by ggml-org
Open source · 189k downloads · 6 likes
The *stories15M MOE* model is an experimental version designed to generate short stories or narratives, ideal for playful or creative applications. It leverages a specialized architecture that combines multiple experts to enhance the diversity of responses, though it is not optimized for professional use. Its primary strengths lie in crafting simple narrative texts with a poetic or fantastical touch, as seen in its Shakespeare-inspired responses when paired with a dedicated LoRA adapter. The model stands out for its lightweight design and potential in artistic or educational projects, such as serving as a companion for bedtime storytelling. However, its use remains limited to non-critical contexts due to its experimental nature.
This model is ModelCloud/tinyllama-15M-stories repeated 4 times to make 4 experts.
The model is used for testing, not intended to be used in production (unless your product is some kind of bedtime story teller)
Weight of router is initialized randomly
A LoRA adapter trained on first 100 paragraphs of shakespeare can be found inside moe_shakespeare15M
With input: Look in thy glass
Look in thy glass was a little girl. She was only three years old and she was three years old. She wasLook in thy glass in love of the eye: That's when when the eye see thy on the sun'