Neural-una-cybertron-7b is an
fblgit/una-cybertron-7b-v2-bf16 model that has been further fine-tuned with Direct Preference Optimization (DPO) using the
Intel/orca_dpo_pairs dataset.
This model was created after examining the procedure of
mlabonne/NeuralHermes-2.5-Mistral-7B model. Special thanks to
@mlabonne.
The total training time was 1 hour and 10 minutes.
<|im_start|>system
{system}<|im_end|>
<|im_start|>user
{user}<|im_end|>
<|im_start|>assistant
{asistant}<|im_end|>
