Roberta Base Thai Spm Upos

🚀 roberta-base-thai-spm-upos

This is a RoBERTa model pre-trained on Thai Wikipedia texts for POS-tagging and dependency-parsing, offering accurate language analysis for Thai text.

🚀 Quick Start

This RoBERTa model is pre - trained on Thai Wikipedia texts and is used for POS - tagging and dependency - parsing. It is derived from roberta-base-thai-spm, and tags every word with UPOS (Universal Part - Of - Speech).

✨ Features

• Language Focus: Focuses on the Thai language, trained on Thai Wikipedia texts.
• Task - Oriented: Specifically designed for POS - tagging and dependency - parsing tasks.
• Standard Tagging: Uses UPOS for word tagging.

📦 Installation

The installation process is mainly about setting up the necessary Python libraries. You can install the transformers and esupar libraries via pip if they are not already installed:

pip install transformers esupar

💻 Usage Examples

Basic Usage

import torch
from transformers import AutoTokenizer,AutoModelForTokenClassification
tokenizer=AutoTokenizer.from_pretrained("KoichiYasuoka/roberta-base-thai-spm-upos")
model=AutoModelForTokenClassification.from_pretrained("KoichiYasuoka/roberta-base-thai-spm-upos")
s="หลายหัวดีกว่าหัวเดียว"
t=tokenizer.tokenize(s)
p=[model.config.id2label[q] for q in torch.argmax(model(tokenizer.encode(s,return_tensors="pt"))["logits"],dim=2)[0].tolist()[1:-1]]
print(list(zip(t,p)))

Advanced Usage

import esupar
nlp=esupar.load("KoichiYasuoka/roberta-base-thai-spm-upos")
print(nlp("หลายหัวดีกว่าหัวเดียว"))

📚 Documentation

esupar is a Tokenizer, POS - tagger, and Dependency - parser that uses BERT/RoBERTa/DeBERTa models. You can refer to it for more related information.

📄 License

This project is licensed under the "apache - 2.0" license.

Information Table