Some thoughts on our eng blog on agent RAG over sci literature: 1/3 A big finding is that the strong effect of chunk size, parsing quality, and embedding choice vanishes with LLM rerank/context summary (RCS). It's slower and more compute, but it makes the systems very robust.

One week from today, the Nobel Prize in Medicine/Physiology will be announced. Here are the 79 most likely awardees, each of whom has won two or more pre-Nobel “predictor” prizes:

life goals: give an interview in the style of Warren McCulloch (the interview itself is also partially worth watching, in particular from min 6:30 or so where he presents an early version of AI-doomer-optimism, albeit fully clothed) youtu.be/wawMjJUCMVw?si…

We’ve just finished writing the missing 15,616 Wikipedia articles to get complete coverage of all 19,255 human genes. We used PaperQA2, which has higher accuracy than existing human-written Wikipedia articles, as judged by blinded biology PhD students and postdocs. 1/5

All languages covey information at a similar rate when spoken (39bits/s). Languages that are spoken faster have less information density per syllable! One of the coolest results in linguistics.

Today, I reached a key point in my tour: speaking at the 4th Young Israeli CRS meeting in Tel Aviv University. Being in Israel now holds special meaning for me. After the Hamas attack, I felt compelled to support Israel and speak against antisemitism. Meeting Israeli scientists and

Alexander Kabanov Tel Aviv University Inspiring talk by Alexander Kabanov about his path in developing polymers for drug delivery Controlled Release Society Israel Tel Aviv University Ronit Satchi-Fainaro Assaf Zinger Zvi Yaari

A huge honor to host the Kabanovs ( Marina & Alexander Kabanov) here at Technion Israel sharing life and professional tips with tour students. You are true friends!🙌🏻 Controlled Release Society Young Scientist Committee YSC-CRS Global Young Academy Lindau Nobel Laureate Meetings

Excited to have visited and spoken at Technion this Monday. Phenomenal new faculty with programs in advanced materials, polymers, AI, and drug delivery. Highlight: 1.5-hour Q&A with students, postdocs, and faculty on making a difference in science, creativity, careers, and

You and Marina were such an inspiration! The students felt the ❤️ and support and really appreciate it. It was really meaningful for them and for us as well! Thank you for being such an amazing person!!

END OF YEAR TOP ALL CANCER TRIALS in 2024! JUST in on 1 JAN 2025: Paper 1: #neoadjuvant checkpoint inhibitor Phase 3 trial in melanoma! #NADINA trial shows ipi+nivo significantly improves EFS vs. adjuvant nivo alone (HR 0.32, p<0.0001). #ASCO24 plenary #Melanoma ASCO

We are working on a meta-analysis of single-cell data. We found a paper relevant to our work with 'data available upon request' in a good journal. We contacted the authors (European group, ERC funded) 8 months ago. After weeks of back-and-forth with their legal we got this

This is crazy!!

#GI25 Is it the hardest to pronounce name of any antibody? Yes. It is also probably the best new treatment for GI malignancies? Also, probably yes. Daraxonrasib, pan-KRASi Amazing responses in second line PDAC Making the undruggable druggable. (Someone please teach me how to

my unfinished paper watching me scrolling twitter

What is considered to be the holy grail of biology—the virtual cell—now has potential for being actualized. erictopol.substack.com/p/the-holy-gra…

Some perspective amidst the news

When 1+1 > 2 ✌️ In this Analysis, we show that (1) nanomedicines are useful to improve the efficacy of multi-drug cancer therapy, and that (2) co-delivering two drugs in one NP produces better outcomes than co-administering two single-drug-loaded NP 👉 rdcu.be/el86N

My lab at Weizmann Institute was completely demolished by the Iranian missile a few days ago. This is hard and sad as nothing was left to save. We study heart regeneration and regenerative medicine. As such, we will regenerate, regrow and rebuild. The one thing that wasn't gone is

When we came to see our destroyed labs hit by an Iranian missile, we found that a piano had survived Weizmann Institute