Dataset research report
Legalbench research report
A reproducible data report with schema notes, generated chart evidence, suggested follow-up questions, and export-ready Helix queries.
Executive Summary
Dataset Card for Dataset Name Homepage: https://hazyresearch.stanford.edu/legalbench/ Repository: https://github.com/HazyResearch/legalbench/ Paper: https://arxiv.org/abs/2308.11462 Dataset Description Dataset Summary The LegalBench project is an ongoing open science effort to collaboratively curate tasks for evaluating legal reasoning in English large language models (LLMs). The benchmark currently consists of 162 tasks gathered from 40… See the full description on the dataset page: https://huggingface.co/datasets/nguha/legalbench.
Research Context
Legalbench: 5 rows by 3 columns. These exploratory charts are generated automatically from the data - open the dataset in Helix to ask your own questions.
Data Profile
Chart Evidence
These views are generated from the dataset profile. Each chart is paired with a Helix query so it can be opened, adjusted, and exported.
Total column_1 by answer
Top answer values ranked by summed column_1.
Open and export this chartFollow-Up Queries
Preview Rows
| # | column_1integer | answertext | texttext |
|---|---|---|---|
| 1 | 0 | generic | The mark "Ivory" for a product made of elephant tusks. |
| 2 | 1 | descriptive | The mark "Tasty" for bread. |
| 3 | 2 | suggestive | The mark "Caress" for body soap. |
| 4 | 3 | arbitrary | The mark "Virgin" for wireless communications. |
| 5 | 4 | fanciful | The mark "Aswelly" for a taxi service. |
Data Dictionary
- index numeric
- answer text
- text text
Method And Limits
- Load the catalog entry and preview rows from the processed dataset file.
- Infer numeric, categorical, time, and location fields from real columns.
- Generate a small set of defensive Plotly chart specifications from that profile.
- Expose each chart idea as a query link so the report can be rerun or exported in Helix.
This report is intentionally reproducible. It uses the local catalog metadata and generated chart specifications rather than claiming external conclusions beyond the dataset.