|
- GitHub - MoonshotAI Kimi-K2: Kimi K2 is the large language model series . . .
Trained with the Muon optimizer, Kimi K2 achieves exceptional performance across frontier knowledge, reasoning, and coding tasks while being meticulously optimized for agentic capabilities
- Kimi K2: Open Agentic Intelligence - moonshotai. github. io
Kimi K2 was designed to further scale up Moonlight, which employs an architecture similar to DeepSeek-V3 Based on scaling-law analysis, we reduce the number of heads for long-context efficiency, and increase MoE sparsity for greater token efficiency
- Kimi K2 - Advanced AI Model by Moonshot AI
How can I access Kimi K2? You can access Kimi K2 through its official website, Moonshot AI's API platform, or via local deployment using GitHub repositories and Hugging Face
- 在 software agents 中使用 kimi k2 模型 - Moonshot AI 开放 . . .
kimi-k2 是一款具备超强代码和 Agent 能力的 MoE 架构基础模型，我们以 VS Code Cline RooCode 为示例，说明如何使用 kimi-k2-0711-preview 模型。
- Kimi K2 - Open Source AI Model | 1T Parameters | Agentic
Kimi K2 is MoonshotAI's 1T parameter open-source AI model with agentic capabilities Supports 128K context, excels in reasoning coding
- 月之暗面发布 Kimi K2 技术报告 - 开源资讯说 - 博客园
月之暗面（Moonshot AI）发布了其 Kimi K2 模型的完整技术报告，相关文档已上传至 GitHub。 一、Kimi K2 是什么？ 换句话说，Kimi K2 是一位既懂编程又会推理、还能灵活使用工具的超级 AI 助手。 二、为什么说它「强」？ 三、它是怎么训练出来的？ Kimi K2 在预训练与后训练阶段均做了大量创新设计。 架构采用 超稀疏 MoE + 多头隐式注意力（MLA），比 DeepSeek V3 更轻更强。 数学重写：将数学文档转写为 “学习笔记” 风格，提高理解与推理能力。 此外，还加入跨语种数学翻译数据，让模型更具多语言泛化能力。 采用人类评审 + LLM 评估，保障生成数据的质量与多样性。
- kimi-k2-ai - GitHub
Kimi K2 API is a comprehensive API platform providing access to the revolutionary Kimi K2 AI model developed by Moonshot AI The Kimi K2 model features a sophisticated mixture-of-experts (MoE) architecture with 1 trillion total parameters and 32 billion activated parameters
- moonshotai Kimi-K2-Instruct · Hugging Face
Kimi-K2-Base: The foundation model, a strong start for researchers and builders who want full control for fine-tuning and custom solutions Kimi-K2-Instruct: The post-trained model best for drop-in, general-purpose chat and agentic experiences
|
|
|