use decode to inspect ds

use tokenizer to batch_decode

debug dataset

ifx overwriting var

print preds for debugging

fix casting single dataset

use take instead of select

fix typo

debug decoding

Merge branch 'main' of https://huggingface.co/marinone94/whisper-tiny-sv

add readme only

skip config ds

add readme only

Merge branch 'main' of https://huggingface.co/marinone94/whisper-tiny-sv into main

fix redme issues

add readme only

test model card

fix model cards

login wandb

upgrade torchaudio

run 50 steps

use multiple datasets

WIP: mix datasets

fix raw dataset features

reduce max steps

allow multiple languages and datasets

skip push to hub during training

allowing multiple datasets

handle PermissionError

get also wandb and email creds from file

get token from creds if not set in venv

force train + eval

add logs

fix path

fix cmd

update run sm py

use python script to clone repo and trigger train

debug

add git init manual command in script

update sm

update dockerfile with CACHEBUST

update dockerfile

first setup

use low max sample for debugging

add results in email

Add logic to init env and send email

initial commit