1Model categories: what are Encoder-only models (examples: BERT, RoBERTa) and when are they used (typically for classification and embeddings)?
2Model categories: what are Decoder-only models (examples: GPT, Llama) and when are they used (typically for text generation)?
3Model categories: what are Encoder-decoder models (examples: T5, BART) and when are they used (typically for translation and summarization)?