AI Briefing#nlp
The latest AI and developer news about #nlp, with the original source and a short summary.
The latest AI and developer news about #nlp, with the original source and a short summary.
Transformers v5 significantly strengthens modularity and customization convenience by separating tokenizer design from trained vocabulary.
Sentence Transformers, the popular embedding library, is transferring from TU Darmstadt's UKP Lab to Hugging Face.
To boost user engagement on OpenChat, a technique was developed to extract trending keywords from message text.
mmBERT, a next-generation multilingual encoder based on the ModernBERT architecture and supporting over 1,800 languages, has been released.
A new benchmark called 3LM has been released to evaluate the STEM and code generation capabilities of Arabic LLMs.
TII has released Falcon-Arabic, a 7B-parameter model based on the Falcon 3 architecture, maximized for Arabic performance.
Hugging Face is updating and overhauling its existing NLP course into an LLM-focused, up-to-date curriculum.
The Sentence Transformers v4.0 update reveals an efficient way to train and fine-tune Reranker models.
The Open Arabic LLM Leaderboard 2 has been released to transparently evaluate the performance of Arabic LLMs.
ModernBERT, a next-generation BERT replacement model with an 8,192-token long context and improved performance, has been released.
This introduces Matryoshka Embedding technology, which concentrates key information in the earlier dimensions to reduce embedding dimensionality without performance loss.
This defines the concept of Explainable AI (XAI) and introduces use cases for improving the performance and user experience of Computer Vision and NLP models.
An open-source leaderboard for evaluating factuality and faithfulness hallucinations in LLMs has been released.
Prodigy-HF, a plugin that directly connects the data annotation tool Prodigy with Hugging Face, has been released.
Meta AI's fastText models and word vectors for 157 languages have been integrated into the Hugging Face Hub.
An integration feature has been added that allows BERTopic models to be saved to and loaded directly from the Hugging Face Hub.
Witty Works implemented a technical solution using SetFit to classify non-inclusive language based on context with only a small amount of data.
Sentence Transformers creator Nils Reimers shares how to evaluate embeddings using MTEB and BEIR.
Harambee boosted sign-up conversion by 20% with an AI21-based chatbot.
MTEB, a large-scale text embedding benchmark supporting 8 tasks and 112 languages, has been released.
It explained how to build a dashboard for topic-based sentiment analysis of hotel reviews using AI21 Jurassic-1.
SetFit, a few-shot learning framework that achieves high performance with little data and no prompts, has been released.
Kakao Pay introduces 'PayProfile,' a user profiling system that leverages vast amounts of data to analyze user behavior and provide personalized services.
APEACH, a Korean hate speech evaluation dataset generated via crowdsourcing, was released to address the bias and copyright issues of existing web-crawled data.