Instructions to use Finesse4002/yahoo-answers-bert with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers.js
How to use Finesse4002/yahoo-answers-bert with Transformers.js:
// npm i @huggingface/transformers import { pipeline } from '@huggingface/transformers'; // Allocate pipeline const pipe = await pipeline('text-classification', 'Finesse4002/yahoo-answers-bert');
Yahoo! Answers topic classification (Base BERT, fine-tuned)
10-class classifier for Yahoo! Answers questions (title + content): Society & Culture, Science & Mathematics, Health, Education & Reference, Computers & Internet, Sports, Business & Finance, Entertainment, Family & Relationships, Politics & Government.
Winner of 11 models trained in the CSE440 lab project notebook:
test 70.02% accuracy / 0.696 macro-F1 (10k stratified official-test subsample).
This copy is int8-quantized ONNX for in-browser inference with
transformers.js (dtype q8),
reproducing the torch model within 0.1 pt (0.6990 measured).
Training: lr=2e-5, batch=32, 2 epochs, seed 42, lemmatized text, max_len=96.
- Downloads last month
- 23
Model tree for Finesse4002/yahoo-answers-bert
Base model
google-bert/bert-base-uncased