¿Big data? 🤗 ¡Datasets al rescate!¶

Install the Transformers, Datasets, and Evaluate libraries to run this notebook.

In [ ]:

!pip install datasets evaluate transformers[sentencepiece]

In [ ]:

!pip install zstandard

In [ ]:

from datasets import load_dataset

# Esto toma algunos minutos para ejecutarse, así que ve por un te o un café mientras esperas :)
data_files = "https://mystic.the-eye.eu/public/AI/pile_preliminary_components/PUBMED_title_abstracts_2019_baseline.jsonl.zst"
pubmed_dataset = load_dataset("json", data_files=data_files, split="train")
pubmed_dataset

Out[ ]:

Dataset({
    features: ['meta', 'text'],
    num_rows: 15518009
})

In [ ]:

pubmed_dataset[0]

Out[ ]:

{'meta': {'pmid': 11409574, 'language': 'eng'},
 'text': 'Epidemiology of hypoxaemia in children with acute lower respiratory infection.\nTo determine the prevalence of hypoxaemia in children aged under 5 years suffering acute lower respiratory infections (ALRI), the risk factors for hypoxaemia in children under 5 years of age with ALRI, and the association of hypoxaemia with an increased risk of dying in children of the same age ...'}

In [ ]:

!pip install psutil

In [ ]:

import psutil

# Process.memory_info está expresado en bytes, así que lo convertimos en megabytes
print(f"RAM used: {psutil.Process().memory_info().rss / (1024 * 1024):.2f} MB")

Out[ ]:

RAM used: 5678.33 MB

In [ ]:

print(f"Number of files in dataset : {pubmed_dataset.dataset_size}")
size_gb = pubmed_dataset.dataset_size / (1024**3)
print(f"Dataset size (cache file) : {size_gb:.2f} GB")

Out[ ]:

Number of files in dataset : 20979437051
Dataset size (cache file) : 19.54 GB

In [ ]:

import timeit

code_snippet = """batch_size = 1000

for idx in range(0, len(pubmed_dataset), batch_size):
    _ = pubmed_dataset[idx:idx + batch_size]
"""

time = timeit.timeit(stmt=code_snippet, number=1, globals=globals())
print(
    f"Iterated over {len(pubmed_dataset)} examples (about {size_gb:.1f} GB) in "
    f"{time:.1f}s, i.e. {size_gb/time:.3f} GB/s"
)

Out[ ]:

'Iterated over 15518009 examples (about 19.5 GB) in 64.2s, i.e. 0.304 GB/s'

In [ ]:

pubmed_dataset_streamed = load_dataset(
    "json", data_files=data_files, split="train", streaming=True
)

In [ ]:

next(iter(pubmed_dataset_streamed))

Out[ ]:

{'meta': {'pmid': 11409574, 'language': 'eng'},
 'text': 'Epidemiology of hypoxaemia in children with acute lower respiratory infection.\nTo determine the prevalence of hypoxaemia in children aged under 5 years suffering acute lower respiratory infections (ALRI), the risk factors for hypoxaemia in children under 5 years of age with ALRI, and the association of hypoxaemia with an increased risk of dying in children of the same age ...'}

In [ ]:

from transformers import AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
tokenized_dataset = pubmed_dataset_streamed.map(lambda x: tokenizer(x["text"]))
next(iter(tokenized_dataset))

Out[ ]:

{'input_ids': [101, 4958, 5178, 4328, 6779, ...], 'attention_mask': [1, 1, 1, 1, 1, ...]}

In [ ]:

shuffled_dataset = pubmed_dataset_streamed.shuffle(buffer_size=10_000, seed=42)
next(iter(shuffled_dataset))

Out[ ]:

{'meta': {'pmid': 11410799, 'language': 'eng'},
 'text': 'Randomized study of dose or schedule modification of granulocyte colony-stimulating factor in platinum-based chemotherapy for elderly patients with lung cancer ...'}

In [ ]:

dataset_head = pubmed_dataset_streamed.take(5)
list(dataset_head)

Out[ ]:

[{'meta': {'pmid': 11409574, 'language': 'eng'},
  'text': 'Epidemiology of hypoxaemia in children with acute lower respiratory infection ...'},
 {'meta': {'pmid': 11409575, 'language': 'eng'},
  'text': 'Clinical signs of hypoxaemia in children with acute lower respiratory infection: indicators of oxygen therapy ...'},
 {'meta': {'pmid': 11409576, 'language': 'eng'},
  'text': "Hypoxaemia in children with severe pneumonia in Papua New Guinea ..."},
 {'meta': {'pmid': 11409577, 'language': 'eng'},
  'text': 'Oxygen concentrators and cylinders ...'},
 {'meta': {'pmid': 11409578, 'language': 'eng'},
  'text': 'Oxygen supply in rural africa: a personal experience ...'}]

In [ ]:

# Skip the first 1,000 examples and include the rest in the training set
train_dataset = shuffled_dataset.skip(1000)
# Take the first 1,000 examples for the validation set
validation_dataset = shuffled_dataset.take(1000)

In [ ]:

law_dataset_streamed = load_dataset(
    "json",
    data_files="https://mystic.the-eye.eu/public/AI/pile_preliminary_components/FreeLaw_Opinions.jsonl.zst",
    split="train",
    streaming=True,
)
next(iter(law_dataset_streamed))

Out[ ]:

{'meta': {'case_ID': '110921.json',
  'case_jurisdiction': 'scotus.tar.gz',
  'date_created': '2010-04-28T17:12:49Z'},
 'text': '\n461 U.S. 238 (1983)\nOLIM ET AL.\nv.\nWAKINEKONA\nNo. 81-1581.\nSupreme Court of United States.\nArgued January 19, 1983.\nDecided April 26, 1983.\nCERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT\n*239 Michael A. Lilly, First Deputy Attorney General of Hawaii, argued the cause for petitioners. With him on the brief was James H. Dannenberg, Deputy Attorney General...'}

In [ ]:

from itertools import islice
from datasets import interleave_datasets

combined_dataset = interleave_datasets([pubmed_dataset_streamed, law_dataset_streamed])
list(islice(combined_dataset, 2))

Out[ ]:

[{'meta': {'pmid': 11409574, 'language': 'eng'},
  'text': 'Epidemiology of hypoxaemia in children with acute lower respiratory infection ...'},
 {'meta': {'case_ID': '110921.json',
   'case_jurisdiction': 'scotus.tar.gz',
   'date_created': '2010-04-28T17:12:49Z'},
  'text': '\n461 U.S. 238 (1983)\nOLIM ET AL.\nv.\nWAKINEKONA\nNo. 81-1581.\nSupreme Court of United States.\nArgued January 19, 1983.\nDecided April 26, 1983.\nCERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT\n*239 Michael A. Lilly, First Deputy Attorney General of Hawaii, argued the cause for petitioners. With him on the brief was James H. Dannenberg, Deputy Attorney General...'}]

In [ ]:

base_url = "https://mystic.the-eye.eu/public/AI/pile/"
data_files = {
    "train": [base_url + "train/" + f"{idx:02d}.jsonl.zst" for idx in range(30)],
    "validation": base_url + "val.jsonl.zst",
    "test": base_url + "test.jsonl.zst",
}
pile_dataset = load_dataset("json", data_files=data_files, streaming=True)
next(iter(pile_dataset["train"]))

Out[ ]:

{'meta': {'pile_set_name': 'Pile-CC'},
 'text': 'It is done, and submitted. You can play “Survival of the Tastiest” on Android, and on the web...'}