Danial Kamali

Danial Kamali

Curious AI Researcher

Sitemap

A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.

Pages

Posts

Aligning AI with Humanity: The Role of Reinforcement Learning in Language Model Alignment

34 minute read

Published:

In this work, we look into the prominent applications of Reinforcement Learning (RL) in the field of Natural Language Processing (NLP) with a focus on Language Models (LM). First, we examine one of the initial applications of Reinforcement Learning with Human Feedback (RLHF) in NLP. Then, we discuss how this method evolves to be applied in a more general AI and becomes a fundamental aspect of Large Language Model (LLM) training. Also, we discuss the risks, challenges, and potential problems associated with RLHF, offering insights into how these issues might be addressed and mitigated. Furthermore, we explore the emerging field of Reinforcement Learning with AI Feedback (RLAIF), assessing its position in current research. Our investigation shows that RLHF training is a very effective tool for language model alignment. This method cannot only improve the performance of the overall model in NLP benchmarks but also help with problems such as hallucination. In addition, we showed that methods like Constitutional AI can improve the LLMs’ safety by increasing harmlessness while keeping high levels of helpfulness. Read more

portfolio

publications

Syntax-Guided Transformers: Elevating Compositional Generalization and Grounding in Multimodal Environments

The Conference on Empirical Methods in Natural Language Processing Genbench Workshop, 2023

In this work, we proposed the syntax guided transformer to improve the compositional generalization in grounding. Read more

Recommended citation: Kamali, Danial, and Parisa Kordjamshidi. "Syntax-Guided Transformers: Elevating Compositional Generalization and Grounding in Multimodal Environments." GenBench: The first workshop on generalisation (benchmarking) in NLP. 2023. https://aclanthology.org/2023.genbench-1.10.pdf

        @inproceedings{kamali2023syntax \n ,title={Syntax-Guided Transformers: Elevating Compositional Generalization and Grounding in Multimodal Environments} \n,author={Kamali, Danial and Kordjamshidi, Parisa} \n,booktitle={GenBench: The first workshop on generalisation (benchmarking) in NLP}\n,pages={130}\n,year={2023}\n}

Using Persuasive Writing Strategies to Explain and Detect Health Misinformation

Joint International Conference on Computational Linguistics, Language Resources and Evaluation, 2024

In this work we introduce a persuasive strategy detection dataset and show using their labels can improve misinformation detection and explanation. Read more

Recommended citation: Kamali, D., Romain, J., Liu, H., Peng, W., Meng, J., & Kordjamshidi, P. (2023). Using Persuasive Writing Strategies to Explain and Detect Health Misinformation. arXiv preprint arXiv:2211.05985. https://aclanthology.org/2024.lrec-main.1501.pdf

        @inproceedings{kamali-etal-2024-using, title = "Using Persuasive Writing Strategies to Explain and Detect Health Misinformation", author = "Kamali, Danial  and Romain, Joseph D.  and Liu, Huiyi  and Peng, Wei  and Meng, Jingbo  and Kordjamshidi, Parisa", editor = "Calzolari, Nicoletta  and Kan, Min-Yen  and Hoste, Veronique  and Lenci, Alessandro  and Sakti, Sakriani  and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.1501", pages = "17285--17309", }

talks

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post. Read more

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post. Read more