.gitignore

*.txt

################################################################################
# Universal Python / Data / ML / Web / Frontend / Docs .gitignore (Extremely Comprehensive)
# Start by keeping this whole file; prune later once your project stabilizes.
# Use .env.example or config.example.* for tracked templates.
################################################################################

########################################
# User-requested explicit patterns
########################################
*.zip
*.tar
*.tar.gz
*.tgz
*.pth
*.ckpt
*.mp4
*.bin
*.pdf

########################################
# Python bytecode / interpreter artefacts
########################################
__pycache__/
*.py[cod]
*$py.class

########################################
# Compiled / extension modules
########################################
*.so
*.pyd
*.dll
*.dylib
*.a
*.o
*.obj
*.out
*.exe

########################################
# Build / packaging
########################################
build/
dist/
sdist/
wheels/
wheelhouse/
*.egg-info/
*.egg
.eggs/
pip-wheel-metadata/
MANIFEST
*.spec          # PyInstaller spec files (regenerate as needed)
*.manifest

########################################
# Virtual environments / dependency managers
########################################
.venv/
venv/
env/
ENV/
.conda/
condaenv.*
*.conda
# Pipenv / Poetry / PDM / Hatch / Rye / UV
Pipfile.lock
poetry.lock
.poetry/
.pdm-build/
.pdm-python
pdm.lock
.hatch/
.hatch_build/
.rye/
.uv/
# pyenv / direnv
.python-version
.envrc
# PEP 582
__pypackages__/

########################################
# Environment / secrets (track only templates)
########################################
.env
.env.*
*.env
*.secret
secrets/
secret/
*.key
*.pem
*.crt
*.cer
*.der
*.p12
*.pfx
*.csr
*.enc

########################################
# Logs / runtime / temp
########################################
*.log
log/
logs/
eval_log/
*.pid
*.pid.lock
*.seed
*.tmp
*.temp
*.bak
*.orig
*.swp
*.swo
*.lock
*.DS_Store?

########################################
# Testing / coverage / lint / type-check
########################################
.pytest_cache/
.cache/
.coverage
.coverage.*
coverage.xml
htmlcov/
.tox/
.nox/
.mypy_cache/
.dmypy.json
dmypy.json
.pyre/
.pytype/
.pyright/
ruff_cache/
.ruff_cache/
.pylint.d/
.bandit
bandit.json
.hypothesis/
.prof
*.prof
*.lprof
*.cprof
*.profraw

########################################
# Jupyter / IPython
########################################
.ipynb_checkpoints/
*.nbconvert.ipynb
.ipython/
.jupyter/
.python_history
.history

########################################
# Data science / ML / training artefacts
########################################
# Common working dirs
data/
datasets/
raw_data/
processed_data/
cache/
cache_dir/
cached/
artifacts/
artifacts_tmp/
workspace/
workdir/
outputs/
output/
output_*/
results/
runs/
experiments/
tensorboard/
tb_logs/
lightning_logs/
wandb/
mlruns/
ray_results/
comet_ml/
neptune/
optuna/
checkpoints/

# Common data file formats (uncomment any you DO want to track)
*.csv
*.tsv
*.parquet
*.feather
*.arrow
*.orc
*.h5
*.hdf5
*.npz
*.npy
*.pkl
*.pickle
*.joblib
*.mat
*.rds
*.jsonl
*.msgpack

# Model / weight formats
*.pt
*.pth
*.ckpt
*.weights
*.safetensors
*.onnx
*.tflite
*.pb
*.mlmodel
*.bin
*.model

# Tokenizer & vocab
tokenizer.model
sentencepiece.bpe.model
merges.txt
vocab.json

# Hugging Face / Transformers caches
.huggingface/
huggingface/
huggingface_cache/
hf_cache/
transformers_cache/
.cache/huggingface/
cache/huggingface/

# Diffusion / SD style models
diffusers_cache/
sd_models/
loras/
vae/
embeddings/

########################################
# Media / large asset files (adjust as needed)
########################################
media/
assets/
static/
staticfiles/
public/
site_media/
uploads/
*.png
*.jpg
*.jpeg
*.gif
*.svg
*.bmp
*.webp
*.tiff
*.ico
*.mp4
*.mov
*.avi
*.mkv
*.mpg
*.mpeg
*.flv
*.wav
*.mp3
*.ogg

########################################
# Documentation build outputs
########################################
docs/_build/
_build/
build_docs/
site/
mkdocs/site/
mkdocs_build/
.book/
.docusaurus/
.next/
.out/
.vitepress/dist/
.storybook/
storybook-static/
.storybook-static/
coverage-docs/

########################################
# Web frameworks / backends
########################################
*.pot
local_settings.py
db.sqlite3
db.sqlite3-journal
instance/
*.db
*.sqlite
*.sqlite3

########################################
# Frontend / Node ecosystem (if present)
########################################
node_modules/
bower_components/
*.tsbuildinfo
npm-debug.log*
yarn-debug.log*
yarn-error.log*
pnpm-debug.log*
.pnpm-store/
dist-ssr/
.eslintcache
.stylelintcache

########################################
# Archives / compressed (beyond those explicitly listed)
########################################
*.tar
*.tar.gz
*.tgz
*.gz
*.bz2
*.xz
*.7z
*.rar

########################################
# Containers / Docker
########################################
Dockerfile.*.tmp
.docker/
.docker_tmp/
docker-compose.override.yml

########################################
# Sphinx / LaTeX / doc tool artefacts
########################################
*.aux
*.toc
*.idx
*.lof
*.lot
*.log
*.fls
*.out
*.fdb_latexmk
*.synctex.gz
*.bbl
*.blg
*.nav
*.snm
*.vrb

########################################
# Generated / codegen / API clients
########################################
generated/
gen/
autogen/
.codegen/
.openapi/
openapi_generated/
swagger_generated/
.huggingface_cache/

########################################
# Distributed / cluster training logs
########################################
slurm-*.out
*.event*
events.out.tfevents.*
distributed_logs/
cluster_logs/
parallel_logs/

########################################
# Patches / backups
########################################
*.bak
*.backup
*.old
*.orig
*.rej
*.diff
patches/
*.sync-conflict-*

########################################
# OS / filesystem noise
########################################
.DS_Store
.AppleDouble
.LSOverride
Icon?
._*
.Spotlight-V100
.Trashes
Thumbs.db
ehthumbs.db
Desktop.ini
$RECYCLE.BIN/
.fseventsd
.turd
.nfs*
.metadata/

########################################
# Editors / IDE
########################################
# VS Code
.vscode/
.vscode-test/
# JetBrains
.idea/
*.iml
*.ipr
*.iws
# Sublime
*.sublime-project
*.sublime-workspace
# Eclipse / NetBeans
.project
.classpath
.cproject
.settings/
.nb-gradle/
# Vim
*.swp
*.swo
.session
# Emacs
*~
\#*\#
.\#*
.tramp
# Misc
.lldb
*.tmproj
*.tmproject
*.code-workspace

########################################
# Merge / conflict artefacts
########################################
*.BACKUP.*
*.BASE.*
*.LOCAL.*
*.REMOTE.*
*.orig
*.rej

########################################
# Misc junk / transient
########################################
*.tmp
*.temp
tmp/
temp/
~$*
*.part
*.crdownload
*.download
*.partial
*.cache
.cache/
cache/

########################################
# Helpful reminders (NOT enforced)
# Keep these tracked (do NOT ignore):
#   requirements.txt / requirements/*.txt
#   pyproject.toml / setup.cfg / setup.py
#   README.md / CHANGELOG.md / LICENSE
#   .env.example / config.example.yaml
#   docs/ (except build outputs)
########################################

################################################################################
# To keep an otherwise ignored file or empty directory:
#   !path/to/file.ext
#   Place a .gitkeep (or .keep) inside the directory and negate it:
#   !data/.gitkeep
################################################################################

# Examples (uncomment if you decide to preserve folder structure):
# !data/.gitkeep
# !models/.gitkeep
# !outputs/.gitkeep
# !checkpoints/.gitkeep

list_*
tmp
*.jsonmodel_factory/vit_aim_v2_packing_hf_old.py


ckpts/
ckpts/**
ckpts
.gitginore

_codeql*

# ===========================================
# Allow example training data demo files
# ===========================================
!llava_next/examples/training_data_demo/output/
!llava_next/examples/training_data_demo/output/**
!llava_next/examples/training_data_demo/videos/
!llava_next/examples/training_data_demo/videos/*.mp4