Davit Melikidze

Andrei Panferov, Davit Melikidze, Martin Jaggi et al.

The wide adoption of LLMs has led to their use in great variety of applications and scenarios, such as chatbot assistants and data annotation, creating the need for the models to satisfy certain budget and hardware constraints. This has led to the trend of LLMs being released in batches consisting of similar models of various sizes for the family of models to adhere to as wide of a range of constraints as possible. In this paper, we validate distillation and quantization as a cost-effective way to expand model families to new sizes and hardware formats. Based on the open-recipe Apertus 8B LLM, we produce Apertus-v1.1 - a distilled family of models with up to 4B parameters trained on 1.7T permissive license tokens. We demonstrate cost-efficiency and strong accuracy performance of our approach for covering large ranges of hardware and systems requirements.

Davit Melikidze, Marian Schneider, Jessica Lam et al.

Reinforcement Learning from Human Feedback (RLHF) has become the standard for aligning Large Language Models (LLMs), yet its efficacy is bottlenecked by the high cost of acquiring preference data, especially in low-resource and expert domains. To address this, we introduce ACTIVEULTRAFEEDBACK, a modular active learning pipeline that leverages uncertainty estimates to dynamically identify the most informative responses for annotation. Our pipeline facilitates the systematic evaluation of standard response selection methods alongside DOUBLE REVERSE THOMPSON SAMPLING (DRTS) and DELTAUCB, two novel methods prioritizing response pairs with large predicted quality gaps, leveraging recent results showing that such pairs provide good signals for fine-tuning. Our experiments demonstrate that ACTIVEULTRAFEEDBACK yields high-quality datasets that lead to significant improvements in downstream performance, notably achieving comparable or superior results with as little as one-sixth of the annotated data relative to static baselines. Our pipeline is available at https://github.com/lasgroup/ActiveUltraFeedback and our preference datasets at https://huggingface.co/ActiveUltraFeedback.

Davit Melikidze, Alexander Gamkrelidze

This research proposes a novel approach to the Word Sense Disambiguation (WSD) task in the Georgian language, based on supervised fine-tuning of a pre-trained Large Language Model (LLM) on a dataset formed by filtering the Georgian Common Crawls corpus. The dataset is used to train a classifier for words with multiple senses. Additionally, we present experimental results of using LSTM for WSD. Accurately disambiguating homonyms is crucial in natural language processing. Georgian, an agglutinative language belonging to the Kartvelian language family, presents unique challenges in this context. The aim of this paper is to highlight the specific problems concerning homonym disambiguation in the Georgian language and to present our approach to solving them. The techniques discussed in the article achieve 95% accuracy for predicting lexical meanings of homonyms using a hand-classified dataset of over 7500 sentences.