LiquidAI/d1-3B
Image-Text-to-Text • 3B • Updated • 9.09k • 253
Thomas Gauthier-Caron
thomasgauthier
AI & ML interests
None yet
Recent Activity
liked a model 2 days ago
LiquidAI/d1-3B updated a model 8 days ago
thomasgauthier/clef-flash-GGUF published a model 8 days ago
thomasgauthier/clef-flash-GGUFOrganizations
liked a
model 2 days ago
updated a
model 8 days ago
published a
model 8 days ago
liked a
model 8 days ago
liked 2
models 9 days ago
upvoted an article 9 days ago
Article
Qwen3.8-27B-pi: Effort-Ordered Reasoning for Agentic Coding
bytkim
• • 16 liked a
model 9 days ago
upvoted an article 11 days ago
Article
NVIDIA Kumo Tabular Sets a New Accuracy-Efficiency Frontier for Tabular Prediction
nvidia
• • 88 liked a
model 11 days ago
Thank You So Much! | GGUF Compatibility?
2
#1 opened 15 days ago
by 
mfielding92
updated a
model 15 days ago
published a
model 15 days ago
liked a
model 15 days ago
liked 3
models 16 days ago
reacted to sergiopaniego's post with 🔥 18 days ago
Post
2195
Can you do RL over taste?
I've spent some time reproducing, in the open, Surya N's idea of training a model to paint with code. It's a coding model that learns to paint watercolours by writing JS code, trained with GRPO. I used TRL and OpenEnv for this, with the whole pipeline running on Hugging Face.
The interesting part is that the reward has no correct answer, unlike a math problem. In this case it's based on the artistic preferences of the person who builds the dataset.
Everything is published: the environment, the reference pool, the trained adapters, every painting of every run with the code that made it, and a write-up with all the decisions, including the ones that went wrong.
Blog post: https://huggingface.co/blog/train-to-paint-with-code
I've spent some time reproducing, in the open, Surya N's idea of training a model to paint with code. It's a coding model that learns to paint watercolours by writing JS code, trained with GRPO. I used TRL and OpenEnv for this, with the whole pipeline running on Hugging Face.
The interesting part is that the reward has no correct answer, unlike a math problem. In this case it's based on the artistic preferences of the person who builds the dataset.
Everything is published: the environment, the reference pool, the trained adapters, every painting of every run with the code that made it, and a write-up with all the decisions, including the ones that went wrong.
Blog post: https://huggingface.co/blog/train-to-paint-with-code
liked a
model 18 days ago
liked a
Space 21 days ago