# Expected input batch\_size (2048) to match target batch\_size (4)

**URL:** <https://discuss.huggingface.co/t/expected-input-batch-size-2048-to-match-target-batch-size-4/15610>\
**Category:** Beginners\
**Created:** [March 11, 2022, 2:40pm UTC](https://discuss.huggingface.co/t/expected-input-batch-size-2048-to-match-target-batch-size-4/15610 "2022-03-11T14:40:53Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![AgelZin](https://avatars.discourse-cdn.com/v4/letter/a/6bbea6/32.png) [@AgelZin](https://discuss.huggingface.co/u/AgelZin)\
**Post date:** [March 11, 2022, 2:40pm UTC](https://discuss.huggingface.co/t/expected-input-batch-size-2048-to-match-target-batch-size-4/15610/1 "2022-03-11T14:40:53Z")

</div>

> !pip install -q transformers datasets sklearn
> 
> # imports
> 
> import numpy as np  
> import pandas as pd  
> import datasets  
> import torch
> 
> from transformers import BertTokenizer, BertForTokenClassification, Trainer, TrainingArguments, AutoConfig  
> from sklearn.preprocessing import LabelEncoder  
> from datasets import load\_dataset  
> from random import shuffle  
> from sklearn.metrics import f1\_score  
> from random import shuffle

> dataset = datasets.load\_dataset(‘yelp\_polarity’,split=‘train[800:1000]’) # [:100] [:1%]

> tokenizer = BertTokenizer.from\_pretrained(‘bert-base-uncased’, do\_lower\_case=True)
> 
> encoded\_dataset = [tokenizer(item[‘text’], return\_tensors=“pt”, padding=‘max\_length’, truncation=True, max\_length=512) for item in dataset]
> 
> import torch  
> for enc\_item, item in zip(encoded\_dataset, dataset):  
> enc\_item[‘labels’] = torch.LongTensor([item[‘label’]])
> 
> for key, val in encoded\_dataset[3].items():  
> print(f’key: {key}, content: {val.size()}')
> 
> model = BertForTokenClassification.from\_pretrained(‘bert-base-uncased’)

> for item in encoded\_dataset:  
> for key in item:  
> item[key] = torch.squeeze(item[key])
> 
> train\_set = encoded\_dataset[:100]  
> test\_set = encoded\_dataset[100:]
> 
> training\_args = TrainingArguments(  
> num\_train\_epochs=5,  
> per\_device\_train\_batch\_size=4,  
> per\_device\_eval\_batch\_size=4,  
> output\_dir=‘results’,  
> logging\_dir=‘logs’,  
> no\_cuda=True,  
> )
> 
> trainer = Trainer(  
> model=model,  
> tokenizer=tokenizer,  
> args=training\_args,  
> train\_dataset=train\_set  
> )  
> trainer.train()

Getting this error when i train  
Expected input batch\_size (2048) to match target batch\_size (4)

---

<div class="post-metadata">

**Author:** ![anwarika](https://avatars.discourse-cdn.com/v4/letter/a/e0b2c6/32.png) [@anwarika](https://discuss.huggingface.co/u/anwarika)\
**Post date:** [March 11, 2022, 5:05pm UTC](https://discuss.huggingface.co/t/expected-input-batch-size-2048-to-match-target-batch-size-4/15610/2 "2022-03-11T17:05:55Z")

</div>

can you supply stacktrace of the error?

---

<div class="post-metadata">

**Author:** ![AgelZin](https://avatars.discourse-cdn.com/v4/letter/a/6bbea6/32.png) [@AgelZin](https://discuss.huggingface.co/u/AgelZin)\
**Post date:** [March 11, 2022, 5:09pm UTC](https://discuss.huggingface.co/t/expected-input-batch-size-2048-to-match-target-batch-size-4/15610/3 "2022-03-11T17:09:24Z")

</div>

I am trying other methods now so i dont have it

---

<div class="post-metadata">

**Author:** ![kullackaan](https://avatars.discourse-cdn.com/v4/letter/k/ce7236/32.png) [@kullackaan](https://discuss.huggingface.co/u/kullackaan)\
**Post date:** [May 23, 2022, 2:55pm UTC](https://discuss.huggingface.co/t/expected-input-batch-size-2048-to-match-target-batch-size-4/15610/4 "2022-05-23T14:55:02Z")

</div>

Which method did you use instead?