{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "99d3910b",
   "metadata": {},
   "source": [
    "# 13 Pi0.5 随机位置、coherent recovery 与 EEF-delta\n",
    "\n",
    "这一节复现四个判断：\n",
    "\n",
    "1. 为什么逐帧反事实 oracle 标签不能直接拼成 action chunk；\n",
    "2. 为什么世界坐标 `eef_abs` 会把接近杯子和搬运到盘子两个相反阶段混在一起；\n",
    "3. 如何用 MuJoCo FK 把原轨迹精确转换成局部 `eef_delta`；\n",
    "4. 如何把离线 MAE、首帧方向和全新随机种子 strict rollout 放在同一张证据表里。\n",
    "\n",
    "这里的策略输入只有双相机图像、语言和 6D robot proprio。评估不使用 target/plate 坐标、GT phase、oracle prefix、外置 head 或 scripted finisher。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "24218d37",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.198752Z",
     "iopub.status.busy": "2026-07-11T22:38:34.198609Z",
     "iopub.status.idle": "2026-07-11T22:38:34.584881Z",
     "shell.execute_reply": "2026-07-11T22:38:34.584307Z"
    }
   },
   "outputs": [],
   "source": [
    "import json\n",
    "from pathlib import Path\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "\n",
    "plt.rcParams[\"figure.dpi\"] = 130\n",
    "plt.rcParams[\"axes.unicode_minus\"] = False\n",
    "plt.rcParams[\"font.sans-serif\"] = [\"Noto Sans CJK SC\", \"Microsoft YaHei\", \"DejaVu Sans\"]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f78bebb0",
   "metadata": {},
   "source": [
    "## 1. Action-chunk DAgger 必须保持时间连续\n",
    "\n",
    "错误采法让 student action 推进环境，却只在每个 student state 上计算一个没有执行的 oracle 单步标签。每个标签单独看都合理，但相邻标签不属于同一条状态转移链，不能作为 Pi0.5 的未来 50 步动作块。\n",
    "\n",
    "修正后，Pi0.5 先执行短 prefix，专家从 takeover state 开始真正接管；训练集只保存连续执行的 oracle suffix。失败轨迹单独保留为诊断，不默认混入成功 BC 数据。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "28107c3a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.586834Z",
     "iopub.status.busy": "2026-07-11T22:38:34.586675Z",
     "iopub.status.idle": "2026-07-11T22:38:34.594442Z",
     "shell.execute_reply": "2026-07-11T22:38:34.594115Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>采集方式</th>\n",
       "      <th>下一状态执行者</th>\n",
       "      <th>标签状态</th>\n",
       "      <th>结论</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>反事实逐帧标签</td>\n",
       "      <td>student</td>\n",
       "      <td>oracle 未执行</td>\n",
       "      <td>不能组成未来 action chunk</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>连续专家接管</td>\n",
       "      <td>oracle</td>\n",
       "      <td>oracle 连续执行</td>\n",
       "      <td>可以监督 chunk policy</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      采集方式  下一状态执行者         标签状态                   结论\n",
       "0  反事实逐帧标签  student   oracle 未执行  不能组成未来 action chunk\n",
       "1   连续专家接管   oracle  oracle 连续执行    可以监督 chunk policy"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dagger_rows = [\n",
    "    (\"反事实逐帧标签\", \"student\", \"oracle 未执行\", \"不能组成未来 action chunk\"),\n",
    "    (\"连续专家接管\", \"oracle\", \"oracle 连续执行\", \"可以监督 chunk policy\"),\n",
    "]\n",
    "pd.DataFrame(dagger_rows, columns=[\"采集方式\", \"下一状态执行者\", \"标签状态\", \"结论\"])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2f437c77",
   "metadata": {},
   "source": [
    "## 2. 审计 `eef_abs` 的阶段方向\n",
    "\n",
    "随机位置完整轨迹中，蓝杯始终在正 Y，盘子始终在负 Y。前段应朝杯子移动，后段则必须转向盘子。若首帧预测直接为负 Y，模型学到的是错误阶段，不是简单的动作噪声。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "45380077",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.595349Z",
     "iopub.status.busy": "2026-07-11T22:38:34.595261Z",
     "iopub.status.idle": "2026-07-11T22:38:34.735778Z",
     "shell.execute_reply": "2026-07-11T22:38:34.735453Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>区间</th>\n",
       "      <th>absolute action Y 均值 (m)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0-20%</td>\n",
       "      <td>0.153594</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20-40%</td>\n",
       "      <td>0.265544</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>40-60%</td>\n",
       "      <td>0.195898</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>60-80%</td>\n",
       "      <td>-0.093735</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>80-100%</td>\n",
       "      <td>-0.182327</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        区间  absolute action Y 均值 (m)\n",
       "0    0-20%                  0.153594\n",
       "1   20-40%                  0.265544\n",
       "2   40-60%                  0.195898\n",
       "3   60-80%                 -0.093735\n",
       "4  80-100%                 -0.182327"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 832x624 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>前80帧朝目标比例</th>\n",
       "      <th>首帧预测为负Y</th>\n",
       "      <th>首帧probe总数</th>\n",
       "      <th>首帧平均Y偏差(m)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.0</td>\n",
       "      <td>6</td>\n",
       "      <td>6</td>\n",
       "      <td>-0.108572</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   前80帧朝目标比例  首帧预测为负Y  首帧probe总数  首帧平均Y偏差(m)\n",
       "0        1.0        6          6   -0.108572"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "direction = {'episodes': 64, 'frames': 26014, 'target_y_positive_fraction': 1.0, 'plate_y_negative_fraction': 1.0, 'early_toward_target_fraction': 1.0, 'progress_action_y_mean': {'0-20%': 0.15359357784407415, '20-40%': 0.2655442285989148, '40-60%': 0.19589836334684946, '60-80%': -0.09373453054878639, '80-100%': -0.18232691485479655}, 'negative_start_probe_count': 6, 'start_probe_count': 6, 'mean_start_y_bias': -0.10857185496327777}\n",
    "direction_table = pd.DataFrame({\n",
    "    '区间': list(direction['progress_action_y_mean']),\n",
    "    'absolute action Y 均值 (m)': list(direction['progress_action_y_mean'].values()),\n",
    "})\n",
    "display(direction_table)\n",
    "ax = direction_table.plot.bar(x='区间', y='absolute action Y 均值 (m)', legend=False, color=['#3377aa', '#3377aa', '#3377aa', '#cc6644', '#cc6644'])\n",
    "ax.axhline(0, color='black', linewidth=0.8)\n",
    "ax.set_title('Pi0.5 EEF-absolute 标签随任务进度发生方向反转')\n",
    "ax.set_xlabel('episode 进度')\n",
    "plt.tight_layout()\n",
    "plt.show()\n",
    "pd.DataFrame([{\n",
    "    '前80帧朝目标比例': direction['early_toward_target_fraction'],\n",
    "    '首帧预测为负Y': direction['negative_start_probe_count'],\n",
    "    '首帧probe总数': direction['start_probe_count'],\n",
    "    '首帧平均Y偏差(m)': direction['mean_start_y_bias'],\n",
    "}])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "92e0845f",
   "metadata": {},
   "source": [
    "## 3. 用当前关节状态做 FK，而不是相邻 action 相减\n",
    "\n",
    "每帧都有动作执行前的 6D 关节状态 `q_t`。精确转换为：\n",
    "\n",
    "```text\n",
    "tcp_t = FK(q_t)\n",
    "delta_xyz_t = eef_abs_target_xyz_t - tcp_t\n",
    "```\n",
    "\n",
    "直接计算 `action[t] - action[t-1]` 会把控制跟踪误差、接触和物理步进误差混入标签。转换后必须重新计算 LeRobot quantile stats，旧的绝对坐标 normalizer 不能复用。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "c2dc5c14",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.736621Z",
     "iopub.status.busy": "2026-07-11T22:38:34.736547Z",
     "iopub.status.idle": "2026-07-11T22:38:34.740607Z",
     "shell.execute_reply": "2026-07-11T22:38:34.740361Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>episodes</th>\n",
       "      <th>frames</th>\n",
       "      <th>first_action</th>\n",
       "      <th>max_spatial_abs</th>\n",
       "      <th>max_reconstruction_error</th>\n",
       "      <th>source_probe_hashes_unchanged</th>\n",
       "      <th>has_quantile_stats</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>64</td>\n",
       "      <td>26014</td>\n",
       "      <td>[0.003999997396022081, 0.004000022076070309, -...</td>\n",
       "      <td>0.004076</td>\n",
       "      <td>2.328282e-10</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   episodes  frames                                       first_action  \\\n",
       "0        64   26014  [0.003999997396022081, 0.004000022076070309, -...   \n",
       "\n",
       "   max_spatial_abs  max_reconstruction_error  source_probe_hashes_unchanged  \\\n",
       "0         0.004076              2.328282e-10                           True   \n",
       "\n",
       "   has_quantile_stats  \n",
       "0                True  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "conversion = {'episodes': 64, 'frames': 26014, 'first_action': [0.003999997396022081, 0.004000022076070309, -0.0040000565350055695, 0.0, 0.0, 0.0, 0.0], 'max_spatial_abs': 0.0040756394155323505, 'max_reconstruction_error': 2.3282820116321545e-10, 'source_probe_hashes_unchanged': True, 'has_quantile_stats': True}\n",
    "pd.DataFrame([conversion])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "1ba9f77d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.741301Z",
     "iopub.status.busy": "2026-07-11T22:38:34.741229Z",
     "iopub.status.idle": "2026-07-11T22:38:34.742874Z",
     "shell.execute_reply": "2026-07-11T22:38:34.742637Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "export PROJECT_ROOT=/path/to/04mujoco复现ACT、Pi0、SmolVLA\n",
      "export TOPIC_ROOT=/path/to/every-embodied/16-专题组队学习/04-AMD-ROCm策略复刻专题\n",
      "export ABS_ROOT=/path/to/lerobot_eef_abs_v30\n",
      "export DELTA_ROOT=/path/to/lerobot_eef_delta_v30\n",
      "\n",
      "cd \"$PROJECT_ROOT\"\n",
      "DISPLAY=\"${DISPLAY:-:0}\" PYTHONPATH=\"$PROJECT_ROOT:${PYTHONPATH:-}\" python \"$TOPIC_ROOT/code/pi0/pi05_convert_eef_abs_to_delta.py\"   --source-root \"$ABS_ROOT\"   --output-root \"$DELTA_ROOT\"   --repo-id datawhale_eai_pnp_pi05_eef_delta_v30   --scene \"$PROJECT_ROOT/asset/example_scene_y2.xml\"\n"
     ]
    }
   ],
   "source": [
    "print(r'''export PROJECT_ROOT=/path/to/04mujoco复现ACT、Pi0、SmolVLA\n",
    "export TOPIC_ROOT=/path/to/every-embodied/16-专题组队学习/04-AMD-ROCm策略复刻专题\n",
    "export ABS_ROOT=/path/to/lerobot_eef_abs_v30\n",
    "export DELTA_ROOT=/path/to/lerobot_eef_delta_v30\n",
    "\n",
    "cd \"$PROJECT_ROOT\"\n",
    "DISPLAY=\"${DISPLAY:-:0}\" PYTHONPATH=\"$PROJECT_ROOT:${PYTHONPATH:-}\" python \"$TOPIC_ROOT/code/pi0/pi05_convert_eef_abs_to_delta.py\"   --source-root \"$ABS_ROOT\"   --output-root \"$DELTA_ROOT\"   --repo-id datawhale_eai_pnp_pi05_eef_delta_v30   --scene \"$PROJECT_ROOT/asset/example_scene_y2.xml\"''')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b94597a7",
   "metadata": {},
   "source": [
    "## 4. 训练结果先看首帧方向，再看 closed-loop\n",
    "\n",
    "`eef_abs` 和 `eef_delta` 语义不同，delta 分支必须从官方 Pi0.5 基座重新适配 action expert。训练日志还应确认 strict 权重加载、action head 确实更新、冻结 VLM 的参数变化为 0。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "edaa3430",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.743627Z",
     "iopub.status.busy": "2026-07-11T22:38:34.743563Z",
     "iopub.status.idle": "2026-07-11T22:38:34.876169Z",
     "shell.execute_reply": "2026-07-11T22:38:34.875627Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>step</th>\n",
       "      <th>loss</th>\n",
       "      <th>loss_window_mean</th>\n",
       "      <th>learning_rate</th>\n",
       "      <th>grad_norm</th>\n",
       "      <th>memory_peak_gib</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>710</td>\n",
       "      <td>0.236005</td>\n",
       "      <td>0.441694</td>\n",
       "      <td>3.511176e-07</td>\n",
       "      <td>3.969491</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72</th>\n",
       "      <td>720</td>\n",
       "      <td>0.383331</td>\n",
       "      <td>0.436856</td>\n",
       "      <td>2.781181e-07</td>\n",
       "      <td>2.439316</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73</th>\n",
       "      <td>730</td>\n",
       "      <td>0.323361</td>\n",
       "      <td>0.414101</td>\n",
       "      <td>2.134025e-07</td>\n",
       "      <td>3.009227</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>74</th>\n",
       "      <td>740</td>\n",
       "      <td>0.241016</td>\n",
       "      <td>0.371807</td>\n",
       "      <td>1.570842e-07</td>\n",
       "      <td>3.651368</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>750</td>\n",
       "      <td>0.289641</td>\n",
       "      <td>0.420056</td>\n",
       "      <td>1.092620e-07</td>\n",
       "      <td>2.682659</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>76</th>\n",
       "      <td>760</td>\n",
       "      <td>1.074834</td>\n",
       "      <td>0.514304</td>\n",
       "      <td>7.001981e-08</td>\n",
       "      <td>15.589719</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>770</td>\n",
       "      <td>0.436091</td>\n",
       "      <td>0.379699</td>\n",
       "      <td>3.942649e-08</td>\n",
       "      <td>2.827284</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>780</td>\n",
       "      <td>0.380316</td>\n",
       "      <td>0.419703</td>\n",
       "      <td>1.753570e-08</td>\n",
       "      <td>3.039821</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>790</td>\n",
       "      <td>0.395599</td>\n",
       "      <td>0.404411</td>\n",
       "      <td>4.385850e-09</td>\n",
       "      <td>2.863014</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>800</td>\n",
       "      <td>0.181156</td>\n",
       "      <td>0.356420</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>3.997414</td>\n",
       "      <td>11.142135</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    step      loss  loss_window_mean  learning_rate  grad_norm  \\\n",
       "71   710  0.236005          0.441694   3.511176e-07   3.969491   \n",
       "72   720  0.383331          0.436856   2.781181e-07   2.439316   \n",
       "73   730  0.323361          0.414101   2.134025e-07   3.009227   \n",
       "74   740  0.241016          0.371807   1.570842e-07   3.651368   \n",
       "75   750  0.289641          0.420056   1.092620e-07   2.682659   \n",
       "76   760  1.074834          0.514304   7.001981e-08  15.589719   \n",
       "77   770  0.436091          0.379699   3.942649e-08   2.827284   \n",
       "78   780  0.380316          0.419703   1.753570e-08   3.039821   \n",
       "79   790  0.395599          0.404411   4.385850e-09   2.863014   \n",
       "80   800  0.181156          0.356420   0.000000e+00   3.997414   \n",
       "\n",
       "    memory_peak_gib  \n",
       "71        11.142135  \n",
       "72        11.142135  \n",
       "73        11.142135  \n",
       "74        11.142135  \n",
       "75        11.142135  \n",
       "76        11.142135  \n",
       "77        11.142135  \n",
       "78        11.142135  \n",
       "79        11.142135  \n",
       "80        11.142135  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1040x520 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "training = {'base': 'official lerobot/pi05_base', 'dataset': 'canonical91 EEF-delta', 'steps': 800, 'chunk_size': 10, 'batch_size': 4, 'learning_rate': 1e-05, 'action_loss_weights': [1.0, 1.0, 1.0, 0.0, 0.0, 0.0, 2.0], 'trainable_parameters': 430098464, 'trainable_head_max_delta': 0.0005358410999178886, 'frozen_vlm_max_delta': 0.0, 'memory_peak_gib': 11.14213514328003, 'dataset_alignment': {'status': 'passed', 'frames': 35284, 'episodes': 91, 'physical_index_mismatch_rows': 0, 'chunk0_probes': 273, 'chunk0_max_abs_error': 0.0}, 'current_observations_sampled': 3200, 'frame_equivalent_fraction': 0.09069266523069947, 'baseline': {'flow_loss_mean': 1.8661300887664158, 'action_mae': 0.07254758477210999, 'xyz_mae': 0.0030045867897570133, 'gripper_mae': 0.4988192915916443, 'valid_action_rows': 240}, 'final': {'flow_loss_mean': 0.7276507802307606, 'action_mae': 0.045883774757385254, 'xyz_mae': 0.0023318433668464422, 'gripper_mae': 0.3141908645629883, 'valid_action_rows': 240}, 'curve': [{'step': 1, 'loss': 1.4417556524276733, 'loss_window_mean': 1.4417556524276733, 'learning_rate': 2.0000000000000002e-07, 'grad_norm': 17.92433738708496, 'memory_peak_gib': 11.13605785369873}, {'step': 10, 'loss': 1.2654118537902832, 'loss_window_mean': 1.4192016005516053, 'learning_rate': 2e-06, 'grad_norm': 10.447264671325684, 'memory_peak_gib': 11.14213514328003}, {'step': 20, 'loss': 1.8144220113754272, 'loss_window_mean': 1.5600367546081544, 'learning_rate': 4e-06, 'grad_norm': 22.87356185913086, 'memory_peak_gib': 11.14213514328003}, {'step': 30, 'loss': 1.165132761001587, 'loss_window_mean': 1.714900255203247, 'learning_rate': 6e-06, 'grad_norm': 6.110684394836426, 'memory_peak_gib': 11.14213514328003}, {'step': 40, 'loss': 1.1618106365203857, 'loss_window_mean': 1.394715392589569, 'learning_rate': 8e-06, 'grad_norm': 10.046595573425293, 'memory_peak_gib': 11.14213514328003}, {'step': 50, 'loss': 1.1253511905670166, 'loss_window_mean': 1.134506142139435, 'learning_rate': 1e-05, 'grad_norm': 4.444893836975098, 'memory_peak_gib': 11.14213514328003}, {'step': 60, 'loss': 1.3185391426086426, 'loss_window_mean': 1.154559999704361, 'learning_rate': 9.995614150494293e-06, 'grad_norm': 5.843959808349609, 'memory_peak_gib': 11.14213514328003}, {'step': 70, 'loss': 0.950001060962677, 'loss_window_mean': 1.0648394346237182, 'learning_rate': 9.982464296247523e-06, 'grad_norm': 5.329283714294434, 'memory_peak_gib': 11.14213514328003}, {'step': 80, 'loss': 0.8659399747848511, 'loss_window_mean': 1.0005968928337097, 'learning_rate': 9.960573506572391e-06, 'grad_norm': 4.88935661315918, 'memory_peak_gib': 11.14213514328003}, {'step': 90, 'loss': 1.247843861579895, 'loss_window_mean': 0.9402445971965789, 'learning_rate': 9.929980185352525e-06, 'grad_norm': 6.4207305908203125, 'memory_peak_gib': 11.14213514328003}, {'step': 100, 'loss': 0.8038835525512695, 'loss_window_mean': 0.8031426101922989, 'learning_rate': 9.890738003669029e-06, 'grad_norm': 5.595571041107178, 'memory_peak_gib': 11.14213514328003}, {'step': 110, 'loss': 0.7756472826004028, 'loss_window_mean': 0.7853406429290771, 'learning_rate': 9.842915805643156e-06, 'grad_norm': 5.915187835693359, 'memory_peak_gib': 11.14213514328003}, {'step': 120, 'loss': 0.6373303532600403, 'loss_window_mean': 0.7213600903749466, 'learning_rate': 9.786597487660336e-06, 'grad_norm': 6.097359657287598, 'memory_peak_gib': 11.14213514328003}, {'step': 130, 'loss': 0.44265979528427124, 'loss_window_mean': 0.7085215389728546, 'learning_rate': 9.721881851187406e-06, 'grad_norm': 3.655480146408081, 'memory_peak_gib': 11.14213514328003}, {'step': 140, 'loss': 0.6982430815696716, 'loss_window_mean': 0.6520547837018966, 'learning_rate': 9.648882429441258e-06, 'grad_norm': 4.649959564208984, 'memory_peak_gib': 11.14213514328003}, {'step': 150, 'loss': 0.8665914535522461, 'loss_window_mean': 0.6195504486560821, 'learning_rate': 9.567727288213005e-06, 'grad_norm': 5.1642584800720215, 'memory_peak_gib': 11.14213514328003}, {'step': 160, 'loss': 0.4535171091556549, 'loss_window_mean': 0.629642778635025, 'learning_rate': 9.478558801197065e-06, 'grad_norm': 3.602992057800293, 'memory_peak_gib': 11.14213514328003}, {'step': 170, 'loss': 0.3743523955345154, 'loss_window_mean': 0.6044069200754165, 'learning_rate': 9.381533400219319e-06, 'grad_norm': 3.9225196838378906, 'memory_peak_gib': 11.14213514328003}, {'step': 180, 'loss': 0.582123339176178, 'loss_window_mean': 0.5756960898637772, 'learning_rate': 9.276821300802535e-06, 'grad_norm': 5.6062331199646, 'memory_peak_gib': 11.14213514328003}, {'step': 190, 'loss': 0.48226675391197205, 'loss_window_mean': 0.5429364055395126, 'learning_rate': 9.164606203550498e-06, 'grad_norm': 4.538710594177246, 'memory_peak_gib': 11.14213514328003}, {'step': 200, 'loss': 0.36315223574638367, 'loss_window_mean': 0.6303666234016418, 'learning_rate': 9.045084971874738e-06, 'grad_norm': 3.1310110092163086, 'memory_peak_gib': 11.14213514328003}, {'step': 210, 'loss': 0.5272535681724548, 'loss_window_mean': 0.44980175644159315, 'learning_rate': 8.9184672866292e-06, 'grad_norm': 2.7016584873199463, 'memory_peak_gib': 11.14213514328003}, {'step': 220, 'loss': 0.6997479796409607, 'loss_window_mean': 0.5202830523252487, 'learning_rate': 8.784975278258783e-06, 'grad_norm': 3.6506481170654297, 'memory_peak_gib': 11.14213514328003}, {'step': 230, 'loss': 0.473248153924942, 'loss_window_mean': 0.6495455771684646, 'learning_rate': 8.644843137107058e-06, 'grad_norm': 4.24538516998291, 'memory_peak_gib': 11.14213514328003}, {'step': 240, 'loss': 0.5092160701751709, 'loss_window_mean': 0.5006179541349411, 'learning_rate': 8.498316702566828e-06, 'grad_norm': 3.5401418209075928, 'memory_peak_gib': 11.14213514328003}, {'step': 250, 'loss': 0.34540748596191406, 'loss_window_mean': 0.442427733540535, 'learning_rate': 8.345653031794292e-06, 'grad_norm': 4.2529616355896, 'memory_peak_gib': 11.14213514328003}, {'step': 260, 'loss': 0.3908185660839081, 'loss_window_mean': 0.5765176087617874, 'learning_rate': 8.18711994874345e-06, 'grad_norm': 3.1925251483917236, 'memory_peak_gib': 11.14213514328003}, {'step': 270, 'loss': 0.40556472539901733, 'loss_window_mean': 0.49329228699207306, 'learning_rate': 8.022995574311876e-06, 'grad_norm': 5.005031108856201, 'memory_peak_gib': 11.14213514328003}, {'step': 280, 'loss': 0.44332557916641235, 'loss_window_mean': 0.5532789498567581, 'learning_rate': 7.85356783842216e-06, 'grad_norm': 4.756326198577881, 'memory_peak_gib': 11.14213514328003}, {'step': 290, 'loss': 0.5760244727134705, 'loss_window_mean': 0.4679155319929123, 'learning_rate': 7.679133974894984e-06, 'grad_norm': 4.0674638748168945, 'memory_peak_gib': 11.14213514328003}, {'step': 300, 'loss': 0.45710864663124084, 'loss_window_mean': 0.48870720267295836, 'learning_rate': 7.500000000000001e-06, 'grad_norm': 3.8178486824035645, 'memory_peak_gib': 11.14213514328003}, {'step': 310, 'loss': 0.8459497690200806, 'loss_window_mean': 0.5000417187809945, 'learning_rate': 7.31648017559931e-06, 'grad_norm': 8.430974006652832, 'memory_peak_gib': 11.14213514328003}, {'step': 320, 'loss': 0.5228030681610107, 'loss_window_mean': 0.6109683305025101, 'learning_rate': 7.128896457825364e-06, 'grad_norm': 6.358067512512207, 'memory_peak_gib': 11.14213514328003}, {'step': 330, 'loss': 0.5134061574935913, 'loss_window_mean': 0.3894116789102554, 'learning_rate': 6.9375779322605154e-06, 'grad_norm': 5.882976531982422, 'memory_peak_gib': 11.14213514328003}, {'step': 340, 'loss': 0.7557446956634521, 'loss_window_mean': 0.53605537712574, 'learning_rate': 6.7428602366090764e-06, 'grad_norm': 10.347887992858887, 'memory_peak_gib': 11.14213514328003}, {'step': 350, 'loss': 0.38439640402793884, 'loss_window_mean': 0.494500470161438, 'learning_rate': 6.545084971874738e-06, 'grad_norm': 3.290647506713867, 'memory_peak_gib': 11.14213514328003}, {'step': 360, 'loss': 0.25496339797973633, 'loss_window_mean': 0.4361968904733658, 'learning_rate': 6.344599103076329e-06, 'grad_norm': 2.3100244998931885, 'memory_peak_gib': 11.14213514328003}, {'step': 370, 'loss': 0.4255277216434479, 'loss_window_mean': 0.4905035555362701, 'learning_rate': 6.141754350553279e-06, 'grad_norm': 3.3459222316741943, 'memory_peak_gib': 11.14213514328003}, {'step': 380, 'loss': 1.0703399181365967, 'loss_window_mean': 0.542897954583168, 'learning_rate': 5.936906572928625e-06, 'grad_norm': 12.507841110229492, 'memory_peak_gib': 11.14213514328003}, {'step': 390, 'loss': 0.43372321128845215, 'loss_window_mean': 0.39251488745212554, 'learning_rate': 5.730415142812059e-06, 'grad_norm': 3.360889434814453, 'memory_peak_gib': 11.14213514328003}, {'step': 400, 'loss': 0.5051335692405701, 'loss_window_mean': 0.5216175526380539, 'learning_rate': 5.522642316338268e-06, 'grad_norm': 6.198289394378662, 'memory_peak_gib': 11.14213514328003}, {'step': 410, 'loss': 0.5980904698371887, 'loss_window_mean': 0.5554435983300209, 'learning_rate': 5.3139525976465675e-06, 'grad_norm': 4.662997245788574, 'memory_peak_gib': 11.14213514328003}, {'step': 420, 'loss': 1.0505011081695557, 'loss_window_mean': 0.5237489521503449, 'learning_rate': 5.1047120994167855e-06, 'grad_norm': 9.173433303833008, 'memory_peak_gib': 11.14213514328003}, {'step': 430, 'loss': 0.461292028427124, 'loss_window_mean': 0.39654761999845506, 'learning_rate': 4.895287900583216e-06, 'grad_norm': 3.7262845039367676, 'memory_peak_gib': 11.14213514328003}, {'step': 440, 'loss': 0.4652067720890045, 'loss_window_mean': 0.5328009217977524, 'learning_rate': 4.686047402353433e-06, 'grad_norm': 10.385488510131836, 'memory_peak_gib': 11.14213514328003}, {'step': 450, 'loss': 0.5061522722244263, 'loss_window_mean': 0.3967305332422256, 'learning_rate': 4.477357683661734e-06, 'grad_norm': 3.6560938358306885, 'memory_peak_gib': 11.14213514328003}, {'step': 460, 'loss': 0.49948805570602417, 'loss_window_mean': 0.43527990877628325, 'learning_rate': 4.269584857187942e-06, 'grad_norm': 3.402996778488159, 'memory_peak_gib': 11.14213514328003}, {'step': 470, 'loss': 0.4548255503177643, 'loss_window_mean': 0.43546119928359983, 'learning_rate': 4.063093427071376e-06, 'grad_norm': 2.596640110015869, 'memory_peak_gib': 11.14213514328003}, {'step': 480, 'loss': 0.21186473965644836, 'loss_window_mean': 0.4617734611034393, 'learning_rate': 3.8582456494467214e-06, 'grad_norm': 1.7170941829681396, 'memory_peak_gib': 11.14213514328003}, {'step': 490, 'loss': 0.34907472133636475, 'loss_window_mean': 0.5340840175747872, 'learning_rate': 3.655400896923672e-06, 'grad_norm': 9.269362449645996, 'memory_peak_gib': 11.14213514328003}, {'step': 500, 'loss': 0.3384384214878082, 'loss_window_mean': 0.3850263386964798, 'learning_rate': 3.4549150281252635e-06, 'grad_norm': 2.6124203205108643, 'memory_peak_gib': 11.14213514328003}, {'step': 510, 'loss': 0.4014032483100891, 'loss_window_mean': 0.4897883325815201, 'learning_rate': 3.2571397633909252e-06, 'grad_norm': 4.532131671905518, 'memory_peak_gib': 11.14213514328003}, {'step': 520, 'loss': 0.4053635001182556, 'loss_window_mean': 0.40699050575494766, 'learning_rate': 3.0624220677394854e-06, 'grad_norm': 2.9826910495758057, 'memory_peak_gib': 11.14213514328003}, {'step': 530, 'loss': 0.5464410185813904, 'loss_window_mean': 0.4697120666503906, 'learning_rate': 2.871103542174637e-06, 'grad_norm': 7.868485450744629, 'memory_peak_gib': 11.14213514328003}, {'step': 540, 'loss': 0.1598701775074005, 'loss_window_mean': 0.5457518607378006, 'learning_rate': 2.683519824400693e-06, 'grad_norm': 2.382927894592285, 'memory_peak_gib': 11.14213514328003}, {'step': 550, 'loss': 0.5085831880569458, 'loss_window_mean': 0.36985157430171967, 'learning_rate': 2.5000000000000015e-06, 'grad_norm': 3.4796156883239746, 'memory_peak_gib': 11.14213514328003}, {'step': 560, 'loss': 0.1951654851436615, 'loss_window_mean': 0.35639636814594267, 'learning_rate': 2.320866025105016e-06, 'grad_norm': 1.9940476417541504, 'memory_peak_gib': 11.14213514328003}, {'step': 570, 'loss': 0.2957967519760132, 'loss_window_mean': 0.3672603234648705, 'learning_rate': 2.146432161577842e-06, 'grad_norm': 3.245014190673828, 'memory_peak_gib': 11.14213514328003}, {'step': 580, 'loss': 0.3202844262123108, 'loss_window_mean': 0.4245576798915863, 'learning_rate': 1.977004425688126e-06, 'grad_norm': 2.6708970069885254, 'memory_peak_gib': 11.14213514328003}, {'step': 590, 'loss': 0.2762383818626404, 'loss_window_mean': 0.541545957326889, 'learning_rate': 1.8128800512565514e-06, 'grad_norm': 2.283701181411743, 'memory_peak_gib': 11.14213514328003}, {'step': 600, 'loss': 0.2439223974943161, 'loss_window_mean': 0.39494886845350263, 'learning_rate': 1.6543469682057105e-06, 'grad_norm': 2.3808577060699463, 'memory_peak_gib': 11.14213514328003}, {'step': 610, 'loss': 0.39529117941856384, 'loss_window_mean': 0.39307540059089663, 'learning_rate': 1.5016832974331725e-06, 'grad_norm': 12.575508117675781, 'memory_peak_gib': 11.14213514328003}, {'step': 620, 'loss': 0.48883605003356934, 'loss_window_mean': 0.4381705656647682, 'learning_rate': 1.3551568628929434e-06, 'grad_norm': 3.0441272258758545, 'memory_peak_gib': 11.14213514328003}, {'step': 630, 'loss': 0.27479487657546997, 'loss_window_mean': 0.3908743917942047, 'learning_rate': 1.2150247217412186e-06, 'grad_norm': 2.226179838180542, 'memory_peak_gib': 11.14213514328003}, {'step': 640, 'loss': 0.3105739951133728, 'loss_window_mean': 0.4727252021431923, 'learning_rate': 1.0815327133708015e-06, 'grad_norm': 1.8943754434585571, 'memory_peak_gib': 11.14213514328003}, {'step': 650, 'loss': 0.6749321818351746, 'loss_window_mean': 0.5321788474917412, 'learning_rate': 9.549150281252633e-07, 'grad_norm': 6.666606903076172, 'memory_peak_gib': 11.14213514328003}, {'step': 660, 'loss': 0.4653055667877197, 'loss_window_mean': 0.464268296957016, 'learning_rate': 8.353937964495029e-07, 'grad_norm': 3.4764044284820557, 'memory_peak_gib': 11.14213514328003}, {'step': 670, 'loss': 0.37903326749801636, 'loss_window_mean': 0.4639122486114502, 'learning_rate': 7.23178699197467e-07, 'grad_norm': 3.6408205032348633, 'memory_peak_gib': 11.14213514328003}, {'step': 680, 'loss': 0.3517443835735321, 'loss_window_mean': 0.4014388144016266, 'learning_rate': 6.184665997806832e-07, 'grad_norm': 2.1739017963409424, 'memory_peak_gib': 11.14213514328003}, {'step': 690, 'loss': 0.3245919346809387, 'loss_window_mean': 0.48749532252550126, 'learning_rate': 5.214411988029355e-07, 'grad_norm': 2.402776002883911, 'memory_peak_gib': 11.14213514328003}, {'step': 700, 'loss': 0.5193167924880981, 'loss_window_mean': 0.492905730009079, 'learning_rate': 4.322727117869951e-07, 'grad_norm': 2.9502995014190674, 'memory_peak_gib': 11.14213514328003}, {'step': 710, 'loss': 0.23600494861602783, 'loss_window_mean': 0.4416937083005905, 'learning_rate': 3.511175705587433e-07, 'grad_norm': 3.9694910049438477, 'memory_peak_gib': 11.14213514328003}, {'step': 720, 'loss': 0.3833305835723877, 'loss_window_mean': 0.4368560403585434, 'learning_rate': 2.7811814881259503e-07, 'grad_norm': 2.4393162727355957, 'memory_peak_gib': 11.14213514328003}, {'step': 730, 'loss': 0.3233606219291687, 'loss_window_mean': 0.41410118341445923, 'learning_rate': 2.134025123396638e-07, 'grad_norm': 3.0092265605926514, 'memory_peak_gib': 11.14213514328003}, {'step': 740, 'loss': 0.2410159409046173, 'loss_window_mean': 0.3718071937561035, 'learning_rate': 1.5708419435684463e-07, 'grad_norm': 3.6513679027557373, 'memory_peak_gib': 11.14213514328003}, {'step': 750, 'loss': 0.2896406650543213, 'loss_window_mean': 0.42005575597286227, 'learning_rate': 1.0926199633097156e-07, 'grad_norm': 2.682659149169922, 'memory_peak_gib': 11.14213514328003}, {'step': 760, 'loss': 1.074833631515503, 'loss_window_mean': 0.5143037632107734, 'learning_rate': 7.001981464747565e-08, 'grad_norm': 15.58971881866455, 'memory_peak_gib': 11.14213514328003}, {'step': 770, 'loss': 0.43609103560447693, 'loss_window_mean': 0.37969904989004133, 'learning_rate': 3.9426493427611177e-08, 'grad_norm': 2.8272836208343506, 'memory_peak_gib': 11.14213514328003}, {'step': 780, 'loss': 0.38031595945358276, 'loss_window_mean': 0.41970276534557344, 'learning_rate': 1.753570375247815e-08, 'grad_norm': 3.039820671081543, 'memory_peak_gib': 11.14213514328003}, {'step': 790, 'loss': 0.3955989181995392, 'loss_window_mean': 0.40441102385520933, 'learning_rate': 4.385849505708084e-09, 'grad_norm': 2.863013744354248, 'memory_peak_gib': 11.14213514328003}, {'step': 800, 'loss': 0.18115566670894623, 'loss_window_mean': 0.3564197301864624, 'learning_rate': 0.0, 'grad_norm': 3.9974138736724854, 'memory_peak_gib': 11.14213514328003}]}\n",
    "pd.DataFrame([\n",
    "    {'阶段': '训练前', **training['baseline']},\n",
    "    {'阶段': '训练后', **training['final']},\n",
    "])\n",
    "curve = pd.DataFrame(training.get('curve', []))\n",
    "if not curve.empty:\n",
    "    display(curve.tail(10))\n",
    "    ax = curve.plot(x='step', y=['loss', 'loss_window_mean'], figsize=(8, 4))\n",
    "    ax.set_title('Pi0.5 canonical EEF-delta 训练曲线')\n",
    "    ax.set_ylabel('flow matching loss')\n",
    "    ax.grid(alpha=0.25)\n",
    "    plt.tight_layout()\n",
    "    plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "69b78d9c",
   "metadata": {},
   "source": [
    "## 5. 全新随机种子 strict fair-VLA\n",
    "\n",
    "离线误差只能说明训练健康。最终仍使用 lift、upright、plate height、plate displacement 和 stable-place 五组物理条件，并保存四视角视频复核。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "c5e08a34",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.877572Z",
     "iopub.status.busy": "2026-07-11T22:38:34.877444Z",
     "iopub.status.idle": "2026-07-11T22:38:34.883652Z",
     "shell.execute_reply": "2026-07-11T22:38:34.883352Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>评估面板</th>\n",
       "      <th>episodes</th>\n",
       "      <th>physical_success_count</th>\n",
       "      <th>physical_success_rate</th>\n",
       "      <th>legacy_success_count</th>\n",
       "      <th>legacy_success_rate</th>\n",
       "      <th>execution_horizon</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>训练内种子</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>全新随机种子</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     评估面板  episodes  physical_success_count  physical_success_rate  \\\n",
       "0   训练内种子         4                       0                    0.0   \n",
       "1  全新随机种子         4                       0                    0.0   \n",
       "\n",
       "   legacy_success_count  legacy_success_rate  execution_horizon  \n",
       "0                     0                  0.0                 10  \n",
       "1                     0                  0.0                 10  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "evaluation = {'episodes': 4, 'physical_success_count': 0, 'physical_success_rate': 0.0, 'legacy_success_count': 0, 'legacy_success_rate': 0.0, 'seeds': [2180, 2181, 2182, 2183], 'position_profile': 'pnp_generalization_v1', 'fair_vla': True, 'action_mode': 'eef_delta', 'execution_horizon': 10, 'fresh_env_per_episode': True, 'hard_reset_sim_data': True, 'physics_steps_per_action': 25}\n",
    "training_seed_evaluation = {'episodes': 4, 'physical_success_count': 0, 'physical_success_rate': 0.0, 'legacy_success_count': 0, 'legacy_success_rate': 0.0, 'seeds': [1000, 1001, 1002, 1003], 'position_profile': 'pnp_generalization_v1', 'fair_vla': True, 'action_mode': 'eef_delta', 'execution_horizon': 10, 'fresh_env_per_episode': True, 'hard_reset_sim_data': True, 'physics_steps_per_action': 25}\n",
    "summary_fields = ['episodes', 'physical_success_count', 'physical_success_rate', 'legacy_success_count', 'legacy_success_rate', 'execution_horizon']\n",
    "rows = [\n",
    "    {'评估面板': '全新随机种子', **{key: evaluation[key] for key in summary_fields}},\n",
    "]\n",
    "if training_seed_evaluation:\n",
    "    rows.insert(0, {'评估面板': '训练内种子', **{key: training_seed_evaluation[key] for key in summary_fields}})\n",
    "pd.DataFrame(rows)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0aa7b37e",
   "metadata": {},
   "source": [
    "训练内 seed 和全新 seed 同时失败时，先检查是否真正覆盖过足够多的当前观测。这里的 `step × batch_size` 只是抽样次数；加权采样还会重复起步、尾段和夹爪切换窗口，不能把一次 action chunk 里的 10 个未来动作误算成 10 个独立视觉状态。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "1ad944f7",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.884489Z",
     "iopub.status.busy": "2026-07-11T22:38:34.884419Z",
     "iopub.status.idle": "2026-07-11T22:38:34.887400Z",
     "shell.execute_reply": "2026-07-11T22:38:34.887201Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>训练step</th>\n",
       "      <th>batch size</th>\n",
       "      <th>当前观测抽样次数</th>\n",
       "      <th>相对全数据帧数</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>800</td>\n",
       "      <td>4</td>\n",
       "      <td>3200</td>\n",
       "      <td>0.090693</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   训练step  batch size  当前观测抽样次数   相对全数据帧数\n",
       "0     800           4      3200  0.090693"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame([{\n",
    "    '训练step': training['steps'],\n",
    "    'batch size': training['batch_size'],\n",
    "    '当前观测抽样次数': training['current_observations_sampled'],\n",
    "    '相对全数据帧数': training['frame_equivalent_fraction'],\n",
    "}])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c9cb0fc3",
   "metadata": {},
   "source": [
    "## 6. 从逐步轨迹定位“靠近之后发生了什么”\n",
    "\n",
    "`0/4` 仍然太粗。逐步日志要回答三个问题：TCP 最接近杯子时有多远、当时比杯体高多少、夹爪是否已经闭合。若 TCP 先在高处闭合，再继续平移，最后才在杯子外下降，就属于阶段顺序错误，不是“完全没有识别杯子”。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "2e968755",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.888213Z",
     "iopub.status.busy": "2026-07-11T22:38:34.888151Z",
     "iopub.status.idle": "2026-07-11T22:38:34.892511Z",
     "shell.execute_reply": "2026-07-11T22:38:34.892275Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>seed</th>\n",
       "      <th>nearest_logged_step</th>\n",
       "      <th>nearest_tcp_target_xy_m</th>\n",
       "      <th>tcp_target_z_gap_at_nearest_m</th>\n",
       "      <th>gripper_at_nearest</th>\n",
       "      <th>lowest_logged_step</th>\n",
       "      <th>lowest_tcp_target_z_gap_m</th>\n",
       "      <th>tcp_target_xy_at_lowest_m</th>\n",
       "      <th>max_target_lift_m</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2180</td>\n",
       "      <td>155</td>\n",
       "      <td>0.054835</td>\n",
       "      <td>0.1271</td>\n",
       "      <td>0.026198</td>\n",
       "      <td>360</td>\n",
       "      <td>-0.0384</td>\n",
       "      <td>0.342680</td>\n",
       "      <td>0.000001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2181</td>\n",
       "      <td>75</td>\n",
       "      <td>0.053994</td>\n",
       "      <td>0.1381</td>\n",
       "      <td>0.020329</td>\n",
       "      <td>225</td>\n",
       "      <td>-0.0381</td>\n",
       "      <td>0.292981</td>\n",
       "      <td>0.000001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2182</td>\n",
       "      <td>105</td>\n",
       "      <td>0.066632</td>\n",
       "      <td>0.1082</td>\n",
       "      <td>0.009427</td>\n",
       "      <td>215</td>\n",
       "      <td>-0.0375</td>\n",
       "      <td>0.252800</td>\n",
       "      <td>0.000003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2183</td>\n",
       "      <td>125</td>\n",
       "      <td>0.054027</td>\n",
       "      <td>0.1041</td>\n",
       "      <td>-0.000742</td>\n",
       "      <td>400</td>\n",
       "      <td>-0.0382</td>\n",
       "      <td>0.246220</td>\n",
       "      <td>0.001622</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   seed  nearest_logged_step  nearest_tcp_target_xy_m  \\\n",
       "0  2180                  155                 0.054835   \n",
       "1  2181                   75                 0.053994   \n",
       "2  2182                  105                 0.066632   \n",
       "3  2183                  125                 0.054027   \n",
       "\n",
       "   tcp_target_z_gap_at_nearest_m  gripper_at_nearest  lowest_logged_step  \\\n",
       "0                         0.1271            0.026198                 360   \n",
       "1                         0.1381            0.020329                 225   \n",
       "2                         0.1082            0.009427                 215   \n",
       "3                         0.1041           -0.000742                 400   \n",
       "\n",
       "   lowest_tcp_target_z_gap_m  tcp_target_xy_at_lowest_m  max_target_lift_m  \n",
       "0                    -0.0384                   0.342680           0.000001  \n",
       "1                    -0.0381                   0.292981           0.000001  \n",
       "2                    -0.0375                   0.252800           0.000003  \n",
       "3                    -0.0382                   0.246220           0.001622  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "phase_diagnostics = [{'seed': 2180, 'nearest_logged_step': 155, 'nearest_tcp_target_xy_m': 0.054835207668066696, 'tcp_target_z_gap_at_nearest_m': 0.1271, 'gripper_at_nearest': 0.02619794015488298, 'lowest_logged_step': 360, 'lowest_tcp_target_z_gap_m': -0.03839999999999999, 'tcp_target_xy_at_lowest_m': 0.34268009863428023, 'max_target_lift_m': 1.2853031164938855e-06}, {'seed': 2181, 'nearest_logged_step': 75, 'nearest_tcp_target_xy_m': 0.05399351812949405, 'tcp_target_z_gap_at_nearest_m': 0.1381, 'gripper_at_nearest': 0.020328813205147046, 'lowest_logged_step': 225, 'lowest_tcp_target_z_gap_m': -0.03810000000000002, 'tcp_target_xy_at_lowest_m': 0.29298071608896037, 'max_target_lift_m': 1.2855770460395988e-06}, {'seed': 2182, 'nearest_logged_step': 105, 'nearest_tcp_target_xy_m': 0.06663249957790868, 'tcp_target_z_gap_at_nearest_m': 0.10819999999999996, 'gripper_at_nearest': 0.009426771461153876, 'lowest_logged_step': 215, 'lowest_tcp_target_z_gap_m': -0.03749999999999998, 'tcp_target_xy_at_lowest_m': 0.25280031645549816, 'max_target_lift_m': 2.7370485115030263e-06}, {'seed': 2183, 'nearest_logged_step': 125, 'nearest_tcp_target_xy_m': 0.0540268451790404, 'tcp_target_z_gap_at_nearest_m': 0.10409999999999997, 'gripper_at_nearest': -0.0007415589700601158, 'lowest_logged_step': 400, 'lowest_tcp_target_z_gap_m': -0.03820000000000001, 'tcp_target_xy_at_lowest_m': 0.2462196174150224, 'max_target_lift_m': 0.001621553991664193}]\n",
    "pd.DataFrame(phase_diagnostics)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "91236851",
   "metadata": {},
   "source": [
    "![Pi0.5 canonical91 训练内 seed1003 失败序列](../assets/pi05_canonical_s800_seed1003_montage.png)\n",
    "\n",
    "图中策略先向蓝杯靠近，夹爪在杯子上方/侧方闭合，随后将杯子推倒并继续越过目标。这个视频是 raw Pi0.5 的失败诊断，不是成功演示。\n",
    "\n",
    "<video controls muted preload=\"metadata\" width=\"100%\">\n",
    "  <source src=\"../assets/pi05_canonical_s800_seed1003.mp4\" type=\"video/mp4\">\n",
    "</video>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "028b15b6",
   "metadata": {},
   "source": [
    "## 7. 缩短执行 horizon 为什么没有救回\n",
    "\n",
    "执行 horizon 只决定一个已预测 action chunk 实际执行多少步后重新看图。它能减轻开环累积，但不能修正模型已经排错的“接近、下降、闭合”顺序。对照时必须保持 checkpoint、种子、物理步进、语言和成功判定不变。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "c84f8492",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.893196Z",
     "iopub.status.busy": "2026-07-11T22:38:34.893132Z",
     "iopub.status.idle": "2026-07-11T22:38:34.896581Z",
     "shell.execute_reply": "2026-07-11T22:38:34.896449Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>horizon</th>\n",
       "      <th>episodes</th>\n",
       "      <th>strict_success</th>\n",
       "      <th>success_rate</th>\n",
       "      <th>seeds</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[2130, 2131, 2132, 2133]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[2130, 2131, 2132, 2133]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>[2133]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   horizon  episodes  strict_success  success_rate                     seeds\n",
       "0       10         4               0           0.0  [2130, 2131, 2132, 2133]\n",
       "1        5         4               0           0.0  [2130, 2131, 2132, 2133]\n",
       "2        1         1               0           0.0                    [2133]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "horizon_diagnostics = [{'horizon': 10, 'episodes': 4, 'strict_success': 0, 'success_rate': 0.0, 'seeds': [2130, 2131, 2132, 2133]}, {'horizon': 5, 'episodes': 4, 'strict_success': 0, 'success_rate': 0.0, 'seeds': [2130, 2131, 2132, 2133]}, {'horizon': 1, 'episodes': 1, 'strict_success': 0, 'success_rate': 0.0, 'seeds': [2133]}]\n",
    "pd.DataFrame(horizon_diagnostics)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "223f1bdf",
   "metadata": {},
   "source": [
    "![Pi0.5 EEF-delta horizon5 失败序列](../assets/pi05_eefdelta_h5_seed2133_montage.png)\n",
    "\n",
    "seed `2133` 的四视角关键帧显示，策略在杯子上方和周围移动，但没有进入稳定抓取高度。关键帧要和 TCP-target xy、TCP z、gripper 逐步日志共同判断。\n",
    "\n",
    "<video controls muted preload=\"metadata\" width=\"100%\">\n",
    "  <source src=\"../assets/pi05_eefdelta_h5_seed2133.mp4\" type=\"video/mp4\">\n",
    "</video>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "adbfbd2b",
   "metadata": {},
   "source": [
    "逐步轨迹还要同时看 `TCP-target xy`、TCP z 和 gripper。只看最终 gripper 会误判成“模型没有学会闭合”：本轮模型确实会闭合，但常在杯子外或杯子上方闭合；真正靠近杯子时又可能重新打开。\n",
    "\n",
    "## 8. 下一批数据：连续 on-policy recovery\n",
    "\n",
    "新的纠偏数据使用当前 Pi0.5 执行不同长度的 prefix，再让专家从实际 takeover state 连续执行到严格成功。只保存专家真实执行过的连续后缀，失败候选不默认进入 BC 数据。prefix 分成 40、80、120，是为了覆盖早期漂移、接近阶段和闭合前后的不同状态，而不是把 episode 序号或 GT phase 喂给模型。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "3da11ee4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.897550Z",
     "iopub.status.busy": "2026-07-11T22:38:34.897455Z",
     "iopub.status.idle": "2026-07-11T22:38:34.900729Z",
     "shell.execute_reply": "2026-07-11T22:38:34.900473Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>prefix</th>\n",
       "      <th>candidates</th>\n",
       "      <th>strict_saved</th>\n",
       "      <th>frames</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>40</td>\n",
       "      <td>10</td>\n",
       "      <td>9</td>\n",
       "      <td>3076</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>80</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>3859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>120</td>\n",
       "      <td>10</td>\n",
       "      <td>7</td>\n",
       "      <td>2335</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   prefix  candidates  strict_saved  frames\n",
       "0      40          10             9    3076\n",
       "1      80          11            11    3859\n",
       "2     120          10             7    2335"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "recovery = {'episodes': 27, 'frames': 9270, 'by_prefix': [{'prefix': 40, 'candidates': 10, 'strict_saved': 9, 'frames': 3076}, {'prefix': 80, 'candidates': 11, 'strict_saved': 11, 'frames': 3859}, {'prefix': 120, 'candidates': 10, 'strict_saved': 7, 'frames': 2335}]}\n",
    "pd.DataFrame(recovery.get('by_prefix', []))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1a93bc38",
   "metadata": {},
   "source": [
    "## 9. 聚合后验证 parquet 物理行序\n",
    "\n",
    "`index` 排序后连续，不代表 Hugging Face Dataset 的物理第 `i` 行就是 index `i`。LeRobot 用物理 `idx + delta` 查询未来 action；一旦文件顺序与全局 index 脱节，当前 observation 和 action chunk 会静默错配。\n",
    "\n",
    "修复脚本不覆盖源数据，而是每个 episode 单独写一个 parquet，按 `frame_index` 排序，重建全局 index、episode 边界、文件映射和数值统计。随后必须通过真实 `delta_timestamps` 接口验证 `chunk[0]`、完整未来动作和 episode 末尾 padding。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "1060bb80",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.901804Z",
     "iopub.status.busy": "2026-07-11T22:38:34.901705Z",
     "iopub.status.idle": "2026-07-11T22:38:34.910965Z",
     "shell.execute_reply": "2026-07-11T22:38:34.910511Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>数据</th>\n",
       "      <th>data_files</th>\n",
       "      <th>rows</th>\n",
       "      <th>physical_index_mismatch_rows</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>聚合源数据</td>\n",
       "      <td>77</td>\n",
       "      <td>35284</td>\n",
       "      <td>4909</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>canonical 副本</td>\n",
       "      <td>91</td>\n",
       "      <td>35284</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             数据  data_files   rows  physical_index_mismatch_rows\n",
       "0         聚合源数据          77  35284                          4909\n",
       "1  canonical 副本          91  35284                             0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>horizon</th>\n",
       "      <th>positions_checked</th>\n",
       "      <th>action_values_checked</th>\n",
       "      <th>padded_steps_checked</th>\n",
       "      <th>max_abs_error</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10</td>\n",
       "      <td>1318</td>\n",
       "      <td>92260</td>\n",
       "      <td>2566</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>50</td>\n",
       "      <td>1318</td>\n",
       "      <td>461300</td>\n",
       "      <td>25674</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   horizon  positions_checked  action_values_checked  padded_steps_checked  \\\n",
       "0       10               1318                  92260                  2566   \n",
       "1       50               1318                 461300                 25674   \n",
       "\n",
       "   max_abs_error  \n",
       "0            0.0  \n",
       "1            0.0  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "canonicalization = {'episodes': 102, 'frames': 32792, 'layout': {'output': {'data_files': 102, 'rows': 32792, 'physical_index_mismatch_rows': 0}}, 'one_episode_per_file': True, 'prefix80_alignment': {'horizon': 10, 'positions_checked': 153, 'max_abs_error': 0.0}, 'status': 'STRUCTURAL_PASS'}\n",
    "layout = canonicalization.get('layout', {})\n",
    "display(pd.DataFrame([\n",
    "    {'数据': '聚合源数据', **layout.get('source', {})},\n",
    "    {'数据': 'canonical 副本', **layout.get('output', {})},\n",
    "]))\n",
    "pd.DataFrame(canonicalization.get('horizon_results', []))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "f06df3ac",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-11T22:38:34.912249Z",
     "iopub.status.busy": "2026-07-11T22:38:34.912077Z",
     "iopub.status.idle": "2026-07-11T22:38:34.914652Z",
     "shell.execute_reply": "2026-07-11T22:38:34.914299Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "python \"$TOPIC_ROOT/code/pi0/pi05_canonicalize_lerobot_v3.py\" \\\n",
      "  --source-root \"$AGGREGATED_ROOT\" \\\n",
      "  --source-repo-id \"$AGGREGATED_REPO_ID\" \\\n",
      "  --output-root \"$CANONICAL_ROOT\" \\\n",
      "  --output-repo-id \"$CANONICAL_REPO_ID\"\n",
      "\n",
      "HF_HUB_OFFLINE=1 PYTHONPATH=\"/path/to/lerobot/src:${PYTHONPATH:-}\" \\\n",
      "python \"$TOPIC_ROOT/code/pi0/pi05_validate_chunk_alignment.py\" \\\n",
      "  --root \"$CANONICAL_ROOT\" \\\n",
      "  --repo-id \"$CANONICAL_REPO_ID\" \\\n",
      "  --horizons 10 50 \\\n",
      "  --uniform-samples 512\n"
     ]
    }
   ],
   "source": [
    "print(r'''python \"$TOPIC_ROOT/code/pi0/pi05_canonicalize_lerobot_v3.py\" \\\n",
    "  --source-root \"$AGGREGATED_ROOT\" \\\n",
    "  --source-repo-id \"$AGGREGATED_REPO_ID\" \\\n",
    "  --output-root \"$CANONICAL_ROOT\" \\\n",
    "  --output-repo-id \"$CANONICAL_REPO_ID\"\n",
    "\n",
    "HF_HUB_OFFLINE=1 PYTHONPATH=\"/path/to/lerobot/src:${PYTHONPATH:-}\" \\\n",
    "python \"$TOPIC_ROOT/code/pi0/pi05_validate_chunk_alignment.py\" \\\n",
    "  --root \"$CANONICAL_ROOT\" \\\n",
    "  --repo-id \"$CANONICAL_REPO_ID\" \\\n",
    "  --horizons 10 50 \\\n",
    "  --uniform-samples 512''')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9b6918d4",
   "metadata": {},
   "source": [
    "## 结论\n",
    "\n",
    "本轮在 canonical102（102 episodes / 32792 frames，EEF-delta 7D，chunk10，物理行序 mismatch=0）上，从干净 S400 expert-vision 基座完成了 800 steps。部署输入仍然只有双相机图像、语言和 robot proprio；严格评估没有使用 target/plate 坐标、oracle prefix、GT phase、scripted gripper 或外置 finisher。\n",
    "\n",
    "第一次训练更新虽然完成，但保存阶段因 /dev/shm 空间不足产生了截断 checkpoint：文件头需要约 7.47 GB，实际只有约 6.87 GB，safe_open 报 file not fully covered。清理本轮临时缓存后重跑，最终 checkpoint 完整，safe_open 成功读取 812 个 tensors。训练峰值统一内存约 33.56 GiB，没有 OOM、NaN、SIGSEGV 或新的 kernel error。\n",
    "\n",
    "训练前后离线指标为：flow loss 0.90623→0.87770，action MAE 0.011642→0.011985，xyz MAE 0.002580→0.002533，gripper MAE 0.073754→0.076300。xyz 只有轻微改善，action 和 gripper 没有同步改善，因此 800 steps 不能解释为策略收敛。\n",
    "\n",
    "在 10 个 fresh seeds（1000、1009、1018、1023、1037、1044、1055、1060、1068、1078）上的 raw strict 结果是 legacy 0/10、physical 0/10，visual contact head active steps 为 0。大多数轨迹没有有效抬升；少数轨迹出现高位闭合、倾斜或接近抬升门槛后倒置。\n",
    "\n",
    "本轮已经排除训练崩溃、checkpoint 未加载、数据行序错位和评估脚本误判，剩下的瓶颈是视觉接近、接触、闭爪、抬升之间的闭环时序。下一步应采集 Pi0.5 自己跑偏后的连续 on-policy correction，或实现只读取图像、语言和 robot proprio 的 learned phase/progress/contact、EEF residual 与 temporal aggregation；候选仍必须用 raw strict 和独立 seeds 验收。\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}