MatthiasPicard's picture
Update README.md
17f802c verified
metadata
title: Submission Template
emoji: 🔥
colorFrom: yellow
colorTo: green
sdk: docker
pinned: false

Model Description

This space is dedicated to the text task of the Frugal AI Challenge. The final model employed is a Qwen2.5-3B-Instruct with LoRA adapters trained on a diverse mix of approximately 95,000 samples, encompassing both real and synthetic data. The dataset was open-sourced at MatthiasPicard/Frugal-AI-Train-Data-88k. The fine-tuned model, along with training logs, was open-sourced at MatthiasPicard/ModernBERT_frugal_88k.

To optimize inference time, the model was quantized to 8 bits to reduce memory usage and increase performance speed.

Note: The inference script includes both model and tokenizer loading. As a result, the first evaluation of our model in the submission space will consume more energy than subsequent evaluations.

Labels

  1. No relevant claim detected
  2. Global warming is not happening
  3. Not caused by humans
  4. Not bad or beneficial
  5. Solutions harmful/unnecessary
  6. Science is unreliable
  7. Proponents are biased
  8. Fossil fuels are needed