iterative_refinement_sampler
Run a deterministic SR3-inspired reverse denoising trajectory conditioned on low-resolution input for proxy recovery.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- iterative_refinement_sampler
- description
- Run a deterministic SR3-inspired reverse denoising trajectory conditioned on low-resolution input for proxy recovery.
02 资源文件 2
- 脚本scripts/sampler.py536 B
- 脚本tests/test_sampler.py261 B
03 SKILL.md 正文
Iterative Refinement Sampler
Use this skill after a denoising objective has produced a parameter estimate. It tests the SR3 generation pattern: start from noise and repeatedly refine while conditioning on the low-resolution input.
Inputs
- Initial noisy scalar state.
- Conditioning scalar and scale factor.
- Learned scalar denoiser weight.
- Number of reverse refinement steps.
Outputs
- A trajectory list with per-step states.
- Final proxy super-resolved estimate.
Workflow
- Initialize from a noisy state rather than the clean target.
- At every step, combine the current state, condition-derived target prior, and learned denoiser correction.
- Save the full trajectory for recovery analysis.
- Compare final distance to the target against the initial distance outside this skill.
Validation
Run python tests/test_sampler.py or validate the skill tree with --run-tests.
Limitations
This scalar trajectory cannot measure perceptual image quality; it checks iterative conditioning mechanics 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).