🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

"In this article, we challenge the conventional 'write-only' CoT reasoning paradigm that dominates current LLM architectures, from both theoretical and practical perspectives." Chenxiao Yang presents insights based on their recent paper. towardsdatascience.com/empowering-llm…

"Your goal in problem solving is not to solve the problem, but to raise your understanding of it.. to a level where the problem is almost trivial" -- my old CS prof

When I applied for PhD (23 Fall), my goal was to develop new math to understand AI, so I chose a theory-focused school. Now, as a PhD, my focus has shifted to studying AI’s "psychology" scientifically. This article echoes many of my earlier reflections. 👍

📢 (1/16) Introducing PaTH 🛣️ — a RoPE-free contextualized position encoding scheme, built for stronger state tracking, better extrapolation, and hardware-efficient training. PaTH outperforms RoPE across short and long language modeling benchmarks arxiv.org/abs/2505.16381

🤣 If the “illusion of thinking” just means “context not large enough to think deeply”, here’s a simple path to true intelligence: ERASE thoughts! We proved that LLMs can solve much larger-sized problems (like Tower of Hanoi which requires exponential steps) by recursively