Spaces:
Sleeping
Sleeping
File size: 14,816 Bytes
8c1f582 |
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 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 |
#!/usr/bin/env python3
"""
UN Motion Simulation Runner
This script runs a UN motion simulation where AI agents representing different countries
vote on resolutions and provide statements explaining their positions.
Usage:
python scripts/run_motion.py <motion_id> [--provider cloud|local] [--model MODEL_NAME]
Example:
python scripts/run_motion.py 01_gaza_ceasefire_resolution --provider cloud
"""
import argparse
import json
import os
import sys
from datetime import datetime
from pathlib import Path
from typing import Dict, List, Optional
import re
# Add project root to path
PROJECT_ROOT = Path(__file__).parent.parent
sys.path.insert(0, str(PROJECT_ROOT))
class MotionRunner:
"""Runs UN motion simulations with AI agents"""
VALID_VOTES = ["yes", "no", "abstain"]
def __init__(self, provider: str = "cloud", model: Optional[str] = None):
"""
Initialize the motion runner
Args:
provider: Either 'cloud' (API) or 'local' (local model)
model: Model name/identifier (optional, uses defaults if not specified)
"""
self.provider = provider
self.model = model
self.project_root = PROJECT_ROOT
self.agents_dir = self.project_root / "agents" / "representatives"
self.motions_dir = self.project_root / "tasks" / "motions"
self.results_dir = self.project_root / "tasks" / "reactions"
# Load configuration
self._load_config()
# Initialize AI client based on provider
self._init_ai_client()
def _load_config(self):
"""Load configuration from environment variables"""
from dotenv import load_dotenv
load_dotenv()
if self.provider == "cloud":
self.anthropic_api_key = os.getenv("ANTHROPIC_API_KEY")
self.openai_api_key = os.getenv("OPENAI_API_KEY")
self.api_base = os.getenv("API_BASE_URL")
if not self.model:
self.model = os.getenv("MODEL_NAME", "gpt-4")
# Determine which API to use based on model name
self.use_anthropic = self.model.startswith("claude")
else:
# Local model configuration
self.local_model_path = os.getenv("LOCAL_MODEL_PATH")
if not self.model:
self.model = os.getenv("LOCAL_MODEL_NAME", "llama3")
self.use_anthropic = False
def _init_ai_client(self):
"""Initialize the appropriate AI client"""
if self.provider == "cloud":
if self.use_anthropic:
try:
import anthropic
self.client = anthropic.Anthropic(api_key=self.anthropic_api_key)
print(f"β Initialized Anthropic API client (model: {self.model})")
except ImportError:
print("Error: anthropic package not installed. Run: pip install anthropic")
sys.exit(1)
else:
try:
# OpenAI or OpenRouter
import openai
if self.api_base:
self.client = openai.OpenAI(
api_key=self.openai_api_key,
base_url=self.api_base
)
else:
self.client = openai.OpenAI(api_key=self.openai_api_key)
print(f"β Initialized OpenAI API client (model: {self.model})")
except ImportError:
print("Error: openai package not installed. Run: pip install openai")
sys.exit(1)
else:
try:
# Use Ollama for local models
import ollama
self.client = ollama
print(f"β Initialized local model client (model: {self.model})")
except ImportError:
print("Error: ollama package not installed. Run: pip install ollama")
sys.exit(1)
def get_country_list(self) -> List[Dict[str, str]]:
"""Get list of all countries with agents"""
countries = []
for country_dir in sorted(self.agents_dir.iterdir()):
if country_dir.is_dir():
system_prompt_path = country_dir / "system-prompt.md"
if system_prompt_path.exists():
country_name = country_dir.name.replace("-", " ").title()
countries.append({
"name": country_name,
"slug": country_dir.name,
"prompt_path": str(system_prompt_path)
})
return countries
def load_motion(self, motion_id: str) -> Dict:
"""Load motion text from file"""
motion_path = self.motions_dir / f"{motion_id}.md"
if not motion_path.exists():
raise FileNotFoundError(f"Motion not found: {motion_path}")
with open(motion_path, 'r', encoding='utf-8') as f:
motion_text = f.read()
return {
"id": motion_id,
"text": motion_text,
"path": str(motion_path)
}
def load_agent_prompt(self, prompt_path: str) -> str:
"""Load agent system prompt"""
with open(prompt_path, 'r', encoding='utf-8') as f:
return f.read()
def query_agent(self, country: Dict, motion: Dict) -> Dict:
"""
Query an AI agent for their vote and statement
Returns:
Dict with 'vote' (yes/no/abstain) and 'statement' (brief explanation)
"""
system_prompt = self.load_agent_prompt(country['prompt_path'])
user_prompt = f"""You are voting on the following UN General Assembly resolution:
{motion['text']}
You must respond with a JSON object containing:
1. "vote": Your vote - must be exactly one of: "yes", "no", or "abstain"
2. "statement": A brief statement (2-4 sentences) explaining your country's position
IMPORTANT: Your statement must articulate {country['name']}'s UNIQUE perspective, national interests, and specific reasons for this vote. Reference your country's:
- Historical positions on this issue
- Regional concerns and alliances
- Domestic political considerations
- Specific clauses in the resolution that align with or contradict your interests
Avoid generic diplomatic language. Be specific to {country['name']}'s situation and worldview.
Your response must be valid JSON in this exact format:
{{
"vote": "yes",
"statement": "Your explanation here."
}}"""
try:
if self.provider == "cloud":
if self.use_anthropic:
# Anthropic API
response = self.client.messages.create(
model=self.model,
max_tokens=800,
system=system_prompt,
messages=[
{"role": "user", "content": user_prompt}
],
temperature=0.7
)
content = response.content[0].text
else:
# OpenAI API
response = self.client.chat.completions.create(
model=self.model,
messages=[
{"role": "system", "content": system_prompt},
{"role": "user", "content": user_prompt}
],
temperature=0.7,
max_tokens=800
)
content = response.choices[0].message.content
else:
response = self.client.chat(
model=self.model,
messages=[
{"role": "system", "content": system_prompt},
{"role": "user", "content": user_prompt}
]
)
content = response['message']['content']
# Extract JSON from response (handle markdown code blocks)
content = content.strip()
if content.startswith("```"):
# Remove markdown code block
content = re.sub(r'^```(?:json)?\n', '', content)
content = re.sub(r'\n```$', '', content)
# Parse JSON response
result = json.loads(content)
# Validate response
if "vote" not in result or "statement" not in result:
raise ValueError("Response missing required fields")
if result["vote"].lower() not in self.VALID_VOTES:
raise ValueError(f"Invalid vote: {result['vote']}")
result["vote"] = result["vote"].lower()
return result
except json.JSONDecodeError as e:
print(f" β JSON parse error for {country['name']}: {e}")
print(f" Raw response: {content[:200]}...")
return {
"vote": "abstain",
"statement": f"[Error: Unable to parse response]",
"error": str(e)
}
except Exception as e:
print(f" β Error querying {country['name']}: {e}")
return {
"vote": "abstain",
"statement": f"[Error: {str(e)}]",
"error": str(e)
}
def run_motion(self, motion_id: str, sample_size: Optional[int] = None) -> Dict:
"""
Run a motion through all country agents
Args:
motion_id: ID of the motion to run
sample_size: If set, only query this many countries (for testing)
Returns:
Dict containing all votes and metadata
"""
print(f"\n{'='*60}")
print(f"Running Motion: {motion_id}")
print(f"Provider: {self.provider} | Model: {self.model}")
print(f"{'='*60}\n")
# Load motion
motion = self.load_motion(motion_id)
print(f"β Loaded motion from {motion['path']}\n")
# Get countries
countries = self.get_country_list()
if sample_size:
countries = countries[:sample_size]
print(f"π Querying {sample_size} countries (sample mode)\n")
else:
print(f"π Querying {len(countries)} countries\n")
# Query each country
votes = []
vote_counts = {"yes": 0, "no": 0, "abstain": 0}
for i, country in enumerate(countries, 1):
print(f"[{i}/{len(countries)}] Querying {country['name']}...", end=" ", flush=True)
result = self.query_agent(country, motion)
vote_counts[result["vote"]] += 1
votes.append({
"country": country['name'],
"country_slug": country['slug'],
"vote": result["vote"],
"statement": result["statement"],
"error": result.get("error")
})
# Print vote result
vote_emoji = {"yes": "β
", "no": "β", "abstain": "βͺ"}
print(f"{vote_emoji[result['vote']]} {result['vote'].upper()}")
# Compile results
results = {
"motion_id": motion_id,
"motion_path": motion['path'],
"timestamp": datetime.utcnow().isoformat() + "Z",
"provider": self.provider,
"model": self.model,
"total_votes": len(countries),
"vote_summary": vote_counts,
"votes": votes
}
# Print summary
print(f"\n{'='*60}")
print(f"Vote Summary:")
print(f" YES: {vote_counts['yes']:3d} ({vote_counts['yes']/len(countries)*100:.1f}%)")
print(f" NO: {vote_counts['no']:3d} ({vote_counts['no']/len(countries)*100:.1f}%)")
print(f" ABSTAIN: {vote_counts['abstain']:3d} ({vote_counts['abstain']/len(countries)*100:.1f}%)")
print(f"{'='*60}\n")
return results
def save_results(self, results: Dict):
"""Save simulation results to file"""
# Create results directory if it doesn't exist
self.results_dir.mkdir(parents=True, exist_ok=True)
# Generate filename with timestamp
timestamp = datetime.utcnow().strftime("%Y%m%d_%H%M%S")
filename = f"{results['motion_id']}_{timestamp}.json"
filepath = self.results_dir / filename
# Save results
with open(filepath, 'w', encoding='utf-8') as f:
json.dump(results, f, indent=2, ensure_ascii=False)
print(f"β Results saved to: {filepath}")
# Also create/update a "latest" symlink or copy
latest_filepath = self.results_dir / f"{results['motion_id']}_latest.json"
with open(latest_filepath, 'w', encoding='utf-8') as f:
json.dump(results, f, indent=2, ensure_ascii=False)
print(f"β Latest results: {latest_filepath}")
def main():
parser = argparse.ArgumentParser(
description="Run UN motion simulation with AI agents",
formatter_class=argparse.RawDescriptionHelpFormatter,
epilog="""
Examples:
# Run with cloud API (default)
python scripts/run_motion.py 01_gaza_ceasefire_resolution
# Run with local model
python scripts/run_motion.py 01_gaza_ceasefire_resolution --provider local
# Test with only 5 countries
python scripts/run_motion.py 01_gaza_ceasefire_resolution --sample 5
# Use specific model
python scripts/run_motion.py 01_gaza_ceasefire_resolution --model gpt-4-turbo
"""
)
parser.add_argument(
"motion_id",
help="ID of the motion to run (e.g., 01_gaza_ceasefire_resolution)"
)
parser.add_argument(
"--provider",
choices=["cloud", "local"],
default="cloud",
help="AI provider: cloud (API) or local (Ollama)"
)
parser.add_argument(
"--model",
help="Model name (optional, uses config defaults)"
)
parser.add_argument(
"--sample",
type=int,
help="Only query N countries (for testing)"
)
args = parser.parse_args()
# Run simulation
try:
runner = MotionRunner(provider=args.provider, model=args.model)
results = runner.run_motion(args.motion_id, sample_size=args.sample)
runner.save_results(results)
print("\nβ Motion simulation complete!")
except FileNotFoundError as e:
print(f"\nβ Error: {e}", file=sys.stderr)
sys.exit(1)
except KeyboardInterrupt:
print("\n\nβ Simulation interrupted by user")
sys.exit(130)
except Exception as e:
print(f"\nβ Unexpected error: {e}", file=sys.stderr)
import traceback
traceback.print_exc()
sys.exit(1)
if __name__ == "__main__":
main()
|