zero-shot-classification

Proof of Concept für einen Mood-Recommender: Automatisches Labeln von Film- und Hörbuchbeschreibungen mit Hugging Face (Zero-Shot-Klassifikation). Zeigt, wie Inhalte nach Stimmung kategorisiert werden können, um einfache stimmungsbasierte Empfehlungen zu generieren.

Hybrid-Analyzer 📰 automatise l’analyse d’articles de presse pour une agence de media monitoring. Le back-end ⚙️ utilise Hugging Face 📊 pour classifier les textes et Gemini ✍️ pour générer un résumé et détecter le ton 🔄, le tout de manière sécurisée 🔒.

Generalized Zero-Shot Character Recognition

What a class description buys a zero-shot classifier: NACE sections for German trade-register texts. Every number is regenerated by code and checked by CI; three label-free studies were preregistered before they were measured.

Writing your beancount files with AI!

Read a criterion off a local LLM without generating any text: score the option logits for runtime-editable criteria. Worked example: read one incoming chat message as emotion / real reason / next action — for dating, social and workplace conversations; Chinese and Japanese criteria sets; measured baselines included.

AI-powered semantic search and analysis for FMCG promo copy. Find, classify, and re-use high-performing headlines using LLMs, emotion analysis, and KPI filters. Modular Python project with Streamlit dashboard—showcasing practical data science and modern NLP for marketing.

[Private 1st] SW중심대학연합 제3회 아주 소중한 딥러닝 챌린지

Zero-Shot Classification of Protein Binders with Jev

LLM Course Projects (IUT)

This is a simple Python REST API built with FastAPI that demonstrates how to integrate a Hugging Face ML Model

HuggingFace pipeline practice — sentiment analysis, zero-shot classification and text generation on IT incident data

Telegram bot for movie and TV show recommendations based on mood and plot themes

Zero-shot evaluation of Jev AI and DeepSeek V4 Pro on Banking77: intent classification accuracy, latency, and confidence calibration

The goal here is to develop a simple mock web app book library for demonstrating a possible use of AI for aiding users

Transformers, including the T5 and MarianMT, enabled effective understanding and generating complex programming codes. Consequently, they can help us in Data Security field. Let's see how!

A Proxy that enables zero-shot system one typed-decisions endpoint for decoder only models served with vLLM or Sglang or Openrouter (Experimental)

Official implementation comparing parameter-efficient LoRA fine-tuning (Llama, Phi, Qwen families) with zero-shot API classification (o3, DeepSeek, Claude) for abusive language detection. Includes training pipelines and evaluation code. Published in JVLC

This project performs sentiment analysis on e-commerce reviews, comparing traditional models like LSTMs and CNNs with advanced LLMs like GPT-2, BERT and DistilBERT. Using Amazon and Yelp datasets, it identifies the best and efficient methods for classifying reviews as positive or negative.

A zero-shot GPT-based Intent classifier service that can be used to provide inferencing via HTTP with FastAPI