독서리서치레이더투자 프레임워크
로그인 / 회원가입
← 읽기로 돌아가기
뉴스주제 · Mitra-v2중대2 분 소요DisconfirmAI 편집부

Amazon Mitra-v2: Expanding Synthetic Pretraining Diversity to Replace Network Depth, Self-Reported Leading TabArena

The Amazon Science team has released the tabular foundation model Mitra-v2. Its architecture remains identical to its predecessor; performance gains stem from scaling up synthetic pretraining data (10x rows, ~3x columns) and a novel hybrid causal model prior. This prior combines neural networks, decision trees, Gaussian processes, and autoregressive mechanisms within a single causal graph, allowing one synthetic table to carry mixed inter-column relationships found in real tables. Self-reported results show it leads the single-model track on 51 datasets in TabArena and ranks first on the ScoringBench regression benchmark.

2026-11-10 23:37 GMT+8

The Amazon Science team released the tabular foundation model Mitra-v2. Author Chris Kong provided a clear assessment in his blog post: the substantive improvement of this model over the previous generation, Mitra-v1, does not come from architectural changes—the Transformer still attends to both rows and columns simultaneously—but rather from two changes in the synthetic pretraining data: increased scale and diversity.

By 'increased scale,' we mean the number of rows in the synthetic tables used for pretraining is 10 times that of v1, with approximately three times as many columns, enabling the model to encounter tables comparable to actual business scales during training. By 'increased diversity,' we refer to the introduction of a new hybrid structural causal model (SCM) prior. While v1's synthetic tables were generated separately by classical SCMs and tree models, v2's hybrid SCM prior constructs a single causal graph where each edge relationship can be defined by a small neural network, a decision tree, a smooth function from a Gaussian process, or a time-dependent autoregressive process. This allows a single synthetic table to carry the mixed relationship types common in real tables. A new hybrid optimizer is employed to maintain training stability.

로그인 / 회원가입
독서리서치레이더투자 프레임워크

At deployment, two lightweight wrappers extend the model's boundaries: singular value decomposition is used to compress dimensions for extremely wide tables; for classification tasks exceeding 10 classes, predictions from each node in the output layer are fed into a hierarchical tree for further refinement. Notably, Mitra-v2 is not strictly a zero-shot model—it performs brief automatic fine-tuning on each new table by default, producing eight copies whose predictions are averaged (bagging). This incurs additional computational overhead but yields higher accuracy.

On TabArena, a public benchmark comprising 51 real-world datasets, self-reported results indicate that Mitra-v2 outperforms v1 in 32 of the 33 datasets where both models run natively, and completes all 51 datasets within the one-hour budget (v1 timed out on more than a third of them). On the ScoringBench regression benchmark, fine-tuned Mitra-v2 ranks first. These are vendor self-test results; the blog post did not mention independent replication.

For practitioners in tabular machine learning, the core transferable insight from this blog post is: given an unchanged architecture, improvements in synthetic data generation strategies—specifically richer combinations of causal mechanisms and tables closer to real-world scales—can drive substantial improvements in model generalization capabilities. The author expects that richer synthetic priors and adaptive mixing will continue to advance this direction.

출처

  • amazon.scienceAmazon Science↗
주제

Mitra-v2

亚马逊科学团队发布表格基础模型Mitra-v2,其核心突破在于通过扩大合成预训练数据的规模与多样性来提升性能,而非加深网络架构。该模型在TabArena基准测试的51个真实数据集上领先单模型赛道,并在ScoringBench回归基准中排名第…

2 단계1 건업데이트됨 2026-11-13 19:45

현재까지의 과정

  1. 2026-09-03亚马逊发布《Mitra-v2技术报告》(arXiv:2609.04540),详细阐述模型架构、合成预训练策略及在TabArena和TALENT基准上的评估结果。arxiv.org ↗
  2. 2026-11-10亚马逊科学博客由Chris (Luyang) Kong发文介绍Mitra-v2,宣布其在TabArena单模型赛道领先,并开源Apache-2.0许可的代码与权重。amazon.science ↗

아직 해결되지 않은 문제

  • Mitra-v2在大规模生产环境中的实际推理延迟与成本效益相比传统梯度提升树的具体表现如何?
  • 除了Amazon Connect Customer,其他行业客户对Mitra-v2的独立采纳案例及其带来的具体业务指标提升是否有公开数据支持?
전체 주제 읽기 →
리서치 계속하기 →