AI300

Siyan Zhao / UCLA PhD
1861.
Thanks AK for sharing our work! Unlike autoregressive LLMs, diffusion LLMs can be conditioned on future reasoning hints during generation through inpainting 🧩, enabling guided exploration toward correct solutions.›
LLM score 92 · 9 months ago
Noam Brown / OpenAI Research Scientist
1862.
When we at @OpenAI released o1-preview a year ago, it would think for seconds.›
LLM score 75 · 9 months ago
Noam Brown / OpenAI Research Scientist
1863.
@isthisnessicary @emollick Some evals are harder to beat even with targeted data.›
LLM score 70 · 9 months ago
Noam Brown / OpenAI Research Scientist
1864.
@Qutossar @emollick Maximizing this benchmark would probably be useful for improving a model's ability to read clocks.›
LLM score 75 · 9 months ago
Noam Brown / OpenAI Research Scientist
1865.
@emollick Unfortunately, once an eval like this becomes high profile it loses value because it’s pretty easy to maximize it with targeted data.›
LLM score 80 · 9 months ago
Shuchao Bi / Meta Researcher
1866.
Similar to the AMC12 grading rule (6 points for each correct answer, 0 for incorrect answer, and 1.5 for unanswered question), we should give points (>= p(random guess is correct), 1/1000 for AIME) for unanswered questions in benchmarks and during RL.›
LLM score 70 · 9 months ago
Noam Brown / OpenAI Research Scientist
1867.
@sriramk Though by "long time" I mean 3+ years.›
LLM score 20 · 9 months ago
Noam Brown / OpenAI Research Scientist
1868.
@sriramk I think these models will quickly improve at verifying their own output, but I agree that AI will be worse than humans in some ways for a long time.›
LLM score 70 · 9 months ago
Noam Brown / OpenAI Research Scientist
1869.
@erikbryn There are real limitations to what AI models can do today, but I think it’s important to consider the slope of progress.›
LLM score 70 · 9 months ago
Noam Brown / OpenAI Research Scientist
1870.
@erikbryn From what I've seen, a lot of critics don't have a good understanding of where the frontier really is.›
LLM score 65 · 9 months ago
Noam Brown / OpenAI Research Scientist
1871.
@emollick Also, those forecasts were for *any* AI system to get an IMO gold.›
LLM score 70 · 9 months ago
Noam Brown / OpenAI Research Scientist
1872.
@conitzer Intro to ML introduces a lot of concepts that are built upon in future ML classes, so I think it makes more sense as the intro class.
LLM score 40 · 9 months ago
Noam Brown / OpenAI Research Scientist
1873.
@rishicomplex I've spoken with profs at a lot of universities about this and almost all agree Intro to AI should cover more ML.›
LLM score 60 · 9 months ago
Tri Dao / Chief Scientist at Together
1874.
Clarification: I was comparing A @ B + C here, where the cute-dsl version is quite good at overlapping the epilogue.›
LLM score 85 · 9 months ago
Sheryl Hsu / OpenAI Researcher
1875.
We officially entered the 2025 International Olympiad in Informatics (IOI) online competition track and adhered to the same restrictions as the human contestants, including submissions and time limits, but without direct supervision from the contest organizers.
LLM score 70 · 10 months ago
Alexander Wei / OpenAI Researcher
1876.
3/ We’ve come a long way since last summer.›
LLM score 70 · 10 months ago
Sheryl Hsu / OpenAI Researcher
1877.
5/n It’s been really exciting to see the progress of our newest research methods at OpenAI, with our successes at the AtCoder World Finals, IMO, and IOI over the last couple weeks.›
LLM score 20 · 10 months ago
Sheryl Hsu / OpenAI Researcher
1878.
4/n This result demonstrates a huge improvement over @OpenAI’s attempt at IOI last year where we finished just shy of a bronze medal with a significantly more handcrafted test-time strategy.›
LLM score 80 · 10 months ago
Juntang Zhuang / MTS at xAI (pre-training lead)
1879.
It’s extremely fun though tough to train the first natively multimodal model ever in xAI.›
LLM score 92 · 10 months ago
Geoffrey Hinton
1880.
A major cut to the funding of the National Science Foundation would be very bad for the future of the US.›
LLM score 20 · 10 months ago
Ted Sanders / OpenAI Researcher
1881.
a cool thing you get to see building AI products: ›
LLM score 75 · 10 months ago
Ted Sanders / OpenAI Researcher
1882.
GPT-5 is here! it's way better at coding - not just in pointless evals, but real usage.›
LLM score 70 · 10 months ago
Jeremy Bernstein / Thinking Machines Researcher
1883.
I had wondered why there was no official Dion implementation by the authors...›
LLM score 75 · 10 months ago
Sally Zhu / Researcher at Flapping Airplanes
1884.
Interesting that you need the teacher + student to share the same base model; reminds me of linear mode connectivity stuff.›
LLM score 85 · 11 months ago
Tri Dao / Chief Scientist at Together
1885.
Hierarchical layout is super elegant.›
LLM score 85 · 11 months ago
Jason Wei / AI Researcher at Meta
1886.
Becoming an RL diehard in the past year and thinking about RL for most of my waking hours inadvertently taught me an important lesson about how to live my own life.›
LLM score 60 · 11 months ago
Jason Wei / AI Researcher at Meta
1887.
New blog post about asymmetry of verification and "verifier's law": https://t.co/bvS8HrX1jP›
LLM score 80 · 11 months ago
Jakub Pachocki / OpenAI Chief Scientist
1888.
I am extremely excited about the potential of chain-of-thought faithfulness & interpretability.›
LLM score 80 · 11 months ago
Lilian Weng / Thinking Machines Cofounder
1889.
I still find it mysterious whether and how intelligence and capabilities transfer between domains and skills - from meta learning during early days to more recent question like whether solving maths helps writing a good essay.›
LLM score 75 · 11 months ago
Tri Dao / Chief Scientist at Together
1890.
I played w it for 1h. Went through my usual prompts (math derivations, floating point optimizations, …).›
LLM score 35 · 11 months ago