File size: 1,476 Bytes
5d11d07 f8d6756 5d11d07 e5e2b12 5d11d07 71c1cb6 5d11d07 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 |
# Masked Autoencoders are Scalable Learners of Cellular Morphology
Official repo for Recursion's accepted spotlight paper at [NeurIPS 2023 Generative AI & Biology workshop](https://openreview.net/group?id=NeurIPS.cc/2023/Workshop/GenBio).
Paper: https://arxiv.org/abs/2309.16064
## Provided code
The baseline Vision Transformer architecture backbone used in this work can be built with the following code snippet from Timm:
```
import timm.models.vision_transformer as vit
def vit_base_patch16_256(**kwargs):
default_kwargs = dict(
img_size=256,
in_chans=6,
num_classes=0,
fc_norm=None,
class_token=True,
drop_path_rate=0.1,
init_values=0.0001,
block_fn=vit.ParallelScalingBlock,
qkv_bias=False,
qk_norm=True,
)
for k, v in kwargs.items():
default_kwargs[k] = v
return vit.vit_base_patch16_224(**default_kwargs)
```
Additional code will be released as the date of the workshop gets closer.
**While we cannot share all the internal code we've written training and evaluation of these models, it would be very useful if interested persons could raise an Issue in this repo to inform us as to what the most useful aspects of the code for this project would be of interest to the broader community.**
## Provided models
Stay tuned...
|