language: "en"
tags:
- buy-intent
- sell-intent
- consumer-intent
widget:
- text: "Flutoprazepam (Restas) is a drug which is a benzodiazepine. It was patented in Japan by Sumitomo."
Chemical domain language model finetuned on 13K Chemical, and 14K Pharma Wikipedia articles broken into paragraphs.
Train Loss | Validation Acc. | Test Acc. |
---|---|---|
0.17 | 0.928 | 0.927 |
# Dataset | ||
Dataset with splits can be found @ https://www.kaggle.com/shahrukhkhan/pharma-vs-chemicals-domain-classification |
LABEL_0 => "PHARMACEUTICAL"
LABEL_1 => "CHEMICAL"
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("recobo/chemical-bert-uncased-pharmaceutical-chemical-classifier")
model = AutoModelForSequenceClassification.from_pretrained("recobo/chemical-bert-uncased-pharmaceutical-chemical-classifier")