File size: 2,521 Bytes
8cd3fa7 511f04a 8cd3fa7 a541c0b 8cd3fa7 511f04a 8cd3fa7 28e7c64 8cd3fa7 8520614 8cd3fa7 b978fbd 8cd3fa7 | 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 88 89 | name: "PolicyEvolverEnv"
description: "Policy Design and Evolution Sandbox — agents refine their strategy to evolve real-world governance frameworks through meta-reasoning"
version: "1.0.0"
author: "PolicyEvolution Team"
tags:
- "policy"
- "governance"
- "meta-reasoning"
- "content-moderation"
- "AI-safety"
environment:
module: "server.environment"
class: "PolicyEvolverEnvironment"
variables:
HF_TOKEN:
description: "API key for LLM inference provider (Groq recommended)"
required: true
API_BASE_URL:
description: "OpenAI-compatible endpoint. Default: Groq"
default: "https://api.groq.com/openai/v1"
MODEL_NAME:
description: "Model identifier for the inference provider"
default: "llama-3.1-8b-instant"
observation_schema:
type: "object"
description: "Policy context, data corpus, and system state"
action_schema:
type: "object"
description: "Discriminated union on action_type field"
discriminator: "action_type"
variants:
- action_type: "propose_clarification"
schema: "ProposeClarificationAction"
- action_type: "propose_new_rule"
schema: "ProposeNewRuleAction"
- action_type: "evolve_policy"
schema: "EvolveProcessAction"
description: "Hard task metric keys: fraud_rate (aliases: fraud_detection, fraud), revenue_velocity (aliases: queue_overload, revenue), seller_trust (aliases: seller_confidence, trust)."
reward_range: [0.0, 1.0]
runtime:
max_steps: 5
timeout_seconds: 1200
vcpu: 2
memory_gb: 8
tasks:
- id: "task_easy"
difficulty: "easy"
description: "Identify and clarify ambiguous policy terms in a social media community guidelines"
expected_min_score: 0.70
- id: "task_medium"
difficulty: "medium"
description: "Detect policy gaps in corporate HR policies and propose new rules for emerging scenarios"
expected_min_score: 0.55
- id: "task_hard"
difficulty: "hard"
description: "Holistically evolve an e-commerce Trust & Safety framework with trade-off reasoning"
expected_min_score: 0.40
grading:
module: "server.grader"
function: "grade"
return_range: [0.0, 1.0]
endpoints:
required:
- path: "/reset"
method: "POST"
- path: "/step"
method: "POST"
- path: "/state"
method: "GET"
- path: "/tasks"
method: "GET"
- path: "/grader"
method: "POST"
- path: "/baseline"
method: "GET"
optional:
- path: "/health"
method: "GET"
|