|
| | ScaledDotProductAttention (const Tensor query, const Tensor key, const Tensor value, const std::optional< Tensor > attn_mask, double dropout_p, bool is_causal, const std::optional< double > scale, bool enable_gqa, Tensor out) |
| |
| | ScaledDotProductAttention (const Tensor query, const Tensor key, const Tensor value, Tensor out) |
| |
| virtual void | operator() (const Tensor query, const Tensor key, const Tensor value, const std::optional< Tensor > attn_mask, double dropout_p, bool is_causal, const std::optional< double > scale, bool enable_gqa, Tensor out) const =0 |
| |
| void | operator() (const Tensor query, const Tensor key, const Tensor value, Tensor out) const |
| |
| void | operator() (const Handle &handle, const Args &... args) |
| |
| void | operator() (const Args &... args) const |
| |
| void | operator() (const Handle &handle, const Args &... args) |
| |
| void | operator() (const Args &... args) const |
| |
| virtual | ~OperatorBase ()=default |
| |
| virtual std::size_t | workspace_size_in_bytes () const |
| |
| void | set_handle (const Handle &handle) |
| |
| void | set_config (const Config &config) |
| |
| void | set_stream (void *stream) |
| |
| void | set_workspace (void *workspace) |
| |
| void | set_workspace_size_in_bytes (std::size_t workspace_size_in_bytes) |
| |
| virtual | ~OperatorBase ()=default |
| |
| virtual std::size_t | workspace_size_in_bytes () const |
| |
| void | set_handle (const Handle &handle) |
| |
| void | set_config (const Config &config) |
| |
| void | set_stream (void *stream) |
| |
| void | set_workspace (void *workspace) |
| |
| void | set_workspace_size_in_bytes (std::size_t workspace_size_in_bytes) |
| |
|
| template<typename TensorLike > |
| static auto | MakeReturnValue (const TensorLike &query, const TensorLike &key, const TensorLike &value, const std::optional< TensorLike > attn_mask=std::nullopt, double dropout_p=0.0, bool is_causal=false, const std::optional< double > scale=std::nullopt, bool enable_gqa=false) |
| |
| static void | clear_cache () |
| |
| static void | clear_cache () |
| |
| static std::unique_ptr< Operator > | Make (const Config &config, const Tensor tensor, Args &&... args) |
| |
| static std::unique_ptr< Operator > | Make (const Tensor tensor, Args &&... args) |
| |
| static std::unique_ptr< Operator > | Make (const Config &config, const std::vector< Tensor > tensors, Args &&... args) |
| |
| static std::unique_ptr< Operator > | Make (const std::vector< Tensor > tensors, Args &&... args) |
| |
| static std::unique_ptr< Operator > | Make (const Config &config, const Tensor tensor, Args &&... args) |
| |
| static std::unique_ptr< Operator > | Make (const Tensor tensor, Args &&... args) |
| |
| static std::unique_ptr< Operator > | Make (const Config &config, const std::vector< Tensor > tensors, Args &&... args) |
| |
| static std::unique_ptr< Operator > | Make (const std::vector< Tensor > tensors, Args &&... args) |
| |
| static void | Call (const Handle &handle, const Config &config, const Args &... args) |
| |
| static void | Call (const Tensor tensor, const Args &... args) |
| |
| static auto | Call (const TensorLike &tensor, const Args &... args) |
| |
| static void | Call (const Handle &handle, const Config &config, const Args &... args) |
| |
| static void | Call (const Tensor tensor, const Args &... args) |
| |
| static auto | Call (const TensorLike &tensor, const Args &... args) |
| |
| static std::vector< std::size_t > | active_implementation_indices (Device::Type dev_type) |
| |
| static std::vector< std::size_t > | active_implementation_indices (Device::Type dev_type) |
| |