{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "# 표 파운데이션 모델을 처음부터 학습하는 작은 실습\n",
        "\n",
        "합성 표 생성 → Transformer 정의 → 사전학습 → 새 표에서 문맥 내 추론까지 직접 실행합니다. **정식 TabPFN 재현이 아닌, 그 핵심 아이디어를 보여주는 교육용 모델**입니다. 하나의 수치형 입력과 하나의 정답을 사용해 과정을 곡선으로 볼 수 있게 했습니다.\n",
        "\n",
        "완성된 결과가 포함되어 있어 설치 없이도 읽을 수 있습니다. 직접 학습하려면 Jupyter 또는 Colab에서 위에서 아래로 실행하세요. PDF는 해설용이며 실행 코드는 이 노트북 하나에 들어 있습니다.\n",
        "\n",
        "이 노트북의 모델은 MSE로 점 예측을 학습합니다. 예측 분포나 보정된 불확실성을 출력하지 않으므로 정식 PFN의 확률적 추론 성질을 그대로 주장할 수 없습니다. [TabPFN 원 논문](https://arxiv.org/abs/2207.01848)의 합성 문제 사전학습과 ICL 아이디어를 참고했습니다."
      ],
      "id": "cell-00"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 필요한 패키지 설치\n",
        "이미 설치되어 있다면 건너뜁니다. 아래 줄의 주석을 제거해 실행하세요. CPU만으로도 전체 학습이 가능합니다."
      ],
      "id": "cell-01"
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {},
      "outputs": [],
      "source": [
        "# %pip install torch numpy pandas matplotlib"
      ],
      "id": "cell-02"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 1 실행 환경과 재현성\n",
        "Python 3.10 이상, PyTorch, NumPy, pandas, matplotlib을 사용합니다. 기본 설정은 CPU이며 사전학습 가중치를 다운로드하지 않습니다. 새로 초기화한 모델을 직접 학습합니다. 난수 시드는 초기 가중치·학습 표·검증 표·평가 표를 구분합니다. 하드웨어와 패키지 버전에 따라 시간과 결과가 조금 달라질 수 있습니다.\n",
        "\n",
        "아래 `results` 폴더에 학습 기록과 가중치가 저장됩니다. GPU를 쓰려면 PyTorch에 맞는 CUDA 환경을 준비한 뒤 DEVICE를 바꾸세요."
      ],
      "id": "cell-03"
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {},
      "outputs": [],
      "source": [
        "import copy\n",
        "import json\n",
        "import math\n",
        "import time\n",
        "from pathlib import Path\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import torch\n",
        "from torch import nn\n",
        "\n",
        "SEED = 7\n",
        "np.random.seed(SEED)\n",
        "torch.manual_seed(SEED)\n",
        "torch.set_num_threads(2)\n",
        "DEVICE = 'cpu'  # 학습 흐름을 비교하기 쉽도록 CPU를 기본값으로 사용\n",
        "OUT = Path('results')\n",
        "OUT.mkdir(exist_ok=True)\n"
      ],
      "id": "cell-04"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 2 여러 개의 표 문제 만들기\n",
        "하나의 큰 표를 배치로 나누는 것과 다릅니다. **배치의 각 원소가 서로 다른 규칙을 가진 표 하나**입니다.\n",
        "\n",
        "$$y=a+bx+cx^2+\\epsilon$$\n",
        "\n",
        "각 표에서 a, b를 표준정규분포에서 새로 뽑고 c는 표준편차 0.7인 정규분포에서 뽑습니다. x는 [-1, 1]에서, 잡음 ε는 표준편차 0.05인 정규분포에서 생성합니다. 같은 표 안에서는 a, b, c를 공유하지만 다음 표에서는 바뀝니다. x는 정규화한 공정 조건, y는 응답이라고 상상할 수 있습니다. 실제 소재 물성식은 아닙니다.\n",
        "\n",
        "**한 표에서** 앞의 행들은 관측용 context, 나머지는 질문용 query로 나눕니다. 함수는 질문의 정답도 반환하지만, 그 정답은 손실 계산에만 씁니다. 모델에 a, b, c는 전달하지 않습니다.\n",
        "\n",
        "| 텐서 | 모양 | 의미 |\n",
        "|---|---|---|\n",
        "| x_context | [B, C, 1] | B개 표의 관측 조건 C개 |\n",
        "| y_context | [B, C, 1] | 관측 결과 |\n",
        "| x_query | [B, Q, 1] | 맞혀야 할 조건 Q개 |\n",
        "| y_query | [B, Q] | 손실 계산용 정답 |\n",
        "\n",
        "B=32이면 32개의 다른 규칙을 동시에 풉니다. 한 표의 행들이 서로 다른 규칙을 갖게 만들면 관측으로 규칙을 추론하는 문제가 성립하지 않습니다."
      ],
      "id": "cell-05"
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {},
      "outputs": [],
      "source": [
        "# 하나의 배치 원소가 하나의 독립적인 '표 문제'입니다.\n",
        "def sample_tasks(batch_size, n_context, n_query, generator):\n",
        "    coefficients = torch.randn(batch_size, 3, generator=generator)\n",
        "    coefficients[:, 2] *= 0.7\n",
        "    x = 2 * torch.rand(batch_size, n_context + n_query, 1, generator=generator) - 1\n",
        "    clean_y = (coefficients[:, 0, None] + coefficients[:, 1, None] * x[..., 0]\n",
        "               + coefficients[:, 2, None] * x[..., 0] ** 2)\n",
        "    y = clean_y + 0.05 * torch.randn(clean_y.shape, generator=generator)\n",
        "    return (x[:, :n_context].to(DEVICE), y[:, :n_context, None].to(DEVICE),\n",
        "            x[:, n_context:].to(DEVICE), y[:, n_context:].to(DEVICE))\n"
      ],
      "id": "cell-06"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "### 한 표의 실제 숫자 확인\n",
        "아래 hidden_y는 학습 프로그램만 아는 채점용 정답입니다. forward 입력에는 전달하지 않습니다."
      ],
      "id": "cell-07"
    },
    {
      "cell_type": "code",
      "metadata": {},
      "execution_count": 4,
      "source": [
        "preview_gen = torch.Generator().manual_seed(505)\n",
        "pxc, pyc, pxq, pyq = sample_tasks(1, 4, 2, preview_gen)\n",
        "print(\"관측 행: 모델에 x와 y를 함께 제공합니다.\")\n",
        "print(pd.DataFrame({\"x\": pxc[0, :, 0].cpu(), \"y\": pyc[0, :, 0].cpu()}).round(3).to_string(index=False))\n",
        "print(\"질문 행: 모델에는 x만, loss 계산에는 정답 y를 사용합니다.\")\n",
        "print(pd.DataFrame({\"x_query\": pxq[0, :, 0].cpu(), \"hidden_y\": pyq[0].cpu()}).round(3).to_string(index=False))\n"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "관측 행: 모델에 x와 y를 함께 제공합니다.\n",
            "     x     y\n",
            " 0.480 0.769\n",
            " 0.911 1.660\n",
            "-0.071 0.476\n",
            "-0.893 1.206\n",
            "질문 행: 모델에는 x만, loss 계산에는 정답 y를 사용합니다.\n",
            " x_query  hidden_y\n",
            "   0.544     0.912\n",
            "  -0.777     1.044\n"
          ]
        }
      ],
      "id": "cell-08"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 3 작은 Transformer 직접 정의하기\n",
        "관측 행은 `[x, y, 1]`, 질문 행은 `[x, 0, 0]`으로 만듭니다. 마지막 숫자는 정답을 알고 있는지 나타내는 표시입니다. 관측 y가 실제로 0인 경우와 질문의 임시 0을 구별할 수 있습니다.\n",
        "\n",
        "3개 숫자를 Linear 층으로 48차원 토큰으로 바꿉니다. 위치 임베딩은 사용하지 않습니다. 관측 행 순서가 바뀌어도 같은 표로 처리하려는 설계입니다.\n",
        "\n",
        "**구조:** 입력 임베딩 → Transformer 블록 2개 → LayerNorm → 행별 숫자 예측. 각 블록은 4개 Attention head와 폭 96의 MLP를 사용합니다. 모델을 작게 하고 관측 순열 검증을 단순하게 만들기 위해 dropout은 0으로 둡니다.\n",
        "\n",
        "$$\\mathrm{Attention}(Q,K,V)=\\mathrm{softmax}(QK^T/\\sqrt{d_k}+M)V$$\n",
        "\n",
        "Attention 마스크는 모든 행이 관측 행만 Key/Value로 보도록 합니다.\n",
        "\n",
        "| 읽는 행 / 참조하는 열 | 관측 행 | 질문 행 |\n",
        "|---|---|---|\n",
        "| 관측 행 | 허용 | 차단 |\n",
        "| 질문 행 | 허용 | 차단 |\n",
        "\n",
        "질문 행은 자기 x를 잔차 경로로 유지합니다. 질문끼리 정보를 주고받지 않으므로 한꺼번에 넣는 질문 수가 달라져도 각 예측이 같아야 합니다. 이 설계는 교육용 선택이며 모든 TFM에 의무적인 구조는 아닙니다.\n",
        "\n",
        "**주의:** 함수 인자에 `y_query`가 없습니다. 마스크만으로 누출을 막으려 하지 않고, 정답이 모델에 들어갈 경로 자체를 없앴습니다. PyTorch에서 이 Boolean 마스크의 True는 차단을 의미합니다. [TransformerEncoderLayer 공식 문서](https://docs.pytorch.org/docs/stable/generated/torch.nn.TransformerEncoderLayer.html)"
      ],
      "id": "cell-09"
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {},
      "outputs": [],
      "source": [
        "class TinyTableTransformer(nn.Module):\n",
        "    def __init__(self, width=48, n_heads=4, n_layers=2):\n",
        "        super().__init__()\n",
        "        self.embedding = nn.Linear(3, width)\n",
        "        layer = nn.TransformerEncoderLayer(\n",
        "            d_model=width, nhead=n_heads, dim_feedforward=width * 2,\n",
        "            dropout=0.0, batch_first=True, norm_first=True,\n",
        "            activation='gelu')\n",
        "        self.encoder = nn.TransformerEncoder(\n",
        "            layer, num_layers=n_layers, enable_nested_tensor=False)\n",
        "        # 복제된 층이 서로 다른 초기 가중치를 갖도록 재초기화합니다.\n",
        "        for block in self.encoder.layers:\n",
        "            for param in block.parameters():\n",
        "                if param.ndim > 1:\n",
        "                    nn.init.xavier_uniform_(param)\n",
        "        self.head = nn.Sequential(nn.LayerNorm(width), nn.Linear(width, 1))\n",
        "\n",
        "    def forward(self, x_context, y_context, x_query):\n",
        "        n_context = x_context.shape[1]\n",
        "        context = torch.cat([x_context, y_context, torch.ones_like(y_context)], dim=-1)\n",
        "        # 질문 정답은 이 함수의 인자로도 받지 않습니다.\n",
        "        query = torch.cat([x_query, torch.zeros_like(x_query),\n",
        "                           torch.zeros_like(x_query)], dim=-1)\n",
        "        tokens = torch.cat([context, query], dim=1)\n",
        "        length = tokens.shape[1]\n",
        "        # True는 '볼 수 없음'. 모든 행은 관측 행만 Key/Value로 봅니다.\n",
        "        mask = torch.zeros(length, length, dtype=torch.bool, device=tokens.device)\n",
        "        mask[:, n_context:] = True\n",
        "        hidden = self.encoder(self.embedding(tokens), mask=mask)\n",
        "        return self.head(hidden[:, n_context:]).squeeze(-1)\n"
      ],
      "id": "cell-10"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 4 사전학습과 역전파\n",
        "한 번의 학습 단계는 다음 순서입니다.\n",
        "\n",
        "1. 서로 다른 규칙의 표 32개를 생성합니다.\n",
        "2. 관측 행과 질문 조건을 모델에 넣습니다.\n",
        "3. 질문 예측과 숨겨 둔 정답 사이의 MSE를 구합니다.\n",
        "4. `zero_grad()`로 이전 기울기를 지우고 `backward()`로 새 기울기를 계산합니다.\n",
        "5. `optimizer.step()`으로 공통 가중치 θ를 업데이트합니다.\n",
        "\n",
        "$$L=\\frac{1}{BQ}\\sum_{b=1}^{B}\\sum_{j=1}^{Q}(\\hat y_{bj}-y_{bj})^2$$\n",
        "\n",
        "손실은 질문 행에서 계산하지만 기울기는 Attention을 거쳐 관측 행을 처리하는 층에도 전달됩니다. 즉, 관측 결과에서 유용한 정보를 추출하도록 임베딩과 Attention, MLP가 함께 바뀝니다. 합성 데이터를 만든 a, b, c를 학습하는 것이 아닙니다.\n",
        "\n",
        "1800단계 × 배치 32 = **57,600개 합성 표 문제**를 만납니다. 관측 수는 2, 4, 8, 16을 순환합니다. 매번 새 표를 생성하므로 정해진 데이터셋을 한 바퀴 도는 epoch보다 optimizer step이 적절한 기준입니다.\n",
        "\n",
        "검증은 별도의 고정된 표 64개로 수행합니다. 검증 데이터로 역전파하지 않으며 이 예제에서는 체크포인트 선택에도 쓰지 않습니다. 학습 배치 손실은 관측 수와 문제가 바뀌어 출렁일 수 있습니다. 최종 1800단계 모델을 평가합니다.\n",
        "\n",
        "`N_STEPS=100`은 빠른 동작 확인용이며 충분한 학습 결과를 기대해서는 안 됩니다. 학습 셀을 다시 실행하면 새 모델부터 시작합니다."
      ],
      "id": "cell-11"
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {},
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "   1/1800  train=1.0140  validation=0.8284\n",
            " 100/1800  train=0.0665  validation=0.1236\n",
            " 200/1800  train=0.0338  validation=0.0780\n",
            " 300/1800  train=0.0310  validation=0.0753\n",
            " 400/1800  train=0.0151  validation=0.0576\n",
            " 500/1800  train=0.0198  validation=0.0586\n",
            " 600/1800  train=0.0154  validation=0.0568\n",
            " 700/1800  train=0.0246  validation=0.0436\n",
            " 800/1800  train=0.0196  validation=0.0574\n",
            " 900/1800  train=0.0160  validation=0.0454\n",
            "1000/1800  train=0.0248  validation=0.0484\n",
            "1100/1800  train=0.0118  validation=0.0459\n",
            "1200/1800  train=0.0091  validation=0.0534\n",
            "1300/1800  train=0.0108  validation=0.0456\n",
            "1400/1800  train=0.0112  validation=0.0442\n",
            "1500/1800  train=0.0133  validation=0.0471\n",
            "1600/1800  train=0.0109  validation=0.0393\n",
            "1700/1800  train=0.0159  validation=0.0411\n",
            "1800/1800  train=0.0104  validation=0.0295\n",
            "Parameters: 38,257; time: 14.8s\n"
          ]
        }
      ],
      "source": [
        "model = TinyTableTransformer().to(DEVICE)\n",
        "untrained = copy.deepcopy(model).eval()\n",
        "optimizer = torch.optim.AdamW(model.parameters(), lr=0.0005, weight_decay=0.0001)\n",
        "train_gen = torch.Generator().manual_seed(101)\n",
        "validation_gen = torch.Generator().manual_seed(202)\n",
        "validation = sample_tasks(64, 8, 16, validation_gen)\n",
        "N_STEPS = 1800  # 처음에는 100으로 줄여 동작만 확인해도 됩니다.\n",
        "BATCH_SIZE = 32\n",
        "history = []\n",
        "start = time.perf_counter()\n",
        "for step in range(1, N_STEPS + 1):\n",
        "    model.train()\n",
        "    n_context = [2, 4, 8, 16][(step - 1) % 4]\n",
        "    xc, yc, xq, yq = sample_tasks(BATCH_SIZE, n_context, 16, train_gen)\n",
        "    prediction = model(xc, yc, xq)\n",
        "    loss = nn.functional.mse_loss(prediction, yq)\n",
        "    optimizer.zero_grad()\n",
        "    loss.backward()\n",
        "    nn.utils.clip_grad_norm_(model.parameters(), 1.0)\n",
        "    optimizer.step()\n",
        "    if step == 1 or step % 100 == 0:\n",
        "        model.eval()\n",
        "        with torch.no_grad():\n",
        "            val_loss = nn.functional.mse_loss(model(*validation[:3]), validation[3]).item()\n",
        "        history.append({'step': step, 'train_batch_mse': loss.item(), 'validation_mse': val_loss})\n",
        "        print(f'{step:4d}/{N_STEPS}  train={loss.item():.4f}  validation={val_loss:.4f}', flush=True)\n",
        "elapsed = time.perf_counter() - start\n",
        "model.eval()\n",
        "print(f'Parameters: {sum(p.numel() for p in model.parameters()):,}; time: {elapsed:.1f}s')\n"
      ],
      "id": "cell-12"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 5 처음 보는 표에서 평가하기\n",
        "여기부터 `no_grad()` 안에서만 예측합니다. 새로운 표 128개를 만들고, 각 표에 같은 질문 32개를 둡니다. 관측은 처음 2개 → 4개 → 8개 → 16개로 누적하므로 같은 문제에서 관측 수의 영향을 비교할 수 있습니다.\n",
        "\n",
        "비교 대상은 초기 무작위 Transformer, 사전학습 Transformer, 관측 y의 평균, 다른 표의 y를 잘못 연결한 문맥, 2차 다항식 ridge 회귀입니다. 마지막 모델은 생성 규칙이 2차식이라는 강한 정보를 미리 받습니다. 이 모델을 이기는 것이 실습 목표가 아닙니다.\n",
        "\n",
        "잘못된 문맥 비교는 x를 유지하고 표 사이에서 y를 순환시킵니다. 성능이 나빠지면 관측 정답에 의존한다는 증거가 되지만, 특정 인과 메커니즘을 이해했다는 증명은 아닙니다.\n",
        "\n",
        "점수는 잡음이 있는 질문 정답에 대한 MSE입니다. 표별 MSE의 평균과 표준오차를 보고합니다. 표준오차는 이 128개 문제에서의 변동이며 여러 번 다시 학습했을 때의 변동을 포함하지 않습니다."
      ],
      "id": "cell-13"
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {},
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            " n_context                         model     MSE  task_MSE_SE\n",
            "         2         Untrained Transformer 1.14902      0.11734\n",
            "         2        Pretrained Transformer 0.16033      0.01978\n",
            "         2                  Context mean 0.54193      0.05740\n",
            "         2          Wrong context labels 3.34738      0.38439\n",
            "         2 Quadratic ridge (known basis) 0.14863      0.01950\n",
            "         4         Untrained Transformer 1.18554      0.11524\n",
            "         4        Pretrained Transformer 0.04774      0.00717\n",
            "         4                  Context mean 0.48360      0.05259\n",
            "         4          Wrong context labels 2.75986      0.33362\n",
            "         4 Quadratic ridge (known basis) 0.03356      0.01030\n",
            "         8         Untrained Transformer 1.19084      0.11632\n",
            "         8        Pretrained Transformer 0.01569      0.00287\n",
            "         8                  Context mean 0.44157      0.04417\n",
            "         8          Wrong context labels 2.50719      0.29063\n",
            "         8 Quadratic ridge (known basis) 0.00485      0.00049\n",
            "        16         Untrained Transformer 1.18423      0.11405\n",
            "        16        Pretrained Transformer 0.01137      0.00265\n",
            "        16                  Context mean 0.41850      0.04185\n",
            "        16          Wrong context labels 2.23477      0.28137\n",
            "        16 Quadratic ridge (known basis) 0.00312      0.00010\n"
          ]
        }
      ],
      "source": [
        "# 학습 및 검증과 독립적인 난수 계열로 새 표를 만듭니다.\n",
        "test_gen = torch.Generator().manual_seed(303)\n",
        "xc_all, yc_all, xq, yq = sample_tasks(128, 16, 32, test_gen)\n",
        "rows = []\n",
        "with torch.no_grad():\n",
        "    for n in [2, 4, 8, 16]:\n",
        "        xc, yc = xc_all[:, :n], yc_all[:, :n]\n",
        "        predictions = {\n",
        "            'Untrained Transformer': untrained(xc, yc, xq),\n",
        "            'Pretrained Transformer': model(xc, yc, xq),\n",
        "            'Context mean': yc.mean(dim=1).expand(-1, xq.shape[1]),\n",
        "            'Wrong context labels': model(xc, yc.roll(1, dims=0), xq),\n",
        "        }\n",
        "        # 합성 규칙의 기저를 아는 강력한 비교 기준입니다. 실무의 만능 모델은 아닙니다.\n",
        "        A = torch.cat([torch.ones_like(xc), xc, xc**2], dim=-1)\n",
        "        Q = torch.cat([torch.ones_like(xq), xq, xq**2], dim=-1)\n",
        "        ridge = 0.01 * torch.eye(3, device=DEVICE)[None]\n",
        "        beta = torch.linalg.solve(A.transpose(1, 2) @ A + ridge, A.transpose(1, 2) @ yc)\n",
        "        predictions['Quadratic ridge (known basis)'] = (Q @ beta).squeeze(-1)\n",
        "        for name, pred in predictions.items():\n",
        "            task_mse = ((pred - yq)**2).mean(dim=1)\n",
        "            rows.append({'n_context': n, 'model': name, 'MSE': task_mse.mean().item(),\n",
        "                         'task_MSE_SE': task_mse.std().item() / np.sqrt(len(task_mse))})\n",
        "scores = pd.DataFrame(rows)\n",
        "print(scores.round(5).to_string(index=False))\n"
      ],
      "id": "cell-14"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 6 가중치를 바꾸지 않았는지 확인하기\n",
        "예측 전후의 `state_dict()`를 비교해 모든 값이 같은지 확인합니다. 또한 관측 행과 정답을 같은 순서로 뒤집어 예측이 유지되는지, 질문 하나만 넣어도 결과가 유지되는지 검사합니다.\n",
        "\n",
        "`eval()`은 dropout 등의 동작 모드를 바꾸고, `no_grad()`는 기울기 기록을 끕니다. 실제 업데이트는 optimizer가 담당합니다. 이 단계에는 `backward()`나 `optimizer.step()`이 없습니다. 새로운 정보는 가중치가 아니라 입력과 내부 활성값에 반영됩니다."
      ],
      "id": "cell-15"
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {},
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{'weights_unchanged': True, 'row_permutation_max_abs_difference': 9.5367431640625e-07, 'query_batch_max_abs_difference': 9.5367431640625e-07}\n"
          ]
        }
      ],
      "source": [
        "before = {k: v.clone() for k, v in model.state_dict().items()}\n",
        "with torch.no_grad():\n",
        "    original = model(xc_all[:, :8], yc_all[:, :8], xq)\n",
        "    permutation = torch.arange(7, -1, -1)\n",
        "    permuted = model(xc_all[:, :8][:, permutation], yc_all[:, :8][:, permutation], xq)\n",
        "    # 다른 질문 행을 제거해도 예측이 같은지 확인합니다.\n",
        "    one_query = model(xc_all[:, :8], yc_all[:, :8], xq[:, :1])\n",
        "checks = {\n",
        "    'weights_unchanged': all(torch.equal(before[k], v) for k, v in model.state_dict().items()),\n",
        "    'row_permutation_max_abs_difference': (original - permuted).abs().max().item(),\n",
        "    'query_batch_max_abs_difference': (original[:, :1] - one_query).abs().max().item(),\n",
        "}\n",
        "assert checks['weights_unchanged']\n",
        "assert checks['row_permutation_max_abs_difference'] < 1e-4\n",
        "assert checks['query_batch_max_abs_difference'] < 1e-4\n",
        "print(checks)\n"
      ],
      "id": "cell-16"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 7 같은 가중치에 관측을 추가하면\n",
        "한 개의 가상 실험 규칙 `y=0.4+0.7x−0.9x²`를 정해 관측을 누적합니다. 회색 점선은 우리가 알고 있는 생성 규칙이며, 모델에는 그 식을 알려주지 않습니다. 주황색 점은 관측, 청록색 선은 고정된 모델의 예측입니다.\n",
        "\n",
        "회색 배경의 x < −1 또는 x > 1은 사전학습 때 보지 않은 입력 범위입니다. 그 영역의 오차는 외삽 한계를 관찰하기 위한 것이며, 위 정량 평가의 MSE에는 포함되지 않습니다. 관측이 늘어도 특정 구간의 오차가 반드시 단조롭게 줄지는 않습니다."
      ],
      "id": "cell-17"
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {},
      "outputs": [
        {
          "output_type": "display_data",
          "metadata": {},
          "data": {
            "image/png": "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"
          }
        }
      ],
      "source": [
        "# 관측값을 누적해서 추가합니다. 실제 곡선은 평가에만 사용합니다.\n",
        "grid = torch.linspace(-1.4, 1.4, 180).reshape(1, -1, 1).to(DEVICE)\n",
        "context_x = torch.tensor([-0.85, 0.75, -0.3, 0.15, -0.65, 0.5, -0.05, 0.92,\n",
        "                          -0.95, 0.35, -0.45, 0.6, 0.02, 0.85, -0.75, 0.25]).reshape(1, -1, 1).to(DEVICE)\n",
        "def demo_function(x):\n",
        "    return 0.4 + 0.7 * x - 0.9 * x**2\n",
        "noise_gen = torch.Generator().manual_seed(404)\n",
        "context_y = demo_function(context_x) + 0.05 * torch.randn(context_x.shape, generator=noise_gen).to(DEVICE)\n",
        "fig, axes = plt.subplots(1, 4, figsize=(13, 3.3), sharey=True)\n",
        "with torch.no_grad():\n",
        "    for ax, n in zip(axes, [2, 4, 8, 16]):\n",
        "        pred = model(context_x[:, :n], context_y[:, :n], grid)[0].cpu().numpy()\n",
        "        ax.plot(grid[0, :, 0].cpu(), demo_function(grid)[0, :, 0].cpu(), color='#64748b', ls='--', label='True rule')\n",
        "        ax.plot(grid[0, :, 0].cpu(), pred, color='#147d92', lw=2, label='Fixed model')\n",
        "        ax.scatter(context_x[0, :n, 0].cpu(), context_y[0, :n, 0].cpu(), c='#e87830', s=22, label='Observed rows')\n",
        "        ax.axvspan(-1.4, -1, color='#e2e8f0', alpha=0.7)\n",
        "        ax.axvspan(1, 1.4, color='#e2e8f0', alpha=0.7)\n",
        "        ax.set_title(f'{n} observed rows'); ax.set_xlabel('Normalized condition x'); ax.grid(alpha=0.15)\n",
        "axes[0].set_ylabel('Response y'); axes[-1].legend(fontsize=7)\n",
        "fig.suptitle('In-context inference: same weights, more observations', fontsize=12)\n",
        "fig.tight_layout(); fig.savefig(OUT/'context_growth.png', dpi=180); plt.show()\n"
      ],
      "id": "cell-18"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 8 학습 곡선 읽기\n",
        "고정 검증 문제에서 오차가 줄어드는지 봅니다. 초기 모델과 사전학습 모델의 평가 차이는 사전학습 효과, 같은 모델에서 관측 수에 따른 차이는 문맥 내 적응 효과입니다. 서로 다른 두 효과를 구분하세요. 종축은 로그 척도입니다."
      ],
      "id": "cell-19"
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {},
      "outputs": [
        {
          "output_type": "display_data",
          "metadata": {},
          "data": {
            "image/png": "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"
          }
        }
      ],
      "source": [
        "h = pd.DataFrame(history)\n",
        "fig, ax = plt.subplots(figsize=(7, 3.6))\n",
        "ax.plot(h.step, h.train_batch_mse, color='#c89242', alpha=0.6, label='Current training batch')\n",
        "ax.plot(h.step, h.validation_mse, color='#147d92', label='Fixed validation tasks')\n",
        "ax.set_yscale('log'); ax.set_xlabel('Optimizer step'); ax.set_ylabel('MSE (log scale)')\n",
        "ax.set_title('Pretraining across synthetic table tasks'); ax.legend(); ax.grid(alpha=0.2)\n",
        "fig.tight_layout(); fig.savefig(OUT/'training_curve.png', dpi=180); plt.show()\n"
      ],
      "id": "cell-20"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 9 결과와 가중치 저장하기\n",
        "학습 기록, 평가 수치, 버전, 시드, 실제 소요 시간, 검사 결과와 작은 모델 가중치를 저장합니다. 결과를 재현할 때는 노트북 전체와 환경 기록을 함께 보관하세요. 동일 시드여도 다른 하드웨어나 라이브러리에서는 작은 수치 차이가 있을 수 있습니다."
      ],
      "id": "cell-21"
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {},
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{\n",
            "  \"torch\": \"2.14.0+cpu\",\n",
            "  \"numpy\": \"2.3.5\",\n",
            "  \"seed\": 7,\n",
            "  \"steps\": 1800,\n",
            "  \"batch_size\": 32,\n",
            "  \"synthetic_training_tasks\": 57600,\n",
            "  \"parameters\": 38257,\n",
            "  \"elapsed_seconds\": 14.750920684999983,\n",
            "  \"device\": \"cpu\",\n",
            "  \"checks\": {\n",
            "    \"weights_unchanged\": true,\n",
            "    \"row_permutation_max_abs_difference\": 9.5367431640625e-07,\n",
            "    \"query_batch_max_abs_difference\": 9.5367431640625e-07\n",
            "  }\n",
            "}\n"
          ]
        }
      ],
      "source": [
        "h.to_csv(OUT/'training_history.csv', index=False)\n",
        "scores.to_csv(OUT/'evaluation.csv', index=False)\n",
        "torch.save({'state_dict': model.state_dict(), 'width': 48, 'n_heads': 4, 'n_layers': 2}, OUT/'tiny_table_transformer.pt')\n",
        "metadata = {'torch':torch.__version__, 'numpy':np.__version__, 'seed':SEED, 'steps':N_STEPS,\n",
        "            'batch_size':BATCH_SIZE, 'synthetic_training_tasks':N_STEPS*BATCH_SIZE,\n",
        "            'parameters':sum(p.numel() for p in model.parameters()), 'elapsed_seconds':elapsed,\n",
        "            'device':DEVICE, 'checks':checks}\n",
        "(OUT/'run_metadata.json').write_text(json.dumps(metadata, indent=2), encoding='utf-8')\n",
        "print(json.dumps(metadata, indent=2))\n"
      ],
      "id": "cell-22"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 10 저장한 모델만 불러와 새 표 예측하기\n",
        "학습을 다시 하지 않고 실행할 때는 환경 설정 셀과 모델 정의 셀을 먼저 실행한 뒤 아래 셀을 실행합니다. ZIP의 `results/tiny_table_transformer.pt`가 있어야 합니다. 자신이 만든 가중치나 제공된 가중치를 사용하세요."
      ],
      "id": "cell-23"
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {},
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "x=0.3에서의 예측: 0.7842\n"
          ]
        }
      ],
      "source": [
        "checkpoint = torch.load(OUT / 'tiny_table_transformer.pt', map_location=DEVICE, weights_only=True)\n",
        "loaded = TinyTableTransformer(checkpoint['width'], checkpoint['n_heads'], checkpoint['n_layers']).to(DEVICE)\n",
        "loaded.load_state_dict(checkpoint['state_dict'])\n",
        "loaded.eval()\n",
        "new_x = torch.tensor([[[-0.8], [0.0], [0.8]]], device=DEVICE)\n",
        "new_y = torch.tensor([[[-0.6], [0.4], [1.3]]], device=DEVICE)\n",
        "question = torch.tensor([[[0.3]]], device=DEVICE)\n",
        "with torch.no_grad():\n",
        "    answer = loaded(new_x, new_y, question).item()\n",
        "print(f'x=0.3에서의 예측: {answer:.4f}')\n"
      ],
      "id": "cell-24"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 11 바꿔 볼 실험\n",
        "\n",
        "- 생성 함수를 2차식에서 sin 함수로 바꾸고, 평가도 같은 분포로 맞춰 다시 학습해 보세요.\n",
        "- 2차식으로 학습한 모델을 sin 함수에 평가해, 분포가 바뀌면 무엇이 달라지는지 확인하세요.\n",
        "- 새로운 관측값만 바꾸고 optimizer를 호출하지 않은 상태에서 예측이 바뀌는지 확인하세요.\n",
        "- 입력 특성을 2개로 늘리려면 데이터 생성, 입력 임베딩 차원, 질문 토큰 생성, 비교 모델을 함께 수정해야 합니다. 숫자 하나만 바꾸면 되는 구조는 아닙니다.\n",
        "\n",
        "이 모델은 실제 소재 예측이나 다음 실험 추천용이 아닙니다. 정식 TFM은 더 다양한 생성 분포, 여러 특성, 결측값과 범주형 처리, 회귀 분포 출력 등 훨씬 많은 설계가 필요합니다.\n",
        "\n",
        "[TabPFN 원 논문](https://arxiv.org/abs/2207.01848) · [PyTorch TransformerEncoderLayer](https://docs.pytorch.org/docs/stable/generated/torch.nn.TransformerEncoderLayer.html)"
      ],
      "id": "cell-25"
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 실행 결과와 해석\n",
        "\n",
        "실행 환경은 CPU, PyTorch 2.14.0+cpu, NumPy 2.3.5입니다. 모델은 38,257개 파라미터이며 1,800번 업데이트했습니다. 측정된 학습 루프 시간은 14.8초입니다. 설치, 임포트, 그래프 작성 시간은 제외하며 다른 컴퓨터에서 같은 시간을 보장하지 않습니다.\n",
        "\n",
        "아래는 새로운 표 128개, 각 표의 질문 32개에 대한 MSE입니다. 입력과 정답은 무차원 합성 값입니다.\n",
        "\n",
        "| 관측 행 수 | 초기 모델 | 사전학습 모델 | 관측 평균 | 2차 ridge | 잘못된 문맥 |\n",
        "|---:|---:|---:|---:|---:|---:|\n",
        "| 2 | 1.1490 | 0.1603 | 0.5419 | 0.1486 | 3.3474 |\n",
        "| 4 | 1.1855 | 0.0477 | 0.4836 | 0.0336 | 2.7599 |\n",
        "| 8 | 1.1908 | 0.0157 | 0.4416 | 0.0049 | 2.5072 |\n",
        "| 16 | 1.1842 | 0.0114 | 0.4185 | 0.0031 | 2.2348 |\n",
        "\n",
        "\n",
        "학습된 모델은 초기 모델과 관측 평균보다 낮은 오차를 보였습니다. 관측이 늘어날 때 평균 오차가 감소했고, 잘못된 문맥을 제공하면 오차가 커졌습니다. 따라서 이 예제에서는 사전학습 효과와 문맥 의존성을 관찰할 수 있습니다.\n",
        "\n",
        "2차 ridge 회귀가 더 잘 맞습니다. 생성 규칙의 기저를 정확히 아는 비교 모델이므로 자연스러운 결과입니다. Transformer가 더 복잡하다는 이유만으로 이 문제에서 우월할 필요는 없습니다.\n",
        "\n",
        "가중치 동일성 검사는 통과했습니다. 관측 행 순서를 뒤집거나 질문을 개별 처리했을 때 최대 절대 차이는 각각 9.5e-07, 9.5e-07로 부동소수점 오차 수준입니다. 관측 수가 달라져도 항상 정답으로 수렴하거나 외삽에 성공한다는 보장은 없습니다.\n",
        "\n",
        "검증 MSE는 첫 업데이트 후 0.8284에서 최종 0.0295로 변했습니다. 초기 무작위 모델은 별도 평가 표에서 비교했으므로 이 검증 곡선의 첫 점과 혼동하지 마세요. 학습 손실 기록은 첫 단계만 관측 2개, 이후 100단계 간격에서는 관측 16개인 배치입니다.\n"
      ],
      "id": "cell-26"
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "name": "python",
      "version": "3.12"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}