Ass prof @DondersInst/@KachmanLab. Simulation and Coffee abuser. Ex:@TechnionLive-)@MIT-)@IBMResearch-)Rhizome-)@AQRCapital--)@AI_Radboud spicy opnions are mine

Joined August 2014
463 Photos and videos
Pinned Tweet
After a suggestion from @LucaAmb a thread on the books I used when moving from being a #MachineLearning researcher in tech to being a ML researcher in #finance. A thread ! 1/n (I might continue updating it as more things come to mind)
8
12
42
We have just advanced from graduate student descent to parallel tempering graduate Monte Carlo. No way am I going back!
loops around agents? this is silly, i just instruct my agent to call itself with a followup task when its done
2
92
When I was doing #moleculardynamics a decade ago, we had loops of Slurm bash scripts that were in charge of starting, restarting, and managing different runs. The OG agent setup gets rediscovered
Just landed nested subagent support in Claude Code Starting to experiment more with agents kicking off agents as a way to better manage context. Capped at depth=5 to start, going out in today’s release. Lmk what you think!
1
69
All the insanity of loops recently, it's just a question of time until someone says, Mabius strip is the new loop LLM vibe coding paradigm.
Everyone keeps asking “what is your loop” but nobody’s asking “how is your loop”
102
Tal Kachman ✈️ somewhere retweeted
הצטערתי מאוד לשמוע את אבישי בן חיים אמש בערוץ 14. שני דברים: ראשית מנחם בגין לא היה זקוק "להגנה", זה חלק מבורות מוחלטת בהבנת האירועים ומאגדה אורבנית אווילית כאילו "השמאל שבר אותו". לא 40 מפגינים תחת ביתו ולא 4000 ולא 400 אלף היו יכולים לשבור אדם זה שחמק מציפורני הנאצים שהשמידו את משפחתו, שעבר את חקירות הנ.ק.ו.ד הידוע לשמצה של סטאלין והוגלה למות בערבות הקרח של פצ'ורה שם אמר לו השומר בכניסה "מפה לא יוצאים". האדם שיצא והגיע לפלסטינה המנדטורית להילחם באימפריה הבריטית להקמת מדינת היהודים, עבר את רדיפות הסזון, את ירי אש התופת על האלטלנה, את המדבר הפוליטי זרוע המוקשים, את ההשמצות, הנידוי, ההדרה, את דיכוי נאמניו עד הגיעו לשלטון עימהם. ה ו א - לא היה זקוק "להגנה". מר בן חיים, מנחם בגין נולד בברזל ומת בברזל ואינך יודע דבר וחצי דבר על נסיבות פרישתו כמו שאר הציבור. שנית סברתי כי שנתיים וחצי של שתיקה היו עדות למסע פנימי של שינוי, אמנם תרמת רבות גם אתה לבטחון ישראל בשרותך המבורך במילואים, אך צר לי, באמת צר לי שאדוני סבור שראש הממשלה שמסית ומדיח כנגד ראשי מערכת הביטחון שנכשלו בשבעה באוקטובר ומפעיל מערכת שלמה של שקרים זדוניים, מופרעים אודות מרידה ובגידה שהגיעו כבר לכל שופטי העליון במסע דה לגיטימציה מוחלט ואינו לוקח ולו קמצוץ אחריות אודות מאורעות היום הנורא וכל השנים שקדמו לו ומסרב להקים ועדת חקירה ממלכתית בכדי להגיע לשורש המחדלים הביטחוניים והמדיניים הפך בעינך למנהיג כה דגול. צר לי מאוד. 64 שניות על מנחם בגין ואחריות. @AvishayBenHaim
108
290
2,133
65,382
How to learn cpp in 21 days
was reminded by this classic today: "Teach Yourself Programming in Ten Years" by Peter Norvig Why is everyone in such a rush?
5
326
"As a boomer cognitive scientist, when I first saw the revival of #reinforcementlearning, I was horrified, didn't we get rid of this in like 1965? Do we all have to wear bell-bottom pants now?" Words to live by. Fantastic talk by @AlisonGopnik ! youtu.be/xJec77L6bUE?t=673
69
I was sure it would be Jetson, the guy is too humble
found jensen huang’s car in the mission
1
2
238
The internet has won today
שינינו את המשוואה מול איראן: הצפון הוא בגדר נעלם, אנחנו פראיירים בריבוע ונתניהו הוא אפס
30
I still hand prompt students that hand prompt codex like god intended
I still hand prompt codex like god intended.
2
103
for eps in np.random.randn(): prompt=prmpt eps
what the fuck does that even mean
145
Epic mentors will always be epic mentors. Every time I have to talk to one of my students, I count my blessings. I was lucky enough to be mentored by Jeremy England.
A book and a letter that changed my life. After I read The Geometry of Biological Time, I wrote to its author and asked if I could come work with him. His reply, scrawled in his characteristic magic marker, led to his becoming my most important mentor.
1
98
Tal Kachman ✈️ somewhere retweeted
bored at the airport so i made this killedbygpt.com
19
7
95
18,620
Shout out to @david_i_spivak, amazing book on Category Theory for the Sciences
Warning: once you learn category theory, you'll never be able or willing to talk with people who don't know category theory.
1
141
This trend is massively worrying. Words matter, and terms matter. This is not a paper; it's a position piece. No peer review, nothing is open-sourced or reproducible, which is all great. For a position piece, not a scientific paper. And of course 25% self citation ...
Another great paper from Google. Shows general LLMs can solve formal math by planning proofs and checking each step. Raised general LLM performance from under 10% to 70%. A general LLM failed badly when asked to write full formal proofs in 1 try, but became much stronger when it planned, split the work into smaller claims, reused past claims, and learned from Lean’s feedback. The paper shows the weakness was not just the model’s math ability, but the way it was being used - the absence of structured interaction with a verifier. The key idea is that the model does not try to write one giant perfect proof at once, because that usually fails on long and tricky problems. Instead, LEAP stores the proof as a graph of goals and subgoals, so useful lemmas can be reused instead of rediscovered every time. The authors tested LEAP on Putnam 2025 and a new Lean benchmark built from 60 IMO-style problems, where ordinary one-shot proof writing did very poorly. LEAP solved all 12 Putnam 2025 problems and raised general LLM performance on the Lean IMO benchmark from under 10% to 70%. ---- Link – arxiv. org/abs/2606.03303 Title: "LEAP: Supercharging LLMs for Formal Mathematics with Agentic Frameworks"
2
1
11
1,950
Please listen to this, Avi is one of the most special researchers I have ever heard. His book is a masterpiece, truly and wholeheartedly
Avi Wigderson is the only person in history to have won both a Turing Award (computer science) and Abel Prize (math). I interviewed him all about his field. We discussed: • His intuition on a proof of P vs NP • Why we use SAT solvers for most NP problems • Zero knowledge proofs and their impact • Quantum computation and implications • Math and computer science's relationship Where to watch: • YouTube: youtu.be/5GUcvSAJcJw • Spotify: open.spotify.com/episode/4JZ… • Apple Podcasts: podcasts.apple.com/us/podcas… • Transcript: developing.dev/p/turing-awar… Thank you to this episode's sponsors for supporting my work: • WorkOS: makes your app Enterprise Ready with easy to use APIs to add SSO, SCIM, RBAC, and more in just a few lines of code, check them out at workos.com/ Timestamps: 00:00 - Intro 01:08 - P vs NP 14:51 - What if you relaxed correctness 25:38 - Why NP complete problems are equivalent 30:33 - Space vs time complexity 43:06 - Why people use SAT solvers 45:53 - Randomness is a resource 55:48 - Randomness depends on computational power 01:21:20 - Zero knowledge proofs and their significance 01:38:30 - Quantum computation and why it matters 01:56:24 - Math vs computer science 02:08:16 - Major breakthroughs and his experience 02:12:31 - Advice for his younger self 02:14:48 - Outro
2
206
Tal Kachman ✈️ somewhere retweeted
Lot of QTs about "IB-ification" and how uncool that is and I agree. We've been resisting this by not giving intern candidates exploding offers and will continue to resist it. Thanks to everyone who reached out in this situation. Our listings will go up in a month or so.
Heard that some firms already making QR intern offers for summer 2027. I'd suggest if you're in that situation, first of all don't let them pressure you. Second, if you're a "rising junior" and you want an HRT offer, get in touch. Willing to talk early for exceptional candidates.
4
1
200
30,166
Tal Kachman ✈️ somewhere retweeted
Heard that some firms already making QR intern offers for summer 2027. I'd suggest if you're in that situation, first of all don't let them pressure you. Second, if you're a "rising junior" and you want an HRT offer, get in touch. Willing to talk early for exceptional candidates.
20
7
561
269,198
Tal Kachman ✈️ somewhere retweeted
[AI] Probabilistic Tiny Recursive Model A Sghaier, A Parviz, A Jolicoeur-Martineau [Mila – Quebec AI Institute] (2026) arxiv.org/abs/2605.19943
6
38
2,016
I feel seen
Your PI juggling their PhD students managing their research projects.
2
155
So...... Neurips 40000 papers ?
2
2
760