Liquid AI Showcase: Frontier Speech Systems, Post-Training RL, and Edge VLMs
A technical showcase by Liquid AI and Tokyo AI covering Japanese Small Language Models, specialized Vision-Language Models for data extraction, and native speech-to-speech architectures.
Liquid AI Showcase: Frontier Speech Systems, Post-Training RL, and Edge VLMs
A technical showcase by Liquid AI and Tokyo AI covering Japanese Small Language Models, specialized Vision-Language Models for data extraction, and native speech-to-speech architectures.
The Tokyo AI (TAI) community, in collaboration with Liquid AI, presents a technical deep dive into frontier multimodal architectures and post-training methodologies. The session features technical presentations from Liquid AI scientists and engineers covering: 1. LLM Post-Training Trends & an Introduction to a State-of-the-Art Japanese SLM (LFM2.5-1.2B-JP-202606) by Kohsei Matsutani. 2. Structured Information Extraction with Small VLMs, focusing on 1.6B and 450M models fine-tuned for JSON extraction by Hongkuan Zhang. 3. Building Speech-to-Speech AI: Models, Evals, and Data for Spoken Interaction, a combo talk by Marc Härkönen, Samuel J. Broughton, and Masao Taketani covering real-time speech-to-speech systems and Japanese audio data challenges. The event concludes with a technical networking session.