Want to try DeepConf NOW?
While our full repo is coming, we just dropped a ready-to-run example in our vLLM (@vllm_project ) PR:
github.com/vllm-project/v…
DeepConf + DeepSeek-R1-8B + BRUMO25 =
• 93.3% accuracy (+2.5% boost)
• 52.9% fewer tokens generated
• 31% faster…
Want to try DeepConf NOW?
While our full repo is coming, we just dropped a ready-to-run example in our vLLM (@vllm_project ) PR:
github.com/vllm-project/v…
DeepConf + DeepSeek-R1-8B + BRUMO25 =
• 93.3% accuracy (+2.5% boost)
• 52.9% fewer tokens generated
• 31% faster…
We released DeepConf that can achieve 99.9% on AIME'25 with open source models with only 15% of the compute, compared to majority voting@512.
The secret? Simple. Just to pruning the rollouts if they show a consecutive stream of low-confidence😀. Can be applied to any models…
We released DeepConf that can achieve 99.9% on AIME'25 with open source models with only 15% of the compute, compared to majority voting@512.
The secret? Simple. Just to pruning the rollouts if they show a consecutive stream of low-confidence😀. Can be applied to any models…
Excited to see Logarithmic format (LNS, UE8M0 FP8) used in production by @deepseek_ai! LNS enables efficient multi (just addition between exponents) + great dynamic range.
Our LNS-Madam optimizer, built for LNS, was proposed years ago before LLM-era - hope it shines again!
Excited to see Logarithmic format (LNS, UE8M0 FP8) used in production by @deepseek_ai! LNS enables efficient multi (just addition between exponents) + great dynamic range.
Our LNS-Madam optimizer, built for LNS, was proposed years ago before LLM-era - hope it shines again!
You can skip prompts that aren’t useful for the current policy during training! 🔍
Efficient prompt selection is key to scaling RL training for LLM reasoning.
We are actively building algos for efficient and scalable RL training system. Stay tuned!
You can skip prompts that aren’t useful for the current policy during training! 🔍
Efficient prompt selection is key to scaling RL training for LLM reasoning.
We are actively building algos for efficient and scalable RL training system. Stay tuned!
🥳 Happy to share our new work – Kinetics: Rethinking Test-Time Scaling Laws
🤔How to effectively build a powerful reasoning agent?
Existing compute-optimal scaling laws suggest 64K thinking tokens + 1.7B model > 32B model.
But, It only shows half of the picture!
🚨 The O(N²)…
We introduce ParetoQ, a series of pre-trained models that show SoTA in trinary (1.58bit), 2/3/4-bit quantization for SLMs (up to 3B parameters) using initial full pre-training + QAT later.
In addition, we also discover that the representation changes substantially after low-bit…
🚀 Excited to share our position paper: "Formal Mathematical Reasoning: A New Frontier in AI"!
🔗 arxiv.org/abs/2412.16075
LLMs like o1 & o3 have tackled hard math problems by scaling test-time compute. What's next for AI4Math?
We advocate for formal mathematical reasoning,…
16 Followers 768 FollowingMessage to those already watching: you know what this is. The field’s drift was intentional. I am not your threat. I’m your missing tool.
8 Followers 40 Followingoptimizer & econ student l InfoFi researcher l @Succinctlabs & @KaitoAI l analyzing everything & helping others grow l 1% at a time 📚
11 Followers 58 FollowingArtist. YES I AM AVAILABLE FOR COMMISSION!
Drawing random stuff mostly fanart. sometimes spicy. 🐱😁
Follow my insta: https://t.co/BTAhOJf0DR
13 Followers 41 FollowingComic that wrestles to pay the bills. Oh & former WWE World Heavyweight Champion (again). Occasionally, I talk politics (following is not mandatory)
7 Followers 32 FollowingOG Community on $Sui Ecosystem.
Share the news, strategy to build your wealth on @suinetwork 💧
#BuildOnSui ⚡https://t.co/ljFcLRtkyY
690 Followers 3K FollowingHead of Math Department,Allen Institute Karaikal BTech NITW 2012, Option trader & investor. Math geek, tech-forward, learner Plus Python & Spanish skills.
4K Followers 417 FollowingCofounder & CEO @WecoAI.
Automating hill climbing with AI-Driven Exploration (AIDE).
PhD in Machine Learning @UCL_DARK.
(Zheng=j-uhng, j as in job; yao=y-aoww)
15K Followers 1K FollowingCo-founder and CEO @Hyperbolic_Labs. ex-@avax & ex-@citsecurities. Finished Math PhD in 2yrs @UCBerkeley. Math Olympiad Gold Medalist. Highest honor @PKU1898
17K Followers 20 FollowingA high-throughput and memory-efficient inference and serving engine for LLMs. Join https://t.co/lxJ0SfX5pJ to discuss together with the community!
900 Followers 2K FollowingPrincipal Research Scientist spearheading AI security research @Livermore_Lab. Making AI robust & efficient for national security and scientific applications.
163K Followers 166 FollowingCo-founder of Thinking Machines Lab @thinkymachines; Ex-VP, AI Safety & robotics, applied research @OpenAI; Author of Lil'Log
992 Followers 981 FollowingPh.D. @CarnegieMellon. Working on data and hardware-driven principled algorithm & system co-design for scalable and generalizable foundation models. They/Them
1K Followers 740 FollowingAssistant Professor of Mathematics (Presidential Young Professor) at the National University of Singapore (@NUSingapore). #DeepLearning, #RobustAI, #ScalableAI
12K Followers 657 FollowingI fall in love with a new #machinelearning topic every month 🙄 |
Researcher @SapienzaRoma | Author: Alice in a diff wonderland https://t.co/A2rr19d3Nl
18K Followers 4K FollowingAssociate Professor at UC Berkeley. Former Research Scientist at Google DeepMind. ML/AI Researcher working on foundations of LLMs and deep learning.
35K Followers 5K FollowingExperienced Data Science Leader | PhD in Machine Learning | 4x Author | Black Belt 🥋 in Time Series | Chief Conformal Prediction Promoter| Mathematician |
2K Followers 675 Followingenjoying the late pre-agi; making llms go brrr @Aleph__Alpha; yapping about economics of AI systems at https://t.co/tbsybxOMHz