Spaces:
Running
Running
File size: 2,440 Bytes
df97e68 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 | spec_version: 1
name: gov-workflow-openenv
version: "0.3.0"
type: space
runtime: fastapi
app: app.main:app
port: 7860
description: >
A real-world OpenEnv environment for government-service workflow control.
The agent optimizes avoidable administrative delay via priority control,
document resolution, staffing, escalation, and fairness balancing.
entrypoint:
module: app.main
object: app
inference_script: inference.py
environment:
class: server.gov_environment.GovWorkflowOpenEnv
observation_model: app.models.ObservationModel
action_model: app.models.ActionModel
reward_model: app.models.RewardModel
state_model: app.models.EpisodeStateModel
step_info_model: app.models.StepInfoModel
tasks:
- id: district_backlog_easy
seed: 11
description: >
Small district office with 3 services and generous SLA windows.
Tests baseline queue control and document handling.
grader: app.graders.grade_easy
- id: mixed_urgency_medium
seed: 22
description: >
Mid-sized office with mixed urgency and tighter fairness requirements.
Tests urgency prioritization and staffing trade-offs.
grader: app.graders.grade_medium
- id: cross_department_hard
seed: 33
description: >
Large cross-department office with high arrivals and strict fairness.
Tests escalation discipline and multi-queue balancing.
grader: app.graders.grade_hard
api:
endpoints:
- method: GET
path: /health
description: Server and session health check
- method: POST
path: /reset
description: Create a new episode session and return initial observation
- method: POST
path: /step
description: Apply one action and advance simulation state
- method: GET
path: /state
description: Return current episode state (query param session_id)
- method: POST
path: /state
description: Return current episode state (body with session_id)
- method: POST
path: /grade
description: Run deterministic task grader for the current episode
metadata:
domain: government-services
real_world: true
reward_type: dense
action_space: discrete
observation_space: structured
deterministic_tasks: true
deterministic_graders: true
num_tasks: 3
framework: fastapi
language: python
deployment:
host: 0.0.0.0
port: 7860
dockerfile: Dockerfile
platform: huggingface-spaces
runtime: docker
|