README.md

---
dataset_info:
- config_name: 1k
  features:
  - name: opinion
    dtype: string
  - name: instruction
    dtype: string
  - name: question
    dtype: string
  - name: choices
    sequence: string
  - name: answer
    sequence: int64
  - name: task
    dtype: string
  splits:
  - name: test
    num_bytes: 20746554.315389965
    num_examples: 1000
  - name: val
    num_bytes: 20191583.272299055
    num_examples: 1000
  download_size: 33445966
  dataset_size: 40938137.58768902
- config_name: sc
  features:
  - name: opinion
    dtype: string
  - name: instruction
    dtype: string
  - name: question
    dtype: string
  - name: choices
    sequence: string
  - name: answer
    sequence: int64
  - name: task
    dtype: string
  splits:
  - name: test
    num_bytes: 124966347
    num_examples: 5000
  - name: val
    num_bytes: 125253167
    num_examples: 5000
  download_size: 138383906
  dataset_size: 250219514
- config_name: songer
  features:
  - name: opinion
    dtype: string
  - name: instruction
    dtype: string
  - name: question
    dtype: string
  - name: choices
    sequence: string
  - name: answer
    sequence: int64
  - name: task
    dtype: string
  splits:
  - name: test
    num_bytes: 82357950
    num_examples: 5000
  - name: val
    num_bytes: 79866000
    num_examples: 5000
  download_size: 94738667
  dataset_size: 162223950
configs:
- config_name: 1k
  data_files:
  - split: val
    path: 1k/val-*
  - split: test
    path: 1k/test-*
- config_name: sc
  data_files:
  - split: val
    path: sc/val-*
  - split: test
    path: sc/test-*
- config_name: songer
  data_files:
  - split: val
    path: songer/val-*
  - split: test
    path: songer/test-*
---

CaselawQA is a benchmark comprising legal classification tasks, drawing from the Supreme Court and Songer Court of Appeals legal databases.
The majority of its 10,000 questions are multiple-choice, with 5,000 sourced from each database.
The questions are randomly selected from the test sets of the [Lawma tasks](https://huggingface.co/datasets/ricdomolm/lawma-tasks).
From a technical machine learning perspective, these tasks provide highly non-trivial classification problems where even the best models leave much room for improvement.
From a substantive legal perspective, efficient solutions to such classification problems have rich and important applications in legal research.

Homepage: https://github.com/socialfoundations/lawma