hls4ml.converters.keras_v3.hgq2 package

Submodules

hls4ml.converters.keras_v3.hgq2.einsum module

class hls4ml.converters.keras_v3.hgq2.einsum.QEinsumHandler

Bases: QLayerHandler

handle(layer: hgq.layers.QEinsum, in_tensors: Sequence[KerasTensor], out_tensors: Sequence[KerasTensor])

handles = ('hgq.layers.ops.einsum.QEinsum',)

hls4ml.converters.keras_v3.hgq2.multi_head_attention module

class hls4ml.converters.keras_v3.hgq2.multi_head_attention.QLinformerAttentionHandler

Bases: QMultiHeadAttentionHandler

handle(layer: hgq.layers.linformer_attention.QLinformerAttention, in_tensors: Sequence[KerasTensor], out_tensors: Sequence[KerasTensor])

handles = ('hgq.layers.linformer_attention.QLinformerAttention',)

class hls4ml.converters.keras_v3.hgq2.multi_head_attention.QMultiHeadAttentionHandler

Bases: QLayerHandler

handle(layer: hgq.layers.QMultiHeadAttention, in_tensors: Sequence[KerasTensor], out_tensors: Sequence[KerasTensor])

handles = ('hgq.layers.multi_head_attention.QMultiHeadAttention',)

hls4ml.converters.keras_v3.hgq2.pooling module

class hls4ml.converters.keras_v3.hgq2.pooling.QPoolingHandler

Bases: PoolingHandler, QLayerHandler

handles = ('hgq.layers.pooling.QMaxPooling1D', 'hgq.layers.pooling.QMaxPooling2D', 'hgq.layers.pooling.QMaxPooling3D', 'hgq.layers.pooling.QAveragePooling1D', 'hgq.layers.pooling.QAveragePooling2D', 'hgq.layers.pooling.QAveragePooling3D', 'hgq.layers.pooling.QGlobalAveragePooling1D', 'hgq.layers.pooling.QGlobalAveragePooling2D', 'hgq.layers.pooling.QGlobalAveragePooling3D', 'hgq.layers.pooling.QGlobalMaxPooling1D', 'hgq.layers.pooling.QGlobalMaxPooling2D', 'hgq.layers.pooling.QGlobalMaxPooling3D')

hls4ml.converters.keras_v3.hgq2.softmax module

class hls4ml.converters.keras_v3.hgq2.softmax.QSoftmaxHandler

Bases: QLayerHandler

handle(layer: hgq.layers.QSoftmax, in_tensors: Sequence[KerasTensor], out_tensors: Sequence[KerasTensor])

handles = ('hgq.layers.softmax.QSoftmax',)

hls4ml.converters.keras_v3.hgq2.softmax.fixed_quantizer_to_hls4ml_t(q: FixedPointQuantizerBase, take_max=False)

hls4ml.converters.keras_v3.hgq2.unary_lut module

class hls4ml.converters.keras_v3.hgq2.unary_lut.QUnaryLUTHandler

Bases: QLayerHandler, KerasV3LayerHandler

handle(layer: hgq.layers.QUnaryFunctionLUT, in_tensors: Sequence[KerasTensor], out_tensors: Sequence[KerasTensor])

handles = ('hgq.layers.activation.QUnaryFunctionLUT',)

Module contents