Autoencoder defect detection.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5015b18e",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_size = 784\n",
    "output_size = 784 \n",
    "hidden1 = 256      \n",
    "hidden2 = 128\n",
    "hidden3 = 64\n",
    "bottle = 2\n",
    "weight_bit_width = 8\n",
    "act_bit_width = 8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "147e8c65",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.utils.data as data\n",
    "import torchvision\n",
    "import torch.optim as optim\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision.transforms import transforms\n",
    "import numpy as np\n",
    "import torchvision.utils as vutils\n",
    "import PIL.Image as Image\n",
    "from torch import nn\n",
    "from torch.nn import Module\n",
    "import torch.nn.functional as F\n",
    "from brevitas.nn import QuantLinear, QuantReLU\n",
    "import brevitas.nn as qnn\n",
    "# Model structure\n",
    "class Encoder(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Encoder, self).__init__()\n",
    "        # Encoder\n",
    "        self.encoder = nn.Sequential(\n",
    "            QuantLinear(784, 256,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            nn.Tanh(),\n",
    "            QuantLinear(256, 128,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            nn.Tanh(),\n",
    "            QuantLinear(128, 64,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            nn.Tanh(),\n",
    "            QuantLinear(64, 2,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            nn.Tanh()\n",
    "        )\n",
    "    def forward(self, inputs):\n",
    "        codes = self.encoder(inputs)\n",
    "        return codes\n",
    "class Decoder(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Decoder, self).__init__()\n",
    "        # Decoder\n",
    "        self.decoder = nn.Sequential(\n",
    "            QuantLinear(2, 64,bias=True, weight_bit_width=weight_bit_width),\n",
    "            nn.Tanh(),\n",
    "            QuantLinear(64, 128,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            nn.Tanh(),\n",
    "            QuantLinear(128, 256,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            nn.Tanh(),\n",
    "            QuantLinear(256, 784,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            nn.Sigmoid()\n",
    "        )\n",
    "    def forward(self, inputs):\n",
    "        outputs = self.decoder(inputs)\n",
    "        return outputs\n",
    "class AutoEncoder(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(AutoEncoder, self).__init__()\n",
    "        # Encoder\n",
    "        self.encoder = Encoder()\n",
    "        # Decoder\n",
    "        self.decoder = Decoder()\n",
    "    def forward(self, inputs):\n",
    "        codes = self.encoder(inputs)\n",
    "        decoded = self.decoder(codes)\n",
    "        return codes, decoded"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "98b2696c",
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'weight_bit_width' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-2-f5057f2b8475>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel_ae\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mAutoEncoder\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[0m\u001b[1;32m      2\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodel_ae\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-1-8d57d60af20b>\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m     53\u001b[0m         \u001b[0msuper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mAutoEncoder\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__init__\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     54\u001b[0m         \u001b[0;31m# Encoder\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 55\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mencoder\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mEncoder\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[0m\u001b[1;32m     56\u001b[0m         \u001b[0;31m# Decoder\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     57\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdecoder\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mDecoder\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[0;32m<ipython-input-1-8d57d60af20b>\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m     20\u001b[0m         \u001b[0;31m# Encoder\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m         self.encoder = nn.Sequential(\n\u001b[0;32m---> 22\u001b[0;31m             \u001b[0mQuantLinear\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m784\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m256\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mbias\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweight_bit_width\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mweight_bit_width\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[0m\u001b[1;32m     23\u001b[0m             \u001b[0mnn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTanh\u001b[0m\u001b[0;34m(\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     24\u001b[0m             \u001b[0mQuantLinear\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m256\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m128\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mbias\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mweight_bit_width\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mweight_bit_width\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[0;31mNameError\u001b[0m: name 'weight_bit_width' is not defined"
     ]
    }
   ],
   "source": [
    "model_ae = AutoEncoder()\n",
    "print(model_ae)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "a965cd2e",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.utils.data as data\n",
    "import torchvision\n",
    "import torch.optim as optim\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision.transforms import transforms\n",
    "import numpy as np\n",
    "import torchvision.utils as vutils\n",
    "import PIL.Image as Image\n",
    "from torch import nn\n",
    "from torch.nn import Module\n",
    "import torch.nn.functional as F\n",
    "from brevitas.nn import QuantLinear, QuantReLU  \n",
    "import brevitas.nn as qnn\n",
    "from brevitas.core.quant import QuantType\n",
    "\n",
    "model = nn.Sequential(\n",
    "            QuantLinear(input_size, hidden1, bias=True, weight_bit_width=weight_bit_width ),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(hidden1, hidden2, bias=True, weight_bit_width=weight_bit_width  ),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(hidden2, hidden3, bias=True, weight_bit_width=weight_bit_width ),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(hidden3, bottle, bias=True, weight_bit_width=weight_bit_width ),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(bottle, hidden3,bias=True, weight_bit_width=weight_bit_width),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(hidden3, hidden2,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(hidden2, hidden1,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(hidden1, output_size,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            qnn.QuantReLU(quant_type=QuantType.INT, bit_width=weight_bit_width , min_val=0, max_val=255),\n",
    ")\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "08048642",
   "metadata": {},
   "source": [
    "From Now Up\n",
    "======"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "57907097",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_size = 400\n",
    "output_size = 400 \n",
    "hidden1 = 128      \n",
    "hidden2 = 64\n",
    "hidden3 = 32\n",
    "bottle = 2\n",
    "weight_bit_width = 8\n",
    "act_bit_width = 8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "9b896827",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.utils.data as data\n",
    "import torchvision\n",
    "import torch.optim as optim\n",
    "from torch.utils.data import DataLoader\n",
    "from torchvision.transforms import transforms\n",
    "import numpy as np\n",
    "import torchvision.utils as vutils\n",
    "import PIL.Image as Image\n",
    "from torch import nn\n",
    "from torch.nn import Module\n",
    "import torch.nn.functional as F\n",
    "from brevitas.nn import QuantLinear, QuantReLU  \n",
    "import brevitas.nn as qnn\n",
    "from brevitas.core.quant import QuantType\n",
    "\n",
    "model = nn.Sequential(\n",
    "            QuantLinear(input_size, hidden1, bias=True, weight_bit_width=weight_bit_width ),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(hidden1, hidden3, bias=True, weight_bit_width=weight_bit_width  ),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "           # QuantLinear(hidden2, hidden3, bias=True, weight_bit_width=weight_bit_width ),\n",
    "           # qnn.QuantReLU(bit_width=act_bit_width),\n",
    "           # QuantLinear(hidden3, hidden2,bias=True, weight_bit_width=weight_bit_width ),\n",
    "           # qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(hidden3, hidden1,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            qnn.QuantReLU(bit_width=act_bit_width),\n",
    "            QuantLinear(hidden1, output_size,bias=True, weight_bit_width=weight_bit_width ),\n",
    "            qnn.QuantReLU(quant_type=QuantType.INT, bit_width=weight_bit_width , min_val=0, max_val=255)\n",
    "            #qnn.QuantIdentity(quant_type=QuantType.FP, bit_width=weight_bit_width , min_val=0, max_val=1),\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "575155ac",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Sequential(\n",
       "  (0): QuantLinear(\n",
       "    in_features=400, out_features=128, bias=True\n",
       "    (input_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (output_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (weight_quant): WeightQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "      (tensor_quant): RescalingIntQuant(\n",
       "        (int_quant): IntQuant(\n",
       "          (float_to_int_impl): RoundSte()\n",
       "          (tensor_clamp_impl): TensorClampSte()\n",
       "          (delay_wrapper): DelayWrapper(\n",
       "            (delay_impl): _NoDelay()\n",
       "          )\n",
       "        )\n",
       "        (scaling_impl): StatsFromParameterScaling(\n",
       "          (parameter_list_stats): _ParameterListStats(\n",
       "            (first_tracked_param): _ViewParameterWrapper(\n",
       "              (view_shape_impl): OverTensorView()\n",
       "            )\n",
       "            (stats): _Stats(\n",
       "              (stats_impl): AbsMax()\n",
       "            )\n",
       "          )\n",
       "          (stats_scaling_impl): _StatsScaling(\n",
       "            (affine_rescaling): Identity()\n",
       "            (restrict_clamp_scaling): _RestrictClampValue(\n",
       "              (clamp_min_ste): Identity()\n",
       "              (restrict_value_impl): FloatRestrictValue()\n",
       "            )\n",
       "            (restrict_scaling_pre): Identity()\n",
       "          )\n",
       "        )\n",
       "        (int_scaling_impl): IntScaling()\n",
       "        (zero_point_impl): ZeroZeroPoint(\n",
       "          (zero_point): StatelessBuffer()\n",
       "        )\n",
       "        (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "          (bit_width): StatelessBuffer()\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (bias_quant): BiasQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "  )\n",
       "  (1): QuantReLU(\n",
       "    (input_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (act_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "      (fused_activation_quant_proxy): FusedActivationQuantProxy(\n",
       "        (activation_impl): ReLU()\n",
       "        (tensor_quant): RescalingIntQuant(\n",
       "          (int_quant): IntQuant(\n",
       "            (float_to_int_impl): RoundSte()\n",
       "            (tensor_clamp_impl): TensorClamp()\n",
       "            (delay_wrapper): DelayWrapper(\n",
       "              (delay_impl): _NoDelay()\n",
       "            )\n",
       "          )\n",
       "          (scaling_impl): ParameterFromRuntimeStatsScaling(\n",
       "            (stats_input_view_shape_impl): OverTensorView()\n",
       "            (stats): _Stats(\n",
       "              (stats_impl): AbsPercentile()\n",
       "            )\n",
       "            (restrict_clamp_scaling): _RestrictClampValue(\n",
       "              (clamp_min_ste): Identity()\n",
       "              (restrict_value_impl): FloatRestrictValue()\n",
       "            )\n",
       "            (restrict_inplace_preprocess): Identity()\n",
       "            (restrict_preprocess): Identity()\n",
       "          )\n",
       "          (int_scaling_impl): IntScaling()\n",
       "          (zero_point_impl): ZeroZeroPoint(\n",
       "            (zero_point): StatelessBuffer()\n",
       "          )\n",
       "          (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "            (bit_width): StatelessBuffer()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "  )\n",
       "  (2): QuantLinear(\n",
       "    in_features=128, out_features=32, bias=True\n",
       "    (input_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (output_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (weight_quant): WeightQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "      (tensor_quant): RescalingIntQuant(\n",
       "        (int_quant): IntQuant(\n",
       "          (float_to_int_impl): RoundSte()\n",
       "          (tensor_clamp_impl): TensorClampSte()\n",
       "          (delay_wrapper): DelayWrapper(\n",
       "            (delay_impl): _NoDelay()\n",
       "          )\n",
       "        )\n",
       "        (scaling_impl): StatsFromParameterScaling(\n",
       "          (parameter_list_stats): _ParameterListStats(\n",
       "            (first_tracked_param): _ViewParameterWrapper(\n",
       "              (view_shape_impl): OverTensorView()\n",
       "            )\n",
       "            (stats): _Stats(\n",
       "              (stats_impl): AbsMax()\n",
       "            )\n",
       "          )\n",
       "          (stats_scaling_impl): _StatsScaling(\n",
       "            (affine_rescaling): Identity()\n",
       "            (restrict_clamp_scaling): _RestrictClampValue(\n",
       "              (clamp_min_ste): Identity()\n",
       "              (restrict_value_impl): FloatRestrictValue()\n",
       "            )\n",
       "            (restrict_scaling_pre): Identity()\n",
       "          )\n",
       "        )\n",
       "        (int_scaling_impl): IntScaling()\n",
       "        (zero_point_impl): ZeroZeroPoint(\n",
       "          (zero_point): StatelessBuffer()\n",
       "        )\n",
       "        (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "          (bit_width): StatelessBuffer()\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (bias_quant): BiasQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "  )\n",
       "  (3): QuantReLU(\n",
       "    (input_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (act_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "      (fused_activation_quant_proxy): FusedActivationQuantProxy(\n",
       "        (activation_impl): ReLU()\n",
       "        (tensor_quant): RescalingIntQuant(\n",
       "          (int_quant): IntQuant(\n",
       "            (float_to_int_impl): RoundSte()\n",
       "            (tensor_clamp_impl): TensorClamp()\n",
       "            (delay_wrapper): DelayWrapper(\n",
       "              (delay_impl): _NoDelay()\n",
       "            )\n",
       "          )\n",
       "          (scaling_impl): ParameterFromRuntimeStatsScaling(\n",
       "            (stats_input_view_shape_impl): OverTensorView()\n",
       "            (stats): _Stats(\n",
       "              (stats_impl): AbsPercentile()\n",
       "            )\n",
       "            (restrict_clamp_scaling): _RestrictClampValue(\n",
       "              (clamp_min_ste): Identity()\n",
       "              (restrict_value_impl): FloatRestrictValue()\n",
       "            )\n",
       "            (restrict_inplace_preprocess): Identity()\n",
       "            (restrict_preprocess): Identity()\n",
       "          )\n",
       "          (int_scaling_impl): IntScaling()\n",
       "          (zero_point_impl): ZeroZeroPoint(\n",
       "            (zero_point): StatelessBuffer()\n",
       "          )\n",
       "          (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "            (bit_width): StatelessBuffer()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "  )\n",
       "  (4): QuantLinear(\n",
       "    in_features=32, out_features=128, bias=True\n",
       "    (input_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (output_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (weight_quant): WeightQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "      (tensor_quant): RescalingIntQuant(\n",
       "        (int_quant): IntQuant(\n",
       "          (float_to_int_impl): RoundSte()\n",
       "          (tensor_clamp_impl): TensorClampSte()\n",
       "          (delay_wrapper): DelayWrapper(\n",
       "            (delay_impl): _NoDelay()\n",
       "          )\n",
       "        )\n",
       "        (scaling_impl): StatsFromParameterScaling(\n",
       "          (parameter_list_stats): _ParameterListStats(\n",
       "            (first_tracked_param): _ViewParameterWrapper(\n",
       "              (view_shape_impl): OverTensorView()\n",
       "            )\n",
       "            (stats): _Stats(\n",
       "              (stats_impl): AbsMax()\n",
       "            )\n",
       "          )\n",
       "          (stats_scaling_impl): _StatsScaling(\n",
       "            (affine_rescaling): Identity()\n",
       "            (restrict_clamp_scaling): _RestrictClampValue(\n",
       "              (clamp_min_ste): Identity()\n",
       "              (restrict_value_impl): FloatRestrictValue()\n",
       "            )\n",
       "            (restrict_scaling_pre): Identity()\n",
       "          )\n",
       "        )\n",
       "        (int_scaling_impl): IntScaling()\n",
       "        (zero_point_impl): ZeroZeroPoint(\n",
       "          (zero_point): StatelessBuffer()\n",
       "        )\n",
       "        (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "          (bit_width): StatelessBuffer()\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (bias_quant): BiasQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "  )\n",
       "  (5): QuantReLU(\n",
       "    (input_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (act_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "      (fused_activation_quant_proxy): FusedActivationQuantProxy(\n",
       "        (activation_impl): ReLU()\n",
       "        (tensor_quant): RescalingIntQuant(\n",
       "          (int_quant): IntQuant(\n",
       "            (float_to_int_impl): RoundSte()\n",
       "            (tensor_clamp_impl): TensorClamp()\n",
       "            (delay_wrapper): DelayWrapper(\n",
       "              (delay_impl): _NoDelay()\n",
       "            )\n",
       "          )\n",
       "          (scaling_impl): ParameterFromRuntimeStatsScaling(\n",
       "            (stats_input_view_shape_impl): OverTensorView()\n",
       "            (stats): _Stats(\n",
       "              (stats_impl): AbsPercentile()\n",
       "            )\n",
       "            (restrict_clamp_scaling): _RestrictClampValue(\n",
       "              (clamp_min_ste): Identity()\n",
       "              (restrict_value_impl): FloatRestrictValue()\n",
       "            )\n",
       "            (restrict_inplace_preprocess): Identity()\n",
       "            (restrict_preprocess): Identity()\n",
       "          )\n",
       "          (int_scaling_impl): IntScaling()\n",
       "          (zero_point_impl): ZeroZeroPoint(\n",
       "            (zero_point): StatelessBuffer()\n",
       "          )\n",
       "          (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "            (bit_width): StatelessBuffer()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "  )\n",
       "  (6): QuantLinear(\n",
       "    in_features=128, out_features=400, bias=True\n",
       "    (input_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (output_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (weight_quant): WeightQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "      (tensor_quant): RescalingIntQuant(\n",
       "        (int_quant): IntQuant(\n",
       "          (float_to_int_impl): RoundSte()\n",
       "          (tensor_clamp_impl): TensorClampSte()\n",
       "          (delay_wrapper): DelayWrapper(\n",
       "            (delay_impl): _NoDelay()\n",
       "          )\n",
       "        )\n",
       "        (scaling_impl): StatsFromParameterScaling(\n",
       "          (parameter_list_stats): _ParameterListStats(\n",
       "            (first_tracked_param): _ViewParameterWrapper(\n",
       "              (view_shape_impl): OverTensorView()\n",
       "            )\n",
       "            (stats): _Stats(\n",
       "              (stats_impl): AbsMax()\n",
       "            )\n",
       "          )\n",
       "          (stats_scaling_impl): _StatsScaling(\n",
       "            (affine_rescaling): Identity()\n",
       "            (restrict_clamp_scaling): _RestrictClampValue(\n",
       "              (clamp_min_ste): Identity()\n",
       "              (restrict_value_impl): FloatRestrictValue()\n",
       "            )\n",
       "            (restrict_scaling_pre): Identity()\n",
       "          )\n",
       "        )\n",
       "        (int_scaling_impl): IntScaling()\n",
       "        (zero_point_impl): ZeroZeroPoint(\n",
       "          (zero_point): StatelessBuffer()\n",
       "        )\n",
       "        (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "          (bit_width): StatelessBuffer()\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (bias_quant): BiasQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "  )\n",
       "  (7): QuantReLU(\n",
       "    (input_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "    )\n",
       "    (act_quant): ActQuantProxyFromInjector(\n",
       "      (_zero_hw_sentinel): StatelessBuffer()\n",
       "      (fused_activation_quant_proxy): FusedActivationQuantProxy(\n",
       "        (activation_impl): ReLU()\n",
       "        (tensor_quant): RescalingIntQuant(\n",
       "          (int_quant): IntQuant(\n",
       "            (float_to_int_impl): RoundSte()\n",
       "            (tensor_clamp_impl): TensorClamp()\n",
       "            (delay_wrapper): DelayWrapper(\n",
       "              (delay_impl): _NoDelay()\n",
       "            )\n",
       "          )\n",
       "          (scaling_impl): ParameterFromRuntimeStatsScaling(\n",
       "            (stats_input_view_shape_impl): OverTensorView()\n",
       "            (stats): _Stats(\n",
       "              (stats_impl): AbsPercentile()\n",
       "            )\n",
       "            (restrict_clamp_scaling): _RestrictClampValue(\n",
       "              (clamp_min_ste): Identity()\n",
       "              (restrict_value_impl): FloatRestrictValue()\n",
       "            )\n",
       "            (restrict_inplace_preprocess): Identity()\n",
       "            (restrict_preprocess): Identity()\n",
       "          )\n",
       "          (int_scaling_impl): IntScaling()\n",
       "          (zero_point_impl): ZeroZeroPoint(\n",
       "            (zero_point): StatelessBuffer()\n",
       "          )\n",
       "          (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "            (bit_width): StatelessBuffer()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "54fb5bb2",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from torchvision import transforms, datasets\n",
    "import os, sys\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "from PIL import Image\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "class my_dataset(Dataset):\n",
    "    def __init__(self, img_path, data_transform = None):\n",
    "        self.img_path = img_path\n",
    "        self.transforms = data_transform\n",
    "        self.img_list = []\n",
    "        \n",
    "        for file in os.listdir(self.img_path):\n",
    "            img_path = os.path.join(self.img_path, file)\n",
    "            self.img_list.append(img_path)\n",
    "    def __getitem__(self, item):\n",
    "        img = Image.open(self.img_list[item]).convert('RGB')\n",
    "        img = img.resize((20,20), Image.ANTIALIAS)\n",
    "        img_np = np.array(img)\n",
    "        \n",
    "        if self.transforms is not None:\n",
    "            img = self.transforms(img_np)\n",
    "        \n",
    "        return img\n",
    "    def __len__(self):\n",
    "        return len(self.img_list)\n",
    "    \n",
    "data_transform = transforms.Compose([\n",
    "    transforms.ToPILImage(),\n",
    "    transforms.ToTensor(),\n",
    "])\n",
    "\n",
    "img_path = 'bottle/train/good'\n",
    "train_dataset = my_dataset(img_path, data_transform)\n",
    "\n",
    "dataset_loader = DataLoader(train_dataset,\n",
    "                           batch_size = 1,\n",
    "                           shuffle = True,\n",
    "                           num_workers = 1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "a98586bb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[255. 255. 255. 255. 254. 254. 255. 255. 255. 245. 245. 255. 255. 255.\n",
      "  254. 254. 255. 255. 255. 255.]\n",
      " [255. 255. 255. 253. 255. 255. 202. 136.  99.  88.  92. 109. 150. 211.\n",
      "  255. 255. 253. 255. 255. 255.]\n",
      " [255. 255. 253. 255. 228. 114.  52.  51.  61.  72.  81.  78.  68.  73.\n",
      "  130. 233. 255. 253. 255. 255.]\n",
      " [255. 253. 255. 216.  81.  60.  85.  91.  84.  87.  95. 100. 106.  91.\n",
      "   71.  93. 219. 255. 254. 255.]\n",
      " [254. 254. 235.  89.  79. 100.  83.  63.  56.  52.  54.  59.  72.  86.\n",
      "   94.  74.  95. 235. 254. 254.]\n",
      " [253. 255. 133.  74. 107.  83.  53.  44.  41.  39.  40.  44.  50.  62.\n",
      "   83. 102.  70. 136. 255. 253.]\n",
      " [255. 222.  77. 101.  92.  56.  44.  43.  35.  32.  32.  34.  38.  49.\n",
      "   63.  96.  95.  75. 220. 255.]\n",
      " [255. 164.  78. 107.  72.  46.  43.  32.  34.  34.  33.  32.  32.  42.\n",
      "   56.  75. 119.  67. 160. 255.]\n",
      " [255. 132.  89.  98.  60.  45.  35.  34.  37.  37.  36.  36.  34.  38.\n",
      "   51.  63. 115.  79. 121. 255.]\n",
      " [255. 122.  94.  88.  57.  44.  33.  35.  37.  39.  42.  40.  37.  35.\n",
      "   47.  60.  94.  85. 107. 254.]\n",
      " [255. 119.  95.  89.  55.  44.  33.  35.  36.  41.  45.  41.  36.  34.\n",
      "   46.  60.  93.  84. 110. 255.]\n",
      " [255. 133.  93.  97.  59.  47.  34.  34.  36.  38.  39.  38.  35.  36.\n",
      "   50.  63.  96.  79. 127. 255.]\n",
      " [255. 169.  85. 112.  73.  52.  42.  32.  34.  35.  35.  34.  32.  42.\n",
      "   55.  72. 102.  70. 162. 255.]\n",
      " [255. 224.  85. 127. 118.  60.  52.  41.  33.  32.  32.  33.  41.  55.\n",
      "   64.  90.  97.  73. 220. 255.]\n",
      " [253. 255. 137.  81. 115.  74.  58.  54.  49.  44.  45.  51.  64.  71.\n",
      "   86. 107.  72. 133. 255. 253.]\n",
      " [254. 254. 233.  84.  77.  94.  70.  59.  59.  56.  55.  64.  77.  94.\n",
      "  119.  83.  92. 234. 254. 254.]\n",
      " [255. 253. 255. 213.  80.  69.  88.  79.  73.  71.  82.  86.  94. 107.\n",
      "   79.  88. 217. 255. 253. 255.]\n",
      " [255. 255. 253. 255. 229. 117.  65.  72.  75.  77.  86.  78.  74.  74.\n",
      "  120. 227. 255. 253. 255. 255.]\n",
      " [255. 255. 255. 253. 255. 255. 205. 145. 103.  92.  89. 101. 140. 204.\n",
      "  255. 255. 253. 255. 255. 255.]\n",
      " [255. 255. 255. 255. 254. 254. 255. 255. 255. 244. 244. 255. 255. 255.\n",
      "  254. 254. 255. 255. 255. 255.]]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-75-648ab9a0185b>:20: DeprecationWarning: ANTIALIAS is deprecated and will be removed in Pillow 10 (2023-07-01). Use Resampling.LANCZOS instead.\n",
      "  img = img.resize((20,20), Image.ANTIALIAS)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f3dbb592ee0>"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "img = train_dataset.__getitem__(0)\n",
    "img = img.numpy()*255\n",
    "print(img[0])\n",
    "img = img.swapaxes(0, 1)\n",
    "img = img.swapaxes(1, 2)\n",
    "plt.imshow(img/255)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "3ac4a249",
   "metadata": {},
   "outputs": [],
   "source": [
    "build_dir = \"/workspace/finn\"\n",
    "\n",
    "epochs = 1\n",
    "batch_size = 64\n",
    "lr = 0.001\n",
    "# model_ae = AutoEncoder()\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=lr)\n",
    "\n",
    "loss_function = nn.MSELoss()\n",
    "\n",
    "#loss_function = nn.BCEWithLogitsLoss()\n",
    "scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer, milestones=[10,40], gamma=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "0b22c76c",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-75-648ab9a0185b>:20: DeprecationWarning: ANTIALIAS is deprecated and will be removed in Pillow 10 (2023-07-01). Use Resampling.LANCZOS instead.\n",
      "  img = img.resize((20,20), Image.ANTIALIAS)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Restructured Image by AE\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1/1] Loss: 11124.7294921875\n",
      "[1/1] Loss: 11124.7294921875\n"
     ]
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "# Settings\n",
    "plt.rcParams['figure.figsize'] = (10.0, 8.0)\n",
    "plt.rcParams['image.interpolation'] = 'nearest'\n",
    "plt.rcParams['image.cmap'] = 'gray'\n",
    "\n",
    "# Show images\n",
    "def show_images(images):\n",
    "    sqrtn = int(np.ceil(np.sqrt(images.shape[0])))\n",
    "    for index, image in enumerate(images):\n",
    "        plt.subplot(sqrtn, sqrtn, index+1)\n",
    "        plt.imshow(image.reshape(20, 20))\n",
    "        plt.axis('off')\n",
    "\n",
    "# Train\n",
    "log_loss=[]\n",
    "for epoch in range(epochs):\n",
    "    total_loss = 0\n",
    "    for idx, data in enumerate(dataset_loader):\n",
    "        inputs = data.view(-1, 400)*255\n",
    "        #model_ae.zero_grad()\n",
    "        model.zero_grad()\n",
    "        # Forward\n",
    "        #codes, decoded = model_ae(inputs)\n",
    "        decoded = model(inputs)\n",
    "        outputs = decoded.detach().cpu()\n",
    "        loss = loss_function(decoded, inputs)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        total_loss+=loss\n",
    "        log_loss.append(loss)\n",
    "        \n",
    "    print('Restructured Image by AE')\n",
    "    show_images(outputs/255)\n",
    "    plt.show()\n",
    "    total_loss /= len(dataset_loader.dataset)\n",
    "    scheduler.step()\n",
    "    \n",
    "    if epoch % 5 ==0:\n",
    "        print('[{}/{}] Loss:'.format(epoch+1, epochs), total_loss.item())\n",
    "print('[{}/{}] Loss:'.format(epoch+1, epochs), total_loss.item())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "a9f292cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(model.state_dict(), build_dir+\"/autoencoder_ln.pth\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "4413e1dd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trained_state_dict = torch.load(build_dir+\"/autoencoder_ln.pth\")\n",
    "\n",
    "model.load_state_dict(trained_state_dict, strict=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "e72a7023",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "inputs = data.view(-1, 400)*255\n",
    "show_images(model(inputs).detach().cpu())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "da192fb7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([  0.0000, 256.6771, 255.6545, 255.6545, 255.6545,   0.0000,   0.0000,\n",
       "        254.6319,   0.0000,   0.0000,   0.0000, 256.6771,   0.0000,   0.0000,\n",
       "          0.0000, 253.6093, 254.6319, 255.6545, 255.6545, 254.6319, 252.5867,\n",
       "        255.6545,   0.0000, 254.6319,   0.0000, 255.6545, 229.0664, 167.7094,\n",
       "        127.8273, 108.3975, 107.3749,   0.0000, 165.6641, 223.9534, 254.6319,\n",
       "          0.0000, 254.6319,   0.0000,   0.0000, 256.6771, 256.6771, 254.6319,\n",
       "        255.6545,   0.0000, 246.4510, 150.3249,  72.6059,  60.3345,  60.3345,\n",
       "         66.4702,   0.0000,  65.4476,  62.3797,  77.7190,   0.0000, 240.3152,\n",
       "          0.0000, 253.6093, 254.6319, 255.6545,   0.0000, 254.6319, 254.6319,\n",
       "        231.1117,   0.0000,  68.5154,  84.8773,  89.9904,   0.0000,  77.7190,\n",
       "         81.8094,  82.8321,   0.0000,   0.0000,  73.6285, 107.3749, 224.9760,\n",
       "          0.0000,   0.0000,   0.0000,   0.0000, 254.6319,   0.0000, 105.3297,\n",
       "          0.0000,  96.1261,   0.0000,  74.6511,  65.4476,   0.0000,  61.3571,\n",
       "         63.4023,  71.5833,  81.8094,  91.0130,  79.7642, 107.3749, 238.2700,\n",
       "          0.0000,   0.0000,   0.0000, 255.6545,   0.0000,   0.0000,  98.1713,\n",
       "          0.0000,   0.0000,  58.2892,  54.1988,  48.0630,   0.0000,  51.1309,\n",
       "          0.0000,  62.3797,  81.8094,  93.0582,  81.8094,   0.0000, 254.6319,\n",
       "        253.6093,   0.0000, 225.9986,   0.0000,   0.0000,  89.9904,  69.5380,\n",
       "          0.0000,   0.0000,   0.0000,  37.8369,   0.0000,  34.7690,  42.9500,\n",
       "         50.1083,  62.3797,  88.9678,   0.0000,  84.8773, 221.9081, 255.6545,\n",
       "          0.0000, 166.6867,   0.0000,  98.1713,   0.0000,  62.3797,  48.0630,\n",
       "          0.0000,   0.0000,   0.0000,  36.8143,   0.0000,  36.8143,   0.0000,\n",
       "         51.1309,   0.0000, 112.4880,  80.7868,   0.0000, 255.6545, 255.6545,\n",
       "        128.8499,   0.0000,   0.0000,  76.6964,  55.2214,   0.0000,  39.8821,\n",
       "          0.0000,  42.9500,   0.0000,   0.0000,  37.8369,  36.8143,  46.0178,\n",
       "          0.0000,   0.0000,   0.0000, 126.8046, 255.6545,   0.0000, 117.6011,\n",
       "         95.1035,   0.0000,  69.5380,  53.1761,   0.0000,  39.8821,  39.8821,\n",
       "         44.9952,  43.9726,   0.0000,  37.8369,  33.7464,  46.0178,  57.2666,\n",
       "         85.8999,  89.9904, 115.5558, 249.5188,   0.0000, 115.5558,  95.1035,\n",
       "          0.0000,  66.4702,  51.1309,  40.9047,  38.8595,   0.0000,  44.9952,\n",
       "          0.0000,  42.9500,  38.8595,   0.0000,  44.9952,   0.0000,  83.8547,\n",
       "          0.0000,   0.0000,   0.0000, 255.6545, 128.8499,   0.0000,   0.0000,\n",
       "         72.6059,  54.1988,   0.0000,  38.8595,   0.0000,  40.9047,  41.9273,\n",
       "         37.8369,   0.0000,  35.7916,   0.0000,   0.0000,  89.9904,  82.8321,\n",
       "          0.0000, 255.6545, 254.6319, 157.4832,  83.8547,   0.0000,  85.8999,\n",
       "         59.3118,   0.0000,   0.0000,   0.0000,  35.7916,   0.0000,  36.8143,\n",
       "          0.0000,  43.9726,  50.1083,  69.5380,  94.0809,   0.0000,   0.0000,\n",
       "          0.0000,   0.0000, 211.6819,   0.0000, 115.5558,   0.0000,   0.0000,\n",
       "         57.2666,  49.0857,   0.0000,  37.8369,   0.0000,   0.0000,   0.0000,\n",
       "          0.0000,  62.3797,   0.0000,   0.0000,   0.0000,   0.0000, 253.6093,\n",
       "        252.5867, 256.6771, 122.7142,  78.7416, 101.2392,   0.0000,   0.0000,\n",
       "         56.2440,   0.0000,   0.0000,   0.0000,  47.0404,  59.3118,   0.0000,\n",
       "         81.8094,  95.1035,   0.0000, 138.0534, 256.6771, 254.6319, 253.6093,\n",
       "        253.6093, 230.0891,  86.9225,  76.6964,  83.8547,  70.5606,  57.2666,\n",
       "         58.2892,  59.3118,   0.0000,   0.0000,  71.5833,  84.8773,  99.1940,\n",
       "         85.8999,  99.1940,   0.0000,   0.0000,   0.0000,   0.0000, 253.6093,\n",
       "        256.6771,   0.0000,   0.0000,   0.0000,  81.8094,  76.6964,   0.0000,\n",
       "         71.5833,   0.0000,  72.6059,  74.6511,  93.0582,   0.0000,   0.0000,\n",
       "        217.8176,   0.0000,   0.0000, 255.6545,   0.0000,   0.0000,   0.0000,\n",
       "        254.6319, 222.9307,   0.0000,  69.5380,  66.4702,   0.0000,  72.6059,\n",
       "         74.6511,  72.6059,  66.4702,   0.0000, 131.9177,   0.0000, 254.6319,\n",
       "        252.5867,   0.0000,   0.0000,   0.0000, 255.6545, 254.6319, 253.6093,\n",
       "        254.6319, 254.6319,   0.0000, 140.0987, 105.3297,   0.0000,   0.0000,\n",
       "        113.5106,   0.0000,   0.0000, 255.6545, 257.6997, 254.6319, 256.6771,\n",
       "          0.0000, 256.6771, 253.6093,   0.0000,   0.0000, 254.6319,   0.0000,\n",
       "        254.6319, 255.6545,   0.0000, 254.6319, 242.3605, 245.4283,   0.0000,\n",
       "        257.6997, 255.6545,   0.0000, 254.6319,   0.0000, 255.6545, 256.6771,\n",
       "        256.6771])"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model(inputs).detach().cpu()[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "8b819cb1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(128, 400)"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from copy import deepcopy\n",
    "\n",
    "modified_model = deepcopy(model)\n",
    "\n",
    "W_orig = modified_model[0].weight.data.detach().numpy()\n",
    "W_orig.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "2fec0b61",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "QuantLinear(\n",
       "  in_features=400, out_features=128, bias=True\n",
       "  (input_quant): ActQuantProxyFromInjector(\n",
       "    (_zero_hw_sentinel): StatelessBuffer()\n",
       "  )\n",
       "  (output_quant): ActQuantProxyFromInjector(\n",
       "    (_zero_hw_sentinel): StatelessBuffer()\n",
       "  )\n",
       "  (weight_quant): WeightQuantProxyFromInjector(\n",
       "    (_zero_hw_sentinel): StatelessBuffer()\n",
       "    (tensor_quant): RescalingIntQuant(\n",
       "      (int_quant): IntQuant(\n",
       "        (float_to_int_impl): RoundSte()\n",
       "        (tensor_clamp_impl): TensorClampSte()\n",
       "        (delay_wrapper): DelayWrapper(\n",
       "          (delay_impl): _NoDelay()\n",
       "        )\n",
       "      )\n",
       "      (scaling_impl): StatsFromParameterScaling(\n",
       "        (parameter_list_stats): _ParameterListStats(\n",
       "          (first_tracked_param): _ViewParameterWrapper(\n",
       "            (view_shape_impl): OverTensorView()\n",
       "          )\n",
       "          (stats): _Stats(\n",
       "            (stats_impl): AbsMax()\n",
       "          )\n",
       "        )\n",
       "        (stats_scaling_impl): _StatsScaling(\n",
       "          (affine_rescaling): Identity()\n",
       "          (restrict_clamp_scaling): _RestrictClampValue(\n",
       "            (clamp_min_ste): Identity()\n",
       "            (restrict_value_impl): FloatRestrictValue()\n",
       "          )\n",
       "          (restrict_scaling_pre): Identity()\n",
       "        )\n",
       "      )\n",
       "      (int_scaling_impl): IntScaling()\n",
       "      (zero_point_impl): ZeroZeroPoint(\n",
       "        (zero_point): StatelessBuffer()\n",
       "      )\n",
       "      (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "        (bit_width): StatelessBuffer()\n",
       "      )\n",
       "    )\n",
       "  )\n",
       "  (bias_quant): BiasQuantProxyFromInjector(\n",
       "    (_zero_hw_sentinel): StatelessBuffer()\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "ec7e9458",
   "metadata": {},
   "outputs": [],
   "source": [
    "from brevitas.core.quant import QuantType\n",
    "from brevitas.nn import QuantIdentity\n",
    "\n",
    "\n",
    "class AutoencoderForExport(nn.Module):\n",
    "    def __init__(self, my_pretrained_model):\n",
    "        super(AutoencoderForExport, self).__init__()\n",
    "        self.pretrained = my_pretrained_model\n",
    "        #self.qnt_output = QuantReLU(quant_type=QuantType.INT, bit_width=8, min_val=0, max_val=255)\n",
    "    \n",
    "    def forward(self, x):\n",
    "        # assume x contains bipolar {-1,1} elems\n",
    "        # shift from {-1,1} -> {0,1} since that is the\n",
    "        # input range for the trained network\n",
    "        # x = (x + torch.tensor([1.0])) / 2.0  \n",
    "        x = x*255\n",
    "        out = self.pretrained(x)\n",
    "        #out_original = self.pretrained(x)\n",
    "        #out = self.qnt_output(out_original)   # output as {-1,1}     \n",
    "        #return out_final\n",
    "\n",
    "        #return out_original\n",
    "        return out\n",
    "model_for_export = AutoencoderForExport(modified_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "25d17c7c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class '__main__.my_dataset'>\n"
     ]
    }
   ],
   "source": [
    "img_path = 'bottle/test/contamination'\n",
    "anomaly_data = my_dataset(img_path, data_transform)\n",
    "print(type(anomaly_data))\n",
    "dataset_loader = DataLoader(anomaly_data,\n",
    "                           batch_size = 1,\n",
    "                           shuffle = True,\n",
    "                           num_workers = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "2b2e0af4",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-75-648ab9a0185b>:20: DeprecationWarning: ANTIALIAS is deprecated and will be removed in Pillow 10 (2023-07-01). Use Resampling.LANCZOS instead.\n",
      "  img = img.resize((20,20), Image.ANTIALIAS)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f3d9a500ac0>"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "img = anomaly_data.__getitem__(6)\n",
    "img = img.numpy()\n",
    "img = img.swapaxes(0, 1)\n",
    "img = img.swapaxes(1, 2)\n",
    "plt.imshow(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "id": "f95986fe",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original Images\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-75-648ab9a0185b>:20: DeprecationWarning: ANTIALIAS is deprecated and will be removed in Pillow 10 (2023-07-01). Use Resampling.LANCZOS instead.\n",
      "  img = img.resize((20,20), Image.ANTIALIAS)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "inputs = anomaly_data.__getitem__(6).view(-1, 400)\n",
    "print('Original Images')\n",
    "show_images(inputs)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "id": "9f8fd85d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "AutoencoderForExport(\n",
       "  (pretrained): Sequential(\n",
       "    (0): QuantLinear(\n",
       "      in_features=400, out_features=128, bias=True\n",
       "      (input_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (output_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (weight_quant): WeightQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "        (tensor_quant): RescalingIntQuant(\n",
       "          (int_quant): IntQuant(\n",
       "            (float_to_int_impl): RoundSte()\n",
       "            (tensor_clamp_impl): TensorClampSte()\n",
       "            (delay_wrapper): DelayWrapper(\n",
       "              (delay_impl): _NoDelay()\n",
       "            )\n",
       "          )\n",
       "          (scaling_impl): StatsFromParameterScaling(\n",
       "            (parameter_list_stats): _ParameterListStats(\n",
       "              (first_tracked_param): _ViewParameterWrapper(\n",
       "                (view_shape_impl): OverTensorView()\n",
       "              )\n",
       "              (stats): _Stats(\n",
       "                (stats_impl): AbsMax()\n",
       "              )\n",
       "            )\n",
       "            (stats_scaling_impl): _StatsScaling(\n",
       "              (affine_rescaling): Identity()\n",
       "              (restrict_clamp_scaling): _RestrictClampValue(\n",
       "                (clamp_min_ste): Identity()\n",
       "                (restrict_value_impl): FloatRestrictValue()\n",
       "              )\n",
       "              (restrict_scaling_pre): Identity()\n",
       "            )\n",
       "          )\n",
       "          (int_scaling_impl): IntScaling()\n",
       "          (zero_point_impl): ZeroZeroPoint(\n",
       "            (zero_point): StatelessBuffer()\n",
       "          )\n",
       "          (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "            (bit_width): StatelessBuffer()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (bias_quant): BiasQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (export_handler): FINNQuantLinearHandler()\n",
       "    )\n",
       "    (1): QuantReLU(\n",
       "      (input_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (act_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "        (fused_activation_quant_proxy): FusedActivationQuantProxy(\n",
       "          (activation_impl): ReLU()\n",
       "          (tensor_quant): RescalingIntQuant(\n",
       "            (int_quant): IntQuant(\n",
       "              (float_to_int_impl): RoundSte()\n",
       "              (tensor_clamp_impl): TensorClamp()\n",
       "              (delay_wrapper): DelayWrapper(\n",
       "                (delay_impl): _NoDelay()\n",
       "              )\n",
       "            )\n",
       "            (scaling_impl): ParameterFromRuntimeStatsScaling(\n",
       "              (stats_input_view_shape_impl): OverTensorView()\n",
       "              (stats): _Stats(\n",
       "                (stats_impl): AbsPercentile()\n",
       "              )\n",
       "              (restrict_clamp_scaling): _RestrictClampValue(\n",
       "                (clamp_min_ste): Identity()\n",
       "                (restrict_value_impl): FloatRestrictValue()\n",
       "              )\n",
       "              (restrict_inplace_preprocess): Identity()\n",
       "              (restrict_preprocess): Identity()\n",
       "            )\n",
       "            (int_scaling_impl): IntScaling()\n",
       "            (zero_point_impl): ZeroZeroPoint(\n",
       "              (zero_point): StatelessBuffer()\n",
       "            )\n",
       "            (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "              (bit_width): StatelessBuffer()\n",
       "            )\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (export_handler): FINNQuantReLUHandler()\n",
       "    )\n",
       "    (2): QuantLinear(\n",
       "      in_features=128, out_features=32, bias=True\n",
       "      (input_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (output_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (weight_quant): WeightQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "        (tensor_quant): RescalingIntQuant(\n",
       "          (int_quant): IntQuant(\n",
       "            (float_to_int_impl): RoundSte()\n",
       "            (tensor_clamp_impl): TensorClampSte()\n",
       "            (delay_wrapper): DelayWrapper(\n",
       "              (delay_impl): _NoDelay()\n",
       "            )\n",
       "          )\n",
       "          (scaling_impl): StatsFromParameterScaling(\n",
       "            (parameter_list_stats): _ParameterListStats(\n",
       "              (first_tracked_param): _ViewParameterWrapper(\n",
       "                (view_shape_impl): OverTensorView()\n",
       "              )\n",
       "              (stats): _Stats(\n",
       "                (stats_impl): AbsMax()\n",
       "              )\n",
       "            )\n",
       "            (stats_scaling_impl): _StatsScaling(\n",
       "              (affine_rescaling): Identity()\n",
       "              (restrict_clamp_scaling): _RestrictClampValue(\n",
       "                (clamp_min_ste): Identity()\n",
       "                (restrict_value_impl): FloatRestrictValue()\n",
       "              )\n",
       "              (restrict_scaling_pre): Identity()\n",
       "            )\n",
       "          )\n",
       "          (int_scaling_impl): IntScaling()\n",
       "          (zero_point_impl): ZeroZeroPoint(\n",
       "            (zero_point): StatelessBuffer()\n",
       "          )\n",
       "          (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "            (bit_width): StatelessBuffer()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (bias_quant): BiasQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (export_handler): FINNQuantLinearHandler()\n",
       "    )\n",
       "    (3): QuantReLU(\n",
       "      (input_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (act_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "        (fused_activation_quant_proxy): FusedActivationQuantProxy(\n",
       "          (activation_impl): ReLU()\n",
       "          (tensor_quant): RescalingIntQuant(\n",
       "            (int_quant): IntQuant(\n",
       "              (float_to_int_impl): RoundSte()\n",
       "              (tensor_clamp_impl): TensorClamp()\n",
       "              (delay_wrapper): DelayWrapper(\n",
       "                (delay_impl): _NoDelay()\n",
       "              )\n",
       "            )\n",
       "            (scaling_impl): ParameterFromRuntimeStatsScaling(\n",
       "              (stats_input_view_shape_impl): OverTensorView()\n",
       "              (stats): _Stats(\n",
       "                (stats_impl): AbsPercentile()\n",
       "              )\n",
       "              (restrict_clamp_scaling): _RestrictClampValue(\n",
       "                (clamp_min_ste): Identity()\n",
       "                (restrict_value_impl): FloatRestrictValue()\n",
       "              )\n",
       "              (restrict_inplace_preprocess): Identity()\n",
       "              (restrict_preprocess): Identity()\n",
       "            )\n",
       "            (int_scaling_impl): IntScaling()\n",
       "            (zero_point_impl): ZeroZeroPoint(\n",
       "              (zero_point): StatelessBuffer()\n",
       "            )\n",
       "            (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "              (bit_width): StatelessBuffer()\n",
       "            )\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (export_handler): FINNQuantReLUHandler()\n",
       "    )\n",
       "    (4): QuantLinear(\n",
       "      in_features=32, out_features=128, bias=True\n",
       "      (input_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (output_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (weight_quant): WeightQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "        (tensor_quant): RescalingIntQuant(\n",
       "          (int_quant): IntQuant(\n",
       "            (float_to_int_impl): RoundSte()\n",
       "            (tensor_clamp_impl): TensorClampSte()\n",
       "            (delay_wrapper): DelayWrapper(\n",
       "              (delay_impl): _NoDelay()\n",
       "            )\n",
       "          )\n",
       "          (scaling_impl): StatsFromParameterScaling(\n",
       "            (parameter_list_stats): _ParameterListStats(\n",
       "              (first_tracked_param): _ViewParameterWrapper(\n",
       "                (view_shape_impl): OverTensorView()\n",
       "              )\n",
       "              (stats): _Stats(\n",
       "                (stats_impl): AbsMax()\n",
       "              )\n",
       "            )\n",
       "            (stats_scaling_impl): _StatsScaling(\n",
       "              (affine_rescaling): Identity()\n",
       "              (restrict_clamp_scaling): _RestrictClampValue(\n",
       "                (clamp_min_ste): Identity()\n",
       "                (restrict_value_impl): FloatRestrictValue()\n",
       "              )\n",
       "              (restrict_scaling_pre): Identity()\n",
       "            )\n",
       "          )\n",
       "          (int_scaling_impl): IntScaling()\n",
       "          (zero_point_impl): ZeroZeroPoint(\n",
       "            (zero_point): StatelessBuffer()\n",
       "          )\n",
       "          (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "            (bit_width): StatelessBuffer()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (bias_quant): BiasQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (export_handler): FINNQuantLinearHandler()\n",
       "    )\n",
       "    (5): QuantReLU(\n",
       "      (input_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (act_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "        (fused_activation_quant_proxy): FusedActivationQuantProxy(\n",
       "          (activation_impl): ReLU()\n",
       "          (tensor_quant): RescalingIntQuant(\n",
       "            (int_quant): IntQuant(\n",
       "              (float_to_int_impl): RoundSte()\n",
       "              (tensor_clamp_impl): TensorClamp()\n",
       "              (delay_wrapper): DelayWrapper(\n",
       "                (delay_impl): _NoDelay()\n",
       "              )\n",
       "            )\n",
       "            (scaling_impl): ParameterFromRuntimeStatsScaling(\n",
       "              (stats_input_view_shape_impl): OverTensorView()\n",
       "              (stats): _Stats(\n",
       "                (stats_impl): AbsPercentile()\n",
       "              )\n",
       "              (restrict_clamp_scaling): _RestrictClampValue(\n",
       "                (clamp_min_ste): Identity()\n",
       "                (restrict_value_impl): FloatRestrictValue()\n",
       "              )\n",
       "              (restrict_inplace_preprocess): Identity()\n",
       "              (restrict_preprocess): Identity()\n",
       "            )\n",
       "            (int_scaling_impl): IntScaling()\n",
       "            (zero_point_impl): ZeroZeroPoint(\n",
       "              (zero_point): StatelessBuffer()\n",
       "            )\n",
       "            (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "              (bit_width): StatelessBuffer()\n",
       "            )\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (export_handler): FINNQuantReLUHandler()\n",
       "    )\n",
       "    (6): QuantLinear(\n",
       "      in_features=128, out_features=400, bias=True\n",
       "      (input_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (output_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (weight_quant): WeightQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "        (tensor_quant): RescalingIntQuant(\n",
       "          (int_quant): IntQuant(\n",
       "            (float_to_int_impl): RoundSte()\n",
       "            (tensor_clamp_impl): TensorClampSte()\n",
       "            (delay_wrapper): DelayWrapper(\n",
       "              (delay_impl): _NoDelay()\n",
       "            )\n",
       "          )\n",
       "          (scaling_impl): StatsFromParameterScaling(\n",
       "            (parameter_list_stats): _ParameterListStats(\n",
       "              (first_tracked_param): _ViewParameterWrapper(\n",
       "                (view_shape_impl): OverTensorView()\n",
       "              )\n",
       "              (stats): _Stats(\n",
       "                (stats_impl): AbsMax()\n",
       "              )\n",
       "            )\n",
       "            (stats_scaling_impl): _StatsScaling(\n",
       "              (affine_rescaling): Identity()\n",
       "              (restrict_clamp_scaling): _RestrictClampValue(\n",
       "                (clamp_min_ste): Identity()\n",
       "                (restrict_value_impl): FloatRestrictValue()\n",
       "              )\n",
       "              (restrict_scaling_pre): Identity()\n",
       "            )\n",
       "          )\n",
       "          (int_scaling_impl): IntScaling()\n",
       "          (zero_point_impl): ZeroZeroPoint(\n",
       "            (zero_point): StatelessBuffer()\n",
       "          )\n",
       "          (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "            (bit_width): StatelessBuffer()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (bias_quant): BiasQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (export_handler): FINNQuantLinearHandler()\n",
       "    )\n",
       "    (7): QuantReLU(\n",
       "      (input_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "      )\n",
       "      (act_quant): ActQuantProxyFromInjector(\n",
       "        (_zero_hw_sentinel): StatelessBuffer()\n",
       "        (fused_activation_quant_proxy): FusedActivationQuantProxy(\n",
       "          (activation_impl): ReLU()\n",
       "          (tensor_quant): RescalingIntQuant(\n",
       "            (int_quant): IntQuant(\n",
       "              (float_to_int_impl): RoundSte()\n",
       "              (tensor_clamp_impl): TensorClamp()\n",
       "              (delay_wrapper): DelayWrapper(\n",
       "                (delay_impl): _NoDelay()\n",
       "              )\n",
       "            )\n",
       "            (scaling_impl): ParameterFromRuntimeStatsScaling(\n",
       "              (stats_input_view_shape_impl): OverTensorView()\n",
       "              (stats): _Stats(\n",
       "                (stats_impl): AbsPercentile()\n",
       "              )\n",
       "              (restrict_clamp_scaling): _RestrictClampValue(\n",
       "                (clamp_min_ste): Identity()\n",
       "                (restrict_value_impl): FloatRestrictValue()\n",
       "              )\n",
       "              (restrict_inplace_preprocess): Identity()\n",
       "              (restrict_preprocess): Identity()\n",
       "            )\n",
       "            (int_scaling_impl): IntScaling()\n",
       "            (zero_point_impl): ZeroZeroPoint(\n",
       "              (zero_point): StatelessBuffer()\n",
       "            )\n",
       "            (msb_clamp_bit_width_impl): BitWidthConst(\n",
       "              (bit_width): StatelessBuffer()\n",
       "            )\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (export_handler): FINNQuantReLUHandler()\n",
       "    )\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_for_export"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "id": "ec23e0d3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Restructured Image by AE\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "outputs = model_for_export(inputs)\n",
    "outputs = outputs.detach().cpu()\n",
    "print('Restructured Image by AE')\n",
    "show_images(outputs)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "id": "119134ea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([0.0000e+00, 4.5646e-41, 0.0000e+00, 4.5646e-41, 1.5461e+34, 3.0925e-41,\n",
       "        1.5461e+34, 3.0925e-41, 0.0000e+00, 4.5644e-41, 1.5414e-44, 0.0000e+00,\n",
       "        0.0000e+00, 0.0000e+00, 1.1210e-44, 0.0000e+00, 0.0000e+00, 7.6301e-06,\n",
       "        0.0000e+00, 0.0000e+00, 1.2612e-44, 0.0000e+00, 0.0000e+00, 4.5644e-41,\n",
       "        2.1453e+30, 3.0925e-41,        nan,        nan, 0.0000e+00, 0.0000e+00,\n",
       "               nan,        nan, 0.0000e+00, 0.0000e+00, 1.4013e-45, 0.0000e+00,\n",
       "        1.0089e-43, 0.0000e+00, 1.5414e-44, 0.0000e+00, 0.0000e+00, 4.5644e-41,\n",
       "        2.1453e+30, 3.0925e-41, 9.8091e-45, 0.0000e+00, 3.9236e-44, 0.0000e+00,\n",
       "               nan,        nan, 0.0000e+00, 0.0000e+00, 1.4013e-45, 0.0000e+00,\n",
       "        2.0179e-43, 0.0000e+00, 1.5414e-44, 0.0000e+00, 0.0000e+00, 4.5644e-41,\n",
       "        2.1453e+30, 3.0925e-41, 0.0000e+00, 0.0000e+00, 3.9236e-44, 0.0000e+00,\n",
       "               nan,        nan, 1.4013e-44, 0.0000e+00, 1.4013e-45, 4.5644e-41,\n",
       "        3.0268e-43, 0.0000e+00, 1.1210e-44, 0.0000e+00, 0.0000e+00, 4.5644e-41,\n",
       "        2.1457e+30, 3.0925e-41, 2.8026e-45, 0.0000e+00, 3.6434e-44, 0.0000e+00,\n",
       "               nan,        nan, 0.0000e+00, 0.0000e+00, 1.4013e-45, 0.0000e+00,\n",
       "        4.0357e-43, 0.0000e+00, 1.5414e-44, 0.0000e+00, 0.0000e+00, 4.5644e-41,\n",
       "        2.1449e+30, 3.0925e-41, 0.0000e+00, 0.0000e+00, 2.2421e-44, 0.0000e+00,\n",
       "               nan,        nan, 0.0000e+00, 0.0000e+00, 1.4013e-45, 0.0000e+00,\n",
       "        5.0447e-43, 0.0000e+00, 9.8091e-45, 0.0000e+00, 0.0000e+00, 4.5644e-41,\n",
       "        2.1452e+30, 3.0925e-41, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,\n",
       "        0.0000e+00, 0.0000e+00, 4.2039e-45, 4.5644e-41, 0.0000e+00, 1.2752e-43,\n",
       "        6.0536e-43, 0.0000e+00, 4.2039e-45, 0.0000e+00, 0.0000e+00, 4.5644e-41,\n",
       "        2.1457e+30, 3.0925e-41, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,\n",
       "        0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 1.4013e-45, 0.0000e+00,\n",
       "        0.0000e+00, 0.0000e+00, 0.0000e+00, 4.5644e-41, 0.0000e+00, 0.0000e+00,\n",
       "        0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,\n",
       "        0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,\n",
       "        0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,\n",
       "        0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00,\n",
       "        0.0000e+00, 4.2039e-45, 0.0000e+00, 0.0000e+00, 0.0000e+00, 4.5644e-41,\n",
       "        0.0000e+00, 0.0000e+00, 6.0536e-43, 0.0000e+00, 2.8026e-45, 0.0000e+00,\n",
       "        0.0000e+00, 4.5644e-41, 2.1452e+30, 3.0925e-41,        nan,        nan,\n",
       "        3.6434e-44, 0.0000e+00,        nan,        nan, 6.7262e-44, 4.5644e-41,\n",
       "        1.4013e-45, 3.0925e-41, 1.2612e-44, 0.0000e+00, 0.0000e+00, 4.5644e-41,\n",
       "        2.1449e+30, 3.0925e-41, 9.8091e-45, 0.0000e+00, 1.4013e-45, 4.5644e-41,\n",
       "        2.0127e+30, 3.0925e-41, 0.0000e+00, 0.0000e+00, 1.4013e-45, 0.0000e+00,\n",
       "        2.8026e-45, 0.0000e+00, 0.0000e+00, 4.5644e-41, 1.4013e-45, 1.0510e-43,\n",
       "        0.0000e+00, 4.5644e-41, 0.0000e+00, 4.5644e-41, 0.0000e+00, 4.5644e-41,\n",
       "        0.0000e+00, 4.5644e-41, 0.0000e+00, 0.0000e+00, 1.1435e-42, 0.0000e+00,\n",
       "        2.8026e-45, 0.0000e+00, 0.0000e+00, 4.5644e-41, 2.0127e+30, 3.0925e-41,\n",
       "        1.3172e-43, 1.3312e-43, 5.6052e-45, 0.0000e+00, 2.8026e-45, 0.0000e+00,\n",
       "        0.0000e+00, 4.5644e-41, 1.4013e-45, 1.4433e-43, 0.0000e+00, 4.5644e-41,\n",
       "        0.0000e+00, 4.5644e-41, 0.0000e+00, 4.5644e-41, 0.0000e+00, 4.5644e-41,\n",
       "        0.0000e+00, 4.5644e-41, 1.3004e-42, 0.0000e+00, 1.5414e-44, 0.0000e+00,\n",
       "        0.0000e+00, 4.5644e-41, 2.0127e+30, 3.0925e-41, 1.4013e-45, 0.0000e+00,\n",
       "        1.7376e-43, 1.7516e-43, 1.7656e-43, 1.7796e-43, 0.0000e+00, 4.5644e-41,\n",
       "        1.4013e-45, 1.8357e-43, 1.4574e-42, 0.0000e+00, 4.2039e-45, 0.0000e+00,\n",
       "        2.4901e-42, 0.0000e+00, 0.0000e+00, 4.5646e-41, 0.0000e+00, 4.5646e-41,\n",
       "        0.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00, 2.0459e-43, 2.0599e-43,\n",
       "        1.4013e-45, 2.0879e-43, 2.1019e-43, 2.1160e-43, 2.1300e-43, 2.1440e-43,\n",
       "        2.1580e-43, 2.1720e-43, 2.1860e-43, 2.2000e-43, 2.2141e-43, 2.2281e-43,\n",
       "        2.2421e-43, 2.2561e-43, 2.2701e-43, 2.2841e-43, 2.2981e-43, 2.3121e-43,\n",
       "        2.3262e-43, 2.3402e-43, 2.3542e-43, 2.3682e-43, 2.3822e-43, 2.3962e-43,\n",
       "        2.4102e-43, 2.4242e-43, 2.4383e-43, 2.4523e-43, 2.4663e-43, 2.4803e-43,\n",
       "        2.4943e-43, 2.5083e-43, 2.5223e-43, 2.5364e-43, 2.5504e-43, 2.5644e-43,\n",
       "        2.5784e-43, 2.5924e-43, 2.6064e-43, 2.6204e-43, 2.6344e-43, 2.6485e-43,\n",
       "        2.6625e-43, 2.6765e-43, 2.6905e-43, 2.7045e-43, 2.7185e-43, 2.7325e-43,\n",
       "        2.7465e-43, 2.7606e-43, 2.7746e-43, 2.7886e-43, 2.8026e-43, 2.8166e-43,\n",
       "        2.8306e-43, 2.8446e-43, 2.8586e-43, 2.8727e-43, 2.8867e-43, 2.9007e-43,\n",
       "        2.9147e-43, 2.9287e-43, 2.9427e-43, 2.9567e-43, 2.9708e-43, 2.9848e-43,\n",
       "        2.9988e-43, 3.0128e-43, 3.0268e-43, 3.0408e-43, 3.0548e-43, 3.0688e-43,\n",
       "        3.0829e-43, 3.0969e-43, 3.1109e-43, 3.1249e-43, 3.1389e-43, 3.1529e-43,\n",
       "        3.1669e-43, 3.1809e-43, 3.1950e-43, 3.2090e-43, 0.0000e+00, 4.5644e-41,\n",
       "        0.0000e+00, 3.2650e-43, 3.2790e-43, 3.2931e-43, 3.3071e-43, 3.3211e-43,\n",
       "        3.3351e-43, 3.3491e-43, 3.3631e-43, 3.3771e-43, 3.3911e-43, 3.4052e-43,\n",
       "        3.4192e-43, 3.4332e-43, 3.4472e-43, 3.4612e-43, 3.4752e-43, 3.4892e-43,\n",
       "        3.5032e-43, 3.5173e-43, 3.5313e-43, 3.5453e-43])"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "outputs = model_for_export(inputs)\n",
    "outputs = outputs.detach().cpu()\n",
    "outputs[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "id": "86cf2fd2",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model saved to autoencoder_ln.onnx\n"
     ]
    }
   ],
   "source": [
    "import brevitas.onnx as bo\n",
    "from brevitas.quant_tensor import QuantTensor\n",
    "\n",
    "ready_model_filename = \"autoencoder_ln.onnx\"\n",
    "input_shape = (1, 400)\n",
    "# create a QuantTensor instance to mark input as bipolar during export\n",
    "input_a = np.random.randint(0, 255, size=input_shape).astype(np.float32)\n",
    "\n",
    "scale = 1.0\n",
    "input_t = torch.from_numpy(input_a * scale)\n",
    "input_qt = QuantTensor(\n",
    "    input_t, scale=torch.tensor(scale), bit_width=torch.tensor(8.0), signed=True\n",
    ")\n",
    "\n",
    "bo.export_finn_onnx(\n",
    "    model_for_export, export_path=ready_model_filename, input_t=input_qt\n",
    ")\n",
    "\n",
    "print(\"Model saved to %s\" % ready_model_filename)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "id": "47ee0648",
   "metadata": {},
   "outputs": [],
   "source": [
    "#import brevitas.onnx as bo\n",
    "\n",
    "#bo.export_finn_onnx(model, input_shape=(1,400), export_path='autoencoder_ln.onnx')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "id": "16e83c09",
   "metadata": {},
   "outputs": [],
   "source": [
    "from finn.util.basic import make_build_dir\n",
    "from finn.util.visualization import showInNetron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "id": "b7a8118a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import onnx\n",
    "from finn.util.test import get_test_model_trained\n",
    "import brevitas.onnx as bo\n",
    "from finn.core.modelwrapper import ModelWrapper\n",
    "from finn.transformation.infer_shapes import InferShapes\n",
    "from finn.transformation.fold_constants import FoldConstants\n",
    "from finn.transformation.general import GiveReadableTensorNames, GiveUniqueNodeNames, RemoveStaticGraphInputs\n",
    "\n",
    "model = ModelWrapper(\"autoencoder_ln.onnx\")\n",
    "model = model.transform(InferShapes())\n",
    "model = model.transform(FoldConstants())\n",
    "model = model.transform(GiveUniqueNodeNames())\n",
    "model = model.transform(GiveReadableTensorNames())\n",
    "model = model.transform(RemoveStaticGraphInputs())\n",
    "model.save(\"autoencoder_ln_w1a1_tidy.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "2f17cf61",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Serving 'autoencoder_ln_w1a1_tidy.onnx' at http://0.0.0.0:8081\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"100%\"\n",
       "            height=\"400\"\n",
       "            src=\"http://localhost:8081/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x7f3da6d63e80>"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "showInNetron(\"autoencoder_ln_w1a1_tidy.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "id": "d9e73158",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/workspace/finn-base/src/finn/transformation/infer_data_layouts.py:124: UserWarning: Assuming 2D input is NC\n",
      "  warnings.warn(\"Assuming 2D input is NC\")\n"
     ]
    }
   ],
   "source": [
    "from finn.transformation.streamline import Streamline\n",
    "from finn.transformation.lower_convs_to_matmul import LowerConvsToMatMul\n",
    "from finn.transformation.bipolar_to_xnor import ConvertBipolarMatMulToXnorPopcount\n",
    "import finn.transformation.streamline.absorb as absorb\n",
    "from finn.transformation.streamline.reorder import MakeMaxPoolNHWC, MoveScalarLinearPastInvariants\n",
    "from finn.transformation.infer_data_layouts import InferDataLayouts\n",
    "from finn.transformation.general import RemoveUnusedTensors\n",
    "\n",
    "model = ModelWrapper(\"autoencoder_ln_w1a1_tidy.onnx\")\n",
    "model = model.transform(MoveScalarLinearPastInvariants())\n",
    "model = model.transform(Streamline())\n",
    "model = model.transform(LowerConvsToMatMul())\n",
    "model = model.transform(MakeMaxPoolNHWC())\n",
    "model = model.transform(absorb.AbsorbTransposeIntoMultiThreshold())\n",
    "\n",
    "model = model.transform(MakeMaxPoolNHWC())\n",
    "model = model.transform(absorb.AbsorbConsecutiveTransposes())\n",
    "\n",
    "model = model.transform(ConvertBipolarMatMulToXnorPopcount())\n",
    "model = model.transform(Streamline())\n",
    "# absorb final add-mul nodes into TopK\n",
    "model = model.transform(absorb.AbsorbScalarMulAddIntoTopK())\n",
    "model = model.transform(InferDataLayouts())\n",
    "model = model.transform(RemoveUnusedTensors())\n",
    "model.save(\"autoencoder_ln_w1a1_streamlined.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "id": "a3ca0ad3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Stopping http://0.0.0.0:8081\n",
      "Serving 'autoencoder_ln_w1a1_streamlined.onnx' at http://0.0.0.0:8081\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"100%\"\n",
       "            height=\"400\"\n",
       "            src=\"http://localhost:8081/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x7f3da6d637f0>"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "showInNetron(\"autoencoder_ln_w1a1_streamlined.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "id": "e1649794",
   "metadata": {},
   "outputs": [],
   "source": [
    "import finn.transformation.fpgadataflow.convert_to_hls_layers as to_hls\n",
    "from finn.transformation.fpgadataflow.create_dataflow_partition import (\n",
    "    CreateDataflowPartition,\n",
    ")\n",
    "from finn.transformation.move_reshape import RemoveCNVtoFCFlatten\n",
    "from finn.custom_op.registry import getCustomOp\n",
    "from finn.transformation.infer_data_layouts import InferDataLayouts\n",
    "\n",
    "# choose the memory mode for the MVTU units, decoupled or const\n",
    "mem_mode = \"decoupled\"\n",
    "\n",
    "model = ModelWrapper(\"autoencoder_ln_w1a1_streamlined.onnx\")\n",
    "model = model.transform(to_hls.InferBinaryStreamingFCLayer(mem_mode))\n",
    "model = model.transform(to_hls.InferQuantizedStreamingFCLayer(mem_mode))\n",
    "# TopK to LabelSelect\n",
    "model = model.transform(to_hls.InferLabelSelectLayer())\n",
    "# input quantization (if any) to standalone thresholding\n",
    "model = model.transform(to_hls.InferThresholdingLayer())\n",
    "model = model.transform(to_hls.InferConvInpGen())\n",
    "model = model.transform(to_hls.InferStreamingMaxPool())\n",
    "# get rid of Reshape(-1, 1) operation between hlslib nodes\n",
    "model = model.transform(RemoveCNVtoFCFlatten())\n",
    "# get rid of Tranpose -> Tranpose identity seq\n",
    "model = model.transform(absorb.AbsorbConsecutiveTransposes())\n",
    "# infer tensor data layouts\n",
    "model = model.transform(InferDataLayouts())\n",
    "\n",
    "model.save(\"autoencoder_ttt.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "id": "4dfcab58",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Stopping http://0.0.0.0:8081\n",
      "Serving 'autoencoder_ttt.onnx' at http://0.0.0.0:8081\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"100%\"\n",
       "            height=\"400\"\n",
       "            src=\"http://localhost:8081/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x7f3dbb61b6a0>"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "showInNetron(\"autoencoder_ttt.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "id": "da1b0c94",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Stopping http://0.0.0.0:8081\n",
      "Serving 'autoencoder_ln_w1a1_dataflow_parent.onnx' at http://0.0.0.0:8081\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"100%\"\n",
       "            height=\"400\"\n",
       "            src=\"http://localhost:8081/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x7f3da6bf85e0>"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "parent_model = model.transform(CreateDataflowPartition())\n",
    "parent_model.save(\"autoencoder_ln_w1a1_dataflow_parent.onnx\")\n",
    "showInNetron(\"autoencoder_ln_w1a1_dataflow_parent.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "id": "abdc782d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Stopping http://0.0.0.0:8081\n",
      "Serving 'autoencoder_ttt.onnx' at http://0.0.0.0:8081\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"100%\"\n",
       "            height=\"400\"\n",
       "            src=\"http://localhost:8081/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x7f3dc20cec40>"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.save(\"autoencoder_ttt.onnx\")\n",
    "showInNetron(\"autoencoder_ttt.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "id": "b784b98b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/ocu8927/finn/build/dataflow_partition_6ftuce9l/partition_0.onnx\n"
     ]
    }
   ],
   "source": [
    "sdp_node = parent_model.get_nodes_by_op_type(\"StreamingDataflowPartition\")[0]\n",
    "sdp_node = getCustomOp(sdp_node)\n",
    "dataflow_model_filename = sdp_node.get_nodeattr(\"model\")\n",
    "print(dataflow_model_filename)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "id": "81e13c85",
   "metadata": {},
   "outputs": [],
   "source": [
    "# save the dataflow partition with a different name for easier access\n",
    "dataflow_model = ModelWrapper(dataflow_model_filename)\n",
    "dataflow_model.save(\"autoencoder_ln_w1a1_dataflow_model.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "id": "f33038ce",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Stopping http://0.0.0.0:8081\n",
      "Serving 'autoencoder_ln_w1a1_dataflow_model.onnx' at http://0.0.0.0:8081\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"100%\"\n",
       "            height=\"400\"\n",
       "            src=\"http://localhost:8081/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x7f3da6e50040>"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "showInNetron(\"autoencoder_ln_w1a1_dataflow_model.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "id": "3f3da941",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = ModelWrapper(\"autoencoder_ln_w1a1_dataflow_model.onnx\")\n",
    "\n",
    "#model = ModelWrapper(\"autoencoder_ln_w1a1_streamlined.onnx\")\n",
    "fc_layers = model.get_nodes_by_op_type(\"StreamingFCLayer_Batch\")\n",
    "# each tuple is (PE, SIMD, in_fifo_depth) for a layer\n",
    "folding = [\n",
    "\n",
    "    (4, 16, 128),\n",
    "    (16, 4, 128),\n",
    "    (40, 32, 128),\n",
    "\n",
    "]\n",
    "for fcl, (pe, simd, ififodepth) in zip(fc_layers, folding):\n",
    "    fcl_inst = getCustomOp(fcl)\n",
    "    fcl_inst.set_nodeattr(\"PE\", pe)\n",
    "    fcl_inst.set_nodeattr(\"SIMD\", simd)\n",
    "    fcl_inst.set_nodeattr(\"inFIFODepth\", ififodepth)\n",
    "\n",
    "# use same SIMD values for the sliding window operators\n",
    "swg_layers = model.get_nodes_by_op_type(\"ConvolutionInputGenerator\")\n",
    "for i in range(len(swg_layers)):\n",
    "    swg_inst = getCustomOp(swg_layers[i])\n",
    "    simd = folding[i][1]\n",
    "    swg_inst.set_nodeattr(\"SIMD\", simd)\n",
    "\n",
    "model = model.transform(GiveUniqueNodeNames())\n",
    "model.save(\"autoencoder_ln_w1a1_folded.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c06ab4c0",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "id": "62f02772",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Stopping http://0.0.0.0:8081\n",
      "Serving 'autoencoder_ln_w1a1_folded.onnx' at http://0.0.0.0:8081\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"100%\"\n",
       "            height=\"400\"\n",
       "            src=\"http://localhost:8081/\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x7f3dc1efb3a0>"
      ]
     },
     "execution_count": 117,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "showInNetron(\"autoencoder_ln_w1a1_folded.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "1910ecbc",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/workspace/finn/src/finn/transformation/fpgadataflow/floorplan.py:107: UserWarning: 8 nodes have no entry in the provided floorplan, SLR was set to -1\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-73-c530397cffb2>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0mmodel\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mModelWrapper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"autoencoder_ln_w1a1_folded.onnx\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;31m#model = ModelWrapper(\"autoencoder_ln_w1a1_dataflow_model.onnx\")\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m \u001b[0mmodel\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtransform\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mZynqBuild\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mplatform\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtest_pynq_board\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mperiod_ns\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtarget_clk_ns\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[0m\u001b[1;32m      8\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msave\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"autoencoder_ln_w1a1_synth.onnx\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/workspace/finn-base/src/finn/core/modelwrapper.py\u001b[0m in \u001b[0;36mtransform\u001b[0;34m(self, transformation, make_deepcopy, cleanup, fix_float64)\u001b[0m\n\u001b[1;32m    139\u001b[0m         \u001b[0mmodel_was_changed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    140\u001b[0m         \u001b[0;32mwhile\u001b[0m \u001b[0mmodel_was_changed\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 141\u001b[0;31m             (transformed_model, model_was_changed) = transformation.apply(\n\u001b[0m\u001b[1;32m    142\u001b[0m                 \u001b[0mtransformed_model\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    143\u001b[0m             )\n",
      "\u001b[0;32m/workspace/finn/src/finn/transformation/fpgadataflow/make_zynq_proj.py\u001b[0m in \u001b[0;36mapply\u001b[0;34m(self, model)\u001b[0m\n\u001b[1;32m    345\u001b[0m                 \u001b[0mPrepareIP\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfpga_part\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mperiod_ns\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    346\u001b[0m             )\n\u001b[0;32m--> 347\u001b[0;31m             \u001b[0mkernel_model\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mkernel_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtransform\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mHLSSynthIP\u001b[0m\u001b[0;34m(\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[0m\u001b[1;32m    348\u001b[0m             kernel_model = kernel_model.transform(\n\u001b[1;32m    349\u001b[0m                 CreateStitchedIP(\n",
      "\u001b[0;32m/workspace/finn-base/src/finn/core/modelwrapper.py\u001b[0m in \u001b[0;36mtransform\u001b[0;34m(self, transformation, make_deepcopy, cleanup, fix_float64)\u001b[0m\n\u001b[1;32m    139\u001b[0m         \u001b[0mmodel_was_changed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    140\u001b[0m         \u001b[0;32mwhile\u001b[0m \u001b[0mmodel_was_changed\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 141\u001b[0;31m             (transformed_model, model_was_changed) = transformation.apply(\n\u001b[0m\u001b[1;32m    142\u001b[0m                 \u001b[0mtransformed_model\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    143\u001b[0m             )\n",
      "\u001b[0;32m/workspace/finn-base/src/finn/transformation/base.py\u001b[0m in \u001b[0;36mapply\u001b[0;34m(self, model)\u001b[0m\n\u001b[1;32m    103\u001b[0m         \u001b[0;31m# Execute transformation in parallel\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    104\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mmp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mPool\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_num_workers\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mp\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 105\u001b[0;31m             \u001b[0mnew_nodes_and_bool\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmap\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mapplyNodeLocal\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mold_nodes\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mchunksize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1\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    106\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    107\u001b[0m         \u001b[0;31m# extract nodes and check if the transformation needs to run again\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.8/multiprocessing/pool.py\u001b[0m in \u001b[0;36mmap\u001b[0;34m(self, func, iterable, chunksize)\u001b[0m\n\u001b[1;32m    362\u001b[0m         \u001b[0;32min\u001b[0m \u001b[0ma\u001b[0m \u001b[0mlist\u001b[0m \u001b[0mthat\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mreturned\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    363\u001b[0m         '''\n\u001b[0;32m--> 364\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_map_async\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0miterable\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmapstar\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mchunksize\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget\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[0m\u001b[1;32m    365\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    366\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mstarmap\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfunc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0miterable\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mchunksize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\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[0;32m/opt/conda/lib/python3.8/multiprocessing/pool.py\u001b[0m in \u001b[0;36mget\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    763\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    764\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtimeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\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[0;32m--> 765\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\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    766\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mready\u001b[0m\u001b[0;34m(\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    767\u001b[0m             \u001b[0;32mraise\u001b[0m \u001b[0mTimeoutError\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.8/multiprocessing/pool.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    760\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    761\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtimeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\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[0;32m--> 762\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_event\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\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    763\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    764\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtimeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\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[0;32m/opt/conda/lib/python3.8/threading.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    556\u001b[0m             \u001b[0msignaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_flag\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    557\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0msignaled\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 558\u001b[0;31m                 \u001b[0msignaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_cond\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\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    559\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0msignaled\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    560\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/opt/conda/lib/python3.8/threading.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    300\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m    \u001b[0;31m# restore state no matter what (e.g., KeyboardInterrupt)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    301\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mtimeout\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 302\u001b[0;31m                 \u001b[0mwaiter\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0macquire\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[0m\u001b[1;32m    303\u001b[0m                 \u001b[0mgotit\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    304\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "test_pynq_board = \"Pynq-Z2\"\n",
    "target_clk_ns = 10\n",
    "\n",
    "from finn.transformation.fpgadataflow.make_zynq_proj import ZynqBuild\n",
    "model = ModelWrapper(\"autoencoder_ln_w1a1_folded.onnx\")\n",
    "#model = ModelWrapper(\"autoencoder_ln_w1a1_dataflow_model.onnx\")\n",
    "model = model.transform(ZynqBuild(platform = test_pynq_board, period_ns = target_clk_ns))\n",
    "model.save(\"autoencoder_ln_w1a1_synth.onnx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "70809536",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5ecb6c66",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}