sentence camembert base

Introduction

The sentence-camembert-base model is designed for sentence similarity tasks, specifically optimized for the French language. It is built on the camembert-base architecture and fine-tuned using the sentence-transformers library. This model is particularly effective for generating sentence embeddings and evaluating text similarity.

Architecture

The model is based on the camembert-base architecture, which is a variant of BERT optimized for the French language. It uses Siamese BERT-Networks to create sentence embeddings, enabling it to evaluate sentence similarity efficiently. The model is fine-tuned on the stsb_multi_mt dataset, a multilingual benchmark for sentence similarity tasks.

Training

The sentence-camembert-base model was fine-tuned using the pre-trained facebook/camembert-base model. The fine-tuning process involved training with the sentence-transformers library on the stsb_multi_mt dataset. The model's performance is evaluated using Pearson and Spearman correlation coefficients, achieving high accuracy on both development and test datasets.

Guide: Running Locally

To use the model locally, follow these steps:

1. Install the Required Library:

   pip install sentence-transformers
   

2. Load the Model:

   from sentence_transformers import SentenceTransformer
   model = SentenceTransformer("dangvantuan/sentence-camembert-base")
   

3. Encode Sentences:

   sentences = ["Un avion est en train de décoller.", "Un homme joue d'une grande flûte."]
   embeddings = model.encode(sentences)
   

4. Evaluation (optional): Use the provided code snippets to evaluate the model's performance using a test dataset.

For enhanced performance and large-scale tasks, consider using cloud GPUs such as AWS EC2, Google Cloud, or Azure.

License

The sentence-camembert-base model is released under the Apache 2.0 license. This allows for both personal and commercial use with proper attribution and adherence to the license terms.