# mT5-base model summary language issue

**URL:** <https://discuss.huggingface.co/t/mt5-base-model-summary-language-issue/16294>\
**Category:** 🤗Transformers\
**Created:** [March 31, 2022, 6:17am UTC](https://discuss.huggingface.co/t/mt5-base-model-summary-language-issue/16294 "2022-03-31T06:17:31Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Doogie](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/doogie/32/12892_2.png) [@Doogie](https://discuss.huggingface.co/u/Doogie)\
**Post date:** [March 31, 2022, 6:17am UTC](https://discuss.huggingface.co/t/mt5-base-model-summary-language-issue/16294/1 "2022-03-31T06:17:31Z")

</div>

Hello  
I’m use load a pre-train mT5 base model from Huggingface.  
I’m adding English and Korean Train datasets this model  
In the process of summarization of mT5 tasks,  
When I only trained English, model’s output has no problem  
But training with a Korean dataset, if i input English, the results will come out in Korean + English.  
I would like to summarize and print them according to each input language, but how should I solve them?

---

<div class="post-metadata">

**Author:** ![mHossain](https://avatars.discourse-cdn.com/v4/letter/m/4491bb/32.png) [@mHossain](https://discuss.huggingface.co/u/mHossain)\
**Post date:** [October 11, 2023, 5:10pm UTC](https://discuss.huggingface.co/t/mt5-base-model-summary-language-issue/16294/2 "2023-10-11T17:10:03Z")

</div>

is your issue solved?

i am trying to cross lingual summarization. like i will give input in english output will be bengali summarization but while pretraining output always are in english. how can i control this issue ?