Prior Labs发布TabPFN-3.5：表格基础模型默认设置击败Kagg
Prior Labs Releases TabPFN-3.5: A Tabular Foundation Model That Beats the Winning Otto Kaggle Solution With Default Settings
表格AI领域的重要突破，用极简配置直接刷新了多年前的Kaggle纪录，对做结构化数据的同学极具参考价值。
Prior Labs has released TabPFN-3.5, the newest version of its tabular foundation model. It predicts on a table in a forward pass, with no per-dataset training or tuning. Prior Labs reports first place across 7 tabular benchmarks. A separate demonstration shows it beating the winning solution of a famous 2015 Kaggle competition.
Prior Labs 发布了 TabPFN-3.5，这是其表格基础模型的最新版本。它在前向传播中直接对表格进行预测，无需针对每个数据集进行训练或微调。Prior Labs 报告称其在 7 个表格基准测试中均获得第一名。单独的演示显示它击败了 2015 年著名 Kaggle 竞赛的获胜方案。
Deployable? Yes, with a license. Open weights run locally for research, evaluation and Kaggle, but production use needs Prior Labs’ API or a commercial license.
可以部署吗？是的，但需要许可证。开放权重模型可在本地用于研究、评估和 Kaggle 竞赛，但生产环境使用需要 Prior Labs 的 API 或商业许可证。
The Otto Result
Otto 结果
The Otto Group Product Classification Challenge ran on Kaggle in 2015. It drew 3,505 teams competing for $10,000. Entrants sorted products into 9 categories using 93 obfuscated count features. Submissions were scored with multi-class log loss, where lower is better.
Otto Group 产品分类挑战赛于 2015 年在 Kaggle 上举行。共有 3,505 支队伍参赛，争夺 10,000 美元奖金。参赛者使用 93 个混淆计数特征将产品分为 9 类。提交结果的评分采用多类对数损失（multi-class log loss），分数越低越好。
The winning solution came from Gilberto Titericz and Stanislav Semenov. Both have held the world #1 Kaggle grandmaster ranking. Their entry was a multi-layer stack of 36 models built on hand-crafted features.
获胜方案来自 Gilberto Titericz 和 Stanislav Semenov。两人曾同时拥有 Kaggle 全球排名第一的大师级排名。他们的参赛作品是一个由手工构建特征支撑的 36 层模型堆叠结构。
Nick Erickson, co-creator of AutoGluon and an AI researcher at Prior Labs, has chased that score for years. According to Erickson, AutoGluon placed rank 23 in its 2020 paper. AutoGluon 1.0 reached rank 14 in 2023, and AutoGluon 1.6 reached rank 9 in August 2026.
AutoGluon 的联合创作者、Prior Labs 的 AI 研究员 Nick Erickson 多年来一直致力于追赶这一分数。据 Erickson 介绍，AutoGluon 在 2020 年的论文中排名第 23 位。AutoGluon 1.0 在 2023 年达到第 14 位，而 AutoGluon 1.6 在 2026 年 8 月达到第 9 位。
The final stretch was the hardest. Moving from rank 50 to rank 10 cut log loss from 0.41 to 0.40. Reaching the winning 0.382 from rank 10 took a further 0.018, nearly double.
最后的冲刺阶段最为艰难。从第 50 名提升到第 10 名，将对数损失从 0.41 降至 0.40。从第 10 名追至获胜分数 0.382，又进一步降低了 0.018，降幅接近一倍。
TabPFN-3.5 scores 0.375 on the private leaderboard. Erickson says it ran on raw data with default settings. It took about a minute on an RTX PRO 6000 GPU. The model was pretrained only on synthetic data and never saw Otto or any Kaggle dataset. A reproducible Kaggle notebook is public.
TabPFN-3.5 在私有排行榜上的得分为 0.375。Erickson 表示，该模型在默认设置下直接在原始数据上运行。在 RTX PRO 6000 GPU 上耗时约一分钟。该模型仅在合成数据上进行过预训练，从未接触过 Otto 或任何 Kaggle 数据集。一个可复现的 Kaggle Notebook 已公开。
Benchmark Results
基准测试结果
The technical report lists first place on TabArena, BeyondArena, STRABLE, MulTaBench, RelArena-α, TALENT and ScoringBench. The top entry is not always the base model. TabPFN-3.5-Thinking holds first on TabArena, BeyondArena, STRABLE and MulTaBench. An internal TabPFN-Rel harness preview takes RelArena-α.
技术报告列出了在 TabArena、BeyondArena、STRABLE、MulTaBench、RelArena-α、TALENT 和 ScoringBench 上获得的冠军成绩。最高分并不总是来自基础模型。TabPFN-3.5-Thinking 在 TabArena、BeyondArena、STRABLE 和 MulTaBench 上位列第一。内部预览版 TabPFN-Rel harness 在 RelArena-α 上取得领先。
On TabArena, a living benchmark of 51 datasets, Thinking reaches 1910 Elo. The base model scores 1866, ahead of TabFM+ at 1823. Prior Labs says the base model beats AutoGluon 1.6 extreme by 130 Elo in a fifth of the time.
在包含 51 个数据集的动态基准 TabArena 上，Thinking 版本达到 1910 Elo 分。基础模型得分为 1866，高于 TabFM+ 的 1823 分。Prior Labs 表示，基础模型在五分之一的时间内比 AutoGluon 1.6 extreme 高出 130 Elo 分。
BeyondArena spans 142 datasets with grouped, temporal, wide, text-rich and high-cardinality data. TabPFN-3.5 finishes about 150 Elo points ahead of the previous overall leader. Tuned and ensembled MLPs still lead on its grouped, temporal and large-data slices.
BeyondArena 涵盖 142 个数据集，包含分组数据、时序数据、宽表数据、文本丰富型数据和高基数数据。TabPFN-3.5 的 Elo 积分比之前的总体领先者高出约 150 分。经过调优和集成的 MLP 在其分组数据、时序数据和大规模数据子集上仍保持领先。
What Changed Under the Hood
底层发生了哪些变化
- Wider model: The in-context transformer grows from 512 to 1024 dimensions. Parameters rise to 220M, up from 53M for TabPFN-3 classification.
- 1 checkpoint: A single multitask checkpoint now covers classification and regression.
- New cell encodings: Values pass through learned Fourier features and in-context ECDF ranks. The ranks do not change under monotonic transforms such as log scaling.
- Simpler preprocessing: Quantile transforms, robust scaling and SVD features are removed.
- Scale: Up to 1M rows, with 6,000 features recommended and 20,000 supported.
- Tuned prior: Synthetic data now emphasizes high-cardinality, wide and grouped tables.
- 更宽的模型：上下文学习（in-context）Transformer 的维度从 512 增加到 1024。参数量上升至 2.2 亿，而 TabPFN-3 分类模型的参数量为 5300 万。
- 单一检查点：现在一个多任务检查点同时覆盖分类和回归任务。
- 新的单元格编码：数值通过学习的傅里叶特征和上下文经验累积分布函数（ECDF）秩进行转换。在单调变换（如对数缩放）下，这些秩值保持不变。
- 简化的预处理：移除了分位数变换、鲁棒缩放和 SVD 特征。
- 规模：支持最多 100 万行数据，推荐特征数为 6000，最大支持 20000 个特征。
- 调优后的先验：合成数据现在更强调高基数、宽表和分组表格。
The KV cache stays roughly the same size as TabPFN-3, despite about 4x more parameters. Cached single-row predictions match TabPFN-3 speed. On large training sets, however, the base model runs up to 2x slower than TabPFN-3.
尽管参数量约为 TabPFN-3 的 4 倍，KV 缓存的大小大致保持不变。缓存的单行预测速度与 TabPFN-3 相当。然而，在大型训练集上，基础模型的运行速度比 TabPFN-3 慢多达 2 倍。
The Model Family
模型家族
- TabPFN-3.5: Open weights, 220M parameters, 8 estimators by default.
- TabPFN-3.5-Fast (alpha): Open weights, 84M parameters, up to 6x faster than the base model.
- TabPFN-3.5-Plus: API and enterprise only. Adds native text handling and FP8 attention.
- TabPFN-3.5-Thinking: Spends extra inference compute, with no LLMs, real data or search. Runs up to 12x faster than TabPFN-3-Thinking.
- TabPFN-3.5：开源权重，2.2 亿参数，默认包含 8 个估计器。
- TabPFN-3.5-Fast（alpha 版）：开源权重，8400 万参数，速度比基础模型快多达 6 倍。
- TabPFN-3.5-Plus：仅限 API 和企业版。新增原生文本处理和 FP8 注意力机制。
- TabPFN-3.5-Thinking：在不使用大语言模型（LLM）、真实数据或搜索的情况下，消耗额外的推理计算资源。其运行速度比 TabPFN-3-Thinking 快多达 12 倍。
Key Takeaways
关键要点
- TabPFN-3.5 scores 0.375 on Otto, beating the 2015 winning 0.382.
- The Otto run used raw data and default settings, in about 1 minute.
- Prior Labs reports rank 1 on 7 tabular benchmarks.
- The base model grows to 220M parameters with Fourier and ECDF encodings.
- Open weights are non-commercial; production needs the API or a license.
- TabPFN-3.5 在 Otto 数据集上得分为 0.375，击败了 2015 年获胜方案的 0.382 分。
- Otto 实验使用原始数据和默认设置，耗时约 1 分钟。
- Prior Labs 报告其在 7 个表格基准测试中排名第一。
- 基础模型通过傅里叶特征和 ECDF 编码将参数量扩展至 2.2 亿。
- 开源权重仅限非商业用途；生产环境需要使用 API 或获取许可证。
Check out the Technical report and Technical details. All credit goes to the researcher of this project. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
请查阅技术报告和技术细节。所有功劳归于本项目的研究人员。也欢迎在 Twitter 上关注我们，别忘了加入我们有 15 万+成员的 ML SubReddit，并订阅我们的新闻通讯。等等！你在 Telegram 上吗？现在你也可以加入我们的 Telegram 群组。
Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us
需要与我们合作推广您的 GitHub 仓库、Hugging Face 页面、产品发布或网络研讨会等吗？请联系我们
The post Prior Labs Releases TabPFN-3.5: A Tabular Foundation Model That Beats the Winning Otto Kaggle Solution With Default Settings appeared first on MarkTechPost.
本文最初发表于 MarkTechPost。标题为《Prior Labs 发布 TabPFN-3.5：一款使用默认设置即可击败 Kaggle Otto 获胜方案的表格基础模型》。
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