Eval error: UnboundLocalError: cannot access local variable 'input_ids' where it is not associated with a value

[apple-1]

After training on local machine, running the eval code:

result= model.eval_model(test_file)

My config:

model_args = LanguageModelingArgs()
model_args.reprocess_input_data = True
model_args.overwrite_output_dir = True
model_args.num_train_epochs = 1
model_args.best_model_dir = "outputs"
model_args.save_best_model =True
model_args.train_batch_size = 64
model_args.dataset_type = "simple"
model_args.mlm = False # mlm must be False for CLM
model_args.vocab_size = 50257

Error message:
-----------------

_UnboundLocalError Traceback (most recent call last)
Cell In[33], line 2
1 # Evaluate the model
----> 2 result= model.eval_model(test_file)

File ~\anaconda3\envs\training_pytorch\Lib\site-packages\simpletransformers\language_modeling\language_modeling_model.py:1367, in LanguageModelingModel.eval_model(self, eval_file, output_dir, evaluate_generated_text, verbose, silent, **kwargs)
1362 eval_dataset = self.load_and_cache_examples(
1363 eval_file, evaluate=True, verbose=verbose, silent=silent
1364 )
1365 os.makedirs(output_dir, exist_ok=True)
-> 1367 result = self.evaluate(
1368 eval_dataset, output_dir, verbose=verbose, silent=silent, **kwargs
1369 )
1370 self.results.update(result)
1372 if verbose:

File ~\anaconda3\envs\training_pytorch\Lib\site-packages\simpletransformers\language_modeling\language_modeling_model.py:1436, in LanguageModelingModel.evaluate(self, eval_dataset, output_dir, prefix, verbose, silent, **kwargs)
1430 if self.args.use_hf_datasets:
1431 input_ids = batch["input_ids"]
1433 inputs, labels = (
1434 mask_tokens(batch, tokenizer, args)
1435 if args.mlm
-> 1436 else (input_ids, input_ids)
1437 )
1438 inputs = inputs.to(self.device)
1439 labels = labels.to(self.device)

UnboundLocalError: cannot access local variable 'input_ids' where it is not associated with a value_