3 ms·
Very cool. It looks like you train your transformer model as normal and then fine tune it some more with a reward model
by imranq 4y ago
Very cool. It looks like you train your transformer model as normal and then fine tune it some more with a reward model