diff --git a/sample/greedy.go b/sample/greedy.go
index 206f5544d..4d110f021 100644
--- a/sample/greedy.go
+++ b/sample/greedy.go
@@ -8,6 +8,6 @@ func Greedy() Sampler {
 	return greedy{}
 }
 
-func (s greedy) Sample(t []float64) ([]float64, error) {
-	return []float64{float64(floats.MaxIdx(t))}, nil
+func (s greedy) Sample(t []float64) (int, error) {
+	return floats.MaxIdx(t), nil
 }
diff --git a/sample/sample.go b/sample/sample.go
index 714d994de..d873e7cee 100644
--- a/sample/sample.go
+++ b/sample/sample.go
@@ -10,13 +10,30 @@ import (
 	"gonum.org/v1/gonum/stat/sampleuv"
 )
 
+type Transform interface {
+	Apply([]float64) ([]float64, error)
+}
+
 type Sampler interface {
-	Sample([]float64) ([]float64, error)
+	Sample([]float64) (int, error)
+}
+
+type SamplerConfig struct {
+	transforms []Transform
+	sampler    Sampler
+}
+
+// NewSampler creates a sampler with the given transforms and sampling method
+func NewSampler(transforms []Transform, sampler Sampler) *SamplerConfig {
+	return &SamplerConfig{
+		transforms: transforms,
+		sampler:    sampler,
+	}
 }
 
 type Temperature float64
 
-func (t Temperature) Sample(logits []float64) ([]float64, error) {
+func (t Temperature) Apply(logits []float64) ([]float64, error) {
 	if t < 0 || t > 2 {
 		return nil, errors.New("temperature must be between 0 and 2")
 	}
@@ -34,15 +51,16 @@ func (t Temperature) Sample(logits []float64) ([]float64, error) {
 
 type softmax struct{}
 
-func Softmax() Sampler {
+func Softmax() Transform {
 	return softmax{}
 }
 
-func (softmax) Sample(logits []float64) ([]float64, error) {
-	return computeSoftmax(logits)
+func (softmax) Apply(logits []float64) ([]float64, error) {
+	return computeSoftmax(logits), nil
 }
 
-func computeSoftmax(logits []float64) ([]float64, error) {
+// TODO: cache softmax values
+func computeSoftmax(logits []float64) []float64 {
 	copiedLogits := make([]float64, len(logits))
 	copy(copiedLogits, logits)
 	for i := range copiedLogits {
@@ -52,12 +70,12 @@ func computeSoftmax(logits []float64) ([]float64, error) {
 	floatSum := floats.Sum(copiedLogits)
 	floats.Scale(1.0/floatSum, copiedLogits)
 
-	return copiedLogits, nil
+	return copiedLogits
 }
 
 type TopK int
 
-func (k TopK) Sample(logits []float64) ([]float64, error) {
+func (k TopK) Apply(logits []float64) ([]float64, error) {
 	if k <= 0 {
 		return nil, errors.New("k must be positive")
 	}
@@ -76,23 +94,20 @@ func (k TopK) Sample(logits []float64) ([]float64, error) {
 	})
 
 	for _, idx := range indices[k:] {
-		logits[idx] = math.NaN()
+		logits[idx] = math.Inf(-1)
 	}
 
 	return logits, nil
 }
 
-type TopP float32
+type TopP float64
 
-func (p TopP) Sample(logits []float64) ([]float64, error) {
+func (p TopP) Apply(logits []float64) ([]float64, error) {
 	if p <= 0 || p >= 1 {
 		return nil, errors.New("p must be between 0 and 1")
 	}
 
-	probs, err := computeSoftmax(logits)
-	if err != nil {
-		return nil, err
-	}
+	probs := computeSoftmax(logits)
 
 	indices := make([]int, len(probs))
 	for i := range indices {
@@ -104,12 +119,12 @@ func (p TopP) Sample(logits []float64) ([]float64, error) {
 		return cmp.Compare(probs[j], probs[i])
 	})
 
-	cumSum := 0.0
+	var cumSum float64
 	for i, idx := range indices {
 		cumSum += probs[idx]
 		if cumSum > float64(p) {
 			for _, idx := range indices[i+1:] {
-				logits[idx] = math.NaN()
+				logits[idx] = math.Inf(-1)
 			}
 			break
 		}
@@ -117,17 +132,14 @@ func (p TopP) Sample(logits []float64) ([]float64, error) {
 	return logits, nil
 }
 
-type MinP float32
+type MinP float64
 
-func (p MinP) Sample(logits []float64) ([]float64, error) {
+func (p MinP) Apply(logits []float64) ([]float64, error) {
 	if p <= 0 || p >= 1 {
 		return nil, errors.New("p must be between 0 and 1")
 	}
 
-	probs, err := computeSoftmax(logits)
-	if err != nil {
-		return nil, err
-	}
+	probs := computeSoftmax(logits)
 	copiedProbs := make([]float64, len(probs))
 	copy(copiedProbs, probs)
 
@@ -138,7 +150,7 @@ func (p MinP) Sample(logits []float64) ([]float64, error) {
 
 	for i := range probs {
 		if probs[i] < probThreshold {
-			logits[i] = math.NaN()
+			logits[i] = math.Inf(-1)
 		}
 	}
 
@@ -151,48 +163,51 @@ func Weighed() Sampler {
 	return weighed{}
 }
 
-func (s weighed) Sample(logits []float64) ([]float64, error) {
+// should return single value
+func (s weighed) Sample(logits []float64) (int, error) {
 	logitsCopy := make([]float64, 0, len(logits))
 	indices := make([]int, 0, len(logits))
 	// the uv sampler does not support NaN values
 	for i, logit := range logits {
-		if !math.IsNaN(logit) {
+		if !math.IsInf(logit, -1) {
 			logitsCopy = append(logitsCopy, logit)
 			indices = append(indices, i)
 		}
 	}
 
 	if len(logitsCopy) == 0 {
-		return nil, errors.New("no valid tokens found")
+		return -1, errors.New("no valid tokens found")
 	}
 
-	softmax, err := computeSoftmax(logitsCopy)
-	if err != nil {
-		return nil, err
-	}
+	softmax := computeSoftmax(logitsCopy)
 	w := sampleuv.NewWeighted(softmax, nil)
-	if v, ok := w.Take(); ok {
+	if idx, ok := w.Take(); ok {
 		// returns the token ID
-		return []float64{float64(indices[v])}, nil
+		return indices[idx], nil
 	}
-	return nil, errors.New("weighed sampler failed")
+	return -1, errors.New("weighed sampler failed")
 }
 
-func Sample(logits []float64, samplers ...Sampler) ([]float64, error) {
+// Sample applies transforms and samples a token ID
+func (s *SamplerConfig) Sample(input []float32) (int, error) {
+	logits := make([]float64, len(input))
+	for i, v := range input {
+		logits[i] = float64(v)
+	}
+
 	var err error
-	for _, sampler := range samplers {
-		if sampler == Temperature(0) {
+	for _, t := range s.transforms {
+		if t == Temperature(0) {
 			// early return with greedy if temperature is 0
-			logits, err = Greedy().Sample(logits)
-			if err != nil {
-				return nil, err
-			}
-			return logits, nil
+			s.sampler = Greedy()
+			break
 		}
-		logits, err = sampler.Sample(logits)
+
+		logits, err = t.Apply(logits)
 		if err != nil {
-			return nil, err
+			return -1, err
 		}
 	}
-	return logits, nil
+
+	return s.sampler.Sample(logits)
 }
diff --git a/sample/sample_test.go b/sample/sample_test.go
index 4039c29cd..78c7209e7 100644
--- a/sample/sample_test.go
+++ b/sample/sample_test.go
@@ -10,7 +10,7 @@ import (
 )
 
 func TestTemperature(t *testing.T) {
-	logits, err := Temperature(0.5).Sample([]float64{-3, -2, -1, 0, 1, 2, 4})
+	logits, err := Temperature(0.5).Apply([]float64{-3, -2, -1, 0, 1, 2, 4})
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -19,16 +19,16 @@ func TestTemperature(t *testing.T) {
 		t.Fatalf("got: %v, want: %v", logits, want)
 	}
 
-	if _, err := Temperature(-1).Sample([]float64{-3, -2, -1, 0, 1, 2, 4}); err == nil {
+	if _, err := Temperature(-1).Apply([]float64{-3, -2, -1, 0, 1, 2, 4}); err == nil {
 		t.Fatalf("expected error for temperature=-1, got %v", logits)
 	}
-	if _, err := Temperature(2.1).Sample([]float64{-3, -2, -1, 0, 1, 2, 4}); err == nil {
+	if _, err := Temperature(2.1).Apply([]float64{-3, -2, -1, 0, 1, 2, 4}); err == nil {
 		t.Fatalf("expected error for temperature=2.1, got %v", logits)
 	}
 }
 
 func TestSoftmax(t *testing.T) {
-	probs, err := Softmax().Sample([]float64{-3, -2, -1, 0, 1, 2, 4})
+	probs, err := Softmax().Apply([]float64{-3, -2, -1, 0, 1, 2, 4})
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -40,95 +40,101 @@ func TestSoftmax(t *testing.T) {
 }
 
 func TestTopK(t *testing.T) {
-	logits, err := TopK(3).Sample([]float64{-3, -2, -1, 0, 1, 2, 4})
+	logits, err := TopK(3).Apply([]float64{-3, -2, -1, 0, 1, 2, 4})
 	if err != nil {
 		t.Fatal(err)
 	}
-	expectedlogits := []float64{math.NaN(), math.NaN(), math.NaN(), math.NaN(), 1, 2, 4}
+	expectedlogits := []float64{math.Inf(-1), math.Inf(-1), math.Inf(-1), math.Inf(-1), 1, 2, 4}
 	if !floats.Same(logits, expectedlogits) {
 		t.Fatalf("logits: %v, expectedlogits: %v", logits, expectedlogits)
 	}
-	logits, err = TopK(0).Sample([]float64{-3, -2, -1, 0, 1, 2, 4})
+	logits, err = TopK(0).Apply([]float64{-3, -2, -1, 0, 1, 2, 4})
 	if err == nil {
 		t.Fatalf("expected error for k=0, got %v", logits)
 	}
+
+	logits, err = TopK(10).Apply([]float64{-3, -2, -1, 0, 1, 2, 4})
+	if err != nil {
+		t.Fatal(err)
+	}
+	expectedlogits = []float64{-3, -2, -1, 0, 1, 2, 4}
+	if !floats.Same(logits, expectedlogits) {
+		t.Fatalf("logits: %v, expectedlogits: %v", logits, expectedlogits)
+	}
 }
 
 func TestTopP(t *testing.T) {
-	logits, err := TopP(0.9).Sample([]float64{-3, -2, -1, 0, 1, 2, 4})
+	logits, err := TopP(0.9).Apply([]float64{-3, -2, -1, 0, 1, 2, 4})
 	if err != nil {
 		t.Fatal(err)
 	}
-	want := []float64{math.NaN(), math.NaN(), math.NaN(), math.NaN(), math.NaN(), 2, 4}
+	want := []float64{math.Inf(-1), math.Inf(-1), math.Inf(-1), math.Inf(-1), math.Inf(-1), 2, 4}
 	if !floats.Same(logits, want) {
 		t.Fatalf("got: %v, want: %v", logits, want)
 	}
-	logits, err = TopP(1.0).Sample([]float64{-3, -2, -1, 0, 1, 2, 4})
+	logits, err = TopP(1.0).Apply([]float64{-3, -2, -1, 0, 1, 2, 4})
 	if err == nil {
 		t.Fatalf("expected error for p=1.0, got %v", logits)
 	}
-	logits, err = TopP(0.0).Sample([]float64{-3, -2, -1, 0, 1, 2, 4})
+	logits, err = TopP(0.0).Apply([]float64{-3, -2, -1, 0, 1, 2, 4})
 	if err == nil {
 		t.Fatalf("expected error for p=0.0, got %v", logits)
 	}
 }
 
 func TestMinP(t *testing.T) {
-	logits, err := MinP(0.2).Sample([]float64{-3, -2, -1, 0, 1, 2, 3, 4})
+	logits, err := MinP(0.2).Apply([]float64{-3, -2, -1, 0, 1, 2, 3, 4})
 	if err != nil {
 		t.Fatal(err)
 	}
-	want := []float64{math.NaN(), math.NaN(), math.NaN(), math.NaN(), math.NaN(), math.NaN(), 3, 4}
+	want := []float64{math.Inf(-1), math.Inf(-1), math.Inf(-1), math.Inf(-1), math.Inf(-1), math.Inf(-1), 3, 4}
 	if !floats.Same(logits, want) {
 		t.Fatalf("got: %v, want: %v", logits, want)
 	}
-	logits, err = MinP(1.0).Sample([]float64{-3, -2, -1, 0, 1, 2, 3, 4})
+	logits, err = MinP(1.0).Apply([]float64{-3, -2, -1, 0, 1, 2, 3, 4})
 	if err == nil {
 		t.Fatalf("expected error for p=1.0, got %v", logits)
 	}
-	logits, err = MinP(0.0).Sample([]float64{-3, -2, -1, 0, 1, 2, 3, 4})
+	logits, err = MinP(0.0).Apply([]float64{-3, -2, -1, 0, 1, 2, 3, 4})
 	if err == nil {
 		t.Fatalf("expected error for p=0.0, got %v", logits)
 	}
 }
 
 func TestWeighed(t *testing.T) {
-	logits, err := Weighed().Sample([]float64{math.NaN(), 2, math.NaN(), math.NaN()})
+	idx, err := Weighed().Sample([]float64{math.Inf(-1), 2, math.Inf(-1), math.Inf(-1)})
 	if err != nil {
 		t.Fatal(err)
 	}
-	want := []float64{1}
-	if !floats.Equal(logits, want) {
-		t.Fatalf("got: %v, want: %v", logits, want)
+	want := 1
+	if idx != want {
+		t.Fatalf("got: %v, want: %v", idx, want)
 	}
-	logits, err = Weighed().Sample([]float64{math.NaN(), math.NaN(), math.NaN()})
+	idx, err = Weighed().Sample([]float64{math.Inf(-1), math.Inf(-1), math.Inf(-1)})
 	if err == nil {
-		t.Fatalf("expected error for no valid tokens, got %v", logits)
+		t.Fatalf("expected error for no valid tokens, got %v", idx)
 	}
 }
 
 func TestSample(t *testing.T) {
-	input := []float64{1, 2, 3, 4}
-	want := []float64{1, 2, 3, 4}
+	input := []float32{1, 2, 3, 4}
 
 	var callOrder []int
-	mock1 := &testSampler{
-		id:         1,
-		callOrder:  &callOrder,
-		returnVals: want,
+	mock1 := &testTransform{
+		id:        1,
+		callOrder: &callOrder,
 	}
-	mock2 := &testSampler{
-		id:         2,
-		callOrder:  &callOrder,
-		returnVals: want,
+	mock2 := &testTransform{
+		id:        2,
+		callOrder: &callOrder,
 	}
-	mock3 := &testSampler{
-		id:         3,
-		callOrder:  &callOrder,
-		returnVals: want,
+	mock3 := &testTransform{
+		id:        3,
+		callOrder: &callOrder,
 	}
+	sampler := NewSampler([]Transform{mock1, mock2, mock3}, Greedy())
 
-	got, err := Sample(input, mock1, mock2, mock3)
+	got, err := sampler.Sample(input)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -137,43 +143,45 @@ func TestSample(t *testing.T) {
 		t.Errorf("got %v, want %v", callOrder, []int{1, 2, 3})
 	}
 
-	if !floats.Equal(got, want) {
+	want := 3 // Greedy sampler should pick highest logit
+	if got != want {
 		t.Errorf("got %v, want %v", got, want)
 	}
 
-	errMock := &testSampler{
+	errMock := &testTransform{
 		returnErr: fmt.Errorf("mock error"),
 	}
-	_, err = Sample(input, mock1, errMock, mock2)
+	sampler = NewSampler([]Transform{mock1, errMock, mock2}, Greedy())
+	_, err = sampler.Sample(input)
 	if err == nil {
 		t.Error("Expected error from sampler")
 	}
 }
 
-type testSampler struct {
-	id         int
-	callOrder  *[]int
-	returnVals []float64
-	returnErr  error
+type testTransform struct {
+	id        int
+	callOrder *[]int
+	returnErr error
 }
 
-func (ts *testSampler) Sample(logits []float64) ([]float64, error) {
+func (ts *testTransform) Apply(logits []float64) ([]float64, error) {
 	if ts.callOrder != nil {
 		*ts.callOrder = append(*ts.callOrder, ts.id)
 	}
 	if ts.returnErr != nil {
 		return nil, ts.returnErr
 	}
-	return ts.returnVals, nil
+	return logits, nil
 }
 
 func TestSampleTemperatureZero(t *testing.T) {
-	logits, err := Sample([]float64{1, 2, 3, 4}, Temperature(0))
+	sampler := NewSampler([]Transform{Temperature(0)}, Greedy())
+	got, err := sampler.Sample([]float32{1, 2, 3, 4})
 	if err != nil {
 		t.Fatal(err)
 	}
-	want := []float64{3}
-	if !floats.Equal(logits, want) {
-		t.Fatalf("got: %v, want: %v", logits, want)
+	want := 3 // Greedy sampler should pick highest logit index
+	if got != want {
+		t.Fatalf("got: %v, want: %v", got, want)
 	}
 }