every time you: • say a model is "highly accurate" but have no evals • finetune a decoder LLM for classification without trying BERT-style classifiers • use only embeddings without trying text for retrieval, matching, dedup • optimize for sota/complexity instead of solving… https://t.co/ePbRaYYb8c
— Eugene Yan (@eugeneyan) Oct 4, 2024
from Twitter https://twitter.com/eugeneyan
October 04, 2024 at 12:39AM
via IFTTT
No comments:
Post a Comment