PIQA

Overview

PIQA (Physical Interaction QA) is a benchmark for evaluating AI models' understanding of physical commonsense - how objects interact in the physical world and what happens when we manipulate them.

Task Description

Task Type: Physical Commonsense Reasoning
Input: Goal/question with two possible solutions
Output: More physically plausible solution (A or B)
Focus: Physical world knowledge and intuitive physics

Key Features

Tests understanding of physical object properties
Binary choice between plausible/implausible solutions
Requires intuitive physics reasoning
Covers everyday physical scenarios
Adversarially filtered to reduce biases

Evaluation Notes

Default configuration uses 0-shot evaluation
Uses simple multiple-choice prompting
Evaluates on validation split
Simple accuracy metric

Properties

Property	Value
Benchmark Name	`piqa`
Dataset ID	extraordinarylab/piqa
Paper	N/A
Tags	`Commonsense`, `MCQ`, `Reasoning`
Metrics	`acc`
Default Shots	0-shot
Evaluation Split	`validation`
Train Split	`train`

Data Statistics

Metric	Value
Total Samples	1,838
Prompt Length (Mean)	426.52 chars
Prompt Length (Min/Max)	220 / 2335 chars

Sample Example

Subset: default

{
  "input": [
    {
      "id": "a0600392",
      "content": "Answer the following multiple choice question. The entire content of your response should be of the following format: 'ANSWER: [LETTER]' (without quotes) where [LETTER] is one of A,B.\n\nHow do I ready a guinea pig cage for it's new occupants?\n ... [TRUNCATED] ... ps, you will also need to supply it with a water bottle and a food dish.\nB) Provide the guinea pig with a cage full of a few inches of bedding made of ripped jeans material, you will also need to supply it with a water bottle and a food dish."
    }
  ],
  "choices": [
    "Provide the guinea pig with a cage full of a few inches of bedding made of ripped paper strips, you will also need to supply it with a water bottle and a food dish.",
    "Provide the guinea pig with a cage full of a few inches of bedding made of ripped jeans material, you will also need to supply it with a water bottle and a food dish."
  ],
  "target": "A",
  "id": 0,
  "group_id": 0,
  "metadata": {}
}

Note: Some content was truncated for display.

Prompt Template

Prompt Template:

Answer the following multiple choice question. The entire content of your response should be of the following format: 'ANSWER: [LETTER]' (without quotes) where [LETTER] is one of {letters}.

{question}

{choices}

Usage

Using CLI

evalscope eval \
    --model YOUR_MODEL \
    --api-url OPENAI_API_COMPAT_URL \
    --api-key EMPTY_TOKEN \
    --datasets piqa \
    --limit 10  # Remove this line for formal evaluation

Using Python

from evalscope import run_task
from evalscope.config import TaskConfig

task_cfg = TaskConfig(
    model='YOUR_MODEL',
    api_url='OPENAI_API_COMPAT_URL',
    api_key='EMPTY_TOKEN',
    datasets=['piqa'],
    limit=10,  # Remove this line for formal evaluation
)

run_task(task_cfg=task_cfg)

3.3 KiB Raw Blame History