eazygrad.functions

Activations

relu	Apply the rectified linear unit activation elementwise.
sigmoid	Apply the sigmoid activation elementwise.
tanh	Apply the hyperbolic tangent activation elementwise.

Math

min	Return the elementwise minimum of two tensors.
exp	Compute the elementwise exponential.
log	Compute the elementwise natural logarithm.
cos	Compute the elementwise cosine.
sin	Compute the elementwise sine.
clip	Clip tensor values to a closed interval.

Special Functions

logsumexp	Compute log(sum(exp(input))) in a numerically stable way.
softmax	Compute the softmax of a tensor along a given axis.
log_softmax	Compute the logarithm of the softmax along a given axis.

Losses

mse_loss	Compute the mean squared error loss.
nll_loss	Compute the negative log-likelihood loss from log-probabilities.
bce_with_logits_loss	Compute binary cross-entropy loss from unnormalized logits.
bce_loss	Compute binary cross-entropy loss from probabilities.
cross_entropy_loss	Compute cross-entropy loss from class logits.