question
Hi,
Thank you so much for releasing this new model!
I noticed that the model was Apache licensed but it looks like it might be based on Llama 3. If so, would the Llama 3 license apply?
Thanks!
Hi - it is based on Llama 3 - so the Llama 3 should in theory apply - this is a mistake on our end - and we'll try to resolve it asap!
Makes sense, thanks for the clarification!
Will the pre training code be released in case we want to experiment with different base models?
Makes sense, thanks for the clarification!
Will the pre training code be released in case we want to experiment with different base models?
I second this – I'm interested in pre-training a German base model and afterwards fine tune it on a German speaker.
I am currently working on a proposal for a Dutch law firm, to do inbound/outbound calling for case onboarding.
So I'm also interested in co-creating a TTS for the Dutch language.
There is initial pretraining code and information now on the GitHub - let me if anything/what is unclear - preferably in a GitHub issue - and I can provide more information there if needed!
So just to clarify, the training code is licensed under Apache while the model weights themselves are licensed under Apache + Llama3?