https://github.com/anishvdev/santa-2024_the-perplexity-permutation-puzzle
This submission uses a fine-tuned GPT-2 language model to rearrange the scrambled text sequences into coherent permutations with minimal perplexity. The model was fine-tuned on competition-like data to adapt to the vocabulary and structure of the problem. A beam search strategy was applied to explore multiple word orders.
https://github.com/anishvdev/santa-2024_the-perplexity-permutation-puzzle
ai chatgpt chatgpt-api perplexity perplexity-ai python
Last synced: 3 months ago
JSON representation
This submission uses a fine-tuned GPT-2 language model to rearrange the scrambled text sequences into coherent permutations with minimal perplexity. The model was fine-tuned on competition-like data to adapt to the vocabulary and structure of the problem. A beam search strategy was applied to explore multiple word orders.
- Host: GitHub
- URL: https://github.com/anishvdev/santa-2024_the-perplexity-permutation-puzzle
- Owner: AnishVDev
- License: mit
- Created: 2024-11-28T04:20:31.000Z (6 months ago)
- Default Branch: main
- Last Pushed: 2024-12-11T19:13:20.000Z (6 months ago)
- Last Synced: 2025-01-27T13:18:32.429Z (4 months ago)
- Topics: ai, chatgpt, chatgpt-api, perplexity, perplexity-ai, python
- Language: Python
- Homepage: https://www.kaggle.com/c/santa-2024
- Size: 14.6 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Santa 2024 - Perplexity Minimization Challenge
This project minimizes perplexity for reordered text in the Santa 2024 competition.
## Steps to Run
1. Install dependencies:
```bash
pip install -r requirements.txt
2. Fine-tune the model:
```bash
python src/fine_tune.py
4. Generate submission
```bash
python src/reorder_text.py
6. Validate evaluation locally
```bash
python src/evaluation.py --solution data/sample_submission.csv --submission results/submission.csv