Utilities¤

Index¤

Device Management / Data Transfer¤

# Lux.cpu — Function.

cpu(x)

Transfer x to CPU.

Warning

This function has been deprecated. Use cpu_device instead.

source

# Lux.gpu — Function.

gpu(x)

Transfer x to GPU determined by the backend set using Lux.gpu_backend!.

Warning

This function has been deprecated. Use gpu_device instead. Using this function inside performance critical code will cause massive slowdowns due to type inference failure.

source

Note

For detailed API documentation on Data Transfer check out the LuxDeviceUtils.jl

Initialization¤

Note

For API documentation on Initialization check out the WeightInitializers.jl

Miscellaneous Utilities¤

# Lux.foldl_init — Function.

foldl_init(op, x)
foldl_init(op, x, init)

Exactly same as foldl(op, x; init) in the forward pass. But, gives gradients wrt init in the backward pass.

source

# Lux.istraining — Function.

istraining(::Val{training})
istraining(st::NamedTuple)

Returns true if training is true or if st contains a training field with value true. Else returns false.

Method undefined if st.training is not of type Val.

source

# Lux.multigate — Function.

multigate(x::AbstractArray, ::Val{N})

Split up x into N equally sized chunks (along dimension 1).

source

# Lux.replicate — Function.

replicate(rng::AbstractRNG)
replicate(rng::CUDA.RNG)

Creates a copy of the rng state depending on its type.

source

Truncated Stacktraces¤

# Lux.disable_stacktrace_truncation! — Function.

disable_stacktrace_truncation!(; disable::Bool=true)

An easy way to update TruncatedStacktraces.VERBOSE without having to load it manually.

Effectively does TruncatedStacktraces.VERBOSE[] = disable

source