Hi Ed,
According to the paper, the How2Sign experiments were initialized with Uni-Sign’s ST-GCN weights pretrained on YT-ASL. Would it be possible to share the exact checkpoint used in your experiments?
I also saw your comment mentioning plans to conduct full hyperbolic pretraining on YT-ASL. I was wondering whether you have made any progress on this and, if so, whether you would be willing to share the resulting checkpoint with our group.
We would use the weights solely for academic research and reproducibility. If redistribution is restricted, we would also greatly appreciate any guidance on obtaining the checkpoint, details of the pretraining configuration, or an introduction to the appropriate Uni-Sign contact.
Thank you very much for your time and help.
Hi Ed,
According to the paper, the How2Sign experiments were initialized with Uni-Sign’s ST-GCN weights pretrained on YT-ASL. Would it be possible to share the exact checkpoint used in your experiments?
I also saw your comment mentioning plans to conduct full hyperbolic pretraining on YT-ASL. I was wondering whether you have made any progress on this and, if so, whether you would be willing to share the resulting checkpoint with our group.
We would use the weights solely for academic research and reproducibility. If redistribution is restricted, we would also greatly appreciate any guidance on obtaining the checkpoint, details of the pretraining configuration, or an introduction to the appropriate Uni-Sign contact.
Thank you very much for your time and help.