| The decoder generates a token sequence given a prompt. The prompt is represented by a special token for each downstream task. For example, document parsing has a special parsing token that is combined with the encoder hidden states to parse the document into a structured output format (JSON). | |
| Reinforcement learning | |
| Decoder[[rl-decoder]] | |
| The Decision and Trajectory Transformer casts the state, action, and reward as a sequence modeling problem. |