| """Pydantic v2 models for the Scheduling Optimisation Environment.""" | |
| from pydantic import BaseModel, Field | |
| class Observation(BaseModel): | |
| """What the agent sees at each step.""" | |
| schedule_instance: str = Field( | |
| description="JSON-encoded scheduling problem instance to evaluate" | |
| ) | |
| task_id: str = Field(description="Current task identifier") | |
| context: str = Field(description="Instructions or hints for the current step") | |
| step_number: int = Field(ge=0, description="Current step in the episode") | |
| class Action(BaseModel): | |
| """What the agent submits as a response.""" | |
| response: str = Field( | |
| description=( | |
| "Agent's answer: 'feasible'/'infeasible', a violation category, " | |
| "or a JSON repair schedule" | |
| ) | |
| ) | |
| task_id: str = Field(description="Task identifier this action is for") | |
| class Reward(BaseModel): | |
| """Grading result returned to the agent.""" | |
| score: float = Field(ge=0.0, le=1.0, description="Reward score in [0.0, 1.0]") | |
| feedback: str = Field(default="", description="Human-readable grading feedback") | |