.xoptim.patterns_ort

modules

• .xoptim.patterns_ort.activation
• .xoptim.patterns_ort.activation_grad
• .xoptim.patterns_ort.batch_normalization
• .xoptim.patterns_ort.fused_conv
• .xoptim.patterns_ort.fused_matmul
• .xoptim.patterns_ort.gather_grad
• .xoptim.patterns_ort.llm_optim
• .xoptim.patterns_ort.simplified_layer_normalization

experimental_experiment.xoptim.patterns_ort.get_onnxruntime_patterns(verbose: int = 0) → List[PatternOptimization]

Returns a default list of optimization patterns for onnxruntime. It is equal to the following list.

<<<

from experimental_experiment.xoptim.patterns_api import pattern_table_doc
from experimental_experiment.xoptim.patterns_ort import get_onnxruntime_patterns

print(pattern_table_doc(get_onnxruntime_patterns(), as_rst=True))

>>>

	name	short_name	priority	doc
0	BiasGeluPattern	BiasGelu	1	Replaces by y = BiasGelu(x, B)
1	BiasSoftmaxPattern	BiasSoftmax	1	Replaces Softmax(Add(x,y), axis=-1) by BiasSoftmax(x,y,axis=-1)
2	ContribRotaryEmbeddingPattern	ContribRotaryEmbedding	2	Very similar to experimental_experiment.xoptim.patterns.onnx_rotary.RotaryEmbeddingPattern.
3	ContribRotaryEmbedding3DPattern	ContribRotaryEmbedding3D	1	Extension to experimental_experiment.xoptim.patterns_ort.llm_optim.ContribRotaryEmbeddingPattern, turn the operator into a 3D operator including the transpose.
4	GeluOrtPattern	GeluOrt	0	Detects the decomposed version of Gelu with Tanh .. math
5	GeluErfPattern	GeluErf	0	Detects the decomposed version of Gelu with Erf.
6	FusedConvPattern	FusedConv	2	Replaces the Conv + Relu into FusedConv.
7	FastGeluPattern	FastGelu	1	Replaces Gelu by FastGelu.
8	FusedMatMulPattern	FusedMatMul	2	Replaces the sequence Transpose, Matmul into FusedMatMul.
9	FusedMatMulx2Pattern	FusedMatMulx2	3	Replaces the sequence Div by a scalar consumed by two FusedMatMul.
10	FusedMatMulDivPattern	FusedMatMulDiv	2	Replaces the Matmul, Div into FusedMatMul.
11	FusedMatMulTransposePattern	FusedMatMulTranspose	3	Replaces the sequence (Fused)Matmul(A,B) + Transpose into FusedMatMul(B.T, A.T).
12	MultiHeadAttention3DPattern	MultiHeadAttention3D	1	Merges multiple nodes into MultiHeadAttention. It assumes pattern experimental_experiment.xoptim.patterns.onnx_attention.FunctionAttentionPattern was triggered before.
13	OrtBatchNormalizationTrainingPattern	OrtBatchNormalizationTraining	1	onnxruntime does not support batch normalization with training=1.
14	QuickGeluPattern	QuickGelu	1	Replaces Mul(x, Sigmoid(x)) by QuickGelu(x, alpha=1)
15	ReshapeGemmPattern	ReshapeGemm	3	Replaces the sequence Reshape(-1, …) + Gemm into FusedMatMul().
16	SimplifiedLayerNormalizationPattern	SimplifiedLayerNormalization	1	Fuses the nodes equivalent to SimplifiedLayerNormalization.
17	SimplifiedLayerNormalizationMulPattern	SimplifiedLayerNormalizationMul	1	Replaces the sequence SimplifiedLayerNormalization + Mul by SimplifiedLayerNormalization.
18	SkipLayerNormalizationPattern	SkipLayerNormalization	1	Replaces the sequence Add + LayerNormalization into SkipLayerNormalization.
19	SkipSimplifiedLayerNormalizationPattern	SkipSimplifiedLayerNormalization	1	Replaces the sequence Add + SimplifiedLayerNormalization by SkipSimplifiedLayerNormalization.
20	SkipSimplifiedLayerNormalizationMulPattern	SkipSimplifiedLayerNormalizationMul	1	Replaces the sequence SkipSimplifiedLayerNormalization + Mul by SkipSimplifiedLayerNormalization.
21	SoftmaxGradPattern	SoftmaxGrad	1	Replaces the sequence Mul, ReduceSum, Mul, Sub by SoftmaxGrad.
22	TransposeFusedMatMulBPattern	TransposeFusedMatMulB	3	Replaces the sequence Transpose(B, [0, 2, 3, 1] + (Fused)Matmul(A,B) into Transpose(A, [0, 2, 1, 3]) + FusedMatMul(A, B, transB=1).