ValueError: Expected input batch_size (8) to match target batch_size (280)

Beginners
1

Hi folks,

I’m newbie and got stuck by this error msg “ValueError: Expected input batch_size (8) to match target batch_size (280)”.

Appreciate any advice, please… Thank you!

from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")

def preprocess_function(examples):
    model_inputs = tokenizer(
        text = examples["text"],
        max_length=max_text_length,
        truncation=True,
        padding = True,
    )
    label = tokenizer(examples["label"], max_length=max_label_length, truncation=True, padding = True)
    model_inputs["label"] = label["input_ids"]
    return model_inputs

tokenized_aicgp = aicgp.map(preprocess_function, batched=True)

from transformers import DataCollatorWithPadding
data_collator = DataCollatorWithPadding(tokenizer=tokenizer)

from transformers import AutoModelForSequenceClassification, TrainingArguments, Trainer
model = AutoModelForSequenceClassification.from_pretrained("distilbert-base-uncased", num_labels=no_of_labels, id2label=id2label, label2id=label2id)


training_args = TrainingArguments(
    output_dir="my_model",
    learning_rate=2e-5,
    per_device_train_batch_size=8, #16
    per_device_eval_batch_size=8, #16
    num_train_epochs=2,
    weight_decay=0.01,
    evaluation_strategy="epoch",
    save_strategy="epoch",
    load_best_model_at_end=True,
    push_to_hub=False,
)
trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=tokenized_aicgp["train"],
    eval_dataset=tokenized_aicgp["test"],
    tokenizer=tokenizer,
    data_collator=data_collator,
    compute_metrics=compute_metrics,
)
trainer.train()


Note:

The program aborted at trainer.train() with the below error message:
ValueError: Expected input batch_size (8) to match target batch_size (280).


tokenized_aicgp

DatasetDict({
    train: Dataset({
        features: ['label', 'text', 'lblId', 'input_ids', 'attention_mask'],
        num_rows: 56
    })
    test: Dataset({
        features: ['label', 'text', 'lblId', 'input_ids', 'attention_mask'],
        num_rows: 14
    })
})

length label 56
length text 56
length lblId 56
length input_ids 56
length attention_mask 56
2

Work for me:

def tokenize_function(examples):
    inputs = tokenizer(examples["text"] , truncation=True , padding="max_length" , max_length=128 )
    inputs['labels'] = inputs['input_ids'].copy()
    return inputs

See an example: Fine-Tuning a Pre-Trained GPT-2 Model and Performing Inference: A Hands-On Guide | by prashanth ramanathan | Medium

1 Like