Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Implement and train ColPaliDuo #48

Draft
wants to merge 70 commits into
base: main
Choose a base branch
from
Draft

Conversation

tonywu71
Copy link
Collaborator

@tonywu71 tonywu71 commented Sep 10, 2024

Description

Implement and train ColPaliDuo. The most noticeable feature of ColPaliDuo is to being able to output both single-vector (dense) and multi-vector (ColBERT-like) embeddings.

Follow-up of #22.

@tonywu71 tonywu71 self-assigned this Sep 10, 2024
@tonywu71 tonywu71 marked this pull request as draft September 10, 2024 15:17
@tonywu71 tonywu71 force-pushed the implement-and-train-colpali-2 branch 4 times, most recently from cc3b31d to 8bf550b Compare September 23, 2024 15:27
@tonywu71 tonywu71 changed the title Implement and train ColPali2 Implement and train ColPaliDuo Oct 2, 2024
@tonywu71 tonywu71 added the model New ColVision model label Oct 10, 2024
@tonywu71 tonywu71 force-pushed the implement-and-train-colpali-2 branch 2 times, most recently from bf24d43 to 7fac8f0 Compare October 16, 2024 14:05
Comment on lines +8 to +11
class ColbertCELoss(BaseColbertLoss):
"""
Cross-entropy loss using the ColBERT scores between the query and document embeddings.
"""
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@ManuelFay FYI I have renamed the ColbertLoss into ColbertCELoss to make it less ambiguous. Normally, all the previous occurence of this class have already been renamed, but it's worth one more check just before merging time.

@ManuelFay ManuelFay force-pushed the implement-and-train-colpali-2 branch from 033fb14 to a173f32 Compare October 21, 2024 13:10
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
model New ColVision model
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants