When will the full code be released?

keyu-tian / SparK

[ICLR'23 Spotlight🔥] The first successful BERT/MAE-style pretraining on any convolutional network; Pytorch impl. of "Designing BERT for Convolutional Networks: Sparse and Hierarchical Masked Modeling"

https://arxiv.org/abs/2301.03580

MIT License

1.42k stars 82 forks source link

When will the full code be released? #2

Closed zhangzjn closed 1 year ago

zhangzjn commented 1 year ago

Great work, but when will the full code be released? Thanks

keyu-tian commented 1 year ago

Thanks! The full codes including downstream finetuning and visualization are expected to be released by next week.

annawoodard commented 1 year ago

Great work-- very exciting paper; thanks for posting the pretraining code. I'm looking forward to the fine-tuning and viz code

songkq commented 1 year ago

@iFighting @keyu-tian Hi, could you please release the visualization code?

keyu-tian commented 1 year ago

@songkq @annawoodard @zhangzjn sorry kept you waiting so long, now our fine-tuning code & viz demo (Colab) is ready! We're also working on:

upload the vis demo to huggingface
upload pretrained weights into huggingface and timm

Once they finish we would comment here again and paste some notes, thank you all!