Rohit Singh

Unlock AI reasoning with Vision Language Models at #esridevtech2025! Our dev portfolio leverages VLMs for smarter spatial insights. Innovate with us - precision meets cutting-edge tech. Here is to the spatial ones!

📰 #ArcGIS Blog: Use vision-language models to optimize object classification esri.com/arcgis-blog/pr… _ #Esri #GIS #Geospatial #MetaMapping

@AbheySinghIt Is this a fake account? In one video he said he's not on X?

@erbmjha Know all about Abhay Singh, the IIT Baba discovered at #KumbhMela2025 youtube.com/c/AbheySinghGr… facebook.com/abheysinghgrew… abhey-singh.medium.com @greywall5996" target="_blank" rel="nofollow noopener">youtube.com/@greywall5996

Meet IITian Baba at the Maha Kumbh, who did Aerospace Engineering from IIT Bombay but left everything for spirituality. Meanwhile, illiterate Leftists and Seculars mock Sanatanis.

@MagnusCarlsen You were making a joke of the rules and how to bend then. This was clear collision and match fixing. You were too scared of losing the title, you came up with this sinister trick and forced Ian into this. The winner couldn't be determined... There can't be two world champions.

I’ve never prearranged a draw in my career. In the video I’m joking with Ian in a situation with lacking decisive tiebreak rules. This was obviously not an attempt to influence FIDE. It was said in the spirit that I thought FIDE would agree to our proposal. If anything it was a bad joke given the gravity of the situation. I think the match itself showed two players playing high level chess, equally matched and both deserving of a win.

@elToroDelRio This is just a screen grab from fide.com/images/stories… 1) It's just a proposal 2) It's from 2013 These arent the official rules.

By all means, FIDE is just bullying Magnus Carlsen at this point. Jeans are allowed in their own dress code. Here's an excerpt of their dresscode: This was a clear humiliation ritual towards magnus after signing with Weissenhaus. FIDE owes an apology. Magnus' Jeans was NOT torn

China trained a model called DeepSeek-v3 that’s better than OpenAI GPT 4o with a small fraction of the budget and open sourced it. Meanwhile, a bipartisan bill is in the works in America to impose severe restrictions on open sourcing frontier models.

@donelianc @fchollet What happenned to MindsAI? It won, but didn't open source their implementation! "Open"AI will achieve AGI - nah?

@fchollet First surprise from the report 👀🍿

Today we're announcing the winners of ARC Prize 2024. We're also publishing an extensive technical report on what we learned from the competition (link in the next tweet). The state-of-the-art went from 33% to 55.5%, the largest single-year increase we've seen since 2020. The benchmark remains unbeaten, but we're happy to see that research progress on the key bottleneck to AGIs (in particular on-the-fly adaptation to novel tasks) has been reignited in 2024 -- in part thanks to ARC Prize. In particular the competition has popularized Test Time Training (TTT), originally pioneered for ARC-AGI by Jack Cole last year. I believe TTT represents the largest jump in LLM generalization capabilities since the initial findings regarding in-context-learning circa 2019-2020. ARC Prize has also led to a considerable surge of research interest towards program synthesis. Competition winners: 🥇 the ARChitects (Daniel Franzen, Jan Disselhoff) 🥈 @guille_bar 🥉 alijs (Agnis Liukis) Paper Award winners: 🥇 "Combining Induction and Transduction For Abstract Reasoning" by @xu3kev et al. 🥈 "The Surprising Effectiveness of Test-Time Training for Abstract Reasoning" by @akyurekekin et al. 🥉 "Searching Latent Program Spaces" by @ClementBonnet16 & @MattVMacfarlane ARC-AGI-Pub Leaderboard (solutions using commercial APIs): 🥇 @jeremyberman 🥈 @ellisk_kellis & @akyurekekin 🥉 @RyanPGreenblatt

@achera23 See @ab3cerra" target="_blank" rel="nofollow noopener">youtube.com/@ab3cerra - seems to be fully AI generated - the missing piece seems to be the animated characters you have.

Check out this AWESOME video I created for one of my articles on 'Impact of Decisions': notebooklm.video/c/player/2093e…

Introducing -- NotebookLM Video v2 A tool to generate videos from NotebookLM podcasts Now it supports: - Instant video processing - Instant sharing - Exports in full-HD MP4 format - Async video creation with email notifications

YouTube channel entirely made from AI - NotebookLM + books + AI generated art => video, post to YouTube. @rougeux is this your work? @ab3cerra" target="_blank" rel="nofollow noopener">youtube.com/@ab3cerra I also see products being built around this - see x.com/achera23/statu…

If you are curious how Multimodal LLMs work, I wrote a new article to explain the two main approaches, decoder-only- and cross-attention-style: magazine.sebastianraschka.com/p/understandin… Plus, I reviewed and summarized the 10 latest research papers to see how it's done in practice. Happy reading!

FeatUp improves the spatial resolution of any model's features by 16-32x without changing their semantics. github.com/mhamilton723/F…

Even while recording these demos, we encountered some amusing moments. In one, Claude accidentally stopped a long-running screen recording, causing all footage to be lost. Later, Claude took a break from our coding demo and began to peruse photos of Yellowstone National Park.

🚀 Our NeurIPS '24 work, Large Spatial Model (LSM), is here! LSM performs semantic 3D reconstruction in just 0.1s, processing unposed data via feed-forward 3D reconstruction. 👉It leverages large-scale 3D datasets with minimal annotations, defining a 3D latent space. We are continuously exploring how this explicit 3D representation can further enhance reasoning and robotic learning. 🔗 Try our online Gradio demo with your own data at largespatialmodel.github.io #NeurIPS2024 #3DReconstruction

3D Building Reconstruction from Point Clouds #b8" target="_blank" rel="nofollow noopener">sciencedirect.com/science/articl… github.com/chenzhaiyu/pol…

Great collection of Transformer sample notebooks - github.com/NielsRogge/Tra…

SegGPT: Segmenting Everything In Context github.com/baaivision/Pai…

and that too in-context!

One Shot SAM Segmentation - [ICLR'24] Matcher: Segment Anything with One Shot Using All-Purpose Feature Matching arxiv.org/abs/2305.13310