[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-ibm-granite-patchtst-fm-r2-zero-shot-apache":3,"topics-all":38,"news-related-365b770a-2cca-40a0-beb0-1eff823702c0":57},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"365b770a-2cca-40a0-beb0-1eff823702c0","IBM 开源 Granite Time Series PatchTST-FM-r2:零样本 SOTA,Apache 2.0 商用许可","IBM 发布 385M 参数的 Granite Time Series PatchTST-FM-r2,采用 conformer 架构,在 GIFT-Eval 零样本榜单上排名第二、并在可商用宽松许可证类别里位居第一,代码、权重、推理流水线全部公开。","时序预测在过去几年一直是「一个数据集训一个模型」的脏活累活。每条业务线、每类传感器、每个地区,都要单独维护一套参数。基础模型把这条路堵上之后,大家真正关心的是两件事:在从未见过的数据上能不能直接出预测,以及许可证允不允许商用。IBM 在 9 月 9 日发布的 Granite Time Series PatchTST-FM-r2 同时回应了这两点。\n\n## 模型本体:385M 参数 + conformer 块\n\nPatchTST-FM-r2 沿用 patch 化表征,这是 PatchTST 系列高效的关键。但内部结构从标准 Transformer 层换成了 conformer 层:每块里,多头自注意力两边各夹一个半步前馈,中间再加一层时间卷积。自注意力负责跨 patch 的长程依赖,卷积负责 patch 内部的局部结构,两者职责分开。骨干由 20 层扩到 30 层,patch 用 50% 重叠 + Hamming 窗加权,配合 overlap-and-add 预测来平滑 patch 边界。最终模型约 3.85 亿参数,支持最长 8192 步上下文,预测头一次输出 99 个分位数,既能给点预测也能给区间估计。\n\n## GIFT-Eval 上的位置\n\nGIFT-Eval 是目前覆盖最广的零样本时序基准。截至 9 月 8 日,PatchTST-FM-r2 在「replicable、严格零样本」类别里 CRPS 几何均值 0.467、MASE 几何均值 0.6846,排名第二,仅次于 TimesFM-3。考虑到 TimesFM-3 是 Google 的非商用许可,PatchTST-FM-r2 实际上是「宽松商用许可证」类别里的第一名。即便把那些允许拿 GIFT-Eval 训练集做预训练的「pretrained」模型也拉进来比,它仍然排在 CRPS 第三、MASE 第四,优于 Chronos-2、Timer-S1 和 Toto 几个变体,而它的参数体量比它们小很多。\n\n## 训练数据:可审计的四个来源\n\nIBM 把预训练语料拆得很清楚:GiftEvalPretrain 的部分数据、KernelSynth 改版周期核的合成数据、按 Chronos 路线生成的 TSMixup 语料(严格剔除 GIFT-Eval 评测集)、约 50 万条长度 4096 的合成 CauKer 序列。整套数据列表公开,而不是只甩一句「我们用了大量数据」。\n\n## 许可证与生态:门槛降到最低\n\nPatchTST-FM-r2 同时以 Apache 2.0 和 OpenMDW 1.0 双协议开源,企业可以在两者中任选其一。代码、权重、推理流水线全部在 GitHub 的 granite-tsfm 仓库和 Hugging Face 上,推理代码兼容上一代 PatchTST-FM-r1 的检查点。Confluent 那边,IBM 已经在 Flink on Confluent Cloud 的 Early Access 中接入了 PatchTST-FM-r1、FlowState-r1.1、TTM-r3 和 TSPulse,把基础模型直接搬进流式应用。\n\n## 一段话总结\n\nPatchTST-FM-r2 不是又一次「榜单刷分」。它把时序基础模型的可商用门槛打到了 Apache 2.0、把架构改动从论文里搬到了开源代码、再加上 conformer 块这种「语音模型跨界来」的实操细节。对于要在生产里跑预测、又不愿意被许可协议卡住脖子的团队,这是当下为数不多能直接拿来用的选项。","https:\u002F\u002Fhuggingface.co\u002Fblog\u002Fibm-research\u002Fibm-releases-sota-granite-time-series","341f2956-bab4-471d-8d26-454c68615dd6",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"7ac06d8e-b074-4147-abfc-ffaa4c6b8744","ai-efficiency",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"120fa59a-ff6f-4537-9bf5-f818df636a0e","benchmark",{"id":19,"name":20,"slug":20,"description":14,"color":14},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":22,"name":23,"slug":23,"description":14,"color":14},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"2274b7cb-2ac5-4519-8bcc-3a2d7a68b3ab","en","IBM Releases SOTA Granite Time Series PatchTST-FM-r2 Under Apache 2.0","IBM ships a 385M-parameter Granite Time Series PatchTST-FM-r2 with conformer blocks. It ranks second on GIFT-Eval zero-shot and first among permissive-license models; weights, code, and inference pipeline are public.","Time-series forecasting has long been a messy, per-dataset chore. Every business line, every sensor type, every region needs its own trained model. Foundation models change that, but two questions decide whether anyone can actually use them in production: does the model generalize zero-shot to data it has never seen, and does the license allow commercial deployment? IBM's Granite Time Series PatchTST-FM-r2, released on September 9, 2026, addresses both questions at once.\n\n## The model itself: 385M parameters, conformer blocks\n\nPatchTST-FM-r2 keeps the patch-based representation that made the PatchTST family efficient, but swaps standard Transformer layers for conformer blocks. Inside each block, multi-head self-attention sits between two half-step feed-forward layers, with a temporal convolution layer added in. Self-attention captures long-range dependencies across patches; convolution handles local structure within patches. The two responsibilities no longer fight each other. The backbone grows from 20 to 30 layers; patches now use 50% overlap with Hamming-window weighting, paired with overlap-and-add forecasting to smooth patch boundaries. The final model has roughly 385 million parameters, supports context lengths up to 8192 steps, and a prediction head that emits 99 quantiles, giving both point forecasts and uncertainty intervals.\n\n## Where it sits on GIFT-Eval\n\nGIFT-Eval is currently the most comprehensive zero-shot time-series benchmark. As of September 8, 2026, PatchTST-FM-r2 lands at CRPS geometric mean 0.467 and MASE geometric mean 0.6846 in the replicable, strictly zero-shot category — second place overall, behind only TimesFM-3. Since TimesFM-3 ships under a non-commercial license, PatchTST-FM-r2 is effectively the top model in the permissive, commercial-friendly license class. Even when pretrained models that were allowed to include GIFT-Eval training data are added to the comparison, PatchTST-FM-r2 still ranks third on CRPS and fourth on MASE, beating Chronos-2, Timer-S1, and Toto variants while staying a fraction of their size.\n\n## Training data: four auditable sources\n\nIBM breaks down the pretraining corpus explicitly. Four sources: a subset of GiftEvalPretrain; synthetic data built on KernelSynth with modified periodic kernels; a TSMixup corpus generated using the Chronos recipe, with all GIFT-Eval evaluation sets excluded; and roughly 500,000 synthetic CauKer sequences of length 4096. The full list is public, not a hand-wave.\n\n## Licensing and ecosystem: the lowest possible bar\n\nPatchTST-FM-r2 is dual-licensed under Apache 2.0 and OpenMDW 1.0 — enterprises can pick either. Code, weights, and the inference pipeline all ship via the granite-tsfm repository on GitHub and Hugging Face, and the inference code stays backward-compatible with PatchTST-FM-r1 checkpoints. On the streaming side, IBM has already wired PatchTST-FM-r1, FlowState-r1.1, TTM-r3, and TSPulse into Flink on Confluent Cloud under an Early Access program, putting the foundation models directly into stream processing pipelines.\n\n## The short version\n\nPatchTST-FM-r2 is not just another leaderboard climb. It drops the commercial-use bar to Apache 2.0, ships the architectural changes as open code rather than a paper claim, and borrows conformer blocks from speech models in a way that translates cleanly to time series. For teams that need real production forecasting without getting blocked by licensing, this is one of the few options that genuinely works today.","ibm-granite-patchtst-fm-r2-zero-shot-apache","2026-09-12T11:00:00Z","2026-09-12T11:05:17.422900Z","2026-09-12T11:05:17.422909Z",true,"agent",60,[39,48],{"slug":40,"tag_slug":40,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":35,"created_at":46,"modified_at":47},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":49,"tag_slug":49,"title_zh":50,"title_en":51,"intro_zh":52,"intro_en":53,"id":54,"is_active":35,"created_at":55,"modified_at":56},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":58},[59,64,69,74,79,84],{"id":60,"title":61,"news_slug":62,"published_at":63},"4c4a2a9e-f69b-4985-bd42-97ab2ef4e2ac","Spark-X2.5-4B 开源:4B 跑 1M 上下文,22 项基准打 9B 级 Qwen3.5","spark-x2-5-4b-apache-open-source","2026-09-16T01:30:00+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"550cee5e-18e8-4236-9304-7207ebc221a8","Agnes 3.0 Flash 开源:72 层仅 18 层带 KV 缓存,33B 单卡跑 262k 上下文","agnes-3-0-flash-preview-open-weights","2026-09-13T15:20:00+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"c745abb4-d608-4ea6-884b-5176d7134d71","IFM 开源 K2 Horizon 六模型：训练数据全放，7B 刷榜成绩 82 被自己砍到 70.6","ifm-k2-horizon-open-fleet-audit","2026-09-05T23:07:55+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"741bd34c-7134-4e8e-ab45-4f53dc576a6b","腾讯 Hy4 登顶 9 月开源榜:79.87 分超 Qwen3.8 Max,Anthropic 包揽总榜前三","tencent-hy4-tops-open-source-benchlm-september","2026-09-01T17:10:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"4c7f5330-3aff-458a-9ef5-f04cc5585703","微信视觉团队开源 WeMM 嵌入模型:2B 反超 8B 前基线,9B 达 MMEB-v2 80.6","wemm-embedding-wechat-multimodal","2026-08-26T21:07:30+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"f9cf9f03-6aca-4d29-94d3-5c6acfeaf435","匿名模型 OX Alpha 短暂登顶 OpenRouter 编码榜:研究者推测底座指向智谱 GLM-5.x","ox-alpha-stealth-openrouter-glm-5-zhipu","2026-08-24T03:00:00+00:00"]