teknium

/

OpenHermes-2.5-Mistral-7B

like 776

Hello Teknium

I am a huge fan of yours. Quick question: what dataset did you use to finetune this SFT model?

You said:

OpenHermes was trained on 1,000,000 entries of primarily GPT-4 generated data, as well as other high quality data from open datasets across the AI landscape. **** [More details soon] ****

Is it possible to tell other open source dataset (if the 1M entries of GPT-4 is not available to us). Thanks!

I'll be releasing the full dataset very soon :)

For those interested, the full dataset for this amazing model got released here: https://huggingface.co/datasets/teknium/OpenHermes-2.5