BERT based models are Encoder only which are well suited for text classification, and Semantic Text similarity search (If fine-tuned via sBERT). I want to know whether decoder only models like Llama2, GPT can be fine-tuned to do well on STS benchmark. If yes, does it perform better than fine-tuning encoder-only models?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.