mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
eba3623a54
* feat/vbws-part4-v1.8: rebase Signed-off-by: wili-65535 <wili-65535@users.noreply.github.com> * feat/vbws-part4-v1.9: fix incorrect output when using short output length Signed-off-by: wili-65535 <wili-65535@users.noreply.github.com> * v1.9.1: remove useless variables Signed-off-by: wili-65535 <wili-65535@users.noreply.github.com> * v1.9.2:fix incorrect output when using short output length Signed-off-by: wili-65535 <wili-65535@users.noreply.github.com> * v1.9.3: rebase Signed-off-by: wili-65535 <wili-65535@users.noreply.github.com> * v1.9.4: rebase Signed-off-by: wili-65535 <wili-65535@users.noreply.github.com> * v1.9.5: remove API change Signed-off-by: wili-65535 <wili-65535@users.noreply.github.com> --------- Signed-off-by: wili-65535 <wili-65535@users.noreply.github.com> Co-authored-by: wili-65535 <wili-65535@users.noreply.github.com>
267 lines
11 KiB
C++
267 lines
11 KiB
C++
/*
|
|
* Copyright (c) 2019-2024, NVIDIA CORPORATION. All rights reserved.
|
|
* Copyright (c) 2021, NAVER Corp. Authored by CLOVA.
|
|
*
|
|
* 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 "decodingLayer.h"
|
|
#include "tensorrt_llm/layers/beamSearchLayer.h"
|
|
#include "tensorrt_llm/layers/decodingParams.h"
|
|
#include "tensorrt_llm/layers/eagleDecodingLayer.h"
|
|
#include "tensorrt_llm/layers/explicitDraftTokensLayer.h"
|
|
#include "tensorrt_llm/layers/externalDraftTokensLayer.h"
|
|
#include "tensorrt_llm/layers/layerUtils.h"
|
|
#include "tensorrt_llm/layers/lookaheadDecodingLayer.h"
|
|
#include "tensorrt_llm/layers/medusaDecodingLayer.h"
|
|
#include "tensorrt_llm/layers/samplingLayer.h"
|
|
|
|
using namespace tensorrt_llm::common;
|
|
using namespace tensorrt_llm::kernels;
|
|
using namespace tensorrt_llm::runtime;
|
|
|
|
namespace tensorrt_llm::layers
|
|
{
|
|
|
|
template <typename T>
|
|
DecodingLayer<T>::DecodingLayer(executor::DecodingMode const& mode, DecoderDomain const& decoderDomain,
|
|
std::shared_ptr<BufferManager> bufferManager)
|
|
: BaseLayer(decoderDomain, bufferManager)
|
|
, mDecodingMode(mode)
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
|
|
if (mDecodingMode.isTopKorTopP())
|
|
{
|
|
mDecodingLayer = std::make_unique<SamplingLayer<T>>(mDecodingMode, decoderDomain, mBufferManager);
|
|
}
|
|
else if (mDecodingMode.isBeamSearch())
|
|
{
|
|
mDecodingLayer = std::make_unique<BeamSearchLayer<T>>(mDecodingMode, decoderDomain, mBufferManager);
|
|
}
|
|
else if (mDecodingMode.isMedusa())
|
|
{
|
|
mDecodingLayer = std::make_unique<MedusaDecodingLayer<T>>(decoderDomain, mBufferManager);
|
|
}
|
|
else if (mDecodingMode.isLookahead())
|
|
{
|
|
mDecodingLayer = std::make_unique<LookaheadDecodingLayer<T>>(mDecoderDomain, mBufferManager);
|
|
}
|
|
else if (mDecodingMode.isExplicitDraftTokens())
|
|
{
|
|
mDecodingLayer = std::make_unique<ExplicitDraftTokensLayer<T>>(decoderDomain, mBufferManager);
|
|
}
|
|
else if (mDecodingMode.isExternalDraftTokens())
|
|
{
|
|
mDecodingLayer = std::make_unique<ExternalDraftTokensLayer<T>>(mDecodingMode, decoderDomain, mBufferManager);
|
|
}
|
|
else if (mDecodingMode.isEagle())
|
|
{
|
|
mDecodingLayer = std::make_unique<EagleDecodingLayer<T>>(decoderDomain, mBufferManager);
|
|
}
|
|
else
|
|
{
|
|
TLLM_CHECK_WITH_INFO(false,
|
|
"Decoding mode is none of the supported {TopK, TopP, TopKTopP, BeamSearch, Medusa, Lookahead, "
|
|
"ExplicitDraftTokens, ExternalDraftTokens, Eagle}");
|
|
}
|
|
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DecodingLayer<T>::setup(SizeType32 batchSize, SizeType32 beamWidth, TensorConstPtr batchSlots,
|
|
std::shared_ptr<BaseSetupParams> const& baseSetupParams,
|
|
std::shared_ptr<runtime::DecodingLayerWorkspace> const& workspace)
|
|
{
|
|
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 (mDecodingMode.isBeamSearch())
|
|
{
|
|
TLLM_CHECK_WITH_INFO(
|
|
beamWidth > 1, "Decoding mode is %s, but beamWidth <= 1 (%d <= 1)", mDecodingMode.getName(), beamWidth);
|
|
}
|
|
else if (mDecodingMode.isTopKorTopP() || mDecodingMode.isMedusa() || mDecodingMode.isLookahead()
|
|
|| mDecodingMode.isExplicitDraftTokens() || mDecodingMode.isExternalDraftTokens() || mDecodingMode.isEagle())
|
|
{
|
|
TLLM_CHECK_WITH_INFO(
|
|
beamWidth == 1, "Decoding mode is %s, but beamWidth != 1 (%d != 1)", mDecodingMode.getName(), beamWidth);
|
|
}
|
|
else
|
|
{
|
|
TLLM_CHECK_WITH_INFO(false,
|
|
"Decoding mode is none of the supported {TopK, TopP, TopKTopP, BeamSearch, Medusa, Lookahead, "
|
|
"ExplicitDraftTokens, ExternalDraftTokens, Eagle}");
|
|
}
|
|
|
|
mDecodingLayer->setup(batchSize, beamWidth, batchSlots, setupParams->decodingParams, workspace);
|
|
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DecodingLayer<T>::forwardAsync(std::shared_ptr<BaseDecodingOutputs> const& baseOutputs,
|
|
std::shared_ptr<BaseDecodingInputs> const& baseInputs,
|
|
std::shared_ptr<runtime::DecodingLayerWorkspace> const& workspace)
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
auto [outputParams, inputParams] = prepareParams(baseOutputs, baseInputs);
|
|
mDecodingLayer->forwardAsync(outputParams, inputParams, workspace);
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
void DecodingLayer<T>::forwardSync(std::shared_ptr<BaseDecodingOutputs> const& baseOutputs,
|
|
std::shared_ptr<BaseDecodingInputs> const& baseInputs,
|
|
std::shared_ptr<runtime::DecodingLayerWorkspace> const& workspace)
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
auto [outputParams, inputParams] = prepareParams(baseOutputs, baseInputs);
|
|
mDecodingLayer->forwardSync(outputParams, inputParams, workspace);
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
}
|
|
|
|
template <typename T>
|
|
size_t DecodingLayer<T>::getWorkspaceSize() const noexcept
|
|
{
|
|
return mDecodingLayer->getWorkspaceSize();
|
|
}
|
|
|
|
template <typename T>
|
|
std::tuple<std::shared_ptr<BaseDecodingOutputs>, std::shared_ptr<BaseDecodingInputs>> DecodingLayer<T>::prepareParams(
|
|
std::shared_ptr<BaseDecodingOutputs> const& baseOutputs,
|
|
std::shared_ptr<BaseDecodingInputs> const& baseInputs) const
|
|
{
|
|
TLLM_LOG_TRACE("%s start", __PRETTY_FUNCTION__);
|
|
auto params = std::dynamic_pointer_cast<DecodingInputs>(baseInputs);
|
|
|
|
auto const localDecoderDomain = getLocalDecoderDomain(params, mDecoderDomain);
|
|
auto const& endIds = params->endIds;
|
|
|
|
std::shared_ptr<BaseDecodingOutputs> preparedOutputs;
|
|
std::shared_ptr<BaseDecodingInputs> preparedInputs;
|
|
|
|
if (mDecodingMode.isBeamSearch())
|
|
{
|
|
preparedInputs = baseInputs;
|
|
preparedOutputs = baseOutputs;
|
|
}
|
|
else if (mDecodingMode.isTopKorTopP())
|
|
{
|
|
auto const ite = params->ite;
|
|
auto const step = params->step;
|
|
auto const localBatchSize = static_cast<int64_t>(params->localBatchSize);
|
|
|
|
TLLM_CHECK_WITH_INFO(localDecoderDomain.getBeamWidth() == 1,
|
|
"Decoding mode is TopK and/or TopP, but beamWidth != 1 (%d != 1)", localDecoderDomain.getBeamWidth());
|
|
|
|
// In sampling, we have supported batch sampling. So, we always compute all
|
|
// sentences once.
|
|
TensorConstPtr logitsSlice = ITensor::slice(*params->logits, 0, localBatchSize);
|
|
TensorConstPtr endIdSlice = ITensor::slice(endIds, 0, localBatchSize);
|
|
auto decodeInputs = std::make_shared<SamplingInputs>(endIdSlice, params->batchSlots, step, ite, localBatchSize);
|
|
|
|
decodeInputs->finished = params->finished;
|
|
|
|
decodeInputs->logits = logitsSlice;
|
|
|
|
if (params->inputLengths)
|
|
{
|
|
auto& inputLengths = params->inputLengths.value();
|
|
decodeInputs->inputLengths = ITensor::slice(inputLengths, 0, localBatchSize);
|
|
}
|
|
preparedInputs = decodeInputs;
|
|
preparedOutputs = baseOutputs;
|
|
}
|
|
else if (mDecodingMode.isMedusa())
|
|
{
|
|
TLLM_CHECK_WITH_INFO(localDecoderDomain.getBeamWidth() == 1,
|
|
"Decoding mode is Medusa, but beamWidth != 1 (%d != 1)", localDecoderDomain.getBeamWidth());
|
|
|
|
preparedInputs = baseInputs;
|
|
preparedOutputs = baseOutputs;
|
|
}
|
|
else if (mDecodingMode.isLookahead())
|
|
{
|
|
preparedInputs = baseInputs;
|
|
preparedOutputs = baseOutputs;
|
|
}
|
|
else if (mDecodingMode.isExplicitDraftTokens())
|
|
{
|
|
preparedInputs = baseInputs;
|
|
preparedOutputs = baseOutputs;
|
|
}
|
|
else if (mDecodingMode.isExternalDraftTokens())
|
|
{
|
|
auto externalDraftTokenParams = std::dynamic_pointer_cast<ExternalDraftTokensInputs>(baseInputs);
|
|
auto const ite = externalDraftTokenParams->ite;
|
|
auto const step = externalDraftTokenParams->step;
|
|
auto const localBatchSize = static_cast<int64_t>(externalDraftTokenParams->localBatchSize);
|
|
|
|
TLLM_CHECK_WITH_INFO(localDecoderDomain.getBeamWidth() == 1,
|
|
"Decoding mode is TopK and/or TopP, but beamWidth != 1 (%d != 1)", localDecoderDomain.getBeamWidth());
|
|
|
|
// Compute all sentences once since batch-sampling is supported
|
|
TensorConstPtr logitsSlice = ITensor::slice(*externalDraftTokenParams->logits, 0, localBatchSize);
|
|
TensorConstPtr endIdSlice = ITensor::slice(endIds, 0, localBatchSize);
|
|
auto decodeInputs = std::make_shared<ExternalDraftTokensInputs>(
|
|
endIdSlice, externalDraftTokenParams->batchSlots, step, ite, localBatchSize);
|
|
|
|
decodeInputs->finished = externalDraftTokenParams->finished;
|
|
|
|
decodeInputs->logits = logitsSlice;
|
|
|
|
if (externalDraftTokenParams->inputLengths)
|
|
{
|
|
auto& inputLengths = externalDraftTokenParams->inputLengths.value();
|
|
decodeInputs->inputLengths = ITensor::slice(inputLengths, 0, localBatchSize);
|
|
}
|
|
decodeInputs->draftLogits = externalDraftTokenParams->draftLogits;
|
|
decodeInputs->draftProbs = externalDraftTokenParams->draftProbs;
|
|
decodeInputs->targetProbs = externalDraftTokenParams->targetProbs;
|
|
decodeInputs->numDraftTokens = externalDraftTokenParams->numDraftTokens;
|
|
decodeInputs->numDraftTokensHost = externalDraftTokenParams->numDraftTokensHost;
|
|
decodeInputs->draftTokenIds = externalDraftTokenParams->draftTokenIds;
|
|
decodeInputs->constantThreshold = externalDraftTokenParams->constantThreshold;
|
|
decodeInputs->useRandomAcceptanceThreshold = externalDraftTokenParams->useRandomAcceptanceThreshold;
|
|
decodeInputs->step = externalDraftTokenParams->step;
|
|
decodeInputs->useDraftLogits = externalDraftTokenParams->useDraftLogits;
|
|
decodeInputs->useDraftLogitsHost = externalDraftTokenParams->useDraftLogitsHost;
|
|
|
|
preparedInputs = decodeInputs;
|
|
preparedOutputs = baseOutputs;
|
|
}
|
|
else if (mDecodingMode.isEagle())
|
|
{
|
|
preparedInputs = baseInputs;
|
|
preparedOutputs = baseOutputs;
|
|
}
|
|
else
|
|
{
|
|
TLLM_CHECK_WITH_INFO(false,
|
|
"Decoding mode is none of the supported {TopK, TopP, TopKTopP, BeamSearch, Medusa, Lookahead, "
|
|
"ExplicitDraftTokens, ExternalDraftTokens, Eagle}");
|
|
}
|
|
TLLM_LOG_TRACE("%s stop", __PRETTY_FUNCTION__);
|
|
return {preparedOutputs, preparedInputs};
|
|
}
|
|
|
|
template class DecodingLayer<float>;
|
|
template class DecodingLayer<half>;
|
|
|
|
} // namespace tensorrt_llm::layers
|