Update programmatic usage example premises

Fix Space link: fol-slm-space -> fol-slm-demo

Remove bold from limitation bullet

Add limitation: predicate hallucination on large premise sets

Update results: 85.8% premises-FOL, 91.1% question exact match (full 66k test set)

Update model card: fix LR schedule, dataset description, limitations, citations

Fix task categories in model card

Add model card

Upload checkpoint_final.pt with huggingface_hub

initial commit