nn.Functional

• cofhe.nn.functional.relu(input) Applies the ReLU activation function to the input tensor, replacing negative values with zero.

• cofhe.nn.functional.softmax(input) Applies the softmax activation function across the input tensor, typically used for classification tasks.

• cofhe.nn.functional.log_softmax(input) Applies the log softmax activation function to the input tensor. Useful when working with log probabilities.

• cofhe.nn.functional.leaky_relu(input, negative_slope) Applies the Leaky ReLU activation function, where negative values are scaled by a negative_slope factor.

• cofhe.nn.functional.tanh(input) Applies the hyperbolic tangent (Tanh) activation function to the input tensor.

• cofhe.nn.functional.sigmoid(input) Applies the sigmoid activation function to the input tensor, mapping the values to the range [0, 1].

• cofhe.nn.functional.gelu(input) Applies the Gaussian Error Linear Unit (GELU) activation function, a smooth version of ReLU, widely used in transformer models.

• cofhe.nn.functional.mse_loss(input, target) Computes the Mean Squared Error (MSE) loss between the input tensor and the target tensor.

• cofhe.nn.functional.cross_entropy(input, target) Computes the Cross-Entropy loss between the input tensor (logits) and the target tensor (class labels).