XOutlier
Download .zipAdd to Chrome

Nathan Lambert

@natolambert106k followers

Open model research, founder, director @trillium_labs. Prev. co-led Olmo at Ai2. Writes @interconnectsai, wrote

Open on X
Top outlier
193×
Best organic post
Peak views
3.2M
Single post
Bookmarks
38k
Across 101 organic posts
Last scan
Oct 16
Community cache
Posts 101 posts
#DatePost↓ OutlierViewsLikesRepostsBookmarks
1Jan 23, '25Meta is definitely not alone in this. And its normally overblown too.193×3.2M9.2k8082.9k
2Oct 21, '25Airbnb CEO Brian Chesky: “We’re relying a lot on Alibaba’s Qwen model. It’s very good. It’s also fast and cheap... We use OpenAI’s latest models, but we typically don’t use them th68.4×1.1M4.4k3351.4k
3Nov 17, '23I'm happy to announce I'm the next CEO of OpenAI and we're going to start doing open source again53.6×882k4k193184
4Aug 17, '25A tier list of China's top 19 open model builders. Who did we miss? At the frontier * DeepSeek * Qwen Close competitors * Moonshot AI (Kimi) * Zhipu / Z AI Noteworthy * StepFun * T34×559k1k154613
5Nov 22, '23TIN HAT TIME on what OpenAI is cooking with Q* RLHF To start, the hilarious quotes from the @Reuters article: "long-time executive Mira Murati told employees on Wednesday that a le31.4×518k1.1k140831
6Nov 24, '23The Q* hypothesis I can stand behind (from literature): 1. Tree of Thoughts reasoning: something to search over 2. Process reward models: rank all the steps of reasoning 3. GPT4 to21.8×359k649120512
7Apr 5, '25It’s very common for leadership at top labs to be in the know of other lab’s release schedules, so the simplest explanation to Meta releasing today is that next week is going to be20.9×343k1.2k36162
8Apr 5, '23Almost everyone I know working in AI these days feels one step away from total burnout. I took the time to take you behind the curtain and know what people on the state-of-the-art 20.5×338k684132372
9Jul 16, '25It is a major policy failure that the US cannot accommodate top AI conferences due to visa issues.20×328k1.2k154107
10Jun 25, '25A sign Google is waking up is them dropping by far and away the biggest free plans across the industry. This time is for the Claude Code competitor. Super excited, next time can be19.8×325k1.7k109709
11Sep 29, '25This is a way bigger deal than the Claude release.18.8×309k1.5k43305
12Sep 5, '25The weirdest VC subsidizing of our time, 10% of the Anthropic series F goes to writers17.6×290k3.5k148443
13Oct 25, '25A new essay on the crazy, all or nothing approach to work happening in AI today, the looming human costs, and the lack of a finish line. I wouldn't say it's okay, but I'm not sure 17.3×285k58650499
14May 13, '24did OpenAI tell me to downgrade to a free plan today?17.3×284k8552157
15Oct 12, '25Recurring frontier lab gossip: OpenAI has best post-training/rl and has pushed it super hard on weaker pretraining. Gemini has spectacular pretraining. Making a reasoning model was16.9×278k1.8k46587
16Jan 10, '25New export controls incoming, Bloomberg reporting: "But if an AI company wants to fine-tune a general-purpose open weight model for a specific purpose, and that process uses a sign15.4×254k806103381
17Oct 2Today we're unveiling Trillium Labs @trillium_labs, a new non-profit to foster the open science of frontier AI. We're building open post-training recipes and will expand into open 15.4×254k3.3k3021.1k
18May 22, '25Anthropic sliding into that code tooling company role instead of AGI race role14.8×243k2.1k73160
19Aug 17, '25Since everyone loved my Chinese lab ranking based on open model releases, here’s the equivalent for the Western companies based on released open models aligned in tiers to comparab14.5×238k87369387
20Apr 25, '25transparency in AI is dying: no evals, no release notes, just vibes and more bad naming14.3×235k2.7k114134
21May 22, '24If you're a student wanting an exciting life, good comp, and impact on the real world: work on LLMs13×214k51918208
22Dec 13, '24ILYA: "PRETRAINING IS DONE. WE ARE NOW IN THE POST TRAINING ERA."12.1×199k1.5k87213
23Apr 6, '25bruh what's on monday11.6×191k87034110
24Jul 30, '25New Zuck post, what a difference a few years makes: Today: "We'll need to be rigorous about mitigating these risks and careful about what we choose to open source." 2024: "Meta is 11.4×188k67361198
25Oct 20, '25Life update, she said yes. 🤩👩‍❤️‍👨🐕‍🦺10.8×178k3.4k1454
26Aug 9, '25In some ways the GPT-5 release feels like the Llama 4 release. They just waited too long to get it out. Feels like some weirdness may be happening behind the scenes. Messy release 10.7×177k7513879
27Feb 15, '25Glad to see DeepSeek team members writing more papers than just their fun tech reports :) (maybe I just missed them in the past)10.2×168k66879447
28Jun 5, '25In my recent trip, the waymo market in SF has converged to ~2-3x the wait time and ~2-3x the cost of uber because that's how much more people are willing to pay for Waymo.9.7×159k1.9k39159
29Jan 26, '25DeepSeek app sitting at number 1 overall in the US Iphone App Store is not on my bingo card and is the biggest sign yet that the ChatGPT moat can maybe be cracked.9.6×158k92175119
30Oct 30, '23Excited! I've started as a Research Scientist at @allen_ai (on @ai2_allennlp) working on all things "RLHF research" - it encompasses a lot. Let's show the world that openness can f9.2×152k6712644
31Oct 4, '23Life update: Today's my last day at @huggingface after 1.5 years. It's been an awesome ride, I'm moving within the open RLHF space to have a slightly more research oriented role, c9.2×151k77035179
32May 20, '25Most important plot from IO today -- AI usage is skyrocketing. This is real.9.1×149k1.6k123327
33Jul 4, '25My latest post: The American DeepSeek Project Build fully open models in the US in the next two years to enable a flourishing, global scientific AI ecosystem to balance China's sur8.8×145k62475145
34Apr 19, '25o3's search abilities are incredible. Can find extremely niche information without a ton of additional context. Just what I would say to a colleague.8.8×145k61223108
35Oct 31, '25I'm convinced to try it asap, we should all try fp16, look at this plot man. FP16 is like perfect in error reduction. "This is precisely why switching to FP16 provides a fundamenta8.8×144k64343332
36Apr 12, '24Watch my @stanford CS 25 lecture next week, "aligning open language models," it'll be good, v excited8.6×141k59186368
37Apr 6, '25> be me > be zuck > need llama 4 to land > send a model/prompt to LMSYS to get a top1 score, cringe be damned > release a different model as "open source" > think8.3×136k79324132
38Jul 18, '23Excited to share my analysis of the LLAMA2 model. In short, this model and paper are incredibly well done. Meta has stepped up the level for open-source and signaled another path f8.2×136k567127168
39Aug 4, '25America needs to take open models more seriously. This summer the early lead in open model adoption of the US via Llama has been overtaken by Chinese models. With The American Trul8.0×132k61499141
40Apr 8, '25vagueposting is the worst part of the ai community here8.0×132k2.6k4163
41Dec 20, '24o3 does not use different training nor inference methods than o1 (at least in pro mode). No special "search". OpenAI just found a hill and very quickly started hillclimbing it. Exc7.7×127k1.3k102452
42Oct 3, '25Who's using GPT-OSS and for what? Was it cheaper, better, faster than other open models? Or just not from China? Download numbers are actually very strong on HuggingFace for first 7.6×125k58636216
43Mar 24, '25Okay okay, spent my weekend gooning around learning GRPO math. Here's some takes. Essentially, this is me yapping through a recap of smaller details on how GRPO is implemented, wha7.5×123k1.4k1651.8k
44Nov 2, '25Claude 4.1 Opus > Claude 4.5 Sonnet7.3×120k7161269
45Apr 21, '25"Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?" This isn't a new intuition, but a nice new set of results. The paper in question 7.2×119k62985539
46Nov 21, '24I've spent the last two years scouring all available resources on RLHF specifically and post training broadly. Today, with the help of a totally cracked team, we bring you the frui7.2×118k787140624
47Jul 19, '25Not falling for OpenAI’s hype-vague posting about the new IMO gold model with “general purpose RL” and whatever else “breakthrough.” Google also got IMO gold (harder than mastering7.1×117k85635190
48Sep 7, '25I'm going to sound like a shill but I describe paying for better AIs right now as a way that you can "pay to win" in your career. Normally dynamics like this are restricted to vide6.8×111k80838194
49Feb 1, '25Since everyone wants to learn RL for language models now post DeepSeek, reminder that I've been working on this book quietly in the background for months. Policy gradient chapter i6.6×108k1.3k1631.6k
50Aug 8, '25I'm going to miss o36.5×107k1.1k4232
51Mar 21, '25Qwen 3 coming imminently! Meta's smart to have locked in LlamaCon, else Llama 4 maybe would've been delayed again 🤭. Really I'm hype for Llama 4, bring it asap.6.1×101k63553111
52Mar 13, '25A very exciting day for open-source AI! We're releasing our biggest open source model yet -- OLMo 2 32B -- and it beats the latest GPT 3.5, GPT 4o mini, and leading open weight mod6.1×100k920142339
53Sep 12, '25They're showcasing the RL to prod pipeline as a good form of continual learning. Totally shocking if you had told me this just 12 months ago.5.9×98k69039292
54Mar 3, '25If you look at most of the models we've received from OpenAI, Anthropic, and Google in the last 18 months you'll hear a lot of "Most of the improvements were in the post-training p5.8×96k59138406
55May 27, '25immortalizing this moment forever when RL is so easy that you can just use random rewards and your benchmarks still go up smh5.8×96k1.1k55498
56Sep 28, '25RL research is becoming like pretraining/modeling. This is a huge vibe shift. Most research published on RL isn't using enough compute to make many of these decisions matter as muc5.6×93k73170508
57Mar 1, '25Tbh I’m happily using GPT-4.5. thanks OpenAI for not being too eval obsessed5.6×93k6161428
58Sep 16, '25So, Dylan, where does Google land on this chart?5.6×92k57016141
59Oct 16, '25The first fantastic paper on scaling RL with LLMs just dropped. I strongly recommend taking a look and will be sharing more thoughts on the blog soon. The Art of Scaling Reinforcem5.6×92k1.2k1871.1k
60Oct 5, '25I gave a talk today at The Curve on the state of open models. Here are the slides, recording soon. Topics include: Chinese ecosystem, reflections on DeepSeek, the demise of Llama, 5.6×92k59477492
61Apr 16, '25First draft online version of The RLHF Book is DONE. Recently I've been creating the advanced discussion chapters on everything from Constitutional AI to evaluation and character t5.5×91k1.2k2071.2k
62Jun 18, '25Here's a recent talk I gave recapping the last 6-12 months of AI progress, why getting perfect models is hard, how labs are likely approaching the next phase of training (for agent5.4×90k1.1k1421.6k
63Feb 3, '25Stoked to get to talk to @lexfridman + my homie @dylan522p for 5+ hours to try and get to the bottom of what is actually happening in AI right now. DeepSeek R1 & V3, China v US5.4×90k63458125
64Oct 1, '24I'm flying so I'm working on my mini RLHF book today :) rlhfbook dot com5.3×88k93079897
65Jan 20, '25R1 making me feel very heard. Will read more thoroughly later. Laughs in continued shock that RL working like this.5.3×86k89042460
66Dec 20, '24OpenAI skips o2, previews o3 scores, and they're truly crazy. Huge progress on the few benchmarks we think are truly hard today. Including ARC AGI. Rip to people who say any of "pr5.2×86k6156090
67Jun 11, '25Major reasoning models so far with technical reports (focused on those w RL): 2025-01-22 — DeepSeek R1 — 2025-01-22 — Kimi 1.5 — 2025-03-31 — Open-Reasoner-Zero — 2025-04-10 — Seed5.1×84k1.1k1831.3k
68Sep 29, '25Thinking machines proving you can be worth $10B with your one product being great content.5.1×84k64525152
69Oct 1, '24going to go on the record and say this is a bad idea5.1×83k6261155
70Oct 29, '25I'd put good money on this being an high-impact finetune of one of the large, Chinese MoE models. I'm very excited to see more companies able to train models that suit their needs.4.9×81k78225125
71Jan 20, '25For those trying to understand DeepSeeks Group Relative Policy Optimization (GRPO): GRPO is just PPO without a value function using monte carlo estimates of the advantage. So, stud4.7×78k1.1k118945
72Nov 27, '24Qwen's first o1 inspired model. Blog: Model weights:4.6×76k656102337
73Mar 31, '25I hear people are pretty into GRPO and RL these days, so I wrote up a pretty comprehensive research survey of recent papers I liked. Kimi 1.5, OpenReasonerZero, DAPO and Dr. GRPO. 4.6×76k65991640
74Sep 6, '25I'm using GPT 5 Pro a lot. Mostly for research in my case, but I am bullish it or Gemini Deep Think are the smartest models available publicly today. You should use one or both of 4.6×76k4674676
75Jan 30, '25Very happy to show that we can do RL finetuning on 405B models with open-source code, beat Llama 405B instruct with their base model, and beat DeepSeek V3 too. Enjoy building off t4.5×74k65575246
76Apr 17, '25Props to Google for including O4-mini in their Flash 2.5 release. A model released *yesterday*, while some companies only compare to their own models. Looking good gemini.4.5×73k1.4k65154
77Jul 5, '25There are like 10-20 Chinese orgs shipping open models that I try and keep a somewhat close eye on and there are like 3-4 in the rest of the world 😳4.4×73k57237130
78Mar 26, '25Gemini 2.5 is amazing -- a bigger jump than the recent releases of Claude 3.7, Grok 3, and GPT 4.5. Google needs to have the same drive for excellence across the product and cloud 4.4×72k6142971
79Aug 15, '25GPT 5o has arrived.4.3×70k7291129
80May 11, '25leaders in ai talk like there's some master plan but it's literally just this4.2×69k1.9k103143
81Apr 7, '25Models that are actually really really good. Way better than what we were using in 2024: Gemini 2.5 Pro o1 pro4.1×68k65221108
82Jul 27, '25I bet pretty soon a Chinese research org drops a LLM scaling laws for RL paper. Closed frontier labs have definitely done this and wont share it, academics havent mastered the data4.1×68k73041177
83Jan 21, '25DeepSeek makes it quite clear how they trained R1. None of these steps alone are super surprising, but how to sequence and blend them together definitely is.3.9×65k60977400
84Aug 10, '25Anthropic is the only leading AI lab to not release a reasonable open weights model. Is notable that pretty much everyone has a touchpoint here now.3.9×64k6121742
85Jul 11, '25Modern licenses so funny. Amazing looking model. MIT-Modified. Marketing is king. "Our only modification part is that, if the Software (or any derivative worksthereof) is used for 3.8×63k58527133
86Feb 3, '25RL keeps cooking 🤣🤡🫡 “Deep research was trained using end-to-end reinforcement learning on hard browsing and reasoning tasks across a range of domains.”3.8×63k57439122
87Mar 27, '25Gemini 2.5 Pro long context goes hard. First model to take an entire paper latex (15k tokens), tell it to ignore comments, find all typos. Does it perfectly. Even o1pro didn't even3.5×58k76040124
88Feb 28, '25oh no3.5×57k7951331
89Jul 10, '25Grok 4 coming soon after Llama 4 with a completely different trajectory should help people finally take in how important culture is to progress in technology generally and AI speci3.4×56k7782668
90Jun 10, '25Zuck just buying Scale to cut off the competition from data obviously3.4×56k6491856
91Jan 20, '25hahahahah there were actually two technical reports for RL reasoning models today, kimi 1.5 also has good stuff on reward shaping + RL infra3.3×55k67375385
92Oct 17, '25New toy!3.3×55k7973148
93Sep 29, '24NotebookLM and OpenAI Advanced Voice Mode feel like we have entirely new ways we need to learn how to work with AI/LLMs again. Normally, when this happens we unlock a bunch of valu3.3×54k66035141
94Jun 20, '25Mastering the GRPO math, its implementation details, and other policy gradient algorithms has made it way easier to understand new research on reasoning algorithms. Read the policy3.1×50k73942577
95Mar 20, '25FINALLY!!!2.9×47k7673856
96Apr 16, '25TLDR on o3 and o4-mini: incremental. pace of progress still really high, no dramatic changes in performance or shocking new features. Pressure to ship fast across the industry has 2.8×47k6143268
97Jul 7, '22I put together all the interview timelines, reflections, and advice from my job search. I focused on RL jobs, varying from applied ML engineer to research scientist! Lot's of peopl2.7×0880121546
98Apr 18, '25rlhfbook also available on arxiv for SEO 😀 happy friday2.4×39k68679405
99Sep 14, '25Getting ready to invest more time into the RLHF book to prepare for print edition. What do you wish was clearer or had more coverage in it?2.3×38k57949567
100Jun 16, '21In a world before derivatives, Fermat solved optimization by assuming two converging points are not equal (so they don’t divide by 0 in algebra) but are approximately 0 so can be s1.2×06189271
101Aug 30, '25Got the essentials, not much more is needed.0.9×14k222641

Scan another creator

@
@karpathy@levelsio@paulg@naval@marclou@illyism

Run this on any account.

XOutlier builds this table for every public profile you open on x.com, with live data and every post, not just the top 50.

Add to Chrome Download .zip