from datasets import load_dataset

spanish_dataset = load_dataset("amazon_reviews_multi", "es")
english_dataset = load_dataset("amazon_reviews_multi", "en")
english_dataset

DatasetDict({
    train: Dataset({
        features: ['review_id', 'product_id', 'reviewer_id', 'stars', 'review_body', 'review_title', 'language', 'product_category'],
        num_rows: 200000
    })
    validation: Dataset({
        features: ['review_id', 'product_id', 'reviewer_id', 'stars', 'review_body', 'review_title', 'language', 'product_category'],
        num_rows: 5000
    })
    test: Dataset({
        features: ['review_id', 'product_id', 'reviewer_id', 'stars', 'review_body', 'review_title', 'language', 'product_category'],
        num_rows: 5000
    })
})

def show_samples(dataset, num_samples=3, seed=42):
    sample = dataset["train"].shuffle(seed=seed).select(range(num_samples))
    for example in sample:
        print(f"\n'>> Title: {example['review_title']}'")
        print(f"'>> Review: {example['review_body']}'")


show_samples(english_dataset)

'>> Title: Worked in front position, not rear'
'>> Review: 3 stars because these are not rear brakes as stated in the item description. At least the mount adapter only worked on the front fork of the bike that I got it for.'

'>> Title: meh'
'>> Review: Does it’s job and it’s gorgeous but mine is falling apart, I had to basically put it together again with hot glue'

'>> Title: Can\'t beat these for the money'
'>> Review: Bought this for handling miscellaneous aircraft parts and hanger "stuff" that I needed to organize; it really fit the bill. The unit arrived quickly, was well packaged and arrived intact (always a good sign). There are five wall mounts-- three on the top and two on the bottom. I wanted to mount it on the wall, so all I had to do was to remove the top two layers of plastic drawers, as well as the bottom corner drawers, place it when I wanted and mark it; I then used some of the new plastic screw in wall anchors (the 50 pound variety) and it easily mounted to the wall. Some have remarked that they wanted dividers for the drawers, and that they made those. Good idea. My application was that I needed something that I can see the contents at about eye level, so I wanted the fuller-sized drawers. I also like that these are the new plastic that doesn\'t get brittle and split like my older plastic drawers did. I like the all-plastic construction. It\'s heavy duty enough to hold metal parts, but being made of plastic it\'s not as heavy as a metal frame, so you can easily mount it to the wall and still load it up with heavy stuff, or light stuff. No problem there. For the money, you can\'t beat it. Best one of these I\'ve bought to date-- and I\'ve been using some version of these for over forty years.'

english_dataset.set_format("pandas")
english_df = english_dataset["train"][:]
# 显示前 20 个产品的数量
english_df["product_category"].value_counts()[:20]

home                      17679
apparel                   15951
wireless                  15717
other                     13418
beauty                    12091
drugstore                 11730
kitchen                   10382
toy                        8745
sports                     8277
automotive                 7506
lawn_and_garden            7327
home_improvement           7136
pet_products               7082
digital_ebook_purchase     6749
pc                         6401
electronics                6186
office_product             5521
shoes                      5197
grocery                    4730
book                       3756
Name: product_category, dtype: int64

def filter_books(example):
    return (
        example["product_category"] == "book"
        or example["product_category"] == "digital_ebook_purchase"
    )

english_dataset.reset_format()

spanish_books = spanish_dataset.filter(filter_books)
english_books = english_dataset.filter(filter_books)
show_samples(english_books)

'>> Title: I\'m dissapointed.'
'>> Review: I guess I had higher expectations for this book from the reviews. I really thought I\'d at least like it. The plot idea was great. I loved Ash but, it just didnt go anywhere. Most of the book was about their radio show and talking to callers. I wanted the author to dig deeper so we could really get to know the characters. All we know about Grace is that she is attractive looking, Latino and is kind of a brat. I\'m dissapointed.'

'>> Title: Good art, good price, poor design'
'>> Review: I had gotten the DC Vintage calendar the past two years, but it was on backorder forever this year and I saw they had shrunk the dimensions for no good reason. This one has good art choices but the design has the fold going through the picture, so it\'s less aesthetically pleasing, especially if you want to keep a picture to hang. For the price, a good calendar'

'>> Title: Helpful'
'>> Review: Nearly all the tips useful and. I consider myself an intermediate to advanced user of OneNote. I would highly recommend.'

from datasets import concatenate_datasets, DatasetDict

books_dataset = DatasetDict()

for split in english_books.keys():
    books_dataset[split] = concatenate_datasets(
        [english_books[split], spanish_books[split]]
    )
    books_dataset[split] = books_dataset[split].shuffle(seed=42)

# 挑选一些样例
show_samples(books_dataset)

'>> Title: Easy to follow!!!!'
'>> Review: I loved The dash diet weight loss Solution. Never hungry. I would recommend this diet. Also the menus are well rounded. Try it. Has lots of the information need thanks.'

'>> Title: PARCIALMENTE DAÑADO'
'>> Review: Me llegó el día que tocaba, junto a otros libros que pedí, pero la caja llegó en mal estado lo cual dañó las esquinas de los libros porque venían sin protección (forro).'

'>> Title: no lo he podido descargar'
'>> Review: igual que el anterior'

books_dataset = books_dataset.filter(lambda x: len(x["review_title"].split()) > 2)

from transformers import AutoTokenizer

model_checkpoint = "google/mt5-small"
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)

inputs = tokenizer("I loved reading the Hunger Games!")
inputs

{'input_ids': [336, 259, 28387, 11807, 287, 62893, 295, 12507, 1], 'attention_mask': [1, 1, 1, 1, 1, 1, 1, 1, 1]}

tokenizer.convert_ids_to_tokens(inputs.input_ids)

['▁I', '▁', 'loved', '▁reading', '▁the', '▁Hung', 'er', '▁Games', '</s>']

max_input_length = 512
max_target_length = 30


def preprocess_function(examples):
    model_inputs = tokenizer(
        examples["review_body"],
        max_length=max_input_length,
        truncation=True,
    )
    labels = tokenizer(
        examples["review_title"], max_length=max_target_length, truncation=True
    )
    model_inputs["labels"] = labels["input_ids"]
    return model_inputs

tokenized_datasets = books_dataset.map(preprocess_function, batched=True)

generated_summary = "I absolutely loved reading the Hunger Games"
reference_summary = "I loved reading the Hunger Games"

!pip install rouge_score

import evaluate

rouge_score = evaluate.load("rouge")

scores = rouge_score.compute(
    predictions=[generated_summary], references=[reference_summary]
)
scores

{'rouge1': AggregateScore(low=Score(precision=0.86, recall=1.0, fmeasure=0.92), mid=Score(precision=0.86, recall=1.0, fmeasure=0.92), high=Score(precision=0.86, recall=1.0, fmeasure=0.92)),
 'rouge2': AggregateScore(low=Score(precision=0.67, recall=0.8, fmeasure=0.73), mid=Score(precision=0.67, recall=0.8, fmeasure=0.73), high=Score(precision=0.67, recall=0.8, fmeasure=0.73)),
 'rougeL': AggregateScore(low=Score(precision=0.86, recall=1.0, fmeasure=0.92), mid=Score(precision=0.86, recall=1.0, fmeasure=0.92), high=Score(precision=0.86, recall=1.0, fmeasure=0.92)),
 'rougeLsum': AggregateScore(low=Score(precision=0.86, recall=1.0, fmeasure=0.92), mid=Score(precision=0.86, recall=1.0, fmeasure=0.92), high=Score(precision=0.86, recall=1.0, fmeasure=0.92))}

scores["rouge1"].mid

Score(precision=0.86, recall=1.0, fmeasure=0.92)

!pip install nltk

import nltk

nltk.download("punkt")

from nltk.tokenize import sent_tokenize


def three_sentence_summary(text):
    return "\n".join(sent_tokenize(text)[:3])


print(three_sentence_summary(books_dataset["train"][1]["review_body"]))

'I grew up reading Koontz, and years ago, I stopped,convinced i had "outgrown" him.'
'Still,when a friend was looking for something suspenseful too read, I suggested Koontz.'
'She found Strangers.'

def evaluate_baseline(dataset, metric):
    summaries = [three_sentence_summary(text) for text in dataset["review_body"]]
    return metric.compute(predictions=summaries, references=dataset["review_title"])

import pandas as pd

score = evaluate_baseline(books_dataset["validation"], rouge_score)
rouge_names = ["rouge1", "rouge2", "rougeL", "rougeLsum"]
rouge_dict = dict((rn, round(score[rn].mid.fmeasure * 100, 2)) for rn in rouge_names)
rouge_dict

{'rouge1': 16.74, 'rouge2': 8.83, 'rougeL': 15.6, 'rougeLsum': 15.96}

from transformers import AutoModelForSeq2SeqLM

model = AutoModelForSeq2SeqLM.from_pretrained(model_checkpoint)

from huggingface_hub import notebook_login

notebook_login()

huggingface-cli login

from transformers import Seq2SeqTrainingArguments

batch_size = 8
num_train_epochs = 8
# 每个训练周期都输出训练损失
logging_steps = len(tokenized_datasets["train"]) // batch_size
model_name = model_checkpoint.split("/")[-1]

args = Seq2SeqTrainingArguments(
    output_dir=f"{model_name}-finetuned-amazon-en-es",
    evaluation_strategy="epoch",
    learning_rate=5.6e-5,
    per_device_train_batch_size=batch_size,
    per_device_eval_batch_size=batch_size,
    weight_decay=0.01,
    save_total_limit=3,
    num_train_epochs=num_train_epochs,
    predict_with_generate=True,
    logging_steps=logging_steps,
    push_to_hub=True,
)

import numpy as np


def compute_metrics(eval_pred):
    predictions, labels = eval_pred
    # 将生成的摘要解码为文本
    decoded_preds = tokenizer.batch_decode(predictions, skip_special_tokens=True)
    # 替换标签中的-100,因为我们无法解码它们
    labels = np.where(labels != -100, labels, tokenizer.pad_token_id)
    # 将参考摘要解码为文本
    decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)
    # ROUGE期望每个句子后都有一个换行符
    decoded_preds = ["\n".join(sent_tokenize(pred.strip())) for pred in decoded_preds]
    decoded_labels = ["\n".join(sent_tokenize(label.strip())) for label in decoded_labels]
    # 计算ROUGE分数
    result = rouge_score.compute(
        predictions=decoded_preds, references=decoded_labels, use_stemmer=True
    )
    # 计算ROUGE分数
    result = {key: value.mid.fmeasure * 100 for key, value in result.items()}
    return {k: round(v, 4) for k, v in result.items()}

from transformers import DataCollatorForSeq2Seq

data_collator = DataCollatorForSeq2Seq(tokenizer, model=model)

tokenized_datasets = tokenized_datasets.remove_columns(
    books_dataset["train"].column_names
)

features = [tokenized_datasets["train"][i] for i in range(2)]
data_collator(features)

{'attention_mask': tensor([[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
         1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0],
        [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
         1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]]), 'input_ids': tensor([[  1494,    259,   8622,    390,    259,    262,   2316,   3435,    955,
            772,    281,    772,   1617,    263,    305,  14701,    260,   1385,
           3031,    259,  24146,    332,   1037,    259,  43906,    305,    336,
            260,      1,      0,      0,      0,      0,      0,      0],
        [   259,  27531,  13483,    259,   7505,    260, 112240,  15192,    305,
          53198,    276,    259,  74060,    263,    260,    459,  25640,    776,
           2119,    336,    259,   2220,    259,  18896,    288,   4906,    288,
           1037,   3931,    260,   7083, 101476,   1143,    260,      1]]), 'labels': tensor([[ 7483,   259,  2364, 15695,     1,  -100],
        [  259, 27531, 13483,   259,  7505,     1]]), 'decoder_input_ids': tensor([[    0,  7483,   259,  2364, 15695,     1],
        [    0,   259, 27531, 13483,   259,  7505]])}

from transformers import Seq2SeqTrainer

trainer = Seq2SeqTrainer(
    model,
    args,
    train_dataset=tokenized_datasets["train"],
    eval_dataset=tokenized_datasets["validation"],
    data_collator=data_collator,
    tokenizer=tokenizer,
    compute_metrics=compute_metrics,
)

trainer.train()

trainer.evaluate()

{'eval_loss': 3.028524398803711,
 'eval_rouge1': 16.9728,
 'eval_rouge2': 8.2969,
 'eval_rougeL': 16.8366,
 'eval_rougeLsum': 16.851,
 'eval_gen_len': 10.1597,
 'eval_runtime': 6.1054,
 'eval_samples_per_second': 38.982,
 'eval_steps_per_second': 4.914}

trainer.push_to_hub(commit_message="Training complete", tags="summarization")

'https://huggingface.co/huggingface-course/mt5-finetuned-amazon-en-es/commit/aa0536b829b28e73e1e4b94b8a5aacec420d40e0'

tokenized_datasets.set_format("torch")

from torch.utils.data import DataLoader

batch_size = 8
train_dataloader = DataLoader(
    tokenized_datasets["train"],
    shuffle=True,
    collate_fn=data_collator,
    batch_size=batch_size,
)
eval_dataloader = DataLoader(
    tokenized_datasets["validation"], collate_fn=data_collator, batch_size=batch_size
)

from torch.optim import AdamW

optimizer = AdamW(model.parameters(), lr=2e-5)

model = AutoModelForSeq2SeqLM.from_pretrained(model_checkpoint)

from accelerate import Accelerator

accelerator = Accelerator()
model, optimizer, train_dataloader, eval_dataloader = accelerator.prepare(
    model, optimizer, train_dataloader, eval_dataloader
)

from transformers import get_scheduler

num_train_epochs = 10
num_update_steps_per_epoch = len(train_dataloader)
num_training_steps = num_train_epochs * num_update_steps_per_epoch

lr_scheduler = get_scheduler(
    "linear",
    optimizer=optimizer,
    num_warmup_steps=0,
    num_training_steps=num_training_steps,
)

def postprocess_text(preds, labels):
    preds = [pred.strip() for pred in preds]
    labels = [label.strip() for label in labels]

    # ROUGE 需要每个句子后有一个换行符
    preds = ["\n".join(nltk.sent_tokenize(pred)) for pred in preds]
    labels = ["\n".join(nltk.sent_tokenize(label)) for label in labels]

    return preds, labels

from huggingface_hub import get_full_repo_name

model_name = "test-bert-finetuned-squad-accelerate"
repo_name = get_full_repo_name(model_name)
repo_name

'lewtun/mt5-finetuned-amazon-en-es-accelerate'

from huggingface_hub import Repository

output_dir = "results-mt5-finetuned-squad-accelerate"
repo = Repository(output_dir, clone_from=repo_name)

from tqdm.auto import tqdm
import torch
import numpy as np

progress_bar = tqdm(range(num_training_steps))

for epoch in range(num_train_epochs):
    # 训练
    model.train()
    for step, batch in enumerate(train_dataloader):
        outputs = model(**batch)
        loss = outputs.loss
        accelerator.backward(loss)

        optimizer.step()
        lr_scheduler.step()
        optimizer.zero_grad()
        progress_bar.update(1)

    # 评估
    model.eval()
    for step, batch in enumerate(eval_dataloader):
        with torch.no_grad():
            generated_tokens = accelerator.unwrap_model(model).generate(
                batch["input_ids"],
                attention_mask=batch["attention_mask"],
            )

            generated_tokens = accelerator.pad_across_processes(
                generated_tokens, dim=1, pad_index=tokenizer.pad_token_id
            )
            labels = batch["labels"]

            # 如果我们没有填充到最大长度,我们需要填充标签
            labels = accelerator.pad_across_processes(
                batch["labels"], dim=1, pad_index=tokenizer.pad_token_id
            )

            generated_tokens = accelerator.gather(generated_tokens).cpu().numpy()
            labels = accelerator.gather(labels).cpu().numpy()

            # 替换标签中的 -100,因为我们无法解码它们
            labels = np.where(labels != -100, labels, tokenizer.pad_token_id)
            if isinstance(generated_tokens, tuple):
                generated_tokens = generated_tokens[0]
            decoded_preds = tokenizer.batch_decode(
                generated_tokens, skip_special_tokens=True
            )
            decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)

            decoded_preds, decoded_labels = postprocess_text(
                decoded_preds, decoded_labels
            )

            rouge_score.add_batch(predictions=decoded_preds, references=decoded_labels)

    # 计算评估的 loss
    result = rouge_score.compute()
    # 提取中位 ROUGE 分数
    result = {key: value.mid.fmeasure * 100 for key, value in result.items()}
    result = {k: round(v, 4) for k, v in result.items()}
    print(f"Epoch {epoch}:", result)

    # 保存和上传
    accelerator.wait_for_everyone()
    unwrapped_model = accelerator.unwrap_model(model)
    unwrapped_model.save_pretrained(output_dir, save_function=accelerator.save)
    if accelerator.is_main_process:
        tokenizer.save_pretrained(output_dir)
        repo.push_to_hub(
            commit_message=f"Training in progress epoch {epoch}", blocking=False
        )

Epoch 0: {'rouge1': 5.6351, 'rouge2': 1.1625, 'rougeL': 5.4866, 'rougeLsum': 5.5005}
Epoch 1: {'rouge1': 9.8646, 'rouge2': 3.4106, 'rougeL': 9.9439, 'rougeLsum': 9.9306}
Epoch 2: {'rouge1': 11.0872, 'rouge2': 3.3273, 'rougeL': 11.0508, 'rougeLsum': 10.9468}
Epoch 3: {'rouge1': 11.8587, 'rouge2': 4.8167, 'rougeL': 11.7986, 'rougeLsum': 11.7518}
Epoch 4: {'rouge1': 12.9842, 'rouge2': 5.5887, 'rougeL': 12.7546, 'rougeLsum': 12.7029}
Epoch 5: {'rouge1': 13.4628, 'rouge2': 6.4598, 'rougeL': 13.312, 'rougeLsum': 13.2913}
Epoch 6: {'rouge1': 12.9131, 'rouge2': 5.8914, 'rougeL': 12.6896, 'rougeLsum': 12.5701}
Epoch 7: {'rouge1': 13.3079, 'rouge2': 6.2994, 'rougeL': 13.1536, 'rougeLsum': 13.1194}
Epoch 8: {'rouge1': 13.96, 'rouge2': 6.5998, 'rougeL': 13.9123, 'rougeLsum': 13.7744}
Epoch 9: {'rouge1': 14.1192, 'rouge2': 7.0059, 'rougeL': 14.1172, 'rougeLsum': 13.9509}

from transformers import pipeline

hub_model_id = "huggingface-course/mt5-small-finetuned-amazon-en-es"
summarizer = pipeline("summarization", model=hub_model_id)

def print_summary(idx):
    review = books_dataset["test"][idx]["review_body"]
    title = books_dataset["test"][idx]["review_title"]
    summary = summarizer(books_dataset["test"][idx]["review_body"])[0]["summary_text"]
    print(f"'>>> Review: {review}'")
    print(f"\n'>>> Title: {title}'")
    print(f"\n'>>> Summary: {summary}'")

print_summary(100)

'>>> Review: Nothing special at all about this product... the book is too small and stiff and hard to write in. The huge sticker on the back doesn’t come off and looks super tacky. I would not purchase this again. I could have just bought a journal from the dollar store and it would be basically the same thing. It’s also really expensive for what it is.'

'>>> Title: Not impressed at all... buy something else'

'>>> Summary: Nothing special at all about this product'

print_summary(0)

'>>> Review: Es una trilogia que se hace muy facil de leer. Me ha gustado, no me esperaba el final para nada'

'>>> Title: Buena literatura para adolescentes'

'>>> Summary: Muy facil de leer'