Devansh Bisla

AI / AV @ NVIDIA | Past: PhD AI/ML @ NYU |

[Saai Arora @SaaiArora]

Drake concerts in sf go crazy

[Devansh Bisla @devansh20la]

What a beautiful day for football! @FIFAWorldCup

[Physical Intelligence @physical_int]

New work with @nvidia: evaluating robot policies entirely inside a world model. The policy acts, the model imagines the consequences, and the imagined evals predict real-world results. 🧵

real vs world-model rollout side by side📷

[Mojtaba Seyedhosseini @mseyed]

Gemini Embedding 2: A Native Multimodal Embedding Model from Gemini 🚀

Today, we’re sharing the @GoogleDeepMind white paper for GE 2, our first native multimodal embedding model. Whether it’s text, audio, video, or image, GE 2 provides a unified representation of the input.

[Devansh Bisla @devansh20la]

@khushkhushkhush @CMayerHQ One of my fav things to do in nyc is to take a day off from work on a random Tue/Wed and go around the city.

[Matteo @MozarellaPesto]

I trained an autoencoder that reconstructs images with zero reconstruction loss.

No MSE. No image space supervision.

The only signal: "According to you, does your output look like your input through your own eyes?"

It works.

Blog link, demo and summary 👇

[Devansh Bisla @devansh20la]

@dwarkesh_sp @ericjang11 Transformer have the global context with attention, they can choose to attend or not attend to local or global information. It’s not an architecture issue but rather optimization.

[Dylan Itzikowitz @ditzikow]

Demo night, @spc NYC, May 20

Lineup

@LumaLabsAI — Ray3 (reasoning video model)
@ElevenLabs — ElevenMusic (studio-grade AI music)
@SentienceCom — personal AI “digital twins”
@sandbar — voice ring for notes/actions
@floraai — unified creative AI

Apply below

[Devansh Bisla @devansh20la]

Probably the lowest piece of argument clinic in this week of @TheEconomist.

The piece poorly tries to wash away the history of Mughal rule over India in guard of giving us paneer and biryani? What? 🤷

[NVIDIA AI Developer @NVIDIAAIDev]

Today, we released Lyra 2.0, a framework for generating persistent, explorable 3D worlds at scale, from NVIDIA Research.

Generating large-scale, complex environments is difficult for AI models. Current models often “forget” what spaces look like and lose track of movement over time, causing objects to shift, blur, or appear inconsistent. This prevents them from creating the reliable 3D environments required for downstream simulations. Lyra 2.0 solves these issues by:

✅ Maintaining per-frame 3D geometry to retrieve past frames and establish spatial correspondences
✅ Using self-augmented training to correct its own temporal drifting.

Lyra 2.0 turns an image into a 3D world you can walk through, look back, and drop a robot into for real-time rendering, simulation, and immersive applications.

➡️ Learn more: https://t.co/ROR7miJeCU
📄 Read the paper: https://t.co/1osU9EGjGD

[Devansh Bisla @devansh20la]

Protecting interest of one group doesn’t automatically mean harming other.

He also didn’t say I will not let robotaxis operate period. Although, it will be a terrible decision.

The question is asked at 15:00
https://t.co/tOLt4xtR5Z

[Sar Haribhakti @sarthakgh]

Congrats to NYC Mayor on yet another push to harm the majority in the interest of protecting a minority interest group

Waymo rollout in NYC has been stopped

[Devansh Bisla @devansh20la]

@goharism @jackieberardo @designerfund @collabfund lol I was gonna say the same thing. I love how they display the dishes as wall art 🍜

[Devansh Bisla @devansh20la]

Damnnit! I just started learning it but seems like we are going back to BASIC now:🤣

https://t.co/UQFpYOHEMF

[Bryce, the CUDA Colonel @blelbach]

Today, NVIDIA is launching the next paradigm shift in GPU programming: cuTile BASIC

Write perf portable BASIC kernels and deploy them at any scale from edge inference devices like your calculator to entire GPU clusters

We're going back to BASIC

https://t.co/meF2T0jUSc

[Devansh Bisla @devansh20la]

My feed is talking about custom CUDA kernels, so I started digging in last week. I’ve put together a "fill in the missing code" guide for reproducing GPT-2 here: https://t.co/JI2MKPUSHl

The boilerplate & paths are wired up-you just implement the forward paths with kernels.

[Devansh Bisla @devansh20la]

My first pass is on the main branch (```git checkout main```). There’s still plenty of memory and synchronization overhead to iron out for anyone looking to get closer to the hardware.

[Dylan Itzikowitz @ditzikow]

NYC is becoming a hotbed for frontier tech.

Hear from a few of the names leading the way including @josephfkrause (@RadicalAI) and Rob Cochran (@faunarobotics) on March 31st @ SPC NYC moderated by our friend @EricNewcomer.

[Devansh Bisla @devansh20la]

Very interesting!

I’m confused by the fact that my timeline is filled with CVPR papers declaring spatial problems like 3d reconstruction, monocular depth estimation, etc completely solved.

Can anyone help me understand why do we still need depth cameras and prebuilt object libraries with meshes to build digital twin? What are the remaining gaps? 🧐

[Devansh Bisla @devansh20la]

@lesDecroissant @sarahookr In theory the manipulation only gets easier instead of paying people you can just buy enough tokens and manipulate the result.