setup-and-models
Guides Interactive Deep Colorization installation, backend choice, model artifact checks, and Docker or display setup before running colorization workflows.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- setup-and-models
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Guides Interactive Deep Colorization installation, backend choice, model artifact checks, and Docker or display setup before running colorization workflows.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/docker-reference.md3.0 KB
- 参考文档references/model-artifacts.md3.9 KB
- 参考文档references/setup-reference.md4.7 KB
- 参考文档references/troubleshooting.md3.3 KB
- 脚本scripts/check_model_artifacts.py5.9 KB
03 SKILL.md 正文
setup-and-models
Use this sub-skill when a task is about preparing Interactive Deep Colorization rather than applying color hints: installing legacy dependencies, choosing Caffe versus PyTorch, checking required model files, deciding whether Docker/PyQt5 is a better route, or diagnosing setup failures.
Route first
- Read references/setup-reference.md to choose a backend, dependency path, and model/runtime prerequisites.
- Read references/model-artifacts.md before downloading, validating, or explaining model weights and expected filenames.
- Read references/docker-reference.md for the repository's Docker/PyQt5 variant and display-server caveats.
- Read references/troubleshooting.md when Caffe, PyTorch, Qt, display, OpenCV, or missing-model errors block a workflow.
Safe bundled script
- Run scripts/check_model_artifacts.py to validate whether a checkout or staged artifact directory has the expected Caffe and PyTorch weight files. The script never downloads; it only checks file presence, size, and known paths.
Boundaries
- For local-hints GUI, notebook-style API, mask/
abtensor, CLI-default, suggested-color, or saved-output questions, route to ../interactive-colorization/SKILL.md. - For global histogram/reference-image transfer, route to ../global-histogram-transfer/SKILL.md.
- This sub-skill documents setup and safe validation. It does not certify that PyCaffe, PyQt GUI launch, Docker build, downloaded model inference, or network downloads work on the user's host.
Key setup decisions
- Decide backend intent:
caffeis the original SIGGRAPH 2017 backend and is required for the global histogram transfer notebook.pytorchuses converted weights for the local-hints GUI path and avoids PyCaffe, but still needs the PyTorch weight file.
- Decide UI path:
- The root GUI script is PyQt4-based.
- The Docker entry script is PyQt5-based and defaults to the PyTorch backend.
- Validate assets before launching:
- Use the bundled model checker for expected paths and missing files.
- Do not launch GUI/notebooks first when model-weight errors are likely; missing weights usually fail only after expensive imports.
- Treat training as out of scope:
- The repository README points training to a separate PyTorch reimplementation repository; this generated skill covers inference/demo usage only.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).