pi05-minimal-validation/notebooks/pi05_minimal_reproduction.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "title",
   "metadata": {},
   "source": [
    "# π0.5 结构级最小复现：从环境到实验结果\n",
    "\n",
    "本 Notebook 对应 Physical Intelligence 的 **π0.5** 层级式 VLA 结构。它不复现参数规模、预训练数据和真实机器人能力，只验证论文中的最小信息流：\n",
    "\n",
    "`π(subtask | observation, task) × π(action | observation, predicted_subtask)`\n",
    "\n",
    "从上到下依次运行所有代码单元，即可完成环境检查、数据准备、测试、训练、推理和结果展示。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "toc",
   "metadata": {},
   "source": [
    "## 目录\n",
    "\n",
    "1. [官方论文与复现边界](#1-官方论文与复现边界)\n",
    "2. [架构介绍](#2-架构介绍)\n",
    "3. [环境安装](#3-环境安装)\n",
    "4. [数据集与权重准备](#4-数据集与权重准备)\n",
    "5. [检查层级 token 与数据接口](#5-检查层级-token-与数据接口)\n",
    "6. [运行单元测试](#6-运行单元测试)\n",
    "7. [生成实验配置](#7-生成实验配置)\n",
    "8. [训练或导入权重](#8-训练或导入权重)\n",
    "9. [推理与实验结果](#9-推理与实验结果)\n",
    "10. [结构与结果验收](#10-结构与结果验收)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "paper",
   "metadata": {},
   "source": [
    "## 1. 官方论文与复现边界\n",
    "\n",
    "- 论文：[*π0.5: a Vision-Language-Action Model with Open-World Generalization*](https://arxiv.org/abs/2504.16054)\n",
    "- PDF：[arXiv PDF](https://arxiv.org/pdf/2504.16054)\n",
    "- 官方研究页面：[Physical Intelligence π0.5](https://www.pi.website/blog/pi05)\n",
    "- 官方开源基础实现：[Physical-Intelligence/openpi](https://github.com/Physical-Intelligence/openpi)\n",
    "\n",
    "本项目保留：离散 state token、总体任务条件下的高层 subtask、预测 subtask 条件下的低层 action、FAST-lite 离散动作、独立 action expert、AdaRMSNorm timestep 注入、flow matching 和 FAST/flow 信息隔离。\n",
    "\n",
    "本项目缩小：视觉编码器、词表、隐藏宽度、层数、图像尺寸、动作维度、动作 horizon 和训练数据。默认从零训练，不需要 PaliGemma 或其他预训练权重。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "architecture",
   "metadata": {},
   "source": [
    "## 2. 架构介绍\n",
    "\n",
    "```text\n",
    "高层调用\n",
    "image + discrete state + overall task\n",
    "                  │\n",
    "                  ▼\n",
    "         shared VLM backbone\n",
    "                  │ autoregressive\n",
    "                  ▼\n",
    "          predicted subtask\n",
    "\n",
    "低层调用\n",
    "image + discrete state + predicted subtask\n",
    "                  │\n",
    "noisy action + flow timestep\n",
    "                  ▼\n",
    "        separate action expert\n",
    "                  │ bidirectional action attention\n",
    "                  ▼\n",
    "       flow velocity → action chunk\n",
    "```\n",
    "\n",
    "离散 FAST-lite 动作由 VLM 自回归预测，用于验证离散动作训练路径；连续 action expert 不读取 teacher-forced FAST token。高层和低层使用同一个模型权重，但采用不同输入序列。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "environment",
   "metadata": {},
   "source": [
    "## 3. 环境安装\n",
    "\n",
    "推荐先在仓库根目录启动 Jupyter：\n",
    "\n",
    "```bash\n",
    "python3 -m venv .venv\n",
    "source .venv/bin/activate\n",
    "python -m pip install -e '.[dev,notebook]'\n",
    "jupyter lab notebooks/pi05_minimal_reproduction.ipynb\n",
    "```\n",
    "\n",
    "如果已经在 Colab 或其他 Jupyter 环境中打开，本节代码会定位仓库并安装项目运行依赖。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "locate-repo",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:04:52.000020Z",
     "iopub.status.busy": "2026-07-12T13:04:51.999659Z",
     "iopub.status.idle": "2026-07-12T13:04:52.188244Z",
     "shell.execute_reply": "2026-07-12T13:04:52.090512Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Repository: <REPO_ROOT>\n",
      "Python: <REPO_ROOT>/.venv/bin/python\n",
      "Version: 3.12.13\n"
     ]
    }
   ],
   "source": [
    "from pathlib import Path\n",
    "import os\n",
    "import subprocess\n",
    "import sys\n",
    "\n",
    "candidates = [Path.cwd(), *Path.cwd().parents]\n",
    "REPO_ROOT = next(\n",
    "    path for path in candidates\n",
    "    if (path / 'pyproject.toml').is_file() and (path / 'src/pi05mini').is_dir()\n",
    ")\n",
    "os.chdir(REPO_ROOT)\n",
    "print('Repository:', REPO_ROOT)\n",
    "print('Python:', sys.executable)\n",
    "print('Version:', sys.version.split()[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "install-project",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:04:52.651839Z",
     "iopub.status.busy": "2026-07-12T13:04:52.649844Z",
     "iopub.status.idle": "2026-07-12T13:05:33.223410Z",
     "shell.execute_reply": "2026-07-12T13:05:33.219967Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Obtaining file://<REPO_ROOT>\n",
      "  Installing build dependencies: started\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Installing build dependencies: finished with status 'done'\n",
      "  Checking if build backend supports build_editable: started\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Checking if build backend supports build_editable: finished with status 'done'\n",
      "  Getting requirements to build editable: started\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Getting requirements to build editable: finished with status 'done'\n",
      "  Preparing editable metadata (pyproject.toml): started\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Preparing editable metadata (pyproject.toml): finished with status 'done'\n",
      "Requirement already satisfied: matplotlib>=3.8 in ./.venv/lib/python3.12/site-packages (from pi05-minimal-validation==0.1.0) (3.11.0)\n",
      "Requirement already satisfied: numpy>=1.26 in ./.venv/lib/python3.12/site-packages (from pi05-minimal-validation==0.1.0) (2.5.1)\n",
      "Requirement already satisfied: Pillow>=10.0 in ./.venv/lib/python3.12/site-packages (from pi05-minimal-validation==0.1.0) (12.3.0)\n",
      "Requirement already satisfied: PyYAML>=6.0 in ./.venv/lib/python3.12/site-packages (from pi05-minimal-validation==0.1.0) (6.0.3)\n",
      "Requirement already satisfied: torch>=2.2 in ./.venv/lib/python3.12/site-packages (from pi05-minimal-validation==0.1.0) (2.13.0)\n",
      "Requirement already satisfied: pytest>=8.0 in ./.venv/lib/python3.12/site-packages (from pi05-minimal-validation==0.1.0) (9.1.1)\n",
      "Requirement already satisfied: contourpy>=1.0.1 in ./.venv/lib/python3.12/site-packages (from matplotlib>=3.8->pi05-minimal-validation==0.1.0) (1.3.3)\n",
      "Requirement already satisfied: cycler>=0.10 in ./.venv/lib/python3.12/site-packages (from matplotlib>=3.8->pi05-minimal-validation==0.1.0) (0.12.1)\n",
      "Requirement already satisfied: fonttools>=4.22.0 in ./.venv/lib/python3.12/site-packages (from matplotlib>=3.8->pi05-minimal-validation==0.1.0) (4.63.0)\n",
      "Requirement already satisfied: kiwisolver>=1.3.1 in ./.venv/lib/python3.12/site-packages (from matplotlib>=3.8->pi05-minimal-validation==0.1.0) (1.5.0)\n",
      "Requirement already satisfied: packaging>=20.0 in ./.venv/lib/python3.12/site-packages (from matplotlib>=3.8->pi05-minimal-validation==0.1.0) (26.2)\n",
      "Requirement already satisfied: pyparsing>=3 in ./.venv/lib/python3.12/site-packages (from matplotlib>=3.8->pi05-minimal-validation==0.1.0) (3.3.2)\n",
      "Requirement already satisfied: python-dateutil>=2.7 in ./.venv/lib/python3.12/site-packages (from matplotlib>=3.8->pi05-minimal-validation==0.1.0) (2.9.0.post0)\n",
      "Requirement already satisfied: iniconfig>=1.0.1 in ./.venv/lib/python3.12/site-packages (from pytest>=8.0->pi05-minimal-validation==0.1.0) (2.3.0)\n",
      "Requirement already satisfied: pluggy<2,>=1.5 in ./.venv/lib/python3.12/site-packages (from pytest>=8.0->pi05-minimal-validation==0.1.0) (1.6.0)\n",
      "Requirement already satisfied: pygments>=2.7.2 in ./.venv/lib/python3.12/site-packages (from pytest>=8.0->pi05-minimal-validation==0.1.0) (2.20.0)\n",
      "Requirement already satisfied: six>=1.5 in ./.venv/lib/python3.12/site-packages (from python-dateutil>=2.7->matplotlib>=3.8->pi05-minimal-validation==0.1.0) (1.17.0)\n",
      "Requirement already satisfied: filelock in ./.venv/lib/python3.12/site-packages (from torch>=2.2->pi05-minimal-validation==0.1.0) (3.29.7)\n",
      "Requirement already satisfied: typing-extensions>=4.10.0 in ./.venv/lib/python3.12/site-packages (from torch>=2.2->pi05-minimal-validation==0.1.0) (4.16.0)\n",
      "Requirement already satisfied: setuptools>=77.0.3 in ./.venv/lib/python3.12/site-packages (from torch>=2.2->pi05-minimal-validation==0.1.0) (83.0.0)\n",
      "Requirement already satisfied: sympy>=1.13.3 in ./.venv/lib/python3.12/site-packages (from torch>=2.2->pi05-minimal-validation==0.1.0) (1.14.0)\n",
      "Requirement already satisfied: networkx>=2.5.1 in ./.venv/lib/python3.12/site-packages (from torch>=2.2->pi05-minimal-validation==0.1.0) (3.6.1)\n",
      "Requirement already satisfied: jinja2 in ./.venv/lib/python3.12/site-packages (from torch>=2.2->pi05-minimal-validation==0.1.0) (3.1.6)\n",
      "Requirement already satisfied: fsspec>=0.8.5 in ./.venv/lib/python3.12/site-packages (from torch>=2.2->pi05-minimal-validation==0.1.0) (2026.6.0)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: mpmath<1.4,>=1.1.0 in ./.venv/lib/python3.12/site-packages (from sympy>=1.13.3->torch>=2.2->pi05-minimal-validation==0.1.0) (1.3.0)\n",
      "Requirement already satisfied: MarkupSafe>=2.0 in ./.venv/lib/python3.12/site-packages (from jinja2->torch>=2.2->pi05-minimal-validation==0.1.0) (3.0.3)\n",
      "Building wheels for collected packages: pi05-minimal-validation\n",
      "  Building editable for pi05-minimal-validation (pyproject.toml): started\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  Building editable for pi05-minimal-validation (pyproject.toml): finished with status 'done'\n",
      "  Created wheel for pi05-minimal-validation: filename=pi05_minimal_validation-0.1.0-0.editable-py3-none-any.whl size=5429 sha256=bf15b87e7ebe8d23c0410e8b62767478b5c20f9c6ef08f4dcb0577eea7baec8e\n",
      "  Stored in directory: /private/var/folders/kn/1kgg956d3f94tf10r577mty80000gn/T/pip-ephem-wheel-cache-79_89p34/wheels/9c/b2/79/aadb78a275779eb8bbd7f70ce6f3b2ad0d33aa75a492838aaf\n",
      "Successfully built pi05-minimal-validation\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Installing collected packages: pi05-minimal-validation\n",
      "  Attempting uninstall: pi05-minimal-validation\n",
      "    Found existing installation: pi05-minimal-validation 0.1.0\n",
      "    Uninstalling pi05-minimal-validation-0.1.0:\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "      Successfully uninstalled pi05-minimal-validation-0.1.0\n",
      "Successfully installed pi05-minimal-validation-0.1.0\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "PyTorch: 2.13.0\n",
      "MPS available: True\n",
      "CUDA available: False\n"
     ]
    }
   ],
   "source": [
    "subprocess.run(\n",
    "    [sys.executable, '-m', 'pip', 'install', '-e', '.[dev]'],\n",
    "    check=True,\n",
    ")\n",
    "\n",
    "import torch\n",
    "import numpy as np\n",
    "import yaml\n",
    "from PIL import Image\n",
    "\n",
    "print('PyTorch:', torch.__version__)\n",
    "print('MPS available:', torch.backends.mps.is_available())\n",
    "print('CUDA available:', torch.cuda.is_available())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "data-weights",
   "metadata": {},
   "source": [
    "## 4. 数据集与权重准备\n",
    "\n",
    "数据按以下优先级解析：\n",
    "\n",
    "1. 环境变量 `PI05_SOURCE_MANIFEST=/absolute/path/overfit_manifest.jsonl`；\n",
    "2. 环境变量 `PI05_DATA_ARCHIVE_URL=https://.../dataset.zip`，下载并解压后搜索 manifest；\n",
    "3. 相邻 Alpamayo 项目的默认本地路径；\n",
    "4. 如果以上均不存在，自动生成一份确定性图片与二维轨迹，用于完整验证代码链。\n",
    "\n",
    "权重默认从零训练。若要跳过训练，可设置 `PI05_CHECKPOINT=/absolute/path/checkpoint.pt`，或设置 `PI05_CHECKPOINT_URL` 下载与当前代码版本兼容的 checkpoint。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "prepare-data",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:05:33.282020Z",
     "iopub.status.busy": "2026-07-12T13:05:33.268673Z",
     "iopub.status.idle": "2026-07-12T13:05:33.381684Z",
     "shell.execute_reply": "2026-07-12T13:05:33.359915Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Source manifest: <LOCAL_DATA_ROOT>/data/processed/overfit_manifest.jsonl\n",
      "{\"clip_id\": \"314f59c2-e420-4fcd-9df2-a51d572e0c20\", \"t0_us\": 1503899, \"reasoning\": \"Steer left following the temporary lane delineated by traffic cones through the construction zone.\", \"image_path\": \"data/processed/images/sample_00_314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899.jpg\", \"trajectory_path\": \"data/processed/trajectories/sample_00_314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899.npz\", \"camera\": \"camera_front_wide_120fov\", \"history_steps\": 8, \"expert_waypoints\": 16}\n"
     ]
    }
   ],
   "source": [
    "import json\n",
    "import shutil\n",
    "import urllib.request\n",
    "\n",
    "ARTIFACT_ROOT = REPO_ROOT / 'notebook_artifacts'\n",
    "ARTIFACT_ROOT.mkdir(parents=True, exist_ok=True)\n",
    "\n",
    "manifest_override = os.environ.get('PI05_SOURCE_MANIFEST')\n",
    "archive_url = os.environ.get('PI05_DATA_ARCHIVE_URL')\n",
    "default_manifest = (\n",
    "    REPO_ROOT / '../../alpamayo/alpamayo-qwen05b-validation/data/processed/overfit_manifest.jsonl'\n",
    ").resolve()\n",
    "\n",
    "if manifest_override:\n",
    "    SOURCE_MANIFEST = Path(manifest_override).expanduser().resolve()\n",
    "elif archive_url:\n",
    "    archive_path = ARTIFACT_ROOT / 'dataset.zip'\n",
    "    extract_path = ARTIFACT_ROOT / 'downloaded_dataset'\n",
    "    print('Downloading dataset:', archive_url)\n",
    "    urllib.request.urlretrieve(archive_url, archive_path)\n",
    "    if extract_path.exists():\n",
    "        shutil.rmtree(extract_path)\n",
    "    shutil.unpack_archive(archive_path, extract_path)\n",
    "    manifests = sorted(extract_path.rglob('overfit_manifest.jsonl'))\n",
    "    if not manifests:\n",
    "        raise FileNotFoundError('Downloaded archive contains no overfit_manifest.jsonl')\n",
    "    SOURCE_MANIFEST = manifests[0]\n",
    "elif default_manifest.is_file():\n",
    "    SOURCE_MANIFEST = default_manifest\n",
    "else:\n",
    "    source_root = ARTIFACT_ROOT / 'synthetic_source'\n",
    "    processed = source_root / 'data/processed'\n",
    "    image_dir = processed / 'images'\n",
    "    trajectory_dir = processed / 'trajectories'\n",
    "    image_dir.mkdir(parents=True, exist_ok=True)\n",
    "    trajectory_dir.mkdir(parents=True, exist_ok=True)\n",
    "\n",
    "    pixels = np.zeros((64, 96, 3), dtype=np.uint8)\n",
    "    pixels[..., 0] = 48\n",
    "    pixels[..., 1] = np.arange(96, dtype=np.uint8)[None, :]\n",
    "    pixels[..., 2] = np.arange(64, dtype=np.uint8)[:, None]\n",
    "    Image.fromarray(pixels).save(image_dir / 'sample.jpg')\n",
    "\n",
    "    history_x = np.linspace(-7.0, 0.0, 8, dtype=np.float32)\n",
    "    history = np.stack((history_x, 0.02 * history_x, np.zeros(8)), axis=-1)\n",
    "    future_x = np.linspace(1.0, 32.0, 16, dtype=np.float32)\n",
    "    future = np.stack((future_x, 0.002 * future_x**2, np.zeros(16)), axis=-1)\n",
    "    np.savez_compressed(\n",
    "        trajectory_dir / 'sample.npz',\n",
    "        history_xyz=history,\n",
    "        future_xyz=future,\n",
    "    )\n",
    "\n",
    "    SOURCE_MANIFEST = processed / 'overfit_manifest.jsonl'\n",
    "    row = {\n",
    "        'clip_id': 'notebook-synthetic',\n",
    "        't0_us': 0,\n",
    "        'task_instruction': 'Follow the visible road.',\n",
    "        'reasoning': 'Steer left through the construction lane.',\n",
    "        'image_path': 'data/processed/images/sample.jpg',\n",
    "        'trajectory_path': 'data/processed/trajectories/sample.npz',\n",
    "    }\n",
    "    SOURCE_MANIFEST.write_text(json.dumps(row) + '\\n', encoding='utf-8')\n",
    "\n",
    "if not SOURCE_MANIFEST.is_file():\n",
    "    raise FileNotFoundError(SOURCE_MANIFEST)\n",
    "print('Source manifest:', SOURCE_MANIFEST)\n",
    "print(SOURCE_MANIFEST.read_text(encoding='utf-8').splitlines()[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "prepare-checkpoint",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:05:33.415691Z",
     "iopub.status.busy": "2026-07-12T13:05:33.412018Z",
     "iopub.status.idle": "2026-07-12T13:05:33.434599Z",
     "shell.execute_reply": "2026-07-12T13:05:33.432723Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "External checkpoint: none; train from scratch\n"
     ]
    }
   ],
   "source": [
    "checkpoint_override = os.environ.get('PI05_CHECKPOINT')\n",
    "checkpoint_url = os.environ.get('PI05_CHECKPOINT_URL')\n",
    "\n",
    "if checkpoint_override:\n",
    "    CHECKPOINT_PATH = Path(checkpoint_override).expanduser().resolve()\n",
    "elif checkpoint_url:\n",
    "    CHECKPOINT_PATH = ARTIFACT_ROOT / 'downloaded_checkpoint.pt'\n",
    "    print('Downloading checkpoint:', checkpoint_url)\n",
    "    urllib.request.urlretrieve(checkpoint_url, CHECKPOINT_PATH)\n",
    "else:\n",
    "    CHECKPOINT_PATH = None\n",
    "\n",
    "if CHECKPOINT_PATH is not None and not CHECKPOINT_PATH.is_file():\n",
    "    raise FileNotFoundError(CHECKPOINT_PATH)\n",
    "print('External checkpoint:', CHECKPOINT_PATH or 'none; train from scratch')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "inspect-contract",
   "metadata": {},
   "source": [
    "## 5. 检查层级 token 与数据接口\n",
    "\n",
    "这一节直接检查高层 prefix 包含总体任务，而低层 prefix 不包含总体任务，只包含预测/teacher-forced subtask。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "inspect-tokenizer",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:05:33.453241Z",
     "iopub.status.busy": "2026-07-12T13:05:33.452808Z",
     "iopub.status.idle": "2026-07-12T13:05:33.984713Z",
     "shell.execute_reply": "2026-07-12T13:05:33.976217Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Samples: 4\n",
      "Task: Follow the visible road.\n",
      "Subtask: steer_left\n",
      "High-level prefix length: 22\n",
      "Low-level prefix length: 21\n",
      "High-level tail: [259, 260, 267]\n",
      "Low-level tail: [268, 270]\n",
      "Hierarchical token contract: PASS\n"
     ]
    }
   ],
   "source": [
    "from pi05mini.data import Pi05OverfitDataset, materialize_subset\n",
    "from pi05mini.tokenizer import MiniTokenizer\n",
    "\n",
    "materialized_manifest = materialize_subset(\n",
    "    SOURCE_MANIFEST, REPO_ROOT / 'data/overfit', max_samples=1\n",
    ")\n",
    "dataset = Pi05OverfitDataset(\n",
    "    materialized_manifest,\n",
    "    image_size=64,\n",
    "    augmentation_factor=4,\n",
    "    action_horizon=16,\n",
    ")\n",
    "sample = dataset[0]\n",
    "tokenizer = dataset.tokenizer\n",
    "\n",
    "print('Samples:', len(dataset))\n",
    "print('Task:', sample['task_instruction'])\n",
    "print('Subtask:', sample['subtask_label'])\n",
    "print('High-level prefix length:', len(sample['prefix_tokens']))\n",
    "print('Low-level prefix length:', len(sample['low_level_prefix_tokens']))\n",
    "print('High-level tail:', sample['prefix_tokens'][-3:].tolist())\n",
    "print('Low-level tail:', sample['low_level_prefix_tokens'][-2:].tolist())\n",
    "\n",
    "assert tokenizer.TASK in sample['prefix_tokens']\n",
    "assert tokenizer.SUBTASK_QUERY == int(sample['prefix_tokens'][-1])\n",
    "assert tokenizer.TASK not in sample['low_level_prefix_tokens']\n",
    "assert tokenizer.SUBTASK_COMMAND == int(sample['low_level_prefix_tokens'][-2])\n",
    "assert int(sample['low_level_prefix_tokens'][-1]) == int(sample['subtask_target'])\n",
    "print('Hierarchical token contract: PASS')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "tests",
   "metadata": {},
   "source": [
    "## 6. 运行单元测试\n",
    "\n",
    "测试覆盖数据接口、task/subtask 层级 token、FAST-lite、blockwise attention、flow matching 和模型输出形状。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "run-tests",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:05:33.999296Z",
     "iopub.status.busy": "2026-07-12T13:05:33.998889Z",
     "iopub.status.idle": "2026-07-12T13:05:52.941689Z",
     "shell.execute_reply": "2026-07-12T13:05:52.939709Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m.\u001b[0m\u001b[32m                                                           [100%]\u001b[0m\n",
      "\u001b[32m\u001b[32m\u001b[1m14 passed\u001b[0m\u001b[32m in 13.15s\u001b[0m\u001b[0m\n",
      "\n"
     ]
    }
   ],
   "source": [
    "test_result = subprocess.run(\n",
    "    [sys.executable, '-m', 'pytest', '-q'],\n",
    "    cwd=REPO_ROOT,\n",
    "    text=True,\n",
    "    capture_output=True,\n",
    ")\n",
    "print(test_result.stdout)\n",
    "if test_result.stderr:\n",
    "    print(test_result.stderr)\n",
    "test_result.check_returncode()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "config",
   "metadata": {},
   "source": [
    "## 7. 生成实验配置\n",
    "\n",
    "默认执行论文结构验证所用的完整最小预算：300 步离散预训练和 500 步联合后训练。若只想快速检查 notebook，可在启动前设置 `PI05_FAST_NOTEBOOK=1`，此时使用 40+80 步。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "write-config",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:05:52.948005Z",
     "iopub.status.busy": "2026-07-12T13:05:52.947688Z",
     "iopub.status.idle": "2026-07-12T13:05:52.980574Z",
     "shell.execute_reply": "2026-07-12T13:05:52.965976Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "seed: 7\n",
      "device: auto\n",
      "output_dir: <REPO_ROOT>/notebook_artifacts/run/outputs\n",
      "data:\n",
      "  source_manifest: <LOCAL_DATA_ROOT>/data/processed/overfit_manifest.jsonl\n",
      "  num_variants: 4\n",
      "  image_size: 64\n",
      "  action_horizon: 16\n",
      "model:\n",
      "  d_model: 96\n",
      "  n_heads: 4\n",
      "  n_layers: 3\n",
      "  mlp_ratio: 3\n",
      "  patch_size: 16\n",
      "  max_text_tokens: 64\n",
      "  dropout: 0.0\n",
      "train:\n",
      "  pretrain_steps: 300\n",
      "  posttrain_steps: 500\n",
      "  batch_size: 4\n",
      "  learning_rate: 0.001\n",
      "  weight_decay: 0.0\n",
      "  grad_clip_norm: 1.0\n",
      "  semantic_loss_weight: 1.0\n",
      "  flow_loss_weight: 2.0\n",
      "  log_every: 20\n",
      "inference:\n",
      "  flow_steps: 10\n",
      "\n"
     ]
    }
   ],
   "source": [
    "FAST_NOTEBOOK = os.environ.get('PI05_FAST_NOTEBOOK', '0') == '1'\n",
    "pretrain_steps, posttrain_steps = (40, 80) if FAST_NOTEBOOK else (300, 500)\n",
    "RUN_ROOT = ARTIFACT_ROOT / 'run'\n",
    "RUN_ROOT.mkdir(parents=True, exist_ok=True)\n",
    "CONFIG_PATH = RUN_ROOT / 'overfit.yaml'\n",
    "\n",
    "notebook_config = {\n",
    "    'seed': 7,\n",
    "    'device': 'auto',\n",
    "    'output_dir': str((RUN_ROOT / 'outputs').resolve()),\n",
    "    'data': {\n",
    "        'source_manifest': str(SOURCE_MANIFEST.resolve()),\n",
    "        'num_variants': 4,\n",
    "        'image_size': 64,\n",
    "        'action_horizon': 16,\n",
    "    },\n",
    "    'model': {\n",
    "        'd_model': 96,\n",
    "        'n_heads': 4,\n",
    "        'n_layers': 3,\n",
    "        'mlp_ratio': 3,\n",
    "        'patch_size': 16,\n",
    "        'max_text_tokens': 64,\n",
    "        'dropout': 0.0,\n",
    "    },\n",
    "    'train': {\n",
    "        'pretrain_steps': pretrain_steps,\n",
    "        'posttrain_steps': posttrain_steps,\n",
    "        'batch_size': 4,\n",
    "        'learning_rate': 1e-3,\n",
    "        'weight_decay': 0.0,\n",
    "        'grad_clip_norm': 1.0,\n",
    "        'semantic_loss_weight': 1.0,\n",
    "        'flow_loss_weight': 2.0,\n",
    "        'log_every': 20,\n",
    "    },\n",
    "    'inference': {'flow_steps': 10},\n",
    "}\n",
    "CONFIG_PATH.write_text(\n",
    "    yaml.safe_dump(notebook_config, sort_keys=False), encoding='utf-8'\n",
    ")\n",
    "print(CONFIG_PATH.read_text(encoding='utf-8'))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "training",
   "metadata": {},
   "source": [
    "## 8. 训练或导入权重\n",
    "\n",
    "没有提供外部 checkpoint 时，本单元从零训练。Stage 1 学习总体任务条件下的 subtask 与 FAST-lite token；Stage 2 以两个调用联合训练同一个模型：高层 token CE 与低层 subtask-conditioned flow loss。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "train-model",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:05:52.989477Z",
     "iopub.status.busy": "2026-07-12T13:05:52.988597Z",
     "iopub.status.idle": "2026-07-12T13:10:54.510191Z",
     "shell.execute_reply": "2026-07-12T13:10:54.509235Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 1, \"loss\": 80.38571166992188, \"text_loss\": 80.38571166992188, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 20, \"loss\": 2.5035173892974854, \"text_loss\": 2.5035173892974854, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 40, \"loss\": 1.3843878507614136, \"text_loss\": 1.3843878507614136, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 60, \"loss\": 0.816001296043396, \"text_loss\": 0.816001296043396, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 80, \"loss\": 0.19024920463562012, \"text_loss\": 0.19024920463562012, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 100, \"loss\": 0.058376140892505646, \"text_loss\": 0.058376140892505646, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 120, \"loss\": 0.005495623219758272, \"text_loss\": 0.005495623219758272, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 140, \"loss\": 0.0006469746003858745, \"text_loss\": 0.0006469746003858745, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 160, \"loss\": 0.0001160374958999455, \"text_loss\": 0.0001160374958999455, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 180, \"loss\": 4.6215052861953154e-05, \"text_loss\": 4.6215052861953154e-05, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 200, \"loss\": 3.21573497785721e-05, \"text_loss\": 3.21573497785721e-05, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 220, \"loss\": 2.635362761793658e-05, \"text_loss\": 2.635362761793658e-05, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 240, \"loss\": 2.2365060431184247e-05, \"text_loss\": 2.2365060431184247e-05, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 260, \"loss\": 1.9384962797630578e-05, \"text_loss\": 1.9384962797630578e-05, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 280, \"loss\": 1.701932524156291e-05, \"text_loss\": 1.701932524156291e-05, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"pretrain\", \"step\": 300, \"loss\": 1.5113296285562683e-05, \"text_loss\": 1.5113296285562683e-05, \"flow_loss\": null}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 1, \"loss\": 3.92230224609375, \"text_loss\": 1.5026464097900316e-05, \"flow_loss\": 1.9611436128616333}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 20, \"loss\": 4.334129810333252, \"text_loss\": 6.012214271322591e-06, \"flow_loss\": 2.1670618057250977}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 40, \"loss\": 2.9611411094665527, \"text_loss\": 1.9754215827560984e-06, \"flow_loss\": 1.48056960105896}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 60, \"loss\": 1.8074257373809814, \"text_loss\": 4.7257489654839446e-07, \"flow_loss\": 0.9037126302719116}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 80, \"loss\": 1.9522995948791504, \"text_loss\": 3.6688338695967104e-06, \"flow_loss\": 0.9761479496955872}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 100, \"loss\": 2.660996913909912, \"text_loss\": 5.381422170103178e-07, \"flow_loss\": 1.330498218536377}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 120, \"loss\": 1.4003163576126099, \"text_loss\": 3.1249706466951466e-07, \"flow_loss\": 0.7001579999923706}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 140, \"loss\": 0.40199264883995056, \"text_loss\": 2.946158588201797e-07, \"flow_loss\": 0.20099617540836334}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 160, \"loss\": 0.4382443428039551, \"text_loss\": 4.0020243119442966e-08, \"flow_loss\": 0.21912215650081635}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 180, \"loss\": 0.19535040855407715, \"text_loss\": 9.536735490200954e-08, \"flow_loss\": 0.09767515957355499}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 200, \"loss\": 0.6647185683250427, \"text_loss\": 3.746574961382976e-08, \"flow_loss\": 0.332359254360199}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 220, \"loss\": 0.1602100431919098, \"text_loss\": 7.067399820925857e-08, \"flow_loss\": 0.08010498434305191}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 240, \"loss\": 0.4512874484062195, \"text_loss\": 7.32284206605982e-08, \"flow_loss\": 0.22564369440078735}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 260, \"loss\": 0.42879992723464966, \"text_loss\": 6.130751728505857e-08, \"flow_loss\": 0.21439993381500244}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 280, \"loss\": 0.5335817933082581, \"text_loss\": 1.8307034110875975e-07, \"flow_loss\": 0.26679080724716187}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 300, \"loss\": 0.24511851370334625, \"text_loss\": 3.295255339708092e-07, \"flow_loss\": 0.1225590929389}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 320, \"loss\": 0.2449100911617279, \"text_loss\": 6.301056032498309e-08, \"flow_loss\": 0.12245501577854156}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 340, \"loss\": 0.09267646819353104, \"text_loss\": 7.067376373015577e-08, \"flow_loss\": 0.04633820056915283}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 360, \"loss\": 0.11380708962678909, \"text_loss\": 1.6178400841226903e-08, \"flow_loss\": 0.05690353736281395}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 380, \"loss\": 0.7959573268890381, \"text_loss\": 1.1069432481747299e-08, \"flow_loss\": 0.39797866344451904}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 400, \"loss\": 0.0984947681427002, \"text_loss\": 7.663453871487036e-09, \"flow_loss\": 0.0492473803460598}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 420, \"loss\": 0.13179005682468414, \"text_loss\": 1.192092735635697e-08, \"flow_loss\": 0.06589502096176147}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 440, \"loss\": 0.13048851490020752, \"text_loss\": 1.1410003253331524e-07, \"flow_loss\": 0.06524419784545898}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 460, \"loss\": 0.13017220795154572, \"text_loss\": 7.663453871487036e-09, \"flow_loss\": 0.06508609652519226}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 480, \"loss\": 0.1828930377960205, \"text_loss\": 1.0217936718959209e-08, \"flow_loss\": 0.09144651144742966}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\"stage\": \"posttrain\", \"step\": 500, \"loss\": 0.21158601343631744, \"text_loss\": 1.4475405762937044e-08, \"flow_loss\": 0.10579299926757812}\n",
      "Checkpoint: <REPO_ROOT>/notebook_artifacts/run/outputs/checkpoint.pt\n"
     ]
    }
   ],
   "source": [
    "from pi05mini.train import train\n",
    "\n",
    "if CHECKPOINT_PATH is None:\n",
    "    CHECKPOINT_PATH = train(str(CONFIG_PATH))\n",
    "else:\n",
    "    print('Skipping training; using:', CHECKPOINT_PATH)\n",
    "\n",
    "CHECKPOINT_PATH = Path(CHECKPOINT_PATH).resolve()\n",
    "assert CHECKPOINT_PATH.is_file()\n",
    "print('Checkpoint:', CHECKPOINT_PATH)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "inference",
   "metadata": {},
   "source": [
    "## 9. 推理与实验结果\n",
    "\n",
    "推理先从总体任务分支生成 subtask，再用预测 subtask 构造低层 prefix。最终同时比较 ground truth、VLM/FAST-lite 离散轨迹和 flow matching 连续轨迹。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "run-inference",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:10:54.514710Z",
     "iopub.status.busy": "2026-07-12T13:10:54.514174Z",
     "iopub.status.idle": "2026-07-12T13:11:04.860027Z",
     "shell.execute_reply": "2026-07-12T13:11:04.859236Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[{'sample_id': '314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899:aug0', 'target_subtask': 'steer_left', 'task_instruction': 'Follow the visible road.', 'predicted_subtask': 'steer_left', 'vlm_action_tokens_valid': True, 'vlm_trajectory_mse': 0.07854792475700378, 'diffusion_trajectory_mse': 0.08387580513954163}, {'sample_id': '314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899:aug1', 'target_subtask': 'steer_left', 'task_instruction': 'Follow the visible road.', 'predicted_subtask': 'steer_left', 'vlm_action_tokens_valid': True, 'vlm_trajectory_mse': 0.025470610707998276, 'diffusion_trajectory_mse': 0.08775432407855988}, {'sample_id': '314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899:aug2', 'target_subtask': 'steer_left', 'task_instruction': 'Follow the visible road.', 'predicted_subtask': 'steer_left', 'vlm_action_tokens_valid': True, 'vlm_trajectory_mse': 0.013774838298559189, 'diffusion_trajectory_mse': 0.045165739953517914}, {'sample_id': '314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899:aug3', 'target_subtask': 'steer_left', 'task_instruction': 'Follow the visible road.', 'predicted_subtask': 'steer_left', 'vlm_action_tokens_valid': True, 'vlm_trajectory_mse': 0.059195056557655334, 'diffusion_trajectory_mse': 0.019245963543653488}]\n",
      "[\n",
      "  {\n",
      "    \"sample_id\": \"314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899:aug0\",\n",
      "    \"target_subtask\": \"steer_left\",\n",
      "    \"task_instruction\": \"Follow the visible road.\",\n",
      "    \"predicted_subtask\": \"steer_left\",\n",
      "    \"vlm_action_tokens_valid\": true,\n",
      "    \"vlm_trajectory_mse\": 0.07854792475700378,\n",
      "    \"diffusion_trajectory_mse\": 0.08387580513954163\n",
      "  },\n",
      "  {\n",
      "    \"sample_id\": \"314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899:aug1\",\n",
      "    \"target_subtask\": \"steer_left\",\n",
      "    \"task_instruction\": \"Follow the visible road.\",\n",
      "    \"predicted_subtask\": \"steer_left\",\n",
      "    \"vlm_action_tokens_valid\": true,\n",
      "    \"vlm_trajectory_mse\": 0.025470610707998276,\n",
      "    \"diffusion_trajectory_mse\": 0.08775432407855988\n",
      "  },\n",
      "  {\n",
      "    \"sample_id\": \"314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899:aug2\",\n",
      "    \"target_subtask\": \"steer_left\",\n",
      "    \"task_instruction\": \"Follow the visible road.\",\n",
      "    \"predicted_subtask\": \"steer_left\",\n",
      "    \"vlm_action_tokens_valid\": true,\n",
      "    \"vlm_trajectory_mse\": 0.013774838298559189,\n",
      "    \"diffusion_trajectory_mse\": 0.045165739953517914\n",
      "  },\n",
      "  {\n",
      "    \"sample_id\": \"314f59c2-e420-4fcd-9df2-a51d572e0c20_1503899:aug3\",\n",
      "    \"target_subtask\": \"steer_left\",\n",
      "    \"task_instruction\": \"Follow the visible road.\",\n",
      "    \"predicted_subtask\": \"steer_left\",\n",
      "    \"vlm_action_tokens_valid\": true,\n",
      "    \"vlm_trajectory_mse\": 0.059195056557655334,\n",
      "    \"diffusion_trajectory_mse\": 0.019245963543653488\n",
      "  }\n",
      "]\n",
      "Summary: {\n",
      "  \"device\": \"mps\",\n",
      "  \"samples\": 4,\n",
      "  \"parameters\": 817346,\n",
      "  \"initial\": {\n",
      "    \"stage\": \"pretrain\",\n",
      "    \"step\": 1,\n",
      "    \"loss\": 80.38571166992188,\n",
      "    \"text_loss\": 80.38571166992188,\n",
      "    \"flow_loss\": null\n",
      "  },\n",
      "  \"final\": {\n",
      "    \"stage\": \"posttrain\",\n",
      "    \"step\": 500,\n",
      "    \"loss\": 0.21158601343631744,\n",
      "    \"text_loss\": 1.4475405762937044e-08,\n",
      "    \"flow_loss\": 0.10579299926757812\n",
      "  }\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "from pi05mini.infer import evaluate\n",
    "\n",
    "PLOT_PATH = evaluate(str(CHECKPOINT_PATH), manifest=str(materialized_manifest))\n",
    "OUTPUT_DIR = CHECKPOINT_PATH.parent\n",
    "EVALUATION_PATH = OUTPUT_DIR / 'evaluation.json'\n",
    "SUMMARY_PATH = OUTPUT_DIR / 'summary.json'\n",
    "evaluation = json.loads(EVALUATION_PATH.read_text(encoding='utf-8'))\n",
    "summary = json.loads(SUMMARY_PATH.read_text(encoding='utf-8')) if SUMMARY_PATH.exists() else {}\n",
    "\n",
    "print(json.dumps(evaluation, indent=2, ensure_ascii=False))\n",
    "print('Summary:', json.dumps(summary, indent=2, ensure_ascii=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "display-plot",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:11:04.863087Z",
     "iopub.status.busy": "2026-07-12T13:11:04.862791Z",
     "iopub.status.idle": "2026-07-12T13:11:04.872207Z",
     "shell.execute_reply": "2026-07-12T13:11:04.871634Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from IPython.display import Image as DisplayImage, display\n",
    "\n",
    "display(DisplayImage(filename=str(PLOT_PATH)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "acceptance",
   "metadata": {},
   "source": [
    "## 10. 结构与结果验收\n",
    "\n",
    "下面的断言检查层级连接、预测标签、轨迹指标、checkpoint 和结果图。全部通过表示这份缩小实现能够端到端运行并记忆最小数据；不代表论文规模下的泛化或真实机器人能力。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "accept-results",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-12T13:11:04.876068Z",
     "iopub.status.busy": "2026-07-12T13:11:04.875871Z",
     "iopub.status.idle": "2026-07-12T13:11:04.881710Z",
     "shell.execute_reply": "2026-07-12T13:11:04.880933Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "PASS: overall task -> predicted subtask -> low-level flow action\n",
      "FAST MSE: [0.0785, 0.0255, 0.0138, 0.0592]\n",
      "Flow MSE: [0.0839, 0.0878, 0.0452, 0.0192]\n"
     ]
    }
   ],
   "source": [
    "assert CHECKPOINT_PATH.is_file()\n",
    "assert Path(PLOT_PATH).is_file()\n",
    "assert evaluation\n",
    "assert all(row['predicted_subtask'] == row['target_subtask'] for row in evaluation)\n",
    "assert all(row['vlm_action_tokens_valid'] for row in evaluation)\n",
    "assert all(np.isfinite(row['diffusion_trajectory_mse']) for row in evaluation)\n",
    "assert all(np.isfinite(row['vlm_trajectory_mse']) for row in evaluation)\n",
    "\n",
    "print('PASS: overall task -> predicted subtask -> low-level flow action')\n",
    "print('FAST MSE:', [round(row['vlm_trajectory_mse'], 4) for row in evaluation])\n",
    "print('Flow MSE:', [round(row['diffusion_trajectory_mse'], 4) for row in evaluation])"
   ]
  }
 ],
 "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": "3.12.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}