# DialoGPT fine-tuning dataset format

**URL:** <https://discuss.huggingface.co/t/dialogpt-fine-tuning-dataset-format/5682>\
**Category:** Models\
**Created:** [April 22, 2021, 10:26am UTC](https://discuss.huggingface.co/t/dialogpt-fine-tuning-dataset-format/5682 "2021-04-22T10:26:41Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![HatterTheMad](https://avatars.discourse-cdn.com/v4/letter/h/91b2a8/32.png) [@HatterTheMad](https://discuss.huggingface.co/u/HatterTheMad)\
**Post date:** [April 22, 2021, 10:26am UTC](https://discuss.huggingface.co/t/dialogpt-fine-tuning-dataset-format/5682/1 "2021-04-22T10:26:42Z")

</div>

I experience issues with training the model directly from the .csv file, I think I am doing something wrong with the formatting. Can somebody share an example of the format?

**@patrickvonplaten** by any chance?

---

<div class="post-metadata">

**Author:** ![HatterTheMad](https://avatars.discourse-cdn.com/v4/letter/h/91b2a8/32.png) [@HatterTheMad](https://discuss.huggingface.co/u/HatterTheMad)\
**Post date:** [April 22, 2021, 11:10pm UTC](https://discuss.huggingface.co/t/dialogpt-fine-tuning-dataset-format/5682/2 "2021-04-22T23:10:52Z")

</div>

May be you can help @julien-c ?

---

<div class="post-metadata">

**Author:** ![julien-c](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/julien-c/32/41937_2.png) [@julien-c](https://discuss.huggingface.co/u/julien-c)\
**Post date:** [April 23, 2021, 10:16am UTC](https://discuss.huggingface.co/t/dialogpt-fine-tuning-dataset-format/5682/3 "2021-04-23T10:16:33Z")

</div>

Which script or service (AutoNLP?) are you using to train?

---

<div class="post-metadata">

**Author:** ![HatterTheMad](https://avatars.discourse-cdn.com/v4/letter/h/91b2a8/32.png) [@HatterTheMad](https://discuss.huggingface.co/u/HatterTheMad)\
**Post date:** [April 27, 2021, 4:59pm UTC](https://discuss.huggingface.co/t/dialogpt-fine-tuning-dataset-format/5682/4 "2021-04-27T16:59:03Z")

</div>

Sorry. I am not sure I understand it correctly. I am trying to appropriate this [git](https://github.com/Xirider/finetune-gpt2xl) for my model’s training. I use “DeepSpeed” is that what you are looking for?