Instructions to use rolf-mozilla/tinybert-address-autofill with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers.js
How to use rolf-mozilla/tinybert-address-autofill with Transformers.js:
// npm i @huggingface/transformers import { pipeline } from '@huggingface/transformers'; // Allocate pipeline const pipe = await pipeline('text-classification', 'rolf-mozilla/tinybert-address-autofill');
File size: 599 Bytes
f06031d 9d7f96b f06031d 9d7f96b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 | ---
license: mpl-2.0
base_model: huawei-noah/TinyBERT_General_4L_312D
tags: [autofill, form-field-classification, onnx, transformers.js]
---
# tinybert-address-autofill (v0.2.2)
Firefox address-autofill field classifier, retrained from the jt4qd v2 baseline
**without the `**` input-type marker** (`strip_type_marker=True`) so training
matches Firefox inference. jt4qd params (bb, encoder_layers=0, pooling=cls,
lr=3e-5, batch=64, epochs=25). **Training data:** current/refreshed
train/val/test. **Offline (no-`**`):** total 0.8897 / close 0.9133.
Only `onnx/model_quantized.onnx` (q8) is shipped.
|