SauerkrautLM-LFM2.5-GLiNER — Zero-Shot NER
A compact 350M bidirectional LFM2.5 model for zero-shot Named Entity Recognition. Enter any text and a comma-separated list of entity types — the model extracts matching spans without retraining.
Supports English, French, German, Italian, and Spanish. Strong on general NER, PII/privacy, and biomedical entities.
Links
- Model: VAGOsolutions/SauerkrautLM-LFM2.5-GLiNER
- GLiNER library: github.com/urchade/GLiNER
- Paper: arxiv.org/abs/2311.08526
0 1
Allow overlapping entity spans.
Examples