Class: Aws::SageMakerTrainingSessionRuntime::Types::AdamParamsConfig

Inherits:
Struct
  • Object
show all
Defined in:
gems/aws-sdk-sagemakertrainingsessionruntime/lib/aws-sdk-sagemakertrainingsessionruntime/types.rb

Overview

The Adam optimizer hyperparameters used during an optimizer step.

See Also:

Constant Summary collapse

SENSITIVE = 
[]

Instance Attribute Summary collapse

Instance Attribute Details

#beta1 ⇒ Float

The exponential decay rate for the first moment estimates.

Returns:

  • (Float) 


48
49
50
51
52
53
54
55
56
57
58
 
# File 'gems/aws-sdk-sagemakertrainingsessionruntime/lib/aws-sdk-sagemakertrainingsessionruntime/types.rb', line 48

class AdamParamsConfig < Struct.new(
  :learning_rate,
  :beta1,
  :beta2,
  :eps,
  :weight_decay,
  :grad_clip_norm,
  :lora_plus_lr_ratio)
  SENSITIVE = []
  include Aws::Structure
end

#beta2 ⇒ Float

The exponential decay rate for the second moment estimates.

Returns:

  • (Float) 


48
49
50
51
52
53
54
55
56
57
58
 
# File 'gems/aws-sdk-sagemakertrainingsessionruntime/lib/aws-sdk-sagemakertrainingsessionruntime/types.rb', line 48

class AdamParamsConfig < Struct.new(
  :learning_rate,
  :beta1,
  :beta2,
  :eps,
  :weight_decay,
  :grad_clip_norm,
  :lora_plus_lr_ratio)
  SENSITIVE = []
  include Aws::Structure
end

#eps ⇒ Float

A small constant added to the denominator for numerical stability.

Returns:

  • (Float) 


48
49
50
51
52
53
54
55
56
57
58
 
# File 'gems/aws-sdk-sagemakertrainingsessionruntime/lib/aws-sdk-sagemakertrainingsessionruntime/types.rb', line 48

class AdamParamsConfig < Struct.new(
  :learning_rate,
  :beta1,
  :beta2,
  :eps,
  :weight_decay,
  :grad_clip_norm,
  :lora_plus_lr_ratio)
  SENSITIVE = []
  include Aws::Structure
end

#grad_clip_norm ⇒ Float

The maximum gradient norm for gradient clipping. Gradients are scaled down if their norm exceeds this value.

Returns:

  • (Float) 


48
49
50
51
52
53
54
55
56
57
58
 
# File 'gems/aws-sdk-sagemakertrainingsessionruntime/lib/aws-sdk-sagemakertrainingsessionruntime/types.rb', line 48

class AdamParamsConfig < Struct.new(
  :learning_rate,
  :beta1,
  :beta2,
  :eps,
  :weight_decay,
  :grad_clip_norm,
  :lora_plus_lr_ratio)
  SENSITIVE = []
  include Aws::Structure
end

#learning_rate ⇒ Float

The learning rate for the optimizer. Must be greater than 0.

Returns:

  • (Float) 


48
49
50
51
52
53
54
55
56
57
58
 
# File 'gems/aws-sdk-sagemakertrainingsessionruntime/lib/aws-sdk-sagemakertrainingsessionruntime/types.rb', line 48

class AdamParamsConfig < Struct.new(
  :learning_rate,
  :beta1,
  :beta2,
  :eps,
  :weight_decay,
  :grad_clip_norm,
  :lora_plus_lr_ratio)
  SENSITIVE = []
  include Aws::Structure
end

#lora_plus_lr_ratio ⇒ Float

The learning rate multiplier for LoRA+ B matrices relative to A matrices. When set, B matrices use learningRate * loraPlusLrRatio while A matrices use the base learning rate.

Returns:

  • (Float) 


48
49
50
51
52
53
54
55
56
57
58
 
# File 'gems/aws-sdk-sagemakertrainingsessionruntime/lib/aws-sdk-sagemakertrainingsessionruntime/types.rb', line 48

class AdamParamsConfig < Struct.new(
  :learning_rate,
  :beta1,
  :beta2,
  :eps,
  :weight_decay,
  :grad_clip_norm,
  :lora_plus_lr_ratio)
  SENSITIVE = []
  include Aws::Structure
end

#weight_decay ⇒ Float

The weight decay (L2 regularization) coefficient.

Returns:

  • (Float) 


48
49
50
51
52
53
54
55
56
57
58
 
# File 'gems/aws-sdk-sagemakertrainingsessionruntime/lib/aws-sdk-sagemakertrainingsessionruntime/types.rb', line 48

class AdamParamsConfig < Struct.new(
  :learning_rate,
  :beta1,
  :beta2,
  :eps,
  :weight_decay,
  :grad_clip_norm,
  :lora_plus_lr_ratio)
  SENSITIVE = []
  include Aws::Structure
end