Today, we are open-sourcing our pipeline to deduplicate large-scale image datasets.
On one GPU, we can deduplicate 10k images against 1M indexed test images in ~60 seconds. But how?
On one GPU, we can deduplicate 10k images against 1M indexed test images in ~60 seconds. But how?
July 2, 2025 at 2:08 PM
Today, we are open-sourcing our pipeline to deduplicate large-scale image datasets.
On one GPU, we can deduplicate 10k images against 1M indexed test images in ~60 seconds. But how?
On one GPU, we can deduplicate 10k images against 1M indexed test images in ~60 seconds. But how?
New blog post for nanoVLM is live!
“nanoVLM – The simplest way to train a VLM in pure PyTorch”
We break down the full stack: architecture (SigLIP + SmolLM2), pixel shuffle, training pipeline, and inference.
With Colab + HF Space to try it out.
“nanoVLM – The simplest way to train a VLM in pure PyTorch”
We break down the full stack: architecture (SigLIP + SmolLM2), pixel shuffle, training pipeline, and inference.
With Colab + HF Space to try it out.
May 21, 2025 at 1:19 PM
New blog post for nanoVLM is live!
“nanoVLM – The simplest way to train a VLM in pure PyTorch”
We break down the full stack: architecture (SigLIP + SmolLM2), pixel shuffle, training pipeline, and inference.
With Colab + HF Space to try it out.
“nanoVLM – The simplest way to train a VLM in pure PyTorch”
We break down the full stack: architecture (SigLIP + SmolLM2), pixel shuffle, training pipeline, and inference.
With Colab + HF Space to try it out.
We just dropped full DDP support in native PyTorch for nanoVLM! Huge thanks to @geronimo-ai.bsky.social for the majority of this PR, check out his blog that explains how to do it: medium.com/@geronimo7/g...
May 19, 2025 at 10:07 AM
We just dropped full DDP support in native PyTorch for nanoVLM! Huge thanks to @geronimo-ai.bsky.social for the majority of this PR, check out his blog that explains how to do it: medium.com/@geronimo7/g...
Reposted by Luis
Real-time SmolVLM in a web browser with transformers.js.
All running locally with no installs. Just open the website.
All running locally with no installs. Just open the website.
May 14, 2025 at 3:39 PM
Real-time SmolVLM in a web browser with transformers.js.
All running locally with no installs. Just open the website.
All running locally with no installs. Just open the website.