Tag: moe
-
Qwen3‑235B‑A22B‑Instruct‑2507: Alibaba’s Updated 235B-Parameter Instruction-Tuned LLM
Alibaba’s Qwen team has just released Qwen3‑235B‑A22B‑Instruct‑2507, an enhanced instruction-tuned large language model with 235B total parameters (22B active), designed for chat, reasoning, coding, and long-context understanding. Key points Alibaba’s Qwen team released Qwen3-235B-A22B-Instruct-2507, an instruction-tuned model with 235B total parameters and 22B active, designed for chat, reasoning, coding and long-context understanding. It is a […]
-
MiniMax M1: The World’s First Open-Weight, Million-Token Context AI Model
MiniMax, a Shanghai-based AI company founded in 2021, has released MiniMax M1, an open-weight reasoning model designed to handle extremely long inputs—up to one million tokens—while maintaining high inference efficiency. This marks a significant milestone in large-scale model design, positioning M1 as a leader in applications requiring deep, multi-step reasoning over vast contexts (arxiv.org, en.wikipedia.org). […]
-
Understanding Apple’s Parallel-Track MoE Architecture
Apple recently introduced a novel Parallel-Track Mixture‑of‑Experts (PT‑MoE) architecture for their server-side language models, aiming to dramatically improve scalability and efficiency while preserving model quality (machinelearning.apple.com). Key points Apple recently introduced a Parallel-Track Mixture-of-Experts (PT-MoE) architecture for its server-side language models, aiming to dramatically improve scalability and efficiency while preserving model quality. Rather than a […]