Spaces:
Sleeping
Sleeping
File size: 5,146 Bytes
7414383 5fcd0ee 639b641 7414383 5fcd0ee 7414383 639b641 7414383 | 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 | #!/usr/bin/env python3
import argparse
import json
import sys
from pathlib import Path
from dotenv import load_dotenv
from rich.console import Console
from rich.table import Table
from rich import box
from rich.panel import Panel
load_dotenv()
sys.path.insert(0, str(Path(__file__).parent.parent.parent))
from viral_script_engine.agents.critic import CriticAgent, CriticParseError
from viral_script_engine.evaluation.critic_evaluator import CriticEvaluator
console = Console()
BASE_DIR = Path(__file__).parent.parent
def load_scripts(dry_run: bool) -> list:
scripts_path = BASE_DIR / "data" / "test_scripts" / "scripts.json"
with open(scripts_path) as f:
scripts = json.load(f)
if dry_run:
scripts = scripts[:2]
return scripts
def run_gate(max_retries: int = 3, dry_run: bool = False, backend: str = "groq", model_name: str = "llama-3.3-70b-versatile") -> bool:
agent = CriticAgent(backend=backend, model_name=model_name)
evaluator = CriticEvaluator()
scripts = load_scripts(dry_run)
table = Table(title="Critic Gate Results", box=box.ROUNDED)
table.add_column("Script ID", style="cyan", no_wrap=True)
table.add_column("Claims", justify="center")
table.add_column("Specificity", justify="center")
table.add_column("Falsifiability", justify="center")
table.add_column("Gate", justify="center")
outputs = []
all_results = []
with console.status("[bold green]Running CriticAgent on scripts...") as status:
for entry in scripts:
sid = entry["script_id"]
status.update(f"[bold green]Processing {sid}...")
critique_output = None
for attempt in range(max_retries):
try:
critique_output = agent.critique(
script=entry["script_text"],
region=entry["region"],
platform=entry["platform"],
niche=entry["niche"],
)
break
except CriticParseError as e:
if attempt == max_retries - 1:
console.print(f"[red]FAILED {sid} after {max_retries} attempts: {e}")
else:
console.print(f"[yellow]Retry {attempt + 1} for {sid}")
if critique_output is None:
continue
result = evaluator.evaluate(critique_output, entry["script_text"], script_id=sid)
all_results.append(result)
outputs.append((sid, entry, critique_output))
gate_str = "[green]PASS[/green]" if result.passes_gate else "[red]FAIL[/red]"
table.add_row(
sid,
str(result.claim_count),
f"{result.specificity_score:.2f}",
f"{result.falsifiability_score:.2f}",
gate_str,
)
console.print(table)
pass_count = sum(1 for r in all_results if r.passes_gate)
pass_rate = pass_count / len(all_results) if all_results else 0.0
overall_pass = pass_rate >= 0.8
if overall_pass:
fixtures_dir = BASE_DIR / "data" / "golden_fixtures"
fixtures_dir.mkdir(exist_ok=True)
for sid, entry, critique_output in outputs:
fixture_path = fixtures_dir / f"fixture_{sid}.json"
fixture_data = {
"script_id": sid,
"region": entry["region"],
"platform": entry["platform"],
"niche": entry["niche"],
"critique": critique_output.model_dump(),
}
with open(fixture_path, "w") as f:
json.dump(fixture_data, f, indent=2)
console.print(f"[green]Golden fixtures saved to {fixtures_dir}")
failing = [r.script_id for r in all_results if not r.passes_gate]
if failing:
console.print(f"[red]Failing scripts: {', '.join(failing)}")
for r in all_results:
if not r.passes_gate:
console.print(
f" [yellow]{r.script_id}[/yellow]: "
f"claims={r.claim_count}, specificity={r.specificity_score:.2f}, "
f"falsifiability={r.falsifiability_score:.2f}"
)
gate_label = f"PHASE 0 GATE: {'PASS' if overall_pass else 'FAIL'}"
style = "bold green" if overall_pass else "bold red"
console.print(Panel(f"[{style}]{gate_label}[/{style}] ({pass_count}/{len(all_results)} scripts passed)"))
return overall_pass
def main():
parser = argparse.ArgumentParser(description="Run Critic quality gate")
parser.add_argument("--max-retries", type=int, default=3)
parser.add_argument("--dry-run", action="store_true")
parser.add_argument("--backend", default="groq", choices=["groq", "qwen", "anthropic", "openai"])
parser.add_argument("--model-name", default="llama-3.3-70b-versatile")
args = parser.parse_args()
passed = run_gate(max_retries=args.max_retries, dry_run=args.dry_run, backend=args.backend, model_name=args.model_name)
sys.exit(0 if passed else 1)
if __name__ == "__main__":
main()
|