Samuel Lavoie
@lavoiems.bsky.social
PhD candidate @Mila_quebec, @UMontreal. Ex: FAIR @AIatMeta.
Learning representations, minimizing free energy, running.
Learning representations, minimizing free energy, running.
Reposted by Samuel Lavoie
Compositionality is a central desideratum for intelligent systems...but it's a fuzzy concept and difficult to quantify. In this blog post, lab member @ericelmoznino.bsky.social outlines ideas toward formalizing it & surveys recent work. A must-read for interested researchers in AI and Neuro
Very excited to release a new blog post that formalizes what it means for data to be compositional, and shows how compositionality can exist at multiple scales. Early days, but I think there may be significant implications for AI. Check it out! ericelmoznino.github.io/blog/2025/08...
Defining and quantifying compositional structure
What is compositionality? For those of us working in AI or cognitive neuroscience this question can appear easy at first, but becomes increasingly perplexing the more we think about it. We aren’t shor...
ericelmoznino.github.io
August 19, 2025 at 1:51 PM
Compositionality is a central desideratum for intelligent systems...but it's a fuzzy concept and difficult to quantify. In this blog post, lab member @ericelmoznino.bsky.social outlines ideas toward formalizing it & surveys recent work. A must-read for interested researchers in AI and Neuro
🧵 Everyone is chasing new diffusion models—but what about the representations they model from?
We introduce Discrete Latent Codes (DLCs):
- Discrete representation for diffusion models
- Uncond. gen. SOTA FID (1.59 on ImageNet)
- Compositional generation
- Integrates with LLM
🧱
We introduce Discrete Latent Codes (DLCs):
- Discrete representation for diffusion models
- Uncond. gen. SOTA FID (1.59 on ImageNet)
- Compositional generation
- Integrates with LLM
🧱
July 22, 2025 at 2:41 PM
🧵 Everyone is chasing new diffusion models—but what about the representations they model from?
We introduce Discrete Latent Codes (DLCs):
- Discrete representation for diffusion models
- Uncond. gen. SOTA FID (1.59 on ImageNet)
- Compositional generation
- Integrates with LLM
🧱
We introduce Discrete Latent Codes (DLCs):
- Discrete representation for diffusion models
- Uncond. gen. SOTA FID (1.59 on ImageNet)
- Compositional generation
- Integrates with LLM
🧱
The code and model weights for Llip are finally out! I hope you will find this model useful!
Paper: arxiv.org/abs/2405.00740
Code: github.com/facebookrese...
Models:
- ViT-G: huggingface.co/lavoies/llip...
- ViT-B: huggingface.co/lavoies/llip...
Paper: arxiv.org/abs/2405.00740
Code: github.com/facebookrese...
Models:
- ViT-G: huggingface.co/lavoies/llip...
- ViT-B: huggingface.co/lavoies/llip...
Modeling Caption Diversity in Contrastive Vision-Language Pretraining
There are a thousand ways to caption an image. Contrastive Language Pretraining (CLIP) on the other hand, works by mapping an image and its caption to a single vector -- limiting how well CLIP-like mo...
arxiv.org
July 17, 2025 at 1:59 PM
The code and model weights for Llip are finally out! I hope you will find this model useful!
Paper: arxiv.org/abs/2405.00740
Code: github.com/facebookrese...
Models:
- ViT-G: huggingface.co/lavoies/llip...
- ViT-B: huggingface.co/lavoies/llip...
Paper: arxiv.org/abs/2405.00740
Code: github.com/facebookrese...
Models:
- ViT-G: huggingface.co/lavoies/llip...
- ViT-B: huggingface.co/lavoies/llip...