last fix

src/brevitas/quant/base.py

@@ -302,7 +302,8 @@ class WeightPerTensorFloatDecoupledL2Param(SolveWeightScalingStatsInputDimsFromM
     signed = True
 
 
-class WeightPerChannelFloatDecoupled(SolveWeightScalingStatsInputDimsFromModule,
+class WeightPerChannelFloatDecoupled(SolveStatsReduceDimFromEnum,
+                                     SolveWeightScalingStatsInputDimsFromModule,
                                      SolveWeightScalingPerOutputChannelShapeFromModule,
                                      SolveParameterScalingShape):
     """

src/brevitas/quant/experimental/float_quant_fnuz.py

@@ -8,6 +8,8 @@
 from brevitas.quant.base import MSESymmetricScale
 from brevitas.quant.experimental.float_base import FloatActBase
 from brevitas.quant.experimental.float_base import FloatWeightBase
+from brevitas.quant.experimental.float_base import Fp8e4m3Mixin
+from brevitas.quant.experimental.float_base import Fp8e5m2Mixin
 from brevitas.quant.experimental.float_base import ScaledFloatActBase
 from brevitas.quant.experimental.float_base import ScaledFloatWeightBase
 
@@ -36,7 +38,7 @@ class FpFNUZAct(FpFNUZMixin, FloatActBase):
 
 class FpFNUZWeightPerTensorFloat(FpFNUZMixin, ScaledFloatWeightBase):
     """
-    FP8 signed E3M4 weight quantizer with per-tensor absmax-based scaling.
+    FP8 signed weight quantizer with per-tensor absmax-based scaling.
     """
     scaling_output_type = ScalingPerOutputType.TENSOR
 
@@ -50,7 +52,7 @@ class FpFNUZActPerTensorFloat(FpFNUZMixin, ScaledFloatActBase):
 
 class FpFNUZWeightPerChannelFloat(FpFNUZMixin, ScaledFloatWeightBase):
     """
-    FP8 signed E3M4 weight quantizer with per-channel absmax-based scaling.
+    FP8 signed weight quantizer with per-channel absmax-based scaling.
     """
     scaling_output_type = ScalingPerOutputType.CHANNEL
 
@@ -80,13 +82,128 @@ class FpFNUZActPerChannelFloat2dMSE(FpFNUZMixin, MSESymmetricScale, ScaledFloatA
 
 class FpFNUZWeightPerChannelFloatMSE(FpFNUZMixin, MSESymmetricScale, ScaledFloatWeightBase):
     """
-    FP8 signed E3M4 weight quantizer with per-channel MSE-based scaling.
+    FP8 signed weight quantizer with per-channel MSE-based scaling.
     """
     scaling_output_type = ScalingPerOutputType.CHANNEL
 
 
 class FpFNUZWeightPerTensorFloatMSE(FpFNUZMixin, MSESymmetricScale, ScaledFloatWeightBase):
     """
-    FP8 signed E3M4 weight quantizer with per-tensor MSE-based scaling.
+    FP8 signed weight quantizer with per-tensor MSE-based scaling.
     """
     scaling_output_type = ScalingPerOutputType.TENSOR
+
+
+## Predefined FP8 Quantizers
+
+
+class Fp8e4m3FNUZWeight(FpFNUZWeight, Fp8e4m3Mixin):
+    """
+    FNUZ FP8 E4M3 signed weight quantizer.
+    """
+    pass
+
+
+class Fp8e4m3FNUZAct(FpFNUZAct, Fp8e4m3Mixin):
+    """
+    FNUZ FP8 E4M3 signed act quantizer.
+    """
+    pass
+
+
+class Fp8e4m3FNUZWeightPerTensorFloat(FpFNUZWeightPerTensorFloat, Fp8e4m3Mixin):
+    """
+    FNUZ FP8 E4M3 per-tensor scaled signed weight quantizer.
+    """
+    pass
+
+
+class Fp8e4m3FNUZWeightPerChannelFloat(FpFNUZWeightPerChannelFloat, Fp8e4m3Mixin):
+    """
+    FNUZ FP8 E4M3 per-channel scaled signed weight quantizer.
+    """
+    pass
+
+
+class Fp8e4m3FNUZActPerTensorFloat(FpFNUZActPerTensorFloat, Fp8e4m3Mixin):
+    """
+    FNUZ FP8 E4M3 scaled signed act quantizer.
+    """
+    pass
+
+
+class Fp8e4m3FNUZActPerTensorFloatMSE(FpFNUZActPerTensorFloatMSE, Fp8e4m3Mixin):
+    """
+    FNUZ FP8 E4M3 MSE-based scaled signed act quantizer.
+    """
+    pass
+
+
+class Fp8e4m3FNUZWeightPerTensorFloatMSE(FpFNUZWeightPerTensorFloatMSE, Fp8e4m3Mixin):
+    """
+    FNUZ FP8 E4M3 MSE-based per-tensor scaled signed weight quantizer.
+    """
+    pass
+
+
+class Fp8e4m3FNUZWeightPerChannelFloatMSE(FpFNUZWeightPerChannelFloatMSE, Fp8e4m3Mixin):
+    """
+    FNUZ FP8 E4M3 MSE-based per-channel scaled signed weight quantizer.
+    """
+    pass
+
+
+class Fp8e5m2FNUZWeight(FpFNUZWeight, Fp8e5m2Mixin):
+    """
+    FNUZ FP8 e5m2 signed weight quantizer.
+    """
+    pass
+
+
+class Fp8e5m2FNUZAct(FpFNUZAct, Fp8e5m2Mixin):
+    """
+    FNUZ FP8 e5m2 signed act quantizer.
+    """
+    pass
+
+
+class Fp8e5m2FNUZWeightPerTensorFloat(FpFNUZWeightPerTensorFloat, Fp8e5m2Mixin):
+    """
+    FNUZ FP8 e5m2 per-tensor scaled signed weight quantizer.
+    """
+    pass
+
+
+class Fp8e5m2FNUZWeightPerChannelFloat(FpFNUZWeightPerChannelFloat, Fp8e5m2Mixin):
+    """
+    FNUZ FP8 e5m2 per-channel scaled signed weight quantizer.
+    """
+    pass
+
+
+class Fp8e5m2FNUZActPerTensorFloat(FpFNUZActPerTensorFloat, Fp8e5m2Mixin):
+    """
+    FNUZ FP8 e5m2 scaled signed act quantizer.
+    """
+    pass
+
+
+class Fp8e5m2FNUZActPerTensorFloatMSE(FpFNUZActPerTensorFloatMSE, Fp8e5m2Mixin):
+    """
+    FNUZ FP8 e5m2 MSE-based scaled signed act quantizer.
+    """
+    pass
+
+
+class Fp8e5m2FNUZWeightPerTensorFloatMSE(FpFNUZWeightPerTensorFloatMSE, Fp8e5m2Mixin):
+    """
+    FNUZ FP8 e5m2 MSE-based per-tensor scaled signed weight quantizer.
+    """
+    pass
+
+
+class Fp8e5m2FNUZWeightPerChannelFloatMSE(FpFNUZWeightPerChannelFloatMSE, Fp8e5m2Mixin):
+    """
+    FNUZ FP8 e5m2 MSE-based per-channel scaled signed weight quantizer.
+    """
+    pass