This model seems to be closest to Hermes1 and gpt4-x-vicuna-13B LLAMA1 models
#1
by
boqsc
- opened
This model seems to be closest to Hermes1 and gpt4-x-vicuna-13B LLAMA1 models,
if they were merged one day.
Model feels uncensored, but still lacks abilities in good concise detailed storytelling.
However a good start. As most LLAMA2 models are just horrible at it.
Also unlike gpt4-x-vicuna-13B LLAMA1, this model is not as good at math and numbers.
Unlike both LLAMA1 models previously mentioned, it does not have any cultural information like in depth understanding of Lord of the Rings world.
It also not as good at guiding through story unlike gpt4-x-vicuna-13B LLAMA1 model.
Overall story generation is short and avoiding on what's asked, however if questioned it will provide some content that you would expect from a good concise story.