torch.distributions.transforms.SigmoidTransform

class SigmoidTransform extends Transform

new SigmoidTransform(options?: TransformOptions)

readonlydomain(unknown)
readonlycodomain(unknown)
readonlybijective(unknown)
readonlysign(unknown)

Sigmoid transform: y = sigmoid(x) = 1/(1 + exp(-x)) (maps ℝ → (0, 1)).

Transform via the mapping y = sigmoid(x) = 1 / (1 + exp(-x)). This transform maps real numbers to the interval (0, 1). The inverse is the logit function: x = log(y / (1 - y)).

\begin{aligned} \text{Forward: } y = \sigma(x) = \frac{1}{1 + e^{-x}} \\ \text{Inverse: } x = \text{logit}(y) = \ln\left(\frac{y}{1-y}\right) \\ \text{Log-det Jacobian: } \log\left|\frac{dy}{dx}\right| = \ln(y) + \ln(1-y) \\ \text{Derivative: } \frac{dy}{dx} = \sigma(x)(1-\sigma(x)) = y(1-y) \\ \text{Domain: } \mathbb{R}, \text{Codomain: } (0, 1) \end{aligned}

torch.distributions.transforms.SigmoidTransform

readonlydomain(unknown)

readonlycodomain(unknown)

readonlybijective(unknown)

readonlysign(unknown)

Sigmoid transform: y = sigmoid(x) = 1/(1 + exp(-x)) (maps ℝ → (0, 1)).

Transform via the mapping y = sigmoid(x) = 1 / (1 + exp(-x)). This transform maps real numbers to the interval (0, 1). The inverse is the logit function: x = log(y / (1 - y)).

\begin{aligned} \text{Forward: } y = \sigma(x) = \frac{1}{1 + e^{-x}} \\ \text{Inverse: } x = \text{logit}(y) = \ln\left(\frac{y}{1-y}\right) \\ \text{Log-det Jacobian: } \log\left|\frac{dy}{dx}\right| = \ln(y) + \ln(1-y) \\ \text{Derivative: } \frac{dy}{dx} = \sigma(x)(1-\sigma(x)) = y(1-y) \\ \text{Domain: } \mathbb{R}, \text{Codomain: } (0, 1) \end{aligned}