typeof/openhermes-2.5-mistral-lora

Model Details

$$W_{mistral}+LoR{A}_{hermes}=W_{hermes}{W}_{hermes}-LoR{A}_{hermes}=W_{mistral}W\_\{mistral\}\; +\; LoRA\_\{hermes\}\; =\; W\_\{hermes\}\; \backslash \backslash \; W\_\{hermes\}\; -\; LoRA\_\{hermes\}\; =\; W\_\{mistral\}$$

Why Though?

unfortunately this is not as simple as typeof/zephyr-7b-beta-lora due to the way in which OpenHermes-2.5-Mistral-7B was trained... by adding tokens, the corresponance is not 1-to-1 with mistralai/Mistral-7B-v0.1 as is the case with typeof/zephyr-7b-beta-lora ... nevertheless, if you have found yourself here, I'm sure you can figure out how to use it... if not, open up an issue!

photo courtesy @teknium OpenHermes-2.5-Mistral-7B was trained...

How to Get Started with the Model

Use the code below to get started with the model.

[More Information Needed]

Summary

A fine-tuned version of mistralai/Mistral-7B-v0.1

LoRA

QLoRA