IFM发布K2 Horizon系列6款开源模型及完整训练数据
Institute of Foundation Models (IFM) released K2 Horizon, 6 open AI models from…
全链路开源的旗舰级模型发布，包含罕见的数据配方与训练日志，对复现和研究极具价值。
Institute of Foundation Models (IFM) released K2 Horizon, 6 open AI models from 0.9B to 375B, covering tiny on-device models through large models for reasoning, coding and AI agents.
基础模型研究所（IFM）发布了 K2 Horizon，包含 6 个开源 AI 模型，参数量从 0.9B 到 375B 不等，涵盖用于设备端的小型模型以及用于推理、编码和 AI 智能体的大型模型。
For every model, IFM is releasing weights, training code, configs, data or detailed data recipes, intermediate checkpoints, training logs and evaluations, covering the full process from pre-training through reasoning and agent training.
对于每个模型，IFM 均发布了权重、训练代码、配置、数据或详细的数据配方、中间检查点、训练日志和评估结果，覆盖了从头预训练到推理和智能体训练的全过程。
The 0.9B, 3.7B and 7B models set new state of the art in their size classes, while the sparse 36B-A4B gets close to the dense 32B while activating only about 4B parameters per token using IFM's new Mixture-of-Value Attention architecture.
0.9B、3.7B 和 7B 模型在其尺寸类别中创造了新的最先进水平（SOTA），而稀疏的 36B-A4B 模型在使用 IFM 全新的 Mixture-of-Value Attention 架构、每 token 仅激活约 4B 参数的情况下，性能已接近稠密的 32B 模型。
Each model was trained on roughly 20T tokens, while released checkpoints and logs give researchers an unusually detailed record of how the models developed during training.
每个模型均在大约 20T token 上进行了训练，而发布的检查点和日志为研究人员提供了关于模型在训练期间发展过程的异常详细的记录。
@IFM_AI is also releasing xLLM, the production training infrastructure used to build K2 Horizon, plus the full agentic post-training codebase including the RL pipeline. (Github link in the last post to this thread).
@IFM_AI 还发布了 xLLM，这是构建 K2 Horizon 所使用的生产级训练基础设施，以及完整的智能体后训练代码库，包括强化学习（RL）管道。（该线程最后一条帖子中包含 Github 链接）。
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力