Trainer does not show epochs or steps just 1 line without numbers

Course
1 
args = TrainingArguments(
output_dir=tokenizer_path / "model",
per_device_train_batch_size=32,
per_device_eval_batch_size=32,
evaluation_strategy="steps",
eval_steps=5000,
logging_steps=5000,
gradient_accumulation_steps=8,
num_train_epochs=100, # Először egy epochot nézzünk meg, hogy minden stimmel-e
weight_decay=0.1,
warmup_steps=1000,
lr_scheduler_type="cosine",
learning_rate=5e-4,
save_steps=5000,
fp16=True, # Lower precision only allowed on Cuda devices
#push_to_hub=True, Nem akarom a hub-ra feltennie
push_to_hub= False
)

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

output = trainer.train()

![image|617x292](upload://jnQMc6SchOY5E2j3Hpk5pyUjPD7.png)


trainer.state.log_history
>>> [{'train_runtime': 2240.0086,
  'train_samples_per_second': 200.758,
  'train_steps_per_second': 0.759,
  'total_flos': 1.1068167168e+16,
  'train_loss': 1.500028435202206,
  'epoch': 96.45,
  'step': 1700}]

It is clear that there is only one element of the log_history. What did I make wrong?

Thanks in advance