# Python
__pycache__/
*.py[cod]
*.egg-info/
.venv/

# Data artifacts (generated datasets are large and fetched via HF Hub)
data/*.png
data/*.jpg
data/*.jpeg
data/goodscents_html_cache/
data/goodscents_ambiguous.json
data/*.pdf
*.bak
*.bak*
data/matrix_coverage_v3.json
logs/
logs*/

# Model checkpoints and training outputs
models/
runs/
*.pt
*.pth
*.ckpt

# Reports
reports/

# Local environment
.env
.wisemoor/
.DS_Store
.idea/
.vscode/
*.swp
*~

# SQLite registry cache (pushed to HF dataset repo, not source repo)
src/pino/registry.db
data/empirical_dataset_v8.jsonl
data/empirical_dataset_v8_raw.jsonl
data/empirical_dataset_v8_synth.jsonl
data/tgsc_training_formulas*.jsonl
data/openpom_curated_4983.csv
*.bak*
src/pino/registry.db*

# Restricted answer key for the blinded v2 genre challenge. Keep outside Git;
# disclose only after evaluator predictions and their hashes are frozen.
artifacts/genre_challenge_v2_release/label_key.json

# WiseMoor authenticated/raw artifacts contain account-specific download data.
data/wisemoor_cookies.txt
data/wisemoor_free_formulas.jsonl
data/wisemoor_free_formulas.summary.json
data/wisemoor_free_pdfs/
data/wisemoor_store_products.json

# PIMT v8 source-text and derived-label extracts need release/licensing review
# before adding to the source repo.
data/arctander_cas_bridge.jsonl
data/arctander_pubchem_resolution_cache_v8.json
data/literature_profiles_poucher_cleaned_v8.jsonl
data/odour_character_labels_v8*.jsonl
data/odour_taxonomy_26.json
data/per_material_signal_inspection_v8*.jsonl
data/substantivity_targets_poucher_enriched_v8_phase1.jsonl
data/substantivity_targets_poucher_tiers_v8.jsonl

# v11.6 / A/B session additions (2026-07-17)
.venv-pom/
data/aroma.db
*.log
data/empirical_dataset_v9_plus_wisemoor_clean.jsonl
data/empirical_dataset_v9_plus_wisemoor_clean.repaired.jsonl
data/pimt_candidate_20260716_a_train.jsonl
data/repaired_formulas_v11_6*.jsonl
data/tgsc_demo_formulas_v2.repaired.jsonl
data/material_profiles_v11_6.jsonl

# Large generated artifacts — live in HF data repo (pino-synthetic-dataset-v3) + local backup, not source repo
data/empirical_dataset_v*.jsonl
data/synthetic_dataset_v2*.jsonl
data/material_profiles_v11_*.jsonl
data/pubchem_material_catalog_v11_*.json
data/wisemoor_training_records_v9.jsonl
data/tgsc_demo_formulas_enriched.jsonl
data/literature_flat/literature_pages.jsonl
training.log
training_formula_disjoint_final.log