CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutSign UpSign In
huggingface

Real-time collaboration for Jupyter Notebooks, Linux Terminals, LaTeX, VS Code, R IDE, and more,
all in one place. Commercial Alternative to JupyterHub.

GitHub Repository: huggingface/notebooks
 Path: blob/main/course/fr/chapter8/section4.ipynb
Views: 2555
Kernel: Python 3

Déboguer le pipeline d'entraînement

Ce chapitre portant sur le débogage, la langue nous importe peu ici. Nous nous intéressons surtout à la logique du code pour comprendre d'où provient l'erreur.

Installez les bibliothèques 🤗 Transformers et 🤗 Datasets pour exécuter ce notebook.

!pip install datasets transformers[sentencepiece]

from datasets import load_dataset, load_metric
from transformers import (
    AutoTokenizer,
    AutoModelForSequenceClassification,
    TrainingArguments,
    Trainer,
)

raw_datasets = load_dataset("glue", "mnli")

model_checkpoint = "distilbert-base-uncased"
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)


def preprocess_function(examples):
    return tokenizer(examples["premise"], examples["hypothesis"], truncation=True)


tokenized_datasets = raw_datasets.map(preprocess_function, batched=True)
model = AutoModelForSequenceClassification.from_pretrained(model_checkpoint)

args = TrainingArguments(
    f"distilbert-finetuned-mnli",
    evaluation_strategy="epoch",
    save_strategy="epoch",
    learning_rate=2e-5,
    num_train_epochs=3,
    weight_decay=0.01,
)

metric = load_metric("glue", "mnli")


def compute_metrics(eval_pred):
    predictions, labels = eval_pred
    return metric.compute(predictions=predictions, references=labels)


trainer = Trainer(
    model,
    args,
    train_dataset=raw_datasets["train"],
    eval_dataset=raw_datasets["validation_matched"],
    compute_metrics=compute_metrics,
)
trainer.train()

trainer.train_dataset[0]

from datasets import load_dataset, load_metric
from transformers import (
    AutoTokenizer,
    AutoModelForSequenceClassification,
    TrainingArguments,
    Trainer,
)

raw_datasets = load_dataset("glue", "mnli")

model_checkpoint = "distilbert-base-uncased"
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)


def preprocess_function(examples):
    return tokenizer(examples["premise"], examples["hypothesis"], truncation=True)


tokenized_datasets = raw_datasets.map(preprocess_function, batched=True)
model = AutoModelForSequenceClassification.from_pretrained(model_checkpoint)

args = TrainingArguments(
    f"distilbert-finetuned-mnli",
    evaluation_strategy="epoch",
    save_strategy="epoch",
    learning_rate=2e-5,
    num_train_epochs=3,
    weight_decay=0.01,
)

metric = load_metric("glue", "mnli")


def compute_metrics(eval_pred):
    predictions, labels = eval_pred
    return metric.compute(predictions=predictions, references=labels)


trainer = Trainer(
    model,
    args,
    train_dataset=tokenized_datasets["train"],
    eval_dataset=tokenized_datasets["validation_matched"],
    compute_metrics=compute_metrics,
)
trainer.train()

tokenizer.decode(trainer.train_dataset[0]["input_ids"])

trainer.train_dataset[0].keys()

type(trainer.model)

trainer.train_dataset[0]["attention_mask"]

len(trainer.train_dataset[0]["attention_mask"]) == len(
    trainer.train_dataset[0]["input_ids"]
)

trainer.train_dataset[0]["label"]

trainer.train_dataset.features["label"].names

for batch in trainer.get_train_dataloader():
    break

data_collator = trainer.get_train_dataloader().collate_fn
data_collator

from datasets import load_dataset, load_metric
from transformers import (
    AutoTokenizer,
    AutoModelForSequenceClassification,
    DataCollatorWithPadding,
    TrainingArguments,
    Trainer,
)

raw_datasets = load_dataset("glue", "mnli")

model_checkpoint = "distilbert-base-uncased"
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)


def preprocess_function(examples):
    return tokenizer(examples["premise"], examples["hypothesis"], truncation=True)


tokenized_datasets = raw_datasets.map(preprocess_function, batched=True)
model = AutoModelForSequenceClassification.from_pretrained(model_checkpoint)

args = TrainingArguments(
    f"distilbert-finetuned-mnli",
    evaluation_strategy="epoch",
    save_strategy="epoch",
    learning_rate=2e-5,
    num_train_epochs=3,
    weight_decay=0.01,
)

metric = load_metric("glue", "mnli")


def compute_metrics(eval_pred):
    predictions, labels = eval_pred
    return metric.compute(predictions=predictions, references=labels)


data_collator = DataCollatorWithPadding(tokenizer=tokenizer)

trainer = Trainer(
    model,
    args,
    train_dataset=tokenized_datasets["train"],
    eval_dataset=tokenized_datasets["validation_matched"],
    compute_metrics=compute_metrics,
    data_collator=data_collator,
    tokenizer=tokenizer,
)
trainer.train()

data_collator = trainer.get_train_dataloader().collate_fn
batch = data_collator([trainer.train_dataset[i] for i in range(4)])

data_collator = trainer.get_train_dataloader().collate_fn
actual_train_set = trainer._remove_unused_columns(trainer.train_dataset)
batch = data_collator([actual_train_set[i] for i in range(4)])

for batch in trainer.get_train_dataloader():
    break

outputs = trainer.model.cpu()(**batch)

trainer.model.config.num_labels

from datasets import load_dataset, load_metric
from transformers import (
    AutoTokenizer,
    AutoModelForSequenceClassification,
    DataCollatorWithPadding,
    TrainingArguments,
    Trainer,
)

raw_datasets = load_dataset("glue", "mnli")

model_checkpoint = "distilbert-base-uncased"
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)


def preprocess_function(examples):
    return tokenizer(examples["premise"], examples["hypothesis"], truncation=True)


tokenized_datasets = raw_datasets.map(preprocess_function, batched=True)
model = AutoModelForSequenceClassification.from_pretrained(model_checkpoint, num_labels=3)

args = TrainingArguments(
    f"distilbert-finetuned-mnli",
    evaluation_strategy="epoch",
    save_strategy="epoch",
    learning_rate=2e-5,
    num_train_epochs=3,
    weight_decay=0.01,
)

metric = load_metric("glue", "mnli")


def compute_metrics(eval_pred):
    predictions, labels = eval_pred
    return metric.compute(predictions=predictions, references=labels)


data_collator = DataCollatorWithPadding(tokenizer=tokenizer)

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

for batch in trainer.get_train_dataloader():
    break

outputs = trainer.model.cpu()(**batch)

import torch

device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
batch = {k: v.to(device) for k, v in batch.items()}

outputs = trainer.model.to(device)(**batch)

loss = outputs.loss
loss.backward()

trainer.create_optimizer()
trainer.optimizer.step()

# This will take a long time and error out, so you shouldn't run this cell
trainer.train()

trainer.evaluate()

for batch in trainer.get_eval_dataloader():
    break

batch = {k: v.to(device) for k, v in batch.items()}

with torch.no_grad():
    outputs = trainer.model(**batch)

predictions = outputs.logits.cpu().numpy()
labels = batch["labels"].cpu().numpy()

compute_metrics((predictions, labels))

predictions.shape, labels.shape

import numpy as np


def compute_metrics(eval_pred):
    predictions, labels = eval_pred
    predictions = np.argmax(predictions, axis=1)
    return metric.compute(predictions=predictions, references=labels)


compute_metrics((predictions, labels))

import numpy as np
from datasets import load_dataset, load_metric
from transformers import (
    AutoTokenizer,
    AutoModelForSequenceClassification,
    DataCollatorWithPadding,
    TrainingArguments,
    Trainer,
)

raw_datasets = load_dataset("glue", "mnli")

model_checkpoint = "distilbert-base-uncased"
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)


def preprocess_function(examples):
    return tokenizer(examples["premise"], examples["hypothesis"], truncation=True)


tokenized_datasets = raw_datasets.map(preprocess_function, batched=True)
model = AutoModelForSequenceClassification.from_pretrained(model_checkpoint, num_labels=3)

args = TrainingArguments(
    f"distilbert-finetuned-mnli",
    evaluation_strategy="epoch",
    save_strategy="epoch",
    learning_rate=2e-5,
    num_train_epochs=3,
    weight_decay=0.01,
)

metric = load_metric("glue", "mnli")


def compute_metrics(eval_pred):
    predictions, labels = eval_pred
    predictions = np.argmax(predictions, axis=1)
    return metric.compute(predictions=predictions, references=labels)


data_collator = DataCollatorWithPadding(tokenizer=tokenizer)

trainer = Trainer(
    model,
    args,
    train_dataset=tokenized_datasets["train"],
    eval_dataset=tokenized_datasets["validation_matched"],
    compute_metrics=compute_metrics,
    data_collator=data_collator,
    tokenizer=tokenizer,
)
trainer.train()

for batch in trainer.get_train_dataloader():
    break

batch = {k: v.to(device) for k, v in batch.items()}
trainer.create_optimizer()

for _ in range(20):
    outputs = trainer.model(**batch)
    loss = outputs.loss
    loss.backward()
    trainer.optimizer.step()
    trainer.optimizer.zero_grad()

with torch.no_grad():
    outputs = trainer.model(**batch)
preds = outputs.logits
labels = batch["labels"]

compute_metrics((preds.cpu().numpy(), labels.cpu().numpy()))