# Reproducibility of LongFormer Model

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Freproducibility-of-longformer-model%2F104619%3E%5C
**Category:** 🤗Transformers\
**Created:** [August 30, 2024, 6:51am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Freproducibility-of-longformer-model%2F104619 "2024-08-30T06:51:01Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![ambrishgupta](https://avatars.discourse-cdn.com/v4/letter/a/46a35a/32.png) [@ambrishgupta](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fambrishgupta)%5C
**Post date:** [August 30, 2024, 6:51am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Freproducibility-of-longformer-model%2F104619%2F1 "2024-08-30T06:51:02Z")

</div>

I am unable to reproduce the same results over different model runs.

When using `bert-base-uncased` , the generated results (and loss values) will be _exactly_ the same across runs. However, when using `allenai/longformer-base-4096` results (and loss values) will vary across runs.

Please help me. This is urgent.