{"cells":[{"metadata":{},"cell_type":"markdown","source":"## About this notebook\n\nThis notebook focus on: Gradient Accumulation with Tensorflow & TPU.\n\nFrom [Distributions of no. of tokens](#Distributions-of-no.-of-tokens), we can see that there are non-negligible number of comments which have more than 256 tokens. It is probably worth training with a higher value for `MAX_LEN`. However, with a large model like `XLM Roberta Large`, we can't have the same batch size for `MAX_LEN = 512` as for `MAX_LEN = 192`. So I looked how [OpenNMT/OpenNMT-tf](https://github.com/OpenNMT/OpenNMT-tf/) implements gradient accumulation and copy the code with modification to this kernel.\n\nThe following picture shows the results of training on 65536 examples with different batch configurations (use the optimized training loop).\n\n![batch.PNG](attachment:batch.PNG)\n\nFor `MAX_LEN = 192`, we can train with `batch_size_per_replica = 16` (without grandient accumulatioin). However, for `MAX_LEN = 256` or higher, `batch_size_per_replica = 16` no longer works. In order to keep the same effective batch size, it is necessary to use a smaller value for `batch_size_per_replica` with gradient accumulation.\n\nThe training configuation (fewer epochs / training examples, larger learning rate and `MAX_LEN = 192`) in this version is only for demonstrating the gradient accumulation works. Currently, my best LB score is given by:\n\n    * MAX_LEN = 512\n    * EPOCHS = 8\n    * WARMUP_EPOCHS = 4\n    * LEARNING_RATE_SCALING = 1\n    * N_NEGATIVE_EXAMPLES_FACTOR = 2\n    \nHowever, it takes too long to train, and it can't be trained on Kaggle. I am still looking for a faster training configuation that can achieve the same result.","attachments":{"batch.PNG":{"image/png":"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"}}},{"metadata":{},"cell_type":"markdown","source":"## Distributions of no. of tokens\n\nWe only look the comments with no. of tokens in the range (256, 512].\n\nThe tokens are obtained from the `bert-base-multilingual-cased` tokenizer. However, the distributions should be almost the same even other tokenizers are used.\n\n| ![ccc](attachment:train_long.png) | ![valid_long.png](attachment:valid_long.png) | ![test_long.png](attachment:test_long.png) |\n|:-|:-|:-|  \n|training dataset|validation dataset|test dataset| \n","attachments":{"train_long.png":{"image/png":"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"},"valid_long.png":{"image/png":"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"},"test_long.png":{"image/png":"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"}}},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import os\n\nimport numpy as np\nimport pandas as pd\nimport tensorflow as tf\nfrom tensorflow.keras.layers import Dense, Input\nfrom tensorflow.keras.optimizers import Adam\nfrom tensorflow.keras.models import Model\nfrom tensorflow.keras.callbacks import ModelCheckpoint\nfrom kaggle_datasets import KaggleDatasets\nimport transformers\nfrom transformers import TFAutoModel, AutoTokenizer\nfrom tqdm import tqdm\nfrom tokenizers import BertWordPieceTokenizer\nimport datetime\nfrom matplotlib import pyplot as plt\nimport sklearn\n\nAUTO = tf.data.experimental.AUTOTUNE\n\nprint(\"Tensorflow version \" + tf.__version__)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Seed"},{"metadata":{"trusted":true},"cell_type":"code","source":"# Set `DETERMINISTIC` to `True`\n# if you want to have more stable results among different trainings.\nDETERMINISTIC = True\n\nSEED = 0\n\nGLOBAL_SEED = None\nOP_SEED = None\n\nif DETERMINISTIC:\n\n    GLOBAL_SEED = SEED\n    OP_SEED = SEED\n\n    tf.random.set_seed(seed=GLOBAL_SEED)\n    np.random.seed(GLOBAL_SEED)\n\n    os.environ['PYTHONHASHSEED'] = str(SEED)\n    os.environ['TF_DETERMINISTIC_OPS'] = '1'","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## TPU or GPU detection"},{"metadata":{"trusted":true},"cell_type":"code","source":"# Detect hardware, return appropriate distribution strategy\ntry:\n    tpu = tf.distribute.cluster_resolver.TPUClusterResolver()  # TPU detection. No parameters necessary if TPU_NAME environment variable is set. On Kaggle this is always the case.\n    print('Running on TPU ', tpu.master())\nexcept ValueError:\n    tpu = None\n\nif tpu:\n    tf.config.experimental_connect_to_cluster(tpu)\n    tf.tpu.experimental.initialize_tpu_system(tpu)\n    strategy = tf.distribute.experimental.TPUStrategy(tpu)\nelse:\n    strategy = tf.distribute.get_strategy() # default distribution strategy in Tensorflow. Works on CPU and single GPU.\n\nprint(\"REPLICAS: \", strategy.num_replicas_in_sync)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Configuration"},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"# Easier to do experiments.\nCONFIG_DICT = {\n    \"distilbert-base-multilingual-cased\": {\n        'fast_encode': True,\n        'fast_tokenizer_class': BertWordPieceTokenizer,\n        'padding_token': 0\n    },\n    'bert-base-multilingual-cased': {\n        'fast_encode': True,\n        'fast_tokenizer_class': BertWordPieceTokenizer,\n        'padding_token': 0\n    },\n    'jplu/tf-xlm-roberta-base': {\n        'fast_encode': False,\n        'padding_token': 1\n    },\n    'jplu/tf-xlm-roberta-large': {\n        'fast_encode': False,\n        'padding_token': 1\n    }\n}\n\n# TRANSFORMER_TYPE = 'distilbert-base-multilingual-cased'\n# TRANSFORMER_TYPE = 'bert-base-multilingual-cased'\n# TRANSFORMER_TYPE = 'jplu/tf-xlm-roberta-base'\nTRANSFORMER_TYPE = 'jplu/tf-xlm-roberta-large'\n\nCONFIG = CONFIG_DICT[TRANSFORMER_TYPE]\n\nFAST_ENCODE = CONFIG['fast_encode']\nif FAST_ENCODE:\n    FAST_TOKENIZER_CLASS = CONFIG['fast_tokenizer_class']\nPADDING_TOKEN = CONFIG['padding_token']\n\nMAX_LEN = 192  # 512\n\nEPOCHS = 3  # 8\nWARMUP_EPOCHS = 1  # 4\n\n# The number of examples for which the training procedure running on a single replica will compute the gradients once in order to accumulate them.\nBATCH_SIZE_PER_REPLICA = 8  # 2\n\n# Accumulate `BATCHES_PER_UPDATE` of gradients before updating the model's parameters.\nBATCHES_PER_UPDATE = 2  # 8\n\n# The batch size for prediction procedure running on a single replica.\nPREDICTION_BATCH_SIZE_PER_REPLICA = 32\n\nLEARNING_RATE_SCALING = 2  # 1\nSTART_LR = 5e-6\nMAX_LR = 1e-5  # * strategy.num_replicas_in_sync\nENDING_LR = 5e-6\n\nSHUFFLE_BUFFER_SIZE = 4096\n\n# If to convert labels to 0 and 1.\nROUND_LABELS = True\n\n# The number of positive examples to sample from each training .csv files.\n# Set to `None` to use all positive examples.\nN_POSITIVE_EXAMPLES_TO_SAMPLE = None\n\n# The number of negative examples to sample from each training .csv files\n# will be `N_NEGATIVE_EXAMPLES_FACTOR * N_POSITIVE_EXAMPLES_TO_SAMPLE`.\n# Set to `None` to use all negative examples.\nN_NEGATIVE_EXAMPLES_FACTOR = 1  # 2\n\n# If to use optimized training / validation loops.\nOPTIMIZED_LOOP = True\n\n# If to use gradient accumulation.\nUSE_GRADIENT_ACCUMULATION = True\nif BATCHES_PER_UPDATE > 1:\n    USE_GRADIENT_ACCUMULATION = True","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Dataset methods"},{"metadata":{"trusted":true},"cell_type":"code","source":"def fast_encode(texts, tokenizer, chunk_size=256, maxlen=512):\n    \"\"\"\n    https://www.kaggle.com/xhlulu/jigsaw-tpu-distilbert-with-huggingface-and-keras\n    \"\"\"\n\n    tokenizer.enable_truncation(max_length=maxlen)\n    tokenizer.enable_padding(max_length=maxlen)\n    all_ids = []\n\n    for i in tqdm(range(0, len(texts), chunk_size)):\n        text_chunk = texts[i:i+chunk_size].tolist()\n        encs = tokenizer.encode_batch(text_chunk)\n        all_ids.extend([enc.ids for enc in encs])\n\n    return np.array(all_ids)\n\n\ndef regular_encode(texts, tokenizer, maxlen=512):\n    \"\"\"\n    https://www.kaggle.com/xhlulu/jigsaw-tpu-xlm-roberta\n    \"\"\"\n    enc_di = tokenizer.batch_encode_plus(\n        texts,\n        return_attention_masks=False,\n        return_token_type_ids=False,\n        pad_to_max_length=True,\n        max_length=maxlen\n    )\n\n    return np.array(enc_di['input_ids'])\n\n\ndef get_training_dataset(batch_size):\n\n    dataset = tf.data.Dataset.from_tensor_slices((x_train, y_train))\n    dataset = dataset.repeat()\n\n    dataset = dataset.shuffle(SHUFFLE_BUFFER_SIZE, seed=OP_SEED)\n\n    dataset = dataset.batch(batch_size, drop_remainder=True)\n    dataset = dataset.prefetch(AUTO)\n\n    return dataset\n\n\ndef get_validation_dataset(batch_size, repeated=False):\n\n    dataset = tf.data.Dataset.from_tensor_slices((x_valid, y_valid))\n\n    if repeated:\n        dataset = dataset.repeat()\n\n        # If no repetition, don't shuffle validation dataset\n        if not DETERMINISTIC:\n            dataset = dataset.shuffle(SHUFFLE_BUFFER_SIZE, seed=OP_SEED)\n\n    dataset = dataset.batch(batch_size, drop_remainder=True)\n    dataset = dataset.prefetch(AUTO)\n\n    return dataset\n\n\ndef get_test_dataset(batch_size):\n\n    dataset = tf.data.Dataset.from_tensor_slices(x_test)\n    dataset = dataset.batch(batch_size)\n    dataset = dataset.prefetch(AUTO)\n\n    return dataset","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Get datasets"},{"metadata":{"trusted":true},"cell_type":"code","source":"# From https://www.kaggle.com/xhlulu/jigsaw-tpu-distilbert-with-huggingface-and-keras\n\n# First load the real tokenizer\ntokenizer = AutoTokenizer.from_pretrained(TRANSFORMER_TYPE)\n\n# Save the loaded tokenizer locally\ntokenizer.save_pretrained('.')\n\n# Reload it with the huggingface tokenizers library\nif FAST_ENCODE:\n    fast_tokenizer = FAST_TOKENIZER_CLASS('vocab.txt', lowercase=False)\n\n# CSV\ntrain_1 = pd.read_csv(\"/kaggle/input/jigsaw-multilingual-toxic-comment-classification/jigsaw-toxic-comment-train.csv\")\ntrain_2 = pd.read_csv(\"/kaggle/input/jigsaw-multilingual-toxic-comment-classification/jigsaw-unintended-bias-train.csv\")\nvalid = pd.read_csv('/kaggle/input/jigsaw-multilingual-toxic-comment-classification/validation.csv')\ntest = pd.read_csv('/kaggle/input/jigsaw-multilingual-toxic-comment-classification/test.csv')\n\nif ROUND_LABELS:\n    train_2.toxic = train_2.toxic.round().astype(int)\n\nPOSITIVE_EXAMPLES_1 = train_1[['comment_text', 'toxic']].query('toxic > 0.5')\nPOSITIVE_EXAMPLES_2 = train_2[['comment_text', 'toxic']].query('toxic > 0.5')\nNEGATIVE_EXAMPLES_1 = train_1[['comment_text', 'toxic']].query('toxic <= 0.5')\nNEGATIVE_EXAMPLES_2 = train_2[['comment_text', 'toxic']].query('toxic <= 0.5')\n\nPOSITIVE_EXAMPLES_USED_1 = POSITIVE_EXAMPLES_1\nPOSITIVE_EXAMPLES_USED_2 = POSITIVE_EXAMPLES_2\nNEGATIVE_EXAMPLES_USED_1 = NEGATIVE_EXAMPLES_1\nNEGATIVE_EXAMPLES_USED_2 = NEGATIVE_EXAMPLES_2\n\nif N_POSITIVE_EXAMPLES_TO_SAMPLE is not None:\n    POSITIVE_EXAMPLES_USED_1 = POSITIVE_EXAMPLES_USED_1.sample(n=N_POSITIVE_EXAMPLES_TO_SAMPLE, random_state=OP_SEED)\n    POSITIVE_EXAMPLES_USED_2 = POSITIVE_EXAMPLES_USED_2.sample(n=N_POSITIVE_EXAMPLES_TO_SAMPLE, random_state=OP_SEED)\n\nif N_NEGATIVE_EXAMPLES_FACTOR is not None:\n    NEGATIVE_EXAMPLES_USED_1 = NEGATIVE_EXAMPLES_1.sample(n=int(len(POSITIVE_EXAMPLES_USED_1) * N_NEGATIVE_EXAMPLES_FACTOR), random_state=OP_SEED)    \n    NEGATIVE_EXAMPLES_USED_2 = NEGATIVE_EXAMPLES_2.sample(n=int(len(POSITIVE_EXAMPLES_USED_2) * N_NEGATIVE_EXAMPLES_FACTOR), random_state=OP_SEED)    \n\n# Combine train_1 with a subset of train_2\ntrain = pd.concat([\n    POSITIVE_EXAMPLES_USED_1,\n    NEGATIVE_EXAMPLES_USED_1,\n    POSITIVE_EXAMPLES_USED_2,\n    NEGATIVE_EXAMPLES_USED_2\n])\n# Shuffle\ntrain = sklearn.utils.shuffle(train, random_state=OP_SEED)\n\nvalid = valid[['comment_text', 'toxic']]\n\nN_TRAINING_EXAMPLES = len(train)\nN_VALIDATION_EXAMPLES = len(valid)\nN_TEST_EXAMPLES = len(test)\n\nprint(f'N_TRAINING_EXAMPLES = {N_TRAINING_EXAMPLES}')\nprint(f'N_POSITIVE_EXAMPLES = {len(POSITIVE_EXAMPLES_USED_1) + len(POSITIVE_EXAMPLES_USED_2)}')\nprint(f'N_NEGATIVE_EXAMPLES = {len(NEGATIVE_EXAMPLES_USED_1) + len(NEGATIVE_EXAMPLES_USED_2)}')\nprint(f'N_VALIDATION_EXAMPLES = {N_VALIDATION_EXAMPLES}')\nprint(f'N_TEST_EXAMPLES = {N_TEST_EXAMPLES}')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"if FAST_ENCODE:\n    x_train = fast_encode(train.comment_text.values, fast_tokenizer, maxlen=MAX_LEN)\n    x_valid = fast_encode(valid.comment_text.values, fast_tokenizer, maxlen=MAX_LEN)\n    x_test = fast_encode(test.content.values, fast_tokenizer, maxlen=MAX_LEN)\nelse:\n    x_train = regular_encode(train.comment_text.values, tokenizer, maxlen=MAX_LEN)\n    x_valid = regular_encode(valid.comment_text.values, tokenizer, maxlen=MAX_LEN)\n    x_test = regular_encode(test.content.values, tokenizer, maxlen=MAX_LEN)\n\n#  Labels needs to be reshaped\ny_train = train['toxic'].values.reshape(len(x_train), 1)\ny_valid = valid['toxic'].values.reshape(len(x_valid), 1)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Peek the datasets"},{"metadata":{"trusted":true},"cell_type":"code","source":"training_dataset = get_training_dataset(batch_size=2)\nvalidation_dataset = get_validation_dataset(batch_size=2)\ntest_dataset = get_test_dataset(batch_size=2)\n\nfor batch in training_dataset.take(1):\n    print(batch)\n\nfor batch in validation_dataset.take(1):\n    print(batch)\n\nfor batch in test_dataset.take(1):\n    print(batch)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Gradient Accumulator\n\nCopy from the [OpenNMT/OpenNMT-tf](https://github.com/OpenNMT/OpenNMT-tf/) with modification.\n"},{"metadata":{"trusted":true},"cell_type":"code","source":"# https://github.com/OpenNMT/OpenNMT-tf/blob/f14c05a7cb8b1b8f3a692d6fea3c12067bc3eb2c/opennmt/optimizers/utils.py#L64\n\nclass GradientAccumulator(object):\n    \"\"\"Gradient accumulation utility.\n    When used with a distribution strategy, the accumulator should be called in a\n    replica context. Gradients will be accumulated locally on each replica and\n    without synchronization. Users should then call ``.gradients``, scale the\n    gradients if required, and pass the result to ``apply_gradients``.\n    \"\"\"\n\n    # We use the ON_READ synchronization policy so that no synchronization is\n    # performed on assignment. To get the value, we call .value() which returns the\n    # value on the current replica without synchronization.\n\n    def __init__(self):\n        \"\"\"Initializes the accumulator.\"\"\"\n\n        self._gradients = []\n\n    def gradients(self):\n        \"\"\"The accumulated gradients on the current replica.\"\"\"\n\n        if not self._gradients:\n            raise ValueError(\"The accumulator should be called first to initialize the gradients\")\n\n        # return list(gradient.value() for gradient in self._gradients)\n        return self._gradients\n\n    def __call__(self, gradients):\n        \"\"\"Accumulates :obj:`gradients` on the current replica.\"\"\"\n\n        if not self._gradients:\n\n            self._gradients.extend(\n                [\n                    tf.Variable(\n                        tf.zeros_like(gradient),\n                        trainable=False,\n                        synchronization=tf.VariableSynchronization.ON_READ\n                    ) if gradient is not None else None for gradient in gradients\n                ]\n            )\n\n        if len(gradients) != len(self._gradients):\n            raise ValueError(\"Expected %s gradients, but got %d\" % (\n                    len(self._gradients), len(gradients)))\n\n        for accum_gradient, gradient in zip(self._gradients, gradients):\n            if gradient is not None:\n                accum_gradient.assign_add(gradient)\n\n    def reset(self):\n        \"\"\"Resets the accumulated gradients on the current replica.\"\"\"\n\n        if not self._gradients:\n            return\n\n        for gradient in self._gradients:\n            if gradient is not None:\n                gradient.assign(tf.zeros_like(gradient))","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## batch configurations"},{"metadata":{"trusted":true},"cell_type":"code","source":"# The total number of examples for which the training procedure will compute the gradients once in order to accumulate them.\n# This is also used for validation step.\nBATCH_SIZE = BATCH_SIZE_PER_REPLICA * strategy.num_replicas_in_sync\n\n# The number of examples for which the training procedure will update the model's parameters once.\n# This is the `effective` batch size, which will be used in tf.data.Dataset.\nUPDATE_SIZE = BATCH_SIZE * BATCHES_PER_UPDATE\n\n# The number of parameter updates in 1 epoch\nUPDATES_PER_EPOCH = N_TRAINING_EXAMPLES // UPDATE_SIZE\n\nPREDICTION_BATCH_SIZE = PREDICTION_BATCH_SIZE_PER_REPLICA * strategy.num_replicas_in_sync\n\n# The number of batches for a validation step.\nVALID_BATCHES_PER_EPOCH = N_VALIDATION_EXAMPLES // PREDICTION_BATCH_SIZE","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Model"},{"metadata":{"trusted":true},"cell_type":"code","source":"class Toxic_Classifier(tf.keras.models.Model):\n\n    def __init__(self, transformer):\n\n        super(Toxic_Classifier, self).__init__()\n\n        self.transformer = transformer\n        self.dropout = tf.keras.layers.Dropout(rate=0.25, name='dropout')\n        self.average_pooling_layer = tf.keras.layers.GlobalAveragePooling1D(name='average_pooling_layer')\n        self.max_pooling_layer = tf.keras.layers.GlobalMaxPool1D(name='max_pooling_layer')\n        self.dense_layer = tf.keras.layers.Dense(\n            1, name='probabilities', activation='sigmoid',\n            kernel_initializer=tf.keras.initializers.GlorotUniform(seed=OP_SEED),\n            bias_initializer='zeros'\n        )\n\n    def call(self, inputs, **kwargs):\n\n        comments = inputs\n        attention_mask = tf.math.not_equal(comments, PADDING_TOKEN)\n\n        # sequence_outpu: shape = [batch_size, seq_len, hidden_dim]\n        # pooled_output: shape = [batch_size, hidden_dim]\n        sequence_output = self.transformer([comments, attention_mask], **kwargs)[0]\n\n        # shape = [batch_size, hidden_dim]\n        average_pooling = self.average_pooling_layer(sequence_output, mask=attention_mask)\n        \n        # Avoid the padding timestamps to contribute to `max_pooling`\n        sequence_output_masked = sequence_output * tf.cast(attention_mask, tf.float32)[:, :, tf.newaxis] - (1e9) * tf.cast(tf.math.logical_not(attention_mask), tf.float32)[:, :, tf.newaxis]\n        max_pooling = self.max_pooling_layer(sequence_output_masked)\n\n        pooling = tf.concat([average_pooling, max_pooling], axis=1)\n\n        # shape = [batch_size, seq_len, hidden_dim]\n        x = self.dropout(pooling, training=kwargs.get('training', False))\n\n        # shape = [batch_size, 1]\n        probabilities = self.dense_layer(x)\n\n        return probabilities\n\n\nclass CustomExponentialDecaySchedule(tf.keras.optimizers.schedules.ExponentialDecay):\n    \"\"\"\n    Learning rate with exponential decay and linear warmup.\n    \"\"\"\n\n    def __init__(\n            self,\n            start_lr,\n            max_lr,\n            ending_lr,\n            num_training_steps,\n            num_warmup_steps,\n            scaling=1,\n            cycle=False,\n            name=None,\n      ):\n\n        self.start_lr = tf.cast(start_lr, tf.float32)\n        self.max_lr = tf.cast(max_lr, tf.float32)\n        self.ending_lr = tf.cast(ending_lr, tf.float32)\n\n        self.decay_rate = self.ending_lr / self.max_lr\n\n        self.num_training_steps = tf.cast(num_training_steps, tf.float32)\n        self.num_warmup_steps = max(num_warmup_steps, 0)\n\n        self.decay_steps = self.num_training_steps - self.num_warmup_steps\n\n        self.decay_lr_schedule = tf.keras.optimizers.schedules.ExponentialDecay(\n            self.max_lr,\n            self.decay_steps,\n            self.decay_rate,\n            staircase=False,\n            name=name\n        )\n\n        self.scaling = scaling\n\n        self.cycle = tf.constant(cycle, dtype=tf.bool)\n\n    def __call__(self, step):\n\n        step = tf.cond(self.cycle and step >= self.num_training_steps, lambda: step % self.num_training_steps, lambda: tf.cast(step, tf.float32))\n        lr = self.decay_lr_schedule(step)\n        lr = tf.math.maximum(lr, self.ending_lr)\n\n        if self.num_warmup_steps > 0:\n\n            num_warmup_steps = tf.cast(self.num_warmup_steps, tf.float32)\n\n            is_warmup = tf.cast(step < num_warmup_steps, tf.float32)\n\n            warmup_lr = (self.max_lr - self.start_lr) / num_warmup_steps * step + self.start_lr\n\n            decay_lr = self.decay_lr_schedule(step - num_warmup_steps)\n            decay_lr = tf.math.maximum(decay_lr, self.ending_lr)\n\n            lr = (1.0 - is_warmup) * decay_lr + is_warmup * warmup_lr\n\n        return lr * self.scaling","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def set_model(learning_rate_scaling=1):\n\n    with strategy.scope():\n\n        transformer = TFAutoModel.from_pretrained(TRANSFORMER_TYPE)\n        model = Toxic_Classifier(transformer)\n\n        # number of training steps\n        num_training_steps = UPDATES_PER_EPOCH * EPOCHS\n\n        # warmup epochs\n        num_warmup_steps = UPDATES_PER_EPOCH * WARMUP_EPOCHS\n\n        lr_schedule = CustomExponentialDecaySchedule(\n            start_lr=START_LR,\n            max_lr=MAX_LR,\n            ending_lr=ENDING_LR,\n            num_training_steps=num_training_steps,\n            num_warmup_steps=num_warmup_steps,\n            scaling=learning_rate_scaling,\n            cycle=False,\n            name=None\n        )\n\n        optimizer = tf.keras.optimizers.Adam(learning_rate=lr_schedule)\n\n        gradient_accumulator = None\n        if USE_GRADIENT_ACCUMULATION:\n            gradient_accumulator = GradientAccumulator()\n            gradient_accumulator.reset()\n\n        # Instantiate metrics\n        train_accuracy = tf.keras.metrics.BinaryAccuracy()\n        valid_accuracy = tf.keras.metrics.BinaryAccuracy()\n        train_loss = tf.keras.metrics.Sum()\n        valid_loss = tf.keras.metrics.Sum()\n\n        loss_fn = tf.keras.losses.BinaryCrossentropy(from_logits=False, reduction=tf.keras.losses.Reduction.SUM)\n\n        return model, loss_fn, optimizer, gradient_accumulator, train_loss, train_accuracy, valid_loss, valid_accuracy","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Training routines"},{"metadata":{"trusted":true},"cell_type":"code","source":"def set_routines():\n\n    with strategy.scope():\n\n        def train_step_1_forward_backward(comments, labels):\n\n            with tf.GradientTape() as tape:\n\n                probabilities = model(comments, training=True)\n                loss = loss_fn(labels, probabilities)\n\n                # Take into account the fact that 1 parameter update for `UPDATE_SIZE` training examples.\n                scaled_loss = loss / UPDATE_SIZE\n\n            grads = tape.gradient(scaled_loss, model.trainable_variables)\n\n            # Accumulated already scaled gradients\n            if USE_GRADIENT_ACCUMULATION:\n                gradient_accumulator(grads)\n\n            # update metrics\n            train_accuracy.update_state(labels, probabilities)\n            train_loss.update_state(loss)\n\n            if not USE_GRADIENT_ACCUMULATION:\n                return grads\n\n        def train_step_1_update(comments, labels):\n\n            if not USE_GRADIENT_ACCUMULATION:\n\n                grads = train_step_1_forward_backward(comments, labels)\n                optimizer.apply_gradients(list(zip(grads, model.trainable_variables)))\n\n            else:\n\n                for _ in tf.range(BATCHES_PER_UPDATE):\n\n                    # Take the 1st `BATCH_SIZE_PER_REPLICA` examples.\n                    small_comments = comments[:BATCH_SIZE_PER_REPLICA]\n                    small_labels = labels[:BATCH_SIZE_PER_REPLICA]\n\n                    train_step_1_forward_backward(small_comments, small_labels)\n\n                    # Move the leading part to the end, so the shape is not changed.\n                    comments = tf.concat([comments[BATCH_SIZE_PER_REPLICA:], small_comments], axis=0)\n                    labels = tf.concat([labels[BATCH_SIZE_PER_REPLICA:], small_labels], axis=0)\n\n                # Update the model's parameters\n                gradients = gradient_accumulator.gradients()\n                optimizer.apply_gradients(list(zip(gradients, model.trainable_variables)))\n                gradient_accumulator.reset()\n\n        @tf.function\n        def dist_train_step(batch):\n            strategy.experimental_run_v2(train_step_1_update, batch)\n\n        @tf.function\n        def dist_train_1_epoch(data_iter):\n\n            for _ in tf.range(UPDATES_PER_EPOCH):\n                dist_train_step(next(data_iter))\n\n        def valid_step(comments, labels):\n\n            probabilities = model(comments, training=False)\n            loss = loss_fn(labels, probabilities)\n\n            # update metrics\n            valid_accuracy.update_state(labels, probabilities)\n            valid_loss.update_state(loss)\n\n        @tf.function\n        def dist_valid_step(batch):\n            strategy.experimental_run_v2(valid_step, batch)\n\n        @tf.function\n        def dist_valid(data_iter):\n\n            for _ in tf.range(VALID_BATCHES_PER_EPOCH):\n                dist_valid_step(next(data_iter))\n\n        return dist_train_1_epoch, dist_train_step, dist_valid, dist_valid_step","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Datasets"},{"metadata":{"trusted":true},"cell_type":"code","source":"print(\"NUM_TRAINING_EXAMPLES: {}\".format(N_TRAINING_EXAMPLES))\nprint(\"BATCH_SIZE_PER_REPLICA: {}\".format(BATCH_SIZE_PER_REPLICA))\nprint(\"BATCH_SIZE: {}\".format(BATCH_SIZE))\nprint(\"BATCHES_PER_UPDATE: {}\".format(BATCHES_PER_UPDATE))\nprint(\"UPDATE_SIZE: {}\".format(UPDATE_SIZE))\nprint(\"UPDATES_PER_EPOCH: {}\".format(UPDATES_PER_EPOCH))\nprint(\"NUM_VALIDATION_EXAMPLES: {}\".format(N_VALIDATION_EXAMPLES))\nprint(\"PREDICTION_BATCH_SIZE: {}\".format(PREDICTION_BATCH_SIZE))\nprint(\"VALID_BATCHES_PER_EPOCH: {}\".format(VALID_BATCHES_PER_EPOCH))\n\ntrain_ds = get_training_dataset(batch_size=UPDATE_SIZE)\ntrain_dist_ds = strategy.experimental_distribute_dataset(train_ds)\n\nvalid_ds = get_validation_dataset(batch_size=PREDICTION_BATCH_SIZE, repeated=True)\nvalid_dist_ds = strategy.experimental_distribute_dataset(valid_ds)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Usual custom training loop\nSlower, but training information could be shown more often."},{"metadata":{"trusted":true},"cell_type":"code","source":"if not OPTIMIZED_LOOP:\n\n    with strategy.scope():\n    \n        model, loss_fn, optimizer, gradient_accumulator, train_loss, train_accuracy, valid_loss, valid_accuracy = set_model(learning_rate_scaling=LEARNING_RATE_SCALING)\n        dist_train_1_epoch, dist_train_step, dist_valid, dist_valid_step = set_routines()\n\n        for epoch_idx in range(EPOCHS):\n\n            s = datetime.datetime.now()\n\n            for batch_idx, batch in enumerate(train_dist_ds):\n\n                if batch_idx >= UPDATES_PER_EPOCH:\n                    break\n\n                dist_train_step(batch)\n\n                loss = train_loss.result() / ((batch_idx + 1) * UPDATE_SIZE)\n                acc = train_accuracy.result()\n\n                if (batch_idx + 1) == UPDATES_PER_EPOCH or (batch_idx + 1) % (UPDATES_PER_EPOCH // 10) == 0:\n                    print(\"epoch: {} | batch: {}\".format(epoch_idx + 1, batch_idx + 1))\n                    print(\"train loss: {}\".format(loss))\n                    print(\"train accuracy: {}\".format(acc))\n\n            train_loss.reset_states()\n            train_accuracy.reset_states()\n\n            e = datetime.datetime.now()\n            print(\"training time: {}\".format((e-s).total_seconds()))\n\n            s = datetime.datetime.now()\n\n            for batch_idx, batch in enumerate(valid_dist_ds):\n\n                if batch_idx >= VALID_BATCHES_PER_EPOCH:\n                    break\n\n                dist_valid_step(batch)\n\n                val_loss = valid_loss.result() / (VALID_BATCHES_PER_EPOCH * PREDICTION_BATCH_SIZE)\n                val_acc = valid_accuracy.result()\n\n            print(\"valid loss: {}\".format(val_loss))\n            print(\"valid accuracy: {}\".format(val_acc))\n\n            valid_loss.reset_states()\n            valid_accuracy.reset_states()\n\n            e = datetime.datetime.now()\n            print(\"validation time: {}\".format((e-s).total_seconds()))\n\n            print(\"-\" * 80)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Optimized custom training loop\n\nOptimized by calling the TPU less often and performing more steps per call.\nTraining information is shown after each epoch.\n"},{"metadata":{"trusted":true},"cell_type":"code","source":"if OPTIMIZED_LOOP:\n\n    with strategy.scope():\n\n        model, loss_fn, optimizer, gradient_accumulator, train_loss, train_accuracy, valid_loss, valid_accuracy = set_model(learning_rate_scaling=LEARNING_RATE_SCALING)\n        dist_train_1_epoch, dist_train_step, dist_valid, dist_valid_step = set_routines()\n\n        train_data_iter = iter(train_dist_ds)\n        valid_data_iter = iter(valid_dist_ds)\n\n        for epoch_idx in range(EPOCHS):\n\n            s = datetime.datetime.now()\n\n            dist_train_1_epoch(train_data_iter)\n\n            loss = train_loss.result() / (UPDATES_PER_EPOCH * UPDATE_SIZE)\n            acc = train_accuracy.result()\n\n            print(\"epoch: {}\".format(epoch_idx + 1))\n            print(\"train loss: {}\".format(loss))\n            print(\"train accuracy: {}\".format(acc))\n\n            train_loss.reset_states()\n            train_accuracy.reset_states()\n\n            e = datetime.datetime.now()\n            print(\"training time: {}\".format((e-s).total_seconds()))\n\n            s = datetime.datetime.now()\n\n            dist_valid(valid_data_iter)\n\n            val_loss = valid_loss.result() / (VALID_BATCHES_PER_EPOCH * PREDICTION_BATCH_SIZE)\n            val_acc = valid_accuracy.result()\n\n            print(\"valid loss: {}\".format(val_loss))\n            print(\"valid accuracy: {}\".format(val_acc))\n\n            valid_loss.reset_states()\n            valid_accuracy.reset_states()\n\n            e = datetime.datetime.now()\n            print(\"validation time: {}\".format((e-s).total_seconds()))\n\n            print(\"-\" * 80)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Predict and submit"},{"metadata":{"trusted":true},"cell_type":"code","source":"with strategy.scope():\n\n    def predict_step(comments):\n\n        probabilities = model(comments, training=False)\n\n        return probabilities\n\n    @tf.function\n    def dist_predict_step(batch):\n\n        probabilities = strategy.experimental_run_v2(predict_step, args=[batch])\n\n        return probabilities\n\ntest_ds = get_test_dataset(batch_size=PREDICTION_BATCH_SIZE)\ndist_test_ds = strategy.experimental_distribute_dataset(test_ds)\nprint(\"PREDICTION_BATCH_SIZE: {}\".format(PREDICTION_BATCH_SIZE))\nPREDICTION_BATCHES = N_TEST_EXAMPLES // PREDICTION_BATCH_SIZE + int(N_TEST_EXAMPLES % PREDICTION_BATCH_SIZE > 0)\nprint(\"PREDICTION_BATCHES: {}\".format(PREDICTION_BATCHES))\n\nall_probabilities = []\nfor batch in tqdm(dist_test_ds):\n\n    # PerReplica object\n    probabilities = dist_predict_step(batch)\n\n    # Tuple of tensors\n    probabilities = strategy.experimental_local_results(probabilities)\n\n    # tf.Tensor\n    probabilities = tf.concat(probabilities, axis=0)\n\n    all_probabilities.append(probabilities)\n\n# tf.Tensor\npredictions = tf.concat(all_probabilities, axis=0)\n# numpy\npredictions = predictions.numpy()\n\nsubmission = pd.read_csv('/kaggle/input/jigsaw-multilingual-toxic-comment-classification/sample_submission.csv')\nsubmission['toxic'] = predictions\nsubmission.to_csv('submission.csv', index=False)\nsubmission.head(10)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Save model weights"},{"metadata":{"trusted":true},"cell_type":"code","source":"model.save_weights(\"model.h5\")","execution_count":null,"outputs":[]}],"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":4,"nbformat_minor":4}