# ═══════════════════════════════════════════════════════════════
# Paritok .gitignore
# Scope: 4B SFT only. GRPO / DPO / 3B / 7B kept locally but not on GitHub.
# ═══════════════════════════════════════════════════════════════

# ── Python ──
__pycache__/
*.pyc
*.pyo
*.egg-info/
.venv/
venv/
env/
build/
dist/

# ── OS ──
.DS_Store
Thumbs.db

# ── IDE ──
.vscode/
.idea/
*.swp
*~

# ── Training data (not committed to GitHub; users regenerate) ──
# Any jsonl inside data_pipeline/ is treated as training data
data_pipeline/*.jsonl
data_pipeline/**/*.jsonl
# Old update/ dir kept locally (contents were moved out; only training-data jsonl remains)
update/

# ── Training outputs ──
training/models/

# ── GRPO / DPO / 3B / 7B (kept locally, not on GitHub) ──
training/train_grpo.py
training/reward.py
training/configs/grpo_config.yaml
deploy_grpo.sh
analysis/measure_r_ce_curve.py
deploy_sft_3b.sh
deploy_sft_7b.sh
training/configs/sft_config_coder3b.yaml
training/configs/sft_config_coder7b.yaml
training/train_dpo.py
training/configs/dpo_config_*
training/models/*.log
training/data/
training/scripts/

# ── Eval (internal proxy metrics; a different basis from the README SWE-bench numbers, not on GitHub) ──
eval/

# ── Backup files ──
*.trl-autogen-backup
*.backup

# ── Environment variables ──
.env
*.env
!.env.example

# ── Temp / scratch ──
tool-results/
scratchpad/
tmp/
*.log
!README*.md
!docs/*.md
compress_trace.jsonl
paritok.yaml

# ── Large model files uploaded to HF Hub (GitHub 100MB limit; keep only adapter_config.json) ──
training/models/checkpoint-2000/adapter_model.safetensors
training/models/checkpoint-2000/added_tokens.json
training/models/checkpoint-2000/chat_template.jinja
training/models/checkpoint-2000/merges.txt
training/models/checkpoint-2000/README.md
training/models/checkpoint-2000/special_tokens_map.json
training/models/checkpoint-2000/tokenizer.json
training/models/checkpoint-2000/tokenizer_config.json
training/models/checkpoint-2000/vocab.json
