# Unsupervised fine-tuning

A sentence embedding model optimized for Ukrainian, based on the multilingual MPNet architecture, suitable for semantic similarity and feature extraction tasks

Bonito is an open-source conditional task generation model focused on transforming unannotated text into task-specific instruction-tuning training datasets.

Vision-language model initialized with OpenAI CLIP, enhanced with unsupervised adversarial fine-tuning for improved robustness

Zephyr 7B Beta is a 7B parameter model based on the Mistral architecture of Hugging Face H4. It is optimized through AWQ quantization and suitable for efficient inference tasks.