AnalysisAI ModelsNovember 21, 2026

PhiloLM: a 112M model trained only on philosophy texts

Read original source →reddit.com

PhiloLM is a 112M-parameter base model trained from scratch on PhiloText, a 0.5B-token corpus of open philosophy encyclopedias including SEP and IEP plus public-domain texts. Weights are on Hugging Face under gustavecortal.

1 source

More stories today

Open the live feed