Spaces:
Runtime error
Runtime error
File size: 9,312 Bytes
69ee6e4 | 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 | from __future__ import annotations
import json
from pathlib import Path
import typer
from rich.console import Console
from rich.table import Table
from llm_eval.connectivity import (
check_connectivity,
diagnose_openai_endpoints,
get_key_debug_info,
)
from llm_eval.config import build_run_manifest, load_run_config, resolve_provider_keys
from llm_eval.reporting import write_reports
from llm_eval.runner import run_evaluation
from llm_eval.scoring import load_results, load_summary, score_results
from llm_eval.stats import add_confidence_intervals, pairwise_significance
app = typer.Typer(help="LLM multi-model evaluation framework CLI.")
console = Console()
@app.command("validate-config")
def validate_config(
config_path: str = typer.Option(
"configs/run.example.yaml", "--config", "-c", help="Path to run config YAML."
),
) -> None:
"""Validate a run configuration and print a manifest preview."""
config = load_run_config(config_path)
manifest = build_run_manifest(config)
console.print("[green]Config is valid.[/green]")
console.print_json(data=manifest.model_dump())
@app.command("check-keys")
def check_keys(
config_path: str = typer.Option(
"configs/run.example.yaml", "--config", "-c", help="Path to run config YAML."
),
env_path: str = typer.Option(".env", "--env", help="Path to .env file."),
) -> None:
"""Check whether required provider keys are present in env."""
config = load_run_config(config_path)
key_status = resolve_provider_keys(config=config, env_path=env_path)
table = Table(title="Provider Key Availability")
table.add_column("Provider")
table.add_column("Key Available")
for provider_name, is_available in key_status.items():
table.add_row(provider_name, "yes" if is_available else "no")
console.print(table)
@app.command("scaffold-config")
def scaffold_config(
output_path: str = typer.Option(
"configs/run.example.yaml", "--out", "-o", help="Output path for scaffold config."
)
) -> None:
"""Write a starter run config for Anthropic and Gemini."""
target = Path(output_path)
target.parent.mkdir(parents=True, exist_ok=True)
if target.exists():
raise typer.BadParameter(f"{output_path} already exists.")
template = {
"run_name": "phase1-baseline",
"seed": 42,
"benchmark": {
"name": "mmlu_subset",
"split": "dev",
"dataset_path": "data/benchmarks/mmlu_subset/dev.jsonl",
"max_samples": 50,
},
"providers": [
{
"provider": "anthropic",
"model": "claude-3-5-haiku-latest",
"api_key_env": "ANTHROPIC_API_KEY",
"temperature": 0.0,
"max_tokens": 512,
},
{
"provider": "gemini",
"model": "gemini-2.0-flash",
"api_key_env": "GEMINI_API_KEY",
"temperature": 0.0,
"max_tokens": 512,
},
],
}
import yaml
target.write_text(yaml.safe_dump(template, sort_keys=False), encoding="utf-8")
console.print(f"[green]Created[/green] {output_path}")
@app.command("print-policy")
def print_policy(policy_path: str = typer.Option("configs/policy.yaml", "--policy")) -> None:
"""Print policy YAML as JSON for easy inspection."""
import yaml
data = yaml.safe_load(Path(policy_path).read_text(encoding="utf-8")) or {}
console.print_json(data=json.loads(json.dumps(data)))
@app.command("run")
def run(
config_path: str = typer.Option(
"configs/run.example.yaml", "--config", "-c", help="Path to run config YAML."
),
policy_path: str = typer.Option(
"configs/policy.yaml", "--policy", help="Path to policy YAML."
),
artifacts_root: str = typer.Option(
"artifacts", "--artifacts-root", help="Directory for run artifacts."
),
env_path: str = typer.Option(".env", "--env", help="Path to environment file."),
) -> None:
"""Execute a benchmark run and persist artifacts."""
config = load_run_config(config_path)
summary = run_evaluation(
config=config,
policy_path=policy_path,
artifacts_root=artifacts_root,
env_path=env_path,
)
table = Table(title=f"Run Summary ({summary.run_id})")
table.add_column("System")
table.add_column("Attempted")
table.add_column("Correct")
table.add_column("Errors")
table.add_column("Accuracy")
for system_id, metrics in summary.provider_metrics.items():
table.add_row(
system_id,
str(metrics.get("attempted", 0)),
str(metrics.get("correct", 0)),
str(metrics.get("errors", 0)),
f"{metrics.get('accuracy', 0.0):.3f}",
)
console.print(table)
console.print(f"Artifacts written under [bold]{artifacts_root}/runs/{summary.run_id}[/bold]")
@app.command("check-connectivity")
def check_connectivity_command(
config_path: str = typer.Option(
"configs/run.example.yaml", "--config", "-c", help="Path to run config YAML."
),
env_path: str = typer.Option(".env", "--env", help="Path to environment file."),
) -> None:
"""Execute lightweight provider calls to validate live connectivity."""
config = load_run_config(config_path)
results = check_connectivity(config=config, env_path=env_path)
table = Table(title="Provider Connectivity")
table.add_column("Provider")
table.add_column("Model")
table.add_column("Status")
table.add_column("Detail")
for result in results:
table.add_row(
result.provider,
result.model,
"PASS" if result.ok else "FAIL",
result.detail,
)
console.print(table)
@app.command("debug-key")
def debug_key(
env_var: str = typer.Option(
"OPENAI_API_KEY", "--env-var", help="Environment variable name to inspect."
),
env_path: str = typer.Option(".env", "--env", help="Path to environment file."),
) -> None:
"""Print masked key metadata (never prints full key)."""
info = get_key_debug_info(env_var=env_var, env_path=env_path)
table = Table(title=f"Key Debug ({info.env_var})")
table.add_column("Field")
table.add_column("Value")
masked = (
f"{info.starts_with}...{info.ends_with}"
if info.present and info.length > 11
else ("(short)" if info.present else "(missing)")
)
table.add_row("present", "yes" if info.present else "no")
table.add_row("length", str(info.length))
table.add_row("fingerprint", masked)
table.add_row("has_whitespace", "yes" if info.has_whitespace else "no")
console.print(table)
@app.command("diagnose-openai")
def diagnose_openai(
env_var: str = typer.Option(
"OPENAI_API_KEY", "--env-var", help="Environment variable containing OpenAI key."
),
model: str = typer.Option("gpt-4o-mini", "--model", help="Model for completion check."),
env_path: str = typer.Option(".env", "--env", help="Path to environment file."),
project_id_env_var: str = typer.Option(
"OPENAI_PROJECT_ID",
"--project-id-env-var",
help="Env var name containing OpenAI project id.",
),
organization_id_env_var: str = typer.Option(
"OPENAI_ORG_ID",
"--organization-id-env-var",
help="Env var name containing OpenAI organization id.",
),
) -> None:
"""Diagnose OpenAI key behavior on models vs chat completions endpoints."""
diagnostics = diagnose_openai_endpoints(
env_var=env_var,
model=model,
env_path=env_path,
project_id_env_var=project_id_env_var,
organization_id_env_var=organization_id_env_var,
)
table = Table(title="OpenAI Endpoint Diagnostics")
table.add_column("Endpoint")
table.add_column("Status")
table.add_column("HTTP")
table.add_column("Detail")
for item in diagnostics:
table.add_row(
item.endpoint,
"PASS" if item.ok else "FAIL",
str(item.status_code) if item.status_code is not None else "-",
item.detail,
)
console.print(table)
@app.command("report")
def report(
run_id: str = typer.Option(..., "--run-id", help="Run id from artifacts/runs/<run_id>."),
artifacts_root: str = typer.Option("artifacts", "--artifacts-root"),
reports_root: str = typer.Option("reports", "--reports-root"),
) -> None:
"""Generate markdown/html/json reports from run artifacts."""
run_dir = Path(artifacts_root) / "runs" / run_id
if not run_dir.exists():
raise typer.BadParameter(f"Run directory does not exist: {run_dir}")
results = load_results(run_dir)
summary = load_summary(run_dir)
scored = score_results(results, summary)
scored = add_confidence_intervals(scored)
pairwise = pairwise_significance(results)
outputs = write_reports(
run_id=run_id,
scored=scored,
pairwise=pairwise,
reports_root=reports_root,
)
table = Table(title=f"Report Outputs ({run_id})")
table.add_column("Format")
table.add_column("Path")
for fmt, path in outputs.items():
table.add_row(fmt, path)
console.print(table)
|