NEML2 2.0.0
Loading...
Searching...
No Matches
[Settings]

Note
Clicking on the option with a triangle bullet ▸ next to it will expand/collapse its detailed information.
Type name written in PascalCase typically refer to a NEML2 object type, oftentimes a primitive tensor type.
The 🔗 symbol denotes that the option can cross-reference another object.
You can always use Ctrl+F or Cmd+F to search the entire page.

Available objects and their input file syntax

Refer to System Documentation for detailed explanation about this system.

Settings

Global settings for tensors, models, etc.

buffer_name_separator Nested buffer name separator. The default is '_'. For example, a sub-model 'foo' which declares a buffer 'bar' will have a buffer named 'foo_bar'.

  • Type: string
  • Default: _
default_device Default device on which tensors are created and models are evaluated. The string supplied must follow the following schema: (cpu|cuda)[:<device-index>] where cpu or cuda specifies the device type, and :<device-index> optionally specifies a device index. For example, device='cpu' sets the target compute device to be CPU, and device='cuda:1' sets the target compute device to be CUDA with device ID 1.

  • Type: string
  • Default: cpu
default_floating_point_type Default floating point type for tensors. Options are Float64 Float32 Float16

  • Type: EnumSelection
  • Default: Float64
default_integer_type Default integer type for tensors. Options are Int64 Int32 Int16 Int8

  • Type: EnumSelection
  • Default: Int64
interop_threads Number threads used for inter-ops parallelism. If set to 0, defaults to number of CPU cores.

  • Type: number
intraop_threads Number threads used for intra-ops parallelism. If set to 0, defaults to number of CPU cores.

  • Type: number
machine_precision Machine precision used at various places to workaround singularities like division-by-zero.

  • Type: number
  • Default: 1e-15
parameter_name_separator Parameter name separator. The default is '_'. For example, a sub-model 'foo' which declares a parameter 'bar' will have a parameter named 'foo_bar'.

  • Type: string
  • Default: _
tighter_tolerance A tighter tolerance used in various algorithms.

  • Type: number
  • Default: 1e-12
tolerance Tolerance used in various algorithms.

  • Type: number
  • Default: 1e-06

Detailed documentation link