| Modifier and Type | Field and Description |
|---|---|
protected TrainingDriver<? extends TrainingMessage> |
VoidParameterServer.trainer |
| Modifier and Type | Method and Description |
|---|---|
void |
VoidParameterServer.init(VoidConfiguration voidConfiguration,
Transport transport,
TrainingDriver<? extends TrainingMessage> trainer)
This method starts ParameterServer instance
PLEASE NOTE: This method is blocking for first caller only
|
void |
VoidParameterServer.setTrainingDriver(TrainingDriver<? extends TrainingMessage> trainer) |
| Modifier and Type | Field and Description |
|---|---|
protected TrainingDriver<? extends TrainingMessage> |
Frame.trainer |
protected TrainingDriver<? extends TrainingMessage> |
BaseVoidMessage.trainer |
| Modifier and Type | Method and Description |
|---|---|
void |
VoidMessage.attachContext(VoidConfiguration voidConfiguration,
TrainingDriver<? extends TrainingMessage> trainer,
Clipboard clipboard,
Transport transport,
Storage storage,
NodeRole role,
short shardIndex)
This method initializes message for further processing
|
void |
Frame.attachContext(VoidConfiguration voidConfiguration,
TrainingDriver<? extends TrainingMessage> trainer,
Clipboard clipboard,
Transport transport,
Storage storage,
NodeRole role,
short shardIndex) |
void |
BaseVoidMessage.attachContext(VoidConfiguration voidConfiguration,
TrainingDriver<? extends TrainingMessage> trainer,
Clipboard clipboard,
Transport transport,
Storage storage,
NodeRole role,
short shardIndex) |
| Modifier and Type | Class and Description |
|---|---|
class |
BaseTrainer<T extends TrainingMessage> |
| Modifier and Type | Field and Description |
|---|---|
protected Map<String,TrainingDriver<?>> |
TrainerProvider.trainers |
| Modifier and Type | Method and Description |
|---|---|
protected <T extends TrainingMessage> |
TrainerProvider.getTrainer(T message) |
| Modifier and Type | Class and Description |
|---|---|
class |
CbowTrainer |
class |
SkipGramTrainer
Distributed SkipGram trainer
TrainingDriver idea is simple:
1) We get request from Client
2) We initiate training by issuing DotRequest
3) Each Shard does Dot accumulation
4) As soon as Dot aggregated, we calculate gradients independently
5) As soon as they are ready - we just apply them to appropriate
|
Copyright © 2017. All rights reserved.