Liquid AI Showcase: Frontier Speech Systems, Post-Training RL, and Edge VLMs

The Tokyo AI (TAI) community, in collaboration with Liquid AI, hosts a technical deep dive into frontier multimodal architectures and post-training methods. Liquid AI scientists and engineers present on three fronts: LLM post-training paradigms behind a state-of-the-art Japanese small language model (SLM), structured information extraction with compact vision-language models (VLMs), and the systems and evaluation work behind real-time speech-to-speech models. The agenda runs from text-based foundations through specialized VLMs to native speech-to-speech systems, with sessions on Japanese SLM training, small VLMs for structured output, and audio language models including evaluation and data engineering for Japanese speech. A technical networking block closes out the evening.