Learning from Naturally Occurring Feedback - AI Tinkerers x HFF Paper Club
Virtual Meeting with Authors of Learning from Naturally Occurring Feedback, feat. Shachar Don-Yehiya and Leshem Choshen (Hebrew University of Jerusalem, IBM) produced by the Human Feedback Foundation and AI Tinkerers.
About the Paper (https://arxiv.org/abs/2407.10944)
This groundbreaking paper proposes a scalable method for extracting and leveraging naturally occurring feedback in user interactions with chat models. The authors demonstrate significant performance improvements in model alignment to human preferences using this approach.
Extraction of naturally occurring feedback from user interactions
Manual annotation confirming feedback presence in 30% of chats
Application to over 1M conversations, yielding hundreds of thousands of feedback samples
Significant performance improvements in models trained with extracted feedback
What is Paper Club?
Paper Club is a virtual event series brought to you by the Human Feedback Foundation in collaboration with AI Tinkerers, featuring authors of cutting-edge AI and machine learning papers. These online meetups allow attendees to hear about groundbreaking research directly from the authors, participate in live Q&A sessions, and engage in discussions. Open to all, Paper Club offers a regular opportunity to learn and interact with leaders in the rapidly evolving field of artificial intelligence. Learn more at: https://paperclub.aitinkerers.org/
About Human Feedback Foundation:
The Human Feedback Foundation is a nonprofit organization advocating for human-centered AI. They work to integrate public input into AI systems, curate human feedback datasets, and educate about machine learning. As part of the Linux Foundation, they partner with various sectors to make AI more accessible and aligned with human values. Learn more: https://humanfeedback.io/
About AI Tinkerers:
AI Tinkerers is a curated, global community for active builders in AI. It brings together practitioners with machine learning and entrepreneurial backgrounds who are working on cutting-edge projects involving foundation models, large language models (LLMs), and generative AI. This carefully selected network spans multiple cities worldwide, providing a platform for members to share their latest AI innovations, exchange technical insights, and learn from fellow experts in the field. Learn more: https://aitinkerers.org/
NLP research, human-AI interaction, feedback taxonomy, model training, human feedback, AI annotation, LLM bias, user interaction, conversational AI, naturalistic feedback, implicit feedback, AI improvement, open-source AI, AI evaluation, AI experiments, sentiment analysis, AI moderation, interaction data, AI ethics, user preferences, conversational models, AI alignment, human annotations, AI bias, AI research, automated feedback, AI taxonomy, AI precision, AI recall, AI models.