legal_t5_small_trans_fr_de_small_finetuned model

Model on translating legal text from French to Deustch. It was first released in this repository. This model is first pretrained all the translation data over some unsupervised task. Then the model is trained on three parallel corpus from jrc-acquis, europarl and dcep.

Model description

legal_t5_small_trans_fr_de_small_finetuned is initially pretrained on unsupervised task with the all of the data of the training set. The unsupervised task was "masked language modelling". legal_t5_small_trans_fr_de_small_finetuned is based on the t5-small model and was trained on a large corpus of parallel text. This is a smaller model, which scales the baseline model of t5 down by using dmodel = 512, dff = 2,048, 8-headed attention, and only 6 layers each in the encoder and decoder. This variant has about 60 million parameters.

Intended uses & limitations

The model could be used for translation of legal texts from French to Deustch.

How to use

Here is how to use this model to translate legal text from French to Deustch in PyTorch:

from transformers import AutoTokenizer, AutoModelWithLMHead, TranslationPipeline

pipeline = TranslationPipeline(
tokenizer=AutoTokenizer.from_pretrained(pretrained_model_name_or_path = "SEBIS/legal_t5_small_trans_fr_de", do_lower_case=False, 

fr_text = "7. demande instamment à la Commission de veiller à ce que l'objectif d'une part de 20% d'énergie renouvelable soit rendue contraignante pour les États membres par des dispositions législatives à cet effet et soit mis en œuvre d'une manière conséquente, et à ce que les États membres qui n'honorent pas leurs engagements soient frappés de lourdes sanctions; souligne la nécessité de plans d'action nationaux dans le cadre desquels chaque État membre se fixe un objectif contraignant pour chaque secteur en fonction de ses possibilités spécifiques météorologiques, géographiques et géologiques et de ses réalisations dans le passé; demande instamment à la Commission de procéder à une évaluation préalable puis intermédiaire de ces plans d'action;"

pipeline([fr_text], max_length=512)

Training data

The legal_t5_small_trans_fr_de_small_finetuned (the supervised task which involved only the corresponding langauge pair and as well as unsupervised task where all of the data of all language pairs were available) model was trained on JRC-ACQUIS, EUROPARL, and DCEP dataset consisting of 8 Million parallel texts.

Training procedure

The model was trained on a single TPU Pod V3-8 for 250K steps in total, using sequence length 512 (batch size 4096). It has a total of approximately 220M parameters and was trained using the encoder-decoder architecture. The optimizer used is AdaFactor with inverse square root learning rate schedule for pre-training.


An unigram model trained with 88M lines of text from the parallel corpus (of all possible language pairs) to get the vocabulary (with byte pair encoding), which is used with this model.


The pre-training data was the combined data from all the 42 language pairs. The task for the model was to predict the portions of a sentence which were masked randomly.

Evaluation results

When the model is used for translation test dataset, achieves the following results:

Test results :

Model BLEU score
legal_t5_small_trans_fr_de_small_finetuned 41.085

BibTeX entry and citation info

Created by Ahmed Elnaggar/@Elnaggar_AI | LinkedIn


Select AutoNLP in the “Train” menu to fine-tune this model automatically.

Downloads last month
Hosted inference API
Text2Text Generation
This model can be loaded on the Inference API on-demand.