amigo-mini

SmolLM2-360M-Instruct, LoRA fine-tuned for Amigo, a free study buddy. Runs in the browser with transformers.js (dtype: "q8"). Trained only on Amigo's own rewritten study notes and answers written by an open-weight teacher (Qwen2.5-7B-Instruct, Apache-2.0) from commercial-safe sources. No output of any proprietary model was used. Small model: it can make mistakes, so Amigo always shows the sources it used.

Downloads last month
114
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for maphaha7/amigo-mini

Quantized
(134)
this model