hi. i try your fas-tune-float.traineddata. it was very good.
i try to training new one but it can't be better than your file.
can you explain what data you used in this file?
are you trained with best fas.traineddata or ara.traineddata ?
you put english text to it too. how you do that?
are you use tesstrain.sh or python tesstrain?
please answer this.
i need information to train new one.
thanks.
hi. i try your fas-tune-float.traineddata. it was very good. i try to training new one but it can't be better than your file. can you explain what data you used in this file? are you trained with best fas.traineddata or ara.traineddata ? you put english text to it too. how you do that? are you use tesstrain.sh or python tesstrain? please answer this. i need information to train new one. thanks.