# Pre-training fine-tuning

This is a cross-encoder model based on the Transformer architecture, primarily used for text ranking tasks.

French sentence embedding model based on FlauBERT for calculating sentence similarity

A sentiment analysis model based on DeepPavlov/rubert-base-cased-conversational architecture, trained on the RuSentiment dataset, capable of identifying neutral, positive, and negative emotions in Russian text.

A multilingual machine translation model fine-tuned based on mBART-large-50, supporting translation tasks from 50 languages to English

A Chinese extractive QA model based on the RoBERTa architecture, suitable for tasks that extract answers from given texts.

CodeT5-small is a Transformer-based code understanding and generation model developed by Salesforce, specifically fine-tuned for Python code-to-text tasks.