Commit
·
d07878e
1
Parent(s):
0cc253f
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,19 +1,107 @@
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
---
|
| 17 |
-
#
|
| 18 |
|
| 19 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- fr
|
| 5 |
+
multilinguality:
|
| 6 |
+
- monolingual
|
| 7 |
+
tags:
|
| 8 |
+
- finetuning
|
| 9 |
+
- legal
|
| 10 |
+
- french law
|
| 11 |
+
- droit français
|
| 12 |
+
- Code du sport
|
| 13 |
+
source_datasets:
|
| 14 |
+
- original
|
| 15 |
+
pretty_name: Code du sport
|
| 16 |
+
task_categories:
|
| 17 |
+
- text-generation
|
| 18 |
+
- table-question-answering
|
| 19 |
+
- summarization
|
| 20 |
+
- conversational
|
| 21 |
+
size_categories:
|
| 22 |
+
- 1K<n<10K
|
| 23 |
---
|
| 24 |
+
# Code du sport, non-instruct (11-12-2023)
|
| 25 |
|
| 26 |
+
This project focuses on fine-tuning pre-trained language models to create efficient and accurate models for legal practice.
|
| 27 |
+
|
| 28 |
+
Fine-tuning is the process of adapting a pre-trained model to perform specific tasks or cater to particular domains. It involves adjusting the model's parameters through a further round of training on task-specific or domain-specific data. While conventional fine-tuning strategies involve supervised learning with labeled data, instruction-based fine-tuning introduces a more structured and interpretable approach.
|
| 29 |
+
|
| 30 |
+
Instruction-based fine-tuning leverages the power of human-provided instructions to guide the model's behavior. These instructions can be in the form of text prompts, prompts with explicit task descriptions, or a combination of both. This approach allows for a more controlled and context-aware interaction with the LLM, making it adaptable to a multitude of specialized tasks.
|
| 31 |
+
|
| 32 |
+
Instruction-based fine-tuning significantly enhances the performance of LLMs in the following ways:
|
| 33 |
+
|
| 34 |
+
- Task-Specific Adaptation: LLMs, when fine-tuned with specific instructions, exhibit remarkable adaptability to diverse tasks. They can switch seamlessly between translation, summarization, and question-answering, guided by the provided instructions.
|
| 35 |
+
- Reduced Ambiguity: Traditional LLMs might generate ambiguous or contextually inappropriate responses. Instruction-based fine-tuning allows for a clearer and more context-aware generation, reducing the likelihood of nonsensical outputs.
|
| 36 |
+
- Efficient Knowledge Transfer: Instructions can encapsulate domain-specific knowledge, enabling LLMs to benefit from expert guidance. This knowledge transfer is particularly valuable in fields like tax practice, law, medicine, and more.
|
| 37 |
+
- Interpretability: Instruction-based fine-tuning also makes LLM behavior more interpretable. Since the instructions are human-readable, it becomes easier to understand and control model outputs.
|
| 38 |
+
- Adaptive Behavior: LLMs, post instruction-based fine-tuning, exhibit adaptive behavior that is responsive to both explicit task descriptions and implicit cues within the provided text.
|
| 39 |
+
|
| 40 |
+
## Dataset generation
|
| 41 |
+
|
| 42 |
+
This JSON file is a list of dictionaries, each dictionary contains the following fields:
|
| 43 |
+
|
| 44 |
+
- `instruction`: `string`, presenting the instruction linked to the element.
|
| 45 |
+
- `input`: `string`, signifying the input details for the element.
|
| 46 |
+
- `output`: `string`, indicating the output information for the element.
|
| 47 |
+
|
| 48 |
+
We used the following list of instructions for generating the dataset:
|
| 49 |
+
```python
|
| 50 |
+
instructions = [
|
| 51 |
+
"Compose l'intégralité de l'article sous forme écrite.",
|
| 52 |
+
"Écris la totalité du contenu de l'article.",
|
| 53 |
+
"Formule la totalité du texte présent dans l'article.",
|
| 54 |
+
"Produis l'intégralité de l'article en écriture.",
|
| 55 |
+
"Développe l'article dans son ensemble par écrit.",
|
| 56 |
+
"Génère l'ensemble du texte contenu dans l'article.",
|
| 57 |
+
"Formule le contenu intégral de l'article en entier.",
|
| 58 |
+
"Rédige la totalité du texte de l'article en entier.",
|
| 59 |
+
"Compose l'intégralité du contenu textuel de l'article.",
|
| 60 |
+
"Rédige l'ensemble du texte qui constitue l'article.",
|
| 61 |
+
"Formule l'article entier dans son contenu écrit.",
|
| 62 |
+
"Composez l'intégralité de l'article sous forme écrite.",
|
| 63 |
+
"Écrivez la totalité du contenu de l'article.",
|
| 64 |
+
"Formulez la totalité du texte présent dans l'article.",
|
| 65 |
+
"Développez l'article dans son ensemble par écrit.",
|
| 66 |
+
"Générez l'ensemble du texte contenu dans l'article.",
|
| 67 |
+
"Formulez le contenu intégral de l'article en entier.",
|
| 68 |
+
"Rédigez la totalité du texte de l'article en entier.",
|
| 69 |
+
"Composez l'intégralité du contenu textuel de l'article.",
|
| 70 |
+
"Écrivez l'article dans son intégralité en termes de texte.",
|
| 71 |
+
"Rédigez l'ensemble du texte qui constitue l'article.",
|
| 72 |
+
"Formulez l'article entier dans son contenu écrit.",
|
| 73 |
+
"Composer l'intégralité de l'article sous forme écrite.",
|
| 74 |
+
"Écrire la totalité du contenu de l'article.",
|
| 75 |
+
"Formuler la totalité du texte présent dans l'article.",
|
| 76 |
+
"Produire l'intégralité de l'article en écriture.",
|
| 77 |
+
"Développer l'article dans son ensemble par écrit.",
|
| 78 |
+
"Générer l'ensemble du texte contenu dans l'article.",
|
| 79 |
+
"Formuler le contenu intégral de l'article en entier.",
|
| 80 |
+
"Rédiger la totalité du texte de l'article en entier.",
|
| 81 |
+
"Composer l'intégralité du contenu textuel de l'article.",
|
| 82 |
+
"Rédiger l'ensemble du texte qui constitue l'article.",
|
| 83 |
+
"Formuler l'article entier dans son contenu écrit.",
|
| 84 |
+
"Quelles sont les dispositions de l'article ?",
|
| 85 |
+
"Quelles dispositions sont incluses dans l'article ?",
|
| 86 |
+
"Quelles sont les dispositions énoncées dans l'article ?",
|
| 87 |
+
"Quel est le texte intégral de l'article ?",
|
| 88 |
+
"Quelle est la lettre de l'article ?"
|
| 89 |
+
]
|
| 90 |
+
```
|
| 91 |
+
|
| 92 |
+
## Citing this project
|
| 93 |
+
|
| 94 |
+
If you use this code in your research, please use the following BibTeX entry.
|
| 95 |
+
|
| 96 |
+
```BibTeX
|
| 97 |
+
@misc{louisbrulenaudet2023,
|
| 98 |
+
author = {Louis Brulé Naudet},
|
| 99 |
+
title = {Code du sport, non-instruct (11-12-2023)},
|
| 100 |
+
howpublished = {\url{https://huggingface.co/datasets/louisbrulenaudet/code-procedure-penale}},
|
| 101 |
+
year = {2023}
|
| 102 |
+
}
|
| 103 |
+
```
|
| 104 |
+
|
| 105 |
+
## Feedback
|
| 106 |
+
|
| 107 |
+
If you have any feedback, please reach out at [[email protected]](mailto:[email protected]).
|