CPUExecutionProvider
    unchanged

CUDAExecutionProvider
    unchanged

CoreMLExecutionProvider
    +12 Slice
    +11 MatMul
    +9 Add
    +2 Mul
    -1 Attention is_causal=0 qk_matmul_output_mode=0 softcap=0.0
    -1 ReduceL2 keepdims=1 noop_with_empty_axes=0
    +1 ReduceSum keepdims=1
    +1 Softmax axis=-1
    +1 Sqrt
    +1 Transpose perm=(0, 1, 3, 2)
    -1 Transpose perm=(0, 3, 1, 2)
    535 nodes, weights 92f96a725b3d
    DecomposeAttention x5 b00b8abe
    DecomposeReduceL2 x3 a689cbf8
    FuseBatchNormIntoConv, RemoveOptionalBiasFromConv x199 323cfe83
    SplitLargeReduction x33 46040459
    _ConstantifyReshapeTarget x9 8c4eb3c1
    _ScalarGatherToSlice x1 c2f58daf
    unstamped x285 8079f14d

MIGraphXExecutionProvider
    +2 MatMul
    -1 Attention is_causal=0 qk_matmul_output_mode=0 softcap=0.0
    +1 Mul
    +1 Softmax axis=-1
    +1 Transpose perm=(0, 1, 3, 2)
    504 nodes, weights 95b5f0ed0d20
    DecomposeAttention x5 b00b8abe
    FuseBatchNormIntoConv, RemoveOptionalBiasFromConv x199 8d4edbb1
    _ConstantifyReshapeTarget x9 8c4eb3c1
    _ScalarGatherToSlice x1 c2f58daf
    unstamped x290 20772a87

NvTensorRTRTXExecutionProvider
    +2 MatMul
    -1 Attention is_causal=0 qk_matmul_output_mode=0 softcap=0.0
    +1 Mul
    +1 Softmax axis=-1
    +1 Transpose perm=(0, 1, 3, 2)
    504 nodes, weights 95b5f0ed0d20
    DecomposeAttention x5 b00b8abe
    FuseBatchNormIntoConv, RemoveOptionalBiasFromConv x199 8d4edbb1
    _ConstantifyReshapeTarget x9 8c4eb3c1
    _ScalarGatherToSlice x1 c2f58daf
    unstamped x290 20772a87

OpenVINOExecutionProvider
    +2 MatMul
    -1 Attention is_causal=0 qk_matmul_output_mode=0 softcap=0.0
    +1 Mul
    +1 Softmax axis=-1
    +1 Transpose perm=(0, 1, 3, 2)
    504 nodes, weights 95b5f0ed0d20
    DecomposeAttention x5 b00b8abe
    FuseBatchNormIntoConv, RemoveOptionalBiasFromConv x199 8d4edbb1
    _ConstantifyReshapeTarget x9 8c4eb3c1
    _ScalarGatherToSlice x1 c2f58daf
    unstamped x290 20772a87

RKNPU static
    +200 Slice
    +142 Add
    +112 Conv auto_pad=NOTSET dilations=(1, 1) group=1 pads=(1, 1, 1, 1) strides=(1, 1)
    +23 MatMul
    +9 Conv auto_pad=NOTSET dilations=(1, 1) group=1 pads=(0, 0, 0, 0) strides=(1, 1)
    -1 Attention is_causal=0 qk_matmul_output_mode=0 softcap=0.0
    -1 Cast to=1
    +1 Mul
    +1 Softmax axis=-1
    +1 Transpose perm=(0, 1, 3, 2)
    987 nodes, weights 7fc618c42854
    contract {"dims":[{}]}
    opset 19
    DecomposeAttention x5 b00b8abe
    FuseBatchNormIntoConv, RemoveOptionalBiasFromConv x144 84f30084
    SplitLargeConvReduction, FuseBatchNormIntoConv, RemoveOptionalBiasFromConv x473 45af76e7
    SplitLargeReduction x69 2f748d6a
    _ConstantifyReshapeTarget x9 8c4eb3c1
    _ScalarGatherToSlice x1 c2f58daf
    unstamped x286 2bdc87b7

TensorrtExecutionProvider
    +2 MatMul
    -1 Attention is_causal=0 qk_matmul_output_mode=0 softcap=0.0
    +1 Mul
    +1 Softmax axis=-1
    +1 Transpose perm=(0, 1, 3, 2)
    504 nodes, weights 95b5f0ed0d20
    DecomposeAttention x5 b00b8abe
    FuseBatchNormIntoConv, RemoveOptionalBiasFromConv x199 8d4edbb1
    _ConstantifyReshapeTarget x9 8c4eb3c1
    _ScalarGatherToSlice x1 c2f58daf
    unstamped x290 20772a87
