/AWS1/CL_SGIOPERATIONCONFIG¶
The configuration for a training operation. Exactly one member must be set to specify the operation type and its parameters.
CONSTRUCTOR¶
IMPORTING¶
Optional arguments:¶
io_forward TYPE REF TO /AWS1/CL_SGIFORWARDCONFIG /AWS1/CL_SGIFORWARDCONFIG¶
The configuration for a forward-only pass through the model without gradient computation.
io_forwardbackward TYPE REF TO /AWS1/CL_SGIFORWARDBACKWARDCFG /AWS1/CL_SGIFORWARDBACKWARDCFG¶
The configuration for a forward and backward pass that computes gradients.
io_optimizerstep TYPE REF TO /AWS1/CL_SGIOPTSTEPCONFIG /AWS1/CL_SGIOPTSTEPCONFIG¶
The configuration for an optimizer step that updates model weights using accumulated gradients.
io_saveweights TYPE REF TO /AWS1/CL_SGISAVEWEIGHTSCONFIG /AWS1/CL_SGISAVEWEIGHTSCONFIG¶
The configuration for saving model weights as a checkpoint.
io_savestate TYPE REF TO /AWS1/CL_SGISAVESTATECONFIG /AWS1/CL_SGISAVESTATECONFIG¶
The configuration for saving full training state, including model weights and optimizer state.
io_loadstate TYPE REF TO /AWS1/CL_SGILOADSTATECONFIG /AWS1/CL_SGILOADSTATECONFIG¶
The configuration for loading training state from a previously saved checkpoint.
Queryable Attributes¶
forward¶
The configuration for a forward-only pass through the model without gradient computation.
Accessible with the following methods¶
| Method | Description |
|---|---|
GET_FORWARD() |
Getter for FORWARD |
forwardBackward¶
The configuration for a forward and backward pass that computes gradients.
Accessible with the following methods¶
| Method | Description |
|---|---|
GET_FORWARDBACKWARD() |
Getter for FORWARDBACKWARD |
optimizerStep¶
The configuration for an optimizer step that updates model weights using accumulated gradients.
Accessible with the following methods¶
| Method | Description |
|---|---|
GET_OPTIMIZERSTEP() |
Getter for OPTIMIZERSTEP |
saveWeights¶
The configuration for saving model weights as a checkpoint.
Accessible with the following methods¶
| Method | Description |
|---|---|
GET_SAVEWEIGHTS() |
Getter for SAVEWEIGHTS |
saveState¶
The configuration for saving full training state, including model weights and optimizer state.
Accessible with the following methods¶
| Method | Description |
|---|---|
GET_SAVESTATE() |
Getter for SAVESTATE |
loadState¶
The configuration for loading training state from a previously saved checkpoint.
Accessible with the following methods¶
| Method | Description |
|---|---|
GET_LOADSTATE() |
Getter for LOADSTATE |