| /aosp_15_r20/external/pytorch/torch/_inductor/fx_passes/ |
| H A D | fuse_attention.py | 77 def _sfdp_pattern_3(query, key, value, inv_scale_factor, dropout_p): argument
86 def _sfdp_replacement_3(query, key, value, inv_scale_factor, dropout_p): argument
99 def _sfdp_pattern_4(query, key, value, scale_factor, dropout_p): argument
106 def _sfdp_replacement_4(query, key, value, scale_factor, dropout_p): argument
139 def _sfdp_pattern_6(query, key, value, attn_mask, dropout_p): argument
147 def _sfdp_replacement_6(query, key, value, attn_mask, dropout_p): argument
159 def _sfdp_pattern_7(query, key, value, dropout_p): argument
174 def _sfdp_replacement_7(query, key, value, dropout_p): argument
221 def _sfdp_pattern_9(query, key, value, dropout_p): argument
234 def _sfdp_replacement_9(query, key, value, dropout_p): argument
[all …]
|
| /aosp_15_r20/external/executorch/extension/llm/custom_ops/ |
| H A D | op_sdpa_test.cpp | 27 double dropout_p, in op_scaled_dot_product_attention()
98 double dropout_p = 0; in TEST() local
141 double dropout_p = 0.0; in TEST() local
171 double dropout_p = 0.0; in TEST() local
239 double dropout_p = 0.0; in TEST() local
315 double dropout_p = 0.0; in TEST() local
502 double dropout_p = 0.0; in TEST() local
|
| H A D | op_sdpa_with_kv_cache_test.cpp | 30 double dropout_p, in op_sdpa_with_kv_cache()
146 double dropout_p = 0; in TEST() local
388 double dropout_p = 0; in TEST() local
591 double dropout_p = 0; in TEST() local
836 double dropout_p = 0; in TEST() local
|
| H A D | op_sdpa_aot.cpp | 31 const double dropout_p, in sdpa_with_kv_cache_out_no_context()
64 const double dropout_p, in sdpa_with_kv_cache_aten()
93 const double dropout_p, in custom_sdpa_out_no_context()
120 const double dropout_p, in custom_sdpa_aten()
|
| H A D | op_sdpa.cpp | 222 double dropout_p, in cpu_flash_attention()
764 const double dropout_p, in flash_attention_kernel_out()
847 const double dropout_p, in custom_sdpa_out()
996 const double dropout_p, in sdpa_with_kv_cache_out()
|
| /aosp_15_r20/external/pytorch/aten/src/ATen/native/transformers/ |
| H A D | attention.cpp | 427 …const std::optional<Tensor>& attn_mask_, double dropout_p, bool is_causal, std::optional<double> s… in _fused_sdp_choice_cpp()
450 double dropout_p, in _fused_sdp_choice_meta()
484 double dropout_p, in validate_sdpa_input()
602 …const std::optional<Tensor>& attn_mask_, double dropout_p, bool is_causal, std::optional<double> s… in handle_private_use()
654 double dropout_p, in scaled_dot_product_attention()
717 const std::optional<Tensor>& attn_mask_, double dropout_p, bool is_causal, in _scaled_dot_product_attention_math()
775 double dropout_p, in _scaled_dot_product_flash_attention_cpu()
823 double dropout_p, in _scaled_dot_product_flash_attention_cpu_backward()
|
| /aosp_15_r20/external/pytorch/aten/src/ATen/native/transformers/cuda/ |
| H A D | attention.cu | 688 double dropout_p, in _scaled_dot_product_flash_attention_cuda()
742 double dropout_p, in _scaled_dot_product_cudnn_attention_cuda()
795 double dropout_p, in _scaled_dot_product_efficient_attention_cuda()
830 …const std::optional<Tensor>& attn_mask_, double dropout_p, bool is_causal, std::optional<double> s… in _fused_sdp_choice_cuda()
851 double dropout_p, in _flash_attention_forward()
966 double dropout_p, // attention matrix dropout probability in _efficient_attention_forward()
1314 …r triton_scaled_dot_attention(const Tensor& q, const Tensor& k, const Tensor& v, double dropout_p){ in triton_scaled_dot_attention()
1378 double dropout_p, in _fill_mem_eff_dropout_mask_()
|
| H A D | attention_backward.cu | 72 double dropout_p, in _flash_attention_backward()
181 double dropout_p, in _scaled_dot_product_cudnn_attention_backward_cuda()
266 double dropout_p, // dropout probability in _efficient_attention_backward()
743 double dropout_p, in _scaled_dot_product_flash_attention_backward_cuda()
794 double dropout_p, in _scaled_dot_product_efficient_attention_backward_cuda()
|
| /aosp_15_r20/external/pytorch/aten/src/ATen/native/ |
| H A D | RNN.cpp | 1082 int64_t num_layers, double dropout_p, bool train) { in apply_layer_stack()
1112 int64_t num_layers, double dropout_p, bool train, bool bidirectional) { in _rnn_impl()
1129 int64_t num_layers, double dropout_p, bool train, bool bidirectional) { in _rnn_impl_with_concat()
1138 int64_t num_layers, double dropout_p, bool train, bool bidirectional) { in _lstm_impl()
1405 double dropout_p, in quantized_gru_data_legacy()
1433 int64_t num_layers, double dropout_p, bool train, bool bidirectional, bool batch_first) { in lstm()
1494 int64_t num_layers, double dropout_p, bool train, bool bidirectional) { in lstm()
1694 double dropout_p, in quantized_lstm_input()
1734 double dropout_p, in quantized_lstm_input_legacy()
1754 double dropout_p, in quantized_lstm_data()
[all …]
|
| /aosp_15_r20/external/pytorch/aten/src/ATen/native/nested/cuda/ |
| H A D | NestedTensorTransformerFunctions.cpp | 233 double dropout_p, in _scaled_dot_product_flash_attention_nestedtensor_cuda()
288 double dropout_p, in _scaled_dot_product_efficient_attention_nestedtensor_cuda()
337 double dropout_p, in _scaled_dot_product_flash_attention_backward_nested()
|
| /aosp_15_r20/external/executorch/backends/vulkan/test/op_tests/ |
| H A D | sdpa_test.cpp | 43 const double dropout_p, in sdpa_with_kv_cache_out_no_context()
76 const double dropout_p, in sdpa_with_kv_cache_aten()
163 const double dropout_p, in sdpa_reference_impl()
|
| /aosp_15_r20/external/pytorch/aten/src/ATen/native/cpu/ |
| H A D | FlashAttentionKernel.cpp | 180 double dropout_p, in cpu_flash_attention()
430 double dropout_p, in cpu_flash_attention_backward()
735 double dropout_p, in flash_attention_kernel_impl()
786 double dropout_p, in flash_attention_backward_kernel_impl()
|
| /aosp_15_r20/external/pytorch/torch/nested/_internal/ |
| H A D | sdpa.py | 28 dropout_p=0.0, argument
622 dropout_p=0.0, argument
|
| /aosp_15_r20/external/pytorch/aten/src/ATen/native/cudnn/ |
| H A D | RNN.cpp | 135 auto dropout_p = train ? dropout : 0; in descriptor() local
2214 auto dropout_p = train ? dropout : 0; in _cudnn_init_dropout_state() local
2348 double dropout_p, in get_dropout_state()
2506 double dropout_p, in _cudnn_impl()
2579 double dropout_p, in _cudnn_impl()
2729 double dropout_p, in lstm_packed_cudnn()
|
| /aosp_15_r20/external/pytorch/test/cpp_extensions/ |
| H A D | open_registration_extension.cpp | 131 …const std::optional<at::Tensor> & attn_mask, double dropout_p, bool is_causal, std::optional<doubl… in _fused_sdp_choice_privateuse1()
450 double dropout_p, in custom_scaled_dot_product_fused_attention_overrideable()
485 double dropout_p, in custom_scaled_dot_product_fused_attention_overrideable_backward()
|
| /aosp_15_r20/external/pytorch/aten/src/ATen/native/mkldnn/ |
| H A D | RNN.cpp | 458 int64_t num_layers, bool has_biases, bool batch_first, double dropout_p, in mkldnn_rnn()
538 int64_t num_layers, double dropout_p, bool train, bool bidirectional, bool batch_first) { in mkldnn_impl()
553 int64_t num_layers, double dropout_p, bool train, bool bidirectional, bool batch_first) { in lstm_mkldnn()
|
| /aosp_15_r20/external/pytorch/torch/testing/_internal/distributed/_tensor/ |
| H A D | common_dtensor.py | 98 dropout_p: float = 0.1 variable in ModelArgs
143 def __init__(self, dim, hidden_dim, dropout_p): argument
|
| /aosp_15_r20/external/pytorch/torch/csrc/inductor/aoti_torch/ |
| H A D | shim_common.cpp | 516 double dropout_p, in aoti_torch__scaled_dot_product_flash_attention_v2()
566 double dropout_p, in aoti_torch__scaled_dot_product_flash_attention()
606 double dropout_p, in aoti_torch__scaled_dot_product_efficient_attention()
|
| /aosp_15_r20/external/pytorch/aten/src/ATen/functorch/ |
| H A D | BatchRulesLinearAlgebra.cpp | 494 double dropout_p, in _scaled_dot_product_flash_attention_batch_rule()
544 double dropout_p, in _scaled_dot_product_efficient_attention_batch_rule()
585 double dropout_p, in _scaled_dot_product_cudnn_attention_batch_rule()
|
| /aosp_15_r20/external/pytorch/test/ |
| H A D | test_transformers.py | 969 …ef test_scaled_dot_product_attention(self, device, input_dim, attn_mask_dim, is_causal, dropout_p): argument
975 dropout_p=0.0): argument
2995 …def _get_mem_eff_drop_mask(batch_size, n_heads, q_len, kv_len, dropout_p, seed, offset, device=dev… argument
3001 …def get_dropout_mask(output, fused_kernel, batch_size, n_heads, q_len, kv_len, dropout_p, device=d… argument
|
| /aosp_15_r20/external/pytorch/torch/nn/attention/ |
| H A D | _utils.py | 47 dropout_p=0.0, argument
|
| /aosp_15_r20/external/pytorch/aten/src/ATen/native/miopen/ |
| H A D | RNN_miopen.cpp | 825 int64_t num_layers, double dropout_p, bool train, bool bidirectional) { in _miopen_impl()
847 int64_t num_layers, double dropout_p, bool train, bool bidirectional, bool batch_first) { in _miopen_impl()
900 int64_t num_layers, double dropout_p, bool train, bool bidirectional) { in lstm_packed_miopen()
|
| /aosp_15_r20/external/pytorch/torch/csrc/api/src/nn/options/ |
| H A D | activation.cpp | 48 double dropout_p, in MultiheadAttentionForwardFuncOptions()
|
| /aosp_15_r20/external/executorch/backends/vulkan/runtime/graph/ops/impl/ |
| H A D | SDPA.cpp | 191 const ValueRef dropout_p = args[arg_idx++]; in sdpa_with_kv_cache_impl() local
|
| /aosp_15_r20/external/pytorch/torch/ |
| H A D | _meta_registrations.py | 5261 dropout_p=0.0, argument
|