Training setup
Num train steps 10000
Max seq len 256
Batch size 512
Total data points seen 5.1 mil
Total tokens seen 450 mil
Checkpoint step 10000
Learning rate 1e-3
Metric Val Test
BLEU 30.5 28.3
chrf++ 48.3 46.8
Downloads last month
5
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support