🎉 Our last invited speaker announcement — TokShop is next week! Welcome Artidoro Pagnoni @artidoro.bsky.social (Meta Superintelligence, FAIR), lead author of the Byte Latent Transformer (BLT) and co-creator of QLoRA. Best paper award & orals at ACL/NeurIPS.

Audio is stochastic & dense: naive reps need 1000s of steps/sec. Alexandre will show how discrete bottlenecks cut that to a few hundred, and close with advances in continuous tokenization.

🎉 Another invited speaker announcement for TokShop! Welcome @honualx.bsky.social , co-founder of @kyutai-labs.bsky.social, speaking on: "Discrete and continuous tokenization for audio-text modeling" 🧵

🎙️ Excited to announce our second invited speaker: Amir Zamir, Assistant Professor at EPFL! Talk: "(Multimodal) Flexible-Length Tokenization" Looking forward to a talk that pushes tokenization beyond text 👀

Tiago covers: why tokenisation shouldn't matter in theory but does in practice, why finding optimal tokenisers is NP-complete, and how tokenisers can even reveal a text's language. He's a postdoc at ETH Zürich working on NLP, interpretability & psycholinguistics.

Announcing our first invited speaker at TokShop! We're thrilled to welcome Tiago Pimentel @tpimentel.bsky.social (ETH Zürich) for a talk on: "How much does tokenisation impact language models?" 🧵