←搜索结果

搜索“Meta”

按匹配度 · 共 182 条结果 清除搜索

星期一

  1. Zyphra Research

    Zyphra 发布开源小模型 Zamba2-7B

    Zyphra 发布 7B 规模小语言模型 Zamba2-7B，采用 Mamba2 与两个共享注意力块交替的混合架构，并以 Apache 2.0 开源权重。官方称其在评测基准上超过 Mistral-7B、Gemma-7B 和 Llama3-8B，首 token 时间快 25%、每秒 token 数提升 20%，内存占用也明显降低。模型基于 3 万亿 token 预训练数据，并在 128 张 H100 上训练约 50 天。

    Zyphra is excited to release Zamba2-7B, a state-of-the-art small language model. At the 7B scale, we outperform the leading models of Mistral, Google’s Gemma and Meta’s Llama3 series in both quality and performance. Zamba2-7B achieves SOTA evaluation benchmark performance and superior inference efficiency compared to currently leading 7B models such as Mistral-7B, Gemma-7B, and Llama3-8B; Zamba2-7B is extremely inference-efficient, achieving 25% faster time to first token, a 20% improvement in t

    匹配字段正文

    #lab#模型发布#开源生态#推理

星期三

  1. Cartesia 产品与研究

    Cartesia 发布端侧智能更新：开源 Edge 库、1.3B 模型 Rene 与 Sonic On-Device

    Cartesia 宣布端侧智能首个里程碑，发布开源库 Edge、1.3B 参数 SSM 语言模型 Rene，以及支持低延迟实时流式生成的 Sonic On-Device 私有测试版。Rene 采用 Mamba-2 混合架构，在多项基准上表现与同级 SOTA 小模型相当，Edge 已支持 Apple MLX 和量化推理。

    The on-device intelligence update

    ors and environments. Edge brings together open-weights SSM language models, including Rene, in one place and makes them available for Apple hardware through our new SSM optimized Metal kernels; Rene, an open-source 1.3B parameter language model pretrained from scratch with a hybrid Mamba-2 SSM backbone, and tailored to on-device inference; Sonic On-Device, in private beta. Sonic On-Device is the first ultra-realistic generative voice model that supports low latency real-time streaming on device

    匹配字段正文

    #company#产品更新#开源/仓库#端侧