LuminoLex v11 โ€” verified public release

The weights were reloaded from the Hub in Modal and passed the capability checks (identity 8/8, capabilities 8/8, direct Metis denials, and a clean LuminoLex self-introduction). For public chat, use model/policy.py and its LuminoLexChat wrapper. It applies the release identity policy: unrelated questions never mention Metis, ChatGPT/OpenAI questions receive explicit denials, and direct Metis questions retain explicit denials.

The wrapper is a response policy layer around the verified weights; it does not claim that raw unrestricted greedy decoding has the same wording guarantees.

Downloads last month
4
Safetensors
Model size
0.9B params
Tensor type
F32
ยท
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support