JAX_FLAX_NNX_save_single_nnxjit.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/jecampagne/JaxTutos/blob/main/JAX_FLAX_NNX_save_single_nnxjit.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "id": "04e4f884-1a4c-4fc8-a685-b6e431431382",
      "metadata": {
        "id": "04e4f884-1a4c-4fc8-a685-b6e431431382"
      },
      "outputs": [],
      "source": [
        "import matplotlib as mpl\n",
        "import matplotlib.pyplot as plt\n",
        "mpl.rc('image', cmap='jet')\n",
        "mpl.rcParams['font.size'] = 16"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "id": "75b28140-7ad1-4dbd-b3dc-1c2b4569d975",
      "metadata": {
        "id": "75b28140-7ad1-4dbd-b3dc-1c2b4569d975"
      },
      "outputs": [],
      "source": [
        "import jax\n",
        "import flax\n",
        "import orbax.checkpoint as ocp"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Test of the nnx.jit\n",
        "it is very slow compared to the jax.jit\n",
        "[see flax #4224](https://github.com/google/flax/discussions/4224)"
      ],
      "metadata": {
        "id": "1GqA1QomYdv2"
      },
      "id": "1GqA1QomYdv2"
    },
    {
      "cell_type": "code",
      "source": [
        "print(jax.__version__, flax.__version__, ocp.__version__)\n",
        "#0.4.33 0.10.3 0.6.4 on Colab 25 Fev 0k\n",
        "#0.5.0 0.10.3 0.11.5 on Jean Zay 25 Fev Ok\n",
        "#change the restore section if needed\n",
        "\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bJjP-vIL_xNx",
        "outputId": "fd45518e-8620-4267-d50a-d50c537ebfb1"
      },
      "id": "bJjP-vIL_xNx",
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.4.33 0.10.3 0.6.4\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "id": "2267ac67-69ad-444d-9e3b-befa2d5ca2f3",
      "metadata": {
        "id": "2267ac67-69ad-444d-9e3b-befa2d5ca2f3"
      },
      "outputs": [],
      "source": [
        "import jax.numpy as jnp"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "id": "bb8a79fb-f84f-4e19-b98d-4652df0c5880",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bb8a79fb-f84f-4e19-b98d-4652df0c5880",
        "outputId": "73c439c3-300f-45ae-df00-1035d9890765"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "1"
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ],
      "source": [
        "from functools import partial\n",
        "\n",
        "import numpy as np\n",
        "num_devices = jax.device_count()\n",
        "num_devices # 1"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "id": "ddb60fc7-24e8-411a-bfb6-9794a2ec87c4",
      "metadata": {
        "id": "ddb60fc7-24e8-411a-bfb6-9794a2ec87c4"
      },
      "outputs": [],
      "source": [
        "args = {\n",
        "    'x_dim': 2,\n",
        "    'seed': 2025,\n",
        "}\n",
        "args['num_blobs'] = 4\n",
        "args['x_lim'] = 1\n",
        "\n",
        "# Set the parallel batch size.\n",
        "args['batch_size'] = num_devices * 512 if num_devices > 1 else  512\n",
        "args['epochs'] = 1500  # min: 500. More <=> less samples between the blobs...\n",
        "args['batch_size_p'] = args['batch_size'] //  num_devices\n",
        "\n",
        "args['sigma_max'] = 15.0"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "id": "923f6f1e-a1a4-454d-9fbd-c40ec1a52065",
      "metadata": {
        "id": "923f6f1e-a1a4-454d-9fbd-c40ec1a52065"
      },
      "outputs": [],
      "source": [
        "import optax\n",
        "\n",
        "from flax import nnx\n",
        "from flax.training import train_state\n",
        "\n",
        "from typing import Any, Callable, Sequence, Optional, Tuple\n",
        "PRNGKey = jnp.ndarray"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "0fa298f6-78e1-4cc2-ac44-d6467318ee69",
      "metadata": {
        "id": "0fa298f6-78e1-4cc2-ac44-d6467318ee69"
      },
      "source": [
        "# Dataset"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "id": "da549dfb-928c-47c0-aca4-758e91a4c9e2",
      "metadata": {
        "id": "da549dfb-928c-47c0-aca4-758e91a4c9e2"
      },
      "outputs": [],
      "source": [
        "def make_dataset_gaussian_blobs(\n",
        "    seed: int,\n",
        "    batch_size: int = 1,\n",
        "    num_blobs: int = 8,\n",
        "    r: float = 1.0,\n",
        "    SCALE: float = 0.1\n",
        "):\n",
        "    key = jax.random.PRNGKey(seed)\n",
        "    offset = 0  # jnp.pi / num_blobs\n",
        "    thetas = jnp.linspace(offset, 2 * jnp.pi + offset, num_blobs, endpoint=False)\n",
        "    x = r * jnp.cos(thetas)\n",
        "    y = r * jnp.sin(thetas)\n",
        "    loc = jnp.stack([x, y], axis=-1)\n",
        "\n",
        "    kay, key1 = jax.random.split(key,2)\n",
        "    iblob = jax.random.randint(key1,minval=0, maxval=num_blobs, shape=(batch_size,))\n",
        "    keys = jax.random.split(key,batch_size)\n",
        "    dataGen = jax.vmap(lambda x: jax.random.multivariate_normal(keys[x],\n",
        "            mean=loc[iblob[x]],\n",
        "            cov=SCALE**2*jnp.array([[1,0],[0,1]]), shape=(1,)))\n",
        "\n",
        "    data = dataGen(jnp.arange(batch_size))\n",
        "    data = data.squeeze()\n",
        "\n",
        "    return data/(2*jnp.max(jnp.abs(data))) + jnp.ones((batch_size,2))*0.5\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "id": "56fca03e-b501-45e3-845f-0320f5be0d44",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "56fca03e-b501-45e3-845f-0320f5be0d44",
        "outputId": "6a2b35c1-1688-4bd9-a8a9-37061bbf49e7"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "99840 195\n"
          ]
        }
      ],
      "source": [
        "dataset_size = 100_000\n",
        "batches_in_epoch = dataset_size // args['batch_size']\n",
        "dataset_size = batches_in_epoch * args['batch_size']\n",
        "\n",
        "print(dataset_size, batches_in_epoch)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "id": "5b4a2097-7a95-46f8-b13e-cb69e813286a",
      "metadata": {
        "id": "5b4a2097-7a95-46f8-b13e-cb69e813286a"
      },
      "outputs": [],
      "source": [
        "X_train= make_dataset_gaussian_blobs(seed=args['seed'],num_blobs=args['num_blobs'],\n",
        "                                     batch_size=dataset_size)\n",
        "xlim, ylim = [0,args['x_lim'] ], [0,args['x_lim'] ]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "id": "9b559f56-1aca-4359-81b3-136054e81968",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 490
        },
        "id": "9b559f56-1aca-4359-81b3-136054e81968",
        "outputId": "277dfb35-1822-49b7-8524-1dc541c192ba"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "plt.scatter(X_train[:5000, 0], X_train[:5000, 1], s=0.1, color='red',label='Train')\n",
        "plt.title(\"Training set\")\n",
        "plt.xlim(xlim)\n",
        "plt.ylim(ylim)\n",
        "plt.xlabel(\"$x_1$\")\n",
        "plt.ylabel(\"$x_2$\")\n",
        "plt.gca().set_aspect('equal')\n",
        "plt.grid()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "id": "df6d3c5a-079a-4d1c-aaf2-f7f0890ebb30",
      "metadata": {
        "id": "df6d3c5a-079a-4d1c-aaf2-f7f0890ebb30"
      },
      "outputs": [],
      "source": [
        "import itertools\n",
        "data_gen = itertools.cycle(jnp.array(jnp.array_split(jax.random.permutation(jax.random.PRNGKey(42),X_train), batches_in_epoch )))"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "788d0abd-df79-408f-b9f2-9b5c69c13a9d",
      "metadata": {
        "id": "788d0abd-df79-408f-b9f2-9b5c69c13a9d"
      },
      "source": [
        "# ScoreNet simple architecture"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "id": "48eb5091-fb3e-4790-8949-56021c7035e1",
      "metadata": {
        "id": "48eb5091-fb3e-4790-8949-56021c7035e1"
      },
      "outputs": [],
      "source": [
        "def marginal_prob_std(t, sigma):\n",
        "  \"\"\"Compute the mean and standard deviation of $p_{0t}(x(t) | x(0))$.\n",
        "\n",
        "  Args:\n",
        "    t: A vector of time steps.\n",
        "    sigma: The $\\sigma$ in our SDE.\n",
        "\n",
        "  Returns:\n",
        "    The standard deviation.\n",
        "  \"\"\"\n",
        "  return jnp.sqrt((sigma**(2 * t) - 1.) / 2. / jnp.log(sigma))\n",
        "\n",
        "marginal_prob_std_fn = partial(marginal_prob_std, sigma=args['sigma_max'])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "id": "742bbbcc-44af-4658-a4bf-ecb9c365f782",
      "metadata": {
        "id": "742bbbcc-44af-4658-a4bf-ecb9c365f782"
      },
      "outputs": [],
      "source": [
        "class GaussianFourierProjection(nnx.Module):\n",
        "  \"\"\"Gaussian random features for encoding time steps.\"\"\"\n",
        "  #embed_dim: int\n",
        "  #scale: float = 30.\n",
        "\n",
        "  def __init__(self, embed_dim: int, scale: float, *, rngs: nnx.Rngs):\n",
        "    key = rngs.params()\n",
        "    dout = embed_dim // 2\n",
        "    self.W = nnx.Variable(jax.random.normal(key, (dout,))  *  scale)\n",
        "\n",
        "  def __call__(self, x):\n",
        "    x_proj = x[:, None] * self.W[None, :] * 2 * jnp.pi\n",
        "    return jnp.concatenate([jnp.sin(x_proj), jnp.cos(x_proj)], axis=-1)\n",
        "\n",
        "class ScoreNet(nnx.Module):\n",
        "  channels: Tuple[int] =  (32, 64, 128, 256)\n",
        "  embed_dim: int = 256\n",
        "  scale: float = 30.\n",
        "\n",
        "  def __init__(self, marginal_prob_std:Any, din_t:int, rngs: nnx.Rngs):\n",
        "    self.act = nnx.swish\n",
        "    self.marginal_prob_std = marginal_prob_std\n",
        "    #time embedding\n",
        "    self.embed = GaussianFourierProjection(embed_dim=self.embed_dim,\n",
        "                                           scale=self.scale,\n",
        "                                           rngs=rngs)\n",
        "    self.LayerEmb = nnx.Linear(self.embed_dim, self.embed_dim, rngs=rngs)\n",
        "    #encoding part\n",
        "    self.Layer1  = nnx.Linear(args['x_dim'],self.channels[0],use_bias=False, rngs=rngs)\n",
        "    self.Layer1e = nnx.Linear(self.embed_dim,self.channels[0], rngs=rngs)\n",
        "    self.Norm1   = nnx.GroupNorm(self.channels[0],num_groups=4, rngs=rngs)\n",
        "\n",
        "    self.Layer2  = nnx.Linear(self.channels[0],self.channels[1],use_bias=False, rngs=rngs)\n",
        "    self.Layer2e = nnx.Linear(self.embed_dim,self.channels[1], rngs=rngs)\n",
        "    self.Norm2   = nnx.GroupNorm(self.channels[1], rngs=rngs) # num_groups=32 by default\n",
        "\n",
        "    self.Layer3  = nnx.Linear(self.channels[1],self.channels[2],use_bias=False, rngs=rngs)\n",
        "    self.Layer3e = nnx.Linear(self.embed_dim,self.channels[2], rngs=rngs)\n",
        "    self.Norm3   = nnx.GroupNorm(self.channels[2], rngs=rngs) # num_groups=32 by default\n",
        "\n",
        "    self.Layer4  = nnx.Linear(self.channels[2],self.channels[3],use_bias=False, rngs=rngs)\n",
        "    self.Layer4e = nnx.Linear(self.embed_dim,self.channels[3], rngs=rngs)\n",
        "    self.Norm4   = nnx.GroupNorm(self.channels[3], rngs=rngs) # num_groups=32 by default\n",
        "\n",
        "    #decoding part\n",
        "    self.Layer5  = nnx.Linear(self.channels[3],self.channels[2],use_bias=False, rngs=rngs)\n",
        "    self.Layer5e = nnx.Linear(self.embed_dim,self.channels[2], rngs=rngs)\n",
        "    self.Norm5   = nnx.GroupNorm(self.channels[2], rngs=rngs) # num_groups=32 by default\n",
        "\n",
        "    self.Layer6  = nnx.Linear(2*self.channels[2],self.channels[1],use_bias=False, rngs=rngs)\n",
        "    self.Layer6e = nnx.Linear(self.embed_dim,self.channels[1], rngs=rngs)\n",
        "    self.Norm6   = nnx.GroupNorm(self.channels[1], rngs=rngs) # num_groups=32 by default\n",
        "\n",
        "    self.Layer7  = nnx.Linear(2*self.channels[1],self.channels[0],use_bias=False, rngs=rngs)\n",
        "    self.Layer7e = nnx.Linear(self.embed_dim,self.channels[0], rngs=rngs)\n",
        "    self.Norm7   = nnx.GroupNorm(self.channels[0], rngs=rngs) # num_groups=32 by default\n",
        "\n",
        "    self.Layer7  = nnx.Linear(2*self.channels[1],self.channels[0],use_bias=False, rngs=rngs)\n",
        "    self.Layer7e = nnx.Linear(self.embed_dim,self.channels[0], rngs=rngs)\n",
        "    self.Norm7   = nnx.GroupNorm(self.channels[0], rngs=rngs) # num_groups=32 by default\n",
        "\n",
        "    self.Layer8  = nnx.Linear(2*self.channels[0],args['x_dim'], rngs=rngs)\n",
        "\n",
        "\n",
        "  def __call__ (self,x,t):\n",
        "    # time embding\n",
        "    embed = self.act(self.LayerEmb(self.embed(t)))\n",
        "\n",
        "    # encoding\n",
        "    h1 = self.Layer1(x)\n",
        "    h1 += self.Layer1e(embed)\n",
        "    h1 = self.Norm1(h1)\n",
        "    h1 = self.act(h1)\n",
        "\n",
        "    h2 = self.Layer2(h1)\n",
        "    h2 += self.Layer2e(embed)\n",
        "    h2 = self.Norm2(h2)\n",
        "    h2 = self.act(h2)\n",
        "\n",
        "    h3 = self.Layer3(h2)\n",
        "    h3 += self.Layer3e(embed)\n",
        "    h3 = self.Norm3(h3)\n",
        "    h3 = self.act(h3)\n",
        "\n",
        "    h4 = self.Layer4(h3)\n",
        "    h4 += self.Layer4e(embed)\n",
        "    h4 = self.Norm4(h4)\n",
        "    h4 = self.act(h4)\n",
        "\n",
        "    # decondig\n",
        "    h = self.Layer5(h4)\n",
        "    h += self.Layer5e(embed)\n",
        "    h = self.Norm5(h)\n",
        "    h = self.act(h)\n",
        "\n",
        "    h = self.Layer6(jnp.concatenate([h, h3], axis=-1))\n",
        "    h += self.Layer6e(embed)\n",
        "    h = self.Norm6(h)\n",
        "    h = self.act(h)\n",
        "\n",
        "    h = self.Layer7(jnp.concatenate([h, h2], axis=-1))\n",
        "    h += self.Layer7e(embed)\n",
        "    h = self.Norm7(h)\n",
        "    h = self.act(h)\n",
        "\n",
        "    h = self.Layer8(jnp.concatenate([h, h1], axis=-1))\n",
        "\n",
        "    # normalisation\n",
        "    h = h / self.marginal_prob_std(t)[:, None]\n",
        "    return h"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "f5fa7018-5dd2-41a7-b83b-6da9e90ac9cf",
      "metadata": {
        "id": "f5fa7018-5dd2-41a7-b83b-6da9e90ac9cf"
      },
      "source": [
        "# Training\n",
        "restart from here if you want to change training parameters\n",
        "\n",
        "Here we use nnx.jit [see flax #4224](https://github.com/google/flax/discussions/4224)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "id": "2a1cf0ba-286d-461c-af04-34a9253dad01",
      "metadata": {
        "id": "2a1cf0ba-286d-461c-af04-34a9253dad01"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import shutil\n",
        "path = os.getcwd()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "id": "69678666-f3e1-4d16-a487-55c66b0daf01",
      "metadata": {
        "id": "69678666-f3e1-4d16-a487-55c66b0daf01"
      },
      "outputs": [],
      "source": [
        "import orbax.checkpoint as ocp\n",
        "\n",
        "# Create the checkpoint manager\n",
        "ckpt_dir = f\"{path}/tmp\"\n",
        "ckpt_dir = ocp.test_utils.erase_and_create_empty(ckpt_dir)\n",
        "\n",
        "options = ocp.CheckpointManagerOptions(\n",
        "    save_interval_steps=1,\n",
        "    max_to_keep=1,\n",
        "    # other options\n",
        ")\n",
        "mngr = ocp.CheckpointManager(\n",
        "    ckpt_dir,\n",
        "    options=options,\n",
        ")\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "id": "1f819af2-28cf-47ec-8996-98dae9f907a5",
      "metadata": {
        "id": "1f819af2-28cf-47ec-8996-98dae9f907a5"
      },
      "outputs": [],
      "source": [
        "@nnx.jit\n",
        "def train_step(model: ScoreNet,\n",
        "                optimizer: nnx.Optimizer,\n",
        "                x: jnp.ndarray,\n",
        "                key: PRNGKey):\n",
        "\n",
        "\n",
        "  key, key_std, key_noise = jax.random.split(key,3)\n",
        "  random_t = jax.random.uniform(key_std, (x.shape[0],), minval=1e-5, maxval=1.)\n",
        "  std = marginal_prob_std_fn(random_t)\n",
        "  z = jax.random.normal(key_noise, x.shape)\n",
        "  perturbed_x = x + z * std[:,None]\n",
        "\n",
        "  def loss_fn(model: ScoreNet):\n",
        "    score = model(perturbed_x, random_t)\n",
        "    loss = jnp.mean(jnp.sum((score * std[:, None] + z)**2, axis=1))\n",
        "    return loss\n",
        "\n",
        "  loss, grads = nnx.value_and_grad(loss_fn)(model)\n",
        "  optimizer.update(grads)\n",
        "\n",
        "  return loss"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "id": "f789b21f-4ed1-465f-954a-240dc51dbf22",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f789b21f-4ed1-465f-954a-240dc51dbf22",
        "outputId": "4cd09456-fd9c-493c-95f7-a0163f105e03"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "292695\n"
          ]
        }
      ],
      "source": [
        "key = jax.random.PRNGKey(seed=2025)\n",
        "key, key_scorenet = jax.random.split(key, 2)\n",
        "\n",
        "total_steps = (args['epochs']+1) * batches_in_epoch\n",
        "print(total_steps)\n",
        "schedule_fn = optax.piecewise_constant_schedule(init_value=1e-3,\n",
        "        boundaries_and_scales={int(total_steps*0.33):0.1,\n",
        "                               int(total_steps*0.66):0.1})\n",
        "\n",
        "model = ScoreNet(marginal_prob_std_fn,din_t=1,rngs=nnx.Rngs(args['seed']))\n",
        "optimizer = nnx.Optimizer(model, optax.adam(learning_rate=schedule_fn))# implicitly does not upate W variables"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "id": "e4105939-1def-46c0-b249-535c898ed562",
      "metadata": {
        "id": "e4105939-1def-46c0-b249-535c898ed562"
      },
      "outputs": [],
      "source": [
        "from IPython.display import clear_output\n",
        "import orbax"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "id": "d4238405-16b8-40bb-be72-c7b01471fda2",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 301
        },
        "id": "d4238405-16b8-40bb-be72-c7b01471fda2",
        "outputId": "acad31bd-e780-4cb2-b8ca-197bda9dd088"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 440/1500: loss = 0.17912\n"
          ]
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-22-aa1c1cde5973>\u001b[0m in \u001b[0;36m<cell line: 0>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     14\u001b[0m       \u001b[0mbatch_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata_gen\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m       \u001b[0;31m# Take a step with the denoiser.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 16\u001b[0;31m       loss =  train_step(model, optimizer,\n\u001b[0m\u001b[1;32m     17\u001b[0m                         batch_data, key_scorenet)\n\u001b[1;32m     18\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.11/dist-packages/flax/nnx/graph.py\u001b[0m in \u001b[0;36mupdate_context_manager_wrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m   1080\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mupdate_context_manager_wrapper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1081\u001b[0m       \u001b[0;32mwith\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1082\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1083\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1084\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mupdate_context_manager_wrapper\u001b[0m  \u001b[0;31m# type: ignore\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.11/dist-packages/flax/nnx/transforms/compilation.py\u001b[0m in \u001b[0;36mjit_wrapper\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    343\u001b[0m       \u001b[0mctxtag\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'jit'\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    344\u001b[0m     )\n\u001b[0;32m--> 345\u001b[0;31m     pure_args_out, pure_kwargs_out, pure_out = jitted_fn(\n\u001b[0m\u001b[1;32m    346\u001b[0m       \u001b[0;34m*\u001b[0m\u001b[0mpure_args\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mpure_kwargs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    347\u001b[0m     )\n",
            "\u001b[0;32m/usr/local/lib/python3.11/dist-packages/flax/nnx/graph.py\u001b[0m in \u001b[0;36m__eq__\u001b[0;34m(self, other)\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.11/dist-packages/flax/nnx/graph.py\u001b[0m in \u001b[0;36m__eq__\u001b[0;34m(self, other)\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.11/dist-packages/flax/nnx/graph.py\u001b[0m in \u001b[0;36m__eq__\u001b[0;34m(self, other)\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.11/dist-packages/flax/nnx/graph.py\u001b[0m in \u001b[0;36m__eq__\u001b[0;34m(self, other)\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ],
      "source": [
        "info_losses=[]\n",
        "loss_best = jnp.inf\n",
        "\n",
        "eval_every = 10\n",
        "\n",
        "for epoch in range(1, args['epochs'] + 1):\n",
        "    loss_mean = 0.\n",
        "    for batch in range(batches_in_epoch):\n",
        "      # Generate RNG keys for denoiser\n",
        "      key, key_scorenet = jax.random.split(key, 2)\n",
        "      #key_scorenet = shard_prng_key(key_scorenet)\n",
        "\n",
        "      # Shard the data to possible devices.\n",
        "      batch_data = next(data_gen)\n",
        "      # Take a step with the denoiser.\n",
        "      loss =  train_step(model, optimizer,\n",
        "                        batch_data, key_scorenet)\n",
        "\n",
        "      loss_mean += loss\n",
        "\n",
        "    loss_mean = loss_mean/batches_in_epoch\n",
        "    info_losses.append(loss_mean)\n",
        "    #save best\n",
        "    if loss_mean < loss_best:\n",
        "      loss_best = loss_mean\n",
        "      epoch_best = epoch\n",
        "      _, best_state = nnx.split(model)\n",
        "      mngr.save(epoch_best, args=ocp.args.StandardSave(best_state))\n",
        "\n",
        "    if epoch > 0 and (epoch % eval_every == 0 or epoch == args['epochs']):\n",
        "        clear_output(wait=True)\n",
        "        print(f\"Epoch {epoch}/{args['epochs']}: loss = {loss_mean:.5f}\")\n",
        "#        clear_output(wait=True)\n",
        "#        # Plot loss and accuracy in subplots\n",
        "#        plt.plot(info_losses,label='Loss')\n",
        "#        plt.scatter(epoch_best,loss_best,s=10,c='r',label='best')\n",
        "#        plt.yscale('log')\n",
        "#        plt.legend();\n",
        "#        plt.show()\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "epoch_best, loss_best"
      ],
      "metadata": {
        "id": "0qjXuykvWQhq"
      },
      "id": "0qjXuykvWQhq",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "082c4719-967d-48a3-8b77-4ae374715bad",
      "metadata": {
        "id": "082c4719-967d-48a3-8b77-4ae374715bad"
      },
      "outputs": [],
      "source": [
        "mngr.wait_until_finished()"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "e54a0769-2c30-4fe2-aef4-4dea8e3d8e84",
      "metadata": {
        "id": "e54a0769-2c30-4fe2-aef4-4dea8e3d8e84"
      },
      "source": [
        "# Plot"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "a085aaa6-6d19-4bb5-be48-acc7a184c42a",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 436
        },
        "id": "a085aaa6-6d19-4bb5-be48-acc7a184c42a",
        "outputId": "ddd5ae83-8d6a-4de3-a36f-77901a901cc2"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "import numpy as np\n",
        "\n",
        "info_losses = np.array(info_losses)\n",
        "plt.plot(info_losses,label='Loss')\n",
        "plt.scatter(epoch_best,loss_best,s=10,c='r',label='best')\n",
        "plt.yscale('log')\n",
        "#plt.xlim([0,10])\n",
        "#plt.ylim([info_den_losses.min()/2,1])\n",
        "plt.legend();"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "29228c5d-4822-4e64-9859-30dbfaec4faa",
      "metadata": {
        "id": "29228c5d-4822-4e64-9859-30dbfaec4faa"
      },
      "source": [
        "# Sampling"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "709bf37f-39fc-43a9-afbc-49200b7d7572",
      "metadata": {
        "id": "709bf37f-39fc-43a9-afbc-49200b7d7572"
      },
      "outputs": [],
      "source": [
        "def diffusion_coeff(t, sigma):\n",
        "  \"\"\"Compute the diffusion coefficient of our SDE.\n",
        "\n",
        "  Args:\n",
        "    t: A vector of time steps.\n",
        "    sigma: The $\\sigma$ in our SDE.\n",
        "\n",
        "  Returns:\n",
        "    The vector of diffusion coefficients.\n",
        "  \"\"\"\n",
        "  return sigma**t\n",
        "diffusion_coeff_fn = partial(diffusion_coeff, sigma=args['sigma_max'])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "e38ba0e3-3daa-49ac-9b89-e507ab10147f",
      "metadata": {
        "id": "e38ba0e3-3daa-49ac-9b89-e507ab10147f"
      },
      "outputs": [],
      "source": [
        "from scipy import integrate\n",
        "\n",
        "## The error tolerance for the black-box ODE solver\n",
        "error_tolerance = 1e-5\n",
        "def ode_sampler(rng,\n",
        "                model,   # model\n",
        "                marginal_prob_std,\n",
        "                diffusion_coeff,\n",
        "                batch_size=64,\n",
        "                atol=error_tolerance,\n",
        "                rtol=error_tolerance,\n",
        "                z=None,\n",
        "                eps=1e-5):\n",
        "  \"\"\"Generate samples from score-based models with black-box ODE solvers.\n",
        "\n",
        "  Args:\n",
        "    rng: A JAX random state.\n",
        "    score_model: A `flax.linen.Module` object  that represents architecture\n",
        "      of the score-based model.\n",
        "    params: A dictionary that contains model parameters.\n",
        "    marginal_prob_std: A function that returns the standard deviation\n",
        "      of the perturbation kernel.\n",
        "    diffusion_coeff: A function that returns the diffusion coefficient of the SDE.\n",
        "    batch_size: The number of samplers to generate by calling this function once.\n",
        "    atol: Tolerance of absolute errors.\n",
        "    rtol: Tolerance of relative errors.\n",
        "    z: The latent code that governs the final sample. If None, we start from p_1;\n",
        "      otherwise, we start from the given z.\n",
        "    eps: The smallest time step for numerical stability.\n",
        "  \"\"\"\n",
        "\n",
        "  time_shape = (batch_size,)\n",
        "  sample_shape = (*time_shape, args['x_dim'])\n",
        "\n",
        "  # Create the latent code\n",
        "  if z is None:\n",
        "    rng, step_rng = jax.random.split(rng)\n",
        "    z = jax.random.normal(step_rng, sample_shape)\n",
        "    init_x = z * marginal_prob_std(1.)\n",
        "  else:\n",
        "    init_x = z\n",
        "\n",
        "  shape = init_x.shape\n",
        "\n",
        "  def score_eval_wrapper(sample, time_steps):\n",
        "    \"\"\"A wrapper of the score-based model for use by the ODE solver.\"\"\"\n",
        "    sample = jnp.asarray(sample, dtype=jnp.float32).reshape(sample_shape)\n",
        "    time_steps = jnp.asarray(time_steps).reshape(time_shape)\n",
        "\n",
        "    #score, (graphdef, new_state) = state.apply_fn(\n",
        "    #  state.params,\n",
        "    #  state.other_variables\n",
        "    #)( sample, time_steps)\n",
        "    score = model(sample, time_steps)\n",
        "\n",
        "    return np.asarray(score).reshape((-1,)).astype(np.float64)\n",
        "\n",
        "  def ode_func(t, x):\n",
        "    \"\"\"The ODE function for use by the ODE solver.\"\"\"\n",
        "    time_steps = np.ones(time_shape) * t\n",
        "    g = diffusion_coeff(t)\n",
        "    return  -0.5 * (g**2) * score_eval_wrapper(x, time_steps)\n",
        "\n",
        "  # Run the black-box ODE solver.\n",
        "  res = integrate.solve_ivp(ode_func, (1., eps), np.asarray(init_x).reshape(-1),\n",
        "                            rtol=rtol, atol=atol, method='RK45', max_step=200)\n",
        "  print(f\"Number of function evaluations: {res.nfev}\")\n",
        "  x = jnp.asarray(res.y[:, -1]).reshape(shape)\n",
        "\n",
        "  return x"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "a498b3c2-9af7-4280-a8fb-4b28f02216d0",
      "metadata": {
        "id": "a498b3c2-9af7-4280-a8fb-4b28f02216d0"
      },
      "outputs": [],
      "source": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "678d2d3d-b324-4dce-bd8a-7ff4eaa2dda8",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "678d2d3d-b324-4dce-bd8a-7ff4eaa2dda8",
        "outputId": "062831ef-01bc-4f41-dbd0-d5d38e745605"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:absl:CheckpointManagerOptions.read_only=True, setting save_interval_steps=0.\n",
            "WARNING:absl:Given directory is read only=/content/tmp\n"
          ]
        }
      ],
      "source": [
        "#model = ScoreNet(marginal_prob_std_fn,din_t=1,rngs=nnx.Rngs(args['seed']))\n",
        "#optimizer = nnx.Optimizer(model, optax.adamw(learning_rate=schedule_fn))# implicitly does not upate W variables\n",
        "#nnx.update((model, optimizer), best_state)\n",
        "# Restore the checkpoint back to its `nnx.State` structure - need an abstract reference.\n",
        "\n",
        "options = ocp.CheckpointManagerOptions(\n",
        "    read_only=True,\n",
        "    create=False,\n",
        ")\n",
        "\n",
        "ckpt_dir = f\"{path}/tmp\"\n",
        "\n",
        "mngr = ocp.CheckpointManager(\n",
        "    ckpt_dir,\n",
        "    options=options,\n",
        ")\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "ocp.__version__"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 37
        },
        "id": "Jd8tkzERGzjy",
        "outputId": "34da24b9-a460-41b3-8b60-baed31fdba73"
      },
      "id": "Jd8tkzERGzjy",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'0.6.4'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 32
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "abstract_model = nnx.eval_shape(lambda: ScoreNet(marginal_prob_std_fn,din_t=1,rngs=nnx.Rngs(0)))\n",
        "graphdef, abstract_state = nnx.split(abstract_model)\n",
        "\n",
        "epoch_best= mngr.latest_step()\n",
        "\n",
        "if ocp.__version__ == '0.6.4':\n",
        "  state_restored = mngr.restore(epoch_best,  args=ocp.args.StandardSave(abstract_state)) # ocp 0.6.4\n",
        "elif ocp.__version__ == '0.11.5':\n",
        "  state_restored = mngr.restore(epoch_best, items=abstract_state)\n",
        "else:\n",
        "  state_restored= None\n",
        "  print(\"HELP! do not know how to proceed yet\")\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bddee708JDIq",
        "outputId": "ac64fbed-5c24-40ae-da2f-f2722a437e8f"
      },
      "id": "bddee708JDIq",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.11/dist-packages/orbax/checkpoint/type_handlers.py:1330: UserWarning: Couldn't find sharding info under RestoreArgs. Populating sharding info from sharding file. Please note restoration time will be slightly increased due to reading from file instead of directly from RestoreArgs. Note also that this option is unsafe when restoring on a different topology than the checkpoint was saved with.\n",
            "  warnings.warn(\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "KOQ5yAsPJDSY"
      },
      "id": "KOQ5yAsPJDSY",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "0370cbfa-bd19-4e40-b8d2-8a103936015e",
      "metadata": {
        "id": "0370cbfa-bd19-4e40-b8d2-8a103936015e"
      },
      "outputs": [],
      "source": [
        "model = nnx.merge(graphdef, state_restored)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "77f9bafa-ab7d-44f5-a229-475d5cf96139",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "77f9bafa-ab7d-44f5-a229-475d5cf96139",
        "outputId": "f555e58b-ddfa-4521-d8d1-fe4d6a32258f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Number of function evaluations: 254\n"
          ]
        }
      ],
      "source": [
        "sample_batch_size = 1_024\n",
        "sampler = ode_sampler\n",
        "rng = jax.random.PRNGKey(0)\n",
        "\n",
        "## Generate samples\n",
        "rng, step_rng = jax.random.split(rng)\n",
        "samples = sampler(rng,\n",
        "                  model,\n",
        "                  marginal_prob_std_fn,\n",
        "                  diffusion_coeff_fn,\n",
        "                  sample_batch_size)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "3ae2596d-e728-422f-b8aa-d41a6825069c",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 469
        },
        "id": "3ae2596d-e728-422f-b8aa-d41a6825069c",
        "outputId": "1b1e201d-88a3-4def-9431-6d422cda2273"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "plt.scatter(X_train[:samples.shape[0], 0], X_train[:samples.shape[0], 1], s=0.3, color='red',label='Train')\n",
        "plt.scatter(samples[:,0],samples[:,1],s=0.3,color='blue',label='Gen')\n",
        "plt.xlim(xlim)\n",
        "plt.ylim(ylim)\n",
        "plt.xlabel(\"$x_1$\")\n",
        "plt.ylabel(\"$x_2$\")\n",
        "plt.gca().set_aspect('equal')\n",
        "plt.grid()\n",
        "lgnd = plt.legend(loc=\"lower left\", scatterpoints=1, fontsize=10)\n",
        "lgnd.legend_handles[0]._sizes = [30]\n",
        "lgnd.legend_handles[1]._sizes = [30]\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "d27331a1-2d66-4746-ac01-6f8ad08a5201",
      "metadata": {
        "id": "d27331a1-2d66-4746-ac01-6f8ad08a5201"
      },
      "outputs": [],
      "source": []
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "pytorch-gpu-2.6.0_py3.12.8",
      "language": "python",
      "name": "module-conda-env-pytorch-gpu-2.6.0_py3.12.8"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.12.8"
    },
    "colab": {
      "provenance": [],
      "toc_visible": true,
      "gpuType": "T4",
      "include_colab_link": true
    },
    "accelerator": "GPU"
  },
  "nbformat": 4,
  "nbformat_minor": 5
}