Reasoning Models - Connect Session

🧵 ResearchTrend.AI Reasoning Model Connect Session: Unlimited Memory & Optimal Thinking Styles!

Meet the Speakers 🧑‍🔬
Hongyin Luo: Co-founder / CTO at Subconscious and Research Scientist at Massachusetts Institute of Technology
Junyu Guo: PhD student at University of California, Berkeley

This virtual session 💻 features two essential presentations from leading researchers, diving into the critical limits and strategies that govern complex LLM reasoning.

Agenda (UTC) - December 10th

16:00 - 16:30: Hongyin Luo
📄 Paper: Beyond Context Limits: Subconscious Threads for Long-Horizon Reasoning
💡 Abstract: The context limit is a major bottleneck for LLM reasoning. Hongyin will introduce the Thread Inference Model (TIM) and its runtime, TIMRUN. This pioneering system supports virtually unlimited working memory and multi-hop tool calls by modeling language as a reasoning tree . TIMRUN sustains high throughput and delivers accurate reasoning on mathematical and information retrieval tasks that demand long-horizon capabilities.

16:30 - 17:00: Junyu Guo
📄 Paper: StyleBench: Evaluating thinking styles in Large Language Models
💡 Abstract: The performance of LLMs depends heavily on the reasoning style (CoT, ToT, AoT, etc.) used in prompting. Junyu will present StyleBench, a comprehensive benchmark evaluating five major reasoning styles. The analysis reveals that no single style is universally optimal; efficacy is highly contingent on both model scale and task type, providing a crucial roadmap for strategy selection.

🌟 This is a fantastic opportunity to engage directly with research that is fundamentally addressing the architectural and strategic bottlenecks of current LLMs.

🗓️ Time: 4:00 PM - 5:00 PM UTC 📍 Location: Virtual
👉 Register for this event here: https://lnkd.in/eF6gzFus

Don't miss our future sessions! 📆 Find out more about upcoming events: https://lnkd.in/g7-iczUp