# Adding additional features to BERT model

**URL:** <https://discuss.huggingface.co/t/adding-additional-features-to-bert-model/20528>\
**Category:** Models\
**Created:** [July 18, 2022, 7:02pm UTC](https://discuss.huggingface.co/t/adding-additional-features-to-bert-model/20528 "2022-07-18T19:02:12Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Apple11](https://avatars.discourse-cdn.com/v4/letter/a/9fc29f/32.png) [@Apple11](https://discuss.huggingface.co/u/Apple11)\
**Post date:** [July 18, 2022, 7:02pm UTC](https://discuss.huggingface.co/t/adding-additional-features-to-bert-model/20528/1 "2022-07-18T19:02:12Z")

</div>

Hello All,

I am beginner and have really less experience in training transformers.I have gone through the huggingface course for training transformers and now I am trying to use 3 text input features to finetune a BERT-cased model for sequence classification. I tried using the 2 text features and feeding them to the auto tokenizer and then feeding the output to the model for training. For 2 input features, the function of the tokenizer looks as below -  
def tokenize\_function(examples):  
return tokenizer(examples[“feature\_1”], examples[“feature\_2”], padding=“max\_length”, truncation=True)

I was trying to add the third feature like -  
def tokenize\_function(examples):  
return tokenizer(examples[“feature\_1”],examples[“feature\_2”],  
examples[“feature\_3”],padding=“max\_length”, truncation=True)  
but it was returning an error that says, "cant convert [“feature\_3”].

Could you please help me understand, if there is a simple way to add the 3 text inputs to bert model for training.