TWIML

Today, we're joined by Julie Kallini, PhD student at Stanford University to discuss her recent papers, “MrT5: Dynamic Token Merging for Efficient Byte-level Language Models” and “Mission: Impossible Language Models.” For the MrT5 paper, we explore the importance and failings of tokenization in large language models—including inefficient compression rates for under-resourced languages—and dig into byte-level modeling as an alternative. We discuss the architecture of MrT5, its ability to learn language-specific compression rates, its performance on multilingual benchmarks and character-level manipulation tasks, and its performance and efficiency. For the “Mission: Impossible Language Models” paper, we review the core idea behind the research, the definition and creation of impossible languages, the creation of impossible language training datasets, and explore the bias of language model architectures towards natural language. ?? / ?? Listen or watch the full episode on our page: https://twimlai.com/go/724.

Our Generative AI study group is about to start in 1 hour! ? Don’t miss it and discuss LLM distillation techniques, share efficient agentic reasoning workflows, explore Large Action Models (LAMs), and more! Feel free to pass this along with your colleagues and be sure to join us in an hour! #llms #generativeai #studygroup #TWIMLMeetup #TWIML #machinelearning #ml #ai #artificialintelligence

Catch our Generative AI study group tomorrow at 8am PT and share training recipes for agents, explore latent reasoning techniques, dig into federated generative AI, and more! Sign up at https://lnkd.in/eGp6wMa and we’ll meet you there! #llms #generativeai #studygroup #TWIMLMeetup #TWIML #machinelearning #ml #ai #artificialintelligence

Don't miss out on the opportunity to learn, share, and connect with fellow AI enthusiasts in our Generative AI study group! Register at https://lnkd.in/eGp6wMa and meet us every Friday at 8am PT.

Today, we're joined by Jonas Geiping, research group leader at ELLIS Institute Tübingen and the Max Planck Institute for Intelligent Systems to discuss his recent paper, “Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach.” This paper proposes a novel language model architecture which uses recurrent depth to enable “thinking in latent space.” We dig into “internal reasoning” versus “verbalized reasoning”—analogous to non-verbalized and verbalized thinking in humans, and discuss how the model searches in latent space to predict the next token and dynamically allocates more compute based on token difficulty. We also explore how the recurrent depth architecture simplifies LLMs, the parallels to diffusion models, the model's performance on reasoning tasks, the challenges of comparing models with varying compute budgets, and architectural advantages such as zero-shot adaptive exits and natural speculative decoding. ?? / ?? Listen or watch the full episode on our page: https://twimlai.com/go/723.

Join us in 1 hour for another knowledge-packed session in our Generative AI study group! Share demos of new models and platforms, collaborate on the TWIML-RAG project, explore practical generative model applications, and more! We’re looking forward to having you join us! #llms #generativeai #studygroup #TWIMLMeetup #TWIML #machinelearning #ml #ai #artificialintelligence

Come and join our Generative AI study group where we discuss, share, and explore topics on generative AI, LLMs, RAG, agents, and more! If you’re interested in joining, register at https://lnkd.in/eGp6wMa and meet us tomorrow at 8am PT. The discussions are informal, and participants are welcome to jump in and out anytime during the session. See you there! #llms #generativeai #studygroup #TWIMLMeetup #TWIML #machinelearning #ml #ai #artificialintelligence

Today, we're joined by Chengzu Li, PhD student at the University of Cambridge to discuss his recent paper, “Imagine while Reasoning in Space: Multimodal Visualization-of-Thought.” We explore the motivations behind MVoT, its connection to prior work like TopViewRS, and its relation to cognitive science principles such as dual coding theory. We dig into the MVoT framework along with its various task environments—maze, mini-behavior, and frozen lake. We explore token discrepancy loss, a technique designed to align language and visual embeddings, ensuring accurate and meaningful visual representations. Additionally, we cover the data collection and training process, reasoning over relative spatial relations between different entities, and dynamic spatial reasoning. Lastly, Chengzu shares insights from experiments with MVoT, focusing on the lessons learned and the potential for applying these models in real-world scenarios like robotics and architectural design. ?? / ?? Listen or watch the full episode on our page: https://twimlai.com/go/722.

Our Generative AI study group starts in just one hour! Don’t miss out on the opportunity to chat about the latest in LLM news, exchange generative AI insights, share ideas for the RAG chatbot project, and more! If you haven’t registered yet, go to https://lnkd.in/eGp6wMa. Looking forward to you joining us in an hour! #llms #generativeai #studygroup #TWIMLMeetup #TWIML #machinelearning #ml #ai #artificialintelligence

Make sure you’re registered at https://lnkd.in/eGp6wMa to attend tomorrow’s session of our Generative AI study group! Discuss deep research agents, exchange supervised fine-tuning techniques, share thoughts on AI governance, and more. See you there at 8am PT! #llms #generativeai #studygroup #TWIMLMeetup #TWIML #machinelearning #ml #ai #artificialintelligence