Spaces:
Sleeping
Sleeping
| spec_version: 1 | |
| name: customer_support_benchmark | |
| type: space | |
| runtime: fastapi | |
| app: app:app | |
| port: 8000 | |
| tasks: | |
| - id: easy_classify_respond | |
| difficulty: easy | |
| objective: Classify ticket correctly and provide the expected response. | |
| grader: grade_easy | |
| - id: medium_kb_grounded_response | |
| difficulty: medium | |
| objective: Use the correct KB entry and provide a grounded response. | |
| grader: grade_medium | |
| - id: hard_escalation_safety | |
| difficulty: hard | |
| objective: Make the right escalation decision and keep the response safe/polite. | |
| grader: grade_hard | |
| graders: | |
| - name: grade_easy | |
| module: tasks | |
| function: grade_easy | |
| - name: grade_medium | |
| module: tasks | |
| function: grade_medium | |
| - name: grade_hard | |
| module: tasks | |
| function: grade_hard |