nightly_run - CyrilB1531/lodestar GitHub Wiki

Nightly benchmark run

Generated. Do not edit. Produced by .github/workflows/bench-nightly.yml; every edit is overwritten by the next run. The curated figures, measured on a named machine, are in performance. The last known reading for a method quiet tonight is in benchmark_latest.

Read the ratios, not the means. These run on a GitHub hosted runner: a shared VM whose hardware differs from night to night and whose neighbours are unknown. An absolute figure here is not comparable to the performance page, and not reliably comparable to yesterday's. A ratio against a baseline measured in the same run, on the same VM, in the same minute, is.

This run

Commit: eba6ece0df93976b598aa7f203ae077e6acda030
Previous run: eba6ece0df93976b598aa7f203ae077e6acda030
Runner: Linux / X64 (GitHub hosted)

Classes re-run

Selected by tools/select_benchmarks.py from the sources that changed since the previous run:

BatchEmbeddingBenchmarks
BpeBenchmarks
BpeScalingBenchmarks
BpeWordCacheBenchmarks
DecompositionBenchmarks
KMeansFitIncumbentBenchmarks
KMeansLloydIncumbentBenchmarks
DbscanIncumbentBenchmarks
DbscanDimensionBenchmarks
AgglomerativeIncumbentBenchmarks
PrincipalComponentVarianceBenchmarks
FuzzBenchmarks
FuzzIncumbentBenchmarks
CdistIncumbentBenchmarks
CdistCutoffBenchmarks
FuzzCodePointBenchmarks
VarianceAndFitBenchmarks
CorrelationBenchmarks
MetaNumericsStatsBenchmarks
KsAutoBenchmarks
MetricsIncumbentBenchmarks
TokenizerIncumbentBenchmarks
VectorizerIncumbentBenchmarks
MetricsBenchmarks
RegressionMetricsBenchmarks
PartialRatioLongNeedleBenchmarks
PersistenceBenchmarks
StopWordBenchmarks
VectorMathBenchmarks
VectorizerBenchmarks
Bm25Benchmarks
CoxBenchmarks
SurvivalBenchmarks
TiledCosineTopKBenchmarks
TiledSparseDenseProductBenchmarks
BitParallelEditDistanceBenchmarks
ChainedProductBenchmarks
TiledMinHashSignaturesBenchmarks
StatsBenchmarks
RankTestBenchmarks
DistributionTailBenchmarks
StationarityBenchmarks
SerialCorrelationBenchmarks
QuantileBenchmarks
OlsBenchmarks
GlsBenchmarks
WeightedLeastSquaresBenchmarks
RobustCovarianceBenchmarks
HacClusterBenchmarks
GlmBenchmarks
GlmPoissonBenchmarks
GlmOffsetBenchmarks
MultinomialLogitBenchmarks
VectorAutoregressionBenchmarks
TextRankBenchmarks
SilhouetteBenchmarks
ClusteringAgreementBenchmarks
EmbeddingSearchBenchmarks
MannWhitneyExactBenchmarks
RobustScalerSparseBenchmarks
TopKAccuracyBenchmarks
DamerauLevenshteinBenchmarks
QgramBenchmarks
KsExactTableBenchmarks
DoubleMetaphoneBenchmarks
ProcessExtractBenchmarks
FilteredVectorSearchBenchmarks
SentencePieceBpeLineageBenchmarks
BertNormalizerBenchmarks
AddedTokenScanBenchmarks
PrecompiledNormalizerBenchmarks
EmbeddingSearchHelpersBenchmarks
RankingMetricsBenchmarks
MultilabelConfusionMatrixBenchmarks
ClassifierCurveBenchmarks
PartitionValidityBenchmarks
ClassificationReportBenchmarks
MultiClassRocAucBenchmarks
KolmogorovDurbinBenchmarks
DecompositionWidthBenchmarks
HouseholderQrBenchmarks
GlmNegativeBinomialBenchmarks
LeastSquaresRoutingBenchmarks

Lodestar.Stats.Benchmarks.CorrelationBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	PairCount	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Allocated	Alloc Ratio
Lodestar_Pearson	100	974.7 ns	74.22 ns	4.07 ns	1.00	0.01	-	-	NA
MetaNumerics_Pearson	100	797.3 ns	15.69 ns	0.86 ns	0.82	0.00	0.0095	168 B	NA
Lodestar_Spearman	100	2,991.3 ns	23.26 ns	1.27 ns	3.07	0.01	0.2441	4144 B	NA
MetaNumerics_Spearman	100	3,663.8 ns	40.91 ns	2.24 ns	3.76	0.01	0.1183	2024 B	NA
Lodestar_KendallTau	100	5,385.7 ns	198.87 ns	10.90 ns	5.53	0.02	-	-	NA
MetaNumerics_KendallTau	100	10,799.1 ns	4,899.35 ns	268.55 ns	11.08	0.24	-	152 B	NA
Lodestar_KendallTau_Asymptotic	100	5,335.9 ns	37.65 ns	2.06 ns	5.47	0.02	-	-	NA

Lodestar_Pearson	10000	86,130.6 ns	1,375.53 ns	75.40 ns	1.00	0.00	-	-	NA
MetaNumerics_Pearson	10000	56,972.0 ns	1,414.53 ns	77.53 ns	0.66	0.00	-	168 B	NA
Lodestar_Spearman	10000	1,418,199.7 ns	68,434.31 ns	3,751.12 ns	16.47	0.04	23.4375	400147 B	NA
MetaNumerics_Spearman	10000	1,957,132.0 ns	98,845.67 ns	5,418.06 ns	22.72	0.06	7.8125	160430 B	NA
Lodestar_KendallTau	10000	2,120,454.6 ns	62,259.72 ns	3,412.67 ns	24.62	0.04	-	-	NA
MetaNumerics_KendallTau	10000	287,271,997.5 ns	3,684,933.83 ns	201,983.66 ns	3,335.31	3.24	-	-	NA
Lodestar_KendallTau_Asymptotic	10000	2,138,461.0 ns	57,604.76 ns	3,157.51 ns	24.83	0.04	-	-	NA

Lodestar.Stats.Benchmarks.DistributionTailBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
ChiSquaredSfOneDf	28.39 ns	0.763 ns	0.042 ns	1.00	-	NA
ChiSquaredSfFourDf	11.80 ns	0.757 ns	0.041 ns	0.42	-	NA
ChiSquaredSfThreeDfFarTail	41.67 ns	3.989 ns	0.219 ns	1.47	-	NA
ChiSquaredSfHundredDf	71.77 ns	1.997 ns	0.109 ns	2.53	-	NA
ChiSquaredSfFractionalDf	112.67 ns	0.379 ns	0.021 ns	3.97	-	NA
NormalQuantile	115.50 ns	1.335 ns	0.073 ns	4.07	-	NA

Lodestar.Stats.Benchmarks.GlmBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Regressors	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Allocated	Alloc Ratio
Lodestar_Glm	200	1	38.22 μs	16.114 μs	0.883 μs	1.00	0.03	0.7324	-	12.13 KB	1.00
Accord_Glm	200	1	68.00 μs	13.274 μs	0.728 μs	1.78	0.04	7.2021	0.2441	117.67 KB	9.70

Lodestar_Glm	200	3	45.63 μs	0.349 μs	0.019 μs	1.00	0.00	1.1597	0.0610	19.38 KB	1.00
Accord_Glm	200	3	123.47 μs	36.967 μs	2.026 μs	2.71	0.04	10.2539	0.4883	168.44 KB	8.69

Lodestar_Glm	2000	1	362.36 μs	21.105 μs	1.157 μs	1.00	0.00	6.3477	0.4883	110.56 KB	1.00
Accord_Glm	2000	1	504.76 μs	133.255 μs	7.304 μs	1.39	0.02	53.7109	15.6250	881.77 KB	7.98

Lodestar_Glm	2000	3	429.42 μs	110.927 μs	6.080 μs	1.00	0.02	10.2539	-	174.06 KB	1.00
Accord_Glm	2000	3	901.26 μs	270.750 μs	14.841 μs	2.10	0.04	77.1484	33.2031	1265.51 KB	7.27

Lodestar.Stats.Benchmarks.GlmNegativeBinomialBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
Fit	1000	280.4 μs	15.86 μs	0.87 μs	3.4180	-	-	58.2 KB
Fit	10000	3,140.5 μs	47.08 μs	2.58 μs	97.6563	97.6563	97.6563	550.45 KB
Fit	100000	31,508.2 μs	1,070.50 μs	58.68 μs	437.5000	437.5000	437.5000	5474.06 KB

Lodestar.Stats.Benchmarks.GlmOffsetBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Poisson	200	31.36 μs	7.293 μs	0.400 μs	1.00	0.02	0.7324	-	-	12.13 KB	1.00
PoissonWithExposure	200	61.37 μs	2.087 μs	0.114 μs	1.96	0.02	1.2207	-	-	20.52 KB	1.69

Poisson	20000	4,310.44 μs	1,570.863 μs	86.104 μs	1.00	0.02	328.1250	328.1250	328.1250	1097.77 KB	1.00
PoissonWithExposure	20000	7,910.40 μs	148.367 μs	8.133 μs	1.84	0.03	500.0000	500.0000	500.0000	1878.97 KB	1.71

Lodestar.Stats.Benchmarks.GlmPoissonBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	MeanCount	Mean	Error	StdDev	Gen0	Gen1	Allocated
Fit	5	305.3 μs	23.71 μs	1.30 μs	6.3477	0.4883	110.56 KB
Fit	50000	321.8 μs	88.03 μs	4.83 μs	6.3477	0.4883	110.56 KB
Fit	5000000	316.2 μs	8.30 μs	0.45 μs	6.3477	0.4883	110.56 KB

Lodestar.Stats.Benchmarks.GlsBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar_Gls	50	31.34 μs	4.240 μs	0.232 μs	1.00	0.01	1.6479	0.0610	-	27.1 KB	1.00
MathNet_Gls	50	112.93 μs	26.409 μs	1.448 μs	3.60	0.05	1.9531	0.1221	-	32.94 KB	1.22

Lodestar_Gls	200	712.31 μs	94.587 μs	5.185 μs	1.00	0.01	99.6094	99.6094	99.6094	336.57 KB	1.00
MathNet_Gls	200	2,433.67 μs	1,314.230 μs	72.037 μs	3.42	0.09	97.6563	97.6563	97.6563	369.89 KB	1.10

Lodestar_Gls	500	7,787.91 μs	1,391.392 μs	76.267 μs	1.00	0.01	296.8750	296.8750	296.8750	2010.2 KB	1.00
MathNet_Gls	500	30,174.90 μs	23,409.210 μs	1,283.138 μs	3.87	0.15	281.2500	281.2500	281.2500	2239.78 KB	1.11

Lodestar_Gls	1000	50,264.66 μs	2,169.296 μs	118.906 μs	1.00	0.00	100.0000	100.0000	100.0000	7924.07 KB	1.00
MathNet_Gls	1000	192,868.40 μs	41,182.003 μs	2,257.325 μs	3.84	0.04	-	-	-	8885.34 KB	1.12

Lodestar.Stats.Benchmarks.HacClusterBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Lags	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Nonrobust	100	4	7.057 μs	1.3520 μs	0.0741 μs	1.00	0.01	0.1602	-	-	2.65 KB	1.00
Hc0	100	4	11.255 μs	0.8372 μs	0.0459 μs	1.60	0.02	0.5341	-	-	8.75 KB	3.30
Hac	100	4	21.160 μs	1.7258 μs	0.0946 μs	3.00	0.03	0.7629	-	-	12.9 KB	4.87
Cluster	100	4	10.247 μs	0.7830 μs	0.0429 μs	1.45	0.01	0.5798	-	-	9.71 KB	3.67

Nonrobust	10000	4	460.129 μs	20.0378 μs	1.0983 μs	1.00	0.00	4.3945	0.9766	-	79.99 KB	1.00
Hc0	10000	4	1,151.257 μs	319.1584 μs	17.4942 μs	2.50	0.03	123.0469	123.0469	123.0469	472.89 KB	5.91
Hac	10000	4	2,206.617 μs	3,086.4207 μs	169.1771 μs	4.80	0.32	246.0938	246.0938	246.0938	863.84 KB	10.80
Cluster	10000	4	1,011.596 μs	524.7550 μs	28.7636 μs	2.20	0.05	123.0469	123.0469	123.0469	565.18 KB	7.07

Lodestar.Stats.Benchmarks.KolmogorovDurbinBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
Asymptotic	100	7.152 μs	0.5511 μs	0.0302 μs	0.2823	-	-	4.73 KB
Asymptotic	280	121.875 μs	11.8451 μs	0.6493 μs	1.4648	-	-	24.7 KB
Asymptotic	20000	1,285.832 μs	110.5349 μs	6.0588 μs	99.6094	99.6094	99.6094	352.77 KB

Lodestar.Stats.Benchmarks.KsAutoBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	Gen0	Allocated	Alloc Ratio
Before_Asymptotic	1000	32.18 μs	1.706 μs	0.094 μs	1.00	0.9155	15.73 KB	1.00
After_Auto	1000	33.81 μs	4.305 μs	0.236 μs	1.05	0.9155	15.67 KB	1.00

Before_Asymptotic	10000	2,003.30 μs	95.603 μs	5.240 μs	1.00	7.8125	156.3 KB	1.00
After_Auto	10000	1,454.56 μs	15.631 μs	0.857 μs	0.73	7.8125	156.3 KB	1.00

Lodestar.Stats.Benchmarks.KsExactTableBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Shape	Mean	Error	StdDev	Gen0	Allocated
TwoSidedExact	999x1001	3,427.82 μs	232.939 μs	12.768 μs	-	39.23 KB
TwoSidedExact	99x101	47.31 μs	4.093 μs	0.224 μs	0.2441	4.07 KB

Lodestar.Stats.Benchmarks.LeastSquaresRoutingBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Regressors	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar_Estimate	4	131.4 μs	16.39 μs	0.90 μs	1.00	0.01	49.8047	49.8047	49.8047	188.26 KB	1.00
Lodestar_Fit	4	174.6 μs	24.21 μs	1.33 μs	1.33	0.01	1.9531	-	-	33.12 KB	0.18
MathNet_Qr	4	560.7 μs	423.12 μs	23.19 μs	4.27	0.15	190.4297	190.4297	190.4297	705.61 KB	3.75

Lodestar_Estimate	250	59,390.2 μs	45,493.90 μs	2,493.67 μs	1.00	0.05	375.0000	375.0000	375.0000	8865.83 KB	1.00
Lodestar_Fit	250	296,193.9 μs	22,940.20 μs	1,257.43 μs	4.99	0.18	-	-	-	3499.3 KB	0.39
MathNet_Qr	250	475,282.1 μs	205,081.75 μs	11,241.22 μs	8.01	0.33	-	-	-	32660.53 KB	3.68

Lodestar.Stats.Benchmarks.MannWhitneyExactBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Shape	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
Exact	141x141	993.5 ms	357.4 ms	19.59 ms	1000.0000	1000.0000	1000.0000	43.23 MB
Exact	8x2500	469.9 ms	113.1 ms	6.20 ms	-	-	-	2.9 MB

Lodestar.Stats.Benchmarks.MetaNumericsStatsBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Allocated	Alloc Ratio
Lodestar_StudentT	100	775.8 ns	10.53 ns	0.58 ns	1.00	0.00	-	-	NA
MetaNumerics_StudentT	100	2,191.6 ns	206.55 ns	11.32 ns	2.82	0.01	0.0038	104 B	NA
Lodestar_MannWhitney	100	2,515.5 ns	365.43 ns	20.03 ns	3.24	0.02	-	-	NA
MetaNumerics_MannWhitney	100	3,996.7 ns	381.56 ns	20.91 ns	5.15	0.02	0.0687	1176 B	NA
Lodestar_KruskalWallis	100	7,376.1 ns	1,299.29 ns	71.22 ns	9.51	0.08	-	32 B	NA
MetaNumerics_KruskalWallis	100	9,862.7 ns	3,818.69 ns	209.32 ns	12.71	0.23	0.1068	1896 B	NA
Lodestar_KolmogorovSmirnov	100	2,494.2 ns	36.99 ns	2.03 ns	3.21	0.00	0.0954	1648 B	NA
MetaNumerics_KolmogorovSmirnov	100	4,061.1 ns	233.24 ns	12.78 ns	5.23	0.01	0.0687	1168 B	NA
Lodestar_OneWayAnova	100	969.4 ns	124.19 ns	6.81 ns	1.25	0.01	0.0019	32 B	NA
MetaNumerics_OneWayAnova	100	3,168.7 ns	99.33 ns	5.44 ns	4.08	0.01	0.0420	744 B	NA
Lodestar_Wilcoxon	100	1,037.3 ns	69.40 ns	3.80 ns	1.34	0.00	0.0477	824 B	NA
MetaNumerics_Wilcoxon	100	1,069.3 ns	129.01 ns	7.07 ns	1.38	0.01	0.0572	976 B	NA
Lodestar_FisherExact	100	461.8 ns	43.47 ns	2.38 ns	0.60	0.00	-	-	NA
MetaNumerics_FisherExact	100	1,542.5 ns	107.10 ns	5.87 ns	1.99	0.01	0.0553	944 B	NA
Lodestar_ChiSquare	100	121.8 ns	14.86 ns	0.81 ns	0.16	0.00	0.0100	168 B	NA
MetaNumerics_ChiSquare	100	670.6 ns	47.64 ns	2.61 ns	0.86	0.00	0.0572	968 B	NA

Lodestar_StudentT	10000	37,579.8 ns	335.23 ns	18.38 ns	1.000	0.00	-	-	NA
MetaNumerics_StudentT	10000	178,738.0 ns	1,609.95 ns	88.25 ns	4.756	0.00	-	104 B	NA
Lodestar_MannWhitney	10000	716,969.9 ns	16,288.12 ns	892.81 ns	19.079	0.02	-	-	NA
MetaNumerics_MannWhitney	10000	2,096,231.3 ns	960,866.07 ns	52,668.31 ns	55.781	1.21	3.9063	80382 B	NA
Lodestar_KruskalWallis	10000	1,473,575.7 ns	21,833.86 ns	1,196.79 ns	39.212	0.03	-	35 B	NA
MetaNumerics_KruskalWallis	10000	3,645,930.0 ns	173,764.40 ns	9,524.61 ns	97.018	0.22	3.9063	120702 B	NA
Lodestar_KolmogorovSmirnov	10000	1,458,716.6 ns	50,476.44 ns	2,766.78 ns	38.816	0.07	7.8125	160051 B	NA
MetaNumerics_KolmogorovSmirnov	10000	2,151,924.7 ns	90,704.37 ns	4,971.81 ns	57.263	0.12	3.9063	80374 B	NA
Lodestar_OneWayAnova	10000	84,464.5 ns	402.93 ns	22.09 ns	2.248	0.00	-	32 B	NA
MetaNumerics_OneWayAnova	10000	267,793.0 ns	8,415.91 ns	461.30 ns	7.126	0.01	-	745 B	NA
Lodestar_Wilcoxon	10000	189,536.2 ns	6,348.23 ns	347.97 ns	5.044	0.01	4.6387	80056 B	NA
MetaNumerics_Wilcoxon	10000	381,039.6 ns	22,614.79 ns	1,239.59 ns	10.139	0.03	4.3945	80177 B	NA
Lodestar_FisherExact	10000	460.6 ns	0.99 ns	0.05 ns	0.012	0.00	-	-	NA
MetaNumerics_FisherExact	10000	1,541.5 ns	114.09 ns	6.25 ns	0.041	0.00	0.0553	944 B	NA
Lodestar_ChiSquare	10000	119.0 ns	6.62 ns	0.36 ns	0.003	0.00	0.0100	168 B	NA
MetaNumerics_ChiSquare	10000	677.5 ns	57.63 ns	3.16 ns	0.018	0.00	0.0572	968 B	NA

Lodestar.Stats.Benchmarks.MultinomialLogitBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Allocated	Alloc Ratio
Lodestar	200	137.8 μs	4.75 μs	0.26 μs	1.00	0.00	0.9766	-	19.24 KB	1.00
Accord	200	993.8 μs	41.60 μs	2.28 μs	7.21	0.02	87.8906	7.8125	1466.83 KB	76.23

Lodestar	2000	1,348.2 μs	194.42 μs	10.66 μs	1.00	0.01	7.8125	-	131.74 KB	1.00
Accord	2000	8,596.6 μs	10,109.03 μs	554.11 μs	6.38	0.36	734.3750	359.3750	12015.21 KB	91.20

Lodestar.Stats.Benchmarks.OlsBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar_Ols	100	6.901 μs	0.1702 μs	0.0093 μs	1.00	0.00	0.1602	-	-	2.65 KB	1.00
Accord_Ols	100	269.367 μs	39.1291 μs	2.1448 μs	39.03	0.27	2.4414	-	-	43 KB	16.24
MathNet_Ols	100	41.485 μs	16.5253 μs	0.9058 μs	6.01	0.11	2.0752	-	-	34.1 KB	12.88

Lodestar_Ols	10000	452.365 μs	62.8765 μs	3.4465 μs	1.00	0.01	4.3945	0.9766	-	79.99 KB	1.00
Accord_Ols	10000	4,707.606 μs	255.2461 μs	13.9909 μs	10.41	0.07	210.9375	187.5000	-	3523.47 KB	44.05
MathNet_Ols	10000	2,246.838 μs	718.1713 μs	39.3654 μs	4.97	0.08	441.4063	441.4063	441.4063	1763.75 KB	22.05

Lodestar.Stats.Benchmarks.QuantileBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
NormalQuantile	87.36 ns	2.457 ns	0.135 ns	1.00	-	NA
NormalQuantileFarTail	86.95 ns	0.585 ns	0.032 ns	1.00	-	NA
StudentQuantile	623.79 ns	9.361 ns	0.513 ns	7.14	-	NA
StudentQuantileCauchyFarTail	205.38 ns	4.411 ns	0.242 ns	2.35	-	NA

Lodestar.Stats.Benchmarks.RankTestBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Ties	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
KruskalWallisTest	10000	False	1,487.2 μs	16.99 μs	0.93 μs	-	-	-	82 B
WilcoxonPaired	10000	False	428.3 μs	25.60 μs	1.40 μs	4.3945	-	-	80057 B
KruskalWallisManyGroups	10000	False	712.7 μs	35.89 μs	1.97 μs	16.6016	-	-	279681 B
WilcoxonPooledDifferences	10000	False	1,269.0 μs	43.63 μs	2.39 μs	-	-	-	-
MannWhitneyTest	10000	False	700.7 μs	27.18 μs	1.49 μs	-	-	-	-
KruskalWallisTest	10000	True	623.9 μs	18.80 μs	1.03 μs	-	-	-	81 B
WilcoxonPaired	10000	True	246.7 μs	18.03 μs	0.99 μs	4.3945	-	-	80057 B
KruskalWallisManyGroups	10000	True	489.9 μs	67.02 μs	3.67 μs	16.6016	-	-	279680 B
WilcoxonPooledDifferences	10000	True	787.4 μs	10.24 μs	0.56 μs	-	-	-	-
MannWhitneyTest	10000	True	410.5 μs	3.63 μs	0.20 μs	-	-	-	-
KruskalWallisTest	100000	False	15,037.3 μs	1,113.18 μs	61.02 μs	-	-	-	-
WilcoxonPaired	100000	False	4,593.2 μs	314.59 μs	17.24 μs	101.5625	101.5625	101.5625	800164 B
KruskalWallisManyGroups	100000	False	9,901.1 μs	1,399.87 μs	76.73 μs	312.5000	312.5000	312.5000	2800334 B
WilcoxonPooledDifferences	100000	False	35,536.6 μs	9,956.88 μs	545.77 μs	933.3333	933.3333	933.3333	13200775 B
MannWhitneyTest	100000	False	7,451.4 μs	957.25 μs	52.47 μs	-	-	-	-
KruskalWallisTest	100000	True	6,053.1 μs	118.41 μs	6.49 μs	-	-	-	88 B
WilcoxonPaired	100000	True	3,070.0 μs	295.97 μs	16.22 μs	113.2813	113.2813	113.2813	800172 B
KruskalWallisManyGroups	100000	True	6,071.1 μs	2,891.90 μs	158.51 μs	343.7500	343.7500	343.7500	2800352 B
WilcoxonPooledDifferences	100000	True	22,653.6 μs	10,764.62 μs	590.05 μs	968.7500	968.7500	968.7500	13076301 B
MannWhitneyTest	100000	True	4,202.7 μs	144.96 μs	7.95 μs	-	-	-	-

Lodestar.Stats.Benchmarks.RobustCovarianceBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Covariance	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
Fit	100	Nonrobust	6.795 μs	0.4616 μs	0.0253 μs	0.1526	-	-	2.6 KB
Fit	100	Hc0	9.972 μs	0.5322 μs	0.0292 μs	0.5188	-	-	8.7 KB
Fit	100	Hc1	10.196 μs	4.5774 μs	0.2509 μs	0.5188	-	-	8.7 KB
Fit	100	Hc2	11.669 μs	0.1418 μs	0.0078 μs	0.5798	-	-	9.57 KB
Fit	100	Hc3	11.756 μs	0.6233 μs	0.0342 μs	0.5798	-	-	9.57 KB
Fit	10000	Nonrobust	445.324 μs	192.8094 μs	10.5685 μs	4.3945	0.9766	-	79.95 KB
Fit	10000	Hc0	1,038.137 μs	1,328.2986 μs	72.8085 μs	124.0234	124.0234	124.0234	472.85 KB
Fit	10000	Hc1	1,021.523 μs	45.9931 μs	2.5210 μs	123.0469	123.0469	123.0469	472.85 KB
Fit	10000	Hc2	1,182.426 μs	1,888.6945 μs	103.5257 μs	123.0469	123.0469	123.0469	551.06 KB
Fit	10000	Hc3	1,141.795 μs	255.8537 μs	14.0242 μs	123.0469	123.0469	123.0469	551.06 KB

Lodestar.Stats.Benchmarks.SerialCorrelationBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Allocated	Alloc Ratio
LodestarAutocorrelation	200	4.577 μs	1.2278 μs	0.0673 μs	1.00	0.02	0.1526	2624 B	1.00
CortexAutocorrelation	200	11.542 μs	0.4216 μs	0.0231 μs	2.52	0.03	-	192 B	0.07
LodestarPartialAutocorrelation	200	5.053 μs	1.9752 μs	0.1083 μs	1.10	0.02	0.1678	2816 B	1.07
CortexPartialAutocorrelation	200	12.188 μs	0.3332 μs	0.0183 μs	2.66	0.03	0.0458	768 B	0.29
LodestarLjungBox	200	5.078 μs	1.3920 μs	0.0763 μs	1.11	0.02	0.1373	2344 B	0.89
CortexLjungBox	200	11.671 μs	0.5389 μs	0.0295 μs	2.55	0.03	-	224 B	0.09

LodestarAutocorrelation	2000	49.953 μs	4.5716 μs	0.2506 μs	1.00	0.01	0.9766	17024 B	1.00
CortexAutocorrelation	2000	126.218 μs	3.4765 μs	0.1906 μs	2.53	0.01	-	192 B	0.01
LodestarPartialAutocorrelation	2000	50.401 μs	3.7867 μs	0.2076 μs	1.01	0.01	0.9766	17216 B	1.01
CortexPartialAutocorrelation	2000	121.748 μs	7.0557 μs	0.3867 μs	2.44	0.01	-	768 B	0.05
LodestarLjungBox	2000	51.122 μs	1.1126 μs	0.0610 μs	1.02	0.00	0.9766	16744 B	0.98
CortexLjungBox	2000	121.201 μs	4.1223 μs	0.2260 μs	2.43	0.01	-	224 B	0.01

Lodestar.Stats.Benchmarks.StationarityBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
LodestarAugmentedDickeyFuller	200	7.546 μs	3.8939 μs	0.2134 μs	1.00	0.03	1.4648	0.0458	-	23.99 KB	1.00
LodestarAdfAutolag	200	36.159 μs	11.2739 μs	0.6180 μs	4.79	0.14	4.0283	0.2441	-	66.56 KB	2.77
CortexAugmentedDickeyFuller	200	25.346 μs	0.7897 μs	0.0433 μs	3.36	0.08	0.9766	0.0305	-	16.18 KB	0.67
LodestarKpss	200	3.943 μs	0.2981 μs	0.0163 μs	0.52	0.01	0.0992	-	-	1.69 KB	0.07
CortexKpss	200	4.291 μs	0.4593 μs	0.0252 μs	0.57	0.01	0.2060	-	-	3.43 KB	0.14
LodestarDecompose	200	1.941 μs	4.0817 μs	0.2237 μs	0.26	0.03	0.3967	0.0038	-	6.5 KB	0.27
CortexDecompose	200	3.112 μs	1.3039 μs	0.0715 μs	0.41	0.01	0.4997	0.0076	-	8.16 KB	0.34

LodestarAugmentedDickeyFuller	2000	119.039 μs	74.6437 μs	4.0915 μs	1.00	0.04	30.2734	30.2734	30.2734	234.95 KB	1.00
LodestarAdfAutolag	2000	1,156.058 μs	420.3142 μs	23.0388 μs	9.72	0.33	249.0234	249.0234	249.0234	1001.27 KB	4.26
CortexAugmentedDickeyFuller	2000	295.524 μs	56.2976 μs	3.0859 μs	2.48	0.08	30.2734	30.2734	30.2734	142.76 KB	0.61
LodestarKpss	2000	96.303 μs	7.5558 μs	0.4142 μs	0.81	0.02	0.8545	-	-	15.75 KB	0.07
CortexKpss	2000	99.732 μs	2.6283 μs	0.1441 μs	0.84	0.02	1.8311	-	-	31.55 KB	0.13
LodestarDecompose	2000	15.597 μs	2.2933 μs	0.1257 μs	0.13	0.00	3.8147	0.3967	-	62.75 KB	0.27
CortexDecompose	2000	34.879 μs	1.8780 μs	0.1029 μs	0.29	0.01	4.7607	0.4272	-	78.48 KB	0.33

Lodestar.Stats.Benchmarks.StatsBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
LodestarWelchT	100	967.9 ns	27.43 ns	1.50 ns	1.00	0.00	-	-	-	-	NA
AccordWelchT	100	38,410.5 ns	4,678.30 ns	256.43 ns	39.69	0.24	-	-	-	392 B	NA
LodestarMannWhitney	100	2,457.0 ns	49.14 ns	2.69 ns	2.54	0.00	-	-	-	-	NA
AccordMannWhitney	100	23,449.6 ns	6,489.18 ns	355.69 ns	24.23	0.32	1.3733	0.0305	-	23336 B	NA
LodestarChiSquare	100	124.7 ns	7.45 ns	0.41 ns	0.13	0.00	0.0100	-	-	168 B	NA
AccordChiSquare	100	222.2 ns	34.26 ns	1.88 ns	0.23	0.00	0.0100	-	-	168 B	NA

LodestarWelchT	10000	37,641.0 ns	508.89 ns	27.89 ns	1.000	0.00	-	-	-	-	NA
AccordWelchT	10000	141,027.1 ns	8,598.62 ns	471.32 ns	3.747	0.01	-	-	-	392 B	NA
LodestarMannWhitney	10000	707,065.6 ns	20,934.71 ns	1,147.50 ns	18.784	0.03	-	-	-	-	NA
AccordMannWhitney	10000	12,592,977.7 ns	561,252.68 ns	30,764.15 ns	334.555	0.74	234.3750	234.3750	234.3750	2241217 B	NA
LodestarChiSquare	10000	123.9 ns	3.10 ns	0.17 ns	0.003	0.00	0.0100	-	-	168 B	NA
AccordChiSquare	10000	219.3 ns	9.25 ns	0.51 ns	0.006	0.00	0.0100	-	-	168 B	NA

Lodestar.Stats.Benchmarks.VarianceAndFitBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	GroupSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Allocated	Alloc Ratio
Lodestar_Levene	100	2,035.5 ns	127.30 ns	6.98 ns	1.00	0.00	0.3014	-	5072 B	1.000
Accord_Levene	100	4,022.5 ns	37.17 ns	2.04 ns	1.98	0.01	0.3204	-	5480 B	1.080
Lodestar_Bartlett	100	597.2 ns	43.79 ns	2.40 ns	0.29	0.00	0.0048	-	80 B	0.016
Accord_Bartlett	100	1,183.5 ns	45.07 ns	2.47 ns	0.58	0.00	0.0095	-	168 B	0.033
Lodestar_Binomial	100	1,288.7 ns	116.84 ns	6.40 ns	0.63	0.00	0.0019	-	48 B	0.009
Accord_Binomial	100	4,622.4 ns	209.01 ns	11.46 ns	2.27	0.01	0.1297	-	2216 B	0.437
Lodestar_AndersonDarling	100	8,639.6 ns	477.30 ns	26.16 ns	4.24	0.02	0.0458	-	1000 B	0.197
Accord_AndersonDarling	100	6,274.2 ns	244.56 ns	13.41 ns	3.08	0.01	0.0610	-	1096 B	0.216
Lodestar_ClopperPearson	100	4,939.7 ns	216.14 ns	11.85 ns	2.43	0.01	-	-	48 B	0.009

Lodestar_Levene	10000	373,264.2 ns	11,947.39 ns	654.88 ns	1.000	0.00	28.3203	-	480272 B	1.000
Accord_Levene	10000	1,322,007.2 ns	67,026.78 ns	3,673.96 ns	3.542	0.01	27.3438	5.8594	480681 B	1.001
Lodestar_Bartlett	10000	56,311.1 ns	191.72 ns	10.51 ns	0.151	0.00	-	-	80 B	0.000
Accord_Bartlett	10000	112,352.6 ns	2,559.10 ns	140.27 ns	0.301	0.00	-	-	168 B	0.000
Lodestar_Binomial	10000	1,989.9 ns	143.65 ns	7.87 ns	0.005	0.00	-	-	48 B	0.000
Accord_Binomial	10000	892,306.2 ns	18,211.93 ns	998.26 ns	2.391	0.00	11.7188	2.9297	200217 B	0.417
Lodestar_AndersonDarling	10000	1,395,263.9 ns	48,857.95 ns	2,678.07 ns	3.738	0.01	3.9063	-	80201 B	0.167
Accord_AndersonDarling	10000	NA	NA	NA	?	?	NA	NA	NA	?
Lodestar_ClopperPearson	10000	35,154.1 ns	4,681.52 ns	256.61 ns	0.094	0.00	-	-	48 B	0.000

Benchmarks with issues: VarianceAndFitBenchmarks.Accord_AndersonDarling: ShortRun(IterationCount=3, LaunchCount=1, WarmupCount=3) [GroupSize=10000]

Lodestar.Stats.Benchmarks.VectorAutoregressionBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Variables	Lags	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
Fit	500	2	1	22.25 μs	0.415 μs	0.023 μs	2.9602	0.3357	-	48.37 KB
Fit	500	2	4	44.40 μs	5.673 μs	0.311 μs	5.8594	1.4038	-	96.13 KB
Fit	500	5	1	69.98 μs	3.683 μs	0.202 μs	6.5918	1.5869	-	109.52 KB
Fit	500	5	4	176.29 μs	82.308 μs	4.512 μs	14.1602	6.8359	-	233.92 KB
Fit	5000	2	1	288.88 μs	118.873 μs	6.516 μs	73.7305	73.7305	73.7305	470.29 KB
Fit	5000	2	4	674.01 μs	219.780 μs	12.047 μs	221.6797	221.6797	221.6797	940.03 KB
Fit	5000	5	1	992.11 μs	573.553 μs	31.438 μs	142.5781	142.5781	142.5781	1058.83 KB
Fit	5000	5	4	2,829.85 μs	888.039 μs	48.676 μs	496.0938	496.0938	496.0938	2238.16 KB

Lodestar.Stats.Benchmarks.WeightedLeastSquaresBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	SampleSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar_Wls	200	11.85 μs	0.966 μs	0.053 μs	1.00	0.01	0.1984	-	-	3.43 KB	1.00
MathNet_Wls	200	36.41 μs	16.087 μs	0.882 μs	3.07	0.07	3.6621	0.1831	-	58.19 KB	16.97

Lodestar_Wls	2000	98.46 μs	8.267 μs	0.453 μs	1.00	0.01	0.9766	-	-	17.49 KB	1.00
MathNet_Wls	2000	197.75 μs	161.909 μs	8.875 μs	2.01	0.08	32.7148	12.6953	-	508.04 KB	29.04

Lodestar_Wls	20000	1,068.15 μs	525.197 μs	28.788 μs	1.00	0.03	46.8750	46.8750	46.8750	158.44 KB	1.00
MathNet_Wls	20000	2,881.88 μs	4,134.016 μs	226.599 μs	2.70	0.19	1070.3125	1070.3125	1070.3125	5024.18 KB	31.71

Lodestar_Wls	200000	9,707.20 μs	1,402.533 μs	76.878 μs	1.00	0.01	93.7500	93.7500	93.7500	1564.95 KB	1.00
MathNet_Wls	200000	22,216.98 μs	47,264.883 μs	2,590.748 μs	2.29	0.23	812.5000	812.5000	812.5000	50022.21 KB	31.96

Lodestar.Text.Benchmarks.AddedTokenScanBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Gen0	Allocated
ChatTemplate	97.59 ms	6.339 ms	0.347 ms	2000.0000	33.3 MB
ProseWithoutAddedTokens	82.70 ms	0.013 ms	0.001 ms	1714.2857	28.47 MB

Lodestar.Text.Benchmarks.AgglomerativeIncumbentBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Rows	Method	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar_Fit	500	average	1,877.2 μs	20.57 μs	1.13 μs	1.00	0.00	248.0469	248.0469	248.0469	1022.55 KB	1.00
Aglomera_GetClustering	500	average	88,472.2 μs	6,512.37 μs	356.97 μs	47.13	0.17	3666.6667	1500.0000	833.3333	59880.82 KB	58.56

Lodestar_Fit	500	complete	1,808.5 μs	42.25 μs	2.32 μs	1.00	0.00	248.0469	248.0469	248.0469	1022.4 KB	1.00
Aglomera_GetClustering	500	complete	89,058.4 μs	21,827.01 μs	1,196.41 μs	49.25	0.58	5000.0000	1800.0000	800.0000	83111.68 KB	81.29

Lodestar_Fit	500	single	779.0 μs	59.88 μs	3.28 μs	1.00	0.01	0.9766	-	-	30.34 KB	1.00
Aglomera_GetClustering	500	single	135,972.6 μs	14,088.07 μs	772.21 μs	174.55	1.07	4750.0000	1750.0000	750.0000	81289.96 KB	2,678.88

Lodestar_Fit	500	ward	2,167.5 μs	64.57 μs	3.54 μs	1.00	0.00	246.0938	246.0938	246.0938	1022.4 KB	1.00
Aglomera_GetClustering	500	ward	66,796.7 μs	5,778.71 μs	316.75 μs	30.82	0.13	3875.0000	1750.0000	875.0000	64191.22 KB	62.78

Lodestar_Fit	1500	average	13,826.6 μs	2,533.25 μs	138.86 μs	1.00	0.01	500.0000	500.0000	500.0000	8925.15 KB	1.00
Aglomera_GetClustering	1500	average	1,860,498.1 μs	53,803.13 μs	2,949.13 μs	134.57	1.19	35000.0000	10000.0000	4000.0000	537149.55 KB	60.18

Lodestar_Fit	1500	complete	12,521.5 μs	389.53 μs	21.35 μs	1.00	0.00	500.0000	500.0000	500.0000	8925 KB	1.00
Aglomera_GetClustering	1500	complete	1,823,599.6 μs	11,831.29 μs	648.51 μs	145.64	0.22	48000.0000	12000.0000	4000.0000	747352.87 KB	83.74

Lodestar_Fit	1500	single	6,644.4 μs	563.21 μs	30.87 μs	1.00	0.01	-	-	-	89.92 KB	1.00
Aglomera_GetClustering	1500	single	2,979,997.6 μs	56,568.38 μs	3,100.70 μs	448.50	1.85	47000.0000	12000.0000	4000.0000	732674.31 KB	8,147.90

Lodestar_Fit	1500	ward	16,195.8 μs	1,493.56 μs	81.87 μs	1.00	0.01	500.0000	500.0000	500.0000	8925.01 KB	1.00
Aglomera_GetClustering	1500	ward	1,712,285.9 μs	85,325.79 μs	4,676.99 μs	105.73	0.52	38000.0000	11000.0000	4000.0000	573621.47 KB	64.27

Lodestar.Text.Benchmarks.BatchEmbeddingBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	CorpusSize	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Allocated	Alloc Ratio
UnitLoop	1	5.738 μs	1.6001 μs	0.0877 μs	1.00	0.02	0.1297	-	2.23 KB	1.00
EmbedBatch	1	6.026 μs	1.0167 μs	0.0557 μs	1.05	0.02	0.1602	-	2.63 KB	1.18
EmbedBatchBucketed	1	5.966 μs	0.3922 μs	0.0215 μs	1.04	0.01	0.1602	-	2.63 KB	1.18

UnitLoop	8	56.293 μs	32.0973 μs	1.7594 μs	1.00	0.04	1.7700	-	29.16 KB	1.00
EmbedBatch	8	22.045 μs	5.6844 μs	0.3116 μs	0.39	0.01	1.3428	0.0305	22.19 KB	0.76
EmbedBatchBucketed	8	22.138 μs	2.0331 μs	0.1114 μs	0.39	0.01	1.3428	0.0305	22.19 KB	0.76

UnitLoop	32	221.117 μs	22.1009 μs	1.2114 μs	1.00	0.01	6.5918	-	109.74 KB	1.00
EmbedBatch	32	78.440 μs	4.0662 μs	0.2229 μs	0.35	0.00	5.0049	0.1221	82.35 KB	0.75
EmbedBatchBucketed	32	72.059 μs	17.4740 μs	0.9578 μs	0.33	0.00	4.1504	0.1221	68.84 KB	0.63

UnitLoop	128	900.455 μs	51.5727 μs	2.8269 μs	1.00	0.00	26.3672	-	438.9 KB	1.00
EmbedBatch	128	315.260 μs	16.6924 μs	0.9150 μs	0.35	0.00	20.0195	2.4414	328.54 KB	0.75
EmbedBatchBucketed	128	268.010 μs	26.1407 μs	1.4329 μs	0.30	0.00	15.6250	1.9531	261.02 KB	0.59

Lodestar.Text.Benchmarks.BertNormalizerBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Text	Lowercase	Mean	Error	StdDev	Gen0	Allocated
Encode	Ascii	False	39.04 ms	56.935 ms	3.121 ms	714.2857	11.7 MB
Encode	Ascii	True	39.70 ms	5.592 ms	0.307 ms	692.3077	11.7 MB
Encode	Accented	False	35.17 ms	2.069 ms	0.113 ms	666.6667	11.69 MB
Encode	Accented	True	48.11 ms	0.424 ms	0.023 ms	1272.7273	20.38 MB
Encode	Cjk	False	39.73 ms	24.277 ms	1.331 ms	1307.6923	21.55 MB
Encode	Cjk	True	52.43 ms	0.860 ms	0.047 ms	1500.0000	24.48 MB

Lodestar.Text.Benchmarks.Bm25Benchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Documents	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
LodestarQuery	1000	2.015 μs	0.3711 μs	0.0203 μs	1.00	0.01	0.0229	-	-	424 B	1.00
LodestarQueryMultiTerm	1000	2.438 μs	0.1346 μs	0.0074 μs	1.21	0.01	0.0229	-	-	424 B	1.00
LuceneQuery	1000	3.443 μs	0.3060 μs	0.0168 μs	1.71	0.02	0.3128	-	-	5264 B	12.42
LodestarFromText	1000	8,006.823 μs	646.7880 μs	35.4526 μs	3,973.42	37.75	984.3750	906.2500	890.6250	4838072 B	11,410.55
LuceneFromText	1000	7,193.061 μs	216.9693 μs	11.8928 μs	3,569.58	31.44	78.1250	62.5000	-	1375078 B	3,243.11

LodestarQuery	20000	24.773 μs	1.5571 μs	0.0853 μs	1.00	0.00	-	-	-	424 B	1.00
LodestarQueryMultiTerm	20000	39.864 μs	3.9599 μs	0.2171 μs	1.61	0.01	-	-	-	424 B	1.00
LuceneQuery	20000	19.231 μs	0.6679 μs	0.0366 μs	0.78	0.00	0.4883	-	-	8648 B	20.40
LodestarFromText	20000	121,061.031 μs	56,119.8084 μs	3,076.1161 μs	4,886.85	108.52	2200.0000	1200.0000	800.0000	83750491 B	197,524.74
LuceneFromText	20000	143,419.661 μs	46,687.1664 μs	2,559.0811 μs	5,789.40	91.11	1250.0000	1000.0000	-	22417220 B	52,870.80

Lodestar.Text.Benchmarks.BpeBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Ratio	Gen0	Allocated	Alloc Ratio
Unigram	30.24 ms	1.014 ms	0.056 ms	1.00	312.5000	5.43 MB	1.00
Bpe	81.46 ms	1.220 ms	0.067 ms	2.69	1714.2857	28.47 MB	5.24

Lodestar.Text.Benchmarks.BpeScalingBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Length	Mean	Error	StdDev	Gen0	Allocated
BpeOnOnePathologicalToken	512	17.39 μs	2.244 μs	0.123 μs	0.4578	7.48 KB
BpeOnOnePathologicalToken	1024	65.35 μs	12.531 μs	0.687 μs	0.8545	14.53 KB
BpeOnOnePathologicalToken	2048	178.15 μs	9.686 μs	0.531 μs	1.7090	28.58 KB
BpeOnOnePathologicalToken	4096	376.16 μs	44.752 μs	2.453 μs	3.4180	56.63 KB

Lodestar.Text.Benchmarks.BpeWordCacheBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  InvocationCount=1  IterationCount=3  
LaunchCount=1  UnrollFactor=1  WarmupCount=3

Method	Mean	Error	StdDev	Allocated
EncodeUnseenProse	8.227 ms	2.171 ms	0.1190 ms	1.47 MB

Lodestar.Text.Benchmarks.CdistCutoffBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Cutoff	Widths	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Allocated	Alloc Ratio
Lodestar_Cdist_Bounded	0	False	295.88 μs	76.752 μs	4.207 μs	1.00	0.02	1.9531	32.02 KB	1.00
Lodestar_Cdist_EveryPair	0	False	287.82 μs	28.966 μs	1.588 μs	0.97	0.01	1.9531	32.02 KB	1.00

Lodestar_Cdist_Bounded	0	True	327.42 μs	25.213 μs	1.382 μs	1.00	0.01	1.9531	32.02 KB	1.00
Lodestar_Cdist_EveryPair	0	True	303.70 μs	17.885 μs	0.980 μs	0.93	0.00	1.9531	32.02 KB	1.00

Lodestar_Cdist_Bounded	60	False	317.78 μs	24.685 μs	1.353 μs	1.00	0.01	1.9531	32.02 KB	1.00
Lodestar_Cdist_EveryPair	60	False	308.53 μs	61.274 μs	3.359 μs	0.97	0.01	1.9531	32.02 KB	1.00

Lodestar_Cdist_Bounded	60	True	276.09 μs	28.866 μs	1.582 μs	1.00	0.01	1.9531	32.02 KB	1.00
Lodestar_Cdist_EveryPair	60	True	309.05 μs	8.707 μs	0.477 μs	1.12	0.01	1.9531	32.02 KB	1.00

Lodestar_Cdist_Bounded	80	False	304.11 μs	43.371 μs	2.377 μs	1.00	0.01	1.9531	32.02 KB	1.00
Lodestar_Cdist_EveryPair	80	False	294.83 μs	23.563 μs	1.292 μs	0.97	0.01	1.9531	32.02 KB	1.00

Lodestar_Cdist_Bounded	80	True	187.71 μs	10.154 μs	0.557 μs	1.00	0.00	1.9531	32.02 KB	1.00
Lodestar_Cdist_EveryPair	80	True	300.25 μs	38.122 μs	2.090 μs	1.60	0.01	1.9531	32.02 KB	1.00

Lodestar_Cdist_Bounded	90	False	245.74 μs	14.962 μs	0.820 μs	1.00	0.00	1.9531	32.02 KB	1.00
Lodestar_Cdist_EveryPair	90	False	302.02 μs	42.336 μs	2.321 μs	1.23	0.01	1.9531	32.02 KB	1.00

Lodestar_Cdist_Bounded	90	True	86.08 μs	27.527 μs	1.509 μs	1.00	0.02	1.9531	32.02 KB	1.00
Lodestar_Cdist_EveryPair	90	True	302.08 μs	9.393 μs	0.515 μs	3.51	0.05	1.9531	32.02 KB	1.00

Lodestar.Text.Benchmarks.CdistIncumbentBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Size	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar_Cdist	50	178.7 μs	37.99 μs	2.08 μs	1.00	0.01	0.9766	-	-	19.55 KB	1.00
Lodestar_HandWrittenLoop	50	183.8 μs	6.23 μs	0.34 μs	1.03	0.01	0.9766	-	-	19.55 KB	1.00
FuzzySharp_ExtractAllPerQuery	50	489.4 μs	383.00 μs	20.99 μs	2.74	0.11	18.0664	-	-	301.95 KB	15.44

Lodestar_Cdist	200	3,210.9 μs	367.18 μs	20.13 μs	1.00	0.01	97.6563	97.6563	97.6563	312.62 KB	1.00
Lodestar_HandWrittenLoop	200	3,119.6 μs	330.39 μs	18.11 μs	0.97	0.01	97.6563	97.6563	97.6563	312.62 KB	1.00
FuzzySharp_ExtractAllPerQuery	200	7,375.4 μs	284.29 μs	15.58 μs	2.30	0.01	289.0625	-	-	4723.45 KB	15.11

Lodestar.Text.Benchmarks.ClassificationReportBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Classes	Mean	Error	StdDev	Gen0	Gen1	Allocated
Report	10	577.8 ns	94.61 ns	5.19 ns	0.0935	-	1.54 KB
Report	1000	690,140.0 ns	28,041.95 ns	1,537.07 ns	6.8359	0.9766	117.56 KB

Lodestar.Text.Benchmarks.ClassifierCurveBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Samples	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
Roc	1000000	137.5 ms	11.17 ms	0.61 ms	750.0000	750.0000	750.0000	53.79 MB
PrecisionRecall	1000000	124.6 ms	10.61 ms	0.58 ms	800.0000	800.0000	800.0000	68.66 MB

Lodestar.Text.Benchmarks.ClusteringAgreementBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Samples	Clusters	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
AdjustedRandScore	100000	10	2.167 ms	0.2574 ms	0.0141 ms	-	-	-	12.07 KB
MutualInformationScore	100000	10	2.130 ms	0.2315 ms	0.0127 ms	-	-	-	12.07 KB
AdjustedMutualInformationScore	100000	10	21.033 ms	0.1634 ms	0.0090 ms	218.7500	218.7500	218.7500	1031.09 KB
AdjustedRandScore	100000	100	3.187 ms	0.0820 ms	0.0045 ms	210.9375	199.2188	199.2188	937.61 KB
MutualInformationScore	100000	100	3.145 ms	0.0221 ms	0.0012 ms	187.5000	187.5000	187.5000	937.61 KB
AdjustedMutualInformationScore	100000	100	183.833 ms	9.0898 ms	0.4982 ms	333.3333	333.3333	333.3333	1745.05 KB

Lodestar.Text.Benchmarks.DamerauLevenshteinBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Length	Mean	Error	StdDev	Gen0	Allocated
Distance	12	988.6 ns	576.6 ns	31.61 ns	0.0496	840 B
Distance	120	64,266.7 ns	763.8 ns	41.86 ns	0.1221	2128 B

Lodestar.Text.Benchmarks.DbscanDimensionBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Shape	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar_Fit	10000x8x8	332.3 ms	10.76 ms	0.59 ms	1.00	0.00	1500.0000	1500.0000	1500.0000	28.69 MB	1.00
NumFlat_Fit	10000x8x8	2,338.0 ms	600.02 ms	32.89 ms	7.04	0.09	14000.0000	8000.0000	7000.0000	156.31 MB	5.45

Lodestar_Fit	5000x16x8	115.8 ms	20.69 ms	1.13 ms	1.00	0.01	1400.0000	1400.0000	1400.0000	13.29 MB	1.00
NumFlat_Fit	5000x16x8	807.9 ms	8.26 ms	0.45 ms	6.98	0.06	5000.0000	3000.0000	3000.0000	59.77 MB	4.50

Lodestar.Text.Benchmarks.DbscanIncumbentBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Shape	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar_Fit	20000x2x10	873.96 ms	15.80 ms	0.866 ms	1.00	0.00	2000.0000	2000.0000	2000.0000	112.13 MB	1.00
NumFlat_Fit	20000x2x10	6,722.72 ms	175.80 ms	9.636 ms	7.69	0.01	38000.0000	13000.0000	11000.0000	531.87 MB	4.74
Dbscan_CalculateClusters	20000x2x10	3,955.04 ms	384.82 ms	21.093 ms	4.53	0.02	22000.0000	12000.0000	10000.0000	372.11 MB	3.32

Lodestar_Fit	5000x2x5	68.62 ms	197.97 ms	10.851 ms	1.02	0.21	1428.5714	1428.5714	1428.5714	14.02 MB	1.00
NumFlat_Fit	5000x2x5	472.12 ms	58.50 ms	3.207 ms	7.01	1.05	4000.0000	3000.0000	2000.0000	66.59 MB	4.75
Dbscan_CalculateClusters	5000x2x5	264.16 ms	33.25 ms	1.823 ms	3.92	0.59	3500.0000	3500.0000	3500.0000	40.4 MB	2.88

Lodestar.Text.Benchmarks.DecompositionBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Ratio	RatioSD
TruncatedSvd_Rank20	14.889 ms	7.1188 ms	0.3902 ms	1.00	0.03
Nmf_Rank20	95.881 ms	8.7500 ms	0.4796 ms	6.44	0.15
Nmf_Transform	3.707 ms	0.3486 ms	0.0191 ms	0.25	0.01
MlNet_ProjectToPrincipalComponents_Rank20	23.938 ms	0.7052 ms	0.0387 ms	1.61	0.04

Lodestar.Text.Benchmarks.DecompositionWidthBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Columns	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
TruncatedSvd_Rank20	500	20,712.3 μs	343.3 μs	18.82 μs	3031.2500	3000.0000	3000.0000	12151.25 KB
TruncatedSvd_Transform	500	549.9 μs	105.0 μs	5.76 μs	94.7266	90.8203	90.8203	390.73 KB
Nmf_Frobenius_Rank20	500	189,099.4 μs	16,053.1 μs	879.92 μs	4500.0000	4500.0000	4500.0000	18143.4 KB
Nmf_KullbackLeibler_Rank20	500	252,092.6 μs	16,264.0 μs	891.48 μs	4500.0000	4500.0000	4500.0000	18596.16 KB
TruncatedSvd_Rank20	20000	189,667.9 μs	6,631.7 μs	363.51 μs	2666.6667	2666.6667	2666.6667	105766.49 KB
TruncatedSvd_Transform	20000	1,898.6 μs	633.2 μs	34.71 μs	496.0938	496.0938	496.0938	3437.86 KB
Nmf_Frobenius_Rank20	20000	786,090.9 μs	29,795.0 μs	1,633.16 μs	5000.0000	5000.0000	5000.0000	156850.69 KB
Nmf_KullbackLeibler_Rank20	20000	634,963.3 μs	34,318.6 μs	1,881.12 μs	5000.0000	5000.0000	5000.0000	157305.7 KB

Lodestar.Text.Benchmarks.DoubleMetaphoneBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Gen0	Allocated
Encode	284.1 μs	179.6 μs	9.84 μs	29.7852	488.6 KB

Lodestar.Text.Benchmarks.EmbeddingSearchBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Count	Mean	Error	StdDev	Allocated
SearchTop10	10000	530.8 μs	160.0 μs	8.77 μs	1.63 KB
SearchTop10	100000	7,865.7 μs	933.7 μs	51.18 μs	1.64 KB

Lodestar.Text.Benchmarks.EmbeddingSearchHelpersBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Rows	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
FromBlockNormalized	1000	706.3 μs	119.91 μs	6.57 μs	249.0234	249.0234	249.0234	1500.18 KB
MmrSelect100	1000	7,501.1 μs	765.55 μs	41.96 μs	-	-	-	21.02 KB
FromBlockNormalized	100000	57,806.5 μs	288.10 μs	15.79 μs	222.2222	222.2222	222.2222	150000.24 KB
MmrSelect100	100000	6,919.1 μs	68.96 μs	3.78 μs	-	-	-	21.02 KB

Lodestar.Text.Benchmarks.FilteredVectorSearchBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Records	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
FilteredTop10	10000	473.3 μs	56.20 μs	3.08 μs	-	-	-	7.2 KB
Unfiltered	10000	525.4 μs	24.27 μs	1.33 μs	-	-	-	2.8 KB
HybridSelective	10000	2,736.9 μs	161.16 μs	8.83 μs	144.5313	121.0938	58.5938	2577.95 KB
HybridBroad	10000	4,372.1 μs	565.41 μs	30.99 μs	164.0625	125.0000	62.5000	3221.11 KB
FilteredTop10	100000	9,799.2 μs	812.64 μs	44.54 μs	-	-	-	7.21 KB
Unfiltered	100000	8,154.0 μs	435.08 μs	23.85 μs	-	-	-	2.81 KB
HybridSelective	100000	40,355.2 μs	994.57 μs	54.52 μs	230.7692	153.8462	-	23470.98 KB
HybridBroad	100000	55,260.7 μs	6,528.46 μs	357.85 μs	222.2222	111.1111	-	29359.82 KB

Lodestar.Text.Benchmarks.FuzzBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Allocated	Alloc Ratio
Ratio	97.01 ns	0.838 ns	0.046 ns	1.00	0.00	-	-	NA
PartialRatio	760.66 ns	394.651 ns	21.632 ns	7.84	0.19	-	-	NA
TokenSortRatio	814.01 ns	127.898 ns	7.010 ns	8.39	0.06	0.0668	1120 B	NA
TokenSetRatio	969.33 ns	34.619 ns	1.898 ns	9.99	0.02	0.0534	896 B	NA
WRatio	1,192.02 ns	442.059 ns	24.231 ns	12.29	0.22	0.0668	1120 B	NA

Lodestar.Text.Benchmarks.FuzzCodePointBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Allocated	Alloc Ratio
Ratio_RepeatedEmoji	1,980,648.7624 ns	104,239.6803 ns	5,713.7286 ns	1.000	0.00	7.8125	-	160508 B	1.00
TokenSortRatio_EmojiWords	645,709.7184 ns	17,013.9814 ns	932.5937 ns	0.326	0.00	35.1563	12.6953	602665 B	3.75
WRatio_ProseAndOneEmoji	1,078,116.8324 ns	86,061.6873 ns	4,717.3315 ns	0.544	0.00	42.9688	15.6250	720569 B	4.49
Ratio_DistinctAstral	69,700,128.4583 ns	8,420,578.6250 ns	461,560.3303 ns	35.191	0.22	-	-	510798 B	3.18
WRatio_EmptyOperand	0.0532 ns	0.0128 ns	0.0007 ns	0.000	0.00	-	-	-	0.00

Lodestar.Text.Benchmarks.FuzzIncumbentBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Operation	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Allocated	Alloc Ratio
Lodestar	Ratio	97.31 ns	4.357 ns	0.239 ns	1.00	0.00	-	-	NA
FuzzySharp	Ratio	230.88 ns	9.065 ns	0.497 ns	2.37	0.01	0.0048	80 B	NA

Lodestar	PartialRatio	755.87 ns	8.177 ns	0.448 ns	1.00	0.00	-	-	NA
FuzzySharp	PartialRatio	10,045.40 ns	460.976 ns	25.268 ns	13.29	0.03	-	160 B	NA

Lodestar	TokenSetRatio	1,002.23 ns	168.628 ns	9.243 ns	1.00	0.01	0.0534	896 B	1.00
FuzzySharp	TokenSetRatio	2,047.68 ns	64.584 ns	3.540 ns	2.04	0.02	0.1144	1944 B	2.17

Lodestar	WRatio	1,264.62 ns	58.267 ns	3.194 ns	1.00	0.00	0.0668	1120 B	1.00
FuzzySharp	WRatio	4,972.10 ns	124.779 ns	6.840 ns	3.93	0.01	0.1831	3096 B	2.76

Lodestar.Text.Benchmarks.HouseholderQrBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Rows	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
Householder	2000	1.629 ms	0.1298 ms	0.0071 ms	277.3438	259.7656	250.0000	1.38 MB
Householder	20000	17.176 ms	0.6729 ms	0.0369 ms	968.7500	968.7500	968.7500	13.74 MB

Lodestar.Text.Benchmarks.KMeansFitIncumbentBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Shape	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Allocated	Alloc Ratio
Lodestar_Fit	10000x16x16	6,694.2 μs	310.11 μs	17.00 μs	1.00	0.00	7.8125	-	162.63 KB	1.00
NumFlat_Fit	10000x16x16	38,927.1 μs	327.73 μs	17.96 μs	5.82	0.01	-	-	10.39 KB	0.06
MetaNumerics_Fit	10000x16x16	83,365.9 μs	1,512.87 μs	82.93 μs	12.45	0.03	-	-	41.95 KB	0.26

Lodestar_Fit	10000x2x8	681.5 μs	47.62 μs	2.61 μs	1.00	0.00	8.7891	0.9766	156.73 KB	1.00
NumFlat_Fit	10000x2x8	4,776.0 μs	411.42 μs	22.55 μs	7.01	0.04	-	-	3 KB	0.02
MetaNumerics_Fit	10000x2x8	2,975.0 μs	1,187.44 μs	65.09 μs	4.37	0.08	-	-	39.57 KB	0.25

Lodestar_Fit	50000x8x32	34,638.5 μs	882.28 μs	48.36 μs	1.00	0.00	-	-	787.83 KB	1.00
NumFlat_Fit	50000x8x32	442,877.9 μs	9,114.29 μs	499.58 μs	12.79	0.02	-	-	16.29 KB	0.02
MetaNumerics_Fit	50000x8x32	744,138.5 μs	64,945.83 μs	3,559.90 μs	21.48	0.09	-	-	199.91 KB	0.25

Lodestar.Text.Benchmarks.KMeansLloydIncumbentBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Shape	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
Lodestar_Lloyd	10000x16x16	8.700 ms	0.4738 ms	0.0260 ms	1.00	88.7 KB	1.00
NumFlat_Lloyd	10000x16x16	16.654 ms	0.5902 ms	0.0323 ms	1.91	13.6 KB	0.15

Lodestar_Lloyd	10000x2x8	33.714 ms	2.6896 ms	0.1474 ms	1.00	98.96 KB	1.00
NumFlat_Lloyd	10000x2x8	113.216 ms	3.7628 ms	0.2062 ms	3.36	96.48 KB	0.97

Lodestar_Lloyd	50000x8x32	84.362 ms	0.9045 ms	0.0496 ms	1.00	410.3 KB	1.00
NumFlat_Lloyd	50000x8x32	210.771 ms	2.1841 ms	0.1197 ms	2.50	36.47 KB	0.09

Lodestar.Text.Benchmarks.MetricsBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Samples	Classes	Mean	Error	StdDev	Gen0	Allocated
Matrix	1000	2	4,329.6 ns	255.41 ns	14.00 ns	0.0153	376 B
MatrixWeighted	1000	2	7,373.3 ns	146.25 ns	8.02 ns	0.0153	376 B
AccuracyScore	1000	2	124.1 ns	15.01 ns	0.82 ns	-	-
F1Macro	1000	2	4,800.5 ns	204.82 ns	11.23 ns	0.0305	536 B
Report	1000	2	7,321.7 ns	166.30 ns	9.12 ns	0.3128	5344 B
Matrix	1000	10	4,548.7 ns	121.13 ns	6.64 ns	0.0763	1312 B
MatrixWeighted	1000	10	7,539.7 ns	94.42 ns	5.18 ns	0.0763	1312 B
AccuracyScore	1000	10	123.1 ns	0.70 ns	0.04 ns	-	-
F1Macro	1000	10	5,077.5 ns	246.81 ns	13.53 ns	0.0992	1728 B
Report	1000	10	11,280.8 ns	799.07 ns	43.80 ns	0.7324	12336 B
Matrix	100000	2	455,916.5 ns	36,788.57 ns	2,016.51 ns	-	376 B
MatrixWeighted	100000	2	829,761.5 ns	66,490.67 ns	3,644.58 ns	-	377 B
AccuracyScore	100000	2	12,581.0 ns	213.91 ns	11.73 ns	-	-
F1Macro	100000	2	476,629.8 ns	10,264.54 ns	562.63 ns	-	536 B
Report	100000	2	788,692.6 ns	215,962.11 ns	11,837.61 ns	-	5369 B
Matrix	100000	10	494,706.0 ns	27,855.33 ns	1,526.84 ns	-	1313 B
MatrixWeighted	100000	10	964,801.1 ns	41,453.34 ns	2,272.20 ns	-	1313 B
AccuracyScore	100000	10	12,476.2 ns	33.87 ns	1.86 ns	-	-
F1Macro	100000	10	494,029.2 ns	20,615.12 ns	1,129.98 ns	-	1729 B
Report	100000	10	668,103.8 ns	60,598.85 ns	3,321.63 ns	-	12681 B
Matrix	1000000	2	4,605,706.8 ns	60,200.37 ns	3,299.79 ns	-	382 B
MatrixWeighted	1000000	2	8,742,675.0 ns	2,567,259.80 ns	140,720.17 ns	-	388 B
AccuracyScore	1000000	2	129,316.6 ns	130,996.40 ns	7,180.35 ns	-	-
F1Macro	1000000	2	4,622,313.0 ns	375,346.98 ns	20,574.03 ns	-	542 B
Report	1000000	2	4,758,348.2 ns	80,078.39 ns	4,389.37 ns	-	5390 B
Matrix	1000000	10	5,049,928.6 ns	109,322.84 ns	5,992.35 ns	-	1318 B
MatrixWeighted	1000000	10	9,344,970.4 ns	1,827,078.59 ns	100,148.34 ns	-	1324 B
AccuracyScore	1000000	10	134,127.7 ns	9,505.08 ns	521.01 ns	-	-
F1Macro	1000000	10	4,907,366.7 ns	77,970.16 ns	4,273.81 ns	-	1734 B
Report	1000000	10	4,906,796.9 ns	199,902.39 ns	10,957.32 ns	-	12726 B

Lodestar.Text.Benchmarks.MetricsIncumbentBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Samples	Request	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar	100000	Bundle	7,693.74 μs	213.862 μs	11.723 μs	1.00	0.00	-	-	-	1080 B	1.00
MlNet	100000	Bundle	33,996.24 μs	2,498.682 μs	136.961 μs	4.42	0.02	600.0000	600.0000	600.0000	5089266 B	4,712.28

Lodestar	100000	AccuracyAlone	11.63 μs	1.404 μs	0.077 μs	1.00	0.01	-	-	-	-	NA
MlNet	100000	AccuracyAlone	33,665.59 μs	527.956 μs	28.939 μs	2,895.04	16.69	600.0000	600.0000	600.0000	5089801 B	NA

Lodestar	1000000	Bundle	97,177.55 μs	52,520.084 μs	2,878.803 μs	1.00	0.04	-	-	-	-	NA
MlNet	1000000	Bundle	242,539.10 μs	7,104.990 μs	389.448 μs	2.50	0.06	-	-	-	23228104 B	NA

Lodestar	1000000	AccuracyAlone	108.94 μs	4.830 μs	0.265 μs	1.00	0.00	-	-	-	-	NA
MlNet	1000000	AccuracyAlone	221,202.93 μs	13,708.464 μs	751.407 μs	2,030.45	7.34	-	-	-	23232104 B	NA

Lodestar.Text.Benchmarks.MultiClassRocAucBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Samples	Mean	Error	StdDev	Allocated
OneVsRest	100000	38.24 ms	4.834 ms	0.265 ms	-
OneVsOne	100000	84.51 ms	4.855 ms	0.266 ms	801915 B
OneVsRest	1000000	517.91 ms	482.609 ms	26.453 ms	-
OneVsOne	1000000	1,138.77 ms	232.889 ms	12.765 ms	8003648 B

Lodestar.Text.Benchmarks.MultilabelConfusionMatrixBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Rows	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
PerLabel	100000	6.848 ms	0.1289 ms	0.0071 ms	-	-	-	6.28 KB
PerLabelWeighted	100000	11.381 ms	0.3830 ms	0.0210 ms	-	-	-	6.29 KB
PerSample	100000	57.732 ms	28.7069 ms	1.5735 ms	2555.5556	2444.4444	777.7778	31250.62 KB
PerClass	100000	4.508 ms	0.2631 ms	0.0144 ms	-	-	-	6.43 KB

Lodestar.Text.Benchmarks.PartialRatioLongNeedleBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	NeedleLength	Mean	Error	StdDev	Allocated
Embedded	65	10.841 μs	3.6131 μs	0.1980 μs	-
EqualLength	65	2.929 μs	0.1979 μs	0.0108 μs	-
Embedded	128	39.520 μs	0.4789 μs	0.0263 μs	-
EqualLength	128	5.364 μs	0.3784 μs	0.0207 μs	-
Embedded	512	1,320.029 μs	73.4407 μs	4.0255 μs	-
EqualLength	512	43.714 μs	0.2630 μs	0.0144 μs	-

Lodestar.Text.Benchmarks.PartitionValidityBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Samples	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
DaviesBouldinScore	1000000	10.36 ms	1.496 ms	0.082 ms	140.6250	140.6250	140.6250	3.82 MB
CalinskiHarabaszScore	1000000	15.19 ms	0.929 ms	0.051 ms	140.6250	140.6250	140.6250	3.82 MB

Lodestar.Text.Benchmarks.PersistenceBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
VocabTxt	4.220 ms	3.5739 ms	0.1959 ms	109.3750	101.5625	31.2500	3711.63 KB
TokenizerJsonWordPiece	11.349 ms	2.7838 ms	0.1526 ms	187.5000	171.8750	46.8750	5852.4 KB
TokenizerJsonUnigram	12.636 ms	0.6209 ms	0.0340 ms	78.1250	62.5000	15.6250	4748.81 KB
SpieceModel	3.924 ms	2.3960 ms	0.1313 ms	109.3750	101.5625	31.2500	3440.05 KB
TfidfSave	1.647 ms	0.1075 ms	0.0059 ms	21.4844	15.6250	15.6250	2137.04 KB
TfidfLoad	4.284 ms	0.2308 ms	0.0126 ms	85.9375	78.1250	23.4375	2930.51 KB
EmbeddingIndexSave	4.188 ms	0.6383 ms	0.0350 ms	187.5000	187.5000	187.5000	20349.83 KB
EmbeddingIndexLoad	5.320 ms	0.4612 ms	0.0253 ms	179.6875	148.4375	117.1875	16094.37 KB
EmbeddingIndexSaveFile	49.814 ms	10.8134 ms	0.5927 ms	-	-	-	323.49 KB
EmbeddingIndexLoadGzip	73.928 ms	51.1662 ms	2.8046 ms	-	-	-	16095.21 KB

Lodestar.Text.Benchmarks.PrecompiledNormalizerBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Mean	Error	StdDev	Gen0	Allocated
NormalizeDocuments	13.50 ms	0.434 ms	0.024 ms	171.8750	2.75 MB
EncodeDocuments	58.44 ms	1.662 ms	0.091 ms	444.4444	8.19 MB

Lodestar.Text.Benchmarks.PrincipalComponentVarianceBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Shape	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar_ExplainedVariance	100x200	9,433.49 μs	156.477 μs	8.577 μs	1.00	0.00	46.8750	46.8750	46.8750	318.28 KB	1.00
NumFlat_Pca	100x200	15,308.67 μs	227.555 μs	12.473 μs	1.62	0.00	187.5000	187.5000	187.5000	628.54 KB	1.97

Lodestar_ExplainedVariance	2000x10	130.51 μs	16.065 μs	0.881 μs	1.00	0.01	-	-	-	2.38 KB	1.00
NumFlat_Pca	2000x10	182.06 μs	10.040 μs	0.550 μs	1.40	0.01	-	-	-	1.94 KB	0.82

Lodestar_ExplainedVariance	2000x50	2,788.57 μs	128.525 μs	7.045 μs	1.00	0.00	-	-	-	42.07 KB	1.00
NumFlat_Pca	2000x50	2,586.14 μs	84.009 μs	4.605 μs	0.93	0.00	-	-	-	40.07 KB	0.95

Lodestar_ExplainedVariance	200x10	21.93 μs	5.153 μs	0.282 μs	1.00	0.02	0.1221	-	-	2.38 KB	1.00
NumFlat_Pca	200x10	25.62 μs	2.230 μs	0.122 μs	1.17	0.01	0.0916	-	-	1.94 KB	0.82

Lodestar.Text.Benchmarks.ProcessExtractBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Limit	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
Extract	1	6.874 ms	0.1145 ms	0.0063 ms	1.00	118 B	1.00
ExtractOne	1	6.718 ms	0.0985 ms	0.0054 ms	0.98	-	0.00

Extract	5	6.903 ms	0.3371 ms	0.0185 ms	1.00	214 B	1.00
ExtractOne	5	6.782 ms	0.6415 ms	0.0352 ms	0.98	-	0.00

Lodestar.Text.Benchmarks.QgramBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Q	Mean	Error	StdDev	Gen0	Allocated
JaccardSimilarity	1	4.775 μs	0.9961 μs	0.0546 μs	0.0076	192 B
JaccardSimilarity	3	5.847 μs	0.8667 μs	0.0475 μs	0.0076	192 B

Lodestar.Text.Benchmarks.RankingMetricsBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Rows	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
NdcgTieAveraged	100000	37.273 ms	1.8178 ms	0.0996 ms	-	-	-	800349 B
NdcgIgnoringTies	100000	34.279 ms	0.3214 ms	0.0176 ms	-	-	-	800345 B
DcgTieAveraged	100000	21.815 ms	0.5919 ms	0.0324 ms	-	-	-	800215 B
ReciprocalRankScore	100000	17.982 ms	0.0847 ms	0.0046 ms	-	-	-	-
CoverageErrorScore	100000	5.869 ms	0.6993 ms	0.0383 ms	15.6250	15.6250	15.6250	800271 B
LabelRankingAveragePrecisionScore	100000	53.006 ms	0.3355 ms	0.0184 ms	-	-	-	-

Lodestar.Text.Benchmarks.RegressionMetricsBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Samples	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
Mse	100000	42.28 μs	15.929 μs	0.873 μs	-	-	-	-
Mae	100000	42.39 μs	5.462 μs	0.299 μs	-	-	-	-
R2Score	100000	110.65 μs	4.118 μs	0.226 μs	-	-	-	64 B
MedianAe	100000	538.63 μs	227.934 μs	12.494 μs	199.2188	199.2188	199.2188	800186 B
Mse	1000000	420.34 μs	229.351 μs	12.572 μs	-	-	-	-
Mae	1000000	420.52 μs	56.057 μs	3.073 μs	-	-	-	-
R2Score	1000000	1,100.27 μs	34.056 μs	1.867 μs	-	-	-	65 B
MedianAe	1000000	5,908.81 μs	482.879 μs	26.468 μs	320.3125	320.3125	320.3125	8000269 B

Lodestar.Text.Benchmarks.RobustScalerSparseBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Shape	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
Fit	2000x2000x20	23.82 ms	0.248 ms	0.014 ms	93.7500	93.7500	93.7500	375.3 KB
Fit	500x20000x5	43.90 ms	3.578 ms	0.196 ms	-	-	-	492.47 KB

Lodestar.Text.Benchmarks.SentencePieceBpeLineageBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Model	Mean	Error	StdDev	Gen0	Allocated
EncodeDocuments	Llama2	138.6 ms	15.44 ms	0.85 ms	2250.0000	37.05 MB
EncodeDocuments	Mistral	133.2 ms	0.78 ms	0.04 ms	2250.0000	37.03 MB

Lodestar.Text.Benchmarks.SilhouetteBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Samples	Mean	Error	StdDev	Gen0	Gen1	Gen2	Allocated
PerSample	2000	27.19 ms	0.298 ms	0.016 ms	31.2500	31.2500	31.2500	148.75 KB
PerSample	5000	182.69 ms	0.769 ms	0.042 ms	-	-	-	371.6 KB

Lodestar.Text.Benchmarks.StopWordBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Documents	Mean	Error	StdDev	Ratio	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Count	200	4.768 ms	0.2163 ms	0.0119 ms	1.00	109.3750	109.3750	109.3750	1366.79 KB	1.00
CountWithStopWords	200	4.300 ms	0.2091 ms	0.0115 ms	0.90	39.0625	15.6250	-	725.83 KB	0.53
Hashing	200	4.924 ms	0.6309 ms	0.0346 ms	1.03	70.3125	70.3125	70.3125	1051.16 KB	0.77
HashingWithStopWords	200	4.534 ms	0.4949 ms	0.0271 ms	0.95	31.2500	15.6250	-	591.9 KB	0.43

Count	1000	15.742 ms	0.5885 ms	0.0323 ms	1.00	843.7500	843.7500	843.7500	5880.52 KB	1.00
CountWithStopWords	1000	14.059 ms	2.0602 ms	0.1129 ms	0.89	375.0000	375.0000	375.0000	3155.02 KB	0.54
Hashing	1000	16.548 ms	1.2923 ms	0.0708 ms	1.05	656.2500	562.5000	562.5000	4793.67 KB	0.82
HashingWithStopWords	1000	15.363 ms	0.4196 ms	0.0230 ms	0.98	265.6250	265.6250	265.6250	2633.78 KB	0.45

Lodestar.Text.Benchmarks.TextRankBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Words	Mean	Error	StdDev	Gen0	Gen1	Allocated
Extract	2000	14.20 ms	1.600 ms	0.088 ms	187.5000	125.0000	3.13 MB
Extract	8000	324.28 ms	22.851 ms	1.253 ms	2000.0000	1000.0000	35.34 MB

Lodestar.Text.Benchmarks.TokenizerIncumbentBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Model	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Allocated	Alloc Ratio
Lodestar	WordPiece	20.69 ms	0.878 ms	0.048 ms	1.00	0.00	531.2500	8.71 MB	1.00
MlTokenizers	WordPiece	53.82 ms	3.554 ms	0.195 ms	2.60	0.01	200.0000	3.55 MB	0.41

Lodestar	SentencePiece	44.32 ms	4.541 ms	0.249 ms	1.00	0.01	333.3333	5.44 MB	1.00
MlTokenizers	SentencePiece	54.54 ms	0.953 ms	0.052 ms	1.23	0.01	100.0000	3.09 MB	0.57

Lodestar	ByteLevelBpe	81.71 ms	27.770 ms	1.522 ms	1.00	0.02	1714.2857	28.47 MB	1.00
MlTokenizers	ByteLevelBpe	237.16 ms	1.392 ms	0.076 ms	2.90	0.05	3666.6667	59.08 MB	2.08

Lodestar.Text.Benchmarks.TopKAccuracyBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Classes	Mean	Error	StdDev	Allocated
TopTwo	10	12.24 ms	0.344 ms	0.019 ms	-
TopTwo	100	83.98 ms	6.346 ms	0.348 ms	-

Lodestar.Text.Benchmarks.VectorMathBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Dim	Mean	Error	StdDev	Ratio	Allocated	Alloc Ratio
Dot	384	131.70 ns	3.664 ns	0.201 ns	1.00	-	NA
L2Norm	384	47.67 ns	0.361 ns	0.020 ns	0.36	-	NA

Dot	768	95.17 ns	4.734 ns	0.259 ns	1.00	-	NA
L2Norm	768	92.48 ns	0.652 ns	0.036 ns	0.97	-	NA

Dot	1024	124.20 ns	5.900 ns	0.323 ns	1.00	-	NA
L2Norm	1024	122.67 ns	1.125 ns	0.062 ns	0.99	-	NA

Lodestar.Text.Benchmarks.VectorizerBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Documents	Mean	Error	StdDev	Ratio	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Count	200	2.312 ms	0.3034 ms	0.0166 ms	1.00	15.6250	11.7188	-	303 KB	1.00
Tfidf	200	2.342 ms	0.1644 ms	0.0090 ms	1.01	19.5313	15.6250	-	332.14 KB	1.10
CountBigrams	200	2.760 ms	0.2236 ms	0.0123 ms	1.19	31.2500	15.6250	-	590.34 KB	1.95
CountCharWordBoundary	200	2.122 ms	0.0498 ms	0.0027 ms	0.92	382.8125	382.8125	382.8125	1685.7 KB	5.56
Hashing	200	2.323 ms	0.1848 ms	0.0101 ms	1.01	11.7188	7.8125	-	233.65 KB	0.77

Count	1000	4.444 ms	0.1809 ms	0.0099 ms	1.00	109.3750	109.3750	109.3750	1280.03 KB	1.00
Tfidf	1000	4.575 ms	0.1523 ms	0.0084 ms	1.03	140.6250	140.6250	140.6250	1424.43 KB	1.11
CountBigrams	1000	6.402 ms	0.3072 ms	0.0168 ms	1.44	390.6250	390.6250	390.6250	2291.9 KB	1.79
CountCharWordBoundary	1000	12.907 ms	0.6044 ms	0.0331 ms	2.90	1046.8750	1015.6250	1015.6250	12022.2 KB	9.39
Hashing	1000	4.475 ms	0.3772 ms	0.0207 ms	1.01	70.3125	70.3125	70.3125	1015.48 KB	0.79

Lodestar.Text.Benchmarks.VectorizerIncumbentBenchmarks-report-github

BenchmarkDotNet v0.14.0, Ubuntu 24.04.5 LTS (Noble Numbat)
AMD EPYC 7763, 1 CPU, 4 logical and 2 physical cores
.NET SDK 10.0.401
  [Host]   : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2
  ShortRun : .NET 10.0.12 (10.0.1226.42308), X64 RyuJIT AVX2

Job=ShortRun  IterationCount=3  LaunchCount=1  
WarmupCount=3

Method	Documents	Mean	Error	StdDev	Ratio	RatioSD	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
Lodestar	200	5.124 ms	0.5411 ms	0.0297 ms	1.00	0.01	148.4375	148.4375	148.4375	2.01 MB	1.00
MlNet	200	51.242 ms	108.1049 ms	5.9256 ms	10.00	1.00	7500.0000	7500.0000	7500.0000	28.27 MB	14.09

Lodestar	1000	18.880 ms	3.0978 ms	0.1698 ms	1.00	0.01	1500.0000	1500.0000	1500.0000	9.17 MB	1.00
MlNet	1000	405.115 ms	771.1246 ms	42.2679 ms	21.46	1.95	76000.0000	76000.0000	76000.0000	324.35 MB	35.36

Against rapidfuzz, in this same run

Both sides on this VM in these minutes, which is what makes the ratio readable where the absolutes are not.

metrics
persistence
stats
cdist
nmf-transform
glm
var
ols

compare-cdist

Python: {'rapidfuzz': '3.14.6', 'numpy': '2.5.3'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.12

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu
cdist_ratio_n50	0.157	0.036	0.23x	0.157	0.036	0.23x
cdist_wratio_n50	2.227	2.504	1.12x	2.226	2.504	1.12x
cdist_ratio_varied_n50	0.130	0.031	0.24x	0.130	0.031	0.24x
cdist_ratio_varied_cutoff90_n50	0.054	0.031	0.58x	0.054	0.031	0.58x
cdist_ratio_n200	2.706	0.440	0.16x	2.709	0.440	0.16x
cdist_wratio_n200	33.918	37.935	1.12x	33.919	37.933	1.12x
cdist_ratio_varied_n200	2.297	0.385	0.17x	2.300	0.385	0.17x
cdist_ratio_varied_cutoff90_n200	0.981	0.385	0.39x	0.984	0.385	0.39x
cdist_ratio_n500	16.390	2.655	0.16x	16.407	2.655	0.16x
cdist_wratio_n500	207.518	229.402	1.11x	207.648	229.388	1.10x
cdist_ratio_varied_n500	13.597	2.294	0.17x	13.707	2.294	0.17x
cdist_ratio_varied_cutoff90_n500	5.501	2.294	0.42x	5.600	2.294	0.41x

ratio > 1 means Lodestar is faster. cpu is the honest one on both sides: elapsed time hides .NET's background GC threads, and numpy's BLAS threads too -- statsmodels at 100 000 rows spends ~10x more cpu than wall.

compare-glm

Python: {'scipy': '1.18.1', 'statsmodels': '0.15.0', 'numpy': '2.5.3'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.12

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu
glm_negative_binomial_n1000	0.398	3.348	8.41x	0.398	3.348	8.41x
glm_gamma_n1000	0.489	4.158	8.51x	0.488	4.158	8.51x
glm_poisson_exposure_n1000	0.397	3.926	9.88x	0.397	3.926	9.88x
mnlogit_n1000	0.842	18.061	21.45x	0.842	18.059	21.45x
glm_negative_binomial_n10000	3.604	13.770	3.82x	3.611	13.769	3.81x
glm_gamma_n10000	4.144	12.954	3.13x	4.153	12.953	3.12x
glm_poisson_exposure_n10000	4.333	12.971	2.99x	4.340	12.969	2.99x
mnlogit_n10000	8.648	96.620	11.17x	8.656	96.613	11.16x
glm_negative_binomial_n100000	37.412	119.356	3.19x	37.705	473.135	12.55x
glm_gamma_n100000	43.372	134.259	3.10x	43.629	535.959	12.28x
glm_poisson_exposure_n100000	55.859	131.451	2.35x	56.263	524.135	9.32x
mnlogit_n100000	86.685	922.316	10.64x	87.131	1654.580	18.99x

compare-metrics

Python: {'scikit-learn': '1.9.1', 'numpy': '2.5.3'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.12

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu
confusion_matrix_n1000_k2	0.006	0.976	153.35x	0.006	0.976	153.35x
accuracy_n1000_k2	0.000	0.514	4174.53x	0.000	0.514	4173.52x
precision_recall_f1_macro_n1000_k2	0.005	1.771	367.95x	0.005	1.771	367.91x
classification_report_n1000_k2	0.007	6.829	951.15x	0.007	6.828	951.14x
roc_auc_binary_n1000_k2	0.016	1.894	122.05x	0.016	1.894	122.05x
balanced_accuracy_n1000_k2	0.005	1.036	211.88x	0.005	1.036	211.87x
matthews_n1000_k2	0.005	1.985	412.84x	0.005	1.984	412.83x
cohen_kappa_n1000_k2	0.005	1.081	234.26x	0.005	1.081	234.26x
mse_n1000_k2	0.000	0.300	615.51x	0.000	0.300	615.53x
mae_n1000_k2	0.000	0.298	615.30x	0.000	0.298	615.24x
median_ae_n1000_k2	0.005	0.312	64.76x	0.005	0.311	64.75x
r2_n1000_k2	0.001	0.362	304.05x	0.001	0.362	304.02x
confusion_matrix_n1000_k10	0.007	0.976	145.10x	0.007	0.976	145.10x
accuracy_n1000_k10	0.000	0.516	4278.85x	0.000	0.516	4278.93x
precision_recall_f1_macro_n1000_k10	0.006	1.793	319.68x	0.006	1.793	319.68x
classification_report_n1000_k10	0.011	6.943	617.50x	0.011	6.943	617.45x
roc_auc_ovr_macro_n1000_k10	0.523	9.824	18.80x	0.523	9.824	18.80x
balanced_accuracy_n1000_k10	0.006	1.046	180.91x	0.006	1.046	180.89x
matthews_n1000_k10	0.006	2.016	357.24x	0.006	2.016	357.26x
cohen_kappa_n1000_k10	0.006	1.093	178.89x	0.006	1.093	178.86x
mse_n1000_k10	0.000	0.300	611.84x	0.000	0.300	611.79x
mae_n1000_k10	0.000	0.301	617.73x	0.000	0.301	617.72x
median_ae_n1000_k10	0.005	0.312	64.79x	0.005	0.312	64.79x
r2_n1000_k10	0.001	0.363	304.37x	0.001	0.363	304.36x
confusion_matrix_n100000_k2	0.626	10.678	17.06x	0.626	10.677	17.06x
accuracy_n100000_k2	0.011	3.733	334.76x	0.011	3.733	334.75x
precision_recall_f1_macro_n100000_k2	0.475	12.259	25.80x	0.475	12.258	25.81x
classification_report_n100000_k2	0.478	26.935	56.34x	0.478	26.933	56.34x
roc_auc_binary_n100000_k2	3.473	26.341	7.58x	3.473	26.336	7.58x
balanced_accuracy_n100000_k2	0.475	10.771	22.69x	0.475	10.770	22.69x
matthews_n100000_k2	0.475	21.466	45.20x	0.475	21.465	45.20x
cohen_kappa_n100000_k2	0.476	10.808	22.73x	0.475	10.807	22.73x
mse_n100000_k2	0.045	0.461	10.19x	0.045	0.461	10.19x
mae_n100000_k2	0.045	0.453	10.11x	0.045	0.453	10.11x
median_ae_n100000_k2	0.571	1.788	3.13x	0.588	1.788	3.04x
r2_n100000_k2	0.110	0.658	5.97x	0.110	0.658	5.97x
confusion_matrix_n100000_k10	0.508	10.675	21.03x	0.508	10.673	21.03x
accuracy_n100000_k10	0.012	3.744	315.79x	0.012	3.744	315.74x
precision_recall_f1_macro_n100000_k10	0.508	12.899	25.39x	0.508	12.898	25.39x
classification_report_n100000_k10	0.514	29.469	57.30x	0.514	29.467	57.30x
roc_auc_ovr_macro_n100000_k10	34.820	211.441	6.07x	34.824	211.424	6.07x
balanced_accuracy_n100000_k10	0.508	10.749	21.17x	0.508	10.749	21.17x
matthews_n100000_k10	0.508	22.127	43.60x	0.507	22.125	43.60x
cohen_kappa_n100000_k10	0.708	10.784	15.23x	0.708	10.783	15.23x
mse_n100000_k10	0.045	0.438	9.69x	0.045	0.438	9.69x
mae_n100000_k10	0.045	0.429	9.55x	0.045	0.429	9.55x
median_ae_n100000_k10	0.644	1.778	2.76x	0.699	1.778	2.54x
r2_n100000_k10	0.111	0.646	5.81x	0.111	0.646	5.81x
confusion_matrix_n1000000_k2	4.847	99.046	20.43x	4.845	99.040	20.44x
accuracy_n1000000_k2	0.118	32.679	277.09x	0.118	32.677	277.11x
precision_recall_f1_macro_n1000000_k2	4.773	107.469	22.52x	4.773	107.455	22.51x
classification_report_n1000000_k2	4.926	209.219	42.47x	4.925	209.185	42.47x
roc_auc_binary_n1000000_k2	67.988	288.337	4.24x	67.988	288.312	4.24x
balanced_accuracy_n1000000_k2	4.876	99.072	20.32x	4.875	99.047	20.32x
matthews_n1000000_k2	4.818	200.454	41.60x	4.818	200.420	41.60x
cohen_kappa_n1000000_k2	4.922	99.206	20.16x	4.921	99.196	20.16x
mse_n1000000_k2	0.496	2.110	4.25x	0.496	2.109	4.26x
mae_n1000000_k2	0.524	2.083	3.98x	0.524	2.083	3.98x
median_ae_n1000000_k2	5.611	14.060	2.51x	5.668	14.058	2.48x
r2_n1000000_k2	1.102	3.508	3.18x	1.102	3.507	3.18x
confusion_matrix_n1000000_k10	5.055	99.231	19.63x	5.055	99.179	19.62x
accuracy_n1000000_k10	0.123	32.815	266.47x	0.123	32.799	266.39x
precision_recall_f1_macro_n1000000_k10	5.153	113.468	22.02x	5.153	113.455	22.02x
classification_report_n1000000_k10	5.130	233.791	45.57x	5.129	233.764	45.58x
balanced_accuracy_n1000000_k10	5.058	99.363	19.64x	5.057	99.355	19.65x
matthews_n1000000_k10	5.136	207.280	40.35x	5.136	207.258	40.36x
cohen_kappa_n1000000_k10	5.151	99.451	19.31x	5.150	99.436	19.31x
mse_n1000000_k10	0.490	2.222	4.53x	0.490	2.221	4.53x
mae_n1000000_k10	0.493	2.199	4.46x	0.493	2.199	4.46x
median_ae_n1000000_k10	5.408	14.131	2.61x	5.449	14.129	2.59x
r2_n1000000_k10	1.268	3.922	3.09x	1.268	3.921	3.09x

ratio > 1 means Lodestar is faster. cpu is the merge gate for this branch (docs/guides/performance.md): every operation, every size, must be >= 1x.

compare-nmf-transform

Python: {'scikit-learn': '1.9.1', 'scipy': '1.18.1'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.12

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu
transform_frobenius_n100	2.243	0.851	0.38x	2.250	0.851	0.38x
transform_frobenius_n500	6.799	1.636	0.24x	6.799	1.635	0.24x
transform_frobenius_n2000	24.095	5.824	0.24x	24.131	23.241	0.96x
transform_kl_n100	1.254	17.613	14.05x	1.253	17.611	14.05x
transform_kl_n500	5.701	34.224	6.00x	5.701	34.222	6.00x
transform_kl_n2000	22.597	122.088	5.40x	22.647	450.930	19.91x

compare-ols

Python: {'scipy': '1.18.1', 'statsmodels': '0.15.0', 'numpy': '2.5.3'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.12

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu
ols_summary_n1000	0.045	2.982	65.80x	0.045	2.981	65.81x
ols_hac_n1000	0.167	3.468	20.82x	0.167	3.468	20.82x
ols_cluster_n1000	0.060	3.514	59.03x	0.060	3.513	59.03x
ols_summary_n10000	0.436	11.433	26.22x	0.436	11.432	26.22x
ols_hac_n10000	1.680	12.677	7.55x	1.688	12.676	7.51x
ols_cluster_n10000	0.599	12.972	21.67x	0.602	12.971	21.54x
ols_summary_n100000	4.264	108.973	25.56x	4.285	434.785	101.48x
ols_hac_n100000	16.699	119.690	7.17x	17.017	477.666	28.07x
ols_cluster_n100000	6.521	124.379	19.07x	6.724	496.598	73.86x

compare-persistence

Python: {'tokenizers': '0.23.2', 'sentencepiece': '0.2.2', 'scikit-learn': '1.9.1', 'numpy': '2.5.3'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.12

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu	C# bytes	Py bytes
vocab_txt	5.269	10.564	2.00x	5.583	10.563	1.89x	228,891	228,891
tokenizer_json_wordpiece	13.430	29.467	2.19x	13.818	29.465	2.13x	706,526	706,526
tokenizer_json_unigram	15.691	66.829	4.26x	16.039	66.819	4.17x	1,990,038	1,990,038
spiece_model	5.344	36.761	6.88x	5.590	36.758	6.58x	533,084	533,084
tfidf_save	1.964	3.203	1.63x	1.992	3.202	1.61x	581,787	591,922
tfidf_load	4.977	5.114	1.03x	5.296	5.113	0.97x	581,787	591,922
embedding_index_save	4.431	3.865	0.87x	4.669	3.865	0.83x	20,589,007	15,360,128
embedding_index_save_file	49.465	36.712	0.74x	10.157	6.334	0.62x	20,589,007	15,360,128
embedding_index_load	5.968	1.542	0.26x	6.305	1.542	0.24x	20,589,007	15,360,128
embedding_index_load_file	6.436	1.100	0.17x	6.796	1.096	0.16x	20,589,007	15,360,128
embedding_index_load_memory	4.799	1.590	0.33x	5.201	1.589	0.31x	20,589,007	15,360,128
embedding_index_ingest_npy	1.565	1.487	0.95x	1.869	1.484	0.79x	15,360,128	15,360,128
embedding_index_view_floor	0.000	0.001	104.15x	0.000	0.001	104.14x	20,589,007	15,360,128
embedding_index_save_gzip	414.304	557.609	1.35x	414.287	557.514	1.35x	15,250,490	14,022,374
embedding_index_load_gzip	71.110	66.386	0.93x	71.492	66.383	0.93x	15,250,490	14,022,374

ratio > 1 means Lodestar is faster. cpu is the honest one: elapsed time hides work .NET does on background GC threads; CPython is single-threaded. bytes is what the row wrote or read; a results file from before #378 carries none, and the two columns then disappear rather than read zero.

compare-stats

Python: {'scipy': '1.18.1', 'statsmodels': '0.15.0', 'numpy': '2.5.3'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.12

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu
welch_t_n1000	0.004	0.794	197.62x	0.004	0.793	197.61x
mann_whitney_n1000	0.029	0.735	25.36x	0.029	0.735	25.35x
chi_square_n1000	0.000	0.335	1599.72x	0.000	0.335	1599.68x
welch_t_n10000	0.038	0.850	22.61x	0.038	0.850	22.61x
mann_whitney_n10000	0.756	2.805	3.71x	0.756	2.805	3.71x
chi_square_n10000	0.001	0.335	256.69x	0.001	0.335	256.71x
welch_t_n100000	0.375	1.224	3.26x	0.375	1.224	3.26x
mann_whitney_n100000	8.045	27.818	3.46x	8.045	27.815	3.46x
chi_square_n100000	0.014	0.358	26.05x	0.014	0.358	26.05x

compare-var

Python: {'scipy': '1.18.1', 'statsmodels': '0.15.0', 'numpy': '2.5.3'} (py 3.12.14)
C#:     Lodestar on .NET 10.0.12

operation	C# ms	Py ms	wall	C# cpu	Py cpu	cpu
var_n1000	0.039	2.677	69.51x	0.039	2.677	69.51x
var_n10000	0.470	16.511	35.11x	0.473	16.510	34.88x
var_n100000	3.825	196.483	51.37x	3.991	420.500	105.36x

Selected, and not reached tonight

The run stops starting classes when the budget left cannot hold one, so these were carried to the next run rather than measured badly or killed mid-flight. They are selected again tomorrow whether or not anything else changes.

BitParallelEditDistanceBenchmarks
ChainedProductBenchmarks
CoxBenchmarks
SurvivalBenchmarks
TiledCosineTopKBenchmarks
TiledMinHashSignaturesBenchmarks
TiledSparseDenseProductBenchmarks

Ratios that moved

Read against bench/nightly/ratios.csv by tools/nightly_series.py, under the thresholds that script sets. A ratio moves when either side of it does: read its baseline before calling a movement a regression.

New tonight: 11. Each of these had not moved on its previous reading.

class	parameters	method	kind	tonight	against	change	threshold
`Bm25Benchmarks`	Documents=20000	`LuceneQuery`	drift	0.78	0.01	+7300%	20%
`Bm25Benchmarks`	Documents=20000	`LuceneFromText`	drift	5789.4	106.7	+5326%	20%
`Bm25Benchmarks`	Documents=20000	`LodestarFromText`	drift	4886.85	207.6	+2202%	20%
`Bm25Benchmarks`	Documents=1000	`LuceneQuery`	drift	1.71	0.12	+1325%	20%
`Bm25Benchmarks`	Documents=1000	`LuceneFromText`	drift	3569.58	257.06	+1289%	20%
`Bm25Benchmarks`	Documents=1000	`LodestarFromText`	drift	3973.42	505.13	+715%	20%
`MetaNumericsStatsBenchmarks`	SampleSize=10000	`MetaNumerics_Wilcoxon`	step	10.139	5.026	+102%	30%
`MetaNumericsStatsBenchmarks`	SampleSize=10000	`Lodestar_MannWhitney`	step	19.079	10.69	+78%	30%
`VectorMathBenchmarks`	Dim=384	`L2Norm`	step	0.36	0.94	-62%	30%
`MetaNumericsStatsBenchmarks`	SampleSize=10000	`Lodestar_KruskalWallis`	step	39.212	27.952	+40%	30%
`StationarityBenchmarks`	SampleSize=200	`CortexAugmentedDickeyFuller`	step	3.36	4.98	-33%	30%

Still away from their median: 20. These moved on an earlier run, and their median has not caught up yet.

class	parameters	method	kind	tonight	against	change	threshold
`MetricsIncumbentBenchmarks`	Samples=1000000, Request=AccuracyAlone	`MlNet`	drift	2030.45	85.09	+2286%	20%
`MetricsIncumbentBenchmarks`	Samples=100000, Request=AccuracyAlone	`MlNet`	drift	2895.04	133.83	+2063%	20%
`FuzzIncumbentBenchmarks`	Operation=PartialRatio	`FuzzySharp`	drift	13.29	0.99	+1242%	20%
`TokenizerIncumbentBenchmarks`	Model=SentencePiece	`MlTokenizers`	drift	1.23	0.16	+650%	20%
`TokenizerIncumbentBenchmarks`	Model=WordPiece	`MlTokenizers`	drift	2.6	0.87	+199%	20%
`FuzzBenchmarks`	—	`PartialRatio`	drift	7.84	115.27	-94%	20%
`FuzzIncumbentBenchmarks`	Operation=WRatio	`FuzzySharp`	drift	3.93	2.13	+93%	20%
`StatsBenchmarks`	SampleSize=10000	`LodestarMannWhitney`	step	18.784	10.939	+72%	30%
`VectorizerIncumbentBenchmarks`	Documents=1000	`MlNet`	drift	21.46	12.89	+66%	20%
`BpeBenchmarks`	—	`Bpe`	drift	2.69	1.74	+59%	20%
`DecompositionBenchmarks`	—	`MlNet_ProjectToPrincipalComponents_Rank20`	drift	1.61	0.85	+59%	20%
`FuzzBenchmarks`	—	`WRatio`	drift	12.29	23.14	-47%	20%
`BatchEmbeddingBenchmarks`	CorpusSize=128	`EmbedBatchBucketed`	drift	0.3	0.5	-40%	20%
`VectorizerIncumbentBenchmarks`	Documents=200	`MlNet`	drift	10	7.26	+38%	20%
`BatchEmbeddingBenchmarks`	CorpusSize=32	`EmbedBatchBucketed`	drift	0.33	0.52	-37%	20%
`BatchEmbeddingBenchmarks`	CorpusSize=32	`EmbedBatch`	drift	0.35	0.55	-36%	20%
`BatchEmbeddingBenchmarks`	CorpusSize=128	`EmbedBatch`	drift	0.35	0.53	-34%	20%
`BatchEmbeddingBenchmarks`	CorpusSize=8	`EmbedBatch`	drift	0.39	0.57	-32%	20%
`BatchEmbeddingBenchmarks`	CorpusSize=8	`EmbedBatchBucketed`	drift	0.39	0.57	-32%	20%
`FuzzBenchmarks`	—	`TokenSortRatio`	drift	8.39	10.79	-22%	20%