| |
| """Rebuild 4-task offline eval dashboard from existing metrics.json files.""" |
| from __future__ import annotations |
|
|
| import json |
| from pathlib import Path |
|
|
| import matplotlib |
|
|
| matplotlib.use("Agg") |
| import matplotlib.pyplot as plt |
| import numpy as np |
|
|
| ROOT = Path(__file__).resolve().parents[1] |
| OUT = ROOT / "outputs/diag/task_offline" |
| TASKS = ["stack", "usb", "cleanup", "place"] |
| CKPT = { |
| "stack": "smolvla_rm65_stack_30ep_s1000", |
| "usb": "smolvla_rm65_usb_100ep_s40000", |
| "cleanup": "smolvla_rm65_cleanup_47ep_s40000", |
| "place": "smolvla_rm65_place_47ep_s40000", |
| } |
| ATTN = { |
| "stack": ROOT / "outputs/diag/smolvla_attn_stack30_f074", |
| "usb": ROOT / "outputs/diag/smolvla_attn_usb_f074", |
| "cleanup": ROOT / "outputs/diag/smolvla_attn_cleanup_f074", |
| "place": ROOT / "outputs/diag/smolvla_attn_place_f074", |
| } |
|
|
|
|
| def main() -> None: |
| plt.rcParams.update( |
| { |
| "figure.facecolor": "#111", |
| "axes.facecolor": "#161616", |
| "axes.edgecolor": "#555", |
| "text.color": "#EEE", |
| "xtick.color": "#BBB", |
| "ytick.color": "#BBB", |
| "axes.labelcolor": "#DDD", |
| "grid.color": "#2A2A2A", |
| "savefig.facecolor": "#111", |
| "font.family": "DejaVu Sans", |
| } |
| ) |
| all_metrics = {} |
| coverage = {} |
| for name in TASKS: |
| mets = json.loads((OUT / name / "metrics.json").read_text()) |
| all_metrics[name] = mets |
| traj_ok = all((OUT / name / f"ep{ep:03d}_trajectory.png").is_file() for ep in (0, 1, 2)) |
| attn_dir = ATTN[name] |
| attn_ok = (attn_dir / "lang2img_camera1.png").is_file() |
| coverage[name] = { |
| "ckpt": CKPT[name], |
| "traj": traj_ok, |
| "attn": attn_ok, |
| "metrics": True, |
| "attn_path": str(attn_dir.relative_to(ROOT / "outputs/diag")), |
| } |
|
|
| names = list(all_metrics) |
| means = [float(np.mean([m["joint_mae_vs_demo_deg"] for m in all_metrics[k]])) for k in names] |
| grips = [float(np.mean([m["gripper_agree"] for m in all_metrics[k]])) * 100 for k in names] |
| fig, axes = plt.subplots(1, 2, figsize=(12.5, 4.4)) |
| fig.suptitle("RM65 SmolVLA — 4-task offline open-loop summary", fontsize=15, fontweight="bold") |
| axes[0].bar(names, means, color=["#50B4FF", "#FFC850", "#B48CFF", "#FF648C"]) |
| axes[0].set_ylabel("mean joint MAE vs demo (deg)") |
| axes[0].grid(True, axis="y", alpha=0.3) |
| for i, v in enumerate(means): |
| axes[0].text(i, v + 0.02, f"{v:.2f}", ha="center", fontsize=10, color="#D2C878") |
| axes[1].bar(names, grips, color=["#50FFA0", "#78DCDC", "#50B4FF", "#FFC850"]) |
| axes[1].set_ylabel("mean gripper agree (%)") |
| axes[1].set_ylim(0, 105) |
| axes[1].grid(True, axis="y", alpha=0.3) |
| for i, v in enumerate(grips): |
| axes[1].text(i, v + 1.5, f"{v:.0f}%", ha="center", fontsize=10, color="#D2C878") |
| fig.savefig(OUT / "cross_task_mae.png", dpi=150, bbox_inches="tight", pad_inches=0.22) |
| plt.close(fig) |
|
|
| summary = { |
| "tasks": { |
| k: { |
| "episodes": mets, |
| "mean_joint_mae_vs_demo_deg": float(np.mean([m["joint_mae_vs_demo_deg"] for m in mets])), |
| "mean_gripper_agree": float(np.mean([m["gripper_agree"] for m in mets])), |
| } |
| for k, mets in all_metrics.items() |
| }, |
| "coverage": coverage, |
| } |
| (OUT / "summary.json").write_text(json.dumps(summary, indent=2, ensure_ascii=False) + "\n") |
|
|
| rows = [] |
| for name, mets in all_metrics.items(): |
| mean_mae = float(np.mean([m["joint_mae_vs_demo_deg"] for m in mets])) |
| mean_grip = float(np.mean([m["gripper_agree"] for m in mets])) |
| cov = coverage[name] |
| rows.append( |
| "<tr>" |
| f"<td><b>{name}</b></td><td>{cov['ckpt']}</td>" |
| f"<td>{'Y' if cov['traj'] else 'N'}</td>" |
| f"<td>{'Y' if cov['attn'] else 'pending'}</td><td>Y</td>" |
| f"<td>{mean_mae:.2f}</td><td>{mean_grip:.0%}</td>" |
| f"<td><a href='{name}/index.html'>{name}/</a> " |
| f"<a href='../{cov['attn_path']}/'>attn</a></td></tr>" |
| ) |
| (OUT / "index.html").write_text( |
| "<html><head><meta charset='utf-8'><title>RM65 SmolVLA task offline eval</title></head>" |
| "<body style='background:#111;color:#eee;font-family:sans-serif;padding:20px;max-width:1100px'>" |
| "<h2>RM65 SmolVLA 四任务离线开环评测汇总</h2>" |
| "<p>open-loop：n_action_steps=16，cleaned demo ep000–002。</p>" |
| "<p><img src='cross_task_mae.png' style='max-width:100%;border:1px solid #333'/></p>" |
| "<table border='1' cellpadding='8' style='border-collapse:collapse;width:100%'>" |
| "<tr><th>任务</th><th>ckpt</th><th>轨迹</th><th>注意力</th><th>指标</th>" |
| "<th>mean MAE°</th><th>grip</th><th>链接</th></tr>" |
| + "".join(rows) |
| + "</table><h3>指标速览</h3><ul>" |
| + "".join( |
| f"<li><b>{k}</b>: MAE {v['mean_joint_mae_vs_demo_deg']:.2f}° / " |
| f"grip {v['mean_gripper_agree']:.0%}</li>" |
| for k, v in summary["tasks"].items() |
| ) |
| + "</ul></body></html>\n", |
| encoding="utf-8", |
| ) |
| for k, v in summary["tasks"].items(): |
| print( |
| f"{k}: MAE={v['mean_joint_mae_vs_demo_deg']:.2f} " |
| f"grip={v['mean_gripper_agree']:.0%} " |
| f"traj={coverage[k]['traj']} attn={coverage[k]['attn']}" |
| ) |
| print("OK", OUT / "index.html") |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|
