#@title
from IPython.display import HTML

HTML('<iframe width="560" height="315" src="https://www.youtube.com/embed/cpzq6ESSM5c?rel=0&amp;controls=0&amp;showinfo=0" frameborder="0" allowfullscreen></iframe>')

! pip install datasets transformers[sentencepiece]

#@title
from IPython.display import HTML

HTML('<iframe width="560" height="315" src="https://www.youtube.com/embed/eKv4rRcCNX0?rel=0&amp;controls=0&amp;showinfo=0" frameborder="0" allowfullscreen></iframe>')

! pip install datasets transformers[sentencepiece]

from datasets import load_dataset
from transformers import AutoTokenizer
import numpy as np

raw_datasets = load_dataset("glue", "mrpc")
checkpoint = "bert-base-uncased"
tokenizer = AutoTokenizer.from_pretrained(checkpoint)

def tokenize_dataset(dataset):
    encoded = tokenizer(
        dataset["sentence1"],
        dataset["sentence2"],
        truncation=True,
    )
    return encoded.data

tokenized_datasets = raw_datasets.map(tokenize_dataset)

train_dataset = tokenized_datasets["train"].to_tf_dataset(
    columns=["input_ids", "attention_mask", "token_type_ids"],
    label_cols=["label"],
    shuffle=True,
    batch_size=8)

validation_dataset = tokenized_datasets["validation"].to_tf_dataset(
    columns=["input_ids", "attention_mask", "token_type_ids"],
    label_cols=["label"],
    shuffle=True,
    batch_size=8)

import tensorflow as tf
from transformers import TFAutoModelForSequenceClassification

checkpoint = 'bert-base-cased'
model = TFAutoModelForSequenceClassification.from_pretrained(checkpoint, num_labels=2)
loss = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True)

from tensorflow.keras.optimizers.schedules import PolynomialDecay

num_epochs = 3
num_train_steps = len(train_dataset) * num_epochs
lr_scheduler = PolynomialDecay(
    initial_learning_rate=5e-5,
    end_learning_rate=0.,
    decay_steps=num_train_steps
)

from tensorflow.keras.optimizers import Adam

opt = Adam(learning_rate=lr_scheduler)
model.compile(loss=loss, optimizer=opt)