Liang Qiu (@aliangment) header
Liang Qiu (@aliangment) profile picture
Joined November 2016
Photo by Liang Qiu (@aliangment)Photo by Liang Qiu (@aliangment)Photo by Liang Qiu (@aliangment)Photo by Liang Qiu (@aliangment)Photo by Liang Qiu (@aliangment)
Something not studied here: the joy human researchers get from working on a problem purely out of curiosity, and from collaborating with others.
New research result: we use Claude to make fully autonomous progress on scalable oversight research, as measured by performance gap recovered (PGR). Claude iterates on a number of different techniques and ends up significantly outperforming human researchers for $18k in credits.
2
3
1
22
3,903
It’s in production! 🎉🇬🇧
178
724
53
14,008
433,111
Liang Qiu retweeted
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use. Its capabilities exceed those of any model we’ve ever made generally available.
4,909
14,063
9,954
103,212
57,743,529
Liang Qiu retweeted
Claude Opus 4.8 showing up the Agent SDK published to npm 20 mins ago: Claude Opus 4.8 — The most capable Claude model to date. Highly autonomous and state-of-the-art for long-horizon agentic work, knowledge work, and memory, with clearer, warmer writing. Uses the same API surface as Opus 4.7: Adaptive thinking only Sampling parameters removed budget_tokens removed 1M context window at standard API pricing No long-context premium See shared/model-migration.md → Migrating to Opus 4.8 A 4.7 → 4.8 migration is a model ID swap plus prompt re-tuning, with no new breaking changes.
40
20
13
373
76,933
Liang Qiu retweeted
Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.
7,849
10,879
7,550
148,701
28,113,535
Liang Qiu retweeted
New post: "Generalization Dynamics of LM Pre-training" Most people (including me) assume that LMs smoothly mature from pattern-matching to generalizing. This mental model is wrong. The true dynamics are stranger, and far more fascinating! We call it Mode-Hopping.
11
82
19
560
102,480
Liang Qiu retweeted
/radio
318
843
258
14,369
1,043,881
New work with @AlecRad and @DavidDuvenaud: Have you ever dreamed of talking to someone from the past? Introducing talkie, a 13B model trained only on pre-1931 text. Vintage models should help us to understand how LMs generalize (e.g., can we teach talkie to code?). Thread:
179
395
173
3,193
1,243,031
Liang Qiu retweeted
New Anthropic Fellows research: developing an Automated Alignment Researcher. We ran an experiment to learn whether Claude Opus 4.6 could accelerate research on a key alignment problem: using a weak AI model to supervise the training of a stronger one. anthropic.com/research/autom…
226
277
113
2,348
455,303
Liang Qiu retweeted
Introducing Project Glasswing: an urgent initiative to help secure the world’s most critical software. It’s powered by our newest frontier model, Claude Mythos Preview, which can find software vulnerabilities better than all but the most skilled humans. anthropic.com/glasswing
1,926
6,503
4,419
43,353
32,028,693
Liang Qiu retweeted
Replying to @wongmjane @BenLesh
Mistakes happen. As a team, the important thing is to recognize it’s never an individuals’s fault — it’s the process, the culture, or the infra. In this case, there was a manual deploy step that should have been better automated. Our team has made a few improvements to the automation for next time, a couple more on the way.
312
813
363
10,884
1,421,853
Liang Qiu retweeted
📰 Excited to share the results of this Anthropic Fellows project! We built an automated alignment agent (A3) for safety finetuning. The project was great fun, and @jifan_zhang was an amazing collaborator to work with - excited to see what he does next!
Sharing my last fellowship work at @AnthropicAI! We are open-sourcing an automated alignment agent for safety and model behavior finetuning. It expedites the timeline in fixing a model behavior issue, and alleviates the pain for researchers.
1
15
4,956
Liang Qiu retweeted
New in Claude Code: Remote Control. Kick off a task in your terminal and pick it up from your phone while you take a walk or join a meeting. Claude keeps running on your machine, and you can control the session from the Claude app or claude.ai/code
1,887
4,523
3,771
43,906
10,163,276
Liang Qiu retweeted
Replying to @Kimi_Moonshot
@Kimi_Moonshot used policy mirror descent (PMD) for RL in Kimi k1.5/k2. Most take it simply as PG+KL with an updating anchor, but this is not the full story. Check our blog for some interesting findings about this algorithm: zhenghaoxu.notion.site/Revis…
1
12
1
38
6,975
Lost for words 🇦🇺
11,879
24,006
3,286
238,519
5,705,408
Liang Qiu retweeted
Demis Hassabis on the final frontier for AI in science 🧬 : it’s not just about solving the math, it’s about having the creativity to ask the question in the first place. While coding and math are verifiable and easier to automate, natural science requires a "missing ingredient" that we haven't fully cracked yet. "Coming up with the question in the first place or coming up with the theory or the hypothesis. I think that's much, much harder and I think that's the highest level of scientific creativity.”
14
19
2
249
12,899
Liang Qiu retweeted
New Anthropic Fellows research: the Assistant Axis. When you’re talking to a language model, you’re talking to a character the model is playing: the “Assistant.” Who exactly is this Assistant? And what happens when this persona wears off?
307
564
297
5,110
1,309,586
Liang Qiu retweeted
looking for a research internship! realized I only know a few industry teams, so I'm reaching out more broadly. I'm interested in - understand LM generalization - scalable oversight / human-AI collaboration in production settings (e.g. coding agents) - transfer superhuman knowledge from AI to humans happy to learn new topics that you find exciting DM me if you think I'd be a good fit for your team!
32
33
734
74,457
We’re opening applications for the next two rounds of the Anthropic Fellows Program, beginning in May and July 2026. We provide funding, compute, and direct mentorship to researchers and engineers to work on real safety and security projects for four months.
107
336
55
3,467
1,177,134
I told my discrete mathematics students about Ramanujan discovering his famous formula for π in a dream and a student in the front row muttered "Dude that's a frickin nightmare"
28
69
8
1,141
37,715