Skip to content

/AWS1/CL_BDKRFTHYPERPARAMETERS

Hyperparameters for controlling the reinforcement fine-tuning training process, including learning settings and evaluation intervals.

CONSTRUCTOR

IMPORTING

Optional arguments:

iv_epochcount TYPE /AWS1/BDKEPOCHCOUNT /AWS1/BDKEPOCHCOUNT

Number of training epochs to run during reinforcement fine-tuning. Higher values may improve performance but increase training time.

iv_batchsize TYPE /AWS1/BDKRFTBATCHSIZE /AWS1/BDKRFTBATCHSIZE

Number of training samples processed in each batch during reinforcement fine-tuning (RFT) training. Larger batches may improve training stability.

iv_learningrate TYPE /AWS1/RT_FLOAT_AS_STRING /AWS1/RT_FLOAT_AS_STRING

Learning rate for the reinforcement fine-tuning. Controls how quickly the model adapts to reward signals.

iv_maxpromptlength TYPE /AWS1/BDKRFTMAXPROMPTLENGTH /AWS1/BDKRFTMAXPROMPTLENGTH

Maximum length of input prompts during RFT training, measured in tokens. Longer prompts allow more context but increase memory usage and training-time.

iv_trainingsampleperprompt TYPE /AWS1/BDKRFTTRNSAMPLEPERPROMPT /AWS1/BDKRFTTRNSAMPLEPERPROMPT

Number of response samples generated per prompt during RFT training. More samples provide better reward signal estimation.

iv_inferencemaxtokens TYPE /AWS1/BDKRFTINFERENCEMAXTOKENS /AWS1/BDKRFTINFERENCEMAXTOKENS

Maximum number of tokens the model can generate in response to each prompt during RFT training.

iv_reasoningeffort TYPE /AWS1/BDKREASONINGEFFORT /AWS1/BDKREASONINGEFFORT

Level of reasoning effort applied during RFT training. Higher values may improve response quality but increase training time.

iv_evalinterval TYPE /AWS1/BDKRFTEVALINTERVAL /AWS1/BDKRFTEVALINTERVAL

Interval between evaluation runs during RFT training, measured in training steps. More frequent evaluation provides better monitoring.


Queryable Attributes

epochCount

Number of training epochs to run during reinforcement fine-tuning. Higher values may improve performance but increase training time.

Accessible with the following methods

Method Description
GET_EPOCHCOUNT() Getter for EPOCHCOUNT, with configurable default
ASK_EPOCHCOUNT() Getter for EPOCHCOUNT w/ exceptions if field has no value
HAS_EPOCHCOUNT() Determine if EPOCHCOUNT has a value

batchSize

Number of training samples processed in each batch during reinforcement fine-tuning (RFT) training. Larger batches may improve training stability.

Accessible with the following methods

Method Description
GET_BATCHSIZE() Getter for BATCHSIZE, with configurable default
ASK_BATCHSIZE() Getter for BATCHSIZE w/ exceptions if field has no value
HAS_BATCHSIZE() Determine if BATCHSIZE has a value

learningRate

Learning rate for the reinforcement fine-tuning. Controls how quickly the model adapts to reward signals.

Accessible with the following methods

Method Description
GET_LEARNINGRATE() Getter for LEARNINGRATE, with configurable default
ASK_LEARNINGRATE() Getter for LEARNINGRATE w/ exceptions if field has no value
STR_LEARNINGRATE() String format for LEARNINGRATE, with configurable default
HAS_LEARNINGRATE() Determine if LEARNINGRATE has a value

maxPromptLength

Maximum length of input prompts during RFT training, measured in tokens. Longer prompts allow more context but increase memory usage and training-time.

Accessible with the following methods

Method Description
GET_MAXPROMPTLENGTH() Getter for MAXPROMPTLENGTH, with configurable default
ASK_MAXPROMPTLENGTH() Getter for MAXPROMPTLENGTH w/ exceptions if field has no val
HAS_MAXPROMPTLENGTH() Determine if MAXPROMPTLENGTH has a value

trainingSamplePerPrompt

Number of response samples generated per prompt during RFT training. More samples provide better reward signal estimation.

Accessible with the following methods

Method Description
GET_TRAININGSAMPLEPERPROMPT() Getter for TRAININGSAMPLEPERPROMPT, with configurable defaul
ASK_TRAININGSAMPLEPERPROMPT() Getter for TRAININGSAMPLEPERPROMPT w/ exceptions if field ha
HAS_TRAININGSAMPLEPERPROMPT() Determine if TRAININGSAMPLEPERPROMPT has a value

inferenceMaxTokens

Maximum number of tokens the model can generate in response to each prompt during RFT training.

Accessible with the following methods

Method Description
GET_INFERENCEMAXTOKENS() Getter for INFERENCEMAXTOKENS, with configurable default
ASK_INFERENCEMAXTOKENS() Getter for INFERENCEMAXTOKENS w/ exceptions if field has no
HAS_INFERENCEMAXTOKENS() Determine if INFERENCEMAXTOKENS has a value

reasoningEffort

Level of reasoning effort applied during RFT training. Higher values may improve response quality but increase training time.

Accessible with the following methods

Method Description
GET_REASONINGEFFORT() Getter for REASONINGEFFORT, with configurable default
ASK_REASONINGEFFORT() Getter for REASONINGEFFORT w/ exceptions if field has no val
HAS_REASONINGEFFORT() Determine if REASONINGEFFORT has a value

evalInterval

Interval between evaluation runs during RFT training, measured in training steps. More frequent evaluation provides better monitoring.

Accessible with the following methods

Method Description
GET_EVALINTERVAL() Getter for EVALINTERVAL, with configurable default
ASK_EVALINTERVAL() Getter for EVALINTERVAL w/ exceptions if field has no value
HAS_EVALINTERVAL() Determine if EVALINTERVAL has a value