench!›
    LLM score 82 · 6 months ago
  13. Jack Clark / Anthropic Cofounder
    2323.
    SNOWSUMMER, the Ultimate Insurance Policy(blog)
    LLM score 95 · 6 months ago
  14. Sebastien Bubeck / OpenAI MTS
    2324.
    Ramsey numbers are one the most basic objects in combinatorics, a beautiful illustration of structure within chaos.›
    LLM score 98 · 6 months ago
  15. Ben Burtenshaw / Hugging Face Researcher
    2325.
    DeepSeek-V4 dropped. 1M context.›
    LLM score 75 · 6 months ago
  16. Merve Noyan / Hugging Face ML Engineer
    2326.
    if you were in a cave, DeepSeek v4 is out, and it's groundbreaking, here's why:›
    LLM score 92 · 6 months ago
  17. Ben Burtenshaw / Hugging Face Researcher
    2327.
    deepseek-v4 is out and solves context rot at 1M tokens by taking on attention for the kv cache.›
    LLM score 92 · 6 months ago
  18. Merve Noyan / Hugging Face ML Engineer
    2328.
    DSv4 genuinely shines in 1M context window and peak efficiency to run many agents/users 😍›
    LLM score 85 · 6 months ago
  19. Clive Chan / OpenAI Hardware
    2329.
    DeepSeek V4 technical report is GOATed as always but the hardware proposals - which had a large impression on the hardware community in V3 - are surprisingly mid/wrong this time.›
    LLM score 92 · 6 months ago
  20. Merve Noyan / Hugging Face ML Engineer
    2330.
    DeepSeek v4 is out with 1M context window 🥵🔥 > Pro (13B/284B) & Flash (49B/1.6T) ›
    LLM score 92 · 6 months ago
  21. Cameron Wolfe / Researcher at Netflix
    2331.
    The idea of training LLMs to manage their own KV cache is super interesting to me.›
    LLM score 92 · 6 months ago
  22. Sherwin Wu / OpenAI API, Head of Engineering
    2332.
    Set Codex to this and never look back. Medium reasoning effort is good enough for me for ~anything I need to do now.›
    LLM score 85 · 6 months ago
  23. Sherwin Wu / OpenAI API, Head of Engineering
    2333.
    It's not every day you hear a frontend engineer of Tyler's caliber praise a model this much.›
    LLM score 75 · 6 months ago
  24. Roland Gavrilescu / ex MTS at xAI
    2334.
    Autoresearch will be the trend of the year https://t.co/xIMiSQi8EL
    LLM score 75 · 6 months ago
  25. Roland Gavrilescu / ex MTS at xAI
    2335.
    Fate loves irony: OpenAI has safer models than Anthropic https://t.co/pZnVVoQMI0
    LLM score 85 · 6 months ago
  26. Clive Chan / OpenAI Hardware
    2336.
    this 38h run was GPT5.5! managing a team of agents is mandatory for researchers now tbh›
    LLM score 92 · 6 months ago
  27. Aidan McLaughlin / OpenAI Research Scientist
    2337.
    over break i dictated to 5.5 for minutes describing a new ambitious rl run.›
    LLM score 85 · 6 months ago
  28. Noam Brown / OpenAI Research Scientist
    2338.
    A hill that I will die on: with today's AI models, intelligence is a function of inference compute.›
    LLM score 92 · 6 months ago
  29. Sebastien Bubeck / OpenAI MTS
    2339.
    GPT-5.5, not fully saturating the TikZ unicorn test yet but getting awfully close ...›
    LLM score 92 · 6 months ago
  30. Jonathan Ross / TPU Creator
    2340.
    I've had early access to this model, and it's 🤯 I've been going back and implementing ideas I had for LPU software over the years that we never had enough resources to do.›
    LLM score 75 · 6 months ago
More