chnages to single feature tree building that should not affect outputs

interpretml · Jan 5, 2025 · f9b42c2 · f9b42c2
1 parent 4480350
commit f9b42c2
Show file tree

Hide file tree

Showing 5 changed files with 103 additions and 94 deletions.
diff --git a/shared/libebm/BoosterCore.cpp b/shared/libebm/BoosterCore.cpp
@@ -823,20 +823,26 @@ ErrorEbm BoosterCore::Create(void* const rng,
          }
 
          const size_t cBytesPerMainBin = GetBinSize<FloatMain, UIntMain>(true, true, bHessian, cScores);
-
-         if(IsAddError(cBytesPerMainBin, sizeof(void*))) {
-            LOG_0(Trace_Warning, "WARNING BoosterCore::Create IsAddError(cBytesPerMainBin, sizeof(void*))");
-            return Error_OutOfMemory;
-         }
-         const size_t cBytesMainBinPlusPointer = cBytesPerMainBin + sizeof(void*);
-         if(IsMultiplyError(cBytesMainBinPlusPointer, cMainBinsMax)) {
+         if(IsMultiplyError(cBytesPerMainBin, cMainBinsMax)) {
             LOG_0(Trace_Warning, "WARNING BoosterCore::Create IsMultiplyError(cBytesPerMainBin, cMainBinsMax)");
             return Error_OutOfMemory;
          }
-         // we also allocate enough space to create an additional array of pointers
-         pBoosterCore->m_cBytesMainBins = cBytesMainBinPlusPointer * cMainBinsMax;
+         size_t cBytesMainBins = cBytesPerMainBin * cMainBinsMax;
 
          if(0 != cSingleDimensionBinsMax) {
+            if(IsAddError(cBytesPerMainBin, sizeof(void*))) {
+               LOG_0(Trace_Warning, "WARNING BoosterCore::Create IsAddError(cBytesPerMainBin, sizeof(void*))");
+               return Error_OutOfMemory;
+            }
+            const size_t cBytesMainBinPlusPointer = cBytesPerMainBin + sizeof(void*);
+            if(IsMultiplyError(cBytesMainBinPlusPointer, cSingleDimensionBinsMax)) {
+               LOG_0(Trace_Warning,
+                     "WARNING BoosterCore::Create IsMultiplyError(cBytesMainBinPlusPointer, cSingleDimensionBinsMax)");
+               return Error_OutOfMemory;
+            }
+            // we also allocate enough space to create an additional array of pointers
+            cBytesMainBins = EbmMax(cBytesMainBins, cBytesMainBinPlusPointer * cSingleDimensionBinsMax);
+
             if(IsOverflowTreeNodeSize(bHessian, cScores) || IsOverflowSplitPositionSize(bHessian, cScores)) {
                LOG_0(Trace_Warning, "WARNING BoosterCore::Create bin tracking size overflow");
                return Error_OutOfMemory;
@@ -881,6 +887,7 @@ ErrorEbm BoosterCore::Create(void* const rng,
             EBM_ASSERT(0 == pBoosterCore->m_cBytesSplitPositions);
             EBM_ASSERT(0 == pBoosterCore->m_cBytesTreeNodes);
          }
+         pBoosterCore->m_cBytesMainBins = cBytesMainBins;
       }
       if(0 != cTerms) {
          error = InitializeTensors(cTerms, pBoosterCore->m_apTerms, cScores, &pBoosterCore->m_apCurrentTermTensors);

diff --git a/shared/libebm/CalcInteractionStrength.cpp b/shared/libebm/CalcInteractionStrength.cpp
@@ -178,16 +178,16 @@ EBM_API_BODY ErrorEbm EBM_CALLING_CONVENTION CalcInteractionStrength(Interaction
       LOG_0(Trace_Warning, "WARNING CalcInteractionStrength maxCardinality can't be less than 0. Turning off.");
    }
 
-   size_t cSamplesLeafMin = size_t{0}; // this is the min value
-   if(IntEbm{0} <= minSamplesLeaf) {
+   size_t cSamplesLeafMin = size_t{1}; // this is the min value
+   if(IntEbm{1} <= minSamplesLeaf) {
       cSamplesLeafMin = static_cast<size_t>(minSamplesLeaf);
       if(IsConvertError<size_t>(minSamplesLeaf)) {
          // we can never exceed a size_t number of samples, so let's just set it to the maximum if we were going to
          // overflow because it will generate the same results as if we used the true number
          cSamplesLeafMin = std::numeric_limits<size_t>::max();
       }
    } else {
-      LOG_0(Trace_Warning, "WARNING CalcInteractionStrength minSamplesLeaf can't be less than 0. Adjusting to 0.");
+      LOG_0(Trace_Warning, "WARNING CalcInteractionStrength minSamplesLeaf can't be less than 1. Adjusting to 1.");
    }
 
    FloatCalc hessianMin = static_cast<FloatCalc>(minHessian);

diff --git a/shared/libebm/GenerateTermUpdate.cpp b/shared/libebm/GenerateTermUpdate.cpp
@@ -781,16 +781,16 @@ EBM_API_BODY ErrorEbm EBM_CALLING_CONVENTION GenerateTermUpdate(void* rng,
       LOG_0(Trace_Warning, "WARNING GenerateTermUpdate learningRate is negative");
    }
 
-   size_t cSamplesLeafMin = size_t{0}; // this is the min value
-   if(IntEbm{0} <= minSamplesLeaf) {
+   size_t cSamplesLeafMin = size_t{1}; // this is the min value
+   if(IntEbm{1} <= minSamplesLeaf) {
       cSamplesLeafMin = static_cast<size_t>(minSamplesLeaf);
       if(IsConvertError<size_t>(minSamplesLeaf)) {
          // we can never exceed a size_t number of samples, so let's just set it to the maximum if we were going to
          // overflow because it will generate the same results as if we used the true number
          cSamplesLeafMin = std::numeric_limits<size_t>::max();
       }
    } else {
-      LOG_0(Trace_Warning, "WARNING GenerateTermUpdate minSamplesLeaf can't be less than 0.  Adjusting to 0.");
+      LOG_0(Trace_Warning, "WARNING GenerateTermUpdate minSamplesLeaf can't be less than 1.  Adjusting to 1.");
    }
 
    FloatCalc hessianMin = static_cast<FloatCalc>(minHessian);
@@ -820,16 +820,16 @@ EBM_API_BODY ErrorEbm EBM_CALLING_CONVENTION GenerateTermUpdate(void* rng,
       deltaStepMax = std::numeric_limits<FloatCalc>::infinity();
    }
 
-   size_t cCategorySamplesMin = size_t{0}; // this is the min value
-   if(IntEbm{0} <= minCategorySamples) {
+   size_t cCategorySamplesMin = size_t{1}; // this is the min value
+   if(IntEbm{1} <= minCategorySamples) {
       cCategorySamplesMin = static_cast<size_t>(minCategorySamples);
       if(IsConvertError<size_t>(minCategorySamples)) {
          // we can never exceed a size_t number of samples, so let's just set it to the maximum if we were going to
          // overflow because it will generate the same results as if we used the true number
          cCategorySamplesMin = std::numeric_limits<size_t>::max();
       }
    } else {
-      LOG_0(Trace_Warning, "WARNING GenerateTermUpdate minSamplesLeaf can't be less than 0.  Adjusting to 0.");
+      LOG_0(Trace_Warning, "WARNING GenerateTermUpdate minSamplesLeaf can't be less than 1.  Adjusting to 1.");
    }
 
    FloatCalc categoricalSmoothingCalc = static_cast<FloatCalc>(categoricalSmoothing);