LHDN Table Classifier · BERT · Production Snapshot
A production snapshot of the acceler-technology team's BERT classifier for LHDN table-related text. It preserves released model and tokenizer files separately from the development checkpoint.
中文简介： 团队 BERT 表格相关文本分类模型的生产快照，保存独立于开发仓库的发布版本。
Purpose and model details
The model predicts a class from input text for the team's table-selection work. It does not directly read document images, detect table boundaries, or extract individual cells.
| Item | Value |
|---|---|
| Role | Production snapshot |
| Architecture | BertForSequenceClassification |
| Recorded source | acceler-technology/dev-lhdn-table-bert-model |
| Task setting | Single-label classification |
| Encoder layers / hidden size | 12 / 768 |
| Maximum input positions | 512 |
| Tokenizer | Lowercasing BertTokenizer |
| Saved Transformers version | 4.48.2 |
Sources: config.json and the saved tokenizer configuration. Use this repository's tokenizer with its weights. Confirm the training label map before interpreting predictions. The prod name identifies a release snapshot; it does not establish a current live deployment.
Load the model
Install torch and transformers first.
from transformers import AutoTokenizer, AutoModelForSequenceClassification
model_id = "acceler-technology/prod-lhdn-table-bert-model"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForSequenceClassification.from_pretrained(model_id)
model.eval()
This example loads the saved artifacts. Input preparation must match training. Long text needs a chunking or truncation policy within the 512-position limit.
Project and limitations
Related project data is held in lhdn-dataset, which remains private. The exact training subset, full recipe, and quantitative evaluation results are not documented here.
Unfamiliar documents and text-extraction errors can change classification results. Review predictions before downstream use. The model does not determine tax or legal outcomes.
Team and links
Developed within acceler-technology. Henry (HenryTeo27) led this project area, executed the data work, and participated in model training and technical discussions. kwyong handled model uploads.
Project showcase · Development BERT model
A license has not been specified for this checkpoint. Public visibility alone does not grant unrestricted reuse rights.
- Downloads last month
- 6