Drug Causality BERT (v1 Legacy Model)
[!IMPORTANT] Production Model Available: This is the version 1.0 (legacy) model weights repository. For all new research, benchmark replication, and production deployments, please use drug-causality-bert-v2-model, which incorporates Optuna-optimized hyperparameters and achieves 97.6% accuracy on the ADE Corpus V2 benchmark.
Model Summary
- Architecture: BioBERT fine-tuned for sequence classification (binary)
- Task: Adverse drug reaction (ADE) causality identification in clinical narratives
- Base Pretrained Model: dmis-lab/biobert-base-cased-v1.2
- Primary Successor: PrashantRGore/drug-causality-bert-v2-model
Usage
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("PrashantRGore/drug-causality-bert-model")
model = AutoModelForSequenceClassification.from_pretrained("PrashantRGore/drug-causality-bert-model")
Related Resources
- 🚀 Latest Model (v2): PrashantRGore/drug-causality-bert-v2-model
- 📊 Benchmark Dataset: PrashantRGore/synthetic-faers-1m-v3
- 🐙 GitHub Repository: PrashantRGore/drug-causality-bert-v2
License
This model is distributed under the Apache-2.0 License.