metadata
language: en
license: apache-2.0
tags:
- materials-science
- electrocatalyst
- ner
- token-classification
base_model: m3rg-iitd/matscibert
BatteryBERT Electrocatalyst NER v2
Fine-tuned MatSciBERT for Named Entity Recognition in electrocatalyst literature.
Accuracy: 99.90% F1 on validation, 100% on real-world test papers
Usage
from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline
tokenizer = AutoTokenizer.from_pretrained("Dmjdxb/batterybert-electrocatalyst-ner-v2")
model = AutoModelForTokenClassification.from_pretrained("Dmjdxb/batterybert-electrocatalyst-ner-v2")
ner = pipeline("ner", model=model, tokenizer=tokenizer, aggregation_strategy="simple")
text = "IrO₂ showed dissolution in 0.5 M H₂SO₄ at pH 0.3"
entities = ner(text)
Project: DurabilityGraph-AI Repository: https://github.com/dmjdxb/ElectrocatalystAI