{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"After reading the great notebook [tensorflow-lstm-model-training-tpu](https://www.kaggle.com/code/rsmits/tensorflow-lstm-model-training-tpu), I decide to try lstm myself, but I'm not familiar with tensorflow so I decide to rewrite it in pytorch. \n\nThen I find my lstm can't learn any thing😭. **In short, I find the gradient of the model is 0**, I can't tell what's wrong with my code so if you have any suggestion, I will apprecate it very much👀\n\n","metadata":{}},{"cell_type":"markdown","source":"Here is what I get when I train locally(I don't train in kaggle enviroments as it cost a lot time)\n\n![image.png](attachment:8645b273-0efb-47bb-ab30-f7a02b6795b7.png)","metadata":{},"attachments":{"8645b273-0efb-47bb-ab30-f7a02b6795b7.png":{"image/png":"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"}}},{"cell_type":"markdown","source":"## How I do and what I meet?\n\nHow I do to convert tensorflow version to pytorch version:\n1. GRU func in pytorch and tensorflow is quiet different, so I make adjustment accordingly\n2. I use [torch.nn.CrossEntropyLoss](https://pytorch.org/docs/stable/generated/torch.nn.CrossEntropyLoss.html?highlight=crossentropyloss#torch.nn.CrossEntropyLoss) to replace `sparse_categorical_crossentropy`\n. I use [torch.nn.utils.rnn.pack_padded_sequence](https://pytorch.org/docs/stable/generated/torch.nn.utils.rnn.pack_padded_sequence.html?highlight=pack_padded_sequence#torch.nn.utils.rnn.pack_padded_sequence) to instead `keras.layers.Masking` and add 1 to col: auxiliary as the default padding value is -1 according to the data preprocessor parts\n\nwhat I meet and my current adjustment:\n1. **the loss val I get descend very slow**, I find the mean and min of loss is \"-6\" which is not good because the loss func contains softmax and **the the grad is not unsuprisly almost 0**! After I trained many epoches, loss value is still above 6. Before training, the origin loss val is nearly 6.35. \n2. Then I try to add \"Batch Normalization layer\" after \"lstm layer\" and use \"leakyrelu\" to instead \"relu\" between two \"linear layer\", it has had some effect and the loss val drop to 5.5 after 15 epoches trains(I mean the lowest loss val is 5.5 whatever the learning rate I use, the difference is only how many epoches it cost), and then the val rapid rise to 7, 10 and final to nan which also looks strange to me.😂\n\n![image.png](attachment:ca4aa43f-e14b-4583-a4f3-71fea70318b1.png)\n\n![image.png](attachment:fa868a1d-57c6-48aa-b8a0-9aac36655ed8.png)\n","metadata":{},"attachments":{"ca4aa43f-e14b-4583-a4f3-71fea70318b1.png":{"image/png":"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"},"fa868a1d-57c6-48aa-b8a0-9aac36655ed8.png":{"image/png":"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"}}},{"cell_type":"code","source":"import os\nimport gc\n\nimport numpy as np\nfrom tqdm import tqdm\nimport pytorch_lightning as pl\n\njoin = os.path.join\ndirname = os.path.dirname\nabspath = os.path.abspath","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:23:51.798640Z","iopub.execute_input":"2023-04-02T06:23:51.799438Z","iopub.status.idle":"2023-04-02T06:24:14.616430Z","shell.execute_reply.started":"2023-04-02T06:23:51.799390Z","shell.execute_reply":"2023-04-02T06:24:14.614463Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import warnings\nwarnings.filterwarnings('ignore')","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:24:17.574942Z","iopub.execute_input":"2023-04-02T06:24:17.575569Z","iopub.status.idle":"2023-04-02T06:24:17.583523Z","shell.execute_reply.started":"2023-04-02T06:24:17.575510Z","shell.execute_reply":"2023-04-02T06:24:17.581482Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"\n# Training\nvalidation_files_amount = 1\ndata_new_load_interval = 6      # Local Training: None\ntrain_files_delta = 2           # Local Training: None\nepochs = 75                     # Local Training: 30\nbatch_size = 8192               # Local Training: 2048\nlearning_rate = 0.0022          # Local Training: 0.0005\nverbose = 0\n\n# Training Batches\ntrain_batch_id_min = 100\ntrain_batch_id_max = 190  # 190\ntrain_batch_ids = [*range(train_batch_id_min, train_batch_id_max+1)]\nnp.random.shuffle(train_batch_ids)\nprint(train_batch_ids)\n\n# Model Parameters\npulse_count = 96\nfeature_count = 6\nlstm_units = 192\nbin_num = 24","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:24:20.309793Z","iopub.execute_input":"2023-04-02T06:24:20.310393Z","iopub.status.idle":"2023-04-02T06:24:20.328275Z","shell.execute_reply.started":"2023-04-02T06:24:20.310342Z","shell.execute_reply":"2023-04-02T06:24:20.325668Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Difference: GRU\n\n[pytorch GRU's docs](https://pytorch.org/docs/stable/generated/torch.nn.GRU.html?highlight=gru#torch.nn.GRU)\n\n[tf.keras.layers.GRU's docs](https://tensorflow.google.cn/api_docs/python/tf/keras/layers/GRU)\n\nas we can see in the docs, GRU in pytorch and tensorflow is quiet difference, tensorflow version is more flexiable:\n1. in tensorflow, we can define the act_func and recurrent_act_func, which we can't define directly in pytorch, but their default setting are the same\n2. in pytorch, return sequences is default however in tensorflow it's optional\n\nfollowing is the model structure","metadata":{}},{"cell_type":"code","source":"batch_size = 8192\ninput_token = 6\nlstm_units = 192\nbin_num = 24\nfeature_count = 6\npulse_count = 96","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:25:18.002597Z","iopub.execute_input":"2023-04-02T06:25:18.003194Z","iopub.status.idle":"2023-04-02T06:25:18.012681Z","shell.execute_reply.started":"2023-04-02T06:25:18.003141Z","shell.execute_reply":"2023-04-02T06:25:18.010615Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import torch\nimport torch.nn as nn\nfrom torch.nn import GRU\nfrom torchinfo import summary\nfrom torch.nn.utils.rnn import pad_sequence, pack_padded_sequence, pad_packed_sequence","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:26:35.387984Z","iopub.execute_input":"2023-04-02T06:26:35.388585Z","iopub.status.idle":"2023-04-02T06:26:35.396801Z","shell.execute_reply.started":"2023-04-02T06:26:35.388538Z","shell.execute_reply":"2023-04-02T06:26:35.395181Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import tensorflow as tf\nfrom tensorflow.keras import Sequential as tf_Sequential\nfrom tensorflow.keras.layers import Bidirectional as tf_Bidirectional\nfrom tensorflow.keras.layers import GRU as tf_GRU","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:26:18.704302Z","iopub.execute_input":"2023-04-02T06:26:18.704994Z","iopub.status.idle":"2023-04-02T06:26:18.719491Z","shell.execute_reply.started":"2023-04-02T06:26:18.704942Z","shell.execute_reply":"2023-04-02T06:26:18.717059Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model_GRUparts = torch.nn.Sequential(torch.nn.GRU(feature_count, lstm_units, bidirectional=True, num_layers=3, batch_first=True))\nsummary(model_GRUparts, input_size=(batch_size, lstm_units, feature_count))","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:26:37.291706Z","iopub.execute_input":"2023-04-02T06:26:37.292412Z","iopub.status.idle":"2023-04-02T06:28:08.556477Z","shell.execute_reply.started":"2023-04-02T06:26:37.292345Z","shell.execute_reply":"2023-04-02T06:28:08.554922Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model_Linearparts = torch.nn.Sequential(torch.nn.Linear(384, 256), torch.nn.ReLU(), torch.nn.Linear(256, bin_num**2))\nsummary(model_Linearparts, input_size=(batch_size, 384))","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:28:11.397419Z","iopub.execute_input":"2023-04-02T06:28:11.398890Z","iopub.status.idle":"2023-04-02T06:28:11.498354Z","shell.execute_reply.started":"2023-04-02T06:28:11.398812Z","shell.execute_reply":"2023-04-02T06:28:11.497342Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"tf_model = tf_Sequential()\ntf_model.add(tf_Bidirectional(tf_GRU(lstm_units, return_sequences=True, activation='tanh')))\ntf_model.add(tf_Bidirectional(tf_GRU(lstm_units, return_sequences=True, activation='tanh')))\ntf_model.add(tf_Bidirectional(tf_GRU(lstm_units, activation='tanh')))\ntf_model.add(tf.keras.layers.Dense(256, activation = 'relu'))\ntf_model.add(tf.keras.layers.Dense(bin_num**2, activation = 'softmax'))\ntf_model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=1e-3), loss='sparse_categorical_crossentropy')\ntf_model.call(inputs=tf.random.normal(shape=(batch_size, 100, input_token)))\ntf_model.summary()","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:28:13.542421Z","iopub.execute_input":"2023-04-02T06:28:13.542917Z","iopub.status.idle":"2023-04-02T06:29:02.372700Z","shell.execute_reply.started":"2023-04-02T06:28:13.542873Z","shell.execute_reply":"2023-04-02T06:29:02.370956Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Difference: loss func\n\nin pytorch, I use [torch.nn.CrossEntropyLoss](https://pytorch.org/docs/stable/generated/torch.nn.CrossEntropyLoss.html?highlight=crossentropyloss#torch.nn.CrossEntropyLoss) to instead `sparse_categorical_crossentropy` in tensorflow. So I don't add final activation func","metadata":{}},{"cell_type":"markdown","source":"## Difference: mask func\n\nI use [torch.nn.utils.rnn.pack_padded_sequence](https://pytorch.org/docs/stable/generated/torch.nn.utils.rnn.pack_padded_sequence.html?highlight=pack_padded_sequence#torch.nn.utils.rnn.pack_padded_sequence) to instead `keras.layers.Masking`","metadata":{}},{"cell_type":"code","source":"from tqdm import tqdm\nimport numpy as np\nimport os\nimport gc\nimport random","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:30:39.503928Z","iopub.execute_input":"2023-04-02T06:30:39.504995Z","iopub.status.idle":"2023-04-02T06:30:39.511503Z","shell.execute_reply.started":"2023-04-02T06:30:39.504942Z","shell.execute_reply":"2023-04-02T06:30:39.509993Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Set Seed\nseed = 4242\ntorch.manual_seed(seed)\nrandom.seed(seed)\nnp.random.seed(seed)\n","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:30:41.814401Z","iopub.execute_input":"2023-04-02T06:30:41.815312Z","iopub.status.idle":"2023-04-02T06:30:41.822962Z","shell.execute_reply.started":"2023-04-02T06:30:41.815261Z","shell.execute_reply":"2023-04-02T06:30:41.821771Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Prepare: Create 'a/z' edge","metadata":{}},{"cell_type":"code","source":"# Data\nbase_dir = \"/kaggle/input/lstmicecubesdata/\"\nfile_format = base_dir + 'pp_mpc96_n7_batch_{batch_id:d}.npz'","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:30:45.525353Z","iopub.execute_input":"2023-04-02T06:30:45.525857Z","iopub.status.idle":"2023-04-02T06:30:45.531344Z","shell.execute_reply.started":"2023-04-02T06:30:45.525815Z","shell.execute_reply":"2023-04-02T06:30:45.530332Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Create Azimuth Edges\nazimuth_edges = np.linspace(0, 2 * np.pi, bin_num + 1)\nprint(azimuth_edges)\n\n# Create Zenith Edges\nzenith_edges = []\nzenith_edges.append(0)\nfor bin_idx in range(1, bin_num):\n    zenith_edges.append(np.arccos(np.cos(zenith_edges[-1]) - 2 / (bin_num)))\nzenith_edges.append(np.pi)\nzenith_edges = np.array(zenith_edges)\nprint(zenith_edges)\n\n\nangle_bin_zenith0 = np.tile(zenith_edges[:-1], bin_num)\nangle_bin_zenith1 = np.tile(zenith_edges[1:], bin_num)\nangle_bin_azimuth0 = np.repeat(azimuth_edges[:-1], bin_num)\nangle_bin_azimuth1 = np.repeat(azimuth_edges[1:], bin_num)\n\nangle_bin_area = (angle_bin_azimuth1 - angle_bin_azimuth0) * (np.cos(angle_bin_zenith0) - np.cos(angle_bin_zenith1))\nangle_bin_vector_sum_x = (np.sin(angle_bin_azimuth1) - np.sin(angle_bin_azimuth0)) * ((angle_bin_zenith1 - angle_bin_zenith0) / 2 - (np.sin(2 * angle_bin_zenith1) - np.sin(2 * angle_bin_zenith0)) / 4)\nangle_bin_vector_sum_y = (np.cos(angle_bin_azimuth0) - np.cos(angle_bin_azimuth1)) * ((angle_bin_zenith1 - angle_bin_zenith0) / 2 - (np.sin(2 * angle_bin_zenith1) - np.sin(2 * angle_bin_zenith0)) / 4)\nangle_bin_vector_sum_z = (angle_bin_azimuth1 - angle_bin_azimuth0) * ((np.cos(2 * angle_bin_zenith0) - np.cos(2 * angle_bin_zenith1)) / 4)\n\nangle_bin_vector_mean_x = angle_bin_vector_sum_x / angle_bin_area\nangle_bin_vector_mean_y = angle_bin_vector_sum_y / angle_bin_area\nangle_bin_vector_mean_z = angle_bin_vector_sum_z / angle_bin_area\n\nangle_bin_vector = np.zeros((1, bin_num * bin_num, 3))\nangle_bin_vector[:, :, 0] = angle_bin_vector_mean_x\nangle_bin_vector[:, :, 1] = angle_bin_vector_mean_y\nangle_bin_vector[:, :, 2] = angle_bin_vector_mean_z\n","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:30:50.594348Z","iopub.execute_input":"2023-04-02T06:30:50.595570Z","iopub.status.idle":"2023-04-02T06:30:50.615014Z","shell.execute_reply.started":"2023-04-02T06:30:50.595521Z","shell.execute_reply":"2023-04-02T06:30:50.613418Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## DataSet&DataLoader","metadata":{}},{"cell_type":"code","source":"from torch.utils.data import Dataset, DataLoader","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:30:54.494661Z","iopub.execute_input":"2023-04-02T06:30:54.496331Z","iopub.status.idle":"2023-04-02T06:30:54.501802Z","shell.execute_reply.started":"2023-04-02T06:30:54.496275Z","shell.execute_reply":"2023-04-02T06:30:54.500817Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"\nclass LSTMDataset(Dataset):\n    def __init__(self, start_batch, mode='train'):\n        \n        print(\"Processing Training Data...\")\n        \n        self.mode = mode\n        if self.mode == 'train':\n            ids_use = train_batch_ids[start_batch:]\n        elif self.mode == 'valid':\n            ids_use = train_batch_ids[:start_batch]\n        else:\n            raise NotImplementedError\n        print(ids_use)\n        \n        self.x = None\n        \n        # Loop\n        for batch_id in tqdm(ids_use):\n            train_data_file = np.load(file_format.format(batch_id=batch_id))\n            \n            if self.x is None:\n                self.x = train_data_file[\"x\"][:, :, [0, 1, 2, 3, 4, 5]]\n                self.y = train_data_file[\"y\"]\n            else:\n                self.x = np.append(self.x, train_data_file[\"x\"][:, :, [0, 1, 2, 3, 4, 5]], axis=0)\n                self.y = np.append(self.y, train_data_file[\"y\"], axis=0)\n            \n            train_data_file.close()\n            del train_data_file\n            _ = gc.collect()\n        \n        # Normalize data\n        self.x = self._normalize_data(self.x)\n        \n        # Shape Summary\n        print(self.x.shape)\n        \n        # Output Encoding\n        if self.mode == 'train':\n            y_anglecode = self._y_to_angle_code(self.y)\n            self.y_code_proced = y_anglecode\n        elif self.mode == 'valid':\n            self.y_code_proced = self.y\n        else:\n            raise NotImplementedError\n    \n    def __len__(self):\n        return len(self.x)\n    \n    def __getitem__(self, idx):\n        \n        batch_x = self.x[idx, :]\n        if self.mode == 'train':\n            batch_y = self.y_code_proced[idx]\n        elif self.mode == 'valid':\n            batch_y = self.y_code_proced[idx, :]\n        else:\n            raise NotImplementedError\n        \n        return (batch_x, batch_y)\n    \n    @staticmethod\n    def _y_to_angle_code(batch_y):\n        azimuth_code = (batch_y[:, 0] > azimuth_edges[1:].reshape((-1, 1))).sum(axis=0)\n        zenith_code = (batch_y[:, 1] > zenith_edges[1:].reshape((-1, 1))).sum(axis=0)\n        angle_code = bin_num * azimuth_code + zenith_code\n        \n        return angle_code\n    \n    @staticmethod\n    def _normalize_data(data):\n        data[:, :, 0] /= 1000  # time\n        data[:, :, 1] /= 300  # charge\n        data[:, :, 3:] /= 600  # space\n        \n        return data","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:30:57.446505Z","iopub.execute_input":"2023-04-02T06:30:57.446964Z","iopub.status.idle":"2023-04-02T06:30:57.464760Z","shell.execute_reply.started":"2023-04-02T06:30:57.446927Z","shell.execute_reply":"2023-04-02T06:30:57.463171Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class LSTMDataLoader(pl.LightningDataModule):\n    def __init__(self, train_dataset, valid_dataset, batch_size=512):\n        super(LSTMDataLoader, self).__init__()\n        self.train_dataset = train_dataset\n        self.val_dataset = valid_dataset\n        \n        self.batch_size = batch_size\n    \n    def train_dataloader(self):\n        return DataLoader(self.train_dataset, batch_size=self.batch_size, shuffle=True)\n    \n    def val_dataloader(self):\n        return DataLoader(self.val_dataset, batch_size=self.batch_size, shuffle=False)\n    \n    # def test_dataloader(self):\n    #     return ''\n","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:31:04.514610Z","iopub.execute_input":"2023-04-02T06:31:04.515005Z","iopub.status.idle":"2023-04-02T06:31:04.525417Z","shell.execute_reply.started":"2023-04-02T06:31:04.514973Z","shell.execute_reply":"2023-04-02T06:31:04.523762Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Define model","metadata":{}},{"cell_type":"code","source":"import torch.nn as nn\nfrom torch.optim import Adam\nimport torch.nn.functional as f\n\nfrom torch.nn.utils.rnn import pad_sequence, pack_padded_sequence, pad_packed_sequence","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:31:06.498304Z","iopub.execute_input":"2023-04-02T06:31:06.498731Z","iopub.status.idle":"2023-04-02T06:31:06.504457Z","shell.execute_reply.started":"2023-04-02T06:31:06.498691Z","shell.execute_reply":"2023-04-02T06:31:06.503065Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n\nprint(device)","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:31:16.305455Z","iopub.execute_input":"2023-04-02T06:31:16.305932Z","iopub.status.idle":"2023-04-02T06:31:16.312926Z","shell.execute_reply.started":"2023-04-02T06:31:16.305893Z","shell.execute_reply":"2023-04-02T06:31:16.311851Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"\nclass IceCubeLSTM(pl.LightningModule):\n    def __init__(self, input_size, lstm_units, lr, batch_size, ts_len):\n        super(IceCubeLSTM, self).__init__()\n        \n        self.save_hyperparameters()\n        \n        self.bidgru_layer = nn.GRU(input_size=input_size, hidden_size=lstm_units, num_layers=3, bias=True,\n                                   batch_first=True)\n        self.linear1 = nn.Linear(lstm_units, 256)\n        self.linear2 = nn.Linear(256, bin_num ** 2)\n        \n        # self.act1 = nn.LeakyReLU(negative_slope=0.1)\n        self.act1 = nn.ReLU()\n        self.lstm_units = lstm_units\n        \n        self.lr = lr\n        self.batch_size = batch_size\n        self.ts_len = ts_len\n        \n        # self.bn = nn.BatchNorm1d(256)\n        # self.bn2 = nn.BatchNorm1d(lstm_units)\n    \n    def forward(self, data):\n        data, h_n = self.bidgru_layer(data, None)\n        \n        # data, _ = pad_packed_sequence(data, batch_first=True)\n        data = data[:, -1, :]  # only use the val of final timestamp\n        # data = self.bn2(data)\n\n        data = self.linear1(data)\n\n        data = self.act1(data)\n        #  data = self.bn(data)\n        data = self.linear2(data)\n        return data\n    \n    def validation_step(self, data, batch_idx):\n        data_x = data[0]\n        data_y = data[1]\n        \n        data_x[:, :, 2] += 1  # default padding val is -1\n        \n        data_y_ = data_y.cpu().detach().numpy()\n        valid_truth_azimuth, valid_truth_zenith = data_y_[:, 0], data_y_[:, 1]\n        valid_truth_anglecode = torch.tensor(self._y_to_angle_code(data_y_), dtype=torch.int64).to(device)\n        \n        data_x_min = data_x.min(axis=-1)[0]\n        data_x_max = data_x.max(axis=-1)[0]\n        data_x_zero = torch.zeros_like(data_x_min)\n        data_x_zero_min = data_x_zero.eq(data_x_min).int()\n        data_x_zero_max = data_x_zero.eq(data_x_max).int()\n        data_x_zero_all = (data_x_zero_min + data_x_zero_max > 1).int()\n        data_x_zero_all = data_x_zero_all.sum(-1)\n        sequence_info_list = self.ts_len - data_x_zero_all\n        sequence_info_list = sequence_info_list.cpu().numpy()\n        \n        pack_padded_data = pack_padded_sequence(data_x, sequence_info_list, batch_first=True, enforce_sorted=False)\n        \n        # pack_padded_data = data_x\n        \n        predict = self.forward(pack_padded_data)\n        \n        loss = f.cross_entropy(predict, valid_truth_anglecode)\n        \n        predict_ = predict.cpu().detach().numpy()\n        valid_pred_azimuth, valid_pred_zenith = self._pred_to_angle(predict_)\n        \n        mae = self._angular_dist_score(valid_truth_azimuth, valid_truth_zenith, valid_pred_azimuth, valid_pred_zenith)\n        \n        self.log('val_mae', mae, on_step=True, on_epoch=True, logger=True)\n        self.log('val_loss', loss, on_step=True, on_epoch=True, logger=True)\n        \n    \n    def training_step(self, data, batch_idx):\n        \n        # tensorboard = self.logger.experiment\n        \n        data_x = data[0]\n        data_y = data[1]\n        \n        data_x[:, :, 2] += 1\n        \n        # to generate sequence_info_list\n        # tell pytorch the length of pedding timestamp\n        data_x_min = data_x.min(axis=-1)[0]\n        data_x_max = data_x.max(axis=-1)[0]\n        data_x_zero = torch.zeros_like(data_x_min)\n        data_x_zero_min = data_x_zero.eq(data_x_min).int()\n        data_x_zero_max = data_x_zero.eq(data_x_max).int()\n        data_x_zero_all = (data_x_zero_min + data_x_zero_max > 1).int()\n        data_x_zero_all = data_x_zero_all.sum(-1)\n        sequence_info_list = self.ts_len - data_x_zero_all\n        sequence_info_list = sequence_info_list.cpu().numpy()\n        \n        pack_padded_data = pack_padded_sequence(data_x, sequence_info_list, batch_first=True, enforce_sorted=False)\n        \n        predict_packed = self.forward(pack_padded_data)\n        \n        loss = f.cross_entropy(predict_packed, data_y)\n        \n        return loss\n    \n    def test_step(self, data, batch_idx):\n        pass\n    \n    def configure_optimizers(self):\n        optimizer = Adam(self.parameters(), lr=learning_rate)\n        \n        return {'optimizer': optimizer}\n    \n    @staticmethod\n    def _pred_to_angle(pred, epsilon=1e-8):\n        # convert prediction to vector\n        pred_vector = (pred.reshape((-1, bin_num * bin_num, 1)) * angle_bin_vector).sum(axis=1)\n        \n        # normalize\n        pred_vector_norm = np.sqrt((pred_vector ** 2).sum(axis=1))\n        mask = pred_vector_norm < epsilon\n        pred_vector_norm[mask] = 1\n        \n        # assign <1, 0, 0> to very small vectors (badly predicted)\n        pred_vector /= pred_vector_norm.reshape((-1, 1))\n        pred_vector[mask] = np.array([1., 0., 0.])\n        \n        # convert to angle\n        azimuth = np.arctan2(pred_vector[:, 1], pred_vector[:, 0])\n        azimuth[azimuth < 0] += 2 * np.pi\n        zenith = np.arccos(pred_vector[:, 2])\n        \n        return azimuth, zenith\n    \n    @staticmethod\n    def _y_to_angle_code(batch_y):\n        azimuth_code = (batch_y[:, 0] > azimuth_edges[1:].reshape((-1, 1))).sum(axis=0)\n        zenith_code = (batch_y[:, 1] > zenith_edges[1:].reshape((-1, 1))).sum(axis=0)\n        angle_code = bin_num * azimuth_code + zenith_code\n        \n        return angle_code\n    \n    @staticmethod\n    def _angular_dist_score(az_true, zen_true, az_pred, zen_pred):\n        if not (np.all(np.isfinite(az_true)) and\n                np.all(np.isfinite(zen_true)) and\n                np.all(np.isfinite(az_pred)) and\n                np.all(np.isfinite(zen_pred))):\n            raise ValueError(\"All arguments must be finite\")\n        \n        # pre-compute all sine and cosine values\n        sa1 = np.sin(az_true)\n        ca1 = np.cos(az_true)\n        sz1 = np.sin(zen_true)\n        cz1 = np.cos(zen_true)\n        \n        sa2 = np.sin(az_pred)\n        ca2 = np.cos(az_pred)\n        sz2 = np.sin(zen_pred)\n        cz2 = np.cos(zen_pred)\n        \n        # scalar product of the two cartesian vectors (x = sz*ca, y = sz*sa, z = cz)\n        scalar_prod = sz1 * sz2 * (ca1 * ca2 + sa1 * sa2) + (cz1 * cz2)\n        \n        # scalar product of two unit vectors is always between -1 and 1, this is against nummerical instability\n        # that might otherwise occure from the finite precision of the sine and cosine functions\n        scalar_prod = np.clip(scalar_prod, -1, 1)\n        \n        # convert back to an angle (in radian)\n        return np.average(np.abs(np.arccos(scalar_prod)))\n","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:31:36.157082Z","iopub.execute_input":"2023-04-02T06:31:36.157662Z","iopub.status.idle":"2023-04-02T06:31:36.191853Z","shell.execute_reply.started":"2023-04-02T06:31:36.157618Z","shell.execute_reply":"2023-04-02T06:31:36.190351Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Model train","metadata":{}},{"cell_type":"code","source":"train = LSTMDataset(90, mode='train')\nvalidation = LSTMDataset(1, mode='valid')","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:31:40.927924Z","iopub.execute_input":"2023-04-02T06:31:40.928622Z","iopub.status.idle":"2023-04-02T06:31:51.733261Z","shell.execute_reply.started":"2023-04-02T06:31:40.928581Z","shell.execute_reply":"2023-04-02T06:31:51.731836Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"lstm_dataloader = LSTMDataLoader(train, validation, batch_size=batch_size)","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:34:13.818249Z","iopub.execute_input":"2023-04-02T06:34:13.818816Z","iopub.status.idle":"2023-04-02T06:34:13.825538Z","shell.execute_reply.started":"2023-04-02T06:34:13.818776Z","shell.execute_reply":"2023-04-02T06:34:13.824226Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model = IceCubeLSTM(input_size=feature_count, lstm_units=lstm_units\n                        , lr=learning_rate, batch_size=batch_size, ts_len=pulse_count)","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:34:16.039621Z","iopub.execute_input":"2023-04-02T06:34:16.040079Z","iopub.status.idle":"2023-04-02T06:34:16.056497Z","shell.execute_reply.started":"2023-04-02T06:34:16.040041Z","shell.execute_reply":"2023-04-02T06:34:16.055307Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"trainer = pl.Trainer(accelerator='auto'\n                     , devices='auto'\n                     , precision=16\n                     , max_epochs=epochs\n                     , val_check_interval=0.1\n                     , fast_dev_run=1)","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:34:19.547675Z","iopub.execute_input":"2023-04-02T06:34:19.548269Z","iopub.status.idle":"2023-04-02T06:34:19.605807Z","shell.execute_reply.started":"2023-04-02T06:34:19.548213Z","shell.execute_reply":"2023-04-02T06:34:19.604936Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# trainer.fit(model, lstm_dataloader.train_dataloader(), lstm_dataloader.val_dataloader())","metadata":{"execution":{"iopub.status.busy":"2023-04-02T06:34:23.125867Z","iopub.execute_input":"2023-04-02T06:34:23.126305Z"},"trusted":true},"execution_count":null,"outputs":[]}]}