proxy_training_recovery
Run bounded proxy PINN training comparisons that exercise residual construction, NTK weighting, and optimizer updates.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- proxy_training_recovery
- description
- Run bounded proxy PINN training comparisons that exercise residual construction, NTK weighting, and optimizer updates.
02 资源文件 3
- 脚本scripts/proxy_train.py3.4 KB
- 脚本tests/test_learning_rate_stress.py248 B
- 脚本tests/test_proxy_training.py406 B
03 SKILL.md 正文
Proxy Training Recovery
Use this skill to produce executable recovery evidence for the NTK-PINN paper when full large-scale experiments are blocked or out of scope under soft recovery mode. The skill compares equal weighting with adaptive NTK weighting on a deterministic Poisson-style proxy.
Inputs
- Problem item from
pinn_residual_problem. - NTK diagnostics from
ntk_block_spectrum. - Adaptive weights from
adaptive_ntk_weighting. - Initial scalar parameters and optimizer settings.
Outputs
- Training trace with
params_before,params_after,loss_before, andloss_after. - Numeric comparison metric
adaptive_loss_ratio_improvement. - Mechanism checks showing residual loss, boundary loss, NTK traces, adaptive weights, and optimizer updates all ran.
Workflow
- Use the problem item to define separated boundary and residual losses.
- Use deterministic Jacobians to compute NTK block traces and Algorithm 1 weights.
- Run equal-weight and adaptive-weight gradient steps from the same initial parameters.
- Measure how close the component-loss ratio is to one after training.
- Save logs that downstream Distiller validators can inspect.
Validation
Run:
python tests/test_proxy_training.py
The tests assert parameter updates, numeric losses, and improved adaptive component balance.
Limitations
This is a soft-mode proxy and must be declared as such. It is not a full reproduction of all paper figures, but it executes the core mechanism in a bounded way.
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).