rescommons/agent-orchestration-dataset

Agent Orchestration Dataset

A dataset for training and evaluating intelligent orchestrator models that route user requests to the correct specialist agents.

Dataset Description

This dataset contains conversational examples where an orchestrator must analyze user requests and determine which specialist agents should handle them. Each example includes:

• System prompt: Defines the orchestrator role and lists available agents
• User message: A natural language request from a user
• Model response (training only): The expected function call with selected agents
• Ground truth agents: The correct agents that should be routed to

Use Cases

• Customer Service: Routing to validation, duplicate detection, case creation, informational queries, transactional updates, sentiment analysis, entity extraction, document verification, and email agents
• HR/Leave Management: Routing to user information retrieval, balance checking, and leave approval agents

Dataset Structure

{
    "messages": [
        {"role": "system", "content": "You are an intelligent orchestrator..."},
        {"role": "user", "content": "User request here..."},
        {"role": "model", "content": "<start_function_call>call:route_to_agents{...}<end_function_call>"}  # Training only
    ],
    "ground_truth_agents": ["agent1", "agent2", ...]
}

Splits

Split	Examples
Train	611
Test	131

Available Agents

Customer Service Scenario

• request_validation_agent
• duplicate_detection_agent
• case_creation_agent
• informational_queries_agent
• transactional_query_responder_agent
• intent_and_sentiment_extraction_agent
• entity_extraction_agent
• document_verification_agent
• email_agent

HR/Leave Scenario

• user_information_retriever_agent
• balance_checking_agent
• leave_approval_agent

Usage

from datasets import load_dataset

dataset = load_dataset("V1rtucious/agent-orchestration-dataset")

# Access splits
train_data = dataset["train"]
test_data = dataset["test"]

# Example
print(train_data[0])

Intended Use

This dataset is designed for:

• Fine-tuning LLMs for multi-agent orchestration
• Benchmarking agent routing accuracy
• Training function-calling models (e.g., FunctionGemma)