Reference¶
panndas.nn¶
- class panndas.nn.AdditiveSkip(block)¶
A Module that applies an additive “skip” connection around the provided Module.
- forward(xs)¶
Minimally, define this method to define a Module.
- show()¶
Displays the Module.
- class panndas.nn.AlphaDropout(p, alpha=0.0)¶
A Module that multiplies its inputs by the weights_df and adds the bias_series.
Input ‘tensors’ can be at most 2-D here: feature (rows) and batch/sequence (columns).
The weights dataframe should have the input feature space as its column index and the output feature space as its row index.
- forward(xs)¶
Minimally, define this method to define a Module.
- show()¶
Displays the Module.
- class panndas.nn.Dropout(p)¶
- class panndas.nn.Identity¶
A Module that returns its inputs unaltered.
- forward(xs)¶
Minimally, define this method to define a Module.
- class panndas.nn.LayerMaxNorm¶
Normalize across the feature dimension with respect to the infinity norm.
- forward(xs)¶
Minimally, define this method to define a Module.
- class panndas.nn.Linear(weights_df, bias_series=- 1)¶
A Module that multiplies its inputs by the weights_df and adds the bias_series.
Input ‘tensors’ can be at most 2-D here: feature (rows) and batch/sequence (columns).
The weights dataframe should have the input feature space as its column index and the output feature space as its row index.
- forward(xs)¶
Minimally, define this method to define a Module.
- show()¶
Displays the Module.
- class panndas.nn.LinearAttention(queries_df, keys_df, values_df)¶
The most basic version of an attention layer.
- forward(xs)¶
Combines queries, keys, and values linearly.
- class panndas.nn.Mish¶
Applies the Mish function, element-wise.
For details, see Mish: A Self-Regularized Non-Monotonic Neural Activation Function.
- forward(xs)¶
Applies the Mish function, element-wise.
- class panndas.nn.Module¶
An object that is callable via its .forward method.
- abstract forward(xs)¶
Minimally, define this method to define a Module.
- show()¶
Displays the Module.
- class panndas.nn.ReLU¶
Ol’ ReLU-iable.
- forward(xs)¶
Minimally, define this method to define a Module.
- class panndas.nn.Sequential(modules)¶
A Module that applies an iterable of Modules sequentially.
- forward(xs)¶
Minimally, define this method to define a Module.
- show()¶
Displays the Module.
- class panndas.nn.Sigmoid¶
Applies the sigmoid function, element-wise.
- forward(xs)¶
Applies the sigmoid function, element-wise.
- class panndas.nn.Softmax¶
Applies softmax function, column-wise.
- forward(xs)¶
Applies softmax function, column-wise.