Class: Google::Apis::AiplatformV1beta1::GoogleCloudAiplatformV1beta1SchemaModelevaluationMetricsPairwiseTextGenerationEvaluationMetrics

Inherits:
Object
  • Object
show all
Includes:
Core::Hashable, Core::JsonObjectSupport
Defined in:
lib/google/apis/aiplatform_v1beta1/classes.rb,
lib/google/apis/aiplatform_v1beta1/representations.rb,
lib/google/apis/aiplatform_v1beta1/representations.rb

Overview

Metrics for general pairwise text generation evaluation results.

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(**args) ⇒ GoogleCloudAiplatformV1beta1SchemaModelevaluationMetricsPairwiseTextGenerationEvaluationMetrics

Returns a new instance of GoogleCloudAiplatformV1beta1SchemaModelevaluationMetricsPairwiseTextGenerationEvaluationMetrics.



25129
25130
25131
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25129

def initialize(**args)
   update!(**args)
end

Instance Attribute Details

#accuracy ⇒ Float

Fraction of cases where the autorater agreed with the human raters. Corresponds to the JSON property accuracy

Returns:

  • (Float) 


25057
25058
25059
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25057

def accuracy
  @accuracy
end

#baseline_model_win_rate ⇒ Float

Percentage of time the autorater decided the baseline model had the better response. Corresponds to the JSON property baselineModelWinRate

Returns:

  • (Float) 


25063
25064
25065
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25063

def baseline_model_win_rate
  @baseline_model_win_rate
end

#cohens_kappa ⇒ Float

A measurement of agreement between the autorater and human raters that takes the likelihood of random agreement into account. Corresponds to the JSON property cohensKappa

Returns:

  • (Float) 


25069
25070
25071
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25069

def cohens_kappa
  @cohens_kappa
end

#f1_score ⇒ Float

Harmonic mean of precision and recall. Corresponds to the JSON property f1Score

Returns:

  • (Float) 


25074
25075
25076
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25074

def f1_score
  @f1_score
end

#false_negative_count ⇒ Fixnum

Number of examples where the autorater chose the baseline model, but humans preferred the model. Corresponds to the JSON property falseNegativeCount

Returns:

  • (Fixnum) 


25080
25081
25082
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25080

def false_negative_count
  @false_negative_count
end

#false_positive_count ⇒ Fixnum

Number of examples where the autorater chose the model, but humans preferred the baseline model. Corresponds to the JSON property falsePositiveCount

Returns:

  • (Fixnum) 


25086
25087
25088
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25086

def false_positive_count
  @false_positive_count
end

#human_preference_baseline_model_win_rate ⇒ Float

Percentage of time humans decided the baseline model had the better response. Corresponds to the JSON property humanPreferenceBaselineModelWinRate

Returns:

  • (Float) 


25091
25092
25093
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25091

def human_preference_baseline_model_win_rate
  @human_preference_baseline_model_win_rate
end

#human_preference_model_win_rate ⇒ Float

Percentage of time humans decided the model had the better response. Corresponds to the JSON property humanPreferenceModelWinRate

Returns:

  • (Float) 


25096
25097
25098
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25096

def human_preference_model_win_rate
  @human_preference_model_win_rate
end

#model_win_rate ⇒ Float

Percentage of time the autorater decided the model had the better response. Corresponds to the JSON property modelWinRate

Returns:

  • (Float) 


25101
25102
25103
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25101

def model_win_rate
  @model_win_rate
end

#precision ⇒ Float

Fraction of cases where the autorater and humans thought the model had a better response out of all cases where the autorater thought the model had a better response. True positive divided by all positive. Corresponds to the JSON property precision

Returns:

  • (Float) 


25108
25109
25110
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25108

def precision
  @precision
end

#recall ⇒ Float

Fraction of cases where the autorater and humans thought the model had a better response out of all cases where the humans thought the model had a better response. Corresponds to the JSON property recall

Returns:

  • (Float) 


25115
25116
25117
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25115

def recall
  @recall
end

#true_negative_count ⇒ Fixnum

Number of examples where both the autorater and humans decided that the model had the worse response. Corresponds to the JSON property trueNegativeCount

Returns:

  • (Fixnum) 


25121
25122
25123
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25121

def true_negative_count
  @true_negative_count
end

#true_positive_count ⇒ Fixnum

Number of examples where both the autorater and humans decided that the model had the better response. Corresponds to the JSON property truePositiveCount

Returns:

  • (Fixnum) 


25127
25128
25129
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25127

def true_positive_count
  @true_positive_count
end

Instance Method Details

#update!(**args) ⇒ Object

Update properties of this object



25134
25135
25136
25137
25138
25139
25140
25141
25142
25143
25144
25145
25146
25147
25148
 
# File 'lib/google/apis/aiplatform_v1beta1/classes.rb', line 25134

def update!(**args)
  @accuracy = args[:accuracy] if args.key?(:accuracy)
  @baseline_model_win_rate = args[:baseline_model_win_rate] if args.key?(:baseline_model_win_rate)
  @cohens_kappa = args[:cohens_kappa] if args.key?(:cohens_kappa)
  @f1_score = args[:f1_score] if args.key?(:f1_score)
  @false_negative_count = args[:false_negative_count] if args.key?(:false_negative_count)
  @false_positive_count = args[:false_positive_count] if args.key?(:false_positive_count)
  @human_preference_baseline_model_win_rate = args[:human_preference_baseline_model_win_rate] if args.key?(:human_preference_baseline_model_win_rate)
  @human_preference_model_win_rate = args[:human_preference_model_win_rate] if args.key?(:human_preference_model_win_rate)
  @model_win_rate = args[:model_win_rate] if args.key?(:model_win_rate)
  @precision = args[:precision] if args.key?(:precision)
  @recall = args[:recall] if args.key?(:recall)
  @true_negative_count = args[:true_negative_count] if args.key?(:true_negative_count)
  @true_positive_count = args[:true_positive_count] if args.key?(:true_positive_count)
end