Sphilberta

🚀 SPhilBerta

SPhilBerta is a Sentence Transformer model based on PhilBERTa, aiming to identify cross - lingual references between Latin and Ancient Greek texts, contributing to the field of Classical Philology.

🚀 Quick Start

The paper Exploring Language Models for Classical Philology is the first attempt to systematically offer state - of - the - art language models for Classical Philology. Leveraging PhilBERTa, we present SPhilBERTa, a Sentence Transformer model designed to identify cross - lingual references between Latin and Ancient Greek texts. We adopt the knowledge distillation method proposed by Reimers and Gurevych (2020). Our paper can be accessed here.

✨ Features

• Multilingual Support: Supports multiple languages including multilingual, Ancient Greek (grc), English (en), and Latin (la).
• Sentence Similarity: Specialized for sentence - similarity tasks.
• Knowledge Distillation: Employs the knowledge distillation method for better performance.

📦 Installation

This README does not provide specific installation steps. However, to use the model, you may need to install relevant libraries such as sentence - transformers or transformers.

💻 Usage Examples

Basic Usage

Sentence - Transformers

When you have sentence - transformers installed, you can use the model as follows:

from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('{MODEL_NAME}')
embeddings = model.encode(sentences)
print(embeddings)

HuggingFace Transformers

Without sentence - transformers, you can use the model in the following way: First, pass your input through the transformer model, then apply the appropriate pooling - operation on top of the contextualized word embeddings.

from transformers import AutoTokenizer, AutoModel
import torch


#Mean Pooling - Take attention mask into account for correct averaging
def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)


# Sentences we want sentence embeddings for
sentences = ['This is an example sentence', 'Each sentence is converted']

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('{MODEL_NAME}')
model = AutoModel.from_pretrained('{MODEL_NAME}')

# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')

# Compute token embeddings
with torch.no_grad():
    model_output = model(**encoded_input)

# Perform pooling. In this case, mean pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])

print("Sentence embeddings:")
print(sentence_embeddings)

📚 Documentation

Contact

If you have any questions or encounter problems, feel free to [reach out](mailto:riemenschneider@cl.uni - heidelberg.de).

Citation

@incollection{riemenschneiderfrank:2023b,
    author = "Riemenschneider, Frederick and Frank, Anette",
    title = "{Graecia capta ferum victorem cepit. Detecting Latin Allusions to Ancient Greek Literature}",
    year = "2023",
    url = "https://arxiv.org/abs/2308.12008",
    note = "to appear",
    publisher = "Association for Computational Linguistics",
    booktitle = "Proceedings of the First Workshop on Ancient Language Processing",
    address = "Varna, Bulgaria"
}

📄 License

This project is licensed under the apache - 2.0 license.