TypeError: TrainingArguments got an unexpected keyword argument ‘evaluation_strategy’

This error gave me a headache. This is for future. I am writing the solution here.

Step 1: Make sure you are using the latest version of Transformers

Step 2: Instead of ‘evaluation_strategy’, Use eval_strategy.

training_args = TrainingArguments(
    output_dir="./results",
    num_train_epochs=5,              
    eval_strategy="epoch",
    eval_steps=100,
    logging_strategy="epoch",
    logging_steps=50,
    save_strategy="epoch",
    learning_rate=2e-5,
    weight_decay=0.01,
    warmup_ratio=0.1,
    gradient_accumulation_steps=2,
    max_grad_norm=1.0,
    fp16=True,
    load_best_model_at_end=True,
    metric_for_best_model="eval_loss",
)

If Above still doesn’t work, then use this one:

# 1. Legacy-style TrainingArguments
training_args = TrainingArguments(
    output_dir="./results",
    num_train_epochs=5,
    per_device_train_batch_size=16,
    per_device_eval_batch_size=16,
    learning_rate=2e-5,
    weight_decay=0.01,
    warmup_ratio=0.1,
    gradient_accumulation_steps=2,
    max_grad_norm=1.0,
    fp16=True,
    load_best_model_at_end=True,
    metric_for_best_model="eval_loss",
    eval_strategy="epoch",       # legacy alias for evaluation_strategy
    eval_steps=100,
    logging_strategy="epoch",    # legacy alias for logging_strategy
    logging_steps=50,
    save_strategy="epoch",       # legacy alias for save_strategy
)