graham

besiktas's activity

• All
• Models
• Datasets
• Spaces
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

New activity in adept/fuyu-8b over 1 year ago

Has anyone tried adding positional embeddings to the image patches to improve the model?

2

#70 opened over 1 year ago by

jchiu1234

Are there special tokens that are ignored during loss computation?

9

#62 opened over 1 year ago by

Nyandwi

Here is a simple multimodal like training script to see model working.

👍 7

3

#60 opened over 1 year ago by

besiktas

fine-tuning using FSDP and non 80GB cards?

8

#43 opened almost 2 years ago by

besiktas

fine-tuning using FSDP and non 80GB cards?

8

#43 opened almost 2 years ago by

besiktas

fine-tuning using FSDP and non 80GB cards?

8

#43 opened almost 2 years ago by

besiktas