Skip to content

A benchmark dataset of electrical engineering problems with step-by-step solutions for evaluating and training reasoning models.

Notifications You must be signed in to change notification settings

Norbera0/EEReasonBench

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 

Repository files navigation

library_name tags license task_categories language pretty_name size_categories
datasets
electrical-engineering
question-answering
problem-solving
reasoning
education
mathematics
physics
apache-2.0
text-generation
question-answering
en
EEReasonBench
n<1K

EEReasonBench: A Reasoning Benchmark for Electrical Engineering

Dataset Description

This dataset provides a collection of electrical engineering problems covering various subjects, including Circuits, Machines, Power Systems, Power Plants, etc. Problems include multiple-choice questions (conceptual and numerical) with detailed, step-by-step solutions formatted in Markdown and LaTeX.

The dataset is designed to serve as both a benchmark for evaluating model performance on electrical engineering reasoning tasks and as a resource for fine-tuning models to improve their capabilities in technical problem-solving within the electrical engineering domain.

Features:

  • Structured JSON format per problem
  • Fields include id, subject, type, problem, choices, given, solution_steps, answer, and concepts
  • Solutions emphasize clear reasoning and link steps to fundamental principles

Dataset Structure

Data Instance

{
    "id": "DC-012",
    "subject": "DC Circuits",
    "type": "Problem Solving",
    "problem": "A 120 V dc motor draws a current of 200 A and is located 1,000 ft from the supply. If the copper transmission wire has a diameter of 0.45 inch and with specific resistance of 10.4 ohm-circular mil per foot, what must be the voltage of the supply source?",
    "choices": {
      "A": "129.10 V",
      "B": "134.30 V",
      "C": "136.90 V",
      "D": "140.50 V"
    },
    "given": "$$V_{motor} = 120\\ V$$\n$$I_{motor} = 200\\ A$$\n$$L = 1000\\ ft$$\n$$D = 0.45\\ inch$$\n$$\\rho = 10.4\\ \\Omega\\text{-CM/ft}$$",
    "solution_steps": "1. **Identify the problem type**: This is a voltage drop problem in a DC circuit with resistance in the transmission lines. We need to apply both Ohm's Law and Kirchhoff's Voltage Law.\n\n2. **Convert diameter to appropriate units**: Wire diameter is given in inches but resistivity in circular mils, so we need consistent units:\n$$D = 0.45 \\text{ inch} \\times 1000 = 450 \\text{ mils}$$\n\n3. **Calculate cross-sectional area**: For circular conductors, area in circular mils is diameter squared:\n$$A_{CM} = D^2 = 450^2 = 202{,}500 \\text{ CM}$$\n\n4. **Calculate resistance of one wire**: Using the resistivity formula which relates material property to physical dimensions:\n$$R_{wire} = \\frac{\\rho \\cdot L}{A_{CM}} = \\frac{10.4 \\times 1000}{202{,}500} \\approx 0.05136\\ \\Omega$$\n\n5. **Determine total circuit resistance**: Since current flows through both supply and return wires:\n$$R_{total} = 2 \\times R_{wire} = 2 \\times 0.05136 = 0.10272\\ \\Omega$$\n\n6. **Calculate voltage drop**: Using Ohm's Law to find voltage lost in transmission:\n$$V_{drop} = I_{motor} \\times R_{total} = 200\\text{ A} \\times 0.10272\\text{ }\\Omega = 20.544\\text{ V}$$\n\n7. **Verification**: We can verify this drop by calculating power loss in the transmission line:\n$$P_{loss} = I^2 \\times R_{total} = (200\\text{ A})^2 \\times 0.10272\\text{ }\\Omega = 4,108.8\\text{ W}$$\nThis significant power loss (about 4.1 kW) explains why high-voltage transmission is preferred for long distances.\n\n8. **Apply Kirchhoff's Voltage Law**: The supply voltage must overcome both the motor voltage and the transmission line drop:\n$$V_{supply} = V_{motor} + V_{drop} = 120\\text{ V} + 20.544\\text{ V} = 140.544\\text{ V}$$\n\n9. **Analyze results**: The supply must provide 17.1% more voltage than what's needed at the motor, highlighting the importance of conductor sizing in electrical distribution systems.\n\n10. **Final Answer**: Comparing our calculated value (140.544 V) with the given choices, the closest match is **D. 140.50 V**.",
    "answer": "D. 140.50 V",
    "concepts": [
        "Ohm's Law",
        "Kirchhoff's Voltage Law",
        "Power Transmission",
        "Voltage Drop",
        "Resistivity"
    ]
}

Data Fields

  • id: (string) Unique problem ID
  • subject: (string) EE subject area
  • type: (string) "Problem Solving" or "Objective Type"
  • problem: (string) Question text
  • choices: (dict) Multiple-choice options (A-D)
  • given: (string) Known values/conditions (LaTeX)
  • solution_steps: (string) Numbered step-by-step solution (Markdown/LaTeX)
  • answer: (string) Correct choice ("Letter. Answer text")
  • concepts: (list[string]) Key concepts

Data Splits

The dataset contains a single train split.

Dataset Creation

The problems were sourced from electrical engineering educational materials. Solutions were refined, formatted, and enhanced for clarity and pedagogical value. This process included structuring the data, applying LaTeX, adding reasoning steps, tagging concepts, and verifying calculations. The dataset contains only technical problem data, with no personal information.

Considerations for Use

  • Scope: Primarily covers core undergraduate EE topics
  • Potential Errors: While curated carefully, errors from source materials or the enhancement process may exist. Users should verify critical applications
  • Rendering: Requires tools capable of rendering Markdown and LaTeX

How to Use

from datasets import load_dataset

# Replace with your Hugging Face repository name
dataset = load_dataset("Noru1/EEReasonBench")

# Access data
print(dataset['train'][0])

Licensing Information

Licensed under the Apache License, Version 2.0.

Citation Information

@misc{ee_problems_dataset_2025,
  author       = {Norbert John Ibera},
  title        = {EEReasonBench: A Reasoning Benchmark for Electrical Engineering},
  year         = {2025},
  publisher    = {Hugging Face},
  journal      = {Hugging Face Hub},
  howpublished = {\url{https://huggingface.co/datasets/Noru1/EEReasonBench}}
}

About

A benchmark dataset of electrical engineering problems with step-by-step solutions for evaluating and training reasoning models.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published