i2sb_bridge_posterior
Compute the analytic Gaussian bridge posterior q(Xt | X0, X1) for I2SB paired clean/degraded endpoints.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- i2sb_bridge_posterior
- description
- Compute the analytic Gaussian bridge posterior q(Xt | X0, X1) for I2SB paired clean/degraded endpoints.
02 资源文件 2
- 脚本scripts/bridge_posterior.py2.4 KB
- 脚本tests/test_bridge_posterior.py1.4 KB
03 SKILL.md 正文
I2SB Bridge Posterior
Use this skill when implementing or checking I2SB training data construction from paired clean and degraded samples. Do not use it for ordinary diffusion noising to Gaussian noise; the terminal endpoint is the degraded paired sample X1.
Inputs
- Clean endpoint
x0as a scalar or vector. - Degraded endpoint
x1with the same shape. - Time
tin[0, 1]. - Positive constant beta, or precomputed accumulated variances.
- Optional seed or explicit standard-normal noise for reproducible sampling.
Outputs
sigma2_t,barsigma2_t, posterior mean, posterior variance.- Optional posterior sample
xt.
Workflow
- Validate endpoint shapes and time bounds.
- Compute accumulated variances: for constant beta,
sigma2_t = beta * tandbarsigma2_t = beta * (1 - t). - Compute the posterior mean using variance weights from I2SB Equation (11).
- Compute covariance
sigma2_t * barsigma2_t / (sigma2_t + barsigma2_t). - Add seeded Gaussian noise only when a sample is requested.
Validation
Run:
python scripts/bridge_posterior.py --demo
python tests/test_bridge_posterior.py
Limitations
This skill implements the scalar/vector contract for recovery and validation. Image tensors use the same broadcasted formulas but require external array/tensor containers.
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).