Zhipu AI

Berita AI terbaru, model, dan rilis dari Zhipu AI. ['GLM', 'GLM 4', 'GLM-4', 'GLM-4-32B-0414', 'GLM-4.5', 'GLM-4.6', 'GLM-4.7', 'glm-4.7-flash', 'GLM-4-9B-0414', 'GLM-4-9B-Chat-1M', 'GLM-4-Air-250414', 'GLM-4-Flash-250414', 'GLM 5', 'GLM-5', 'GLM-5.1', 'GLM5.1', 'GLM 5.2', 'GLM-5.2', 'GLM-5.5', 'GLM5-744B-A40B', 'GLM-Z1-32B-0414', 'GLM-Z1-9B-0414', 'GLM-Z1-Air', 'GLM-Z1-AirX', 'GLM-Z1-Flash', 'GLM-Z1-Rumination-32B-0414', 'not specified']

Riset 🇺🇸

Beyond Standard LLMs: Linear Attention Hybrids, Text Diffusion Models, Code World Models, and Small Recurrent Transformers

The article explores alternatives to standard autoregressive transformer-based LLMs: linear attention hybrids (MiniMax-M1, Qwen3-Next, DeepSeek V3.2, Kimi Linear), text diffusion models, code world models, and small recurrent transformers. The author notes a return to classical attention in MiniMax-M2 and the complexity of linear attention in production.

Moonshot AIMoonshot AI MiniMaxMiniMax Alibaba/QwenAlibaba/Qwen DeepSeekDeepSeek Google/DeepMindGoogle/DeepMind MistralMistral MetaMeta Hugging FaceHugging Face Allen Institute for AIAllen Institute for AI xAIxAI OpenAIOpenAI IBMIBM NVIDIANVIDIA
Sebastian Raschka27.07 · 17:04
Berita terbaru