最新动态

周边新闻待整理

Superalignment Fast Grants

We’re launching $10M in grants to support technical research towards the alignment and safety of superhuman AI systems, including weak-to-strong generalization, interpretability,…

周边新闻待整理

Weak-to-strong generalization

We present a new research direction for superalignment, together with promising initial results: can we leverage the generalization properties of deep learning to control strong…

周边新闻待整理

La Plateforme

The most powerful AI platform for enterprises. Customize, fine-tune, and deploy AI assistants, autonomous agents, and multimodal AI with open models.

周边新闻待整理

Mixtral of experts

The most powerful AI platform for enterprises. Customize, fine-tune, and deploy AI assistants, autonomous agents, and multimodal AI with open models.

模型更新

Mixtral 8x7B Instruct v0.1

收录发布日期:2023-12-11(社区数据库,待与官方公告核对)。能力:推理、开放权重;输入:text/pdf。上下文:32,000 tokens。定位(来源描述):Reasoning model for deliberate analysis, multi-step problem solving, and tool…

模型更新

Mixtral 8x7B Instruct v0.1 (Deep Infra)

收录发布日期:2023-12-11(社区数据库,待与官方公告核对)。能力:工具调用、开放权重;输入:text。上下文:32,768 tokens。定位(来源描述):Open sparse mixture-of-experts Mistral instruction model。实际效果:暂无本站独立评测;能力参数不代表评测成绩。资料来自…

模型更新

Mistral Embed

收录发布日期:2023-12-11(社区数据库,待与官方公告核对)。能力:见模型资料;输入:text。上下文:8,000 tokens。定位(来源描述):Embedding model for semantic search, retrieval, clustering, and ranking…

周边新闻

Google 发布第一代 Gemini

Google 发布 Gemini 系列,作为其新一代原生多模态模型,并按不同规模提供多个版本。主流厂商的竞争由此转向同一模型家族覆盖云端到设备的完整梯队。

模型更新

LlamaGuard-7b

官方机构模型仓库创建于 2023-12-06。仓库创建时间不等于正式发布日期。任务类型:text-generation