Newton2676/physics-solver-cpge

Physics Solver CPGE

A high-quality dataset of solved physics problems from French Classes Préparatoires aux Grandes Écoles (CPGE), designed for training and evaluating reasoning-capable language models on advanced physics problem solving.

Overview

Physics Solver CPGE contains physics exercises and their complete solutions covering the core topics taught in French preparatory classes.

The dataset emphasizes:

• Multi-step reasoning
• Mathematical derivations
• Physical modeling
• Symbolic manipulation
• Rigorous scientific explanations
• Exam-style problem solving

It is intended for:

• Physics reasoning benchmarks
• LLM fine-tuning
• Chain-of-thought research
• Educational AI systems
• Scientific tutoring assistants

---

Subjects Covered

The dataset includes problems from:

Mechanics

• Newtonian dynamics
• Energy methods
• Momentum conservation
• Oscillators
• Rotational dynamics
• Reference frames

Electromagnetism

• Electrostatics
• Magnetostatics
• Electromagnetic induction
• Electrical circuits

Thermodynamics

• First and second laws
• Ideal gases
• Thermodynamic transformations
• Heat transfer

Waves and Optics

• Wave propagation
• Interference
• Diffraction
• Geometrical optics

Modern Physics

• Quantum phenomena
• Relativity fundamentals
• Atomic physics

---

Dataset Structure

Each sample typically follows the structure:

{
  "problem": "...",
  "reasoning": "...",
  "solution": "...",
  "final_answer": "..."
}

Fields

Field	Description
problem	Original physics problem statement
reasoning	Intermediate reasoning steps
solution	Detailed derivation and explanation
final_answer	Final numerical or symbolic answer

---

Example

Problem

A particle of mass (m) moves in a harmonic potential. Determine its period of oscillation.

Reasoning

The restoring force is:

[ F = -kx ]

Applying Newton's second law:

[ m\ddot{x}+kx=0 ]

This differential equation describes a harmonic oscillator.

Solution

The angular frequency is:

[ \omega = \sqrt{\frac{k}{m}} ]

Therefore:

[ T = \frac{2\pi}{\omega} ]

Final Answer

[ T = 2\pi\sqrt{\frac{m}{k}} ]

---

Applications

Supervised Fine-Tuning (SFT)

Train language models to generate complete physics solutions from problem statements.

Reasoning Evaluation

Assess:

• Physical consistency
• Mathematical correctness
• Symbolic manipulation
• Multi-step scientific reasoning

Educational Systems

Develop AI tutors capable of explaining and solving advanced physics exercises.

---

Target Audience

• CPGE students
• Undergraduate physics students
• Engineering students
• AI researchers
• Educational technology developers

---

License

Please refer to the license specified on the Hugging Face dataset page.

---

Citation

@dataset{physics_solver_cpge,
  title={Physics Solver CPGE},
  author={Newton2676},
  year={2025},
  publisher={Hugging Face}
}

---

Acknowledgements

This dataset was created to support research in scientific reasoning, physics education, and advanced problem solving inspired by the standards of the French CPGE curriculum.

---

dataset_info: features: - name: id dtype: string - name: source dtype: string - name: domain dtype: string - name: domain_label dtype: string - name: level dtype: string - name: problem dtype: string - name: solution_steps dtype: string - name: answer dtype: string - name: has_cot dtype: bool - name: text dtype: string splits: - name: train num_bytes: 832568.4 num_examples: 711 - name: test num_bytes: 92507.6 num_examples: 79 download_size: 331328 dataset_size: 925076.0 configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-*