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();
298 mImpl->setBroadcastAcrossBatch(broadcastAcrossBatch);
312 return mImpl->getBroadcastAcrossBatch();
324 return mImpl->getLocation();
343 mImpl->setLocation(location);
366 mImpl->setAllowedFormats(formats);
379 return mImpl->getAllowedFormats();
410 return mImpl->isShapeTensor();
431 return mImpl->isExecutionTensor();
457 mImpl->setDimensionName(index, name);
472 return mImpl->getDimensionName(index);
499 return mLayer->getType();
513 mLayer->setName(name);
523 return mLayer->getName();
531 return mLayer->getNbInputs();
544 return mLayer->getInput(index);
552 return mLayer->getNbOutputs();
562 return mLayer->getOutput(index);
579 return mLayer->setInput(index, tensor);
594 return mLayer->getOutputType(index);
613 mLayer->setMetadata(metadata);
626 return mLayer->getMetadata();
647 return mLayer->setNbRanks(nbRanks);
659 return mLayer->getNbRanks();
664 apiv::VLayer* mLayer;
841 static constexpr int32_t kVALUE = 4;
868 mImpl->setNbOutputMaps(nbOutputMaps);
878 return mImpl->getNbOutputMaps();
898 mImpl->setNbGroups(nbGroups);
908 return mImpl->getNbGroups();
922 mImpl->setKernelWeights(weights);
932 return mImpl->getKernelWeights();
947 mImpl->setBiasWeights(weights);
957 return mImpl->getBiasWeights();
974 mImpl->setPrePadding(padding);
984 return mImpl->getPrePadding();
1001 mImpl->setPostPadding(padding);
1011 return mImpl->getPostPadding();
1025 mImpl->setPaddingMode(paddingMode);
1037 return mImpl->getPaddingMode();
1050 mImpl->setKernelSizeNd(kernelSize);
1060 return mImpl->getKernelSizeNd();
1075 mImpl->setStrideNd(stride);
1085 return mImpl->getStrideNd();
1103 mImpl->setPaddingNd(padding);
1115 return mImpl->getPaddingNd();
1129 mImpl->setDilationNd(dilation);
1139 return mImpl->getDilationNd();
1190 mImpl->setActivationType(type);
1200 return mImpl->getActivationType();
1215 mImpl->setAlpha(alpha);
1229 mImpl->setBeta(beta);
1238 return mImpl->getAlpha();
1247 return mImpl->getBeta();
1277 static constexpr int32_t kVALUE = 3;
1303 mImpl->setPoolingType(type);
1313 return mImpl->getPoolingType();
1328 mImpl->setBlendFactor(blendFactor);
1341 return mImpl->getBlendFactor();
1355 mImpl->setAverageCountExcludesPadding(exclusive);
1366 return mImpl->getAverageCountExcludesPadding();
1384 mImpl->setPrePadding(padding);
1394 return mImpl->getPrePadding();
1412 mImpl->setPostPadding(padding);
1422 return mImpl->getPostPadding();
1435 mImpl->setPaddingMode(paddingMode);
1446 return mImpl->getPaddingMode();
1459 mImpl->setWindowSizeNd(windowSize);
1469 return mImpl->getWindowSizeNd();
1483 mImpl->setStrideNd(stride);
1493 return mImpl->getStrideNd();
1512 mImpl->setPaddingNd(padding);
1524 return mImpl->getPaddingNd();
1557 mImpl->setWindowSize(windowSize);
1567 return mImpl->getWindowSize();
1579 mImpl->setAlpha(alpha);
1589 return mImpl->getAlpha();
1601 mImpl->setBeta(beta);
1611 return mImpl->getBeta();
1633 return mImpl->getK();
1663 static constexpr int32_t kVALUE = 3;
1701 mImpl->setMode(mode);
1711 return mImpl->getMode();
1721 mImpl->setShift(shift);
1731 return mImpl->getShift();
1741 mImpl->setScale(scale);
1751 return mImpl->getScale();
1761 mImpl->setPower(power);
1771 return mImpl->getPower();
1786 return mImpl->getChannelAxis();
1807 mImpl->setChannelAxis(channelAxis);
1862 mImpl->setAxes(axes);
1872 return mImpl->getAxes();
1910 mImpl->setAxis(axis);
1920 return mImpl->getAxis();
1949 mImpl->setNbOutputMaps(nbOutputMaps);
1959 return mImpl->getNbOutputMaps();
1979 mImpl->setNbGroups(nbGroups);
1989 return mImpl->getNbGroups();
2003 mImpl->setKernelWeights(weights);
2013 return mImpl->getKernelWeights();
2028 mImpl->setBiasWeights(weights);
2038 return mImpl->getBiasWeights();
2055 mImpl->setPrePadding(padding);
2065 return mImpl->getPrePadding();
2082 mImpl->setPostPadding(padding);
2092 return mImpl->getPostPadding();
2106 mImpl->setPaddingMode(paddingMode);
2118 return mImpl->getPaddingMode();
2133 mImpl->setKernelSizeNd(kernelSize);
2143 return mImpl->getKernelSizeNd();
2160 mImpl->setStrideNd(stride);
2170 return mImpl->getStrideNd();
2188 mImpl->setPaddingNd(padding);
2200 return mImpl->getPaddingNd();
2226 mImpl->setDilationNd(dilation);
2236 return mImpl->getDilationNd();
2284 static constexpr int32_t kVALUE = 14;
2320 return mImpl->setOperation(op);
2332 return mImpl->getOperation();
2362 static constexpr int32_t kVALUE = 3;
2455 mImpl->setGatherAxis(axis);
2467 return mImpl->getGatherAxis();
2490 mImpl->setNbElementWiseDims(elementWiseDims);
2500 return mImpl->getNbElementWiseDims();
2510 mImpl->setMode(mode);
2520 return mImpl->getMode();
2551 return mImpl->getPlugin();
2580 return mImpl->getPlugin();
2643 static constexpr int32_t kVALUE = 25;
2665 mImpl->setOperation(op);
2675 return mImpl->getOperation();
2726 static constexpr int32_t kVALUE = 6;
2756 static constexpr int32_t kVALUE = 8;
2776 mImpl->setOperation(op);
2786 return mImpl->getOperation();
2796 mImpl->setReduceAxes(reduceAxes);
2806 return mImpl->getReduceAxes();
2816 mImpl->setKeepDimensions(keepDimensions);
2826 return mImpl->getKeepDimensions();
2862 mImpl->setPrePaddingNd(padding);
2874 return mImpl->getPrePaddingNd();
2888 mImpl->setPostPaddingNd(padding);
2900 return mImpl->getPostPaddingNd();
2952 mImpl->setFirstTranspose(permutation);
2964 return mImpl->getFirstTranspose();
2992 mImpl->setReshapeDimensions(dimensions);
3005 return mImpl->getReshapeDimensions();
3052 mImpl->setSecondTranspose(permutation);
3064 return mImpl->getSecondTranspose();
3080 return mImpl->setZeroIsPlaceholder(zeroIsPlaceholder);
3093 return mImpl->getZeroIsPlaceholder();
3127 static constexpr int32_t kVALUE = 5;
3206 mImpl->setStart(start);
3221 return mImpl->getStart();
3235 return mImpl->setSize(size);
3250 return mImpl->getSize();
3264 mImpl->setStride(stride);
3279 return mImpl->getStride();
3289 mImpl->setMode(mode);
3299 return mImpl->getMode();
3342 mImpl->setAxes(axes);
3357 return mImpl->getAxes();
3407 static constexpr int32_t kVALUE = 2;
3431 mImpl->setOperation(op);
3441 return mImpl->getOperation();
3469 return mImpl->getK();
3479 mImpl->setReduceAxes(reduceAxes);
3489 return mImpl->getReduceAxes();
3520 return mImpl->setIndicesType(type);
3532 return mImpl->getIndicesType();
3579 static constexpr int32_t kVALUE = 3;
3620 mImpl->setOperation(index, op);
3632 return mImpl->getOperation(index);
3678 return mImpl->setIndicesType(type);
3690 return mImpl->getIndicesType();
3793 mImpl->setToType(toType);
3804 return mImpl->getToType();
3835 mImpl->setWeights(weights);
3845 return mImpl->getWeights();
3857 mImpl->setDimensions(dimensions);
3869 return mImpl->getDimensions();
3917 static constexpr int32_t kVALUE = 3;
3968 static constexpr int32_t kVALUE = 3;
3995 static constexpr int32_t kVALUE = 2;
4028 static constexpr int32_t kVALUE = 4;
4090 return mImpl->setOutputDimensions(dimensions);
4100 return mImpl->getOutputDimensions();
4128 void setScales(
float const* scales, int32_t nbScales)
noexcept
4130 mImpl->setScales(scales, nbScales);
4147 int32_t
getScales(int32_t size,
float* scales)
const noexcept
4149 return mImpl->getScales(size, scales);
4161 mImpl->setResizeMode(interpolationMode);
4171 return mImpl->getResizeMode();
4206 mImpl->setCoordinateTransformation(coordTransform);
4216 return mImpl->getCoordinateTransformation();
4231 mImpl->setSelectorForSinglePixel(selector);
4241 return mImpl->getSelectorForSinglePixel();
4255 mImpl->setNearestRounding(value);
4265 return mImpl->getNearestRounding();
4287 mImpl->setCubicCoeff(A);
4297 return mImpl->getCubicCoeff();
4310 mImpl->setExcludeOutside(excludeFlag);
4320 return mImpl->getExcludeOutside();
4355 static constexpr int32_t kVALUE = 3;
4378 static constexpr int32_t kVALUE = 2;
4404 return mBoundary->getLoop();
4409 apiv::VLoopBoundaryLayer* mBoundary;
4429 return mBoundary->getConditional();
4434 apiv::VConditionalBoundaryLayer* mBoundary;
4526 return mImpl->setCondition(condition);
4544 return mImpl->addOutput(trueSubgraphOutput, falseSubgraphOutput);
4556 return mImpl->addInput(input);
4571 mImpl->setName(name);
4581 return mImpl->getName();
4655 return mImpl->getLoopOutput();
4672 mImpl->setAxis(axis);
4680 return mImpl->getAxis();
4731 return mImpl->getTripLimit();
4759 mImpl->setAxis(axis);
4767 return mImpl->getAxis();
4781 mImpl->setReverse(reverse);
4791 return mImpl->getReverse();
4822 return mImpl->addRecurrence(initialValue);
4843 return mImpl->addTripLimit(tensor, limit);
4856 return mImpl->addIterator(tensor, axis, reverse);
4869 return mImpl->addLoopOutput(tensor, outputKind, axis);
4884 mImpl->setName(name);
4894 return mImpl->getName();
4953 mImpl->setMessage(message);
4963 return mImpl->getMessage();
5018 static constexpr int32_t kVALUE = 3;
5070 mImpl->setDimensions(dimensions);
5085 return mImpl->getDimensions();
5095 mImpl->setOperation(op);
5105 return mImpl->getOperation();
5124 mImpl->setAlpha(alpha);
5139 return mImpl->getAlpha();
5158 mImpl->setBeta(beta);
5173 return mImpl->getBeta();
5234 mImpl->setAlphaInt64(alpha);
5249 return mImpl->getAlphaInt64();
5268 mImpl->setBetaInt64(beta);
5283 return mImpl->getBetaInt64();
5291 return mImpl->isAlphaBetaInt64();
5309 mImpl->setToType(toType);
5321 return mImpl->getToType();
5418 return mImpl->getAxis();
5429 mImpl->setAxis(axis);
5442 return mImpl->setBlockShape(blockShape);
5453 return mImpl->getBlockShape();
5469 mImpl->setToType(toType);
5481 return mImpl->getToType();
5572 return mImpl->getAxis();
5583 mImpl->setAxis(axis);
5600 return mImpl->setBlockShape(blockShape);
5611 return mImpl->getBlockShape();
5627 mImpl->setToType(toType);
5639 return mImpl->getToType();
5696 mImpl->setToType(toType);
5709 return mImpl->getToType();
5722 mImpl->setScaleType(scaleType);
5735 return mImpl->getScaleType();
5748 mImpl->setAxis(axis);
5758 return mImpl->getAxis();
5771 mImpl->setBlockSize(size);
5781 return mImpl->getBlockSize();
5794 mImpl->setBlockShape(blockShape);
5806 return mImpl->getBlockShape();
5864 return mImpl->setEquation(equation);
5874 return mImpl->getEquation();
5905 static constexpr int32_t kVALUE = 2;
5975 mImpl->setMode(mode);
5985 return mImpl->getMode();
5995 mImpl->setAxis(axis);
6003 return mImpl->getAxis();
6050 mImpl->setAxis(axis);
6058 return mImpl->getAxis();
6089 mImpl->setInterpolationMode(mode);
6101 return mImpl->getInterpolationMode();
6111 mImpl->setAlignCorners(alignCorners);
6123 return mImpl->getAlignCorners();
6135 return mImpl->setSampleMode(mode);
6147 return mImpl->getSampleMode();
6180 static constexpr int32_t kVALUE = 2;
6247 mImpl->setBoundingBoxFormat(fmt);
6259 return mImpl->getBoundingBoxFormat();
6273 mImpl->setTopKBoxLimit(limit);
6283 return mImpl->getTopKBoxLimit();
6318 return mImpl->setIndicesType(type);
6330 return mImpl->getIndicesType();
6365 mImpl->setBatchAxis(batchAxis);
6375 return mImpl->getBatchAxis();
6388 mImpl->setSequenceAxis(sequenceAxis);
6398 return mImpl->getSequenceAxis();
6438 return mImpl->setEpsilon(eps);
6448 return mImpl->getEpsilon();
6458 return mImpl->setAxes(axesMask);
6468 return mImpl->getAxes();
6489 return mImpl->setNbGroups(nbGroups);
6499 return mImpl->getNbGroups();
6510 return mImpl->isV2();
6616 static constexpr int32_t kVALUE = 1;
6660 return mImpl->setOperation(op);
6672 return mImpl->getOperation();
6684 mImpl->setExclusive(exclusive);
6696 return mImpl->getExclusive();
6708 mImpl->setReverse(reverse);
6720 return mImpl->getReverse();
6750 static constexpr int32_t kVALUE = 2;
6787 static constexpr int32_t kVALUE = 3;
6813 static constexpr int32_t kVALUE = 2;
6833 return mBoundary->getAttention();
6838 apiv::VAttentionBoundaryLayer* mBoundary;
6974 return mImpl->setNormalizationOperation(op);
6986 return mImpl->getNormalizationOperation();
7003 return mImpl->setMask(mask);
7015 return mImpl->getMask();
7033 return mImpl->setCausal(isCausal);
7047 return mImpl->getCausal();
7067 return mImpl->setCausalKind(kind);
7079 return mImpl->getCausalKind();
7091 return mImpl->setDecomposable(decomposable);
7104 return mImpl->getDecomposable();
7123 return mImpl->setInput(index, input);
7132 return mImpl->getNbInputs();
7144 return mImpl->getInput(index);
7152 return mImpl->getNbOutputs();
7164 return mImpl->getOutput(index);
7181 return mImpl->setName(name);
7193 return mImpl->getName();
7209 return mImpl->setNormalizationQuantizeScale(tensor);
7220 return mImpl->getNormalizationQuantizeScale();
7233 return mImpl->setNormalizationQuantizeToType(type);
7245 return mImpl->getNormalizationQuantizeToType();
7265 return mImpl->setMetadata(metadata);
7278 return mImpl->getMetadata();
7294 return mImpl->setNbRanks(nbRanks);
7306 return mImpl->getNbRanks();
7323 return mImpl->setQueryForm(form);
7336 return mImpl->getQueryForm();
7353 return mImpl->setKeyValueForm(form);
7366 return mImpl->getKeyValueForm();
7390 return mImpl->setQueryLengths(lengths);
7402 return mImpl->getQueryLengths();
7429 return mImpl->setKeyValueLengths(lengths);
7441 return mImpl->getKeyValueLengths();
7467 mImpl->setInterleaved(interleaved);
7478 return mImpl->getInterleaved();
7489 return mImpl->setRotaryEmbeddingDim(rotaryEmbeddingDim);
7500 return mImpl->getRotaryEmbeddingDim();
7544 static constexpr int32_t kVALUE = 1;
7594 return mImpl->setCacheMode(cacheMode);
7604 return mImpl->getCacheMode();
7622 return mImpl->setUpdateForm(form);
7635 return mImpl->getUpdateForm();
7657 return mImpl->setUpdateLengths(lengths);
7669 return mImpl->getUpdateLengths();
7698 static constexpr int32_t kVALUE = 2;
7833 mImpl->setGatedWeights(fcGateWeights, fcUpWeights, fcDownWeights, activationType);
7845 mImpl->setGatedBiases(fcGateBiases, fcUpBiases, fcDownBiases);
7857 mImpl->setActivationType(activationType);
7869 return mImpl->getActivationType();
7895 mImpl->setQuantizationStatic(fcDownActivationScale, dataType);
7928 mImpl->setQuantizationDynamicDblQ(fcDownActivationDblQScale, dataType, blockShape, dynQOutputScaleType);
7943 mImpl->setQuantizationToType(type);
7955 return mImpl->getQuantizationToType();
7971 mImpl->setQuantizationBlockShape(blockShape);
7983 return mImpl->getQuantizationBlockShape();
7995 mImpl->setDynQOutputScaleType(type);
8007 return mImpl->getDynQOutputScaleType();
8028 mImpl->setSwigluParams(limit, alpha, beta);
8042 mImpl->setSwigluParamLimit(limit);
8054 return mImpl->getSwigluParamLimit();
8068 mImpl->setSwigluParamAlpha(alpha);
8080 return mImpl->getSwigluParamAlpha();
8094 mImpl->setSwigluParamBeta(beta);
8106 return mImpl->getSwigluParamBeta();
8123 mImpl->setInput(index, tensor);
8210 return mImpl->addInput(name, type, dimensions);
8224 mImpl->markOutput(tensor);
8242 return mImpl->markDebug(tensor);
8258 return mImpl->unmarkDebug(tensor);
8268 return mImpl->isDebugTensor(tensor);
8290 return mImpl->markUnfusedTensorsAsDebugTensors();
8304 return mImpl->unmarkUnfusedTensorsAsDebugTensors();
8324 return mImpl->addActivation(input, type);
8343 return mImpl->addLRN(input, window, alpha, beta, k);
8369 return mImpl->addScale(input, mode, shift, scale, power);
8382 return mImpl->addSoftMax(input);
8399 return mImpl->addConcatenation(inputs, nbInputs);
8426 return mImpl->addElementWise(input1, input2, op);
8448 return mImpl->addUnary(input, operation);
8462 return mImpl->addShuffle(input);
8479 return mImpl->addOneHot(indices, values, depth, axis);
8491 return mImpl->getNbLayers();
8505 return mImpl->getLayer(index);
8517 return mImpl->getNbInputs();
8533 return mImpl->getInput(index);
8547 return mImpl->getNbOutputs();
8563 return mImpl->getOutput(index);
8590 return mImpl->addReduce(input, operation, reduceAxes, keepDimensions);
8625 return mImpl->addTopK(input, op, k, reduceAxes);
8658 return mImpl->addTopKV2(input, op, k, reduceAxes, indicesType);
8674 return mImpl->addGather(data, indices, axis);
8690 return mImpl->addGatherV2(data, indices, mode);
8709 return mImpl->addRaggedSoftMax(input, bounds);
8731 return mImpl->addMatrixMultiply(input0, op0, input1, op1);
8749 return mImpl->addNonZero(input);
8765 return mImpl->addNonZeroV2(input, indicesType);
8789 return mImpl->addConstant(dimensions, weights);
8803 return mImpl->addIdentity(input);
8818 return mImpl->addCast(input, toType);
8833 mImpl->removeTensor(tensor);
8845 mImpl->unmarkOutput(tensor);
8866 return mImpl->addPluginV2(inputs, nbInputs, plugin);
8883 int32_t nbShapeInputs,
IPluginV3& plugin)
noexcept
8885 return mImpl->addPluginV3(inputs, nbInputs, shapeInputs, nbShapeInputs, plugin);
8904 return mImpl->addSlice(input, start, size, stride);
8928 mImpl->setName(name);
8942 return mImpl->getName();
8958 return mImpl->addShape(input);
8972 return mImpl->hasImplicitBatchDimension();
8982 return mImpl->getFlags();
8994 return mImpl->getFlag(networkDefinitionCreationFlag);
9011 return mImpl->markOutputForShapes(tensor);
9023 return mImpl->unmarkOutputForShapes(tensor);
9041 return mImpl->addParametricReLU(input, slope);
9064 return mImpl->addConvolutionNd(input, nbOutputMaps, kernelSize, kernelWeights, biasWeights);
9083 return mImpl->addPoolingNd(input, type, windowSize);
9106 return mImpl->addDeconvolutionNd(input, nbOutputMaps, kernelSize, kernelWeights, biasWeights);
9143 return mImpl->addScaleNd(input, mode, shift, scale, power, channelAxis);
9159 return mImpl->addResize(input);
9173 return mImpl->addLoop();
9188 return mImpl->addIfConditional();
9227 return mImpl->addSelect(condition, thenInput, elseInput);
9244 return mImpl->addAssertion(condition, message);
9270 return mImpl->addFillV2(dimensions, op, outputType);
9286 return mImpl->addPaddingNd(input, prePadding, postPadding);
9310 return mImpl->setWeightsName(weights, name);
9329 mImpl->setErrorRecorder(recorder);
9344 return mImpl->getErrorRecorder();
9367 return mImpl->addDequantizeV2(input, scale, outputType);
9387 return mImpl->addScatter(data, indices, updates, mode);
9411 return mImpl->addQuantizeV2(input, scale, outputType);
9439 return mImpl->addDynamicQuantize(input, axis, blockSize, outputType, scaleType);
9463 return mImpl->addDynamicQuantizeV2(input, blockShape, outputType, scaleType);
9478 return mImpl->addEinsum(inputs, nbInputs, equation);
9496 return mImpl->addGridSample(input, grid);
9518 return mImpl->addNMS(boxes, scores, maxOutputBoxesPerClass);
9538 return mImpl->addNMSV2(boxes, scores, maxOutputBoxesPerClass, indicesType);
9555 return mImpl->addReverseSequence(input, sequenceLens);
9587 return mImpl->addNormalization(input, scale, bias, axesMask);
9609 return mImpl->addCumulative(input, axis, operation, exclusive, reverse);
9640 return mImpl->addAttention(query, key, value, normOp, causal);
9671 return mImpl->addAttentionV2(query, key, value, normOp, causalKind);
9695 return mImpl->addRotaryEmbedding(input, cosCache, sinCache, interleaved, rotaryEmbeddingDim);
9730 return mImpl->addKVCacheUpdate(cache, update, writeIndices, cacheMode);
9751 return mImpl->addMoE(hiddenStates, selectedExpertsForTokens, scoresForSelectedExperts);
9782 ReduceOperation reduceOp, int64_t root, int64_t* groups, int64_t groupSize)
noexcept
9784 return mImpl->addDistCollective(input, distCollectiveOp, reduceOp, root, groups, groupSize);
9795 return mImpl->getBuilder();
9810 return mImpl->markWeightsRefittable(name);
9823 return mImpl->unmarkWeightsRefittable(name);
9836 return mImpl->areWeightsMarkedRefittable(name);
9855 return mImpl->addSqueeze(input, axes);
9884 return mImpl->addSqueezeV2(input, axes);
9905 return mImpl->addUnsqueeze(input, axes);
9931 return mImpl->addNormalizationV2(input, scale, bias, axesMask);
9978 static constexpr int32_t kVALUE = 2;
10156 static constexpr int32_t kVALUE = 28;
10192 static constexpr uint64_t kINVALID_TACTIC_HASH = UINT64_MAX;
10225 return mImpl->serialize();
10249 return mImpl->combine(inputCache, ignoreMismatch);
10259 return mImpl->reset();
10276 int64_t
queryKeys(TimingCacheKey* keyBuffer, int64_t capacity)
const noexcept
10278 return mImpl->queryKeys(keyBuffer, capacity);
10293 TimingCacheValue
query(TimingCacheKey
const& key)
const noexcept
10295 return mImpl->query(key);
10315 bool update(TimingCacheKey
const& key, TimingCacheValue
const& value)
noexcept
10317 return mImpl->update(key, value);
10401 static constexpr int32_t kVALUE = 6;
10435 static constexpr int32_t kVALUE = 2;
10487 static constexpr int32_t kVALUE = 3;
10524 static constexpr int32_t kVALUE = 4;
10562 virtual void phaseStart(
char const* phaseName,
char const* parentPhase, int32_t nbSteps)
noexcept = 0;
10635 virtual
void setAvgTimingIterations(int32_t avgTiming) noexcept
10637 mImpl->setAvgTimingIterations(avgTiming);
10649 return mImpl->getAvgTimingIterations();
10662 mImpl->setEngineCapability(capability);
10674 return mImpl->getEngineCapability();
10691 mImpl->setFlags(builderFlags);
10703 return mImpl->getFlags();
10715 mImpl->clearFlag(builderFlag);
10727 mImpl->setFlag(builderFlag);
10739 return mImpl->getFlag(builderFlag);
10756 mImpl->setDeviceType(layer, deviceType);
10766 return mImpl->getDeviceType(layer);
10778 return mImpl->isDeviceTypeSet(layer);
10788 mImpl->resetDeviceType(layer);
10798 return mImpl->canRunOnDLA(layer);
10814 mImpl->setDLACore(dlaCore);
10824 return mImpl->getDLACore();
10835 mImpl->setDefaultDeviceType(deviceType);
10845 return mImpl->getDefaultDeviceType();
10894 return mImpl->setProfileStream(stream);
10906 return mImpl->getProfileStream();
10923 return mImpl->addOptimizationProfile(profile);
10936 return mImpl->getNbOptimizationProfiles();
10948 mImpl->setProfilingVerbosity(verbosity);
10961 return mImpl->getProfilingVerbosity();
10983 return mImpl->setTacticSources(tacticSources);
10998 return mImpl->getTacticSources();
11018 return mImpl->createTimingCache(blob, size);
11041 return mImpl->setTimingCache(cache, ignoreMismatch);
11051 return mImpl->getTimingCache();
11083 mImpl->setMemoryPoolLimit(pool, poolSize);
11102 return mImpl->getMemoryPoolLimit(pool);
11120 mImpl->setPreviewFeature(feature, enable);
11134 return mImpl->getPreviewFeature(feature);
11167 mImpl->setBuilderOptimizationLevel(level);
11179 return mImpl->getBuilderOptimizationLevel();
11196 mImpl->setHardwareCompatibilityLevel(hardwareCompatibilityLevel);
11209 return mImpl->getHardwareCompatibilityLevel();
11222 mImpl->setPluginsToSerialize(paths, nbPaths);
11235 return mImpl->getPluginToSerialize(index);
11245 return mImpl->getNbPluginsToSerialize();
11276 return mImpl->setMaxAuxStreams(nbStreams);
11286 return mImpl->getMaxAuxStreams();
11302 return mImpl->setProgressMonitor(monitor);
11312 return mImpl->getProgressMonitor();
11328 mImpl->setRuntimePlatform(runtimePlatform);
11340 return mImpl->getRuntimePlatform();
11352 mImpl->setMaxNbTactics(maxNbTactics);
11364 return mImpl->getMaxNbTactics();
11380 return mImpl->setTilingOptimizationLevel(level);
11392 return mImpl->getTilingOptimizationLevel();
11408 return mImpl->setL2LimitForTiling(size);
11420 return mImpl->getL2LimitForTiling();
11434 return mImpl->setRemoteAutoTuningConfig(config);
11444 return mImpl->getRemoteAutoTuningConfig();
11467 return mImpl->setBuildRoute(buildRoute);
11481 return mImpl->getBuildRoute();
11517 return mImpl->getAllBuildRoutes();
11572 static constexpr int32_t kVALUE = 3;
11595 int32_t getMaxDLABatchSize() const noexcept
11597 return mImpl->getMaxDLABatchSize();
11605 return mImpl->getNbDLACores();
11623 mImpl->setGpuAllocator(allocator);
11637 return mImpl->createBuilderConfig();
11664 return mImpl->createNetworkV2(flags);
11679 return mImpl->createOptimizationProfile();
11698 mImpl->setErrorRecorder(recorder);
11713 return mImpl->getErrorRecorder();
11741 return mImpl->buildSerializedNetwork(network, config);
11764 return mImpl->buildSerializedNetworkToStream(network, config, writer);
11789 return mImpl->buildSerializedNetworkWithKernelText(network, config, kernelText);
11811 return mImpl->buildEngineWithConfig(network, config);
11837 return mImpl->isNetworkSupported(network, config);
11847 return mImpl->getLogger();
11863 return mImpl->setMaxThreads(maxThreads);
11877 return mImpl->getMaxThreads();
11887 return mImpl->getPluginRegistry();
11902extern "C"
TENSORRTAPI void* createInferBuilder_INTERNAL(
void* logger, int32_t version) noexcept;
#define TRT_DEPRECATED_API
Definition: NvInferRuntimeBase.h:44
#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:224
static constexpr int32_t MAX_DIMS
The maximum rank (number of dimensions) supported for a tensor.
Definition: NvInferRuntimeBase.h:227
An Activation layer in a network definition.
Definition: NvInfer.h:1179
void setBeta(float beta) noexcept
Set the beta parameter (must be finite).
Definition: NvInfer.h:1227
void setActivationType(ActivationType type) noexcept
Set the type of activation to be performed.
Definition: NvInfer.h:1188
ActivationType getActivationType() const noexcept
Get the type of activation to be performed.
Definition: NvInfer.h:1198
float getAlpha() const noexcept
Get the alpha parameter.
Definition: NvInfer.h:1236
float getBeta() const noexcept
Get the beta parameter.
Definition: NvInfer.h:1245
void setAlpha(float alpha) noexcept
Set the alpha parameter (must be finite).
Definition: NvInfer.h:1213
virtual ~IActivationLayer() noexcept=0
An assertion layer in a network.
Definition: NvInfer.h:4941
void setMessage(char const *message) noexcept
Set the message to print if the assertion fails.
Definition: NvInfer.h:4951
char const * getMessage() const noexcept
Return the assertion message.
Definition: NvInfer.h:4961
virtual ~IAssertionLayer() noexcept=0
This is a base class for Attention boundary layers.
Definition: NvInfer.h:6826
IAttention * getAttention() const noexcept
Get a pointer to the IAttention associated with this boundary layer.
Definition: NvInfer.h:6831
virtual ~IAttentionBoundaryLayer() noexcept=0
Helper for constructing an attention that consumes query, key and value tensors.
Definition: NvInfer.h:6963
ITensor * getMask() noexcept
Get the optional mask in attention.
Definition: NvInfer.h:7013
bool setMetadata(char const *metadata) noexcept
Set the metadata for IAttention.
Definition: NvInfer.h:7263
TRT_NODISCARD bool setQueryLengths(ITensor *lengths) noexcept
Set the query lengths tensor.
Definition: NvInfer.h:7388
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:7089
bool setName(char const *name) noexcept
Set the name of the attention.
Definition: NvInfer.h:7179
bool getDecomposable() const noexcept
Get whether the attention can be decomposed to use multiple kernels if no fused kernel support found.
Definition: NvInfer.h:7102
ITensor * getInput(int32_t index) const noexcept
Get the IAttention input corresponding to the given index.
Definition: NvInfer.h:7142
CausalMaskKind getCausalKind() const noexcept
Get the causal mask alignment orientation for the attention.
Definition: NvInfer.h:7077
ITensor * getOutput(int32_t index) const noexcept
Get the IAttention output corresponding to the given index. IAttention has only one output.
Definition: NvInfer.h:7162
int32_t getNbOutputs() const noexcept
Get the number of outputs of a layer. IAttention has one output.
Definition: NvInfer.h:7150
bool setNbRanks(int32_t nbRanks) noexcept
Set the number of ranks for multi-device attention execution.
Definition: NvInfer.h:7292
int32_t getNbInputs() const noexcept
Get the number of inputs of IAttention. IAttention has three inputs.
Definition: NvInfer.h:7130
TRT_NODISCARD bool setKeyValueLengths(ITensor *lengths) noexcept
Set the key-value lengths tensor.
Definition: NvInfer.h:7427
TRT_NODISCARD ITensor * getKeyValueLengths() const noexcept
Get the key-value lengths tensor.
Definition: NvInfer.h:7439
bool setNormalizationOperation(AttentionNormalizationOp op) noexcept
Set the normalization operation for the attention.
Definition: NvInfer.h:6972
TRT_NODISCARD AttentionIOForm getKeyValueForm() const noexcept
Get the key-value form.
Definition: NvInfer.h:7364
char const * getName() const noexcept
Return the name of the attention.
Definition: NvInfer.h:7191
bool setNormalizationQuantizeToType(DataType type) noexcept
Set the datatype the attention normalization is quantized to.
Definition: NvInfer.h:7231
int32_t getNbRanks() const noexcept
Get the number of ranks for multi-device execution.
Definition: NvInfer.h:7304
TRT_DEPRECATED bool getCausal() const noexcept
Get whether the attention will run a causal inference.
Definition: NvInfer.h:7045
AttentionNormalizationOp getNormalizationOperation() const noexcept
Get the normalization operation for the attention.
Definition: NvInfer.h:6984
bool setNormalizationQuantizeScale(ITensor &tensor) noexcept
Set the quantization scale for the attention normalization output.
Definition: NvInfer.h:7207
bool setCausalKind(CausalMaskKind kind) noexcept
Set the causal mask alignment orientation for the attention.
Definition: NvInfer.h:7065
TRT_NODISCARD AttentionIOForm getQueryForm() const noexcept
Get the query form.
Definition: NvInfer.h:7334
char const * getMetadata() const noexcept
Get the metadata of IAttention.
Definition: NvInfer.h:7276
DataType getNormalizationQuantizeToType() const noexcept
Get the datatype the attention normalization is quantized to.
Definition: NvInfer.h:7243
TRT_NODISCARD bool setQueryForm(AttentionIOForm form) noexcept
Set the query form.
Definition: NvInfer.h:7321
virtual ~IAttention() noexcept=0
ITensor * getNormalizationQuantizeScale() const noexcept
Get the quantization scale for the attention normalization output.
Definition: NvInfer.h:7218
bool setInput(int32_t index, ITensor &input) noexcept
Append or replace an input of this layer with a specific tensor.
Definition: NvInfer.h:7121
TRT_NODISCARD ITensor * getQueryLengths() const noexcept
Get the query lengths tensor.
Definition: NvInfer.h:7400
bool setMask(ITensor &mask) noexcept
Set whether a mask will be used for the normalization operation.
Definition: NvInfer.h:7001
TRT_NODISCARD bool setKeyValueForm(AttentionIOForm form) noexcept
Set the key-value form.
Definition: NvInfer.h:7351
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:7031
apiv::VAttention * mImpl
Definition: NvInfer.h:7445
This layer represents an output of an IAttention.
Definition: NvInfer.h:6893
virtual ~IAttentionOutputLayer() noexcept=0
Holds properties for configuring a builder to produce an engine.
Definition: NvInfer.h:10623
void setProfileStream(cudaStream_t const stream) noexcept
Set the CUDA stream that is used to profile this network.
Definition: NvInfer.h:10892
void setMemoryPoolLimit(MemoryPoolType pool, std::size_t poolSize) noexcept
Set the memory size for the memory pool.
Definition: NvInfer.h:11081
nvinfer1::ITimingCache * createTimingCache(void const *blob, std::size_t size) const noexcept
Create timing cache.
Definition: NvInfer.h:11016
bool setMaxAuxStreams(int32_t nbStreams) noexcept
Set the maximum number of auxiliary streams that TRT is allowed to use.
Definition: NvInfer.h:11274
void setPreviewFeature(PreviewFeature feature, bool enable) noexcept
Enable or disable a specific preview feature.
Definition: NvInfer.h:11118
bool getPreviewFeature(PreviewFeature feature) const noexcept
Get status of preview feature.
Definition: NvInfer.h:11132
int32_t getBuilderOptimizationLevel() noexcept
Get builder optimization level.
Definition: NvInfer.h:11177
bool setTacticSources(TacticSources tacticSources) noexcept
Set tactic sources.
Definition: NvInfer.h:10981
void setPluginsToSerialize(char const *const *paths, int32_t nbPaths) noexcept
Set the plugin libraries to be serialized with version-compatible engines.
Definition: NvInfer.h:11220
bool setTilingOptimizationLevel(TilingOptimizationLevel level) noexcept
Set the Tiling optimization level.
Definition: NvInfer.h:11378
bool setL2LimitForTiling(int64_t size) noexcept
Set the L2 cache usage limit for Tiling optimization.
Definition: NvInfer.h:11406
std::size_t getMemoryPoolLimit(MemoryPoolType pool) const noexcept
Get the memory size limit of the memory pool.
Definition: NvInfer.h:11100
int32_t getDLACore() const noexcept
Get the DLA core that the engine executes on.
Definition: NvInfer.h:10822
int32_t getNbPluginsToSerialize() const noexcept
Get the number of plugin library paths to be serialized with version-compatible engines.
Definition: NvInfer.h:11243
void setDeviceType(ILayer const *layer, DeviceType deviceType) noexcept
Set the device that this layer must execute on.
Definition: NvInfer.h:10754
void setEngineCapability(EngineCapability capability) noexcept
Configure the builder to target specified EngineCapability flow.
Definition: NvInfer.h:10660
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:11284
bool getFlag(BuilderFlag builderFlag) const noexcept
Returns true if the build mode flag is set.
Definition: NvInfer.h:10737
void setMaxNbTactics(int32_t maxNbTactics) noexcept
Set the maximum number of tactics to time when there is a choice of tactics.
Definition: NvInfer.h:11350
int64_t getL2LimitForTiling() const noexcept
Get the L2 cache usage limit for tiling optimization.
Definition: NvInfer.h:11418
bool setRemoteAutoTuningConfig(char const *config) noexcept
Set a config string for remote auto tuning.
Definition: NvInfer.h:11432
void setProgressMonitor(IProgressMonitor *monitor) noexcept
Sets the progress monitor for building a network.
Definition: NvInfer.h:11300
void setProfilingVerbosity(ProfilingVerbosity verbosity) noexcept
Set verbosity level of layer information exposed in NVTX annotations and IEngineInspector.
Definition: NvInfer.h:10946
int32_t getNbOptimizationProfiles() const noexcept
Get number of optimization profiles.
Definition: NvInfer.h:10934
nvinfer1::ITimingCache const * getTimingCache() const noexcept
Get the pointer to the timing cache from current IBuilderConfig.
Definition: NvInfer.h:11049
void reset() noexcept
Resets the builder configuration to defaults.
Definition: NvInfer.h:10853
bool setTimingCache(ITimingCache const &cache, bool ignoreMismatch) noexcept
Attach a timing cache to IBuilderConfig.
Definition: NvInfer.h:11039
char const * getPluginToSerialize(int32_t index) const noexcept
Get the plugin library path to be serialized with version-compatible engines.
Definition: NvInfer.h:11233
EngineCapability getEngineCapability() const noexcept
Query EngineCapability flow configured for the builder.
Definition: NvInfer.h:10672
RuntimePlatform getRuntimePlatform() const noexcept
Get the target platform for runtime execution.
Definition: NvInfer.h:11338
DeviceType getDefaultDeviceType() const noexcept
Get the default DeviceType which was set by setDefaultDeviceType.
Definition: NvInfer.h:10843
void setRuntimePlatform(RuntimePlatform runtimePlatform) noexcept
Set the target platform for runtime execution.
Definition: NvInfer.h:11326
int32_t getMaxNbTactics() const noexcept
Query the maximum number of tactics timed when there is a choice.
Definition: NvInfer.h:11362
BuilderFlags getFlags() const noexcept
Get the build mode flags for this builder config. Defaults to 0.
Definition: NvInfer.h:10701
void setFlags(BuilderFlags builderFlags) noexcept
Set the build mode flags to turn on builder options for this network.
Definition: NvInfer.h:10689
TacticSources getTacticSources() const noexcept
Get tactic sources.
Definition: NvInfer.h:10996
char const * getAllBuildRoutes() const noexcept
Get all available build routes.
Definition: NvInfer.h:11515
void resetDeviceType(ILayer const *layer) noexcept
reset the DeviceType for this layer
Definition: NvInfer.h:10786
void setDLACore(int32_t dlaCore) noexcept
Sets the DLA core used by the network. Defaults to -1.
Definition: NvInfer.h:10812
HardwareCompatibilityLevel getHardwareCompatibilityLevel() const noexcept
Get the hardware compatibility level.
Definition: NvInfer.h:11207
char const * getRemoteAutoTuningConfig() const noexcept
Get a config string for remote auto tuning.
Definition: NvInfer.h:11442
bool setBuildRoute(char const *buildRoute) noexcept
Set the build route to be passed to the compiler.
Definition: NvInfer.h:11465
void clearFlag(BuilderFlag builderFlag) noexcept
clear a single build mode flag.
Definition: NvInfer.h:10713
int32_t addOptimizationProfile(IOptimizationProfile const *profile) noexcept
Add an optimization profile.
Definition: NvInfer.h:10921
IProgressMonitor * getProgressMonitor() const noexcept
Definition: NvInfer.h:11310
apiv::VBuilderConfig * mImpl
Definition: NvInfer.h:11521
int32_t getAvgTimingIterations() const noexcept
Query the number of averaging iterations.
Definition: NvInfer.h:10647
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:10833
void setFlag(BuilderFlag builderFlag) noexcept
Set a single build mode flag.
Definition: NvInfer.h:10725
DeviceType getDeviceType(ILayer const *layer) const noexcept
Get the device that this layer executes on.
Definition: NvInfer.h:10764
bool canRunOnDLA(ILayer const *layer) const noexcept
Checks if a layer can run on DLA.
Definition: NvInfer.h:10796
cudaStream_t getProfileStream() const noexcept
Get the default engine-level CUDA stream used to profile this network.
Definition: NvInfer.h:10904
void setHardwareCompatibilityLevel(HardwareCompatibilityLevel hardwareCompatibilityLevel) noexcept
Set the hardware compatibility level.
Definition: NvInfer.h:11194
TilingOptimizationLevel getTilingOptimizationLevel() const noexcept
Get the Tiling optimization level.
Definition: NvInfer.h:11390
ProfilingVerbosity getProfilingVerbosity() const noexcept
Get verbosity level of layer information exposed in NVTX annotations and IEngineInspector.
Definition: NvInfer.h:10959
char const * getBuildRoute() const noexcept
Get the build route string.
Definition: NvInfer.h:11479
bool isDeviceTypeSet(ILayer const *layer) const noexcept
whether the DeviceType has been explicitly set for this layer
Definition: NvInfer.h:10776
void setBuilderOptimizationLevel(int32_t level) noexcept
Set builder optimization level.
Definition: NvInfer.h:11165
Builds an engine from a network definition.
Definition: NvInfer.h:11584
int32_t getNbDLACores() const noexcept
Return the number of DLA engines available to this builder.
Definition: NvInfer.h:11603
virtual ~IBuilder() noexcept=0
IErrorRecorder * getErrorRecorder() const noexcept
get the ErrorRecorder assigned to this interface.
Definition: NvInfer.h:11711
apiv::VBuilder * mImpl
Definition: NvInfer.h:11891
ILogger * getLogger() const noexcept
get the logger with which the builder was created
Definition: NvInfer.h:11845
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:11835
int32_t getMaxThreads() const noexcept
get the maximum number of threads that can be used by the builder.
Definition: NvInfer.h:11875
IPluginRegistry & getPluginRegistry() noexcept
get the local plugin registry that can be used by the builder.
Definition: NvInfer.h:11885
nvinfer1::IOptimizationProfile * createOptimizationProfile() noexcept
Create a new optimization profile.
Definition: NvInfer.h:11677
void setGpuAllocator(IGpuAllocator *allocator) noexcept
Set the GPU allocator.
Definition: NvInfer.h:11621
nvinfer1::INetworkDefinition * createNetworkV2(NetworkDefinitionCreationFlags flags) noexcept
Create a network definition object.
Definition: NvInfer.h:11662
nvinfer1::IBuilderConfig * createBuilderConfig() noexcept
Create a builder configuration object.
Definition: NvInfer.h:11635
void reset() noexcept
Resets the builder state to default values.
Definition: NvInfer.h:11719
bool setMaxThreads(int32_t maxThreads) noexcept
Set the maximum number of threads.
Definition: NvInfer.h:11861
void setErrorRecorder(IErrorRecorder *recorder) noexcept
Set the ErrorRecorder for this interface.
Definition: NvInfer.h:11696
nvinfer1::IHostMemory * buildSerializedNetwork(INetworkDefinition &network, IBuilderConfig &config) noexcept
Builds and serializes a network for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:11739
bool buildSerializedNetworkToStream(INetworkDefinition &network, IBuilderConfig &config, IStreamWriter &writer) noexcept
Builds network using the config configuration, and serializes an engine into writer.
Definition: NvInfer.h:11761
nvinfer1::ICudaEngine * buildEngineWithConfig(INetworkDefinition &network, IBuilderConfig &config) noexcept
Builds a network for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:11809
nvinfer1::IHostMemory * buildSerializedNetwork(INetworkDefinition &network, IBuilderConfig &config, IHostMemory *&kernelText) noexcept
Extended form of buildSerializedNetwork that optionally permits getting the kernel text.
Definition: NvInfer.h:11786
A cast layer in a network.
Definition: NvInfer.h:3782
apiv::VCastLayer * mImpl
Definition: NvInfer.h:3808
DataType getToType() const noexcept
Return cast layer output type.
Definition: NvInfer.h:3802
void setToType(DataType toType) noexcept
Set cast layer output type.
Definition: NvInfer.h:3791
virtual ~ICastLayer() noexcept=0
A concatenation layer in a network definition.
Definition: NvInfer.h:1895
void setAxis(int32_t axis) noexcept
Set the axis along which concatenation occurs.
Definition: NvInfer.h:1908
int32_t getAxis() const noexcept
Get the axis along which concatenation occurs.
Definition: NvInfer.h:1918
virtual ~IConcatenationLayer() noexcept=0
This layer represents a condition input to an IIfConditional.
Definition: NvInfer.h:4445
virtual ~IConditionLayer() noexcept=0
Layer that represents a constant value.
Definition: NvInfer.h:3823
void setWeights(Weights weights) noexcept
Set the weights for the layer.
Definition: NvInfer.h:3833
Weights getWeights() const noexcept
Get the weights for the layer.
Definition: NvInfer.h:3843
virtual ~IConstantLayer() noexcept=0
void setDimensions(Dims const &dimensions) noexcept
Set the dimensions for the layer.
Definition: NvInfer.h:3855
apiv::VConstantLayer * mImpl
Definition: NvInfer.h:3873
Dims getDimensions() const noexcept
Get the dimensions for the layer.
Definition: NvInfer.h:3867
A convolution layer in a network definition.
Definition: NvInfer.h:857
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:982
Weights getBiasWeights() const noexcept
Get the bias weights for the convolution.
Definition: NvInfer.h:955
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:1023
void setDilationNd(Dims const &dilation) noexcept
Set the multi-dimension dilation of the convolution.
Definition: NvInfer.h:1127
virtual ~IConvolutionLayer() noexcept=0
Dims getPaddingNd() const noexcept
Get the multi-dimension padding of the convolution.
Definition: NvInfer.h:1113
Dims getStrideNd() const noexcept
Get the multi-dimension stride of the convolution.
Definition: NvInfer.h:1083
Weights getKernelWeights() const noexcept
Get the kernel weights of the convolution.
Definition: NvInfer.h:930
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride of the convolution.
Definition: NvInfer.h:1073
Dims getDilationNd() const noexcept
Get the multi-dimension dilation of the convolution.
Definition: NvInfer.h:1137
int64_t getNbOutputMaps() const noexcept
Get the number of output maps for the convolution.
Definition: NvInfer.h:876
void setKernelWeights(Weights weights) noexcept
Set the kernel weights for the convolution.
Definition: NvInfer.h:920
Dims getPostPadding() const noexcept
Get the post-padding.
Definition: NvInfer.h:1009
int64_t getNbGroups() const noexcept
Get the number of groups of the convolution.
Definition: NvInfer.h:906
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:1035
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups for a convolution.
Definition: NvInfer.h:896
void setNbOutputMaps(int64_t nbOutputMaps) noexcept
Set the number of output maps for the convolution.
Definition: NvInfer.h:866
void setBiasWeights(Weights weights) noexcept
Set the bias weights for the convolution.
Definition: NvInfer.h:945
Dims getKernelSizeNd() const noexcept
Get the multi-dimension kernel size of the convolution.
Definition: NvInfer.h:1058
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding of the convolution.
Definition: NvInfer.h:1101
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding of the convolution.
Definition: NvInfer.h:972
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding of the convolution.
Definition: NvInfer.h:999
void setKernelSizeNd(Dims const &kernelSize) noexcept
Set the multi-dimension kernel size of the convolution.
Definition: NvInfer.h:1048
An engine for executing inference on a built network, with functionally unsafe features.
Definition: NvInferRuntime.h:3144
Layer that represents a cumulative operation across a tensor.
Definition: NvInfer.h:6647
bool setOperation(CumulativeOperation op) noexcept
Set the cumulative operation for the layer.
Definition: NvInfer.h:6658
void setReverse(bool reverse) noexcept
Specify whether the cumulative operation should be applied backward.
Definition: NvInfer.h:6706
apiv::VCumulativeLayer * mImpl
Definition: NvInfer.h:6724
virtual ~ICumulativeLayer() noexcept=0
bool getExclusive() const noexcept
Get whether it is exclusive accumulation or inclusive accumulation.
Definition: NvInfer.h:6694
bool getReverse() const noexcept
Get the boolean that specifies whether the cumulative operation should be applied backward.
Definition: NvInfer.h:6718
void setExclusive(bool exclusive) noexcept
Set whether it is an exclusive accumulation or inclusive accumulation.
Definition: NvInfer.h:6682
CumulativeOperation getOperation() const noexcept
Get the cumulative operation for the layer.
Definition: NvInfer.h:6670
A deconvolution layer in a network definition.
Definition: NvInfer.h:1938
void setBiasWeights(Weights weights) noexcept
Set the bias weights for the deconvolution.
Definition: NvInfer.h:2026
int64_t getNbGroups() const noexcept
Get the number of groups for a deconvolution.
Definition: NvInfer.h:1987
Weights getKernelWeights() const noexcept
Get the kernel weights for the deconvolution.
Definition: NvInfer.h:2011
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding of the deconvolution.
Definition: NvInfer.h:2053
Dims getStrideNd() const noexcept
Get the multi-dimension stride of the deconvolution.
Definition: NvInfer.h:2168
Dims getDilationNd() const noexcept
Get the multi-dimension dilation of the deconvolution.
Definition: NvInfer.h:2234
virtual ~IDeconvolutionLayer() noexcept=0
Weights getBiasWeights() const noexcept
Get the bias weights for the deconvolution.
Definition: NvInfer.h:2036
void setKernelWeights(Weights weights) noexcept
Set the kernel weights for the deconvolution.
Definition: NvInfer.h:2001
int64_t getNbOutputMaps() const noexcept
Get the number of output feature maps for the deconvolution.
Definition: NvInfer.h:1957
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride of the deconvolution.
Definition: NvInfer.h:2158
Dims getPostPadding() const noexcept
Get the padding.
Definition: NvInfer.h:2090
Dims getKernelSizeNd() const noexcept
Get the multi-dimension kernel size of the deconvolution.
Definition: NvInfer.h:2141
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding of the deconvolution.
Definition: NvInfer.h:2080
void setKernelSizeNd(Dims const &kernelSize) noexcept
Set the multi-dimension kernel size of the deconvolution.
Definition: NvInfer.h:2131
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding of the deconvolution.
Definition: NvInfer.h:2186
void setNbOutputMaps(int64_t nbOutputMaps) noexcept
Set the number of output feature maps for the deconvolution.
Definition: NvInfer.h:1947
Dims getPaddingNd() const noexcept
Get the multi-dimension padding of the deconvolution.
Definition: NvInfer.h:2198
void setDilationNd(Dims const &dilation) noexcept
Set the multi-dimension dilation of the deconvolution.
Definition: NvInfer.h:2224
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:2104
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups for a deconvolution.
Definition: NvInfer.h:1977
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:2063
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:2116
A Dequantize layer in a network definition.
Definition: NvInfer.h:5560
TRT_NODISCARD Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5609
void setToType(DataType toType) noexcept
Set the Dequantize layer output type.
Definition: NvInfer.h:5625
int32_t getAxis() const noexcept
Get the quantization axis.
Definition: NvInfer.h:5570
bool setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5598
virtual ~IDequantizeLayer() noexcept=0
DataType getToType() const noexcept
Return the Dequantize layer output type.
Definition: NvInfer.h:5637
void setAxis(int32_t axis) noexcept
Set the quantization axis.
Definition: NvInfer.h:5581
Definition: NvInfer.h:8143
virtual ~IDistCollectiveLayer() noexcept=0
A network layer to perform dynamic quantization.
Definition: NvInfer.h:5667
virtual ~IDynamicQuantizeLayer() noexcept=0
DataType getScaleType() const noexcept
Return the scale factors data type.
Definition: NvInfer.h:5733
TRT_DEPRECATED void setAxis(int32_t axis) noexcept
Set the axis along which block quantization occurs.
Definition: NvInfer.h:5746
TRT_DEPRECATED void setBlockSize(int32_t size) noexcept
Set the size of the quantization block.
Definition: NvInfer.h:5769
Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5804
void setScaleType(DataType scaleType) noexcept
Set the data type of the scale factors used to quantize the data.
Definition: NvInfer.h:5720
DataType getToType() const noexcept
Return DynamicQuantizeLayer's quantized output type.
Definition: NvInfer.h:5707
TRT_DEPRECATED int32_t getAxis() const noexcept
Get the axis along which blocking occurs.
Definition: NvInfer.h:5756
void setToType(DataType toType) noexcept
Set DynamicQuantizeLayer's quantized output type.
Definition: NvInfer.h:5694
void setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5792
TRT_DEPRECATED int32_t getBlockSize() const noexcept
Get the size of the quantization block.
Definition: NvInfer.h:5779
An Einsum layer in a network.
Definition: NvInfer.h:5851
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:5862
virtual ~IEinsumLayer() noexcept=0
char const * getEquation() const noexcept
Return the equation.
Definition: NvInfer.h:5872
A elementwise layer in a network definition.
Definition: NvInfer.h:2307
apiv::VElementWiseLayer * mImpl
Definition: NvInfer.h:2336
ElementWiseOperation getOperation() const noexcept
Get the binary operation for the layer.
Definition: NvInfer.h:2330
void setOperation(ElementWiseOperation op) noexcept
Set the binary operation for the layer.
Definition: NvInfer.h:2318
virtual ~IElementWiseLayer() noexcept=0
Generate a tensor according to a specified mode.
Definition: NvInfer.h:5057
bool isAlphaBetaInt64() const noexcept
Return true if alpha/beta have type int64, false if they have type double.
Definition: NvInfer.h:5289
FillOperation getOperation() const noexcept
Get the fill operation for the layer.
Definition: NvInfer.h:5103
void setOperation(FillOperation op) noexcept
Set the fill operation for the layer.
Definition: NvInfer.h:5093
DataType getToType() const noexcept
Get the fill layer output type.
Definition: NvInfer.h:5319
void setAlphaInt64(int64_t alpha) noexcept
Set the alpha parameter with int64 datatype.
Definition: NvInfer.h:5232
void setBetaInt64(int64_t beta) noexcept
Set the beta parameter with int64 datatype.
Definition: NvInfer.h:5266
virtual ~IFillLayer() noexcept=0
void setBeta(double beta) noexcept
Set the beta parameter.
Definition: NvInfer.h:5156
int64_t getAlphaInt64() const noexcept
Get the value of alpha parameter with int64 datatype.
Definition: NvInfer.h:5247
int64_t getBetaInt64() const noexcept
Get the value of beta parameter with int64 datatype.
Definition: NvInfer.h:5281
double getAlpha() const noexcept
Get the value of alpha parameter.
Definition: NvInfer.h:5137
void setDimensions(Dims const &dimensions) noexcept
Set the output tensor's dimensions.
Definition: NvInfer.h:5068
void setAlpha(double alpha) noexcept
Set the alpha parameter.
Definition: NvInfer.h:5122
void setToType(DataType toType) noexcept
Set the fill layer output type.
Definition: NvInfer.h:5307
Dims getDimensions() const noexcept
Get the output tensor's dimensions.
Definition: NvInfer.h:5083
double getBeta() const noexcept
Get the value of beta parameter.
Definition: NvInfer.h:5171
A Gather layer in a network definition. Supports several kinds of gathering.
Definition: NvInfer.h:2442
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:2453
void setNbElementWiseDims(int32_t elementWiseDims) noexcept
Set the number of leading dimensions of indices tensor to be handled elementwise.
Definition: NvInfer.h:2488
apiv::VGatherLayer * mImpl
Definition: NvInfer.h:2524
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:2498
void setMode(GatherMode mode) noexcept
Set the gather mode.
Definition: NvInfer.h:2508
int32_t getGatherAxis() const noexcept
Get the axis to gather on.
Definition: NvInfer.h:2465
GatherMode getMode() const noexcept
Get the gather mode.
Definition: NvInfer.h:2518
A GridSample layer in a network definition.
Definition: NvInfer.h:6080
void setInterpolationMode(InterpolationMode mode) noexcept
Set the grid sample interpolation mode.
Definition: NvInfer.h:6087
bool setSampleMode(SampleMode mode) noexcept
Set the sample mode.
Definition: NvInfer.h:6133
void setAlignCorners(bool alignCorners) noexcept
Set the align corners mode.
Definition: NvInfer.h:6109
apiv::VGridSampleLayer * mImpl
Definition: NvInfer.h:6151
SampleMode getSampleMode() const noexcept
Get the sample mode.
Definition: NvInfer.h:6145
virtual ~IGridSampleLayer() noexcept=0
InterpolationMode getInterpolationMode() const noexcept
Get the grid sample interpolation mode.
Definition: NvInfer.h:6099
bool getAlignCorners() const noexcept
Get the align corners mode.
Definition: NvInfer.h:6121
Class to handle library allocated memory that is accessible to the user.
Definition: NvInferRuntime.h:150
A layer that represents the identity function.
Definition: NvInfer.h:3767
virtual ~IIdentityLayer() noexcept=0
apiv::VIdentityLayer * mImpl
Definition: NvInfer.h:3769
This is a base class for Conditional boundary layers.
Definition: NvInfer.h:4422
IIfConditional * getConditional() const noexcept
Get a pointer to the IIfConditional associated with this boundary layer.
Definition: NvInfer.h:4427
virtual ~IIfConditionalBoundaryLayer() noexcept=0
Helper for constructing conditionally-executed subgraphs.
Definition: NvInfer.h:4513
IIfConditionalInputLayer * addInput(ITensor &input) noexcept
Add an If-conditional input.
Definition: NvInfer.h:4554
char const * getName() const noexcept
Return the name of the conditional.
Definition: NvInfer.h:4579
IConditionLayer * setCondition(ITensor &condition) noexcept
Set the condition tensor for this If-Conditional construct.
Definition: NvInfer.h:4524
virtual ~IIfConditional() noexcept=0
IIfConditionalOutputLayer * addOutput(ITensor &trueSubgraphOutput, ITensor &falseSubgraphOutput) noexcept
Add an If-conditional output.
Definition: NvInfer.h:4542
void setName(char const *name) noexcept
Set the name of the conditional.
Definition: NvInfer.h:4569
This layer represents an output of an IIfConditional.
Definition: NvInfer.h:4464
virtual ~IIfConditionalOutputLayer() noexcept=0
A layer to do iterations.
Definition: NvInfer.h:4752
void setReverse(bool reverse) noexcept
Set iteration order to be reverse.
Definition: NvInfer.h:4779
virtual ~IIteratorLayer() noexcept=0
bool getReverse() const noexcept
Check if the iteration order is reverse.
Definition: NvInfer.h:4789
int32_t getAxis() const noexcept
Get axis being iterated over.
Definition: NvInfer.h:4765
void setAxis(int32_t axis) noexcept
Set axis to iterate over.
Definition: NvInfer.h:4757
Layer that represents a KVCacheUpdate operation.
Definition: NvInfer.h:7567
bool setCacheMode(KVCacheMode cacheMode) noexcept
Set the mode of the KVCacheUpdate layer.
Definition: NvInfer.h:7592
TRT_NODISCARD ITensor * getUpdateLengths() const noexcept
Get the update lengths tensor.
Definition: NvInfer.h:7667
virtual ~IKVCacheUpdateLayer() noexcept=0
TRT_NODISCARD AttentionIOForm getUpdateForm() const noexcept
Get the update form.
Definition: NvInfer.h:7633
TRT_NODISCARD bool setUpdateLengths(ITensor *lengths) noexcept
Set the update lengths tensor.
Definition: NvInfer.h:7655
TRT_NODISCARD bool setUpdateForm(AttentionIOForm form) noexcept
Set the update form.
Definition: NvInfer.h:7620
KVCacheMode getCacheMode() const noexcept
Get the mode of the KVCacheUpdate layer.
Definition: NvInfer.h:7602
apiv::VKVCacheUpdateLayer * mImpl
Definition: NvInfer.h:7673
A LRN layer in a network definition.
Definition: NvInfer.h:1544
int64_t getWindowSize() const noexcept
Get the LRN window size.
Definition: NvInfer.h:1565
virtual ~ILRNLayer() noexcept=0
float getAlpha() const noexcept
Get the LRN alpha value.
Definition: NvInfer.h:1587
void setWindowSize(int64_t windowSize) noexcept
Set the LRN window size.
Definition: NvInfer.h:1555
void setK(float k) noexcept
Set the LRN K value.
Definition: NvInfer.h:1621
void setAlpha(float alpha) noexcept
Set the LRN alpha value.
Definition: NvInfer.h:1577
void setBeta(float beta) noexcept
Set the LRN beta value.
Definition: NvInfer.h:1599
float getBeta() const noexcept
Get the LRN beta value.
Definition: NvInfer.h:1609
float getK() const noexcept
Get the LRN K value.
Definition: NvInfer.h:1631
Base class for all layer classes in a network definition.
Definition: NvInfer.h:490
virtual ~ILayer() noexcept=0
void setMetadata(char const *metadata) noexcept
Set the metadata for this layer.
Definition: NvInfer.h:611
void setName(char const *name) noexcept
Set the name of a layer.
Definition: NvInfer.h:511
int32_t getNbInputs() const noexcept
Get the number of inputs of a layer.
Definition: NvInfer.h:529
int32_t getNbRanks() const noexcept
Get the number of ranks for multi-device execution.
Definition: NvInfer.h:657
char const * getMetadata() const noexcept
Get the metadata of the layer.
Definition: NvInfer.h:624
DataType getOutputType(int32_t index) const noexcept
get the output type of this layer
Definition: NvInfer.h:592
char const * getName() const noexcept
Return the name of a layer.
Definition: NvInfer.h:521
int32_t getNbOutputs() const noexcept
Get the number of outputs of a layer.
Definition: NvInfer.h:550
ITensor * getOutput(int32_t index) const noexcept
Get the layer output corresponding to the given index.
Definition: NvInfer.h:560
void setInput(int32_t index, ITensor &tensor) noexcept
Replace an input of this layer with a specific tensor.
Definition: NvInfer.h:577
ITensor * getInput(int32_t index) const noexcept
Get the layer input corresponding to the given index.
Definition: NvInfer.h:542
bool setNbRanks(int32_t nbRanks) noexcept
Set the number of ranks for multi-device execution.
Definition: NvInfer.h:645
LayerType getType() const noexcept
Return the type of a layer.
Definition: NvInfer.h:497
This is a base class for Loop boundary layers.
Definition: NvInfer.h:4397
virtual ~ILoopBoundaryLayer() noexcept=0
ILoop * getLoop() const noexcept
Get a pointer to ILoop associated with this boundary layer.
Definition: NvInfer.h:4402
Helper for creating a recurrent subgraph.
Definition: NvInfer.h:4812
void setName(char const *name) noexcept
Set the name of the loop.
Definition: NvInfer.h:4882
ITripLimitLayer * addTripLimit(ITensor &tensor, TripLimit limit) noexcept
Add a trip-count limiter, based on the given tensor.
Definition: NvInfer.h:4841
IIteratorLayer * addIterator(ITensor &tensor, int32_t axis=0, bool reverse=false) noexcept
Return layer that subscripts tensor by loop iteration.
Definition: NvInfer.h:4854
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:4867
virtual ~ILoop() noexcept=0
char const * getName() const noexcept
Return the name of the loop.
Definition: NvInfer.h:4892
IRecurrenceLayer * addRecurrence(ITensor &initialValue) noexcept
Create a recurrence layer for this loop with initialValue as its first input.
Definition: NvInfer.h:4820
An ILoopOutputLayer is the sole way to get output from a loop.
Definition: NvInfer.h:4648
int32_t getAxis() const noexcept
Get axis being concatenated over.
Definition: NvInfer.h:4678
LoopOutput getLoopOutput() const noexcept
Get which kind a loop output has.
Definition: NvInfer.h:4653
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:4670
Layer that represents a Matrix Multiplication.
Definition: NvInfer.h:3608
apiv::VMatrixMultiplyLayer * mImpl
Definition: NvInfer.h:3636
virtual ~IMatrixMultiplyLayer() noexcept=0
MatrixOperation getOperation(int32_t index) const noexcept
Get the operation for an input tensor.
Definition: NvInfer.h:3630
void setOperation(int32_t index, MatrixOperation op) noexcept
Set the operation for an input tensor.
Definition: NvInfer.h:3618
A MoE layer in a network definition. Mixture of Experts (MoE) is a collection of experts with each ex...
Definition: NvInfer.h:7818
void setSwigluParamLimit(float limit) noexcept
Set the SwiGLU parameter limit.
Definition: NvInfer.h:8040
void setDynQOutputScaleType(DataType type) noexcept
Set the dynamic quantization output scale type.
Definition: NvInfer.h:7993
MoEActType getActivationType() const noexcept
Get the activation type for the MoE layer.
Definition: NvInfer.h:7867
void setQuantizationToType(DataType type) noexcept
Set the data type the mul output is quantized to.
Definition: NvInfer.h:7941
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:7926
void setQuantizationStatic(ITensor &fcDownActivationScale, DataType dataType) noexcept
Configure static quantization after the mul op.
Definition: NvInfer.h:7893
float getSwigluParamLimit() const noexcept
Get the SwiGLU parameter limit.
Definition: NvInfer.h:8052
DataType getQuantizationToType() const noexcept
Get the data type the mul in MoE layer is quantized to.
Definition: NvInfer.h:7953
DataType getDynQOutputScaleType() const noexcept
Get the dynamic quantization output scale type.
Definition: NvInfer.h:8005
virtual ~IMoELayer() noexcept=0
void setActivationType(MoEActType activationType) noexcept
Set the activation type for the MoE layer.
Definition: NvInfer.h:7855
Dims getQuantizationBlockShape() const noexcept
Get the block shape for the quantization of the Mul output.
Definition: NvInfer.h:7981
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:7831
float getSwigluParamBeta() const noexcept
Get the SwiGLU parameter beta.
Definition: NvInfer.h:8104
void setSwigluParamBeta(float beta) noexcept
Set the SwiGLU parameter beta.
Definition: NvInfer.h:8092
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:7843
void setSwigluParams(float limit, float alpha, float beta) noexcept
Set the SwiGLU parameters.
Definition: NvInfer.h:8026
void setQuantizationBlockShape(Dims const &blockShape) noexcept
Set the block shape for the quantization of the Mul output.
Definition: NvInfer.h:7969
void setInput(int32_t index, ITensor &tensor) noexcept
Set the input of the MoE layer.
Definition: NvInfer.h:8121
float getSwigluParamAlpha() const noexcept
Get the SwiGLU parameter alpha.
Definition: NvInfer.h:8078
void setSwigluParamAlpha(float alpha) noexcept
Set the SwiGLU parameter alpha.
Definition: NvInfer.h:8066
A non-maximum suppression layer in a network definition.
Definition: NvInfer.h:6234
void setTopKBoxLimit(int32_t limit) noexcept
Set the TopK box limit parameter for the layer.
Definition: NvInfer.h:6271
void setBoundingBoxFormat(BoundingBoxFormat fmt) noexcept
Set the bounding box format parameter for the layer.
Definition: NvInfer.h:6245
BoundingBoxFormat getBoundingBoxFormat() const noexcept
Get the bounding box format parameter for the layer.
Definition: NvInfer.h:6257
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:6316
apiv::VNMSLayer * mImpl
Definition: NvInfer.h:6334
int32_t getTopKBoxLimit() const noexcept
Get the TopK box limit parameter for the layer.
Definition: NvInfer.h:6281
DataType getIndicesType() const noexcept
Return the NMS layer indices type.
Definition: NvInfer.h:6328
virtual ~INMSLayer() noexcept=0
A network definition for input to the builder.
Definition: NvInfer.h:8169
IConcatenationLayer * addConcatenation(ITensor *const *inputs, int32_t nbInputs) noexcept
Add a concatenation layer to the network.
Definition: NvInfer.h:8397
IShuffleLayer * addShuffle(ITensor &input) noexcept
Add a shuffle layer to the network.
Definition: NvInfer.h:8460
void setName(char const *name) noexcept
Sets the name of the network.
Definition: NvInfer.h:8926
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:8656
bool markDebug(ITensor &tensor) noexcept
Mark a tensor as a debug tensor.
Definition: NvInfer.h:8240
ILRNLayer * addLRN(ITensor &input, int64_t window, float alpha, float beta, float k) noexcept
Add a LRN layer to the network.
Definition: NvInfer.h:8341
ICumulativeLayer * addCumulative(ITensor &input, ITensor &axis, CumulativeOperation operation, bool exclusive, bool reverse) noexcept
Add a cumulative layer to the network.
Definition: NvInfer.h:9607
IAssertionLayer * addAssertion(ITensor &condition, char const *message) noexcept
Add an assertion layer to the network.
Definition: NvInfer.h:9242
TRT_DEPRECATED INonZeroLayer * addNonZero(ITensor &input) noexcept
Add a nonzero layer to the network.
Definition: NvInfer.h:8747
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:9061
ICastLayer * addCast(ITensor &input, DataType toType) noexcept
Add a cast layer.
Definition: NvInfer.h:8816
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:9140
char const * getName() const noexcept
Returns the name associated with the network.
Definition: NvInfer.h:8940
IParametricReLULayer * addParametricReLU(ITensor &input, ITensor &slope) noexcept
Add a parametric ReLU layer to the network.
Definition: NvInfer.h:9039
ITensor * getOutput(int32_t index) const noexcept
Get the output tensor specified by the given index.
Definition: NvInfer.h:8561
ITensor * getInput(int32_t index) const noexcept
Get the input tensor specified by the given index.
Definition: NvInfer.h:8531
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:8623
IDequantizeLayer * addDequantize(ITensor &input, ITensor &scale, DataType outputType) noexcept
Add a dequantization layer to the network.
Definition: NvInfer.h:9365
bool unmarkOutputForShapes(ITensor &tensor) noexcept
Undo markOutputForShapes.
Definition: NvInfer.h:9021
IFillLayer * addFill(Dims const &dimensions, FillOperation op, DataType outputType) noexcept
Add a fill layer to the network.
Definition: NvInfer.h:9268
ILoop * addLoop() noexcept
Add a loop to the network.
Definition: NvInfer.h:9171
bool markUnfusedTensorsAsDebugTensors() noexcept
Mark unfused tensors as debug tensors.
Definition: NvInfer.h:8288
TRT_NODISCARD INormalizationLayer * addNormalizationV2(ITensor &input, ITensor &scale, ITensor &bias, uint32_t axesMask) noexcept
Add a normalization layer to the network.
Definition: NvInfer.h:9929
IActivationLayer * addActivation(ITensor &input, ActivationType type) noexcept
Add an activation layer to the network.
Definition: NvInfer.h:8322
ISliceLayer * addSlice(ITensor &input, Dims const &start, Dims const &size, Dims const &stride) noexcept
Add a slice layer to the network.
Definition: NvInfer.h:8902
virtual IBuilder & getBuilder() const noexcept
Return the builder from which this INetworkDefinition was created.
Definition: NvInfer.h:9793
ILayer * getLayer(int32_t index) const noexcept
Get the layer specified by the given index.
Definition: NvInfer.h:8503
bool isDebugTensor(ITensor const &tensor) const noexcept
Check if a tensor is marked as debug tensor.
Definition: NvInfer.h:8266
bool getFlag(NetworkDefinitionCreationFlag networkDefinitionCreationFlag) const noexcept
Returns true if the network definition creation flag is set.
Definition: NvInfer.h:8992
IIfConditional * addIfConditional() noexcept
Add an if-then-else to the network.
Definition: NvInfer.h:9186
IErrorRecorder * getErrorRecorder() const noexcept
get the ErrorRecorder assigned to this interface.
Definition: NvInfer.h:9342
ISqueezeLayer * addSqueeze(ITensor &input, ITensor &axes) noexcept
Add a squeeze layer to the network.
Definition: NvInfer.h:9853
TRT_DEPRECATED INMSLayer * addNMS(ITensor &boxes, ITensor &scores, ITensor &maxOutputBoxesPerClass) noexcept
Add a non-maximum suppression layer to the network.
Definition: NvInfer.h:9516
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:9668
IReverseSequenceLayer * addReverseSequence(ITensor &input, ITensor &sequenceLens) noexcept
Add a ReverseSequence layer to the network.
Definition: NvInfer.h:9553
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:9436
int32_t getNbInputs() const noexcept
Get the number of inputs in the network.
Definition: NvInfer.h:8515
NetworkDefinitionCreationFlags getFlags() const noexcept
Get the network definition creation flags for this network definition object. Defaults to 0.
Definition: NvInfer.h:8980
IQuantizeLayer * addQuantize(ITensor &input, ITensor &scale, DataType outputType) noexcept
Add a quantization layer to the network.
Definition: NvInfer.h:9409
IDynamicQuantizeLayer * addDynamicQuantizeV2(ITensor &input, Dims const &blockShape, DataType outputType, DataType scaleType) noexcept
Add a dynamic quantization layer to the network.
Definition: NvInfer.h:9460
IReduceLayer * addReduce(ITensor &input, ReduceOperation operation, uint32_t reduceAxes, bool keepDimensions) noexcept
Add a reduce layer to the network.
Definition: NvInfer.h:8587
IUnaryLayer * addUnary(ITensor &input, UnaryOperation operation) noexcept
Add a unary layer to the network.
Definition: NvInfer.h:8446
IGridSampleLayer * addGridSample(ITensor &input, ITensor &grid) noexcept
Add a GridSample layer to the network.
Definition: NvInfer.h:9494
void removeTensor(ITensor &tensor) noexcept
remove a tensor from the network definition.
Definition: NvInfer.h:8831
bool areWeightsMarkedRefittable(char const *name) const noexcept
Whether the weight has been marked as refittable.
Definition: NvInfer.h:9834
ISelectLayer * addSelect(ITensor &condition, ITensor &thenInput, ITensor &elseInput) noexcept
Add a select layer to the network.
Definition: NvInfer.h:9225
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:9385
TRT_DEPRECATED INormalizationLayer * addNormalization(ITensor &input, ITensor &scale, ITensor &bias, uint32_t axesMask) noexcept
Add a normalization layer to the network.
Definition: NvInfer.h:9585
int32_t getNbLayers() const noexcept
Get the number of layers in the network.
Definition: NvInfer.h:8489
TRT_DEPRECATED bool hasImplicitBatchDimension() const noexcept
Query whether the network was created with an implicit batch dimension.
Definition: NvInfer.h:8970
apiv::VNetworkDefinition * mImpl
Definition: NvInfer.h:9935
IKVCacheUpdateLayer * addKVCacheUpdate(ITensor &cache, ITensor &update, ITensor &writeIndices, KVCacheMode cacheMode) noexcept
Add a KVCacheUpdate layer to the network.
Definition: NvInfer.h:9727
bool markOutputForShapes(ITensor &tensor) noexcept
Enable tensor's value to be computed by IExecutionContext::getShapeBinding.
Definition: NvInfer.h:9009
IOneHotLayer * addOneHot(ITensor &indices, ITensor &values, ITensor &depth, int32_t axis) noexcept
Add a OneHot layer to the network.
Definition: NvInfer.h:8477
IScaleLayer * addScale(ITensor &input, ScaleMode mode, Weights shift, Weights scale, Weights power) noexcept
Add a Scale layer to the network.
Definition: NvInfer.h:8367
IPluginV3Layer * addPluginV3(ITensor *const *inputs, int32_t nbInputs, ITensor *const *shapeInputs, int32_t nbShapeInputs, IPluginV3 &plugin) noexcept
Add a plugin layer implementing the IPluginV3 interface to the network.
Definition: NvInfer.h:8882
void unmarkOutput(ITensor &tensor) noexcept
unmark a tensor as a network output.
Definition: NvInfer.h:8843
IIdentityLayer * addIdentity(ITensor &input) noexcept
Add an identity layer.
Definition: NvInfer.h:8801
IGatherLayer * addGatherV2(ITensor &data, ITensor &indices, GatherMode mode) noexcept
Add gather with specified mode, axis=0 and nbElementWiseDims=0.
Definition: NvInfer.h:8688
INonZeroLayer * addNonZero(ITensor &input, DataType indicesType) noexcept
Add a nonzero layer to the network.
Definition: NvInfer.h:8763
IElementWiseLayer * addElementWise(ITensor &input1, ITensor &input2, ElementWiseOperation op) noexcept
Add an elementwise layer to the network.
Definition: NvInfer.h:8424
IConstantLayer * addConstant(Dims const &dimensions, Weights weights) noexcept
Add a constant layer to the network.
Definition: NvInfer.h:8787
void setErrorRecorder(IErrorRecorder *recorder) noexcept
Set the ErrorRecorder for this interface.
Definition: NvInfer.h:9327
IPoolingLayer * addPoolingNd(ITensor &input, PoolingType type, Dims const &windowSize) noexcept
Add a multi-dimension pooling layer to the network.
Definition: NvInfer.h:9081
INMSLayer * addNMS(ITensor &boxes, ITensor &scores, ITensor &maxOutputBoxesPerClass, DataType indicesType) noexcept
Add a non-maximum suppression layer to the network.
Definition: NvInfer.h:9536
IRaggedSoftMaxLayer * addRaggedSoftMax(ITensor &input, ITensor &bounds) noexcept
Add a RaggedSoftMax layer to the network.
Definition: NvInfer.h:8707
IShapeLayer * addShape(ITensor &input) noexcept
Add a shape layer to the network.
Definition: NvInfer.h:8956
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:8672
bool unmarkWeightsRefittable(char const *name) noexcept
Unmark weights as refittable when the builder flag kREFIT_INDIVIDUAL is set.
Definition: NvInfer.h:9821
bool markWeightsRefittable(char const *name) noexcept
Mark weights as refittable when the builder flag kREFIT_INDIVIDUAL is set.
Definition: NvInfer.h:9808
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:9693
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:9103
IResizeLayer * addResize(ITensor &input) noexcept
Add a resize layer to the network.
Definition: NvInfer.h:9157
IUnsqueezeLayer * addUnsqueeze(ITensor &input, ITensor &axes) noexcept
Add an unsqueeze layer to the network.
Definition: NvInfer.h:9903
IMatrixMultiplyLayer * addMatrixMultiply(ITensor &input0, MatrixOperation op0, ITensor &input1, MatrixOperation op1) noexcept
Add a MatrixMultiply layer to the network.
Definition: NvInfer.h:8728
ISoftMaxLayer * addSoftMax(ITensor &input) noexcept
Add a SoftMax layer to the network.
Definition: NvInfer.h:8380
bool unmarkDebug(ITensor &tensor) noexcept
Unmark a tensor as a debug tensor.
Definition: NvInfer.h:8256
TRT_DEPRECATED IAttention * addAttention(ITensor &query, ITensor &key, ITensor &value, AttentionNormalizationOp normOp, bool causal) noexcept
Add an attention to the network.
Definition: NvInfer.h:9637
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:9476
void markOutput(ITensor &tensor) noexcept
Mark a tensor as a network output.
Definition: NvInfer.h:8222
TRT_DEPRECATED IPluginV2Layer * addPluginV2(ITensor *const *inputs, int32_t nbInputs, IPluginV2 &plugin) noexcept
Add a plugin layer to the network using the IPluginV2 interface.
Definition: NvInfer.h:8864
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:9284
int32_t getNbOutputs() const noexcept
Get the number of outputs in the network.
Definition: NvInfer.h:8545
ISqueezeLayer * addSqueeze(ITensor &input, ITensor *axes) noexcept
Add a squeeze layer to the network, with axes given by a tensor or a nullptr.
Definition: NvInfer.h:9882
bool setWeightsName(Weights weights, char const *name) noexcept
Associate a name with all current uses of the given weights.
Definition: NvInfer.h:9308
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:9781
IMoELayer * addMoE(ITensor &hiddenStates, ITensor &selectedExpertsForTokens, ITensor &scoresForSelectedExperts) noexcept
Add a MoE (Mixture of Experts) layer to the network.
Definition: NvInfer.h:9749
bool unmarkUnfusedTensorsAsDebugTensors() noexcept
Undo the marking of unfused tensors as debug tensors.
Definition: NvInfer.h:8302
Forward declaration of IEngineInspector for use by other interfaces.
Definition: NvInferRuntime.h:51
Definition: NvInfer.h:3664
DataType getIndicesType() const noexcept
Return the NonZero layer indices type.
Definition: NvInfer.h:3688
virtual ~INonZeroLayer() noexcept=0
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:3676
A normalization layer in a network definition.
Definition: NvInfer.h:6427
float getEpsilon() const noexcept
Get the epsilon value used for the normalization calculation.
Definition: NvInfer.h:6446
uint32_t getAxes() const noexcept
Get the axes value used for the normalization calculation.
Definition: NvInfer.h:6466
void setEpsilon(float eps) noexcept
Set the epsilon value used for the normalization calculation.
Definition: NvInfer.h:6436
virtual ~INormalizationLayer() noexcept=0
TRT_NODISCARD bool isV2() const noexcept
Returns true if this layer was created through addNormalizationV2().
Definition: NvInfer.h:6508
apiv::VNormalizationLayer * mImpl
Definition: NvInfer.h:6514
int64_t getNbGroups() const noexcept
Get the number of groups used to split the channels for the normalization calculation.
Definition: NvInfer.h:6497
void setAxes(uint32_t axesMask) noexcept
Set the reduction axes for the normalization calculation.
Definition: NvInfer.h:6456
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups used to split the channels in the normalization calculation.
Definition: NvInfer.h:6487
A OneHot layer in a network definition.
Definition: NvInfer.h:6041
apiv::VOneHotLayer * mImpl
Definition: NvInfer.h:6062
void setAxis(int32_t axis) noexcept
Set the axis parameter.
Definition: NvInfer.h:6048
virtual ~IOneHotLayer() noexcept=0
int32_t getAxis() const noexcept
Get the value of the axis parameter.
Definition: NvInfer.h:6056
Optimization profile for dynamic input dimensions and shape tensors.
Definition: NvInferRuntime.h:2667
Layer that represents a padding operation.
Definition: NvInfer.h:2849
Dims getPostPaddingNd() const noexcept
Get the padding that is applied at the end of the tensor.
Definition: NvInfer.h:2898
void setPrePaddingNd(Dims const &padding) noexcept
Set the padding that is applied at the start of the tensor.
Definition: NvInfer.h:2860
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:2886
Dims getPrePaddingNd() const noexcept
Get the padding that is applied at the start of the tensor.
Definition: NvInfer.h:2872
apiv::VPaddingLayer * mImpl
Definition: NvInfer.h:2904
Layer that represents a parametric ReLU operation.
Definition: NvInfer.h:3889
apiv::VParametricReLULayer * mImpl
Definition: NvInfer.h:3891
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:2542
apiv::VPluginV2Layer * mImpl
Definition: NvInfer.h:2555
IPluginV2 & getPlugin() noexcept
Get the plugin for the layer.
Definition: NvInfer.h:2549
virtual ~IPluginV2Layer() noexcept=0
Layer type for V3 plugins.
Definition: NvInfer.h:2571
virtual ~IPluginV3Layer() noexcept=0
IPluginV3 & getPlugin() noexcept
Get the plugin for the layer.
Definition: NvInfer.h:2578
apiv::VPluginV3Layer * mImpl
Definition: NvInfer.h:2584
A Pooling layer in a network definition.
Definition: NvInfer.h:1292
PoolingType getPoolingType() const noexcept
Get the type of activation to be performed.
Definition: NvInfer.h:1311
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:1444
Dims getPostPadding() const noexcept
Get the padding.
Definition: NvInfer.h:1420
bool getAverageCountExcludesPadding() const noexcept
Get whether average pooling uses as a denominator the overlap area between the window and the unpadde...
Definition: NvInfer.h:1364
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:1392
void setPoolingType(PoolingType type) noexcept
Set the type of activation to be performed.
Definition: NvInfer.h:1301
void setWindowSizeNd(Dims const &windowSize) noexcept
Set the multi-dimension window size for pooling.
Definition: NvInfer.h:1457
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:1433
Dims getWindowSizeNd() const noexcept
Get the multi-dimension window size for pooling.
Definition: NvInfer.h:1467
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:1353
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding for pooling.
Definition: NvInfer.h:1510
float getBlendFactor() const noexcept
Get the blending factor for the max_average_blend mode: max_average_blendPool = (1-blendFactor)*maxPo...
Definition: NvInfer.h:1339
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride for pooling.
Definition: NvInfer.h:1481
Dims getStrideNd() const noexcept
Get the multi-dimension stride for pooling.
Definition: NvInfer.h:1491
Dims getPaddingNd() const noexcept
Get the multi-dimension padding for pooling.
Definition: NvInfer.h:1522
virtual ~IPoolingLayer() noexcept=0
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding for pooling.
Definition: NvInfer.h:1410
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding for pooling.
Definition: NvInfer.h:1382
void setBlendFactor(float blendFactor) noexcept
Set the blending factor for the max_average_blend mode: max_average_blendPool = (1-blendFactor)*maxPo...
Definition: NvInfer.h:1326
A Quantize layer in a network definition.
Definition: NvInfer.h:5406
void setToType(DataType toType) noexcept
Set the Quantize layer output type.
Definition: NvInfer.h:5467
bool setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5440
void setAxis(int32_t axis) noexcept
Set the quantization axis.
Definition: NvInfer.h:5427
virtual ~IQuantizeLayer() noexcept=0
TRT_NODISCARD Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5451
int32_t getAxis() const noexcept
Get the quantization axis.
Definition: NvInfer.h:5416
DataType getToType() const noexcept
Return the Quantize layer output type.
Definition: NvInfer.h:5479
A RaggedSoftmax layer in a network definition.
Definition: NvInfer.h:3715
apiv::VRaggedSoftMaxLayer * mImpl
Definition: NvInfer.h:3717
virtual ~IRaggedSoftMaxLayer() noexcept=0
A recurrence layer in a network definition.
Definition: NvInfer.h:4599
virtual ~IRecurrenceLayer() noexcept=0
Layer that represents a reduction across a non-bool tensor.
Definition: NvInfer.h:2767
void setKeepDimensions(bool keepDimensions) noexcept
Set the boolean that specifies whether or not to keep the reduced dimensions for the layer.
Definition: NvInfer.h:2814
virtual ~IReduceLayer() noexcept=0
void setOperation(ReduceOperation op) noexcept
Set the reduce operation for the layer.
Definition: NvInfer.h:2774
ReduceOperation getOperation() const noexcept
Get the reduce operation for the layer.
Definition: NvInfer.h:2784
uint32_t getReduceAxes() const noexcept
Get the axes over which to reduce for the layer.
Definition: NvInfer.h:2804
void setReduceAxes(uint32_t reduceAxes) noexcept
Set the axes over which to reduce.
Definition: NvInfer.h:2794
apiv::VReduceLayer * mImpl
Definition: NvInfer.h:2830
bool getKeepDimensions() const noexcept
Get the boolean that specifies whether or not to keep the reduced dimensions for the layer.
Definition: NvInfer.h:2824
A resize layer in a network definition.
Definition: NvInfer.h:4068
void setSelectorForSinglePixel(ResizeSelector selector) noexcept
Set coordinate selector function when resized to single pixel.
Definition: NvInfer.h:4229
void setNearestRounding(ResizeRoundMode value) noexcept
Set rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4253
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:4147
void setOutputDimensions(Dims const &dimensions) noexcept
Set the output dimensions.
Definition: NvInfer.h:4088
void setCubicCoeff(float A) noexcept
Set the coefficient 'A' used in cubic interpolation.
Definition: NvInfer.h:4285
void setScales(float const *scales, int32_t nbScales) noexcept
Set the resize scales.
Definition: NvInfer.h:4128
virtual ~IResizeLayer() noexcept=0
float getCubicCoeff() const noexcept
Get the coefficient 'A' used in cubic interpolation.
Definition: NvInfer.h:4295
ResizeSelector getSelectorForSinglePixel() const noexcept
Get the coordinate selector function when resized to single pixel.
Definition: NvInfer.h:4239
InterpolationMode getResizeMode() const noexcept
Get resize mode for an input tensor.
Definition: NvInfer.h:4169
void setCoordinateTransformation(ResizeCoordinateTransformation coordTransform) noexcept
Set coordinate transformation function.
Definition: NvInfer.h:4204
void setExcludeOutside(bool excludeFlag) noexcept
Set the state for excluding outside pixels.
Definition: NvInfer.h:4308
void setResizeMode(InterpolationMode interpolationMode) noexcept
Set resize mode for an input tensor.
Definition: NvInfer.h:4159
Dims getOutputDimensions() const noexcept
Get the output dimensions.
Definition: NvInfer.h:4098
ResizeRoundMode getNearestRounding() const noexcept
Get rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4263
bool getExcludeOutside() const noexcept
Get the state for excluding outside pixels.
Definition: NvInfer.h:4318
ResizeCoordinateTransformation getCoordinateTransformation() const noexcept
Get coordinate transformation function.
Definition: NvInfer.h:4214
A ReverseSequence layer in a network definition.
Definition: NvInfer.h:6353
void setSequenceAxis(int32_t sequenceAxis) noexcept
Set the sequence axis. Default is 0.
Definition: NvInfer.h:6386
int32_t getBatchAxis() const noexcept
Return the batch axis. Return 1 if no batch axis was set.
Definition: NvInfer.h:6373
apiv::VReverseSequenceLayer * mImpl
Definition: NvInfer.h:6402
int32_t getSequenceAxis() const noexcept
Return the sequence axis. Return 0 if no sequence axis was set.
Definition: NvInfer.h:6396
void setBatchAxis(int32_t batchAxis) noexcept
Set the batch axis. Default is 1.
Definition: NvInfer.h:6363
virtual ~IReverseSequenceLayer() noexcept=0
Layer that implements Rotary Position Embedding (RoPE) (https://arxiv.org/abs/2104....
Definition: NvInfer.h:7458
TRT_NODISCARD int32_t getRotaryEmbeddingDim() const noexcept
Get the number of hidden dimensions participating in RoPE. The default value is 0,...
Definition: NvInfer.h:7498
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:7465
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:7487
apiv::VRotaryEmbeddingLayer * mImpl
Definition: NvInfer.h:7520
TRT_NODISCARD bool getInterleaved() const noexcept
Get whether the input is in interleaved format. The default value is false.
Definition: NvInfer.h:7476
A Scale layer in a network definition.
Definition: NvInfer.h:1692
Weights getScale() const noexcept
Get the scale value.
Definition: NvInfer.h:1749
Weights getPower() const noexcept
Get the power value.
Definition: NvInfer.h:1769
void setScale(Weights scale) noexcept
Set the scale value.
Definition: NvInfer.h:1739
void setPower(Weights power) noexcept
Set the power value.
Definition: NvInfer.h:1759
ScaleMode getMode() const noexcept
Get the scale mode.
Definition: NvInfer.h:1709
void setShift(Weights shift) noexcept
Set the shift value.
Definition: NvInfer.h:1719
void setChannelAxis(int32_t channelAxis) noexcept
Set the channel axis.
Definition: NvInfer.h:1805
virtual ~IScaleLayer() noexcept=0
Weights getShift() const noexcept
Get the shift value.
Definition: NvInfer.h:1729
void setMode(ScaleMode mode) noexcept
Set the scale mode.
Definition: NvInfer.h:1699
int32_t getChannelAxis() const noexcept
Get the channel axis.
Definition: NvInfer.h:1784
A scatter layer in a network definition. Supports several kinds of scattering.
Definition: NvInfer.h:5966
void setMode(ScatterMode mode) noexcept
Set the scatter mode.
Definition: NvInfer.h:5973
apiv::VScatterLayer * mImpl
Definition: NvInfer.h:6007
void setAxis(int32_t axis) noexcept
Set the axis used by ScatterMode::kELEMENTS.
Definition: NvInfer.h:5993
int32_t getAxis() const noexcept
Get the axis.
Definition: NvInfer.h:6001
ScatterMode getMode() const noexcept
Get the scatter mode.
Definition: NvInfer.h:5983
virtual ~IScatterLayer() noexcept=0
Select elements from two data tensors based on a condition tensor.
Definition: NvInfer.h:4917
virtual ~ISelectLayer() noexcept=0
Layer type for getting shape of a tensor.
Definition: NvInfer.h:3380
virtual ~IShapeLayer() noexcept=0
apiv::VShapeLayer * mImpl
Definition: NvInfer.h:3382
Layer type for shuffling data.
Definition: NvInfer.h:2939
apiv::VShuffleLayer * mImpl
Definition: NvInfer.h:3097
virtual ~IShuffleLayer() noexcept=0
void setFirstTranspose(Permutation permutation) noexcept
Set the permutation applied by the first transpose operation.
Definition: NvInfer.h:2950
void setSecondTranspose(Permutation permutation) noexcept
Set the permutation applied by the second transpose operation.
Definition: NvInfer.h:3050
Dims getReshapeDimensions() const noexcept
Get the reshaped dimensions.
Definition: NvInfer.h:3003
void setReshapeDimensions(Dims const &dimensions) noexcept
Set the reshaped dimensions.
Definition: NvInfer.h:2990
Permutation getFirstTranspose() const noexcept
Get the permutation applied by the first transpose operation.
Definition: NvInfer.h:2962
Permutation getSecondTranspose() const noexcept
Get the permutation applied by the second transpose operation.
Definition: NvInfer.h:3062
bool getZeroIsPlaceholder() const noexcept
Get meaning of 0 in reshape dimensions.
Definition: NvInfer.h:3091
void setZeroIsPlaceholder(bool zeroIsPlaceholder) noexcept
Set meaning of 0 in reshape dimensions.
Definition: NvInfer.h:3078
Slices an input tensor into an output tensor based on the offset and strides.
Definition: NvInfer.h:3193
void setStride(Dims const &stride) noexcept
Set the stride for computing the output slice data.
Definition: NvInfer.h:3262
apiv::VSliceLayer * mImpl
Definition: NvInfer.h:3361
virtual ~ISliceLayer() noexcept=0
void setSize(Dims const &size) noexcept
Set the dimensions of the output slice.
Definition: NvInfer.h:3233
void setAxes(Dims const &axes) noexcept
Set the axes for this ISliceLayer.
Definition: NvInfer.h:3340
void setStart(Dims const &start) noexcept
Set the start offset that the slice layer uses to create the output slice.
Definition: NvInfer.h:3204
Dims getStart() const noexcept
Get the start offset for the slice layer.
Definition: NvInfer.h:3219
void setMode(SampleMode mode) noexcept
Set the slice mode.
Definition: NvInfer.h:3287
Dims getSize() const noexcept
Get dimensions of the output slice.
Definition: NvInfer.h:3248
SampleMode getMode() const noexcept
Get the slice mode.
Definition: NvInfer.h:3297
Dims getStride() const noexcept
Get the stride for the output slice.
Definition: NvInfer.h:3277
Dims getAxes() const noexcept
Get the axes for this ISliceLayer.
Definition: NvInfer.h:3355
A Softmax layer in a network definition.
Definition: NvInfer.h:1838
void setAxes(uint32_t axes) noexcept
Set the axis along which softmax is computed. Currently, only one axis can be set.
Definition: NvInfer.h:1860
virtual ~ISoftMaxLayer() noexcept=0
uint32_t getAxes() const noexcept
Get the axis along which softmax occurs.
Definition: NvInfer.h:1870
Layer that represents a squeeze operation, removing unit dimensions of the first input tensor on a se...
Definition: NvInfer.h:6533
apiv::VSqueezeLayer * mImpl
Definition: NvInfer.h:6553
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:364
TensorLocation getLocation() const noexcept
Get the storage location of a tensor.
Definition: NvInfer.h:322
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:429
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:408
bool isNetworkInput() const noexcept
Whether the tensor is a network input.
Definition: NvInfer.h:271
TRT_DEPRECATED void setBroadcastAcrossBatch(bool broadcastAcrossBatch) noexcept
Set whether to enable broadcast of tensor across the implicit batch dimension.
Definition: NvInfer.h:296
TRT_DEPRECATED bool getBroadcastAcrossBatch() const noexcept
Check if tensor is broadcast across the implicit batch dimension.
Definition: NvInfer.h:310
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:476
void setDimensionName(int32_t index, char const *name) noexcept
Name a dimension of an input tensor.
Definition: NvInfer.h:455
char const * getDimensionName(int32_t index) const noexcept
Get the name of an input dimension.
Definition: NvInfer.h:470
TRT_DEPRECATED void setLocation(TensorLocation location) noexcept
Set the storage location of a tensor.
Definition: NvInfer.h:341
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:377
Class to handle tactic timing info collected from builder.
Definition: NvInfer.h:10210
int64_t queryKeys(TimingCacheKey *keyBuffer, int64_t capacity) const noexcept
Query cache keys from Timing Cache.
Definition: NvInfer.h:10276
virtual ~ITimingCache() noexcept=0
bool combine(ITimingCache const &inputCache, bool ignoreMismatch) noexcept
Combine input timing cache into local instance.
Definition: NvInfer.h:10247
TimingCacheValue query(TimingCacheKey const &key) const noexcept
Query value in a cache entry.
Definition: NvInfer.h:10293
bool update(TimingCacheKey const &key, TimingCacheValue const &value) noexcept
Update values in a cache entry.
Definition: NvInfer.h:10315
apiv::VTimingCache * mImpl
Definition: NvInfer.h:10321
bool reset() noexcept
Empty the timing cache.
Definition: NvInfer.h:10257
Layer that represents a TopK reduction.
Definition: NvInfer.h:3422
void setK(int32_t k) noexcept
Set the static k value for the layer.
Definition: NvInfer.h:3453
void setReduceAxes(uint32_t reduceAxes) noexcept
Set which axes to reduce for the layer.
Definition: NvInfer.h:3477
TopKOperation getOperation() const noexcept
Get the operation for the layer.
Definition: NvInfer.h:3439
apiv::VTopKLayer * mImpl
Definition: NvInfer.h:3536
void setOperation(TopKOperation op) noexcept
Set the operation for the layer.
Definition: NvInfer.h:3429
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:3518
virtual ~ITopKLayer() noexcept=0
int32_t getK() const noexcept
Get the k value for the layer.
Definition: NvInfer.h:3467
uint32_t getReduceAxes() const noexcept
Get the axes to reduce for the layer.
Definition: NvInfer.h:3487
DataType getIndicesType() const noexcept
Return the TopK layer indices type.
Definition: NvInfer.h:3530
A layer that represents a trip-count limiter.
Definition: NvInfer.h:4724
virtual ~ITripLimitLayer() noexcept=0
TripLimit getTripLimit() const noexcept
Get a trip limiter type.
Definition: NvInfer.h:4729
Layer that represents an unary operation.
Definition: NvInfer.h:2654
void setOperation(UnaryOperation op) noexcept
Set the unary operation for the layer.
Definition: NvInfer.h:2663
apiv::VUnaryLayer * mImpl
Definition: NvInfer.h:2679
UnaryOperation getOperation() const noexcept
Get the unary operation for the layer.
Definition: NvInfer.h:2673
virtual ~IUnaryLayer() noexcept=0
Layer that represents an unsqueeze operation, which reshapes the first input tensor by inserting unit...
Definition: NvInfer.h:6568
virtual ~IUnsqueezeLayer() noexcept=0
apiv::VUnsqueezeLayer * mImpl
Definition: NvInfer.h:6586
An Interface class for version control.
Definition: NvInferRuntimeBase.h:284
Version information associated with a TRT interface.
Definition: NvInferRuntimeBase.h:249
An array of weights used as a layer parameter.
Definition: NvInferRuntime.h:132
Definition: NvInferRuntimeBase.h:421
Definition: NvInferRuntime.h:1691
Application-implemented logging interface for the builder, refitter and runtime.
Definition: NvInferRuntime.h:1614
Definition: NvInferPluginBase.h:206
Definition: NvInfer.h:10530
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:654
IBuilder * createInferBuilder(ILogger &logger) noexcept
Create an instance of an IBuilder class.
Definition: NvInfer.h:11916
The TensorRT API version 1 namespace.
Definition: NvInferSafePlugin.h:33
uint32_t TacticSources
Represents a collection of one or more TacticSource values combine using bitwise-OR operations.
Definition: NvInferRuntime.h:2910
ResizeSelector
The coordinate selector when resize to single pixel output.
Definition: NvInfer.h:3979
@ 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:10334
AttentionIOForm
Enumerates the layout of the input/output tensors in an Attention layer.
Definition: NvInfer.h:6796
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:1649
@ 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:9958
HardwareCompatibilityLevel
Describes requirements of compatibility with GPU architectures other than that of the GPU on which th...
Definition: NvInfer.h:10447
@ kSAME_COMPUTE_CAPABILITY
CumulativeOperation
Enumerates the cumulative operations that may be performed by a Cumulative layer.
Definition: NvInfer.h:6604
BoundingBoxFormat
Representation of bounding box data used for the Boxes input tensor in INMSLayer.
Definition: NvInfer.h:6165
@ 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
UnaryOperation
Enumerates the unary operations that may be performed by a Unary layer.
Definition: NvInfer.h:2607
@ 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:4982
ResizeRoundMode
The rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4006
@ 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:6768
@ 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:826
@ 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:4364
@ 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:11533
PreviewFeature
Define preview features.
Definition: NvInfer.h:10413
@ kRUNTIME_ACTIVATION_RESIZE_10_10
@ kALIASED_PLUGIN_IO_10_03
TilingOptimizationLevel
Define the optimization levels for Tiling.
Definition: NvInfer.h:10500
@ 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:151
uint32_t BuilderFlags
Represents one or more BuilderFlag values using binary OR operations, e.g., 1U << BuilderFlag::kDEBUG...
Definition: NvInfer.h:9988
DeviceType
The device that this layer/network will execute on.
Definition: NvInferRuntime.h:1352
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:3109
@ 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:2348
@ 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:7685
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:2922
NetworkDefinitionCreationFlag
List of immutable network properties expressed at network creation time. NetworkDefinitionCreationFla...
Definition: NvInfer.h:11544
@ kPREFER_JIT_PYTHON_PLUGINS
@ kPREFER_AOT_PYTHON_PLUGINS
ElementWiseOperation
Enumerates the binary operations that may be performed by an ElementWise layer.
Definition: NvInfer.h:2259
@ 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:2737
@ kALL_TO_ALL
All-to-all exchange.
@ kREDUCE_SCATTER
Reduce scatter.
InterpolationMode
Enumerates various modes of interpolation.
Definition: NvInfer.h:3903
@ kNEAREST
ND (0 < N <= 8) nearest neighbor resizing.
@ kCUBIC
Supports bicubic (2D) and tricubic (3D) 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:9998
@ kWEIGHT_STREAMING
Enable weight streaming for the current engine.
@ 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.
@ kERROR_ON_TIMING_CACHE_MISS
@ kEDITABLE_TIMING_CACHE
Enable editable timing cache.
@ kDISTRIBUTIVE_INDEPENDENCE
@ 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:3394
ReduceOperation
Enumerates the reduce operations that may be performed by a Reduce layer.
Definition: NvInfer.h:2709
@ kAVG
Average of the elements.
TRT_DEPRECATED_API nvinfer1::safe::IPluginRegistry * getBuilderSafePluginRegistry(nvinfer1::EngineCapability capability) noexcept
Return the plugin registry for building a Safety engine, or nullptr if no registry exists.
ScatterMode
Control form of IScatterLayer.
Definition: NvInfer.h:5892
MatrixOperation
Enumerates the operations that may be performed on a tensor by IMatrixMultiplyLayer before multiplica...
Definition: NvInfer.h:3549
@ kTRANSPOSE
Like kNONE, but transpose the matrix dimensions.
ResizeCoordinateTransformation
The resize coordinate transformation function.
Definition: NvInfer.h:3928
LoopOutput
Enum that describes kinds of loop outputs.
Definition: NvInfer.h:4336
@ 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:7532
PoolingType
The type of pooling to perform in a pooling layer.
Definition: NvInfer.h:1263
@ 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:10613
TensorLocation
The location for tensor data storage, device or host.
Definition: NvInferRuntime.h:214
AttentionNormalizationOp
Enumerates the operations that may be performed by the normalization in the attention subgraph.
Definition: NvInfer.h:6736
Represents a permutation of dimensions.
Definition: NvInfer.h:2916
Declaration of EnumMaxImpl struct to store the exclusive upper bound of an enumeration type.
Definition: NvInferRuntimeBase.h:132
The key to retrieve timing cache entries.
Definition: NvInfer.h:10174
Definition: NvInfer.h:10186
uint64_t tacticHash
Hash of the selected tactic.
Definition: NvInfer.h:10188
float timingMSec
Timing of this tactic in milliseconds. Negative numbers and NaN are invalid values.
Definition: NvInfer.h:10190