Skip to content

data

data

Run data prepared for the interactive interface.

TimelineData(first: int, last: int, steps: tuple[int, ...], bins: tuple[int, ...]) dataclass

A fixed-resolution summary of one run's steps.

RunKey(name: str, nonce: str) dataclass

Identify one stored run independently of its sequence.

PlotSeries(name: str, run: RunKey, points: tuple[tuple[float, float], ...], sequences: tuple[tuple[float, int], ...], checkpoints: tuple[tuple[float, float], ...] = (), checkpoint_sequences: tuple[int, ...] = ()) dataclass

Name one scalar series rendered by a plot.

RunSnapshot(rows: tuple[tuple[Node, ValueRow], ...], checkpoints: frozenset[int], executions: tuple[str, ...] = ()) dataclass

Values and checkpoints read for one run at a single instant.

DataFrameSource

Bases: Protocol

Provide complete tabular data for external exploration.

to_dataframe() -> DataFrame

Return every source row in a tabular representation.

PlotData

Bases: DataFrameSource, Protocol

Provide the fields and series consumed by saved plots.

plot_series(x_key: str, y_key: str, limit: int) -> tuple[PlotSeries, ...]

Return bounded named series for the selected fields.

RunData(store: ObjectReader, name: str, nonce: str)

Materialize one run once and expose UI-shaped projections of it.

comparison: bool property

Return whether plots contain multiple runs.

scalar_keys: tuple[str, ...] property

Return numeric user fields available anywhere in the run.

read(store: ObjectReader) -> RunSnapshot

Read a coherent replacement snapshot without mutating this run.

apply(snapshot: RunSnapshot) -> bool

Apply a snapshot and invalidate cached scalar projections if changed.

timeline(width: int) -> TimelineData

Collapse run events into at most width timeline bins.

nearest(seq: int) -> int

Return the recorded step nearest to seq.

step(seq: int) -> tuple[Node, ValueRow]

Return the identity and folded fields for an exact recorded step.

series(x_key: str, y_key: str, limit: int) -> tuple[tuple[float, float], ...] cached

Return a bounded scalar series, preserving its final point.

plot_series(x_key: str, y_key: str, limit: int) -> tuple[PlotSeries, ...]

Return this run as one named plot series.

to_dataframe() -> DataFrame

Return every run log row with its node identity and checkpoint state.

WorkspaceData(runs: tuple[RunData, ...])

Expose comparable scalar series across a marked set of runs.

comparison: bool property

Return whether plots belong to a cross-cutting workspace.

scalar_keys: tuple[str, ...] property

Return numeric fields available in every currently marked run.

plot_series(x_key: str, y_key: str, limit: int) -> tuple[PlotSeries, ...]

Return one line for each marked run containing both fields.

to_dataframe() -> DataFrame

Return every log row from every run in this workspace.