add VisionTextDualEncoder and CLIP fine-tuning script #15701
patil-suraj
merged
21 commits into
huggingface:master
from
patil-suraj:clip-train-script
begin script
c5a26d22
update script
d0bf3030
fix features and data args
a0ecfa9f
main
11952f81
add requirements
62586058
add column name args
994ab742
fix captions
797a416e
don't jit transforms
e9bbc0bd
fix caption
29301f67
fix labels, handle attention mask
8cc44b88
convert pixel values to numpy
ff1663cd
labels => input_ids
bfd578b3
transform images on the fly
553b8f3a
use AutoModel class, create the hybird model outside of the script
ba7c3642
fix version message
c797ec77
add readme
02dae214
patil-suraj
changed the title 
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
commented
on 2022-02-18
patrickvonplaten
approved these changes
on 2022-02-18
Apply suggestions from code review
f6dec3b3
adderss review comments
af5b523f
Merge branch 'clip-train-script' of https://github.com/patil-suraj/tr…
3ce93e23
add more comments
3542c27e
allow freezing vision and text models
f287c555
patil-suraj
commented
on 2022-02-21
Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub