sbi_posterior_api
Provide observation-conditioned posterior sampling and log-probability APIs for SBI recovery estimators.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- sbi_posterior_api
- description
- Provide observation-conditioned posterior sampling and log-probability APIs for SBI recovery estimators.
02 资源文件 2
- 脚本scripts/posterior_api.py3.4 KB
- 脚本tests/test_posterior_api.py571 B
03 SKILL.md 正文
SBI Posterior API
Use this skill after an SBI estimator has been trained and a task needs to expose distribution-like posterior behavior. It represents the paper's NeuralPosterior abstraction: a common object conditioned on observations that can produce parameter samples and, when supported, density values.
Do not use this skill to run simulators or train estimator parameters. It consumes an estimator and validates posterior behavior.
Inputs
- A trained full or reduced estimator.
- Observation
x_ofor conditioning. - Number of posterior samples and optional random seed.
- Optional candidate
thetavalues for log-probability evaluation.
Outputs
- Posterior summary statistics such as mean and standard deviation.
- Posterior samples conditioned on the observation.
- Optional log-probability values for supported estimators.
Workflow
- Load or receive an estimator with a documented posterior parameterization.
- Condition it on an observation.
- Produce deterministic samples when a seed is supplied.
- Evaluate log probability only when the estimator supports it.
- Save posterior summaries for diagnostics and recovery analysis.
Validation
Run:
python /share/project/yuyang/workspace/Paper2Skills/Distiller/skills/module-to-skill/scripts/validate_skill_tree.py /share/project/yuyang/workspace/Paperbench/record/case15/extracted_skills_attempt_001/sbi_toolkit/sbi_posterior_api --run-tests
For a standalone smoke run:
python scripts/posterior_api.py --demo
Limitations
The bundled script implements a scalar Gaussian reduced posterior. Full sbi posteriors can use richer samplers and conditional density estimators, but recovery code should keep the same contract: condition, sample, summarize, and evaluate density only when valid.
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).