Legal T5 Small Summ En

🚀 legal_t5_small_summ_en Model

This model is designed for summarizing legal texts written in English. It was initially released in this repository and trained on three parallel corpora from jrc - acquis.

🚀 Quick Start

The legal_t5_small_summ_en model can be used to summarize English legal texts. Here's a quick guide on how to use it in PyTorch:

from transformers import AutoTokenizer, AutoModelWithLMHead, TranslationPipeline

pipeline = TranslationPipeline(
model=AutoModelWithLMHead.from_pretrained("SEBIS/legal_t5_small_summ_en"),
tokenizer=AutoTokenizer.from_pretrained(pretrained_model_name_or_path = "SEBIS/legal_t5_small_summ_en", do_lower_case=False, 
                                            skip_special_tokens=True),
    device=0
)

en_text = "THE COMMISSION OF THE EUROPEAN COMMUNITIES, Having regard to the Treaty establishing the European Community, Having regard to Council Regulation (EC) No 1255/1999 of 17 May 1999 on the common organisation of the market in milk and milk products [1], and in particular Article 15 thereof, Whereas: (1) Article 7(1) of Commission Regulation (EC) No 2799/1999 [2] fixes the amount of aid for skimmed milk and skimmed-milk powder intended for animal feed taking into account the factors set out in Article 11(2) of Regulation (EC) No 1255/1999. In view of the developments in the market price of skimmed-milk powder, of the increase in the market prices for competing proteins, and of the reduction of the supply of skimmed-milk powder, the amount of aid should be reduced. (2) Regulation (EC) No 2799/1999 should therefore be amended accordingly. (3) The Management Committee for Milk and Milk Products has not delivered an opinion within the time-limit set by its chairman, HAS ADOPTED THIS REGULATION: Article 1 In Article 7 of Regulation (EC) No 2799/1999, paragraph 1 is replaced by the following: \"1. Aid is fixed at: (a) EUR 1,62 per 100 kg of skimmed milk with a protein content of not less than 35,6 % of the non-fatty dry extract; (b) EUR 1,42 per 100 kg of skimmed milk with a protein content of not less than 31,4 % but less than 35,6 % of the non-fatty dry extract; (c) EUR 20,00 per 100 kg of skimmed-milk powder with a protein content of not less than 35,6 % of the non-fatty dry extract; (d) EUR 17,64 per 100 kg of skimmed-milk powder with a protein content of not less than 31,4 % but less than 35,6 % of the non-fatty dry extract.\" Article 2 This Regulation shall enter into force on the day following its publication in the Official Journal of the European Union. This Regulation shall be binding in its entirety and directly applicable in all Member States. Done at Brussels, 19 April 2006. For the Commission Mariann Fischer Boel Member of the Commission [1] OJ L 160, 26.6.1999, p. 48. Regulation as last amended by Regulation (EC) No 1913/2005 (OJ L 307, 25.11.2005, p. 2). [2] OJ L 340, 31.12.1999, p. 3. Regulation as last amended by Regulation (EC) No 1194/2005 (OJ L 194, 26.7.2005, p. 7). -------------------------------------------------- "

pipeline([en_text], max_length=512)

✨ Features

• Specialized for Legal Texts: The model is specifically designed for summarizing English legal texts.
• Based on t5 - small: It builds on the t5 - small architecture, with a scaled - down configuration for efficiency.

📚 Documentation

Model description

The legal_t5_small_summ_en model is based on the t5 - small model and trained on a large parallel text corpus. It is a smaller model, scaling down the baseline t5 model by using dmodel = 512, dff = 2,048, 8 - headed attention, and only 6 layers each in the encoder and decoder. This variant has approximately 60 million parameters.

Intended uses & limitations

The model is intended for summarizing English legal texts. However, its performance may vary depending on the complexity and domain - specificity of the input text.

💻 Usage Examples

Basic Usage

from transformers import AutoTokenizer, AutoModelWithLMHead, TranslationPipeline

pipeline = TranslationPipeline(
model=AutoModelWithLMHead.from_pretrained("SEBIS/legal_t5_small_summ_en"),
tokenizer=AutoTokenizer.from_pretrained(pretrained_model_name_or_path = "SEBIS/legal_t5_small_summ_en", do_lower_case=False, 
                                            skip_special_tokens=True),
    device=0
)

en_text = "THE COMMISSION OF THE EUROPEAN COMMUNITIES, Having regard to the Treaty establishing the European Community, Having regard to Council Regulation (EC) No 1255/1999 of 17 May 1999 on the common organisation of the market in milk and milk products [1], and in particular Article 15 thereof, Whereas: (1) Article 7(1) of Commission Regulation (EC) No 2799/1999 [2] fixes the amount of aid for skimmed milk and skimmed-milk powder intended for animal feed taking into account the factors set out in Article 11(2) of Regulation (EC) No 1255/1999. In view of the developments in the market price of skimmed-milk powder, of the increase in the market prices for competing proteins, and of the reduction of the supply of skimmed-milk powder, the amount of aid should be reduced. (2) Regulation (EC) No 2799/1999 should therefore be amended accordingly. (3) The Management Committee for Milk and Milk Products has not delivered an opinion within the time-limit set by its chairman, HAS ADOPTED THIS REGULATION: Article 1 In Article 7 of Regulation (EC) No 2799/1999, paragraph 1 is replaced by the following: \"1. Aid is fixed at: (a) EUR 1,62 per 100 kg of skimmed milk with a protein content of not less than 35,6 % of the non-fatty dry extract; (b) EUR 1,42 per 100 kg of skimmed milk with a protein content of not less than 31,4 % but less than 35,6 % of the non-fatty dry extract; (c) EUR 20,00 per 100 kg of skimmed-milk powder with a protein content of not less than 35,6 % of the non-fatty dry extract; (d) EUR 17,64 per 100 kg of skimmed-milk powder with a protein content of not less than 31,4 % but less than 35,6 % of the non-fatty dry extract.\" Article 2 This Regulation shall enter into force on the day following its publication in the Official Journal of the European Union. This Regulation shall be binding in its entirety and directly applicable in all Member States. Done at Brussels, 19 April 2006. For the Commission Mariann Fischer Boel Member of the Commission [1] OJ L 160, 26.6.1999, p. 48. Regulation as last amended by Regulation (EC) No 1913/2005 (OJ L 307, 25.11.2005, p. 2). [2] OJ L 340, 31.12.1999, p. 3. Regulation as last amended by Regulation (EC) No 1194/2005 (OJ L 194, 26.7.2005, p. 7). -------------------------------------------------- "

pipeline([en_text], max_length=512)

🔧 Technical Details

Training data

The legal_t5_small_summ_en model was trained on the [JRC - ACQUIS](https://wt - public.emm4u.eu/Acquis/index_2.2.html) dataset, which consists of 22 Thousand texts.

Training procedure

• Hardware: The model was trained on a single TPU Pod V3 - 8.
• Steps: It was trained for a total of 250K steps.
• Sequence length and batch size: It used a sequence length of 512 and a batch size of 64.
• Parameters: The model has approximately 220M parameters and uses an encoder - decoder architecture.
• Optimizer: The optimizer used is AdaFactor with an inverse square root learning rate schedule for pre - training.

Preprocessing

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

📄 License

No license information is provided in the original document.

Evaluation results

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

BibTeX entry and citation info

Created by Ahmed Elnaggar/@Elnaggar_AI | LinkedIn