@jpgallegoar Could you share more technical details? I'm curious about the hardware setup used for training and how long the training process took. If possible, would you mind sharing the training script?
Made a new video: understanding how a Transformer works, at training time vs. inference time. 🔥
I explain in detail how autoregressive generation works, and explain the difference with "teacher forcing", used during training.
https://t.co/Ie5vRGTn1J
Congratulations to @Aigiz! He worked hard to get his native language in @mozilla Common Voice.
Bashkir is officially included in the dataset now.
https://t.co/iEcF0G76bi