autointent.advisor#

Pre-flight compute feasibility advisor.

Experimental. This subpackage estimates VRAM, RAM, disk, and wall-time for a search space before any training starts. Estimates are heuristic and calibrated against a limited hardware sample, so treat them as guidance rather than guarantees — see the accuracy caveats in the advisor page of the docs. The public surface may change in a minor release.

Two ways in: the autointent-advisor console script, and the functions below. Pipeline.fit(preflight=...) wires the same machinery into a fit, opt-in.

Exceptions#

`PreflightError`	Raised when `Pipeline.fit(preflight="strict")` finds OVER-budget resources.
`ReduceToFitError`	Raised by `reduce_to_fit()` when no subset of the search space fits.

Classes#

`HardwareProfile`	The machine budget every estimate is scored against.
`DatasetStats`	Minimal stats the advisor needs about the user's dataset.
`Finding`	A single advisor finding rendered as one line in the summary.
`PreflightReport`	One report covering all three phases.
`RecommendationResult`	Output of the recommend workflow: ranked per-preset reports plus the pick.
`ResourceEstimate`	Aggregated resource numbers across the search space.
`Severity`	How much headroom a finding leaves against the detected budget.

Functions#

`detect_hardware`(*[, vram_budget_gb, mps_budget_ratio])	Detect the local hardware, with optional manual overrides.
`run_preflight`(config, stats, hardware, *[, ...])	Run all three preflight phases and return one report.
`dataset_stats`(dataset)	Summarize an in-memory `Dataset` for the advisor.
`estimate`(target, *[, stats, budget_vram_gb])	Estimate what a preset (or YAML config path) will cost on the local hardware.
`recommend`(*[, stats, presets, budget_vram_gb, ...])	Walk bundled presets and return the best feasible fit plus all per-preset reports.
`reduce_to_fit`(config, stats, hardware, *[, max_iters, ...])	Iteratively drop the most expensive infeasible module until the search space fits.