126 static constexpr int32_t kVALUE = 57;
168 static constexpr int32_t kVALUE = 14;
205 mImpl->setName(name);
217 return mImpl->getName();
236 mImpl->setDimensions(dimensions);
250 return mImpl->getDimensions();
265 return mImpl->getType();
273 return mImpl->isNetworkInput();
281 return mImpl->isNetworkOutput();
303 mImpl->setAllowedFormats(formats);
316 return mImpl->getAllowedFormats();
347 return mImpl->isShapeTensor();
368 return mImpl->isExecutionTensor();
394 mImpl->setDimensionName(index, name);
409 return mImpl->getDimensionName(index);
436 return mLayer->getType();
450 mLayer->setName(name);
460 return mLayer->getName();
468 return mLayer->getNbInputs();
481 return mLayer->getInput(index);
489 return mLayer->getNbOutputs();
499 return mLayer->getOutput(index);
516 return mLayer->setInput(index, tensor);
531 return mLayer->getOutputType(index);
550 mLayer->setMetadata(metadata);
563 return mLayer->getMetadata();
584 return mLayer->setNbRanks(nbRanks);
596 return mLayer->getNbRanks();
601 apiv::VLayer* mLayer;
778 static constexpr int32_t kVALUE = 4;
805 mImpl->setNbOutputMaps(nbOutputMaps);
815 return mImpl->getNbOutputMaps();
835 mImpl->setNbGroups(nbGroups);
845 return mImpl->getNbGroups();
859 mImpl->setKernelWeights(weights);
869 return mImpl->getKernelWeights();
884 mImpl->setBiasWeights(weights);
894 return mImpl->getBiasWeights();
911 mImpl->setPrePadding(padding);
921 return mImpl->getPrePadding();
938 mImpl->setPostPadding(padding);
948 return mImpl->getPostPadding();
962 mImpl->setPaddingMode(paddingMode);
974 return mImpl->getPaddingMode();
987 mImpl->setKernelSizeNd(kernelSize);
997 return mImpl->getKernelSizeNd();
1012 mImpl->setStrideNd(stride);
1022 return mImpl->getStrideNd();
1040 mImpl->setPaddingNd(padding);
1052 return mImpl->getPaddingNd();
1066 mImpl->setDilationNd(dilation);
1076 return mImpl->getDilationNd();
1127 mImpl->setActivationType(type);
1137 return mImpl->getActivationType();
1152 mImpl->setAlpha(alpha);
1166 mImpl->setBeta(beta);
1175 return mImpl->getAlpha();
1184 return mImpl->getBeta();
1214 static constexpr int32_t kVALUE = 3;
1240 mImpl->setPoolingType(type);
1250 return mImpl->getPoolingType();
1265 mImpl->setBlendFactor(blendFactor);
1278 return mImpl->getBlendFactor();
1292 mImpl->setAverageCountExcludesPadding(exclusive);
1303 return mImpl->getAverageCountExcludesPadding();
1321 mImpl->setPrePadding(padding);
1331 return mImpl->getPrePadding();
1349 mImpl->setPostPadding(padding);
1359 return mImpl->getPostPadding();
1372 mImpl->setPaddingMode(paddingMode);
1383 return mImpl->getPaddingMode();
1396 mImpl->setWindowSizeNd(windowSize);
1406 return mImpl->getWindowSizeNd();
1421 mImpl->setStrideNd(stride);
1431 return mImpl->getStrideNd();
1450 mImpl->setPaddingNd(padding);
1462 return mImpl->getPaddingNd();
1495 mImpl->setWindowSize(windowSize);
1505 return mImpl->getWindowSize();
1517 mImpl->setAlpha(alpha);
1527 return mImpl->getAlpha();
1539 mImpl->setBeta(beta);
1549 return mImpl->getBeta();
1571 return mImpl->getK();
1601 static constexpr int32_t kVALUE = 3;
1639 mImpl->setMode(mode);
1649 return mImpl->getMode();
1659 mImpl->setShift(shift);
1669 return mImpl->getShift();
1679 mImpl->setScale(scale);
1689 return mImpl->getScale();
1699 mImpl->setPower(power);
1709 return mImpl->getPower();
1724 return mImpl->getChannelAxis();
1745 mImpl->setChannelAxis(channelAxis);
1800 mImpl->setAxes(axes);
1810 return mImpl->getAxes();
1848 mImpl->setAxis(axis);
1858 return mImpl->getAxis();
1887 mImpl->setNbOutputMaps(nbOutputMaps);
1897 return mImpl->getNbOutputMaps();
1917 mImpl->setNbGroups(nbGroups);
1927 return mImpl->getNbGroups();
1941 mImpl->setKernelWeights(weights);
1951 return mImpl->getKernelWeights();
1966 mImpl->setBiasWeights(weights);
1976 return mImpl->getBiasWeights();
1993 mImpl->setPrePadding(padding);
2003 return mImpl->getPrePadding();
2020 mImpl->setPostPadding(padding);
2030 return mImpl->getPostPadding();
2044 mImpl->setPaddingMode(paddingMode);
2056 return mImpl->getPaddingMode();
2071 mImpl->setKernelSizeNd(kernelSize);
2081 return mImpl->getKernelSizeNd();
2098 mImpl->setStrideNd(stride);
2108 return mImpl->getStrideNd();
2126 mImpl->setPaddingNd(padding);
2138 return mImpl->getPaddingNd();
2164 mImpl->setDilationNd(dilation);
2174 return mImpl->getDilationNd();
2222 static constexpr int32_t kVALUE = 14;
2258 return mImpl->setOperation(op);
2270 return mImpl->getOperation();
2300 static constexpr int32_t kVALUE = 3;
2393 mImpl->setGatherAxis(axis);
2405 return mImpl->getGatherAxis();
2428 mImpl->setNbElementWiseDims(elementWiseDims);
2438 return mImpl->getNbElementWiseDims();
2448 mImpl->setMode(mode);
2458 return mImpl->getMode();
2489 return mImpl->getPlugin();
2518 return mImpl->getPlugin();
2581 static constexpr int32_t kVALUE = 25;
2603 mImpl->setOperation(op);
2613 return mImpl->getOperation();
2664 static constexpr int32_t kVALUE = 6;
2694 static constexpr int32_t kVALUE = 8;
2714 mImpl->setOperation(op);
2724 return mImpl->getOperation();
2734 mImpl->setReduceAxes(reduceAxes);
2744 return mImpl->getReduceAxes();
2754 mImpl->setKeepDimensions(keepDimensions);
2764 return mImpl->getKeepDimensions();
2800 mImpl->setPrePaddingNd(padding);
2812 return mImpl->getPrePaddingNd();
2826 mImpl->setPostPaddingNd(padding);
2838 return mImpl->getPostPaddingNd();
2890 mImpl->setFirstTranspose(permutation);
2902 return mImpl->getFirstTranspose();
2930 mImpl->setReshapeDimensions(dimensions);
2943 return mImpl->getReshapeDimensions();
2990 mImpl->setSecondTranspose(permutation);
3002 return mImpl->getSecondTranspose();
3018 return mImpl->setZeroIsPlaceholder(zeroIsPlaceholder);
3031 return mImpl->getZeroIsPlaceholder();
3065 static constexpr int32_t kVALUE = 5;
3144 mImpl->setStart(start);
3159 return mImpl->getStart();
3173 return mImpl->setSize(size);
3188 return mImpl->getSize();
3202 mImpl->setStride(stride);
3217 return mImpl->getStride();
3227 mImpl->setMode(mode);
3237 return mImpl->getMode();
3280 mImpl->setAxes(axes);
3295 return mImpl->getAxes();
3345 static constexpr int32_t kVALUE = 2;
3369 mImpl->setOperation(op);
3379 return mImpl->getOperation();
3407 return mImpl->getK();
3417 mImpl->setReduceAxes(reduceAxes);
3427 return mImpl->getReduceAxes();
3458 return mImpl->setIndicesType(type);
3470 return mImpl->getIndicesType();
3517 static constexpr int32_t kVALUE = 3;
3558 mImpl->setOperation(index, op);
3570 return mImpl->getOperation(index);
3616 return mImpl->setIndicesType(type);
3628 return mImpl->getIndicesType();
3731 mImpl->setToType(toType);
3742 return mImpl->getToType();
3773 mImpl->setWeights(weights);
3783 return mImpl->getWeights();
3795 mImpl->setDimensions(dimensions);
3807 return mImpl->getDimensions();
3855 static constexpr int32_t kVALUE = 3;
3906 static constexpr int32_t kVALUE = 3;
3933 static constexpr int32_t kVALUE = 2;
3966 static constexpr int32_t kVALUE = 4;
4028 return mImpl->setOutputDimensions(dimensions);
4038 return mImpl->getOutputDimensions();
4066 void setScales(
float const* scales, int32_t nbScales)
noexcept
4068 mImpl->setScales(scales, nbScales);
4085 int32_t
getScales(int32_t size,
float* scales)
const noexcept
4087 return mImpl->getScales(size, scales);
4099 mImpl->setResizeMode(interpolationMode);
4109 return mImpl->getResizeMode();
4144 mImpl->setCoordinateTransformation(coordTransform);
4154 return mImpl->getCoordinateTransformation();
4169 mImpl->setSelectorForSinglePixel(selector);
4179 return mImpl->getSelectorForSinglePixel();
4193 mImpl->setNearestRounding(value);
4203 return mImpl->getNearestRounding();
4225 mImpl->setCubicCoeff(A);
4235 return mImpl->getCubicCoeff();
4248 mImpl->setExcludeOutside(excludeFlag);
4258 return mImpl->getExcludeOutside();
4293 static constexpr int32_t kVALUE = 3;
4316 static constexpr int32_t kVALUE = 2;
4342 return mBoundary->getLoop();
4347 apiv::VLoopBoundaryLayer* mBoundary;
4367 return mBoundary->getConditional();
4372 apiv::VConditionalBoundaryLayer* mBoundary;
4464 return mImpl->setCondition(condition);
4482 return mImpl->addOutput(trueSubgraphOutput, falseSubgraphOutput);
4494 return mImpl->addInput(input);
4509 mImpl->setName(name);
4519 return mImpl->getName();
4593 return mImpl->getLoopOutput();
4610 mImpl->setAxis(axis);
4618 return mImpl->getAxis();
4669 return mImpl->getTripLimit();
4697 mImpl->setAxis(axis);
4705 return mImpl->getAxis();
4719 mImpl->setReverse(reverse);
4729 return mImpl->getReverse();
4760 return mImpl->addRecurrence(initialValue);
4781 return mImpl->addTripLimit(tensor, limit);
4794 return mImpl->addIterator(tensor, axis, reverse);
4807 return mImpl->addLoopOutput(tensor, outputKind, axis);
4822 mImpl->setName(name);
4832 return mImpl->getName();
4891 mImpl->setMessage(message);
4901 return mImpl->getMessage();
4956 static constexpr int32_t kVALUE = 3;
5008 mImpl->setDimensions(dimensions);
5023 return mImpl->getDimensions();
5033 mImpl->setOperation(op);
5043 return mImpl->getOperation();
5062 mImpl->setAlpha(alpha);
5077 return mImpl->getAlpha();
5096 mImpl->setBeta(beta);
5111 return mImpl->getBeta();
5172 mImpl->setAlphaInt64(alpha);
5187 return mImpl->getAlphaInt64();
5206 mImpl->setBetaInt64(beta);
5221 return mImpl->getBetaInt64();
5229 return mImpl->isAlphaBetaInt64();
5247 mImpl->setToType(toType);
5259 return mImpl->getToType();
5356 return mImpl->getAxis();
5367 mImpl->setAxis(axis);
5380 return mImpl->setBlockShape(blockShape);
5391 return mImpl->getBlockShape();
5407 mImpl->setToType(toType);
5419 return mImpl->getToType();
5510 return mImpl->getAxis();
5521 mImpl->setAxis(axis);
5538 return mImpl->setBlockShape(blockShape);
5549 return mImpl->getBlockShape();
5565 mImpl->setToType(toType);
5577 return mImpl->getToType();
5634 mImpl->setToType(toType);
5647 return mImpl->getToType();
5660 mImpl->setScaleType(scaleType);
5673 return mImpl->getScaleType();
5686 mImpl->setAxis(axis);
5696 return mImpl->getAxis();
5709 mImpl->setBlockSize(size);
5719 return mImpl->getBlockSize();
5732 mImpl->setBlockShape(blockShape);
5744 return mImpl->getBlockShape();
5802 return mImpl->setEquation(equation);
5812 return mImpl->getEquation();
5843 static constexpr int32_t kVALUE = 2;
5913 mImpl->setMode(mode);
5923 return mImpl->getMode();
5933 mImpl->setAxis(axis);
5941 return mImpl->getAxis();
5988 mImpl->setAxis(axis);
5996 return mImpl->getAxis();
6027 mImpl->setInterpolationMode(mode);
6039 return mImpl->getInterpolationMode();
6049 mImpl->setAlignCorners(alignCorners);
6061 return mImpl->getAlignCorners();
6073 return mImpl->setSampleMode(mode);
6085 return mImpl->getSampleMode();
6118 static constexpr int32_t kVALUE = 2;
6185 mImpl->setBoundingBoxFormat(fmt);
6197 return mImpl->getBoundingBoxFormat();
6211 mImpl->setTopKBoxLimit(limit);
6221 return mImpl->getTopKBoxLimit();
6256 return mImpl->setIndicesType(type);
6268 return mImpl->getIndicesType();
6303 mImpl->setBatchAxis(batchAxis);
6313 return mImpl->getBatchAxis();
6326 mImpl->setSequenceAxis(sequenceAxis);
6336 return mImpl->getSequenceAxis();
6376 return mImpl->setEpsilon(eps);
6386 return mImpl->getEpsilon();
6396 return mImpl->setAxes(axesMask);
6406 return mImpl->getAxes();
6427 return mImpl->setNbGroups(nbGroups);
6437 return mImpl->getNbGroups();
6448 return mImpl->isV2();
6548 static constexpr int32_t kVALUE = 1;
6592 return mImpl->setOperation(op);
6604 return mImpl->getOperation();
6616 mImpl->setExclusive(exclusive);
6628 return mImpl->getExclusive();
6640 mImpl->setReverse(reverse);
6652 return mImpl->getReverse();
6682 static constexpr int32_t kVALUE = 2;
6719 static constexpr int32_t kVALUE = 3;
6745 static constexpr int32_t kVALUE = 2;
6765 return mBoundary->getAttention();
6770 apiv::VAttentionBoundaryLayer* mBoundary;
6906 return mImpl->setNormalizationOperation(op);
6918 return mImpl->getNormalizationOperation();
6935 return mImpl->setMask(mask);
6947 return mImpl->getMask();
6965 return mImpl->setCausal(isCausal);
6979 return mImpl->getCausal();
6999 return mImpl->setCausalKind(kind);
7011 return mImpl->getCausalKind();
7023 return mImpl->setDecomposable(decomposable);
7036 return mImpl->getDecomposable();
7055 return mImpl->setInput(index, input);
7064 return mImpl->getNbInputs();
7076 return mImpl->getInput(index);
7084 return mImpl->getNbOutputs();
7096 return mImpl->getOutput(index);
7113 return mImpl->setName(name);
7125 return mImpl->getName();
7141 return mImpl->setNormalizationQuantizeScale(tensor);
7152 return mImpl->getNormalizationQuantizeScale();
7165 return mImpl->setNormalizationQuantizeToType(type);
7177 return mImpl->getNormalizationQuantizeToType();
7197 return mImpl->setMetadata(metadata);
7210 return mImpl->getMetadata();
7226 return mImpl->setNbRanks(nbRanks);
7238 return mImpl->getNbRanks();
7255 return mImpl->setQueryForm(form);
7268 return mImpl->getQueryForm();
7285 return mImpl->setKeyValueForm(form);
7298 return mImpl->getKeyValueForm();
7322 return mImpl->setQueryLengths(lengths);
7334 return mImpl->getQueryLengths();
7361 return mImpl->setKeyValueLengths(lengths);
7373 return mImpl->getKeyValueLengths();
7399 mImpl->setInterleaved(interleaved);
7410 return mImpl->getInterleaved();
7421 return mImpl->setRotaryEmbeddingDim(rotaryEmbeddingDim);
7432 return mImpl->getRotaryEmbeddingDim();
7476 static constexpr int32_t kVALUE = 1;
7526 return mImpl->setCacheMode(cacheMode);
7536 return mImpl->getCacheMode();
7554 return mImpl->setUpdateForm(form);
7567 return mImpl->getUpdateForm();
7589 return mImpl->setUpdateLengths(lengths);
7601 return mImpl->getUpdateLengths();
7630 static constexpr int32_t kVALUE = 2;
7765 mImpl->setGatedWeights(fcGateWeights, fcUpWeights, fcDownWeights, activationType);
7777 mImpl->setGatedBiases(fcGateBiases, fcUpBiases, fcDownBiases);
7789 mImpl->setActivationType(activationType);
7801 return mImpl->getActivationType();
7827 mImpl->setQuantizationStatic(fcDownActivationScale, dataType);
7860 mImpl->setQuantizationDynamicDblQ(fcDownActivationDblQScale, dataType, blockShape, dynQOutputScaleType);
7875 mImpl->setQuantizationToType(type);
7887 return mImpl->getQuantizationToType();
7903 mImpl->setQuantizationBlockShape(blockShape);
7915 return mImpl->getQuantizationBlockShape();
7927 mImpl->setDynQOutputScaleType(type);
7939 return mImpl->getDynQOutputScaleType();
7960 mImpl->setSwigluParams(limit, alpha, beta);
7974 mImpl->setSwigluParamLimit(limit);
7986 return mImpl->getSwigluParamLimit();
8000 mImpl->setSwigluParamAlpha(alpha);
8012 return mImpl->getSwigluParamAlpha();
8026 mImpl->setSwigluParamBeta(beta);
8038 return mImpl->getSwigluParamBeta();
8055 mImpl->setInput(index, tensor);
8142 return mImpl->addInput(name, type, dimensions);
8156 mImpl->markOutput(tensor);
8174 return mImpl->markDebug(tensor);
8190 return mImpl->unmarkDebug(tensor);
8200 return mImpl->isDebugTensor(tensor);
8222 return mImpl->markUnfusedTensorsAsDebugTensors();
8236 return mImpl->unmarkUnfusedTensorsAsDebugTensors();
8256 return mImpl->addActivation(input, type);
8275 return mImpl->addLRN(input, window, alpha, beta, k);
8301 return mImpl->addScale(input, mode, shift, scale, power);
8314 return mImpl->addSoftMax(input);
8331 return mImpl->addConcatenation(inputs, nbInputs);
8358 return mImpl->addElementWise(input1, input2, op);
8380 return mImpl->addUnary(input, operation);
8394 return mImpl->addShuffle(input);
8411 return mImpl->addOneHot(indices, values, depth, axis);
8423 return mImpl->getNbLayers();
8437 return mImpl->getLayer(index);
8449 return mImpl->getNbInputs();
8465 return mImpl->getInput(index);
8479 return mImpl->getNbOutputs();
8495 return mImpl->getOutput(index);
8522 return mImpl->addReduce(input, operation, reduceAxes, keepDimensions);
8557 return mImpl->addTopK(input, op, k, reduceAxes);
8590 return mImpl->addTopKV2(input, op, k, reduceAxes, indicesType);
8606 return mImpl->addGather(data, indices, axis);
8622 return mImpl->addGatherV2(data, indices, mode);
8641 return mImpl->addRaggedSoftMax(input, bounds);
8663 return mImpl->addMatrixMultiply(input0, op0, input1, op1);
8681 return mImpl->addNonZero(input);
8697 return mImpl->addNonZeroV2(input, indicesType);
8721 return mImpl->addConstant(dimensions, weights);
8735 return mImpl->addIdentity(input);
8750 return mImpl->addCast(input, toType);
8765 mImpl->removeTensor(tensor);
8777 mImpl->unmarkOutput(tensor);
8796 return mImpl->addSlice(input, start, size, stride);
8820 mImpl->setName(name);
8834 return mImpl->getName();
8850 return mImpl->addShape(input);
8860 return mImpl->getFlags();
8872 return mImpl->getFlag(networkDefinitionCreationFlag);
8889 return mImpl->markOutputForShapes(tensor);
8901 return mImpl->unmarkOutputForShapes(tensor);
8919 return mImpl->addParametricReLU(input, slope);
8942 return mImpl->addConvolutionNd(input, nbOutputMaps, kernelSize, kernelWeights, biasWeights);
8961 return mImpl->addPoolingNd(input, type, windowSize);
8984 return mImpl->addDeconvolutionNd(input, nbOutputMaps, kernelSize, kernelWeights, biasWeights);
9021 return mImpl->addScaleNd(input, mode, shift, scale, power, channelAxis);
9037 return mImpl->addResize(input);
9051 return mImpl->addLoop();
9066 return mImpl->addIfConditional();
9105 return mImpl->addSelect(condition, thenInput, elseInput);
9122 return mImpl->addAssertion(condition, message);
9148 return mImpl->addFillV2(dimensions, op, outputType);
9164 return mImpl->addPaddingNd(input, prePadding, postPadding);
9188 return mImpl->setWeightsName(weights, name);
9207 mImpl->setErrorRecorder(recorder);
9222 return mImpl->getErrorRecorder();
9245 return mImpl->addDequantizeV2(input, scale, outputType);
9265 return mImpl->addScatter(data, indices, updates, mode);
9289 return mImpl->addQuantizeV2(input, scale, outputType);
9317 return mImpl->addDynamicQuantize(input, axis, blockSize, outputType, scaleType);
9341 return mImpl->addDynamicQuantizeV2(input, blockShape, outputType, scaleType);
9356 return mImpl->addEinsum(inputs, nbInputs, equation);
9374 return mImpl->addGridSample(input, grid);
9396 return mImpl->addNMS(boxes, scores, maxOutputBoxesPerClass);
9416 return mImpl->addNMSV2(boxes, scores, maxOutputBoxesPerClass, indicesType);
9433 return mImpl->addReverseSequence(input, sequenceLens);
9465 return mImpl->addNormalization(input, scale, bias, axesMask);
9487 return mImpl->addCumulative(input, axis, operation, exclusive, reverse);
9518 return mImpl->addAttention(query, key, value, normOp, causal);
9548 return mImpl->addAttentionV2(query, key, value, normOp, causalKind);
9572 return mImpl->addRotaryEmbedding(input, cosCache, sinCache, interleaved, rotaryEmbeddingDim);
9607 return mImpl->addKVCacheUpdate(cache, update, writeIndices, cacheMode);
9628 return mImpl->addMoE(hiddenStates, selectedExpertsForTokens, scoresForSelectedExperts);
9659 ReduceOperation reduceOp, int64_t root, int64_t* groups, int64_t groupSize)
noexcept
9661 return mImpl->addDistCollective(input, distCollectiveOp, reduceOp, root, groups, groupSize);
9672 return mImpl->getBuilder();
9685 return mImpl->markWeightsRefittable(name);
9697 return mImpl->unmarkWeightsRefittable(name);
9710 return mImpl->areWeightsMarkedRefittable(name);
9729 return mImpl->addSqueeze(input, axes);
9750 return mImpl->addUnsqueeze(input, axes);
9776 return mImpl->addNormalizationV2(input, scale, bias, axesMask);
9823 static constexpr int32_t kVALUE = 2;
10001 static constexpr int32_t kVALUE = 29;
10041 static constexpr uint64_t kINVALID_TACTIC_HASH = UINT64_MAX;
10076 return mImpl->serialize();
10100 return mImpl->combine(inputCache, ignoreMismatch);
10110 return mImpl->reset();
10127 int64_t
queryKeys(TimingCacheKey* keyBuffer, int64_t capacity)
const noexcept
10129 return mImpl->queryKeys(keyBuffer, capacity);
10144 TimingCacheValue
query(TimingCacheKey
const& key)
const noexcept
10146 return mImpl->query(key);
10166 bool update(TimingCacheKey
const& key, TimingCacheValue
const& value)
noexcept
10168 return mImpl->update(key, value);
10247 static constexpr int32_t kVALUE = 6;
10281 static constexpr int32_t kVALUE = 2;
10330 static constexpr int32_t kVALUE = 3;
10391 static constexpr int32_t kVALUE = 4;
10429 virtual void phaseStart(
char const* phaseName,
char const* parentPhase, int32_t nbSteps)
noexcept = 0;
10502 virtual
void setAvgTimingIterations(int32_t avgTiming) noexcept
10504 mImpl->setAvgTimingIterations(avgTiming);
10516 return mImpl->getAvgTimingIterations();
10529 mImpl->setEngineCapability(capability);
10541 return mImpl->getEngineCapability();
10558 mImpl->setFlags(builderFlags);
10570 return mImpl->getFlags();
10582 mImpl->clearFlag(builderFlag);
10594 mImpl->setFlag(builderFlag);
10606 return mImpl->getFlag(builderFlag);
10623 mImpl->setDeviceType(layer, deviceType);
10633 return mImpl->getDeviceType(layer);
10645 return mImpl->isDeviceTypeSet(layer);
10655 mImpl->resetDeviceType(layer);
10665 return mImpl->canRunOnDLA(layer);
10681 mImpl->setDLACore(dlaCore);
10691 return mImpl->getDLACore();
10702 mImpl->setDefaultDeviceType(deviceType);
10712 return mImpl->getDefaultDeviceType();
10734 return mImpl->setProfileStream(stream);
10746 return mImpl->getProfileStream();
10763 return mImpl->addOptimizationProfile(profile);
10776 return mImpl->getNbOptimizationProfiles();
10788 mImpl->setProfilingVerbosity(verbosity);
10801 return mImpl->getProfilingVerbosity();
10823 return mImpl->setTacticSources(tacticSources);
10838 return mImpl->getTacticSources();
10860 return mImpl->createTimingCache(blob, size);
10885 return mImpl->setTimingCache(cache, ignoreMismatch);
10897 return mImpl->getTimingCache();
10929 mImpl->setMemoryPoolLimit(pool, poolSize);
10948 return mImpl->getMemoryPoolLimit(pool);
10966 mImpl->setPreviewFeature(feature, enable);
10980 return mImpl->getPreviewFeature(feature);
11013 mImpl->setBuilderOptimizationLevel(level);
11025 return mImpl->getBuilderOptimizationLevel();
11042 mImpl->setHardwareCompatibilityLevel(hardwareCompatibilityLevel);
11055 return mImpl->getHardwareCompatibilityLevel();
11068 mImpl->setPluginsToSerialize(paths, nbPaths);
11081 return mImpl->getPluginToSerialize(index);
11091 return mImpl->getNbPluginsToSerialize();
11122 return mImpl->setMaxAuxStreams(nbStreams);
11132 return mImpl->getMaxAuxStreams();
11148 return mImpl->setProgressMonitor(monitor);
11158 return mImpl->getProgressMonitor();
11174 mImpl->setRuntimePlatform(runtimePlatform);
11186 return mImpl->getRuntimePlatform();
11198 mImpl->setMaxNbTactics(maxNbTactics);
11210 return mImpl->getMaxNbTactics();
11226 return mImpl->setTilingOptimizationLevel(level);
11238 return mImpl->getTilingOptimizationLevel();
11254 return mImpl->setL2LimitForTiling(size);
11266 return mImpl->getL2LimitForTiling();
11285 return mImpl->setNbComputeCapabilities(maxNbComputeCapabilities);
11297 return mImpl->getNbComputeCapabilities();
11315 return mImpl->setComputeCapability(computeCapability, index);
11329 return mImpl->getComputeCapability(index);
11375 static constexpr int32_t kVALUE = 1;
11397 int32_t getMaxDLABatchSize() const noexcept
11399 return mImpl->getMaxDLABatchSize();
11407 return mImpl->getNbDLACores();
11425 mImpl->setGpuAllocator(allocator);
11439 return mImpl->createBuilderConfig();
11465 return mImpl->createNetworkV2(flags);
11480 return mImpl->createOptimizationProfile();
11499 mImpl->setErrorRecorder(recorder);
11514 return mImpl->getErrorRecorder();
11541 return mImpl->buildSerializedNetwork(network, config);
11563 return mImpl->buildSerializedNetworkToStream(network, config, writer);
11590 return mImpl->isNetworkSupported(network, config);
11600 return mImpl->getLogger();
11616 return mImpl->setMaxThreads(maxThreads);
11630 return mImpl->getMaxThreads();
11640 return mImpl->getPluginRegistry();
11655extern "C"
TENSORRTAPI void* createInferBuilder_INTERNAL(
void* logger, int32_t version) noexcept;
#define TENSORRTAPI
Definition: NvInferRuntimeBase.h:70
#define NV_TENSORRT_VERSION
Definition: NvInferRuntimeBase.h:102
#define TRT_NODISCARD
A stand-in for [[nodiscard]] and [[nodiscard(REASON)]] that works with older compilers.
Definition: NvInferRuntimeBase.h:57
#define TRT_DEPRECATED
Definition: NvInferRuntimeBase.h:42
#define TRT_DEPRECATED_ENUM
Definition: NvInferRuntimeBase.h:43
Definition: NvInferRuntimeBase.h:222
static constexpr int32_t MAX_DIMS
The maximum rank (number of dimensions) supported for a tensor.
Definition: NvInferRuntimeBase.h:225
An Activation layer in a network definition.
Definition: NvInfer.h:1116
void setBeta(float beta) noexcept
Set the beta parameter (must be finite).
Definition: NvInfer.h:1164
void setActivationType(ActivationType type) noexcept
Set the type of activation to be performed.
Definition: NvInfer.h:1125
ActivationType getActivationType() const noexcept
Get the type of activation to be performed.
Definition: NvInfer.h:1135
float getAlpha() const noexcept
Get the alpha parameter.
Definition: NvInfer.h:1173
float getBeta() const noexcept
Get the beta parameter.
Definition: NvInfer.h:1182
void setAlpha(float alpha) noexcept
Set the alpha parameter (must be finite).
Definition: NvInfer.h:1150
virtual ~IActivationLayer() noexcept=0
An assertion layer in a network.
Definition: NvInfer.h:4879
void setMessage(char const *message) noexcept
Set the message to print if the assertion fails.
Definition: NvInfer.h:4889
char const * getMessage() const noexcept
Return the assertion message.
Definition: NvInfer.h:4899
virtual ~IAssertionLayer() noexcept=0
This is a base class for Attention boundary layers.
Definition: NvInfer.h:6758
IAttention * getAttention() const noexcept
Get a pointer to the IAttention associated with this boundary layer.
Definition: NvInfer.h:6763
virtual ~IAttentionBoundaryLayer() noexcept=0
Helper for constructing an attention that consumes query, key and value tensors.
Definition: NvInfer.h:6895
ITensor * getMask() noexcept
Get the optional mask in attention.
Definition: NvInfer.h:6945
bool setMetadata(char const *metadata) noexcept
Set the metadata for IAttention.
Definition: NvInfer.h:7195
TRT_NODISCARD bool setQueryLengths(ITensor *lengths) noexcept
Set the query lengths tensor.
Definition: NvInfer.h:7320
bool setDecomposable(bool decomposable) noexcept
Set whether the attention can be decomposed to use multiple kernels if no fused kernel support found.
Definition: NvInfer.h:7021
bool setName(char const *name) noexcept
Set the name of the attention.
Definition: NvInfer.h:7111
bool getDecomposable() const noexcept
Get whether the attention can be decomposed to use multiple kernels if no fused kernel support found.
Definition: NvInfer.h:7034
ITensor * getInput(int32_t index) const noexcept
Get the IAttention input corresponding to the given index.
Definition: NvInfer.h:7074
CausalMaskKind getCausalKind() const noexcept
Get the causal mask alignment orientation for the attention.
Definition: NvInfer.h:7009
ITensor * getOutput(int32_t index) const noexcept
Get the IAttention output corresponding to the given index. IAttention has only one output.
Definition: NvInfer.h:7094
int32_t getNbOutputs() const noexcept
Get the number of outputs of a layer. IAttention has one output.
Definition: NvInfer.h:7082
bool setNbRanks(int32_t nbRanks) noexcept
Set the number of ranks for multi-device attention execution.
Definition: NvInfer.h:7224
int32_t getNbInputs() const noexcept
Get the number of inputs of IAttention. IAttention has three inputs.
Definition: NvInfer.h:7062
TRT_NODISCARD bool setKeyValueLengths(ITensor *lengths) noexcept
Set the key-value lengths tensor.
Definition: NvInfer.h:7359
TRT_NODISCARD ITensor * getKeyValueLengths() const noexcept
Get the key-value lengths tensor.
Definition: NvInfer.h:7371
bool setNormalizationOperation(AttentionNormalizationOp op) noexcept
Set the normalization operation for the attention.
Definition: NvInfer.h:6904
TRT_NODISCARD AttentionIOForm getKeyValueForm() const noexcept
Get the key-value form.
Definition: NvInfer.h:7296
char const * getName() const noexcept
Return the name of the attention.
Definition: NvInfer.h:7123
bool setNormalizationQuantizeToType(DataType type) noexcept
Set the datatype the attention normalization is quantized to.
Definition: NvInfer.h:7163
int32_t getNbRanks() const noexcept
Get the number of ranks for multi-device execution.
Definition: NvInfer.h:7236
TRT_DEPRECATED bool getCausal() const noexcept
Get whether the attention will run a causal inference.
Definition: NvInfer.h:6977
AttentionNormalizationOp getNormalizationOperation() const noexcept
Get the normalization operation for the attention.
Definition: NvInfer.h:6916
bool setNormalizationQuantizeScale(ITensor &tensor) noexcept
Set the quantization scale for the attention normalization output.
Definition: NvInfer.h:7139
bool setCausalKind(CausalMaskKind kind) noexcept
Set the causal mask alignment orientation for the attention.
Definition: NvInfer.h:6997
TRT_NODISCARD AttentionIOForm getQueryForm() const noexcept
Get the query form.
Definition: NvInfer.h:7266
char const * getMetadata() const noexcept
Get the metadata of IAttention.
Definition: NvInfer.h:7208
DataType getNormalizationQuantizeToType() const noexcept
Get the datatype the attention normalization is quantized to.
Definition: NvInfer.h:7175
TRT_NODISCARD bool setQueryForm(AttentionIOForm form) noexcept
Set the query form.
Definition: NvInfer.h:7253
virtual ~IAttention() noexcept=0
ITensor * getNormalizationQuantizeScale() const noexcept
Get the quantization scale for the attention normalization output.
Definition: NvInfer.h:7150
bool setInput(int32_t index, ITensor &input) noexcept
Append or replace an input of this layer with a specific tensor.
Definition: NvInfer.h:7053
TRT_NODISCARD ITensor * getQueryLengths() const noexcept
Get the query lengths tensor.
Definition: NvInfer.h:7332
bool setMask(ITensor &mask) noexcept
Set whether a mask will be used for the normalization operation.
Definition: NvInfer.h:6933
TRT_NODISCARD bool setKeyValueForm(AttentionIOForm form) noexcept
Set the key-value form.
Definition: NvInfer.h:7283
TRT_DEPRECATED bool setCausal(bool isCausal) noexcept
Set whether the attention will run a causal inference. Cannot be used together with setMask().
Definition: NvInfer.h:6963
apiv::VAttention * mImpl
Definition: NvInfer.h:7377
This layer represents an output of an IAttention.
Definition: NvInfer.h:6825
virtual ~IAttentionOutputLayer() noexcept=0
Holds properties for configuring a builder to produce an engine.
Definition: NvInfer.h:10490
void setMemoryPoolLimit(MemoryPoolType pool, std::size_t poolSize) noexcept
Set the memory size for the memory pool.
Definition: NvInfer.h:10927
bool setComputeCapability(ComputeCapability computeCapability, int32_t index) noexcept
Set one compute capability for runtime execution.
Definition: NvInfer.h:11313
bool setNbComputeCapabilities(int32_t maxNbComputeCapabilities) noexcept
Set the number of compute capabilities.
Definition: NvInfer.h:11283
TRT_DEPRECATED bool setTimingCache(ITimingCache const &cache, bool ignoreMismatch) noexcept
Attach a timing cache to IBuilderConfig.
Definition: NvInfer.h:10883
bool setMaxAuxStreams(int32_t nbStreams) noexcept
Set the maximum number of auxiliary streams that TRT is allowed to use.
Definition: NvInfer.h:11120
void setPreviewFeature(PreviewFeature feature, bool enable) noexcept
Enable or disable a specific preview feature.
Definition: NvInfer.h:10964
bool getPreviewFeature(PreviewFeature feature) const noexcept
Get status of preview feature.
Definition: NvInfer.h:10978
int32_t getBuilderOptimizationLevel() noexcept
Get builder optimization level.
Definition: NvInfer.h:11023
bool setTacticSources(TacticSources tacticSources) noexcept
Set tactic sources.
Definition: NvInfer.h:10821
void setPluginsToSerialize(char const *const *paths, int32_t nbPaths) noexcept
Set the plugin libraries to be serialized with version-compatible engines.
Definition: NvInfer.h:11066
bool setTilingOptimizationLevel(TilingOptimizationLevel level) noexcept
Set the Tiling optimization level.
Definition: NvInfer.h:11224
bool setL2LimitForTiling(int64_t size) noexcept
Set the L2 cache usage limit for Tiling optimization.
Definition: NvInfer.h:11252
std::size_t getMemoryPoolLimit(MemoryPoolType pool) const noexcept
Get the memory size limit of the memory pool.
Definition: NvInfer.h:10946
int32_t getDLACore() const noexcept
Get the DLA core that the engine executes on.
Definition: NvInfer.h:10689
int32_t getNbPluginsToSerialize() const noexcept
Get the number of plugin library paths to be serialized with version-compatible engines.
Definition: NvInfer.h:11089
void setDeviceType(ILayer const *layer, DeviceType deviceType) noexcept
Set the device that this layer must execute on.
Definition: NvInfer.h:10621
void setEngineCapability(EngineCapability capability) noexcept
Configure the builder to target specified EngineCapability flow.
Definition: NvInfer.h:10527
virtual ~IBuilderConfig() noexcept=0
int32_t getMaxAuxStreams() const noexcept
Get the maximum number of auxiliary streams that TRT is allowed to use.
Definition: NvInfer.h:11130
bool getFlag(BuilderFlag builderFlag) const noexcept
Returns true if the build mode flag is set.
Definition: NvInfer.h:10604
void setMaxNbTactics(int32_t maxNbTactics) noexcept
Set the maximum number of tactics to time when there is a choice of tactics.
Definition: NvInfer.h:11196
int64_t getL2LimitForTiling() const noexcept
Get the L2 cache usage limit for tiling optimization.
Definition: NvInfer.h:11264
void setProgressMonitor(IProgressMonitor *monitor) noexcept
Sets the progress monitor for building a network.
Definition: NvInfer.h:11146
void setProfilingVerbosity(ProfilingVerbosity verbosity) noexcept
Set verbosity level of layer information exposed in NVTX annotations and IEngineInspector.
Definition: NvInfer.h:10786
int32_t getNbOptimizationProfiles() const noexcept
Get number of optimization profiles.
Definition: NvInfer.h:10774
void reset() noexcept
Resets the builder configuration to defaults.
Definition: NvInfer.h:10720
char const * getPluginToSerialize(int32_t index) const noexcept
Get the plugin library path to be serialized with version-compatible engines.
Definition: NvInfer.h:11079
EngineCapability getEngineCapability() const noexcept
Query EngineCapability flow configured for the builder.
Definition: NvInfer.h:10539
RuntimePlatform getRuntimePlatform() const noexcept
Get the target platform for runtime execution.
Definition: NvInfer.h:11184
DeviceType getDefaultDeviceType() const noexcept
Get the default DeviceType which was set by setDefaultDeviceType.
Definition: NvInfer.h:10710
void setRuntimePlatform(RuntimePlatform runtimePlatform) noexcept
Set the target platform for runtime execution.
Definition: NvInfer.h:11172
int32_t getMaxNbTactics() const noexcept
Query the maximum number of tactics timed when there is a choice.
Definition: NvInfer.h:11208
BuilderFlags getFlags() const noexcept
Get the build mode flags for this builder config. Defaults to 0.
Definition: NvInfer.h:10568
void setFlags(BuilderFlags builderFlags) noexcept
Set the build mode flags to turn on builder options for this network.
Definition: NvInfer.h:10556
TacticSources getTacticSources() const noexcept
Get tactic sources.
Definition: NvInfer.h:10836
void resetDeviceType(ILayer const *layer) noexcept
reset the DeviceType for this layer
Definition: NvInfer.h:10653
ComputeCapability getComputeCapability(int32_t index) const noexcept
Get one compute capability for runtime execution.
Definition: NvInfer.h:11327
void setDLACore(int32_t dlaCore) noexcept
Sets the DLA core used by the network. Defaults to -1.
Definition: NvInfer.h:10679
HardwareCompatibilityLevel getHardwareCompatibilityLevel() const noexcept
Get the hardware compatibility level.
Definition: NvInfer.h:11053
int32_t getNbComputeCapabilities() const noexcept
Get the number of compute capabilities.
Definition: NvInfer.h:11295
void clearFlag(BuilderFlag builderFlag) noexcept
clear a single build mode flag.
Definition: NvInfer.h:10580
int32_t addOptimizationProfile(IOptimizationProfile const *profile) noexcept
Add an optimization profile.
Definition: NvInfer.h:10761
IProgressMonitor * getProgressMonitor() const noexcept
Definition: NvInfer.h:11156
apiv::VBuilderConfig * mImpl
Definition: NvInfer.h:11333
int32_t getAvgTimingIterations() const noexcept
Query the number of averaging iterations.
Definition: NvInfer.h:10514
void setDefaultDeviceType(DeviceType deviceType) noexcept
Sets the default DeviceType to be used by the builder. It ensures that all the layers that can run on...
Definition: NvInfer.h:10700
void setFlag(BuilderFlag builderFlag) noexcept
Set a single build mode flag.
Definition: NvInfer.h:10592
TRT_DEPRECATED nvinfer1::ITimingCache * createTimingCache(void const *blob, std::size_t size) const noexcept
Create timing cache.
Definition: NvInfer.h:10858
DeviceType getDeviceType(ILayer const *layer) const noexcept
Get the device that this layer executes on.
Definition: NvInfer.h:10631
bool canRunOnDLA(ILayer const *layer) const noexcept
Checks if a layer can run on DLA.
Definition: NvInfer.h:10663
cudaStream_t getProfileStream() const noexcept
Get the CUDA stream that is used to profile this network.
Definition: NvInfer.h:10744
void setHardwareCompatibilityLevel(HardwareCompatibilityLevel hardwareCompatibilityLevel) noexcept
Set the hardware compatibility level.
Definition: NvInfer.h:11040
TilingOptimizationLevel getTilingOptimizationLevel() const noexcept
Get the Tiling optimization level.
Definition: NvInfer.h:11236
ProfilingVerbosity getProfilingVerbosity() const noexcept
Get verbosity level of layer information exposed in NVTX annotations and IEngineInspector.
Definition: NvInfer.h:10799
TRT_DEPRECATED nvinfer1::ITimingCache const * getTimingCache() const noexcept
Get the pointer to the timing cache from current IBuilderConfig.
Definition: NvInfer.h:10895
bool isDeviceTypeSet(ILayer const *layer) const noexcept
whether the DeviceType has been explicitly set for this layer
Definition: NvInfer.h:10643
void setBuilderOptimizationLevel(int32_t level) noexcept
Set builder optimization level.
Definition: NvInfer.h:11011
void setProfileStream(const cudaStream_t stream) noexcept
Set the CUDA stream that is used to profile this network.
Definition: NvInfer.h:10732
Builds an engine from a network definition.
Definition: NvInfer.h:11386
int32_t getNbDLACores() const noexcept
Return the number of DLA engines available to this builder.
Definition: NvInfer.h:11405
virtual ~IBuilder() noexcept=0
IErrorRecorder * getErrorRecorder() const noexcept
get the ErrorRecorder assigned to this interface.
Definition: NvInfer.h:11512
apiv::VBuilder * mImpl
Definition: NvInfer.h:11644
ILogger * getLogger() const noexcept
get the logger with which the builder was created
Definition: NvInfer.h:11598
bool isNetworkSupported(INetworkDefinition const &network, IBuilderConfig const &config) const noexcept
Checks that a network is within the scope of the IBuilderConfig settings.
Definition: NvInfer.h:11588
int32_t getMaxThreads() const noexcept
get the maximum number of threads that can be used by the builder.
Definition: NvInfer.h:11628
IPluginRegistry & getPluginRegistry() noexcept
get the local plugin registry that can be used by the builder.
Definition: NvInfer.h:11638
nvinfer1::IOptimizationProfile * createOptimizationProfile() noexcept
Create a new optimization profile.
Definition: NvInfer.h:11478
void setGpuAllocator(IGpuAllocator *allocator) noexcept
Set the GPU allocator.
Definition: NvInfer.h:11423
nvinfer1::INetworkDefinition * createNetworkV2(NetworkDefinitionCreationFlags flags) noexcept
Create a network definition object.
Definition: NvInfer.h:11463
nvinfer1::IBuilderConfig * createBuilderConfig() noexcept
Create a builder configuration object.
Definition: NvInfer.h:11437
void reset() noexcept
Resets the builder state to default values.
Definition: NvInfer.h:11520
bool setMaxThreads(int32_t maxThreads) noexcept
Set the maximum number of threads.
Definition: NvInfer.h:11614
void setErrorRecorder(IErrorRecorder *recorder) noexcept
Set the ErrorRecorder for this interface.
Definition: NvInfer.h:11497
nvinfer1::IHostMemory * buildSerializedNetwork(INetworkDefinition &network, IBuilderConfig &config) noexcept
Builds and serializes a network for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:11539
bool buildSerializedNetworkToStream(INetworkDefinition &network, IBuilderConfig &config, IStreamWriter &writer) noexcept
Builds and serializes a network into stream for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:11560
A cast layer in a network.
Definition: NvInfer.h:3720
apiv::VCastLayer * mImpl
Definition: NvInfer.h:3746
DataType getToType() const noexcept
Return cast layer output type.
Definition: NvInfer.h:3740
void setToType(DataType toType) noexcept
Set cast layer output type.
Definition: NvInfer.h:3729
virtual ~ICastLayer() noexcept=0
A concatenation layer in a network definition.
Definition: NvInfer.h:1833
void setAxis(int32_t axis) noexcept
Set the axis along which concatenation occurs.
Definition: NvInfer.h:1846
int32_t getAxis() const noexcept
Get the axis along which concatenation occurs.
Definition: NvInfer.h:1856
virtual ~IConcatenationLayer() noexcept=0
This layer represents a condition input to an IIfConditional.
Definition: NvInfer.h:4383
virtual ~IConditionLayer() noexcept=0
Layer that represents a constant value.
Definition: NvInfer.h:3761
void setWeights(Weights weights) noexcept
Set the weights for the layer.
Definition: NvInfer.h:3771
Weights getWeights() const noexcept
Get the weights for the layer.
Definition: NvInfer.h:3781
virtual ~IConstantLayer() noexcept=0
void setDimensions(Dims const &dimensions) noexcept
Set the dimensions for the layer.
Definition: NvInfer.h:3793
apiv::VConstantLayer * mImpl
Definition: NvInfer.h:3811
Dims getDimensions() const noexcept
Get the dimensions for the layer.
Definition: NvInfer.h:3805
A convolution layer in a network definition.
Definition: NvInfer.h:794
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:919
Weights getBiasWeights() const noexcept
Get the bias weights for the convolution.
Definition: NvInfer.h:892
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:960
void setDilationNd(Dims const &dilation) noexcept
Set the multi-dimension dilation of the convolution.
Definition: NvInfer.h:1064
virtual ~IConvolutionLayer() noexcept=0
Dims getPaddingNd() const noexcept
Get the multi-dimension padding of the convolution.
Definition: NvInfer.h:1050
Dims getStrideNd() const noexcept
Get the multi-dimension stride of the convolution.
Definition: NvInfer.h:1020
Weights getKernelWeights() const noexcept
Get the kernel weights of the convolution.
Definition: NvInfer.h:867
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride of the convolution.
Definition: NvInfer.h:1010
Dims getDilationNd() const noexcept
Get the multi-dimension dilation of the convolution.
Definition: NvInfer.h:1074
int64_t getNbOutputMaps() const noexcept
Get the number of output maps for the convolution.
Definition: NvInfer.h:813
void setKernelWeights(Weights weights) noexcept
Set the kernel weights for the convolution.
Definition: NvInfer.h:857
Dims getPostPadding() const noexcept
Get the post-padding.
Definition: NvInfer.h:946
int64_t getNbGroups() const noexcept
Get the number of groups of the convolution.
Definition: NvInfer.h:843
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:972
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups for a convolution.
Definition: NvInfer.h:833
void setNbOutputMaps(int64_t nbOutputMaps) noexcept
Set the number of output maps for the convolution.
Definition: NvInfer.h:803
void setBiasWeights(Weights weights) noexcept
Set the bias weights for the convolution.
Definition: NvInfer.h:882
Dims getKernelSizeNd() const noexcept
Get the multi-dimension kernel size of the convolution.
Definition: NvInfer.h:995
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding of the convolution.
Definition: NvInfer.h:1038
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding of the convolution.
Definition: NvInfer.h:909
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding of the convolution.
Definition: NvInfer.h:936
void setKernelSizeNd(Dims const &kernelSize) noexcept
Set the multi-dimension kernel size of the convolution.
Definition: NvInfer.h:985
Layer that represents a cumulative operation across a tensor.
Definition: NvInfer.h:6579
bool setOperation(CumulativeOperation op) noexcept
Set the cumulative operation for the layer.
Definition: NvInfer.h:6590
void setReverse(bool reverse) noexcept
Specify whether the cumulative operation should be applied backward.
Definition: NvInfer.h:6638
apiv::VCumulativeLayer * mImpl
Definition: NvInfer.h:6656
virtual ~ICumulativeLayer() noexcept=0
bool getExclusive() const noexcept
Get whether it is exclusive accumulation or inclusive accumulation.
Definition: NvInfer.h:6626
bool getReverse() const noexcept
Get the boolean that specifies whether the cumulative operation should be applied backward.
Definition: NvInfer.h:6650
void setExclusive(bool exclusive) noexcept
Set whether it is an exclusive accumulation or inclusive accumulation.
Definition: NvInfer.h:6614
CumulativeOperation getOperation() const noexcept
Get the cumulative operation for the layer.
Definition: NvInfer.h:6602
A deconvolution layer in a network definition.
Definition: NvInfer.h:1876
void setBiasWeights(Weights weights) noexcept
Set the bias weights for the deconvolution.
Definition: NvInfer.h:1964
int64_t getNbGroups() const noexcept
Get the number of groups for a deconvolution.
Definition: NvInfer.h:1925
Weights getKernelWeights() const noexcept
Get the kernel weights for the deconvolution.
Definition: NvInfer.h:1949
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding of the deconvolution.
Definition: NvInfer.h:1991
Dims getStrideNd() const noexcept
Get the multi-dimension stride of the deconvolution.
Definition: NvInfer.h:2106
Dims getDilationNd() const noexcept
Get the multi-dimension dilation of the deconvolution.
Definition: NvInfer.h:2172
virtual ~IDeconvolutionLayer() noexcept=0
Weights getBiasWeights() const noexcept
Get the bias weights for the deconvolution.
Definition: NvInfer.h:1974
void setKernelWeights(Weights weights) noexcept
Set the kernel weights for the deconvolution.
Definition: NvInfer.h:1939
int64_t getNbOutputMaps() const noexcept
Get the number of output feature maps for the deconvolution.
Definition: NvInfer.h:1895
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride of the deconvolution.
Definition: NvInfer.h:2096
Dims getPostPadding() const noexcept
Get the padding.
Definition: NvInfer.h:2028
Dims getKernelSizeNd() const noexcept
Get the multi-dimension kernel size of the deconvolution.
Definition: NvInfer.h:2079
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding of the deconvolution.
Definition: NvInfer.h:2018
void setKernelSizeNd(Dims const &kernelSize) noexcept
Set the multi-dimension kernel size of the deconvolution.
Definition: NvInfer.h:2069
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding of the deconvolution.
Definition: NvInfer.h:2124
void setNbOutputMaps(int64_t nbOutputMaps) noexcept
Set the number of output feature maps for the deconvolution.
Definition: NvInfer.h:1885
Dims getPaddingNd() const noexcept
Get the multi-dimension padding of the deconvolution.
Definition: NvInfer.h:2136
void setDilationNd(Dims const &dilation) noexcept
Set the multi-dimension dilation of the deconvolution.
Definition: NvInfer.h:2162
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:2042
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups for a deconvolution.
Definition: NvInfer.h:1915
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:2001
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:2054
A Dequantize layer in a network definition.
Definition: NvInfer.h:5498
TRT_NODISCARD Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5547
void setToType(DataType toType) noexcept
Set the Dequantize layer output type.
Definition: NvInfer.h:5563
int32_t getAxis() const noexcept
Get the quantization axis.
Definition: NvInfer.h:5508
bool setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5536
virtual ~IDequantizeLayer() noexcept=0
DataType getToType() const noexcept
Return the Dequantize layer output type.
Definition: NvInfer.h:5575
void setAxis(int32_t axis) noexcept
Set the quantization axis.
Definition: NvInfer.h:5519
Definition: NvInfer.h:8075
virtual ~IDistCollectiveLayer() noexcept=0
A network layer to perform dynamic quantization.
Definition: NvInfer.h:5605
virtual ~IDynamicQuantizeLayer() noexcept=0
DataType getScaleType() const noexcept
Return the scale factors data type.
Definition: NvInfer.h:5671
TRT_DEPRECATED void setAxis(int32_t axis) noexcept
Set the axis along which block quantization occurs.
Definition: NvInfer.h:5684
TRT_DEPRECATED void setBlockSize(int32_t size) noexcept
Set the size of the quantization block.
Definition: NvInfer.h:5707
Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5742
void setScaleType(DataType scaleType) noexcept
Set the data type of the scale factors used to quantize the data.
Definition: NvInfer.h:5658
DataType getToType() const noexcept
Return DynamicQuantizeLayer's quantized output type.
Definition: NvInfer.h:5645
TRT_DEPRECATED int32_t getAxis() const noexcept
Get the axis along which blocking occurs.
Definition: NvInfer.h:5694
void setToType(DataType toType) noexcept
Set DynamicQuantizeLayer's quantized output type.
Definition: NvInfer.h:5632
void setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5730
TRT_DEPRECATED int32_t getBlockSize() const noexcept
Get the size of the quantization block.
Definition: NvInfer.h:5717
An Einsum layer in a network.
Definition: NvInfer.h:5789
bool setEquation(char const *equation) noexcept
Set the equation. The equation is a comma-separated list of subscript labels, where each label refers...
Definition: NvInfer.h:5800
virtual ~IEinsumLayer() noexcept=0
char const * getEquation() const noexcept
Return the equation.
Definition: NvInfer.h:5810
A elementwise layer in a network definition.
Definition: NvInfer.h:2245
apiv::VElementWiseLayer * mImpl
Definition: NvInfer.h:2274
ElementWiseOperation getOperation() const noexcept
Get the binary operation for the layer.
Definition: NvInfer.h:2268
void setOperation(ElementWiseOperation op) noexcept
Set the binary operation for the layer.
Definition: NvInfer.h:2256
virtual ~IElementWiseLayer() noexcept=0
Generate a tensor according to a specified mode.
Definition: NvInfer.h:4995
bool isAlphaBetaInt64() const noexcept
Return true if alpha/beta have type int64, false if they have type double.
Definition: NvInfer.h:5227
FillOperation getOperation() const noexcept
Get the fill operation for the layer.
Definition: NvInfer.h:5041
void setOperation(FillOperation op) noexcept
Set the fill operation for the layer.
Definition: NvInfer.h:5031
DataType getToType() const noexcept
Get the fill layer output type.
Definition: NvInfer.h:5257
void setAlphaInt64(int64_t alpha) noexcept
Set the alpha parameter with int64 datatype.
Definition: NvInfer.h:5170
void setBetaInt64(int64_t beta) noexcept
Set the beta parameter with int64 datatype.
Definition: NvInfer.h:5204
virtual ~IFillLayer() noexcept=0
void setBeta(double beta) noexcept
Set the beta parameter.
Definition: NvInfer.h:5094
int64_t getAlphaInt64() const noexcept
Get the value of alpha parameter with int64 datatype.
Definition: NvInfer.h:5185
int64_t getBetaInt64() const noexcept
Get the value of beta parameter with int64 datatype.
Definition: NvInfer.h:5219
double getAlpha() const noexcept
Get the value of alpha parameter.
Definition: NvInfer.h:5075
void setDimensions(Dims const &dimensions) noexcept
Set the output tensor's dimensions.
Definition: NvInfer.h:5006
void setAlpha(double alpha) noexcept
Set the alpha parameter.
Definition: NvInfer.h:5060
void setToType(DataType toType) noexcept
Set the fill layer output type.
Definition: NvInfer.h:5245
Dims getDimensions() const noexcept
Get the output tensor's dimensions.
Definition: NvInfer.h:5021
double getBeta() const noexcept
Get the value of beta parameter.
Definition: NvInfer.h:5109
A Gather layer in a network definition. Supports several kinds of gathering.
Definition: NvInfer.h:2380
void setGatherAxis(int32_t axis) noexcept
Set the axis used by GatherMode::kELEMENTS and GatherMode::kDEFAULT The axis must be less than the nu...
Definition: NvInfer.h:2391
void setNbElementWiseDims(int32_t elementWiseDims) noexcept
Set the number of leading dimensions of indices tensor to be handled elementwise.
Definition: NvInfer.h:2426
apiv::VGatherLayer * mImpl
Definition: NvInfer.h:2462
virtual ~IGatherLayer() noexcept=0
int32_t getNbElementWiseDims() const noexcept
Get the number of leading dimensions of indices tensor to be handled elementwise.
Definition: NvInfer.h:2436
void setMode(GatherMode mode) noexcept
Set the gather mode.
Definition: NvInfer.h:2446
int32_t getGatherAxis() const noexcept
Get the axis to gather on.
Definition: NvInfer.h:2403
GatherMode getMode() const noexcept
Get the gather mode.
Definition: NvInfer.h:2456
A GridSample layer in a network definition.
Definition: NvInfer.h:6018
void setInterpolationMode(InterpolationMode mode) noexcept
Set the grid sample interpolation mode.
Definition: NvInfer.h:6025
bool setSampleMode(SampleMode mode) noexcept
Set the sample mode.
Definition: NvInfer.h:6071
void setAlignCorners(bool alignCorners) noexcept
Set the align corners mode.
Definition: NvInfer.h:6047
apiv::VGridSampleLayer * mImpl
Definition: NvInfer.h:6089
SampleMode getSampleMode() const noexcept
Get the sample mode.
Definition: NvInfer.h:6083
virtual ~IGridSampleLayer() noexcept=0
InterpolationMode getInterpolationMode() const noexcept
Get the grid sample interpolation mode.
Definition: NvInfer.h:6037
bool getAlignCorners() const noexcept
Get the align corners mode.
Definition: NvInfer.h:6059
Class to handle library allocated memory that is accessible to the user.
Definition: NvInferRuntime.h:139
A layer that represents the identity function.
Definition: NvInfer.h:3705
virtual ~IIdentityLayer() noexcept=0
apiv::VIdentityLayer * mImpl
Definition: NvInfer.h:3707
This is a base class for Conditional boundary layers.
Definition: NvInfer.h:4360
IIfConditional * getConditional() const noexcept
Get a pointer to the IIfConditional associated with this boundary layer.
Definition: NvInfer.h:4365
virtual ~IIfConditionalBoundaryLayer() noexcept=0
Helper for constructing conditionally-executed subgraphs.
Definition: NvInfer.h:4451
IIfConditionalInputLayer * addInput(ITensor &input) noexcept
Add an If-conditional input.
Definition: NvInfer.h:4492
char const * getName() const noexcept
Return the name of the conditional.
Definition: NvInfer.h:4517
IConditionLayer * setCondition(ITensor &condition) noexcept
Set the condition tensor for this If-Conditional construct.
Definition: NvInfer.h:4462
virtual ~IIfConditional() noexcept=0
IIfConditionalOutputLayer * addOutput(ITensor &trueSubgraphOutput, ITensor &falseSubgraphOutput) noexcept
Add an If-conditional output.
Definition: NvInfer.h:4480
void setName(char const *name) noexcept
Set the name of the conditional.
Definition: NvInfer.h:4507
This layer represents an output of an IIfConditional.
Definition: NvInfer.h:4402
virtual ~IIfConditionalOutputLayer() noexcept=0
A layer to do iterations.
Definition: NvInfer.h:4690
void setReverse(bool reverse) noexcept
Set iteration order to be reverse.
Definition: NvInfer.h:4717
virtual ~IIteratorLayer() noexcept=0
bool getReverse() const noexcept
Check if the iteration order is reverse.
Definition: NvInfer.h:4727
int32_t getAxis() const noexcept
Get axis being iterated over.
Definition: NvInfer.h:4703
void setAxis(int32_t axis) noexcept
Set axis to iterate over.
Definition: NvInfer.h:4695
Layer that represents a KVCacheUpdate operation.
Definition: NvInfer.h:7499
bool setCacheMode(KVCacheMode cacheMode) noexcept
Set the mode of the KVCacheUpdate layer.
Definition: NvInfer.h:7524
TRT_NODISCARD ITensor * getUpdateLengths() const noexcept
Get the update lengths tensor.
Definition: NvInfer.h:7599
virtual ~IKVCacheUpdateLayer() noexcept=0
TRT_NODISCARD AttentionIOForm getUpdateForm() const noexcept
Get the update form.
Definition: NvInfer.h:7565
TRT_NODISCARD bool setUpdateLengths(ITensor *lengths) noexcept
Set the update lengths tensor.
Definition: NvInfer.h:7587
TRT_NODISCARD bool setUpdateForm(AttentionIOForm form) noexcept
Set the update form.
Definition: NvInfer.h:7552
KVCacheMode getCacheMode() const noexcept
Get the mode of the KVCacheUpdate layer.
Definition: NvInfer.h:7534
apiv::VKVCacheUpdateLayer * mImpl
Definition: NvInfer.h:7605
A LRN layer in a network definition.
Definition: NvInfer.h:1482
int64_t getWindowSize() const noexcept
Get the LRN window size.
Definition: NvInfer.h:1503
virtual ~ILRNLayer() noexcept=0
float getAlpha() const noexcept
Get the LRN alpha value.
Definition: NvInfer.h:1525
void setWindowSize(int64_t windowSize) noexcept
Set the LRN window size.
Definition: NvInfer.h:1493
void setK(float k) noexcept
Set the LRN K value.
Definition: NvInfer.h:1559
void setAlpha(float alpha) noexcept
Set the LRN alpha value.
Definition: NvInfer.h:1515
void setBeta(float beta) noexcept
Set the LRN beta value.
Definition: NvInfer.h:1537
float getBeta() const noexcept
Get the LRN beta value.
Definition: NvInfer.h:1547
float getK() const noexcept
Get the LRN K value.
Definition: NvInfer.h:1569
Base class for all layer classes in a network definition.
Definition: NvInfer.h:427
virtual ~ILayer() noexcept=0
void setMetadata(char const *metadata) noexcept
Set the metadata for this layer.
Definition: NvInfer.h:548
void setName(char const *name) noexcept
Set the name of a layer.
Definition: NvInfer.h:448
int32_t getNbInputs() const noexcept
Get the number of inputs of a layer.
Definition: NvInfer.h:466
int32_t getNbRanks() const noexcept
Get the number of ranks for multi-device execution.
Definition: NvInfer.h:594
char const * getMetadata() const noexcept
Get the metadata of the layer.
Definition: NvInfer.h:561
DataType getOutputType(int32_t index) const noexcept
get the output type of this layer
Definition: NvInfer.h:529
char const * getName() const noexcept
Return the name of a layer.
Definition: NvInfer.h:458
int32_t getNbOutputs() const noexcept
Get the number of outputs of a layer.
Definition: NvInfer.h:487
ITensor * getOutput(int32_t index) const noexcept
Get the layer output corresponding to the given index.
Definition: NvInfer.h:497
void setInput(int32_t index, ITensor &tensor) noexcept
Replace an input of this layer with a specific tensor.
Definition: NvInfer.h:514
ITensor * getInput(int32_t index) const noexcept
Get the layer input corresponding to the given index.
Definition: NvInfer.h:479
bool setNbRanks(int32_t nbRanks) noexcept
Set the number of ranks for multi-device execution.
Definition: NvInfer.h:582
LayerType getType() const noexcept
Return the type of a layer.
Definition: NvInfer.h:434
This is a base class for Loop boundary layers.
Definition: NvInfer.h:4335
virtual ~ILoopBoundaryLayer() noexcept=0
ILoop * getLoop() const noexcept
Get a pointer to ILoop associated with this boundary layer.
Definition: NvInfer.h:4340
Helper for creating a recurrent subgraph.
Definition: NvInfer.h:4750
void setName(char const *name) noexcept
Set the name of the loop.
Definition: NvInfer.h:4820
ITripLimitLayer * addTripLimit(ITensor &tensor, TripLimit limit) noexcept
Add a trip-count limiter, based on the given tensor.
Definition: NvInfer.h:4779
IIteratorLayer * addIterator(ITensor &tensor, int32_t axis=0, bool reverse=false) noexcept
Return layer that subscripts tensor by loop iteration.
Definition: NvInfer.h:4792
ILoopOutputLayer * addLoopOutput(ITensor &tensor, LoopOutput outputKind, int32_t axis=0) noexcept
Make an output for this loop, based on the given tensor.
Definition: NvInfer.h:4805
virtual ~ILoop() noexcept=0
char const * getName() const noexcept
Return the name of the loop.
Definition: NvInfer.h:4830
IRecurrenceLayer * addRecurrence(ITensor &initialValue) noexcept
Create a recurrence layer for this loop with initialValue as its first input.
Definition: NvInfer.h:4758
An ILoopOutputLayer is the sole way to get output from a loop.
Definition: NvInfer.h:4586
int32_t getAxis() const noexcept
Get axis being concatenated over.
Definition: NvInfer.h:4616
LoopOutput getLoopOutput() const noexcept
Get which kind a loop output has.
Definition: NvInfer.h:4591
virtual ~ILoopOutputLayer() noexcept=0
void setAxis(int32_t axis) noexcept
Set where to insert the contenation axis. Ignored if getLoopOutput() is kLAST_VALUE.
Definition: NvInfer.h:4608
Layer that represents a Matrix Multiplication.
Definition: NvInfer.h:3546
apiv::VMatrixMultiplyLayer * mImpl
Definition: NvInfer.h:3574
virtual ~IMatrixMultiplyLayer() noexcept=0
MatrixOperation getOperation(int32_t index) const noexcept
Get the operation for an input tensor.
Definition: NvInfer.h:3568
void setOperation(int32_t index, MatrixOperation op) noexcept
Set the operation for an input tensor.
Definition: NvInfer.h:3556
A MoE layer in a network definition. Mixture of Experts (MoE) is a collection of experts with each ex...
Definition: NvInfer.h:7750
void setSwigluParamLimit(float limit) noexcept
Set the SwiGLU parameter limit.
Definition: NvInfer.h:7972
void setDynQOutputScaleType(DataType type) noexcept
Set the dynamic quantization output scale type.
Definition: NvInfer.h:7925
MoEActType getActivationType() const noexcept
Get the activation type for the MoE layer.
Definition: NvInfer.h:7799
void setQuantizationToType(DataType type) noexcept
Set the data type the mul output is quantized to.
Definition: NvInfer.h:7873
void setQuantizationDynamicDblQ(ITensor &fcDownActivationDblQScale, DataType dataType, Dims const &blockShape, DataType dynQOutputScaleType) noexcept
Configure dynamic quantization (with double quantization) after the mul op.
Definition: NvInfer.h:7858
void setQuantizationStatic(ITensor &fcDownActivationScale, DataType dataType) noexcept
Configure static quantization after the mul op.
Definition: NvInfer.h:7825
float getSwigluParamLimit() const noexcept
Get the SwiGLU parameter limit.
Definition: NvInfer.h:7984
DataType getQuantizationToType() const noexcept
Get the data type the mul in MoE layer is quantized to.
Definition: NvInfer.h:7885
DataType getDynQOutputScaleType() const noexcept
Get the dynamic quantization output scale type.
Definition: NvInfer.h:7937
virtual ~IMoELayer() noexcept=0
void setActivationType(MoEActType activationType) noexcept
Set the activation type for the MoE layer.
Definition: NvInfer.h:7787
Dims getQuantizationBlockShape() const noexcept
Get the block shape for the quantization of the Mul output.
Definition: NvInfer.h:7913
void setGatedWeights(ITensor &fcGateWeights, ITensor &fcUpWeights, ITensor &fcDownWeights, MoEActType activationType) noexcept
Set the weights of the experts when each expert is a GLU (gated linear unit). In each GLU,...
Definition: NvInfer.h:7763
float getSwigluParamBeta() const noexcept
Get the SwiGLU parameter beta.
Definition: NvInfer.h:8036
void setSwigluParamBeta(float beta) noexcept
Set the SwiGLU parameter beta.
Definition: NvInfer.h:8024
void setGatedBiases(ITensor &fcGateBiases, ITensor &fcUpBiases, ITensor &fcDownBiases) noexcept
Set the biases of the experts when each expert is a GLU (gated linear unit). In each GLU,...
Definition: NvInfer.h:7775
void setSwigluParams(float limit, float alpha, float beta) noexcept
Set the SwiGLU parameters.
Definition: NvInfer.h:7958
void setQuantizationBlockShape(Dims const &blockShape) noexcept
Set the block shape for the quantization of the Mul output.
Definition: NvInfer.h:7901
void setInput(int32_t index, ITensor &tensor) noexcept
Set the input of the MoE layer.
Definition: NvInfer.h:8053
float getSwigluParamAlpha() const noexcept
Get the SwiGLU parameter alpha.
Definition: NvInfer.h:8010
void setSwigluParamAlpha(float alpha) noexcept
Set the SwiGLU parameter alpha.
Definition: NvInfer.h:7998
A non-maximum suppression layer in a network definition.
Definition: NvInfer.h:6172
void setTopKBoxLimit(int32_t limit) noexcept
Set the TopK box limit parameter for the layer.
Definition: NvInfer.h:6209
void setBoundingBoxFormat(BoundingBoxFormat fmt) noexcept
Set the bounding box format parameter for the layer.
Definition: NvInfer.h:6183
BoundingBoxFormat getBoundingBoxFormat() const noexcept
Get the bounding box format parameter for the layer.
Definition: NvInfer.h:6195
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:6254
apiv::VNMSLayer * mImpl
Definition: NvInfer.h:6272
int32_t getTopKBoxLimit() const noexcept
Get the TopK box limit parameter for the layer.
Definition: NvInfer.h:6219
DataType getIndicesType() const noexcept
Return the NMS layer indices type.
Definition: NvInfer.h:6266
virtual ~INMSLayer() noexcept=0
A network definition for input to the builder.
Definition: NvInfer.h:8101
IConcatenationLayer * addConcatenation(ITensor *const *inputs, int32_t nbInputs) noexcept
Add a concatenation layer to the network.
Definition: NvInfer.h:8329
IShuffleLayer * addShuffle(ITensor &input) noexcept
Add a shuffle layer to the network.
Definition: NvInfer.h:8392
void setName(char const *name) noexcept
Sets the name of the network.
Definition: NvInfer.h:8818
ITopKLayer * addTopK(ITensor &input, TopKOperation op, int32_t k, uint32_t reduceAxes, DataType indicesType) noexcept
Add a TopK layer to the network.
Definition: NvInfer.h:8588
bool markDebug(ITensor &tensor) noexcept
Mark a tensor as a debug tensor.
Definition: NvInfer.h:8172
ILRNLayer * addLRN(ITensor &input, int64_t window, float alpha, float beta, float k) noexcept
Add a LRN layer to the network.
Definition: NvInfer.h:8273
ICumulativeLayer * addCumulative(ITensor &input, ITensor &axis, CumulativeOperation operation, bool exclusive, bool reverse) noexcept
Add a cumulative layer to the network.
Definition: NvInfer.h:9485
IAssertionLayer * addAssertion(ITensor &condition, char const *message) noexcept
Add an assertion layer to the network.
Definition: NvInfer.h:9120
TRT_DEPRECATED INonZeroLayer * addNonZero(ITensor &input) noexcept
Add a nonzero layer to the network.
Definition: NvInfer.h:8679
IConvolutionLayer * addConvolutionNd(ITensor &input, int64_t nbOutputMaps, Dims const &kernelSize, Weights kernelWeights, Weights biasWeights) noexcept
Add a multi-dimension convolution layer to the network.
Definition: NvInfer.h:8939
ICastLayer * addCast(ITensor &input, DataType toType) noexcept
Add a cast layer.
Definition: NvInfer.h:8748
IScaleLayer * addScaleNd(ITensor &input, ScaleMode mode, Weights shift, Weights scale, Weights power, int32_t channelAxis) noexcept
Add a multi-dimension scale layer to the network.
Definition: NvInfer.h:9018
char const * getName() const noexcept
Returns the name associated with the network.
Definition: NvInfer.h:8832
IParametricReLULayer * addParametricReLU(ITensor &input, ITensor &slope) noexcept
Add a parametric ReLU layer to the network.
Definition: NvInfer.h:8917
ITensor * getOutput(int32_t index) const noexcept
Get the output tensor specified by the given index.
Definition: NvInfer.h:8493
ITensor * getInput(int32_t index) const noexcept
Get the input tensor specified by the given index.
Definition: NvInfer.h:8463
TRT_DEPRECATED ITopKLayer * addTopK(ITensor &input, TopKOperation op, int32_t k, uint32_t reduceAxes) noexcept
Add a TopK layer to the network.
Definition: NvInfer.h:8555
IDequantizeLayer * addDequantize(ITensor &input, ITensor &scale, DataType outputType) noexcept
Add a dequantization layer to the network.
Definition: NvInfer.h:9243
bool unmarkOutputForShapes(ITensor &tensor) noexcept
Undo markOutputForShapes.
Definition: NvInfer.h:8899
IFillLayer * addFill(Dims const &dimensions, FillOperation op, DataType outputType) noexcept
Add a fill layer to the network.
Definition: NvInfer.h:9146
ILoop * addLoop() noexcept
Add a loop to the network.
Definition: NvInfer.h:9049
bool markUnfusedTensorsAsDebugTensors() noexcept
Mark unfused tensors as debug tensors.
Definition: NvInfer.h:8220
TRT_NODISCARD INormalizationLayer * addNormalizationV2(ITensor &input, ITensor &scale, ITensor &bias, uint32_t axesMask) noexcept
Add a normalization layer to the network.
Definition: NvInfer.h:9774
IActivationLayer * addActivation(ITensor &input, ActivationType type) noexcept
Add an activation layer to the network.
Definition: NvInfer.h:8254
ISliceLayer * addSlice(ITensor &input, Dims const &start, Dims const &size, Dims const &stride) noexcept
Add a slice layer to the network.
Definition: NvInfer.h:8794
virtual IBuilder & getBuilder() const noexcept
Return the builder from which this INetworkDefinition was created.
Definition: NvInfer.h:9670
ILayer * getLayer(int32_t index) const noexcept
Get the layer specified by the given index.
Definition: NvInfer.h:8435
bool isDebugTensor(ITensor const &tensor) const noexcept
Check if a tensor is marked as debug tensor.
Definition: NvInfer.h:8198
bool getFlag(NetworkDefinitionCreationFlag networkDefinitionCreationFlag) const noexcept
Returns true if the network definition creation flag is set.
Definition: NvInfer.h:8870
IIfConditional * addIfConditional() noexcept
Add an if-then-else to the network.
Definition: NvInfer.h:9064
IErrorRecorder * getErrorRecorder() const noexcept
get the ErrorRecorder assigned to this interface.
Definition: NvInfer.h:9220
ISqueezeLayer * addSqueeze(ITensor &input, ITensor &axes) noexcept
Add a squeeze layer to the network.
Definition: NvInfer.h:9727
TRT_DEPRECATED INMSLayer * addNMS(ITensor &boxes, ITensor &scores, ITensor &maxOutputBoxesPerClass) noexcept
Add a non-maximum suppression layer to the network.
Definition: NvInfer.h:9394
IAttention * addAttentionV2(ITensor &query, ITensor &key, ITensor &value, AttentionNormalizationOp normOp, CausalMaskKind causalKind) noexcept
Add an attention to the network with explicit causal mask kind.
Definition: NvInfer.h:9545
IReverseSequenceLayer * addReverseSequence(ITensor &input, ITensor &sequenceLens) noexcept
Add a ReverseSequence layer to the network.
Definition: NvInfer.h:9431
TRT_DEPRECATED IDynamicQuantizeLayer * addDynamicQuantize(ITensor &input, int32_t axis, int32_t blockSize, DataType outputType, DataType scaleType) noexcept
Add a dynamic quantization layer to the network.
Definition: NvInfer.h:9314
int32_t getNbInputs() const noexcept
Get the number of inputs in the network.
Definition: NvInfer.h:8447
NetworkDefinitionCreationFlags getFlags() const noexcept
Get the network definition creation flags for this network definition object. Defaults to 0.
Definition: NvInfer.h:8858
IQuantizeLayer * addQuantize(ITensor &input, ITensor &scale, DataType outputType) noexcept
Add a quantization layer to the network.
Definition: NvInfer.h:9287
IDynamicQuantizeLayer * addDynamicQuantizeV2(ITensor &input, Dims const &blockShape, DataType outputType, DataType scaleType) noexcept
Add a dynamic quantization layer to the network.
Definition: NvInfer.h:9338
IReduceLayer * addReduce(ITensor &input, ReduceOperation operation, uint32_t reduceAxes, bool keepDimensions) noexcept
Add a reduce layer to the network.
Definition: NvInfer.h:8519
IUnaryLayer * addUnary(ITensor &input, UnaryOperation operation) noexcept
Add a unary layer to the network.
Definition: NvInfer.h:8378
IGridSampleLayer * addGridSample(ITensor &input, ITensor &grid) noexcept
Add a GridSample layer to the network.
Definition: NvInfer.h:9372
void removeTensor(ITensor &tensor) noexcept
remove a tensor from the network definition.
Definition: NvInfer.h:8763
bool areWeightsMarkedRefittable(char const *name) const noexcept
Whether the weight has been marked as refittable.
Definition: NvInfer.h:9708
ISelectLayer * addSelect(ITensor &condition, ITensor &thenInput, ITensor &elseInput) noexcept
Add a select layer to the network.
Definition: NvInfer.h:9103
IScatterLayer * addScatter(ITensor &data, ITensor &indices, ITensor &updates, ScatterMode mode) noexcept
Add a Scatter layer to the network with specified mode and axis=0.
Definition: NvInfer.h:9263
TRT_DEPRECATED INormalizationLayer * addNormalization(ITensor &input, ITensor &scale, ITensor &bias, uint32_t axesMask) noexcept
Add a normalization layer to the network.
Definition: NvInfer.h:9463
int32_t getNbLayers() const noexcept
Get the number of layers in the network.
Definition: NvInfer.h:8421
apiv::VNetworkDefinition * mImpl
Definition: NvInfer.h:9780
IKVCacheUpdateLayer * addKVCacheUpdate(ITensor &cache, ITensor &update, ITensor &writeIndices, KVCacheMode cacheMode) noexcept
Add a KVCacheUpdate layer to the network.
Definition: NvInfer.h:9604
bool markOutputForShapes(ITensor &tensor) noexcept
Enable tensor's value to be computed by IExecutionContext::getShapeBinding.
Definition: NvInfer.h:8887
IOneHotLayer * addOneHot(ITensor &indices, ITensor &values, ITensor &depth, int32_t axis) noexcept
Add a OneHot layer to the network.
Definition: NvInfer.h:8409
IScaleLayer * addScale(ITensor &input, ScaleMode mode, Weights shift, Weights scale, Weights power) noexcept
Add a Scale layer to the network.
Definition: NvInfer.h:8299
void unmarkOutput(ITensor &tensor) noexcept
unmark a tensor as a network output.
Definition: NvInfer.h:8775
IIdentityLayer * addIdentity(ITensor &input) noexcept
Add an identity layer.
Definition: NvInfer.h:8733
IGatherLayer * addGatherV2(ITensor &data, ITensor &indices, GatherMode mode) noexcept
Add gather with specified mode, axis=0 and nbElementWiseDims=0.
Definition: NvInfer.h:8620
INonZeroLayer * addNonZero(ITensor &input, DataType indicesType) noexcept
Add a nonzero layer to the network.
Definition: NvInfer.h:8695
IElementWiseLayer * addElementWise(ITensor &input1, ITensor &input2, ElementWiseOperation op) noexcept
Add an elementwise layer to the network.
Definition: NvInfer.h:8356
IConstantLayer * addConstant(Dims const &dimensions, Weights weights) noexcept
Add a constant layer to the network.
Definition: NvInfer.h:8719
void setErrorRecorder(IErrorRecorder *recorder) noexcept
Set the ErrorRecorder for this interface.
Definition: NvInfer.h:9205
IPoolingLayer * addPoolingNd(ITensor &input, PoolingType type, Dims const &windowSize) noexcept
Add a multi-dimension pooling layer to the network.
Definition: NvInfer.h:8959
INMSLayer * addNMS(ITensor &boxes, ITensor &scores, ITensor &maxOutputBoxesPerClass, DataType indicesType) noexcept
Add a non-maximum suppression layer to the network.
Definition: NvInfer.h:9414
IRaggedSoftMaxLayer * addRaggedSoftMax(ITensor &input, ITensor &bounds) noexcept
Add a RaggedSoftMax layer to the network.
Definition: NvInfer.h:8639
IShapeLayer * addShape(ITensor &input) noexcept
Add a shape layer to the network.
Definition: NvInfer.h:8848
IGatherLayer * addGather(ITensor &data, ITensor &indices, int32_t axis) noexcept
Add gather with mode GatherMode::kDEFAULT and specified axis and nbElementWiseDims=0.
Definition: NvInfer.h:8604
bool unmarkWeightsRefittable(char const *name) noexcept
Unmark weights as refittable when the builder flag kREFIT_INDIVIDUAL is set.
Definition: NvInfer.h:9695
bool markWeightsRefittable(char const *name) noexcept
Mark weights as refittable when the builder flag kREFIT_INDIVIDUAL is set.
Definition: NvInfer.h:9683
IRotaryEmbeddingLayer * addRotaryEmbedding(ITensor &input, ITensor &cosCache, ITensor &sinCache, bool interleaved, int32_t rotaryEmbeddingDim) noexcept
Add a Rotary Position Embedding (RoPE) layer to the network.
Definition: NvInfer.h:9570
IDeconvolutionLayer * addDeconvolutionNd(ITensor &input, int64_t nbOutputMaps, Dims kernelSize, Weights kernelWeights, Weights biasWeights) noexcept
Add a multi-dimension deconvolution layer to the network.
Definition: NvInfer.h:8981
IResizeLayer * addResize(ITensor &input) noexcept
Add a resize layer to the network.
Definition: NvInfer.h:9035
IUnsqueezeLayer * addUnsqueeze(ITensor &input, ITensor &axes) noexcept
Add an unsqueeze layer to the network.
Definition: NvInfer.h:9748
IMatrixMultiplyLayer * addMatrixMultiply(ITensor &input0, MatrixOperation op0, ITensor &input1, MatrixOperation op1) noexcept
Add a MatrixMultiply layer to the network.
Definition: NvInfer.h:8660
ISoftMaxLayer * addSoftMax(ITensor &input) noexcept
Add a SoftMax layer to the network.
Definition: NvInfer.h:8312
bool unmarkDebug(ITensor &tensor) noexcept
Unmark a tensor as a debug tensor.
Definition: NvInfer.h:8188
TRT_DEPRECATED IAttention * addAttention(ITensor &query, ITensor &key, ITensor &value, AttentionNormalizationOp normOp, bool causal) noexcept
Add an attention to the network.
Definition: NvInfer.h:9515
virtual ~INetworkDefinition() noexcept=0
IEinsumLayer * addEinsum(ITensor *const *inputs, int32_t nbInputs, char const *equation) noexcept
Add an Einsum layer to the network.
Definition: NvInfer.h:9354
void markOutput(ITensor &tensor) noexcept
Mark a tensor as a network output.
Definition: NvInfer.h:8154
IPaddingLayer * addPaddingNd(ITensor &input, Dims const &prePadding, Dims const &postPadding) noexcept
Add a padding layer to the network. Only 2D padding is currently supported.
Definition: NvInfer.h:9162
int32_t getNbOutputs() const noexcept
Get the number of outputs in the network.
Definition: NvInfer.h:8477
bool setWeightsName(Weights weights, char const *name) noexcept
Associate a name with all current uses of the given weights.
Definition: NvInfer.h:9186
TRT_NODISCARD IDistCollectiveLayer * addDistCollective(ITensor &input, CollectiveOperation distCollectiveOp, ReduceOperation reduceOp, int64_t root, int64_t *groups, int64_t groupSize) noexcept
Add a DistCollective layer to the network.
Definition: NvInfer.h:9658
IMoELayer * addMoE(ITensor &hiddenStates, ITensor &selectedExpertsForTokens, ITensor &scoresForSelectedExperts) noexcept
Add a MoE (Mixture of Experts) layer to the network.
Definition: NvInfer.h:9626
bool unmarkUnfusedTensorsAsDebugTensors() noexcept
Undo the marking of unfused tensors as debug tensors.
Definition: NvInfer.h:8234
Forward declaration of IEngineInspector for use by other interfaces.
Definition: NvInferRuntime.h:51
Definition: NvInfer.h:3602
DataType getIndicesType() const noexcept
Return the NonZero layer indices type.
Definition: NvInfer.h:3626
virtual ~INonZeroLayer() noexcept=0
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:3614
A normalization layer in a network definition.
Definition: NvInfer.h:6365
float getEpsilon() const noexcept
Get the epsilon value used for the normalization calculation.
Definition: NvInfer.h:6384
uint32_t getAxes() const noexcept
Get the axes value used for the normalization calculation.
Definition: NvInfer.h:6404
void setEpsilon(float eps) noexcept
Set the epsilon value used for the normalization calculation.
Definition: NvInfer.h:6374
virtual ~INormalizationLayer() noexcept=0
TRT_NODISCARD bool isV2() const noexcept
Returns true if this layer was created through addNormalizationV2().
Definition: NvInfer.h:6446
apiv::VNormalizationLayer * mImpl
Definition: NvInfer.h:6452
int64_t getNbGroups() const noexcept
Get the number of groups used to split the channels for the normalization calculation.
Definition: NvInfer.h:6435
void setAxes(uint32_t axesMask) noexcept
Set the reduction axes for the normalization calculation.
Definition: NvInfer.h:6394
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups used to split the channels in the normalization calculation.
Definition: NvInfer.h:6425
A OneHot layer in a network definition.
Definition: NvInfer.h:5979
apiv::VOneHotLayer * mImpl
Definition: NvInfer.h:6000
void setAxis(int32_t axis) noexcept
Set the axis parameter.
Definition: NvInfer.h:5986
virtual ~IOneHotLayer() noexcept=0
int32_t getAxis() const noexcept
Get the value of the axis parameter.
Definition: NvInfer.h:5994
Optimization profile for dynamic input dimensions and shape tensors.
Definition: NvInferRuntime.h:2665
Layer that represents a padding operation.
Definition: NvInfer.h:2787
Dims getPostPaddingNd() const noexcept
Get the padding that is applied at the end of the tensor.
Definition: NvInfer.h:2836
void setPrePaddingNd(Dims const &padding) noexcept
Set the padding that is applied at the start of the tensor.
Definition: NvInfer.h:2798
virtual ~IPaddingLayer() noexcept=0
void setPostPaddingNd(Dims const &padding) noexcept
Set the padding that is applied at the end of the tensor.
Definition: NvInfer.h:2824
Dims getPrePaddingNd() const noexcept
Get the padding that is applied at the start of the tensor.
Definition: NvInfer.h:2810
apiv::VPaddingLayer * mImpl
Definition: NvInfer.h:2842
Layer that represents a parametric ReLU operation.
Definition: NvInfer.h:3827
apiv::VParametricReLULayer * mImpl
Definition: NvInfer.h:3829
virtual ~IParametricReLULayer() noexcept=0
Single registration point for all plugins in an application. It is used to find plugin implementation...
Definition: NvInferRuntimeCommon.h:56
Plugin class for user-implemented layers.
Definition: NvInferRuntimePlugin.h:139
Layer type for pluginV2.
Definition: NvInfer.h:2480
apiv::VPluginV2Layer * mImpl
Definition: NvInfer.h:2493
IPluginV2 & getPlugin() noexcept
Get the plugin for the layer.
Definition: NvInfer.h:2487
virtual ~IPluginV2Layer() noexcept=0
Layer type for V3 plugins.
Definition: NvInfer.h:2509
virtual ~IPluginV3Layer() noexcept=0
IPluginV3 & getPlugin() noexcept
Get the plugin for the layer.
Definition: NvInfer.h:2516
apiv::VPluginV3Layer * mImpl
Definition: NvInfer.h:2522
A Pooling layer in a network definition.
Definition: NvInfer.h:1229
PoolingType getPoolingType() const noexcept
Get the type of activation to be performed.
Definition: NvInfer.h:1248
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:1381
Dims getPostPadding() const noexcept
Get the padding.
Definition: NvInfer.h:1357
bool getAverageCountExcludesPadding() const noexcept
Get whether average pooling uses as a denominator the overlap area between the window and the unpadde...
Definition: NvInfer.h:1301
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:1329
void setPoolingType(PoolingType type) noexcept
Set the type of activation to be performed.
Definition: NvInfer.h:1238
void setWindowSizeNd(Dims const &windowSize) noexcept
Set the multi-dimension window size for pooling.
Definition: NvInfer.h:1394
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:1370
Dims getWindowSizeNd() const noexcept
Get the multi-dimension window size for pooling.
Definition: NvInfer.h:1404
void setAverageCountExcludesPadding(bool exclusive) noexcept
Set whether average pooling uses as a denominator the overlap area between the window and the unpadde...
Definition: NvInfer.h:1290
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding for pooling.
Definition: NvInfer.h:1448
float getBlendFactor() const noexcept
Get the blending factor for the max_average_blend mode: max_average_blendPool = (1-blendFactor)*maxPo...
Definition: NvInfer.h:1276
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride for pooling.
Definition: NvInfer.h:1419
Dims getStrideNd() const noexcept
Get the multi-dimension stride for pooling.
Definition: NvInfer.h:1429
Dims getPaddingNd() const noexcept
Get the multi-dimension padding for pooling.
Definition: NvInfer.h:1460
virtual ~IPoolingLayer() noexcept=0
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding for pooling.
Definition: NvInfer.h:1347
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding for pooling.
Definition: NvInfer.h:1319
void setBlendFactor(float blendFactor) noexcept
Set the blending factor for the max_average_blend mode: max_average_blendPool = (1-blendFactor)*maxPo...
Definition: NvInfer.h:1263
A Quantize layer in a network definition.
Definition: NvInfer.h:5344
void setToType(DataType toType) noexcept
Set the Quantize layer output type.
Definition: NvInfer.h:5405
bool setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5378
void setAxis(int32_t axis) noexcept
Set the quantization axis.
Definition: NvInfer.h:5365
virtual ~IQuantizeLayer() noexcept=0
TRT_NODISCARD Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5389
int32_t getAxis() const noexcept
Get the quantization axis.
Definition: NvInfer.h:5354
DataType getToType() const noexcept
Return the Quantize layer output type.
Definition: NvInfer.h:5417
A RaggedSoftmax layer in a network definition.
Definition: NvInfer.h:3653
apiv::VRaggedSoftMaxLayer * mImpl
Definition: NvInfer.h:3655
virtual ~IRaggedSoftMaxLayer() noexcept=0
A recurrence layer in a network definition.
Definition: NvInfer.h:4537
virtual ~IRecurrenceLayer() noexcept=0
Layer that represents a reduction across a non-bool tensor.
Definition: NvInfer.h:2705
void setKeepDimensions(bool keepDimensions) noexcept
Set the boolean that specifies whether or not to keep the reduced dimensions for the layer.
Definition: NvInfer.h:2752
virtual ~IReduceLayer() noexcept=0
void setOperation(ReduceOperation op) noexcept
Set the reduce operation for the layer.
Definition: NvInfer.h:2712
ReduceOperation getOperation() const noexcept
Get the reduce operation for the layer.
Definition: NvInfer.h:2722
uint32_t getReduceAxes() const noexcept
Get the axes over which to reduce for the layer.
Definition: NvInfer.h:2742
void setReduceAxes(uint32_t reduceAxes) noexcept
Set the axes over which to reduce.
Definition: NvInfer.h:2732
apiv::VReduceLayer * mImpl
Definition: NvInfer.h:2768
bool getKeepDimensions() const noexcept
Get the boolean that specifies whether or not to keep the reduced dimensions for the layer.
Definition: NvInfer.h:2762
A resize layer in a network definition.
Definition: NvInfer.h:4006
void setSelectorForSinglePixel(ResizeSelector selector) noexcept
Set coordinate selector function when resized to single pixel.
Definition: NvInfer.h:4167
void setNearestRounding(ResizeRoundMode value) noexcept
Set rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4191
int32_t getScales(int32_t size, float *scales) const noexcept
Copies resize scales to scales[0, ..., nbScales-1], where nbScales is the number of scales that were ...
Definition: NvInfer.h:4085
void setOutputDimensions(Dims const &dimensions) noexcept
Set the output dimensions.
Definition: NvInfer.h:4026
void setCubicCoeff(float A) noexcept
Set the coefficient 'A' used in cubic interpolation.
Definition: NvInfer.h:4223
void setScales(float const *scales, int32_t nbScales) noexcept
Set the resize scales.
Definition: NvInfer.h:4066
virtual ~IResizeLayer() noexcept=0
float getCubicCoeff() const noexcept
Get the coefficient 'A' used in cubic interpolation.
Definition: NvInfer.h:4233
ResizeSelector getSelectorForSinglePixel() const noexcept
Get the coordinate selector function when resized to single pixel.
Definition: NvInfer.h:4177
InterpolationMode getResizeMode() const noexcept
Get resize mode for an input tensor.
Definition: NvInfer.h:4107
void setCoordinateTransformation(ResizeCoordinateTransformation coordTransform) noexcept
Set coordinate transformation function.
Definition: NvInfer.h:4142
void setExcludeOutside(bool excludeFlag) noexcept
Set the state for excluding outside pixels.
Definition: NvInfer.h:4246
void setResizeMode(InterpolationMode interpolationMode) noexcept
Set resize mode for an input tensor.
Definition: NvInfer.h:4097
Dims getOutputDimensions() const noexcept
Get the output dimensions.
Definition: NvInfer.h:4036
ResizeRoundMode getNearestRounding() const noexcept
Get rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4201
bool getExcludeOutside() const noexcept
Get the state for excluding outside pixels.
Definition: NvInfer.h:4256
ResizeCoordinateTransformation getCoordinateTransformation() const noexcept
Get coordinate transformation function.
Definition: NvInfer.h:4152
A ReverseSequence layer in a network definition.
Definition: NvInfer.h:6291
void setSequenceAxis(int32_t sequenceAxis) noexcept
Set the sequence axis. Default is 0.
Definition: NvInfer.h:6324
int32_t getBatchAxis() const noexcept
Return the batch axis. Return 1 if no batch axis was set.
Definition: NvInfer.h:6311
apiv::VReverseSequenceLayer * mImpl
Definition: NvInfer.h:6340
int32_t getSequenceAxis() const noexcept
Return the sequence axis. Return 0 if no sequence axis was set.
Definition: NvInfer.h:6334
void setBatchAxis(int32_t batchAxis) noexcept
Set the batch axis. Default is 1.
Definition: NvInfer.h:6301
virtual ~IReverseSequenceLayer() noexcept=0
Layer that implements Rotary Position Embedding (RoPE) (https://arxiv.org/abs/2104....
Definition: NvInfer.h:7390
TRT_NODISCARD int32_t getRotaryEmbeddingDim() const noexcept
Get the number of hidden dimensions participating in RoPE. The default value is 0,...
Definition: NvInfer.h:7430
void setInterleaved(bool interleaved) noexcept
Set whether the input is in interleaved format, i.e., whether the 2-d vectors rotated are taken from ...
Definition: NvInfer.h:7397
virtual ~IRotaryEmbeddingLayer() noexcept=0
TRT_NODISCARD bool setRotaryEmbeddingDim(int32_t rotaryEmbeddingDim) noexcept
Set the number of hidden dimensions participating in RoPE. The default value is 0,...
Definition: NvInfer.h:7419
apiv::VRotaryEmbeddingLayer * mImpl
Definition: NvInfer.h:7452
TRT_NODISCARD bool getInterleaved() const noexcept
Get whether the input is in interleaved format. The default value is false.
Definition: NvInfer.h:7408
A Scale layer in a network definition.
Definition: NvInfer.h:1630
Weights getScale() const noexcept
Get the scale value.
Definition: NvInfer.h:1687
Weights getPower() const noexcept
Get the power value.
Definition: NvInfer.h:1707
void setScale(Weights scale) noexcept
Set the scale value.
Definition: NvInfer.h:1677
void setPower(Weights power) noexcept
Set the power value.
Definition: NvInfer.h:1697
ScaleMode getMode() const noexcept
Get the scale mode.
Definition: NvInfer.h:1647
void setShift(Weights shift) noexcept
Set the shift value.
Definition: NvInfer.h:1657
void setChannelAxis(int32_t channelAxis) noexcept
Set the channel axis.
Definition: NvInfer.h:1743
virtual ~IScaleLayer() noexcept=0
Weights getShift() const noexcept
Get the shift value.
Definition: NvInfer.h:1667
void setMode(ScaleMode mode) noexcept
Set the scale mode.
Definition: NvInfer.h:1637
int32_t getChannelAxis() const noexcept
Get the channel axis.
Definition: NvInfer.h:1722
A scatter layer in a network definition. Supports several kinds of scattering.
Definition: NvInfer.h:5904
void setMode(ScatterMode mode) noexcept
Set the scatter mode.
Definition: NvInfer.h:5911
apiv::VScatterLayer * mImpl
Definition: NvInfer.h:5945
void setAxis(int32_t axis) noexcept
Set the axis used by ScatterMode::kELEMENTS.
Definition: NvInfer.h:5931
int32_t getAxis() const noexcept
Get the axis.
Definition: NvInfer.h:5939
ScatterMode getMode() const noexcept
Get the scatter mode.
Definition: NvInfer.h:5921
virtual ~IScatterLayer() noexcept=0
Select elements from two data tensors based on a condition tensor.
Definition: NvInfer.h:4855
virtual ~ISelectLayer() noexcept=0
Layer type for getting shape of a tensor.
Definition: NvInfer.h:3318
virtual ~IShapeLayer() noexcept=0
apiv::VShapeLayer * mImpl
Definition: NvInfer.h:3320
Layer type for shuffling data.
Definition: NvInfer.h:2877
apiv::VShuffleLayer * mImpl
Definition: NvInfer.h:3035
virtual ~IShuffleLayer() noexcept=0
void setFirstTranspose(Permutation permutation) noexcept
Set the permutation applied by the first transpose operation.
Definition: NvInfer.h:2888
void setSecondTranspose(Permutation permutation) noexcept
Set the permutation applied by the second transpose operation.
Definition: NvInfer.h:2988
Dims getReshapeDimensions() const noexcept
Get the reshaped dimensions.
Definition: NvInfer.h:2941
void setReshapeDimensions(Dims const &dimensions) noexcept
Set the reshaped dimensions.
Definition: NvInfer.h:2928
Permutation getFirstTranspose() const noexcept
Get the permutation applied by the first transpose operation.
Definition: NvInfer.h:2900
Permutation getSecondTranspose() const noexcept
Get the permutation applied by the second transpose operation.
Definition: NvInfer.h:3000
bool getZeroIsPlaceholder() const noexcept
Get meaning of 0 in reshape dimensions.
Definition: NvInfer.h:3029
void setZeroIsPlaceholder(bool zeroIsPlaceholder) noexcept
Set meaning of 0 in reshape dimensions.
Definition: NvInfer.h:3016
Slices an input tensor into an output tensor based on the offset and strides.
Definition: NvInfer.h:3131
void setStride(Dims const &stride) noexcept
Set the stride for computing the output slice data.
Definition: NvInfer.h:3200
apiv::VSliceLayer * mImpl
Definition: NvInfer.h:3299
virtual ~ISliceLayer() noexcept=0
void setSize(Dims const &size) noexcept
Set the dimensions of the output slice.
Definition: NvInfer.h:3171
void setAxes(Dims const &axes) noexcept
Set the axes for this ISliceLayer.
Definition: NvInfer.h:3278
void setStart(Dims const &start) noexcept
Set the start offset that the slice layer uses to create the output slice.
Definition: NvInfer.h:3142
Dims getStart() const noexcept
Get the start offset for the slice layer.
Definition: NvInfer.h:3157
void setMode(SampleMode mode) noexcept
Set the slice mode.
Definition: NvInfer.h:3225
Dims getSize() const noexcept
Get dimensions of the output slice.
Definition: NvInfer.h:3186
SampleMode getMode() const noexcept
Get the slice mode.
Definition: NvInfer.h:3235
Dims getStride() const noexcept
Get the stride for the output slice.
Definition: NvInfer.h:3215
Dims getAxes() const noexcept
Get the axes for this ISliceLayer.
Definition: NvInfer.h:3293
A Softmax layer in a network definition.
Definition: NvInfer.h:1776
void setAxes(uint32_t axes) noexcept
Set the axis along which softmax is computed. Currently, only one axis can be set.
Definition: NvInfer.h:1798
virtual ~ISoftMaxLayer() noexcept=0
uint32_t getAxes() const noexcept
Get the axis along which softmax occurs.
Definition: NvInfer.h:1808
Layer that represents a squeeze operation, removing unit dimensions of the first input tensor on a se...
Definition: NvInfer.h:6468
apiv::VSqueezeLayer * mImpl
Definition: NvInfer.h:6485
virtual ~ISqueezeLayer() noexcept=0
A tensor in a network definition.
Definition: NvInfer.h:186
void setAllowedFormats(TensorFormats formats) noexcept
Set allowed formats for an input or output tensor. By default all formats are allowed....
Definition: NvInfer.h:301
void setDimensions(Dims const &dimensions) noexcept
Set the dimensions of a tensor.
Definition: NvInfer.h:234
void setName(char const *name) noexcept
Set the tensor name.
Definition: NvInfer.h:203
bool isExecutionTensor() const noexcept
Whether the tensor is an execution tensor.
Definition: NvInfer.h:366
char const * getName() const noexcept
Get the tensor name.
Definition: NvInfer.h:215
bool isShapeTensor() const noexcept
Whether the tensor is a shape tensor.
Definition: NvInfer.h:345
bool isNetworkInput() const noexcept
Whether the tensor is a network input.
Definition: NvInfer.h:271
bool isNetworkOutput() const noexcept
Whether the tensor is a network output.
Definition: NvInfer.h:279
DataType getType() const noexcept
Get the data type of a tensor.
Definition: NvInfer.h:263
virtual ~ITensor() noexcept=0
apiv::VTensor * mImpl
Definition: NvInfer.h:413
void setDimensionName(int32_t index, char const *name) noexcept
Name a dimension of an input tensor.
Definition: NvInfer.h:392
char const * getDimensionName(int32_t index) const noexcept
Get the name of an input dimension.
Definition: NvInfer.h:407
Dims getDimensions() const noexcept
Get the dimensions of a tensor.
Definition: NvInfer.h:248
TensorFormats getAllowedFormats() const noexcept
Get a bitmask of TensorFormat values that the tensor supports. For a shape tensor,...
Definition: NvInfer.h:314
Class to handle tactic timing info collected from builder.
Definition: NvInfer.h:10061
int64_t queryKeys(TimingCacheKey *keyBuffer, int64_t capacity) const noexcept
Query cache keys from Timing Cache.
Definition: NvInfer.h:10127
virtual ~ITimingCache() noexcept=0
bool combine(ITimingCache const &inputCache, bool ignoreMismatch) noexcept
Combine input timing cache into local instance.
Definition: NvInfer.h:10098
TimingCacheValue query(TimingCacheKey const &key) const noexcept
Query value in a cache entry.
Definition: NvInfer.h:10144
bool update(TimingCacheKey const &key, TimingCacheValue const &value) noexcept
Update values in a cache entry.
Definition: NvInfer.h:10166
apiv::VTimingCache * mImpl
Definition: NvInfer.h:10172
bool reset() noexcept
Empty the timing cache.
Definition: NvInfer.h:10108
Layer that represents a TopK reduction.
Definition: NvInfer.h:3360
void setK(int32_t k) noexcept
Set the static k value for the layer.
Definition: NvInfer.h:3391
void setReduceAxes(uint32_t reduceAxes) noexcept
Set which axes to reduce for the layer.
Definition: NvInfer.h:3415
TopKOperation getOperation() const noexcept
Get the operation for the layer.
Definition: NvInfer.h:3377
apiv::VTopKLayer * mImpl
Definition: NvInfer.h:3474
void setOperation(TopKOperation op) noexcept
Set the operation for the layer.
Definition: NvInfer.h:3367
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:3456
virtual ~ITopKLayer() noexcept=0
int32_t getK() const noexcept
Get the k value for the layer.
Definition: NvInfer.h:3405
uint32_t getReduceAxes() const noexcept
Get the axes to reduce for the layer.
Definition: NvInfer.h:3425
DataType getIndicesType() const noexcept
Return the TopK layer indices type.
Definition: NvInfer.h:3468
A layer that represents a trip-count limiter.
Definition: NvInfer.h:4662
virtual ~ITripLimitLayer() noexcept=0
TripLimit getTripLimit() const noexcept
Get a trip limiter type.
Definition: NvInfer.h:4667
Layer that represents an unary operation.
Definition: NvInfer.h:2592
void setOperation(UnaryOperation op) noexcept
Set the unary operation for the layer.
Definition: NvInfer.h:2601
apiv::VUnaryLayer * mImpl
Definition: NvInfer.h:2617
UnaryOperation getOperation() const noexcept
Get the unary operation for the layer.
Definition: NvInfer.h:2611
virtual ~IUnaryLayer() noexcept=0
Layer that represents an unsqueeze operation, which reshapes the first input tensor by inserting unit...
Definition: NvInfer.h:6500
virtual ~IUnsqueezeLayer() noexcept=0
apiv::VUnsqueezeLayer * mImpl
Definition: NvInfer.h:6518
An Interface class for version control.
Definition: NvInferRuntimeBase.h:282
Version information associated with a TRT interface.
Definition: NvInferRuntimeBase.h:247
An array of weights used as a layer parameter.
Definition: NvInferRuntime.h:121
Definition: NvInferRuntimeBase.h:419
Definition: NvInferRuntime.h:1652
Application-implemented logging interface for the builder, refitter and runtime.
Definition: NvInferRuntime.h:1575
Definition: NvInferPluginBase.h:206
Definition: NvInfer.h:10397
virtual bool stepComplete(char const *phaseName, int32_t step) noexcept=0
Signal that a step of an optimizer phase has finished.
virtual ~IProgressMonitor() noexcept=default
IProgressMonitor()=default
virtual void phaseFinish(char const *phaseName) noexcept=0
Signal that a phase of the optimizer has finished.
virtual void phaseStart(char const *phaseName, char const *parentPhase, int32_t nbSteps) noexcept=0
Signal that a phase of the optimizer has started.
Definition: NvInferRuntime.h:643
IBuilder * createInferBuilder(ILogger &logger) noexcept
Create an instance of an IBuilder class.
Definition: NvInfer.h:11669
The TensorRT API version 1 namespace.
Definition: NvInferPluginBase.h:29
uint32_t TacticSources
Represents a collection of one or more TacticSource values combine using bitwise-OR operations.
Definition: NvInferRuntime.h:2870
ResizeSelector
The coordinate selector when resize to single pixel output.
Definition: NvInfer.h:3917
@ kFORMULA
Use formula to map the original index.
@ kUPPER
Select the upper left pixel.
EngineCapability
List of supported engine capability flows.
Definition: NvInferRuntime.h:76
MemoryPoolType
The type for memory pools used by TensorRT.
Definition: NvInfer.h:10185
AttentionIOForm
Enumerates the layout of the input/output tensors in an Attention layer.
Definition: NvInfer.h:6728
TENSORRTAPI bool setInternalLibraryPath(AsciiChar const *path) noexcept
Set a custom directory path for loading internal TensorRT libraries when building engines.
ScaleMode
Controls how shift, scale and power are applied in a Scale layer.
Definition: NvInfer.h:1587
@ kUNIFORM
Identical coefficients across all elements of the tensor.
@ kCHANNEL
Per-channel coefficients.
RuntimePlatform
Describes the intended runtime platform (operating system and CPU architecture) for the execution of ...
Definition: NvInfer.h:9803
@ kNONE
Tensor is not an input or output.
HardwareCompatibilityLevel
Describes requirements of compatibility with GPU architectures other than that of the GPU on which th...
Definition: NvInfer.h:10293
@ kSAME_COMPUTE_CAPABILITY
CumulativeOperation
Enumerates the cumulative operations that may be performed by a Cumulative layer.
Definition: NvInfer.h:6536
BoundingBoxFormat
Representation of bounding box data used for the Boxes input tensor in INMSLayer.
Definition: NvInfer.h:6103
@ kCENTER_SIZES
(x_center, y_center, width, height) where (x_center, y_center) is the center point of the box
@ kCORNER_PAIRS
(x1, y1, x2, y2) where (x1, y1) and (x2, y2) are any pair of diagonal corners
ComputeCapability
Describes compute capability that an engine will be built for.
Definition: NvInfer.h:10339
@ kSM120
Target NVIDIA Blackwell GPU architecture (SM 12.0).
@ kSM121
Target NVIDIA GB10 GPU (SM 12.1).
@ kSM75
Target NVIDIA Turing GPU architecture (SM 7.5).
@ kSM80
Target NVIDIA Ampere GPU architecture (SM 8.0).
@ kCURRENT
Use the compute capability of the current GPU in the environment.
@ kSM89
Target NVIDIA Ada Lovelace GPU architecture (SM 8.9).
@ kSM86
Target NVIDIA Ampere GPU architecture (SM 8.6).
UnaryOperation
Enumerates the unary operations that may be performed by a Unary layer.
Definition: NvInfer.h:2545
@ kISINF
Return true if input value equals +/- infinity for floating-point data type.
@ kCOSH
Hyperbolic cosine.
@ kACOSH
Inverse hyperbolic cosine.
@ kERF
Gauss error function.
@ kISNAN
Return true if input value is a NaN for floating-point data type.
@ kROUND
Round to nearest even for floating-point data type.
@ kATANH
Inverse hyperbolic tangent.
@ kASINH
Inverse hyperbolic sine.
@ kSIGN
Sign, If input > 0, output 1; if input < 0, output -1; if input == 0, output 0.
ActivationType
Enumerates the types of activation to perform in an activation layer.
Definition: NvInfer.h:143
@ kSELU
Selu activation: x>0 ? beta * x : beta * (alpha*exp(x) - alpha)
@ kSCALED_TANH
Scaled tanh activation: alpha*tanh(beta*x)
@ kRELU
Rectified linear activation.
@ kELU
Elu activation: x>=0 ? x : alpha * (exp(x) - 1).
@ kLEAKY_RELU
LeakyRelu activation: x>=0 ? x : alpha * x.
@ kSOFTSIGN
Softsign activation: x / (1+|x|)
@ kHARD_SIGMOID
Hard sigmoid activation: max(0, min(1, alpha*x+beta))
@ kTHRESHOLDED_RELU
Thresholded ReLU activation: x>alpha ? x : 0.
@ kSIGMOID
Sigmoid activation.
@ kCLIP
Clip activation: max(alpha, min(beta, x))
@ kGELU_TANH
GELU tanh activation: 0.5 * x * (1 + tanh(sqrt(2/pi) * (0.044715F * pow(x, 3) + x)))
@ kGELU_ERF
GELU erf activation: 0.5 * x * (1 + erf(sqrt(0.5) * x))
@ kSOFTPLUS
Parametric softplus activation: alpha*log(exp(beta*x)+1)
FillOperation
Enumerates the tensor fill operations that may performed by a fill layer.
Definition: NvInfer.h:4920
ResizeRoundMode
The rounding mode for nearest neighbor resize.
Definition: NvInfer.h:3944
@ kHALF_DOWN
Round half down.
char_t AsciiChar
Definition: NvInferRuntimeBase.h:116
CausalMaskKind
Enumerates the causal mask alignment orientation for the attention.
Definition: NvInfer.h:6700
@ kUPPER_LEFT
Diagonal anchored at top-left corner (legacy default when causal=true).
@ kLOWER_RIGHT
Diagonal anchored at bottom-right corner (decode-aligned semantics).
PaddingMode
Enumerates the modes of padding to perform in convolution, deconvolution and pooling layer,...
Definition: NvInfer.h:763
@ kSAME_LOWER
Use SAME padding, with prePadding >= postPadding.
@ kEXPLICIT_ROUND_DOWN
Use explicit padding, rounding output size down.
@ kEXPLICIT_ROUND_UP
Use explicit padding, rounding output size up.
@ kSAME_UPPER
Use SAME padding, with prePadding <= postPadding.
TripLimit
Enum that describes kinds of trip limits.
Definition: NvInfer.h:4302
@ kWHILE
Tensor is a scalar of type kBOOL. Loop terminates when value is false.
@ kCOUNT
Tensor is a scalar of type kINT32 or kINT64 that contains the trip count.
uint32_t NetworkDefinitionCreationFlags
Represents one or more NetworkDefinitionCreationFlag flags using binary OR operations....
Definition: NvInfer.h:11345
PreviewFeature
Define preview features.
Definition: NvInfer.h:10259
@ kRUNTIME_ACTIVATION_RESIZE_10_10
@ kALIASED_PLUGIN_IO_10_03
TilingOptimizationLevel
Define the optimization levels for Tiling.
Definition: NvInfer.h:10367
@ kFAST
Use a fast algorithm and heuristic based strategy. Slightly increases engine build time.
@ kFULL
Increase search space even wider. Significantly increases engine build time.
DataType
The type of weights and tensors. The datatypes other than kBOOL, kINT32, and kINT64 are "activation d...
Definition: NvInferRuntimeBase.h:149
uint32_t BuilderFlags
Represents one or more BuilderFlag values using binary OR operations, e.g., 1U << BuilderFlag::kDEBUG...
Definition: NvInfer.h:9832
DeviceType
The device that this layer/network will execute on.
Definition: NvInferRuntime.h:1341
LayerType
The type values of layer classes.
Definition: NvInfer.h:58
@ kGRID_SAMPLE
Grid sample layer.
@ kRAGGED_SOFTMAX
Ragged softmax layer.
@ kDECONVOLUTION
Deconvolution layer.
@ kASSERTION
Assertion layer.
@ kATTENTION_INPUT
Attention Input.
@ kMATRIX_MULTIPLY
Matrix multiply layer.
@ kCONDITION
Condition layer.
@ kCUMULATIVE
Cumulative layer.
@ kCONDITIONAL_INPUT
Conditional Input layer.
@ kIDENTITY
Identity layer.
@ kNORMALIZATION
Normalization layer.
@ kQUANTIZE
Quantize layer.
@ kCONVOLUTION
Convolution layer.
@ kPARAMETRIC_RELU
Parametric ReLU layer.
@ kATTENTION_OUTPUT
Attention Output.
@ kUNSQUEEZE
Unsqueeze Layer.
@ kCONCATENATION
Concatenation layer.
@ kREVERSE_SEQUENCE
Reverse sequence layer.
@ kROTARY_EMBEDDING
Rotary Embedding layer.
@ kRECURRENCE
Loop Recurrence layer.
@ kDEQUANTIZE
Dequantize layer.
@ kPLUGIN_V3
PluginV3 layer.
@ kITERATOR
Loop Iterator layer.
@ kTRIP_LIMIT
Loop Trip limit layer.
@ kDYNAMIC_QUANTIZE
Dynamic Quantize layer.
@ kUNARY
UnaryOp operation Layer.
@ kACTIVATION
Activation layer.
@ kELEMENTWISE
Elementwise layer.
@ kPLUGIN_V2
PluginV2 layer.
@ kLOOP_OUTPUT
Loop output layer.
@ kCONDITIONAL_OUTPUT
Conditional Output layer.
@ kCONSTANT
Constant layer.
@ kNON_ZERO
NonZero layer.
@ kKVCACHE_UPDATE
KV Cache Update layer.
@ kDIST_COLLECTIVE
DistCollective layer.
SampleMode
Controls how ISliceLayer and IGridSample handle out-of-bounds coordinates.
Definition: NvInfer.h:3047
@ kCLAMP
Out of bounds indices are clamped to bounds.
@ kSTRICT_BOUNDS
Fail with error when the coordinates are out of bounds.
@ kWRAP
Coordinates wrap around periodically.
GatherMode
Control form of IGatherLayer.
Definition: NvInfer.h:2286
@ kDEFAULT
Similar to ONNX Gather.
@ kELEMENT
Similar to ONNX GatherElements.
@ kND
Similar to ONNX GatherND.
MoEActType
Enumerates the activation type for the MoE layer.
Definition: NvInfer.h:7617
uint32_t TensorFormats
It is capable of representing one or more TensorFormat by binary OR operations, e....
Definition: NvInfer.h:135
ProfilingVerbosity
List of verbosity levels of layer information exposed in NVTX annotations and in IEngineInspector.
Definition: NvInferRuntime.h:2882
NetworkDefinitionCreationFlag
List of immutable network properties expressed at network creation time. NetworkDefinitionCreationFla...
Definition: NvInfer.h:11356
ElementWiseOperation
Enumerates the binary operations that may be performed by an ElementWise layer.
Definition: NvInfer.h:2197
@ kSUB
Subtract the second element from the first.
@ kSUM
Sum of the two elements.
@ kPROD
Product of the two elements.
@ kFLOOR_DIV
Floor division of the first element by the second.
@ kEQUAL
Check if two elements are equal.
@ kAND
Logical AND of two elements.
@ kOR
Logical OR of two elements.
@ kMIN
Minimum of the two elements.
@ kPOW
The first element to the power of the second element.
@ kLESS
Check if element in first tensor is less than corresponding element in second tensor.
@ kGREATER
Check if element in first tensor is greater than corresponding element in second tensor.
@ kXOR
Logical XOR of two elements.
@ kDIV
Divide the first element by the second.
CollectiveOperation
Enumerates the collective operations that may be performed by a DistCollective layer.
Definition: NvInfer.h:2675
@ kALL_TO_ALL
All-to-all exchange.
@ kREDUCE_SCATTER
Reduce scatter.
InterpolationMode
Enumerates various modes of interpolation.
Definition: NvInfer.h:3841
@ kNEAREST
ND (0 < N <= 8) nearest neighbor resizing.
@ kCUBIC
Supports bicubic (2D) interpolation.
@ kLINEAR
Supports linear (1D), bilinear (2D), and trilinear (3D) interpolation.
BuilderFlag
List of valid modes that the builder can enable when creating an engine from a network definition.
Definition: NvInfer.h:9842
@ kWEIGHT_STREAMING
Enable weight streaming for the current engine.
@ kDEBUG
Enable debugging of layers via synchronizing after every layer.
@ kGPU_FALLBACK
Enable layers marked to execute on GPU if layer cannot execute on DLA.
@ kSPARSE_WEIGHTS
Allow the builder to examine weights and use optimized functions when weights have suitable sparsity.
@ kREQUIRE_USER_ALLOCATION
@ kERROR_ON_TIMING_CACHE_MISS
@ kEDITABLE_TIMING_CACHE
Enable editable timing cache.
@ kDISTRIBUTIVE_INDEPENDENCE
@ kSTRIP_PLAN
Strip the refittable weights from the engine plan file.
@ kMONITOR_MEMORY
Enable memory monitor during build time.
@ kDISABLE_TIMING_CACHE
Disable reuse of timing information across identical layers.
@ kDISABLE_COMPILATION_CACHE
@ kREFIT
Enable building a refittable engine.
TENSORRTAPI nvinfer1::IPluginRegistry * getBuilderPluginRegistry(nvinfer1::EngineCapability capability) noexcept
Return the plugin registry for building a Standard engine, or nullptr if no registry exists.
TopKOperation
Enumerates the operations that may be performed by a TopK layer.
Definition: NvInfer.h:3332
ReduceOperation
Enumerates the reduce operations that may be performed by a Reduce layer.
Definition: NvInfer.h:2647
@ kAVG
Average of the elements.
ScatterMode
Control form of IScatterLayer.
Definition: NvInfer.h:5830
MatrixOperation
Enumerates the operations that may be performed on a tensor by IMatrixMultiplyLayer before multiplica...
Definition: NvInfer.h:3487
@ kTRANSPOSE
Like kNONE, but transpose the matrix dimensions.
ResizeCoordinateTransformation
The resize coordinate transformation function.
Definition: NvInfer.h:3866
LoopOutput
Enum that describes kinds of loop outputs.
Definition: NvInfer.h:4274
@ kLAST_VALUE
Output value is value of tensor for last iteration.
@ kCONCATENATE
Output value is concatenation of values of tensor for each iteration, in forward order.
@ kREVERSE
Output value is concatenation of values of tensor for each iteration, in reverse order.
KVCacheMode
Enumerates the KVCache modes that may be performed by a KVCacheUpdate layer.
Definition: NvInfer.h:7464
PoolingType
The type of pooling to perform in a pooling layer.
Definition: NvInfer.h:1200
@ kAVERAGE
Average over elements. If the tensor is padded, the count includes the padding.
@ kMAX
Maximum over elements.
@ kMAX_AVERAGE_BLEND
Blending between max and average pooling: (1-blendFactor)*maxPool + blendFactor*avgPool.
v_1_0::IProgressMonitor IProgressMonitor
Definition: NvInfer.h:10480
AttentionNormalizationOp
Enumerates the operations that may be performed by the normalization in the attention subgraph.
Definition: NvInfer.h:6668
Represents a permutation of dimensions.
Definition: NvInfer.h:2854
Declaration of EnumMaxImpl struct to store the exclusive upper bound of an enumeration type.
Definition: NvInferRuntimeBase.h:131
The key to retrieve timing cache entries.
Definition: NvInfer.h:10021
Definition: NvInfer.h:10035
uint64_t tacticHash
Hash of the selected tactic.
Definition: NvInfer.h:10037
float timingMSec
Timing of this tactic in milliseconds. Negative numbers and NaN are invalid values.
Definition: NvInfer.h:10039