# Python
__pycache__/
*.py[cod]
*.pyo
*.egg-info/
build/
dist/
wheels/

# Virtual environments
.venv/
venv/

# macOS
.DS_Store

# Environment / secrets
.env
.env.*
!.env.example

# Model checkpoints
checkpoints/
checkpoints_v3/
checkpoints_v4/
checkpoints_v6/
*.pt
*.pth
*.safetensors

# Training outputs
runs/
wandb/
mlruns/

# Python tooling
.pytest_cache/
.mypy_cache/
.ruff_cache/

# Jupyter
.ipynb_checkpoints/

# Large scientific datasets
datasets/pdfs/
datasets/raw/
datasets/extracted/
datasets/cleaned/
datasets/tokens/
datasets/tokens_v2/

# Generated corpora
datasets/scientific_corpus*.jsonl
datasets/scientific_documents_v2*.jsonl

# Large training splits
datasets/train*.jsonl
datasets/val*.jsonl

# Generated tokenizer artifacts
datasets/tokenizer/
datasets/tokenizer_v2/

# Datasets folder (all data)
datasets/

# Evaluation result JSON dumps at root
evaluation_results*.json

# Release zips & folders
*.zip
AstroGPT_50M_RELEASE/

# Kaggle dataset folder
kaggle_dataset/

# Scripts folder (misc shell scripts)
scripts/

# Backup files
*.backup

# Evaluation result files (keep code, not results)
src/evaluation/results/

# Checkpoints (50m and others)
checkpoints_50m/

# Astro-Eval results and experiments
astro_eval_results/
astro_eval_experiments/
