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 means that the tensor value can be cross-reference another object. See Model parameters (revisited) for details.; 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.

additional_libraries Additional dynamic libraries to load at runtime. The Registry from these libraries are merged into the current Registry singleton. This is required for using custom models defined in dynamic libraries not directly linked to libneml2. The paths are either absolute or relative to the current working directory.

Type: list of string

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: _

disable_jit Disable JIT compilation of models. This is useful for debugging or when the JIT compiler is not available. When set to false, each individual model can still selectively enable/disable JIT. When set to true, JIT is disabled globally, and it is an error to explicitly set jit to true for any model.

Type: bool
Default: false

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: _

require_double_precision Require double precision for all computations. An error will be thrown when Model forward operators are called if the default dtype is not Float64. Set this option to false to allow other precisions.

Type: bool
Default: true

Detailed documentation link

Table of Contents

Available objects and their input file syntax

Settings