returns a dictionary of performance metrics: sample centric c.f. https://github.com/ashleyzhou972/CAFA_assessment_tool/blob/master/precrec/precRec

Positional embedding for 1D patches.

Positional embedding for 1D patches.

embed_dim: output dimension for each position pos: a list of positions to be encoded: size (M,) out: (M, D)

grid_size_h: int of the grid height grid_size_w: int of the grid width return: pos_embed: [grid_size_h*grid_size_w, embed_dim] or [1+

override transforms in derived classes if desired

Get the short git revision hash.

Loads the 'meta' dictionary from the memmap meta file associated with the given memmap_filename. Also attempts to load annotation meta if a c

fs: sampling frequency now required argument channel_itos: list of channel names (for mapping to integers) converts list of filenames poi

analogous to npys_to_memmap but processes batches of files before flushing them into memmap for faster processing

Compute A^L and the scan sum_i A^i v_i A: (..., N, N) v: (..., N, L)

Compute A^L and the scan sum_i A^i v_i A: (..., N, N) v: (..., N, L)

calls stratify in batches and collects results afterwards (use only for really large datasets)

BatchNorm layer with `nf` features initialized depending on `norm_type`.

BatchNorm layer with `nf` features and `ndim` initialized depending on `norm_type`.

Returns a linear nn.Module with control over axes order, initialization, and activation

Returns a linear nn.Module with control over axes order, initialization, and activation

accepts three kinds of input: 1) filenames pointing to aligned numpy arrays [timesteps,channels,...] for data and either integer labe

Return the proper conv `ndim` function, potentially `transposed`.

Norm layer with `nf` features initialized depending on `norm_type`.

Norm layer with `nf` features and `ndim` initialized depending on `norm_type`.

Masks extracted features along time axis and/or along feature axis according to [SpecAugment](https://arxiv.org/abs/1904.08779).

Create parameters that allow fast linear stepping of state

Create parameters that allow fast linear stepping of state

Must be called after self.default_state() is used to construct an initial state!

Must be called after self.default_state() is used to construct an initial state!

Convert string target to class object.

Bandpass filtering to remove noise, artifacts etc

returns butterworth filter with given specifications

Pykeops version

Pykeops version

v: (B, N) z: (L) w: (B, N) symmetric: whether to assume that v and w contain complex conjugate pairs, of the form [v_half, v_half

ramp from (e0, t0) -> (e1, t1) through a cosine schedule based on e in [e0, e1]

Model head that takes `nf` features, runs through `lin_ftrs`, and about `nc` classes; added bn and act here