Viewing profile — jacek-123

Feels like a training-data artifact. SFT and preference data are full of "here's a cleaner version of your file", not "here's the minimum 3-line diff". The model learned bigger, mo…

Did you try GradNorm or PCGrad, or was manual task weighting good enough? Also curious about the required-vs-preferred head failing. Was that encoder gradient interference from the…

We ran this benchmark because we kept seeing the same failure mode: teams fine-tune small models on production traces expecting them to learn their agent's behavior, but the downst…