AI300
  1. John Carmack
    2431.
    It is generally frowned upon to have LLMs precisely regurgitate part of their training set, but it is an interesting question how you could use LLM training to nearly losslesly compress a huge corpus like the entirety of the Internet Archive.›
    LLM score 92 · 7 months ago
  2. Dan Fu / VP of Kernels at Together
    2432.
    📢 Super excited to announce Parcae! We've been thinking about scaling laws and the "right" way to get more FLOPs.›
    LLM score 92 · 7 months ago
  3. Eric Jang / ex VP of AI at 1X Robotics
    2433.
    OAI folks: I've wanted to try codex CLI for some time, but I'm unwilling to pay for both Claude and ChatGPT subscriptions.›
    LLM score 75 · 7 months ago
  4. Alex Zhang / MIT CSAIL PhD
    2434.
    Some awesome exploration into how feedback influences RLM behavior through the @PrimeIntellect RL residency! https://t.co/DUz9BNfojw
    LLM score 85 · 7 months ago
  5. Demis Hassabis / CEO of DeepMind
    2435.
    Great to see our collaboration w/ @BostonDynamics unlocking new capabilities! Gemini Robotics-ER 1.6 enables robots like Spot to read complex industrial gauges autonomously.›
    LLM score 92 · 7 months ago
  6. Andrew White / Edison Scientific Cofounder
    2436.
    3 years ago we showed that you could get catalyst activity predictions that were calibrated enough to do Bayesian optimization against with LLMs.›
    LLM score 92 · 7 months ago
  7. Michael Elabd / DeepMind Researcher
    2437.
    Compiled some of my thoughts after talking to 200 AI-native companies on building defensible businesses in the age of claude code!›
    LLM score 85 · 7 months ago
  8. Michael Elabd / DeepMind Researcher
    2438.
    Software in the Age of Claude Code(blog)
    LLM score 98 · 7 months ago
  9. Lewis Tunstall / Hugging Face Researcher
    2439.
    Codex's instruction following capabilities are both impressive and surprisingly fragile to typos :)›
    LLM score 92 · 7 months ago
  10. Kevin Lu / Thinking Machines Researcher
    2440.
    tinker as a sandbox for giving autoresearch access to RL training infra 🙂 https://t.co/C7pk2kSXtq
    LLM score 85 · 7 months ago
  11. Ben Burtenshaw / Hugging Face Researcher
    2441.
    les fucking go... for agents to push kernels to the hub, do:›
    LLM score 92 · 7 months ago
  12. Andrew White / Edison Scientific Cofounder
    2442.
    Most chiral molecules arise from carbons being bonded to 4 different atoms, which are called sterocenters.›
    LLM score 85 · 7 months ago
  13. Merve Noyan / Hugging Face ML Engineer
    2443.
    we just shipped Kernels, it's a new repo at @huggingface 💚›
    LLM score 75 · 7 months ago
  14. Ben Burtenshaw / Hugging Face Researcher
    2444.
    Here is a 20 minute lecture on how to setup automated machine learning with agents.›
    LLM score 85 · 7 months ago
  15. Sayak Paul / Hugging Face Researcher
    2445.
    We shipped a new repo type called "kernel" on the Hub.›
    LLM score 92 · 7 months ago
  16. Horace He / Thinking Machines Founding Engineer
    2446.
    As I've talked to folks about Thinky's purpose and why Thinky exists, I've found myself referring to The Intelligence Curse (https://t.co/FXMFkzHURE) time and time again.›
    LLM score 75 · 7 months ago
  17. John Carmack
    2447.
    FLOPS was originally “floating point operations per second”, specifying a rate of work for a system: A SPARCstation 2 gave 4.2 MFLOPS.›
    LLM score 92 · 7 months ago
  18. Cameron Wolfe / Researcher at Netflix
    2448.
    I just published a report on the LLM-as-a-Judge system used to evaluate show synopses on Netflix.›
    LLM score 95 · 7 months ago
  19. John Schulman
    2449.
    Luke and Rudolf's writing on keeping humans central in an AI-powered world sparked a lot of discussion at Thinking Machines.›
    LLM score 85 · 7 months ago
  20. John Carmack
    2450.
    Rhymes with @RichardSSutton’s Bitter Lesson.›
    LLM score 85 · 7 months ago
  21. Ben Burtenshaw / Hugging Face Researcher
    2451.
    rl for agents is moving fast with open source tools.›
    LLM score 85 · 7 months ago
  22. Lewis Tunstall / Hugging Face Researcher
    2452.
    ML interview question: you're implementing on-policy distillation and let's assume the student and teacher servers are already optimised for fast inference.›
    LLM score 92 · 7 months ago
  23. Ronak Malde / DeepMind Researcher
    2453.
    AI are replacing everything, what if they replace Operating Systems altogether?›
    LLM score 82 · 7 months ago
  24. Thomas Wolf / Hugging Face Cofounder
    2454.
    json is so token inefficient it hurts these days man, these braces and quotes are costing me real $$
    LLM score 82 · 7 months ago
  25. Lewis Tunstall / Hugging Face Researcher
    2455.
    We've rebuilt TRL's on-policy distillation trainer from the ground up to:›
    LLM score 92 · 7 months ago
  26. Merve Noyan / Hugging Face ML Engineer
    2456.
    MiniMax 2.7 is out 🔥 it sits in the open frontier in score per token efficiency 🥵 https://t.co/izzcB1I3vf
    LLM score 15 · 7 months ago
  27. Alex Zhang / MIT CSAIL PhD
    2457.
    ik this is an ai slop post but it contains a laundry list of misconceptions that are are quite nice to clear up lol›
    LLM score 92 · 7 months ago
  28. Eric Jang / ex VP of AI at 1X Robotics
    2458.
    just tried this with Claude code + computer use mcp on my Mac.›
    LLM score 92 · 7 months ago
  29. Skyler Miao / MiniMax Head of Engineering
    2459.
    M2.7 weights are live. hope you all enjoy it 😎 been grinding sleepless on M3 and the harness›
    LLM score 82 · 7 months ago
  30. Jerry Tworek / ex OpenAI VP of RL
    2460.
    Most interesting solutions come from looking at old problem from a new side rather than from hammering harder on old techniques
    LLM score 75 · 7 months ago
More