"""Summarize runs: final/min val loss per run, plus tokens needed to reach target losses.""" import argparse import json from pathlib import Path def load(run_dir): recs = [json.loads(l) for l in (run_dir / "metrics.jsonl").read_text().splitlines() if l.strip()] val = [(r["tokens"], r["val_loss"]) for r in recs if "val_loss" in r] return val def main(): p = argparse.ArgumentParser() p.add_argument("runs_dir") p.add_argument("--targets", default="4.0,3.6,3.4") args = p.parse_args() targets = [float(t) for t in args.targets.split(",")] rows = [] for d in sorted(Path(args.runs_dir).iterdir()): if not (d / "metrics.jsonl").exists(): continue val = load(d) if not val: continue reach = [next((t for t, v in val if v <= tg), None) for tg in targets] rows.append((d.name, "done" if (d / "DONE").exists() else "running", val[0][1], val[-1][0], val[-1][1], reach)) hdr = f"{'run':40s} {'status':8s} {'init_loss':>9s} {'tokens':>8s} {'val':>7s} " + " ".join(f"{'@' + str(t):>8s}" for t in targets) print(hdr) for name, st, v0, tok, v, reach in rows: r = " ".join(f"{(f'{x / 1e6:.0f}M' if x else '-'):>8s}" for x in reach) print(f"{name:40s} {st:8s} {v0:9.3f} {tok / 1e6:7.0f}M {v:7.3f} {r}") if __name__ == "__main__": main()