# FineTuning Reimbursement Documents using 3.5 gpt turbo model

**URL:** <https://discuss.huggingface.co/t/finetuning-reimbursement-documents-using-3-5-gpt-turbo-model/136060>\
**Category:** Beginners\
**Created:** [January 16, 2025, 11:06am UTC](https://discuss.huggingface.co/t/finetuning-reimbursement-documents-using-3-5-gpt-turbo-model/136060 "2025-01-16T11:06:45Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Pashi](https://avatars.discourse-cdn.com/v4/letter/p/f05b48/32.png) [@Pashi](https://discuss.huggingface.co/u/Pashi)\
**Post date:** [January 16, 2025, 11:06am UTC](https://discuss.huggingface.co/t/finetuning-reimbursement-documents-using-3-5-gpt-turbo-model/136060/1 "2025-01-16T11:06:45Z")

</div>

We have been using prompt engineering where prompt works on OCR performed on Reimbursement documents in an organisation.

Although after multiple prompt refining the answer is not always correct.  
We also require additional manual validation of results.

For this we are thinking to explore fine tune model.  
Can you give me some insights regarding the same