/aosp_15_r20/external/executorch/backends/arm/test/ops/ |
H A D | test_layer_norm.py | 52 self.layer_norm = torch.nn.LayerNorm( 59 self.layer_norm.weight = torch.nn.Parameter( 63 self.layer_norm.bias = torch.nn.Parameter( 68 return self.layer_norm(x) 82 .check(["torch.ops.aten.layer_norm.default"]) 86 .check_not(["torch.ops.aten.layer_norm.default"]) 103 .check_not(["torch.ops.aten.layer_norm.default"]) 125 .check_not(["torch.ops.aten.layer_norm.default"])
|
/aosp_15_r20/external/pytorch/test/distributed/_tensor/ |
H A D | test_math_ops.py | 279 layer_norm = torch.nn.LayerNorm( 284 layer_norm_local = copy.deepcopy(layer_norm).to(self.device_type) 294 layer_norm_dist = distribute_module(layer_norm, device_mesh, _replicate_fn) 344 layer_norm = torch.nn.LayerNorm( 349 layer_norm_local = copy.deepcopy(layer_norm).to(self.device_type) 359 layer_norm_dist = distribute_module(layer_norm, device_mesh, _replicate_fn) 472 self.layer_norm = torch.nn.LayerNorm( 481 h = self.layer_norm(h) 489 "layer_norm": SequenceParallel(), 502 "layer_norm": ln_req_grad,
|
/aosp_15_r20/external/pytorch/torch/onnx/ |
H A D | symbolic_opset17.py | 32 __all__ = ["layer_norm", "stft", "quantized_layer_norm"] 37 @_onnx_symbolic("aten::layer_norm") 39 def layer_norm( function 50 # layer_norm normalizes on the last D dimensions, 73 @_onnx_symbolic("quantized::layer_norm") 86 output = layer_norm(g, x, normalized_shape, weight, bias, eps, False)
|
/aosp_15_r20/external/pytorch/torch/csrc/api/include/torch/nn/functional/ |
H A D | normalization.h | 54 inline Tensor layer_norm( in layer_norm() function 60 return torch::layer_norm(input, normalized_shape, weight, bias, eps); in layer_norm() 66 /// https://pytorch.org/docs/main/nn.functional.html#torch.nn.functional.layer_norm 75 /// F::layer_norm(input, F::LayerNormFuncOptions({2, 2}).eps(2e-5)); 77 inline Tensor layer_norm( in layer_norm() function 80 return detail::layer_norm( in layer_norm()
|
/aosp_15_r20/external/pytorch/aten/src/ATen/native/vulkan/ops/ |
H A D | Layernorm.cpp | 74 // We invoke native_layer_norm which returns a tuple of tensors: <layer_norm, in run_layernorm_context() 75 // mean, 1/sqrt(var+eps)>, but we only need the first tensor (layer_norm). in run_layernorm_context() 81 static Tensor layer_norm( in layer_norm() function 98 m.impl(TORCH_SELECTIVE_NAME("aten::layer_norm"), TORCH_FN(layer_norm)); in TORCH_LIBRARY_IMPL()
|
/aosp_15_r20/external/tensorflow/tensorflow/python/keras/layers/normalization/ |
H A D | layer_normalization_test.py | 159 layer_norm = layer_normalization.LayerNormalization(axis=3) 160 layer_norm.build(input_shape=(2, 2, 2)) 165 layer_norm = layer_normalization.LayerNormalization(axis=[-1, -1]) 166 layer_norm.build(input_shape=(2, 2, 2)) 170 layer_norm = layer_normalization.LayerNormalization(axis=[-2, -1]) 171 layer_norm.build(input_shape=(2, 2, 2)) 172 self.assertEqual(layer_norm._fused, True)
|
/aosp_15_r20/external/pytorch/test/onnx/ |
H A D | test_onnxscript_no_runtime.py | 51 # 2. Register layer_norm onnxscript function as custom Op 53 def layer_norm( function 77 layer_norm, input, weight, bias, axes_i=axes, eps_f=eps 81 symbolic_name="aten::layer_norm", 107 self.assertEqual(layer_norm_proto.functions[0].name, "layer_norm")
|
H A D | test_onnxscript_runtime.py | 92 def layer_norm( function 116 layer_norm, input, weight, bias, axes_i=axes, eps_f=eps 120 symbolic_name="aten::layer_norm",
|
H A D | test_pytorch_onnx_onnxruntime_cuda.py | 54 self.layer_norm = torch.nn.LayerNorm([10, 10]) 58 return self.layer_norm(x)
|
/aosp_15_r20/external/executorch/exir/tests/ |
H A D | models.py | 250 self.layer_norm = nn.LayerNorm(input_dim) 262 y = self.layer_norm(x) 498 self.layer_norm = nn.LayerNorm(input_dim) 532 CR_normed = self.layer_norm(CR) 533 # C_normed = self.layer_norm(C) 534 # R_normed = self.layer_norm(R) 563 Z_CR_normed = self.layer_norm(Z_CR) 567 output = self.layer_norm(self.elem_add(ffn_out, Z_CR))
|
/aosp_15_r20/external/pytorch/torch/nn/modules/ |
H A D | normalization.py | 155 >>> layer_norm = nn.LayerNorm(embedding_dim) 157 >>> layer_norm(embedding) 164 >>> layer_norm = nn.LayerNorm([C, H, W]) 165 >>> output = layer_norm(input) 167 .. image:: ../_static/img/nn/layer_norm.jpg 217 return F.layer_norm(
|
/aosp_15_r20/external/pytorch/torch/csrc/jit/passes/ |
H A D | decompose_ops.cpp | 37 …"aten::layer_norm(Tensor input, int[] normalized_shape, Tensor? weight, Tensor? bias, float eps, b… in isDecomposableNorm() 160 …"aten::layer_norm(Tensor input, int[] normalized_shape, Tensor? weight, Tensor? bias, float eps, b… in DecomposeOps() 175 toGraphFunction(decompose_funcs.get_function("layer_norm")).graph(); in DecomposeOps() 210 …def layer_norm(input : Tensor, normalized_shape : List[int], eps : float, cudnn_enable : bool) -> … in DecomposeOps()
|
/aosp_15_r20/external/executorch/examples/cadence/models/ |
H A D | rnnt_encoder.py | 90 self.layer_norm = torch.nn.LayerNorm(output_dim) 103 layer_norm_out = self.layer_norm(output_linear_out) 123 layer_norm_out = self.layer_norm(output_linear_out)
|
H A D | rnnt_predictor.py | 36 self.layer_norm = torch.nn.LayerNorm(symbol_embedding_dim) 47 layer_norm_out = self.layer_norm(linear_out)
|
/aosp_15_r20/external/executorch/backends/arm/_passes/ |
H A D | decompose_layernorm_pass.py | 29 if op == torch.ops.aten.layer_norm.default: 40 raise RuntimeError(f"Can't get layer_norm composition for op {op}") 62 torch.ops.aten.layer_norm.default,
|
/aosp_15_r20/external/executorch/kernels/portable/cpu/ |
H A D | op_native_layer_norm.cpp | 24 void layer_norm( in layer_norm() function 98 …da5b17b086554c6cd0c3ab08a35aeec2a8bad8c]/xplat/caffe2/aten/src/ATen/native/layer_norm.cpp?lines=188 172 layer_norm<CTYPE>( in native_layer_norm_out()
|
/aosp_15_r20/external/pytorch/torch/ao/ns/fx/ |
H A D | mappings.py | 215 # F.layer_norm 217 F.layer_norm, 497 F.layer_norm, 525 toq.layer_norm,
|
/aosp_15_r20/external/executorch/backends/cadence/hifi/operators/ |
H A D | quantized_layer_norm.cpp | 27 // Compute quantized layer_norm. The current implementation assumes that the 52 // layer_norm for each vector. in quantized_layer_norm_per_tensor_() 92 // Compute quantized layer_norm. The current implementation assumes that the
|
/aosp_15_r20/external/executorch/backends/cadence/reference/operators/ |
H A D | quantized_layer_norm.cpp | 24 // Compute quantized layer_norm. The current implementation assumes that the 49 // layer_norm for each vector. in quantized_layer_norm_per_tensor_() 85 // Compute quantized layer_norm. The current implementation assumes that the
|
/aosp_15_r20/external/pytorch/test/ |
H A D | test_nestedtensor.py | 937 layer_norm = torch.nn.LayerNorm(size, device=device, dtype=dtype) 938 nt_result = layer_norm(nt) 940 t_result = layer_norm(t.reshape(1, -1, size).squeeze(0)) 949 layer_norm = torch.nn.LayerNorm(size, device=device, dtype=dtype) 950 nt_result = layer_norm(nt) 952 t_result = layer_norm(t.reshape(1, -1, size).squeeze(0)) 969 layer_norm = torch.nn.LayerNorm( 972 nt_result = layer_norm(nt) 974 t_result = layer_norm(t.reshape(1, -1, size, size, 4).squeeze(0)) 978 layer_norm = torch.nn.LayerNorm((size, 4), device=device, dtype=dtype) [all …]
|
/aosp_15_r20/external/pytorch/torch/nn/utils/_expanded_weights/ |
H A D | layer_norm_expanded_weights.py | 17 @implements_per_sample_grads(F.layer_norm) 48 F.layer_norm(input, normalized_shape, eps=ctx.eps) * grad_output,
|
/aosp_15_r20/external/pytorch/benchmarks/tensorexpr/ |
H A D | pt_engine.py | 39 def layer_norm(self, data, shape): member in TorchTensorEngine 40 return torch.nn.functional.layer_norm(data, shape)
|
/aosp_15_r20/external/executorch/backends/qualcomm/builders/ |
H A D | README.md | 31 self.layer_norm = torch.nn.LayerNorm([768], eps=1e-6) 35 return self.linear(self.layer_norm(x)) 127 …ion](https://github.com/pytorch/pytorch/blob/main/aten/src/ATen/native/layer_norm.cpp) mentioned i…
|
/aosp_15_r20/external/pytorch/aten/src/ATen/native/transformers/ |
H A D | transformer.cpp | 18 #include <ATen/ops/layer_norm.h> 70 return at::layer_norm(input, {embed_dim}, weight, bias, eps, true); in norm()
|
/aosp_15_r20/external/executorch/kernels/optimized/cpu/ |
H A D | op_native_layer_norm.cpp | 27 void layer_norm( in layer_norm() function 160 layer_norm<CTYPE>( in opt_native_layer_norm_out()
|