mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-13 22:18:36 +08:00
70 lines
2.3 KiB
C++
70 lines
2.3 KiB
C++
/*
|
|
* Copyright (c) 2022-2024, NVIDIA CORPORATION. All rights reserved.
|
|
*
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
* you may not use this file except in compliance with the License.
|
|
* You may obtain a copy of the License at
|
|
*
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
*
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
* See the License for the specific language governing permissions and
|
|
* limitations under the License.
|
|
*/
|
|
#pragma once
|
|
|
|
#include "tensorrt_llm/common/config.h"
|
|
#include "tensorrt_llm/common/cudaUtils.h"
|
|
#include "tensorrt_llm/kernels/gptKernels.h"
|
|
#include "tensorrt_llm/runtime/iTensor.h"
|
|
#include <cstdint>
|
|
#include <cuda_fp16.h>
|
|
#include <cuda_runtime.h>
|
|
#include <sstream>
|
|
|
|
namespace tc = tensorrt_llm::common;
|
|
|
|
TRTLLM_NAMESPACE_BEGIN
|
|
|
|
namespace kernels
|
|
{
|
|
|
|
////////////////////////////////////////////////////////////////////////////////////////////////////
|
|
|
|
template <typename MaskDataType>
|
|
struct AttentionMaskParams
|
|
{
|
|
// The attention mask's shape is [batchSize, maxQSeqLen, maxKvSeqLen] when cuQSeqLens is nullptr,
|
|
// otherwise [batchSize, maxQSeqLen] is packed as numTokens.
|
|
MaskDataType* mask = nullptr;
|
|
// The cumulative sequence lengths of Q.
|
|
int* cuQSeqLens = nullptr;
|
|
// The actual q sequence lengths (used to create cuMaskRows).
|
|
int const* actualQSeqLens = nullptr;
|
|
// The actual kv sequence lengths (used to construct the packed mask when full mask is not given).
|
|
int const* actualKvSeqLens = nullptr;
|
|
// The attention mask type.
|
|
AttentionMaskType attentionMaskType = AttentionMaskType::PADDING;
|
|
// Params for block sparse pattern
|
|
BlockSparseParams blockSparseParams;
|
|
// The batch size.
|
|
int batchSize;
|
|
// The maximum q sequence length.
|
|
int maxQSeqLen;
|
|
// The maximum kv sequence length.
|
|
int maxKvSeqLen;
|
|
// The sliding window size.
|
|
int slidingWindowSize;
|
|
};
|
|
|
|
////////////////////////////////////////////////////////////////////////////////////////////////////
|
|
|
|
template <typename MaskDataType>
|
|
void invokeBuildAttentionMask(AttentionMaskParams<MaskDataType> const& params, cudaStream_t stream);
|
|
|
|
} // namespace kernels
|
|
|
|
TRTLLM_NAMESPACE_END
|