issues
search
yk7333
/
d3po
[CVPR 2024] Code for the paper "Using Human Feedback to Fine-tune Diffusion Models without Any Reward Model"
https://arxiv.org/abs/2311.13231
MIT License
168
stars
14
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
After training diffusion model with human preference dataset.
#18
giovanlee
opened
1 month ago
1
Clarification on Human Preference JSON File
#17
jxxtin
opened
1 month ago
1
Can we train D3PO without text prompts?
#16
giovanlee
closed
1 month ago
2
Does D3PO only available for DM which predicts X instead of noise (eps)?
#15
imabackstabber
opened
3 months ago
1
How to organize human feedback results into a JSON file?
#14
Mr-Loevan
opened
3 months ago
3
Pretrained weight release
#13
Mer9ury
opened
4 months ago
0
DPO with existed-images
#12
guoyanan1g
opened
4 months ago
2
Online or Offline?
#11
he-nantian
closed
4 months ago
5
Comparison with DiffusionDPO
#10
samedii
closed
7 months ago
2
Combine with dreambooth
#9
thanhhung0112
opened
8 months ago
1
Releasing human preference data
#8
vishaal27
closed
9 months ago
2
Fixed fatal bug where model failed to save
#7
ADKoishi
opened
9 months ago
1
ValueError : Attemting to unscale fp16 Gradients
#6
ADKoishi
opened
9 months ago
0
Bug fixing and add support for memory cost reducing.
#5
ADKoishi
closed
9 months ago
0
Reproducing Aesthetic Quality results from paper
#4
ritiztambi
closed
7 months ago
2
assert num_timesteps == config.sample.num_steps AssertionError
#3
sdtana
closed
9 months ago
3
LICENSE
#2
jmetzen
closed
11 months ago
1
Can it be used to train controlnet? How?
#1
universewill
closed
11 months ago
3