Calibrated accommodation pairs reward dialect matching in post-training
Across 10 open lineages, adjusted DPO still favored British and Australian spelling over AAE and Singlish.
1 stories, newest first · one card per story
Across 10 open lineages, adjusted DPO still favored British and Australian spelling over AAE and Singlish.
Today 1.5K readers · 2.4K views · 757 reads · all time 189K views