Article
Trained 210M text-to-image model from scratch on one GPU: what actually mattered
ivanmikhnenkov
ā¢ ā¢ 10
M Almenea
malmenea
Ā·
AI & ML interests
None yet
Recent Activity
upvoted an article 5 days ago
Trained 210M text-to-image model from scratch on one GPU: what actually mattered liked a Space 11 days ago
suvadityamuk/3d-representations-guide liked a model 4 months ago
microsoft/GridSFM_OpenOrganizations
upvoted an article 5 days ago
liked a
Space 11 days ago
liked a
model 4 months ago
liked a
model 6 months ago
upvoted a paper 10 months ago
liked a
Space about 1 year ago
reacted to loubnabnl's post with š¤ over 2 years ago
Post
6544
We've just published a detailed blog post on the creation of Cosmopedia dataset. We hope this will provide insights about generating synthetic data at scale for pre-training.
https://huggingface.co/blog/cosmopedia
Here are some key takeaways:
šÆ Prompt curation is crucial: we want to cover many topics with few duplicates.
š You can leverage various resources for diversity: using different seed data, generation formats, and target audiences.
āļø The importance of a good technical stack: for scalable generations with tools like llm-swarm and fast model training and evaluation.
Have a good read!
https://huggingface.co/blog/cosmopedia
Here are some key takeaways:
šÆ Prompt curation is crucial: we want to cover many topics with few duplicates.
š You can leverage various resources for diversity: using different seed data, generation formats, and target audiences.
āļø The importance of a good technical stack: for scalable generations with tools like llm-swarm and fast model training and evaluation.
Have a good read!
- 1 reply
upvoted a paper about 3 years ago
liked a
model about 3 years ago
