@GalChechik

A Sr director of AI research at NVIDIA and a CS Prof. at Bar-Ilan U. I study AI/ML reasoning and perception.

chechiklab.biu.ac.il/
Joined July 2018
  • Tweets205
  • Following403
  • Followers1K
6Media
Gal Chechik retweeted
Uri Berger@uriberger88Oct 13
Do vision-language models look at the image regions they describe? In our new paper, we present Self-Saliency, a method for aligning visual attention with reasoning text. research.nvidia.com/labs/par… w/ @GalChechik @DalalGal

Self-Saliency: Look Where You Say You're Looking

Grounding a VLM's own reasoning chain, and rewarding it for attending to the regions it just described.

research.nvidia.com
4
22
119
3,899
Gal Chechik retweeted
Ido Greenberg@Ido58861641Oct 6
[1/4] EARLI is out in Nature's Communications AI & Computing! EARLI combines learning and genetic algorithms to accelerate routing solvers like cuOpt and HGS, producing high-quality routes in seconds for hundreds of delivery locations.
3
5
8
322
Gal Chechik retweeted
Ido Greenberg@Ido58861641Oct 6
[4/4] Paper: nature.com/articles/s44488-0… Code: github.com/NVlabs/EARLI Data: gitlab.com/igreenberg/olist_… And a big shoutout to the team: Piotr Sielski, @hugolinsen, Rajesh Gandham, @shiemannor, Alex Fender, @GalChechik and @eli_meirom!

Reinforcement learning for initializing genetic algorithms in vehicle routing

Communications AI & Computing - Reinforcement learning for initializing genetic algorithms in vehicle routing

nature.com
2
1
2
140
Gal Chechik retweeted
NVIDIA AI
@NVIDIAAISep 30
We’re looking for our next cohort of graduate fellows. Our 2027–2028 Graduate Fellowship Program is open to Ph.D. students worldwide with awards up to $60,000, plus mentorship and technical support. Apply by October 30: nvda.ws/4rGKDhv
🤖 Made with AI
10
108
8
595
116,561
Gal Chechik retweeted
Or Litany@orlitanyOct 2
Meet DyRAD: Radar NVS for dynamic driving scenes. 📡🚗 Radar directly senses motion through Doppler, but prior radar NVS methods ignore this signal where it matters most: dynamic scenes. 🌐 dyrad-nvs.github.io 📄 arxiv.org/abs/2609.39841 💻 github.com/Dyrad-NVS/DyRAD 🧵
1
20
29
879
Gal Chechik retweeted
TLV Partners
@TLV_PartnersSep 9
Ahead of Rosh Hashanah, we're publishing the TLV AI 50: a list of the 50 most influential Israeli AI researchers and founders. Their ideas shape how modern models think, see, and reason, and their companies turn those ideas into the chips, companies, and products the world runs on. Full list: eu1.hubs.ly/H0y9XRZ0 30 researchers, 20 founders, 7 papers. This is not a ranking, but rather a representation of the most impactful Israelis in the AI research realm. Here’s a few trends that stood out: •⁠  ⁠Israeli neolabs are finally emerging at the model layer •⁠  ⁠AI security: a new cohort of AI natives is defining how systems get hardened and verified - Israel is firmly at the frontier •⁠  ⁠Vision: some of the most cited labs anywhere, with startups now harnessing that talent for physical AI Also featured are 7 noteworthy papers from up-and-coming Israeli researchers, below the list.
3
14
6
111
261,457
Gal Chechik retweeted
Dvir Samuel@dvir_samuelAug 16
🥳 Excited to share UniProbe: A learnable, token-level hallucination detection for large VLMs 🔎 UniProbe looks inside a VLM’s internals to identify hallucinated response tokens and mitigate them before they reach the user. 📉 55% fewer hallucinations ⚡ Just 1.06× latency
1
11
1
25
2,112
Incredible AI parkour simulation from @NVIDIAAI. It learned to be a pro from looking at just 30 seconds of footage! ▶️Full video: youtu.be/8B05cy3UuSE
7
30
1
252
28,858
דברים זזים במטה הלאומי לבינה מלאכותית. זה לא קצב ממשלתי. זה קצב אחר. אתמול פרסם המטה הלאומי לבינה מלאכותית יחד עם החשב הכללי שלושה RFIs חשובים מאוד שמוציאים לדרך את שלושה פרויקטים גדולים של התכנית הלאומית לבינה מלאכותית: 1. הקמת מחשב קוונטי לאומי 2. ⁠הקמת עיר התנסויות ל physical AI 3. ⁠מודל שפה לאומי כחול לבן בדגש לסייבר. 1. פיזילנד: hazira.gpa.gov.il/subdomain/… 2. מודל נולד: hazira.gpa.gov.il/subdomain/… 3. mr.gov.il/ilgstorefront/he/p…
39
39
13
755
126,555
1/ Text-to-video models generate beautiful videos, but they still struggle to follow complex prompts. Relations like left/right, towards/away, on top of, behind, or even multi-stage instructions are often generated incorrectly. In our new paper, accepted to SIGGRAPH Asia 2026, we introduce CVG, an inference-time guidance method that improves compositional video generation without retraining or architectural changes.
8
8
32
76,905
Long due. Next: add small submission fees growing with number of papers
Gautam Kamath@thegautamkamathJul 30
ICLR 2027 has authorship quotas: - No more than 20 submissions per author - No more than 1 submission where no author has a paper previously accepted to a major ML conference
1
1
27
4,162
ICLR 2027 has authorship quotas: - No more than 20 submissions per author - No more than 1 submission where no author has a paper previously accepted to a major ML conference
ICLR@iclr_confJul 30
The ICLR 2027 CfP is up! iclr.cc/Conferences/2027/Cal…
44
65
46
589
348,885
Gal Chechik retweeted
Dvir Samuel@dvir_samuelFeb 3
🚀 Excited to share our new paper: “Fast Autoregressive Video Diffusion & World Models with Temporal Cache Compression & Sparse Attention.” We address attention bottlenecks in auto-regressive video diffusion, enabling ×5–×10 speedup and constant memory over long rollouts.
2
25
6
75
33,305
Gal Chechik retweeted
Dvir Samuel@dvir_samuelJul 6
I'm going to present Fast-AR at #ICML26 tomorrow (Tuesday) @ 2pm - poster #1204! Come and say hi 👋 If you're around this week, feel free to DM me. Project page: dvirsamuel.github.io/fast-au… Details below ⬇️
Dvir Samuel@dvir_samuelFeb 3
🚀 Excited to share our new paper: “Fast Autoregressive Video Diffusion & World Models with Temporal Cache Compression & Sparse Attention.” We address attention bottlenecks in auto-regressive video diffusion, enabling ×5–×10 speedup and constant memory over long rollouts.
3
9
1
23
8,686
Physical AI Day is coming to #SIGGRAPH2026. 🤩 Learn how NVIDIA research is advancing physical AI for robotics, autonomous vehicles, and industrial digital twins. Don't miss sessions including: 👉 OpenUSD for AI Agents, Model-Driven Workflows, and 3D Graphics Pipelines 👉 Building SimReady 3D Worlds for Physical AI 👉 Career Growth in Physical AI: Bringing 3D Graphics Skills into Robotics and Digital Twins Learn more: nvda.ws/4eJxeR0
2
13
2
33
4,086
Gal Chechik retweeted
Chen Tessler
@ChenTesslerJul 2
We've released an update to ProtoMotions! nvlabs.github.io/ProtoMotion… Most importantly, this release was built with Yifeng Jiang, @YiShi_333 , @erwincoumans and @xbpeng4 , whose work shaped everything from the core methods to the final code release. 1/5
7
53
3
298
19,758
Gal Chechik retweeted
Chen Tessler
@ChenTesslerJul 2
If you're in the area, don't miss this event! I'll be there to chat and show interactive demos of our recent work.
Carlos DP ✈️ CoRL 2026
@carlosdponxJul 1
Going to SIGGRAPH 2026? @dollhouserobots is hosting an Entertainment Robotics meetup near the conference on Wednesday July 22! We'll have one of our new humanoid robot characters there, as well as other live and simulated robot demos! Sign up below and tell your friends who are going to SIGGRAPH!
1
3
54
6,985
Gal Chechik retweeted
Or Perel@OrPerelJun 29
🎉 Excited to introduce TRON, a relighting framework for 3D captures. 💡TRON pairs a neural renderer with 3D Gaussian reconstructions, achieving realistic quality, with 3D, material, & lighting control at interactive frame rates. arxiv.org/abs/2606.11314 research.nvidia.com/labs/sil…
4
48
5
152
21,554
Gal Chechik retweeted
Yoad Tewel@YoadTewelJun 9
🎉 Happy to share our latest work: Bootstrap Your Generator: Unpaired Visual Editing with Flow Matching (accepted to #ICML2026)! TL;DR: We train image and video editing models without any paired data. No ground-truth edit data, no external reward models.
7
14
2
40
6,826
Thank you for explaining our work so wonderfully well!
Two Minute Papers@twominutepapersFeb 6
I felt like Indiana Jones unearthing a hidden treasure with this. 🤠 NVIDIA's new AI tech deletes the un-deletable - shadows on grass and more. All this in real time! Full video: youtu.be/RaNay3x0Fmk
4
511
Load more