# Inference/prediction ValueError using BART

**URL:** <https://discuss.huggingface.co/t/inference-prediction-valueerror-using-bart/16912>\
**Category:** 🤗Transformers\
**Created:** [April 17, 2022, 11:22pm UTC](https://discuss.huggingface.co/t/inference-prediction-valueerror-using-bart/16912 "2022-04-17T23:22:18Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Simzalabim](https://avatars.discourse-cdn.com/v4/letter/s/47e85d/32.png) [@Simzalabim](https://discuss.huggingface.co/u/Simzalabim)\
**Post date:** [April 17, 2022, 11:22pm UTC](https://discuss.huggingface.co/t/inference-prediction-valueerror-using-bart/16912/1 "2022-04-17T23:22:18Z")

</div>

Hello there,

I just tried to summarize a text from my dataset using a custom fine-tuned BART model. To this end I followed the pytorch example listed here [Summary of the tasks](https://huggingface.co/docs/transformers/task_summary#summarization)

This is the code that was used. And further below, the error is shown. Any ideas what could be the problem here?

```auto
from transformers import BartForConditionalGeneration, BartTokenizerFast
from transformers import Seq2SeqTrainer, Seq2SeqTrainingArguments
from datasets import load_dataset

raw_dataset = load_dataset(path='parquet', data_files={
    'train': ['2021_1.parquet', '2021_2.parquet', '2021_3.parquet'], 
    'test': ['2021_4.parquet'
]})

model = BartForConditionalGeneration.from_pretrained("bart_nl_sum_17-04_15-50-23/checkpoints/checkpoint-10000")
tokenizer = BartTokenizerFast.from_pretrained("bart_nl_tiny_tz")

model_inputs = tokenizer(raw_dataset['train'][2]['description'], max_length=1024, return_tensors='pt', truncation=True) # padding=max_length

model_outputs = model.generate(
    inputs=model_inputs["input_ids"], 
    max_length=150, 
    min_length=40, 
    length_penalty=2.0, 
    num_beams=4, 
    early_stopping=True
)

print(tokenizer.decode(model_outputs[0]))

```

```auto
Traceback (most recent call last):
  File "legalsum_inf.py", line 22, in <module>
    early_stopping=True
  File "/opt/conda/lib/python3.7/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "/opt/conda/lib/python3.7/site-packages/transformers/generation_utils.py", line 1325, in generate
    **model_kwargs,
  File "/opt/conda/lib/python3.7/site-packages/transformers/generation_utils.py", line 2162, in beam_search
    output_hidden_states=output_hidden_states,
  File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
    return forward_call(*input, **kwargs)
  File "/opt/conda/lib/python3.7/site-packages/transformers/models/bart/modeling_bart.py", line 1363, in forward
    return_dict=return_dict,
  File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
    return forward_call(*input, **kwargs)
  File "/opt/conda/lib/python3.7/site-packages/transformers/models/bart/modeling_bart.py", line 1224, in forward
    return_dict=return_dict,
  File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
    return forward_call(*input, **kwargs)
  File "/opt/conda/lib/python3.7/site-packages/transformers/models/bart/modeling_bart.py", line 793, in forward
    raise ValueError("You have to specify either input_ids or inputs_embeds")
ValueError: You have to specify either input_ids or inputs_embeds

```