limited learning iteration number (#60)

fsharplang-ru · Oct 21, 2024 · 1b846b9 · 1b846b9
1 parent fbcbf60
commit 1b846b9
Show file tree

Hide file tree

Showing 4 changed files with 8 additions and 1 deletion.
diff --git a/.env.example b/.env.example
@@ -31,5 +31,6 @@ ML_TRAINING_SET_FRACTION=0.2
 ML_SPAM_THRESHOLD=0.5
 ML_WARNING_THRESHOLD=0.0
 ML_STOP_WORDS_IN_CHATS={"-123":["word1","word2"]}
+ML_MAX_NUMBER_OF_ITERATIONS=100
 UPDATE_CHAT_ADMINS_INTERVAL_SEC=86400
 UPDATE_CHAT_ADMINS=true
diff --git a/src/VahterBanBot/ML.fs b/src/VahterBanBot/ML.fs
@@ -89,7 +89,11 @@ type MachineLearning(
                 mlContext.Transforms.Text
                     .FeaturizeText(outputColumnName = "TextFeaturized", inputColumnName = "text")
                     .Append(mlContext.Transforms.Concatenate(outputColumnName = "Features", inputColumnNames = [|"TextFeaturized"; "lessThanNMessagesF";|]))
-                    .Append(mlContext.BinaryClassification.Trainers.LbfgsLogisticRegression(labelColumnName = "spam", featureColumnName = "Features"))
+                    .Append(mlContext.BinaryClassification.Trainers.SdcaLogisticRegression(
+                        labelColumnName = "spam",
+                        featureColumnName = "Features",
+                        maximumNumberOfIterations = botConf.MlMaxNumberOfIterations
+                    ))
 
             let trainedModel = dataProcessPipeline.Fit(trainingData)
             predictionEngine <- Some(mlContext.Model.CreatePredictionEngine<SpamOrHam, Prediction>(trainedModel))

diff --git a/src/VahterBanBot/Program.fs b/src/VahterBanBot/Program.fs
@@ -75,6 +75,7 @@ let botConf =
       MlTrainingSetFraction = getEnvOr "ML_TRAINING_SET_FRACTION" "0.2" |> float
       MlSpamThreshold = getEnvOr "ML_SPAM_THRESHOLD" "0.5" |> single
       MlWarningThreshold = getEnvOr "ML_WARNING_THRESHOLD" "0.0" |> single
+      MlMaxNumberOfIterations = getEnvOr "ML_MAX_NUMBER_OF_ITERATIONS" "100" |> int 
       MlStopWordsInChats = getEnvOr "ML_STOP_WORDS_IN_CHATS" "{}" |> fromJson }
 
 let validateApiKey (ctx : HttpContext) =

diff --git a/src/VahterBanBot/Types.fs b/src/VahterBanBot/Types.fs
@@ -41,6 +41,7 @@ type BotConfiguration =
       MlTrainingSetFraction: float
       MlSpamThreshold: single
       MlWarningThreshold: single
+      MlMaxNumberOfIterations: int
       MlStopWordsInChats: Dictionary<int64, string list> }
 
 [<CLIMutable>]