Qwen3.5
Collection
21 items • Updated
• 1.07k
David Bertoin PRO
Bertoin
AI & ML interests
None yet
Recent Activity
upvoted a collection 2 days ago
Qwen3.5 commented on an article 4 days ago
PRX Part 3 — Training a Text-to-Image Model in 24h! upvoted an article 5 days ago
PRX Part 3 — Training a Text-to-Image Model in 24h! Organizations
upvoted a collection 2 days ago
commented on PRX Part 3 — Training a Text-to-Image Model in 24h! 4 days ago
I don’t think we really observed such artifacts, or at least not due to LPIPS. We did see something similar without LPIPS when using JIT on undertrained models, but I don’t think we observe them anymore. It might simply be because the model had not converged yet.
One situation where we started to observe checkerboard patterns was during reward alignment with GRPO or SRPO on one of these models. In that case the artifacts were much more noticeable, but I don’t think they were related to LPIPS since we were not using it at the time.
What type of model were you using?
upvoted an article 5 days ago
Article
PRX Part 3 — Training a Text-to-Image Model in 24h!
•
53
published an
article 6 days ago
Article
PRX Part 3 — Training a Text-to-Image Model in 24h!
•
53
liked a model 11 days ago
updated a Space 24 days ago
updated a model 24 days ago
upvoted 2 articles about 1 month ago
Article
LightOnOCR-2-1B: a lightweight high-performance end-to-end OCR model family
•
87
Article
Training Design for Text-to-Image Models: Lessons from Ablations
•
69
published an
article about 1 month ago
Article
Training Design for Text-to-Image Models: Lessons from Ablations
•
69
Doesn't Work
2
#6 opened 3 months ago
by
qpqpqpqpqpqp
Upload folder using huggingface_hub
#5 opened 3 months ago
by
Smouse
Distorted images
5
#3 opened 4 months ago
by
HarryML
liked a Space 4 months ago
Negative prompting?
4
#1 opened 4 months ago
by
mingyi456
upvoted an article 4 months ago
Article
Text-to-image Architectural Experiments
•
53
published an
article 4 months ago
Article
Text-to-image Architectural Experiments
•
53
upvoted a collection 4 months ago
upvoted an article 4 months ago
Article
We’re open-sourcing our text-to-image model and the process behind it
•
92
published an
article 4 months ago
Article
We’re open-sourcing our text-to-image model and the process behind it
•
92