Second Tokenization Workshop (TokShop) at COLM 2026 opens call for papers

yuvalpi · x · 2026-10-01

The Second Tokenization Workshop (TokShop) will be co-located with COLM 2026 in San Francisco on October 9, 2026. Submissions open May 20, 2026, with a deadline of June 26 (extended, AoE) and notifications July 24. Topics include subword tokenization (BPE, WordPiece, UnigramLM), tokenization for images/audio/video and cross-modal alignment, multilingual tokenization equity and failure modes, and modifying tokenizers after model training. Organizer yuvalpi invites researchers to meet at COLM and the co-located workshop.

Original post →

More from Research

Research channel →