model-loading-inference
Load CLIP models, prepare images/text, and run safe image-text inference.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- model-loading-inference
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Load CLIP models, prepare images/text, and run safe image-text inference.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/api-reference.md5.5 KB
- 参考文档references/troubleshooting.md5.1 KB
- 参考文档references/workflows.md5.8 KB
- 脚本scripts/clip_smoke_check.py4.3 KB
- 脚本scripts/image_text_similarity.py5.8 KB
03 SKILL.md 正文
CLIP Model Loading and Inference
Use this sub-skill when an agent needs to import CLIP, choose a checkpoint, load the model/preprocess pair, prepare PIL images and tokenized text, or run image-text scoring and feature encoding.
Start Here
- For a no-download environment check, run
python scripts/clip_smoke_check.py --jsonfrom this directory or call it by path from any project. - For API details, model names, tensor shapes, Torch Hub entrypoints, and dtype/device expectations, read references/api-reference.md.
- For common loading and inference workflows, including local checkpoints and offline-safe validation, read references/workflows.md.
- For install/import, cache, checksum, JIT, CPU/GPU, preprocessing, and network failures, read references/troubleshooting.md.
- For a runnable one-image helper, use
python scripts/image_text_similarity.py --help; it only downloads weights if the user runs it with a named model that is not already cached.
Scope Boundaries
This sub-skill owns model loading, cache/download choices, preprocessing, device/JIT placement, forward inference, logits/probabilities, and encode_image / encode_text usage.
Do not use this sub-skill for prompt template search, prompt ensembling, or label wording strategy; route those decisions to ../prompt-engineering/. Do not use it for dataset-scale embedding extraction, linear probes, or evaluation loops; route those workflows to ../feature-evaluation/.
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).