モデルアーキテクチャ
Encoder-only Transformers
Let every input token attend to tokens on both sides, producing contextual representations rather than an open-ended generation loop.
理解のためのモデル
Read the entire input, contextualize each position, then pool or attach a task head.
データフロー
- Token sequence
- Bidirectional self-attention blocks
- Contextual token states
- Pooler or task head
- Class, span, score, or vector
学習方法
Masked-token reconstruction is the classic pretraining objective, followed by task-specific classification, token labeling, span prediction, or metric learning.
推論の実行方法
One forward pass processes the available input. It is naturally suited to understanding and representation, but does not by itself generate an arbitrary continuation.
強み
- Rich bidirectional context for classification and extraction
- Parallel processing across the input
- Often efficient for high-volume bounded tasks
トレードオフ
- Not an open-ended generator
- Maximum input length and quadratic attention can constrain long documents
- A task head or pooling strategy is usually required
適する場合
- Classification, named entities, reranking, or embeddings dominate
- The complete input is available before prediction
- A smaller specialized model can meet the rubric
避ける・再検討する場合
- The primary output is long free-form text
- Streaming generation is required
- Inputs routinely exceed the evaluated context strategy
公開された方式の例
- • BERT and RoBERTa research families
- • Encoder-based rerankers and classifiers
よく組み合わせる要素
Embedding modelsVision encodersContrastive / dual encodersRetrieval systems