Esc
Ask AIAnswers may be inaccurate; check the linked pages.Esc
Ask anything about these docs, like how to get started or what a function does.

Scaler protocol for GRU (and future neural models) + RobustScaler.

Scaler

chronax.gru_scaler.Scaler

Per-window scaler. Implementations are pure.

stats(self, x, axis=1, mask=None)

(undocumented)

Parameter Type Default Description
x jnp.ndarray - (undocumented)
axis int 1 (undocumented)
mask jnp.ndarray | None None (undocumented)

Returns: tuple[jnp.ndarray, jnp.ndarray] (undocumented).

transform(self, x, shift, scale)

(undocumented)

Parameter Type Default Description
x jnp.ndarray - (undocumented)
shift jnp.ndarray - (undocumented)
scale jnp.ndarray - (undocumented)

Returns: jnp.ndarray (undocumented).

inverse(self, z, shift, scale)

(undocumented)

Parameter Type Default Description
z jnp.ndarray - (undocumented)
shift jnp.ndarray - (undocumented)
scale jnp.ndarray - (undocumented)

Returns: jnp.ndarray (undocumented).

RobustScaler

chronax.gru_scaler.RobustScaler

Median + MAD scaler with 0.6745*std fallback when MAD=0.

The fallback uses the Gaussian relationship MAD ≈ 0.6745·σ to estimate MAD when the empirical MAD degenerates to zero (e.g. near-constant input).

stats(self, x, axis=1, mask=None)

(undocumented)

Parameter Type Default Description
x jnp.ndarray - (undocumented)
axis int 1 (undocumented)
mask jnp.ndarray | None None (undocumented)

Returns: tuple[jnp.ndarray, jnp.ndarray] (undocumented).

transform(self, x, shift, scale)

(undocumented)

Parameter Type Default Description
x jnp.ndarray - (undocumented)
shift jnp.ndarray - (undocumented)
scale jnp.ndarray - (undocumented)

Returns: jnp.ndarray (undocumented).

inverse(self, z, shift, scale)

(undocumented)

Parameter Type Default Description
z jnp.ndarray - (undocumented)
shift jnp.ndarray - (undocumented)
scale jnp.ndarray - (undocumented)

Returns: jnp.ndarray (undocumented).