model
stringclasses
6 values
teog 2013 puanı
float64
120
403
GPT-4o-mini
403
GPT-3.5-turbo-0125
392
gemma:2b
285.214234
mlx-community/Meta-Llama-3-8B-Instruct-4bit
173.535766
mlx-community/gemma-2-2b-it-4bit
135.758896
mlx-community/SmolLM-135M-Instruct-4bit
120

TEOG Scores Leaderboard

Welcome to the TEOG Scores Leaderboard! This repository contains the results of evaluating various large language models (LLMs) on the TEOG (Temel Eğitimden Ortaöğretime Geçiş) exam dataset. The TEOG exam is a standardized test in Turkey used for high school admissions, and this dataset provides a benchmark for assessing the performance of LLMs in Turkish educational tasks. Please remember that full score for TEOG is 500 points.

More Models Are Coming Really Soon!

Overview

In this repository, you will find:

  • Leaderboard: A ranking of LLMs based on their performance on the TEOG dataset.
  • Dataset: Details about the TEOG dataset used for evaluation.
  • Evaluation Methodology: Explanation of how the models were tested and scored.
  • Future Plans: Information about upcoming model evaluations and updates to the leaderboard.

Dataset

The TEOG dataset consists of multiple-choice questions from past TEOG exams. Each question includes:

  • A question stem
  • Multiple answer choices
  • The correct answer

The dataset is available here.

Evaluation Methodology

The models were evaluated using the following steps:

  1. Data Preprocessing: The TEOG questions were formatted to be compatible with the input requirements of each model.
  2. Model Inference: Each model was run on the TEOG dataset to generate answers for each question.
  3. Scoring: The model-generated answers were compared to the correct answers to calculate the accuracy and overall TEOG score. The evaluation formula has taken from this link For those models that do not give correct choice directly, I used semantic similarity to calculate more accurately. You can see the codes here.

Future Plans

I am continuously working on evaluating more models and improving the dataset. Upcoming updates will include:

  • Additional model evaluations and leaderboard updates.
  • Enhancements to the dataset with more TEOG exams.
  • Improved evaluation methodology for better accuracy.

Stay tuned for more updates and improvements!

License

This project is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) License. You are free to:

  • Share — copy and redistribute the material in any medium or format
  • Adapt — remix, transform, and build upon the material for any purpose, even commercially.

Under the following terms:

  • Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.

For more details, see the license.

Special Thanks

Special thanks to M. Ali Bayram for creating the TEOG dataset

Contact

For questions, suggestions, or contributions, please open an issue or contact me at [ardafincan@icloud.com].


I hope this leaderboard serves as a useful benchmark for the Turkish NLP community and helps in advancing the performance of LLMs on educational tasks.

Downloads last month
51
Edit dataset card