Yoach Lacombe

Hi @lhoestq, I plunged into the code and it should be manageable for me to work on it! #take Also, setting `d1`, `d2` and `d3` as you did raised a...

Hi @lhoestq, I actually have already wrote the code last time [on this commit](https://github.com/ylacombe/datasets/commit/84769db97facc78a33ec53f7b1b395951e1804df) but I still have to change the docs and write some tests though. I'm working on...

Hi @lhoestq, I've finally made some advances in the matter. I've modified the `IterableDataset` behavior so that it aligns with the `Dataset` behavior as we have discussed. The documentation has...

Hey @ruisilvestre, sorry for the long delay, I unfortunately don't have a Mac to try this on. Could you open a PR and we'll discuss this with @sanchit-gandhi ? Also...

Hey @rudransh2004, you can do this but you'd have to retrain the model from scratch!

Hey @rudransh2004, if you want to avoid using the annotations, you could simply use a description column with each samples having empty string "". Note that the model currently doesn't...

Hey @jay2jp, thanks for opening the discussion! Streaming could be done by adapting the [Streamer](https://huggingface.co/spaces/sanchit-gandhi/musicgen-streaming/blob/main/app.py#L52-L169) used for Musicgen with the current code! It'd be great if you could be involved...

Hey @vapemaster-kz, I believe you're misunderstanding what past-key values are used for! You can find on the docs [here](https://huggingface.co/docs/transformers/model_doc/seamless_m4t_v2#transformers.SeamlessM4Tv2ForSpeechToText.forward.past_key_values) that past-key values `can be used to speed up sequential decoding`....

Thanks for working on this @ravenouse, this looks super promising and will clearly be valuable for any audio-related model! I'm indeed considering adding a torch alternative of this numpy implementation!...

Hey @ravenouse, thanks for all the progress so far and happy new year! Could you actually update the code here so that it's easier to review, test and keep track...