These are llama.cpp quants for
Taurus 7B 1.0.
Taurus is an
OpenHermes 2.5 finetune using the
Economicus dataset, an instruct dataset synthetically generated from Economics PhD textbooks.
The model was trained for 2 epochs (QLoRA) using
axolotl. The exact config I used can be found
here.
<|im_start|>system
System message
<|im_start|>user
User message<|im_end|>
<|im_start|>assistant