# Finetuning T5 on SQUADv2 with Seq2SeqTrainer fails

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ffinetuning-t5-on-squadv2-with-seq2seqtrainer-fails%2F43771%3E%5C
**Category:** 🤗Transformers\
**Created:** [June 19, 2023, 10:17am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ffinetuning-t5-on-squadv2-with-seq2seqtrainer-fails%2F43771 "2023-06-19T10:17:42Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![pandemodeus](https://avatars.discourse-cdn.com/v4/letter/p/f1d935/32.png) [@pandemodeus](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fpandemodeus)%5C
**Post date:** [June 19, 2023, 10:17am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ffinetuning-t5-on-squadv2-with-seq2seqtrainer-fails%2F43771%2F1 "2023-06-19T10:17:42Z")

</div>

Hi,  
As specified in the topic name, I’m trying to finetune T5 with HF build-in Seq2SeqTrainer as provided here: [transformers/examples/pytorch/question-answering/README.md at main · huggingface/transformers · GitHub](https://github.com/huggingface/transformers/blob/main/examples/pytorch/question-answering/README.md)  
However, evaluation fails with the KeyError  
 ![Снимок экрана 2023-06-19 в 13.15.48](https://us1.discourse-cdn.com/hellohellohello/original/3X/0/0/004231d2f77da70dd3fac635a256d7db56e21869.png)

I’m running the train process with the command below:  
python run\_seq2seq\_qa.py --model\_name\_or\_path bigscience/mt0-large --dataset\_name squad\_v2 --context\_column context --question\_column question --answer\_column answers --do\_train --do\_eval --per\_device\_train\_batch\_size 4 --learning\_rate 3e-5 --num\_train\_epochs 2 --max\_seq\_length 384 --doc\_stride 128 --output\_dir ~/tmp --pad\_to\_max\_length False --max\_train\_samples 100 --max\_eval\_samples 100 --version\_2\_with\_negative True --n\_best\_size 5 --overwrite\_output\_dir True --evaluation\_strategy steps --prediction\_loss\_only False --log\_level info --logging\_dir ~/mt0\_logs --report\_to tensorboard --logging\_strategy steps --logging\_steps 50 --save\_strategy no --disable\_tqdm False --predict\_with\_generate True

How could I resolve this issue?

---

<div class="post-metadata">

**Author:** ![pandemodeus](https://avatars.discourse-cdn.com/v4/letter/p/f1d935/32.png) [@pandemodeus](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fpandemodeus)%5C
**Post date:** [June 20, 2023, 11:10am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ffinetuning-t5-on-squadv2-with-seq2seqtrainer-fails%2F43771%2F2 "2023-06-20T11:10:37Z")

</div>

@sgugger @valhalla Guys, could you probably help me?