Transformers
Safetensors
English
scientific-claim-verification
scifact
evidence-selection
sentence-reranking
Instructions to use rishhh/verisci-scifact-sentence-reranker-seed123 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use rishhh/verisci-scifact-sentence-reranker-seed123 with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("rishhh/verisci-scifact-sentence-reranker-seed123", device_map="auto") - Notebooks
- Google Colab
- Kaggle
VeriSci SciFact Sentence Reranker
Experimental cross-encoder evidence sentence reranker for SciFact claims.
- Reranker base:
microsoft/deberta-v3-small - Validation candidate policy: retrieved-document lexical candidates only; gold validation sentences are not injected.
- SciFact validation any evidence sentence recall@6: 0.6863
- SciFact validation complete evidence sentence recall@6: 0.6503
- Promotion gate passed:
False - Promotion gate reason:
candidate_any_recall@6_below_target: 0.686 < 0.800
This artifact is experimental unless the repository documentation marks it as promoted.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support