openhermes-7b-dpo/README.md

license

license
apache-2.0

New experimental dpo tune. More details to come...

odel description OpenHermes 2.5 Mistral 7B is a state of the art Mistral Fine-tune, a continuation of OpenHermes 2 model, which trained on additional code datasets.

Potentially the most interesting finding from training on a good ratio (est. of around 7-14% of the total dataset) of code instruction was that it has boosted several non-code benchmarks, including TruthfulQA, AGIEval, and GPT4All suite. It did however reduce BigBench benchmark score, but the net gain overall is significant.

Here, we are finetuning openheremes using DPO with various data meant to improve its abilities.