C4AI Expedition Aya - Closing Ceremony

At this event, we marked the end of Expedition Aya, our six-week global open-build challenge designed to jumpstart multilingual research initiatives. The event featured presentations from all participating teams, with updates on their progress. Judging by Sebastian Ruder, Marzieh Fadaee, and Alice Schoenauer Sebag, awarded top teams with Cohere API credits. Let’s celebrate multilingual research and look ahead to projects to come.

00:00:00 Welcome and Intro
00:04:42 Multi-Agent Aya for Workforce
00:11:33 MCM: Multilingual LLM Capability For Crisis Management
00:15:44 Evaluating Reward Models in Multilingual Settings
00:23:07 Identifying mislabelled samples in the Aya dataset
00:30:26 Synthetic Multilingual Instruction Generation
00:36:35 DistAYA
00:43:28 MLCC: Multilingual Climate Chatbot
00:49:30 Maya: Multimodal Aya
00:57:01 Multilingual Machine Generated Text Portion Detection
01:02:05 Sanskriti Aya: Designing Linguistic Quality Index to analyze conversational capabilities of LLMs
01:11:04 Doclingual: Multilingual Document Understanding
01:19:55 Mechanistically Understanding Aya for Safety Steering
01:25:16 Samskrutha LLM
01:29:50 Enhancing Sinhala NLP
01:35:19 The Language Effect
01:42:07 SpeechAya
01:47:26 SafeAya: Multilingual Adversarial Robustness
01:56:55 Global Exams
02:02:44 StAyaL: Multilingual Style Transfer
02:47:46 Perplexity vs Model Performance Evaluation
02:13:26 Announcing C4AI Research Office Hours
02:17:28 Results from the judges