跳到正文
Hacker News:AI 熱帖·· 11 天前AI 評分53

TimeNet 推出面向時序基礎模型的開源數據基礎設施

TimeNet: Open-source data infrastructure for Temporal Foundation Models

AI 導讀

TimeNet 提供面向時序基礎模型(TFMs)的開源數據基礎設施,將自有或倉庫數據納入統一數據層,並支援多類時序任務。團隊稱,在多個時序數據集上訓練後,OpenTSLM 和 SLIP 的 F1 表現平均分別提升 14% 和 22%。

正文

Image 1Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

TimeNet: Open-source data infrastructure for Temporal Foundation Models (github.com/opentslm) 15 points by rjakob1 hour ago | hide | past | favorite | 3 comments Hi HN, Current frontier LLMs show incredible results across text, video, and audio, but don't do as well with real-world scenarios like production lines, trading, drilling operations, or demand supply forecasting. They are not capable of comprehending this data, especially given a limited context. Models exist that are able to understand temporal information natively—Temporal Foundation Models (TFMs). However, there is a data problem: unlike all the public information that frontier models train on, temporal information is still scattered across organizations without a unified way to access it. That’s why we built TimeNet. It's like ImageNet for temporal data: - One uniform data layer for either our repository or your own data - Define tasks on top temporal data around classification, question answering, scalar prediction, temporal localization, forecasting, generation, editing, and correspondence (more to come) - Simple and performant API to integrate into ML frameworks. - First experimental results: training on multiple temporal datasets increased F1 performance on each dataset respectively by an average of 14% and 22% for OpenTSLM and SLIP respectively. - Install via ‘uv add timenet’ TimeNet is the result of a growing open and global collaboration between researchers from Aionic Labs, Stanford, ETH Zurich, UNSW Sydney, Sun Yat-sen University, Griffith University, Google, UCLA, TimeCopilot, Dartmouth, Harvard, Cambridge, Eindhoven University of Technology and KAIST. Paper: https://arxiv.org/abs/2610.04407 Quickstart: https://docs.timenet.ai/get-started.html TimeNet Registry: https://registry.timenet.ai/ There is currently an online hackathon ongoing on TimeNet (https://challenge.timenet.ai/) with up to $50k in cloud credit prizes, this should fuel your first model training. We would love to see you build on top of it and to hear your thoughts! help Image 2RealLast13 minutes ago | next[–] TimeNet already helped us out a lot accelerating training the next version(s) of OpenTSLM :) reply Image 3chrysander45 minutes ago | prev | next[–] Nice to see work from ETH reply Image 4nickdothutton13 minutes ago | prev[–] I've long said that "time is the neglected dimension", congratulations. reply Image 5 Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact Search:

來源:Hacker News:AI 熱帖 · news.ycombinator.com