NLP

Tools for probing the reasoning ability of LLMs and logging the logprobs (log probabilties) of generated tokens to identify weak points in an LLM's response.

Le Mot Diplomatique trace l'historique de la première apparition d'un mot dans Le Monde Diplomatique.

Natural Language Understanding for math symbols, digits, and words with a Gradio user interface and REST API.

Django application to build and maintain chatbots that combine the best of LLMs and rule-based engines to truly assist your users instead of manipulating them.

Fork of Rubrix for natural language processing machine learning dataset labeling and RLHF (reinforcement learning human feedback) and active learning. Original license was Apache 2.0 which allows Argilla.io to modify the software without sharing it with the community.

A set of python classes to manage corpora aimed at Automatic Text Summarization (ATS) as well as the semi-automatic evaluation process (with ROUGE). Typically, a corpus for ATS has several Original Texts and reference summaries (gold standard)

Une interface web pour verbecc-svc | Web front-end for verbecc-svc

RPA solution designed to search and apply to jobs in LinkedIn. It creates a NLP model using the jobs manually applied, then get keywords to search in LinkedIn, using that model calculates the similarity between texts and apply to those with >40%

Performance Comparison of S-O-T-A, Transformer Based Language Models, like BERT & XLNet in Legal domain. Studying impacts of additional (domain-specific) pre-training on BERT in Legal domain. Exploring impacts of Introducing Legal Vocabulary in BERT.

A Job Recommendation Engine

NLP Lean Programming framework (NLPf)

Domain specific language for generating analogous variations of textual inputs

Un projet personnel de divulgation de l'intelligence artificielle en linguistique.