Closed lvaleriu closed 6 years ago
Hi,
the name of these models describes the method and architecture of the model. We did not use any data from the contest you are mentioning and hence have unfortunately no weights we could share with you.
Thanks for your PR ;)
Is there a reference that describes these architectures? I don't see them mentioned in the paper.
Those models have been used during testing and initial research, but we did not include them in the paper, as they did not perform that well. They are included in the repository for reference. Just in case somebody wants to play around with them.
But is there any reference to where they came from? It seems like from some topcoder contest?
On Fri, Jun 15, 2018, 04:34 Christian Bartz notifications@github.com wrote:
Those models have been used during testing and initial research, but we did not include them in the paper, as they did not perform that well. They are included in the repository for reference. Just in case somebody wants to play around with them.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/HPI-DeepLearning/crnn-lid/issues/2#issuecomment-397552817, or mute the thread https://github.com/notifications/unsubscribe-auth/ACSLV-k8spPHwe_uzdq8vcRjRU6jf-QYks5t83GWgaJpZM4QpCQz .
Hello. I'm seeing some models who are described as "TOPCODER etc". Ex: topcoder_5s_finetune, Topcoder_CRNN, etc.
Are these models from the SpokenLanguage topcoder contest? If this is the case, can you share the weights?
Thanks in advance!
Your project is SUPERB by the way!