How to Efficiently Fine-Tune Models on Custom Datasets with Limited Resources?

Hi everyone,

I’m relatively new to Hugging Face and I’m trying to fine-tune some pre-trained models on my custom dataset. However, I’m running into resource constraints, especially with memory and computational power. I couldn’t find detailed documents on this topic.

Could anyone share some strategies or best practices for efficiently fine-tuning models on limited hardware? Are there specific settings or tools within Hugging Face that can help optimize this process? Any advice or examples would be greatly appreciated! Also, where can I find more documentation on this?

Thanks you.