30 lines
784 B
Markdown
30 lines
784 B
Markdown
# BENCHMARKS
|
|
|
|
|
|
- Use `criterion` for microbenchmarks (eval throughput).
|
|
- Dataset-level timing: fit to Friedman1 and report wall-clock and MSE.
|
|
- Compare feature toggles: scalar vs SIMD, parallel vs single-thread.
|
|
|
|
|
|
|
|
---
|
|
|
|
|
|
## 8) Branching model (with a `porting` branch)
|
|
|
|
|
|
- `main`: stable, tagged releases.
|
|
- `develop`: integration of ready features.
|
|
- `porting`: **all work specific to parity with SymbolicRegression.jl** (feature-by-feature ports, benchmarks, notes). Merge into `develop` via PRs when each slice reaches MVP.
|
|
- Per-feature branches from `porting` (e.g., `porting/const-fitting`, `porting/egg-rewrites`).
|
|
|
|
|
|
**Initial commands:**
|
|
```bash
|
|
git init
|
|
git checkout -b main
|
|
git add . && git commit -m "feat: repo scaffold"
|
|
git checkout -b develop
|
|
git checkout -b porting
|
|
|