TimeRLM - Model & AnomalyXL datasets
Collection
3 items • Updated
The dataset viewer is not available for this split.
Parquet error: Scan size limit exceeded: attempted to read 2053370943 bytes, limit is 300000000 bytes
Make sure that
1. the Parquet files contain a page index to enable random access without loading entire row groups2. otherwise use smaller row-group sizes when serializing the Parquet files
Error code: TooBigContentError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
AnomalyXL
Long-context multivariate time-series anomaly benchmark, in two partitions:
- MCQ: multiple-choice questions over six task categories (presence, start time, categorization, magnitude, identification, indicator), projected from the Localize partition so paired rows share the same underlying series. Gold answers are option text.
- Localize: open evidence-grounded questions (localize, classify-with-evidence, measure magnitude, localize-all-channels, lead/lag) with structured JSON golds scored by IoU / tolerance-banded metrics.
Series are stored per-row (channel_names / channel_values) with injection
ground truth (injection_* columns). test is the seed-42 release grid
(800 Localize / 248 MCQ rows); train/validation are the seed-100 training
grids. Row ids are stable join keys across partitions
(paired__<localize id>__<template> links MCQ rows to their sources).
- Downloads last month
- 130