publication_ready remains false until the canonical full-factor expected plan/checksums and independent release evidence pass. 内置紧凑求解器可以生成完整质量报告，但通过数值验证、完整因子计划、校验和及独立发布证据前，不应作为论文级真值数据。Choose a factor slice or the complete 7 x 4 x 10 benchmark. The page creates a reproducible YAML configuration and copy-ready commands; it never sends your selections or paths to a server. 选择需要的 PDE、边界、数据规模和运行环境后，页面会自动生成配置与命令；所有选择均只在浏览器本地处理。
2. Training per observation type / 按观测类型分别训练
Primary matched-mask comparison: every normal learned baseline gets an independent checkpoint for each PDE and each of the nine observation masks. The model trained on random 3% is reused only for the separate mask-transfer/OOD table; it never replaces matched-mask training in the primary IID table.
主要 matched-mask 对比：每个普通学习基线都按 PDE 与观测掩码分别训练。random 3% 检查点只用于单独的 mask-transfer/OOD 表，不能替代主要 IID 表中的同掩码训练。
Open the complete observation-training protocol / 查看完整观测训练协议
Campaign preset / 计划预设
| Method row | Fit policy | Seeds | Observations | Preparation | Evaluations | Builder status |
|---|
3. Copy the generated code / 复制生成代码
The dataset YAML is the executable source of truth. The campaign tab is a non-executable planning record and lists every missing adapter explicitly.
Loading benchmark contract...4. Quality management for every PDE / 全 PDE 数据质量管理
Every generated sample receives finite-value, geometry, initial/boundary, and family-specific physics diagnostics. Dataset aggregation reports each selected PDE separately; the complete benchmark reports all seven losses together. 每个样本都保存数值、几何、初边值和对应 PDE 物理残差；完整基准会在同一份报告中分别汇总七类 PDE loss。
| PDE | Reported normalized loss | Interpretation |
|---|
Always produced
Gate levels
- Report: report scientific losses without applying an unfrozen PDE threshold; malformed array/geometry contracts are still quarantined.
- Strict: reject structural/BC/IC failures, log them to
*.quality-failures.jsonl, and apply a PDE limit only when calibrated. - Publication candidate: expert-only and intentionally blocked by this Builder until validated solver evidence and a per-stratum threshold table exist. It does not by itself set
publication_ready.