Loading video...

Video Failed to Load

Go Home

Our Latent Encoder-Decoder code base is fully open sourced, you can train and visualize the latent space: Code⚙️: ArXiv 📚: #CVPR2026

xyz2maureen's profile picture

Xueyan Zou

2,056 subscribers

19,323 views • 6 months ago •via X (Twitter)

#CVPR2026
LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

The latent-vs-pixel debate misses the point. GPT Image 2 shows what users notice: pixel-level fidelity. Latent models show what scales: compact semantic structure. We connect them by replacing VAE/RAE decoders with a Pixel Diffusion Decoder. Code and Model available: 🧵(1/N)
0:48

The latent-vs-pixel debate misses the point. GPT Image 2 shows what users notice: pixel-level fidelity. Latent models show what scales: compact semantic structure. We connect them by replacing VAE/RAE decoders with a Pixel Diffusion Decoder. Code and Model available: 🧵(1/N)

Xuanchi Ren

669,535 views • 3 months ago

Agents can now understand physical space and temporality on Dimensional. Input: “Find the model airplane” With SpatioTemporal RAG, our agent has an emergent understanding of objects, stairwells, and obstacles in latent space.
0:35

Agents can now understand physical space and temporality on Dimensional. Input: “Find the model airplane” With SpatioTemporal RAG, our agent has an emergent understanding of objects, stairwells, and obstacles in latent space.

stash

10,569 views • 7 months ago

You can now build AI Agents using a no-code drag and drop interface. Fully open-source.
0:22

You can now build AI Agents using a no-code drag and drop interface. Fully open-source.

Lior Alexander

87,456 views • 1 year ago

Love seeing Silico (Goodfire ) used to probe our EchoJEPA's representations! this is exactly the kind of interpretability work that's been missing for JEPA-style models. One thing that makes EchoJEPA particularly interesting to interpret: unlike MAE-based approaches, it never reconstructs pixels. The model learns entirely in latent space through masked prediction, so you can't just look at decoder outputs to understand what it captured. Attribution onto a temporally aligned 3D mesh is a much more honest probe of what the representations actually encode. What we found in building EchoJEPA: training on 18M echo videos across 300K patients, the model learns to disentangle cardiac anatomy from ultrasound noise (speckle, reverberation artifacts) almost entirely through self-supervision. With 1% labeled data it already outperforms supervised baselines trained on 100%. The latent space is doing real anatomical work, but until you can visualize it like this, "real anatomical work" is mostly a claim. Paper + code: |
0:51

Love seeing Silico (Goodfire ) used to probe our EchoJEPA's representations! this is exactly the kind of interpretability work that's been missing for JEPA-style models. One thing that makes EchoJEPA particularly interesting to interpret: unlike MAE-based approaches, it never reconstructs pixels. The model learns entirely in latent space through masked prediction, so you can't just look at decoder outputs to understand what it captured. Attribution onto a temporally aligned 3D mesh is a much more honest probe of what the representations actually encode. What we found in building EchoJEPA: training on 18M echo videos across 300K patients, the model learns to disentangle cardiac anatomy from ultrasound noise (speckle, reverberation artifacts) almost entirely through self-supervision. With 1% labeled data it already outperforms supervised baselines trained on 100%. The latent space is doing real anatomical work, but until you can visualize it like this, "real anatomical work" is mostly a claim. Paper + code: |

Bo Wang

29,611 views • 4 months ago

We just released our latest NanoGL experience! If you haven’t already, head over: The code is fully open source ( so you can see how we make interactive experiences at Makemepulse using our Webgl framework:
0:29

We just released our latest NanoGL experience! If you haven’t already, head over: The code is fully open source ( so you can see how we make interactive experiences at Makemepulse using our Webgl framework:

makemepulse

11,011 views • 2 years ago

This is insane. We can now detect feelings in real-time through facial expressions using AI. (Open-sourced code in the next tweet)
1:10

This is insane. We can now detect feelings in real-time through facial expressions using AI. (Open-sourced code in the next tweet)

Rowan Cheung

3,157,721 views • 3 years ago

I built an AI data visualization Agent. The AI Agent can automatically generate the code and visualize the data using code interpreter sandbox. 100% Opensource Code with step-by-step tutorial.
1:40

I built an AI data visualization Agent. The AI Agent can automatically generate the code and visualize the data using code interpreter sandbox. 100% Opensource Code with step-by-step tutorial.

Shubham Saboo

68,488 views • 1 year ago

New way to navigate latent space. It preservers the underlying image structure and feels a bit like a powerful style-transfer that can be applied to anything. The trick is to...
0:36

New way to navigate latent space. It preservers the underlying image structure and feels a bit like a powerful style-transfer that can be applied to anything. The trick is to...

Johannes Stelzer

313,952 views • 2 years ago

If AI can code 100x faster, why aren't you shipping 100x faster? Because AI code is not production-ready code, and definitely not code where you understand and can vouch for every line Introducing the Command Center alpha. Support our Product Hunt launch!
0:22

If AI can code 100x faster, why aren't you shipping 100x faster? Because AI code is not production-ready code, and definitely not code where you understand and can vouch for every line Introducing the Command Center alpha. Support our Product Hunt launch!

Jimmy Koppel

72,602 views • 10 months ago

Congratulations to the Cohere team on the release of Cohere Transcribe Arabic! 🎉 Runs natively on mlx-audio (Python + Swift) from day-0 🚀 What's inside: → 2B params, Conformer encoder-decoder (audio-in, text-out) → Large Conformer encoder for acoustic representations + lightweight Transformer decoder for token generation → Built for Arabic dialects and Arabic-English code-switching → Auto-resampling to 16kHz + stereo→mono handling built into preprocessing → Apache 2.0 — fully open for community research Currently topping the Open Universal Arabic ASR Leaderboard 🥇 Get started today: 🐍 Python uv pip install -U mlx-audio 🍎 Swift
1:02

Congratulations to the Cohere team on the release of Cohere Transcribe Arabic! 🎉 Runs natively on mlx-audio (Python + Swift) from day-0 🚀 What's inside: → 2B params, Conformer encoder-decoder (audio-in, text-out) → Large Conformer encoder for acoustic representations + lightweight Transformer decoder for token generation → Built for Arabic dialects and Arabic-English code-switching → Auto-resampling to 16kHz + stereo→mono handling built into preprocessing → Apache 2.0 — fully open for community research Currently topping the Open Universal Arabic ASR Leaderboard 🥇 Get started today: 🐍 Python uv pip install -U mlx-audio 🍎 Swift

Prince Canuma

16,680 views • 2 months ago

We introduce W.A.L.T, a diffusion model for photorealistic video generation. Our model is a transformer trained on image and video generation in a shared latent space. 🧵👇
0:43

We introduce W.A.L.T, a diffusion model for photorealistic video generation. Our model is a transformer trained on image and video generation in a shared latent space. 🧵👇

Agrim Gupta

431,249 views • 2 years ago

I'm releasing a new latent space explorer 🧭🎨 — a creative tool for generating and evolving images. It's tree-based and designed for creative flow and open-ended discovery, using image mixing, branching, and editing. You can try it out now on ArtBreeder!
0:43

I'm releasing a new latent space explorer 🧭🎨 — a creative tool for generating and evolving images. It's tree-based and designed for creative flow and open-ended discovery, using image mixing, branching, and editing. You can try it out now on ArtBreeder!

Joel Simon

19,922 views • 1 year ago

Your Enterprise Notebooks just got an upgrade. You can now use Google Cloud Workbench Instances directly from your favorite IDE with the new VS Code extension. With this new extension you can connect to and run notebooks on managed cloud environments directly within your local IDE. As part of our commitment to the developer ecosystem, the extension is fully open-sourced to support community-driven innovation. 🧵
0:20

Your Enterprise Notebooks just got an upgrade. You can now use Google Cloud Workbench Instances directly from your favorite IDE with the new VS Code extension. With this new extension you can connect to and run notebooks on managed cloud environments directly within your local IDE. As part of our commitment to the developer ecosystem, the extension is fully open-sourced to support community-driven innovation. 🧵

Google for Developers

33,144 views • 2 months ago

You can now edit HyperFrames’ Timeline directly! Drag in preview → video updates When Claude Code struggles with timing, you can give it a hand Experimental, open sourced in @hyperframes/player $ npx hyperframes upgrade RT + Comment "Timeline" for source code (must follow)
0:33

You can now edit HyperFrames’ Timeline directly! Drag in preview → video updates When Claude Code struggles with timing, you can give it a hand Experimental, open sourced in @hyperframes/player $ npx hyperframes upgrade RT + Comment "Timeline" for source code (must follow)

HeyGen

729,337 views • 4 months ago

Someone just built a Claude skill that turns any arxiv paper into working code. It's called paper2code. Drop an arxiv URL and you get a full implementation back with every line cited to the exact paper section and equation. 100% Open Source.
0:49

Someone just built a Claude skill that turns any arxiv paper into working code. It's called paper2code. Drop an arxiv URL and you get a full implementation back with every line cited to the exact paper section and equation. 100% Open Source.

Oliver Prompts

31,647 views • 4 months ago

here's how we did it and the open sourced HyperFrames code in thread 🧵 i did a quick screen recording as a quick explainer our open source framework & motion graphics skills:
0:36

here's how we did it and the open sourced HyperFrames code in thread 🧵 i did a quick screen recording as a quick explainer our open source framework & motion graphics skills:

Bin Liu

32,579 views • 1 month ago

Shadows of The Shining Exploring the latent space of The Shining is to approach the film as one approaches a memory. Here, artificial intelligence acts as a second consciousness: it does not reproduce — it remembers. It reconstructs The Shining from its traces, its echoes, from what has survived the narrative. The latent space becomes a mental territory, an inner Overlook where the ghosts of the image, the logic of dreams, and the mechanisms of visual memory unfold once again. The aim is to turn the exploration of the latent space into a mnemonic experience of cinema itself.
0:45

Shadows of The Shining Exploring the latent space of The Shining is to approach the film as one approaches a memory. Here, artificial intelligence acts as a second consciousness: it does not reproduce — it remembers. It reconstructs The Shining from its traces, its echoes, from what has survived the narrative. The latent space becomes a mental territory, an inner Overlook where the ghosts of the image, the logic of dreams, and the mechanisms of visual memory unfold once again. The aim is to turn the exploration of the latent space into a mnemonic experience of cinema itself.

Benjamin Bardou

37,527 views • 10 months ago

The future of Klever Blockchain is now fully open. The core code is officially open source — audited, transparent, and ready for the world. No black boxes. No compromises. Just secure infrastructure built to scale Web3 globally. Developers, institutions, builders: the door is open. Explore the code →
0:35

The future of Klever Blockchain is now fully open. The core code is officially open source — audited, transparent, and ready for the world. No black boxes. No compromises. Just secure infrastructure built to scale Web3 globally. Developers, institutions, builders: the door is open. Explore the code →

Klever Blockchain

316,826 views • 6 months ago