mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
2d234357c6
* Update TensorRT-LLM --------- Co-authored-by: Altair-Alpha <62340011+Altair-Alpha@users.noreply.github.com>
287 lines
11 KiB
C++
287 lines
11 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.
|
|
*/
|
|
|
|
#include "tensorrt_llm/layers/dynamicDecodeLayer.h"
|
|
#include "tensorrt_llm/common/tensor.h"
|
|
#include "tensorrt_llm/kernels/decodingKernels.h"
|
|
#include "tensorrt_llm/layers/layerUtils.h"
|
|
#include "tensorrt_llm/layers/layersFactory.h"
|
|
#include "tensorrt_llm/runtime/bufferManager.h"
|
|
|
|
#include <optional>
|
|
#include <utility>
|
|
|
|
using namespace tensorrt_llm::common;
|
|
using namespace tensorrt_llm::kernels;
|
|
using namespace tensorrt_llm::runtime;
|
|
|
|
namespace tensorrt_llm::layers
|
|
{
|
|
|
|
template <typename T>
|
|
DynamicDecodeLayer<T>::DynamicDecodeLayer(executor::DecodingMode const& mode, DecoderDomain const& decoderDomain,
|
|
cudaStream_t stream, std::shared_ptr<IAllocator> allocator)
|
|
: BaseLayer(decoderDomain, stream, std::move(allocator))
|
|
, mDecodingMode(mode)
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
|
|
initialize();
|
|
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
DynamicDecodeLayer<T>::~DynamicDecodeLayer()
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
|
|
freeBuffer();
|
|
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DynamicDecodeLayer<T>::initialize()
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
|
|
mIdsPtrHost = runtime::BufferManager::pinned(ITensor::makeShape({}), runtime::TRTDataType<TokenIdType*>::value);
|
|
|
|
allocateBuffer();
|
|
|
|
mCyclicStep = 0;
|
|
mRuntimeMaxSeqLen = 0;
|
|
mConfiguredBeamWidth = -1;
|
|
|
|
if (!mDecodingMode.isAuto())
|
|
{
|
|
mConfiguredBeamWidth = mDecoderDomain.getBeamWidth();
|
|
initializeLayers();
|
|
}
|
|
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DynamicDecodeLayer<T>::allocateBuffer()
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
|
|
mZeroParentIdsDevice
|
|
= mAllocator->reMalloc(mZeroParentIdsDevice, sizeof(TokenIdType*) * 2 * mDecoderDomain.getBatchSize(), false);
|
|
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DynamicDecodeLayer<T>::freeBuffer()
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
|
|
mAllocator->free((void**) &mZeroParentIdsDevice);
|
|
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DynamicDecodeLayer<T>::initializeLayers()
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
|
|
mLayers = createLayers<T>(mDecodingMode, mDecoderDomain, mStream, mAllocator);
|
|
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DynamicDecodeLayer<T>::setup(SizeType32 batchSize, SizeType32 beamWidth, SizeType32 const* batchSlots,
|
|
std::shared_ptr<BaseSetupParams> const& baseSetupParams)
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
|
|
auto setupParams = std::dynamic_pointer_cast<DynamicDecodeSetupParams>(baseSetupParams);
|
|
|
|
TLLM_CHECK_WITH_INFO(setupParams->decodingParams, "decodingParams for setup is not set");
|
|
if (setupParams->decodingParams->outputLogProbs)
|
|
{
|
|
// FIXME(nkorobov): monotonically growing
|
|
mOutputLogProbs = std::any_of(setupParams->decodingParams->outputLogProbs->begin(),
|
|
setupParams->decodingParams->outputLogProbs->end(),
|
|
[this](bool outputLogProbs) { return this->mOutputLogProbs | outputLogProbs; });
|
|
}
|
|
|
|
if (mConfiguredBeamWidth == -1)
|
|
{
|
|
// This code is left only for Python runtime
|
|
// In C++ runtime given maxBeamWidth should always be equal to the runtime beamWidth
|
|
TLLM_CHECK(mDecodingMode.isAuto());
|
|
mConfiguredBeamWidth = beamWidth;
|
|
mDecodingMode
|
|
= mConfiguredBeamWidth == 1 ? executor::DecodingMode::TopKTopP() : executor::DecodingMode::BeamSearch();
|
|
initializeLayers();
|
|
}
|
|
|
|
TLLM_CHECK_WITH_INFO((mConfiguredBeamWidth == 1 && beamWidth == 1)
|
|
|| (mConfiguredBeamWidth > 1 && beamWidth > 1 && beamWidth <= mConfiguredBeamWidth),
|
|
"Decoder is configured with beam width %d, but %d was given", mConfiguredBeamWidth, beamWidth);
|
|
TLLM_CHECK_WITH_INFO(mConfiguredBeamWidth <= mDecoderDomain.getBeamWidth(),
|
|
"Decoder is created with max beam width %d, but %d was given", mDecoderDomain.getBeamWidth(),
|
|
mConfiguredBeamWidth);
|
|
|
|
for (auto& layer : mLayers)
|
|
{
|
|
layer->setup(batchSize, beamWidth, batchSlots, baseSetupParams);
|
|
}
|
|
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DynamicDecodeLayer<T>::forwardAsync(
|
|
std::shared_ptr<BaseDecodingOutputs> const& baseOutputs, std::shared_ptr<BaseDecodingInputs> const& baseInputs)
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
|
|
auto params = std::dynamic_pointer_cast<DecodingInputs>(baseInputs);
|
|
|
|
TLLM_CHECK_WITH_INFO(mDecodingMode.isExplicitDraftTokens() || params->logits || params->logitsVec,
|
|
"If not explicit Draft Tokens mode, either logits or logitsVec have to be specified.");
|
|
TLLM_CHECK_WITH_INFO(
|
|
baseOutputs->sequenceLength.has_value(), "sequenceLength tensor is required in DynamicDecoderLayer.");
|
|
|
|
auto const localDecoderDomain = getLocalDecoderDomain(params, mDecoderDomain);
|
|
auto const maxSeqLen = baseOutputs->outputIds.shape[baseOutputs->outputIds.shape.size() - 1];
|
|
|
|
TLLM_CHECK_WITH_INFO((mConfiguredBeamWidth == 1 && localDecoderDomain.getBeamWidth() == 1)
|
|
|| (mConfiguredBeamWidth > 1 && localDecoderDomain.getBeamWidth() > 1
|
|
&& localDecoderDomain.getBeamWidth() <= mConfiguredBeamWidth),
|
|
"Decoder is configured with beam width %d, but %d was given", mConfiguredBeamWidth,
|
|
localDecoderDomain.getBeamWidth());
|
|
|
|
if (!mIdsPtrHost->data())
|
|
{
|
|
mIdsPtrHost = runtime::BufferManager::pinnedPool(ITensor::makeShape({static_cast<int32_t>(maxSeqLen),
|
|
static_cast<int32_t>(2 * mDecoderDomain.getBatchSize())}),
|
|
runtime::TRTDataType<int32_t*>::value);
|
|
mRuntimeMaxSeqLen = maxSeqLen;
|
|
}
|
|
|
|
std::vector<SizeType32> batchSlotsVec(localDecoderDomain.getBatchSize());
|
|
std::iota(batchSlotsVec.begin(), batchSlotsVec.end(), 0);
|
|
auto batchSlotsHost
|
|
= params->batchSlots ? params->batchSlots->template getPtr<SizeType32 const>() : batchSlotsVec.data();
|
|
auto batchSlots = params->batchSlots ? params->batchSlots->template getPtr<SizeType32 const>() : nullptr;
|
|
|
|
mCyclicStep = mCyclicStep % mRuntimeMaxSeqLen;
|
|
prepareIdsPtrs(
|
|
baseOutputs, batchSlotsHost, localDecoderDomain.getBatchSize(), localDecoderDomain.getBeamWidth(), maxSeqLen);
|
|
|
|
for (auto& layer : mLayers)
|
|
{
|
|
layer->forwardAsync(baseOutputs, baseInputs);
|
|
}
|
|
|
|
// Copy nextIds and transpose logits when needed
|
|
prepareOutputData(baseOutputs, params, mIdsPtrHost, batchSlots, localDecoderDomain.getBatchSize(),
|
|
mDecoderDomain.getBatchSize(), localDecoderDomain.getBeamWidth(), maxSeqLen,
|
|
mDecoderDomain.getMaxDecodingTokens(), mCyclicStep, mOutputLogProbs, mStream);
|
|
|
|
mCyclicStep += 1;
|
|
|
|
sync_check_cuda_error();
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DynamicDecodeLayer<T>::forwardSync(
|
|
std::shared_ptr<BaseDecodingOutputs> const& baseOutputs, std::shared_ptr<BaseDecodingInputs> const& baseInputs)
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
for (auto& layer : mLayers)
|
|
{
|
|
layer->forwardSync(baseOutputs, baseInputs);
|
|
}
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DynamicDecodeLayer<T>::prepareIdsPtrs(std::shared_ptr<BaseDecodingOutputs> const& outputs,
|
|
SizeType32 const* batchSlots, SizeType32 batchSize, SizeType32 beamWidth, SizeType32 maxSeqLen)
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
auto idsPtrHostSlice = ITensor::slice(mIdsPtrHost, mCyclicStep, 1);
|
|
auto idsPtrHost = reinterpret_cast<TokenIdType**>(runtime::bufferCast<int64_t>(*idsPtrHostSlice));
|
|
for (SizeType32 bi = 0; bi < batchSize; bi++)
|
|
{
|
|
auto const batchSlot = batchSlots[bi];
|
|
idsPtrHost[batchSlot]
|
|
= outputs->outputIds.template getPtrWithOffset<TokenIdType>(batchSlot * beamWidth * maxSeqLen);
|
|
}
|
|
for (SizeType32 bi = 0; bi < batchSize; bi++)
|
|
{
|
|
auto const batchSlot = batchSlots[bi];
|
|
if (beamWidth > 1)
|
|
{
|
|
idsPtrHost[mDecoderDomain.getBatchSize() + batchSlot]
|
|
= outputs->parentIds.value().template getPtrWithOffset<SizeType32>(batchSlot * beamWidth * maxSeqLen);
|
|
}
|
|
else
|
|
{
|
|
idsPtrHost[mDecoderDomain.getBatchSize() + batchSlot] = mZeroParentIdsDevice + bi * beamWidth * maxSeqLen;
|
|
}
|
|
}
|
|
|
|
outputs->outputIdsPtr = Tensor(
|
|
MEMORY_GPU, DataType::TYPE_INT32_PTR, {static_cast<size_t>(mDecoderDomain.getBatchSize())}, idsPtrHost);
|
|
outputs->parentIdsPtr = Tensor(MEMORY_GPU, DataType::TYPE_INT32_PTR,
|
|
{static_cast<size_t>(mDecoderDomain.getBatchSize())}, idsPtrHost + mDecoderDomain.getBatchSize());
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DynamicDecodeLayer<T>::prepareOutputData(std::shared_ptr<BaseDecodingOutputs> const& outputs,
|
|
std::shared_ptr<DecodingInputs> const& params, runtime::ITensor::SharedPtr const& idsPtrsHost,
|
|
SizeType32 const* batchSlots, SizeType32 batchSize, SizeType32 maxBatchSize, SizeType32 beamWidth,
|
|
SizeType32 maxSeqLen, SizeType32 maxTokensPerStep, SizeType32 cyclicStep, bool outputLogProbs, cudaStream_t stream)
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
auto idsPtrHostSlice = ITensor::slice(idsPtrsHost, cyclicStep, 1);
|
|
auto idsPtrHost = reinterpret_cast<TokenIdType**>(runtime::bufferCast<int64_t>(*idsPtrHostSlice));
|
|
auto const numNewTokens
|
|
= outputs->numNewTokens ? outputs->numNewTokens->template getPtr<SizeType32 const>() : nullptr;
|
|
invokeCopyNextStepIds(outputs->newTokens.template getPtr<TokenIdType>(), idsPtrHost,
|
|
outputs->sequenceLength->template getPtr<SizeType32>(), numNewTokens, batchSlots, batchSize, maxBatchSize,
|
|
beamWidth, maxSeqLen, maxTokensPerStep, stream);
|
|
|
|
// Transpose output log probs from [maxSeqLen, batchSize, beamWidth] to [batchSize, beamWidth, maxSeqLen]
|
|
if (outputLogProbs && outputs->outputLogProbsTiled)
|
|
{
|
|
auto logProbsMaxSeqLen = outputs->outputLogProbsTiled.value().shape[0];
|
|
|
|
invokeTransposeLogProbs(outputs->outputLogProbs.value().template getPtr<float>(),
|
|
outputs->outputLogProbsTiled.value().template getPtr<float>(),
|
|
outputs->sequenceLength->template getPtr<SizeType32>(), batchSlots, batchSize, maxBatchSize, beamWidth,
|
|
logProbsMaxSeqLen, stream);
|
|
}
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template class DynamicDecodeLayer<float>;
|
|
template class DynamicDecodeLayer<half>;
|
|
|
|
} // namespace tensorrt_llm::layers
|