Unverified50% confidenceReleaseExact time
谷歌发布了两款基础模型TabFM（面向表格数据）和TimesFM（面向时间序列）
1
Sources
50%
Confidence
Long-term
Relevance
7/14/2026
First Seen
Sources
Zer0Fit：用MCP将谷歌TabFM与TimesFM接入LLM聊天界面
redditr/MachineLearning7/12/2026
Related Claims
UnverifiedGoogle officially disclosed that TimesFM was pre-trained on 200B time points of data, covering multiple domains including finance, retail, energy, and healthcare.77% similarUnverifiedGoogle's TimesFM is built on the PatchTST approach and performs well on multiple public benchmarks73% similarUnverifiedGoogle Research推出TabFM表格数据零样本基础模型，采用交替行列注意力、行压缩和上下文学习Transformer三大机制，在Tab Arena基准（38个分类数据集与13个回归数据集）上TabFM Ensemble分类与回归双任务均位列第一66% similarVerifiedThe DORA framework proposed by Google contains four key metrics: deployment frequency, change lead time, change failure rate, and service recovery time62% similarUnverifiedGoogle open-sourced the Gemma series of small models, drawing on the successful path of Meta's open release of the LLaMA series.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/503945API
curl https://kongchang.com/api/v1/knowledge/claims/503945MCP
get_claim(id=503945)