# ValueError with Trainer

**URL:** <https://discuss.huggingface.co/t/valueerror-with-trainer/52678>\
**Category:** Beginners\
**Created:** [August 29, 2023, 2:47pm UTC](https://discuss.huggingface.co/t/valueerror-with-trainer/52678 "2023-08-29T14:47:12Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![deadbod-81](https://avatars.discourse-cdn.com/v4/letter/d/c2a13f/32.png) [@deadbod-81](https://discuss.huggingface.co/u/deadbod-81)\
**Post date:** [August 29, 2023, 2:47pm UTC](https://discuss.huggingface.co/t/valueerror-with-trainer/52678/1 "2023-08-29T14:47:12Z")

</div>

Getting the following error when running trainer.train()

ValueError: You have to specify either input\_ids or inputs\_embeds

Here is my trainer

trainer = Trainer(  
model=model,  
args=training\_args,  
train\_dataset=train\_test\_valid\_dataset[“train”],  
eval\_dataset=train\_test\_valid\_dataset[“test”],  
compute\_metrics=compute\_metrics  
)

Using the AutoTkenizer

from transformers import AutoTokenizer

tokenizer = AutoTokenizer.from\_pretrained(“distilbert-base-uncased”)

def tokenize\_function(examples):  
return tokenizer(examples[“title”], padding=“max\_length”, truncation=True)

tokenized\_datasets = train\_test\_valid\_dataset.map(tokenize\_function, batched=True)

Can someone help out here ?