3 stories tagged bert, newest first.
33 points by surprisetalk | 1 comments
Marc Brooker's blog post explaining why the strands-decider-2B classifier uses a modified decoder LLM rather than a BERT-style encoder, comparing five designs.
435 points by nathan-barry | 102 comments
Blog post arguing BERT-style masked language modeling is a single step of discrete text diffusion, with experiments fine-tuning RoBERTa as a diffusion text generator.
341 points by ks2048 | 109 comments
Blog post arguing the "gzip beats BERT" text classification paper reports top-2 accuracy instead of kNN(k=2) accuracy due to a code bug, inflating results.