I am trying to Finetune model.
Here is the train-test split of my dataset – Train – 4746 (80%)
Test – 1188 (20%)
Here is my code snippet:

training_args = TrainingArguments(
    bf16=True, # specify bf16=True instead when training on GPUs that support bf16
    do_eval=True,
    evaluation_strategy="epoch", #ch
    gradient_accumulation_steps=64,
    gradient_checkpointing=True, #ch
    gradient_checkpointing_kwargs={"use_reentrant": False},
    learning_rate=2.0e-05,
    log_level="info",
    logging_steps=5,
    logging_strategy="steps",
    lr_scheduler_type="cosine", #ch
    max_steps=-1,
    num_train_epochs=3,
    output_dir=output_dir,
    overwrite_output_dir=True,
    per_device_eval_batch_size=2, # originally set to 8
    per_device_train_batch_size=2, # originally set to 8
    save_strategy="no", #ch
    save_total_limit=None,
    seed=42,
)
peft_config = LoraConfig(
        r=64,
        lora_alpha=16,
        lora_dropout=0.1,
        bias="none",
        task_type="CAUSAL_LM",
        target_modules=["q_proj", "k_proj", "v_proj", "o_proj"],
)
trainer = SFTTrainer(
        model=model_id,
        model_init_kwargs=model_kwargs,
        args=training_args,
        train_dataset=train_set,
        eval_dataset=val_set,
        dataset_text_field="text",
        tokenizer=tokenizer,
        packing=True,
        peft_config=peft_config,
        max_seq_length=tokenizer.model_max_length,
    )

But when I run training, this is what is logged –

***** Running training *****
  Num examples = 1,060
  Num Epochs = 3
  Instantaneous batch size per device = 2
  Total train batch size (w. parallel, distributed & accumulation) = 128
  Gradient Accumulation steps = 64
  Total optimization steps = 24
  Number of trainable parameters = 54,525,952
***** Running Evaluation *****
  Num examples = 264
  Batch size = 2

Why are my actual train-test sizes not matching with the logged ones?