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();
1484 mImpl->setStrideNd(stride);
1494 return mImpl->getStrideNd();
1513 mImpl->setPaddingNd(padding);
1525 return mImpl->getPaddingNd();
1558 mImpl->setWindowSize(windowSize);
1568 return mImpl->getWindowSize();
1580 mImpl->setAlpha(alpha);
1590 return mImpl->getAlpha();
1602 mImpl->setBeta(beta);
1612 return mImpl->getBeta();
1634 return mImpl->getK();
1664 static constexpr int32_t kVALUE = 3;
1702 mImpl->setMode(mode);
1712 return mImpl->getMode();
1722 mImpl->setShift(shift);
1732 return mImpl->getShift();
1742 mImpl->setScale(scale);
1752 return mImpl->getScale();
1762 mImpl->setPower(power);
1772 return mImpl->getPower();
1787 return mImpl->getChannelAxis();
1808 mImpl->setChannelAxis(channelAxis);
1863 mImpl->setAxes(axes);
1873 return mImpl->getAxes();
1911 mImpl->setAxis(axis);
1921 return mImpl->getAxis();
1950 mImpl->setNbOutputMaps(nbOutputMaps);
1960 return mImpl->getNbOutputMaps();
1980 mImpl->setNbGroups(nbGroups);
1990 return mImpl->getNbGroups();
2004 mImpl->setKernelWeights(weights);
2014 return mImpl->getKernelWeights();
2029 mImpl->setBiasWeights(weights);
2039 return mImpl->getBiasWeights();
2056 mImpl->setPrePadding(padding);
2066 return mImpl->getPrePadding();
2083 mImpl->setPostPadding(padding);
2093 return mImpl->getPostPadding();
2107 mImpl->setPaddingMode(paddingMode);
2119 return mImpl->getPaddingMode();
2134 mImpl->setKernelSizeNd(kernelSize);
2144 return mImpl->getKernelSizeNd();
2161 mImpl->setStrideNd(stride);
2171 return mImpl->getStrideNd();
2189 mImpl->setPaddingNd(padding);
2201 return mImpl->getPaddingNd();
2227 mImpl->setDilationNd(dilation);
2237 return mImpl->getDilationNd();
2285 static constexpr int32_t kVALUE = 14;
2321 return mImpl->setOperation(op);
2333 return mImpl->getOperation();
2363 static constexpr int32_t kVALUE = 3;
2456 mImpl->setGatherAxis(axis);
2468 return mImpl->getGatherAxis();
2491 mImpl->setNbElementWiseDims(elementWiseDims);
2501 return mImpl->getNbElementWiseDims();
2511 mImpl->setMode(mode);
2521 return mImpl->getMode();
2552 return mImpl->getPlugin();
2581 return mImpl->getPlugin();
2644 static constexpr int32_t kVALUE = 25;
2666 mImpl->setOperation(op);
2676 return mImpl->getOperation();
2727 static constexpr int32_t kVALUE = 6;
2757 static constexpr int32_t kVALUE = 8;
2777 mImpl->setOperation(op);
2787 return mImpl->getOperation();
2797 mImpl->setReduceAxes(reduceAxes);
2807 return mImpl->getReduceAxes();
2817 mImpl->setKeepDimensions(keepDimensions);
2827 return mImpl->getKeepDimensions();
2863 mImpl->setPrePaddingNd(padding);
2875 return mImpl->getPrePaddingNd();
2889 mImpl->setPostPaddingNd(padding);
2901 return mImpl->getPostPaddingNd();
2953 mImpl->setFirstTranspose(permutation);
2965 return mImpl->getFirstTranspose();
2993 mImpl->setReshapeDimensions(dimensions);
3006 return mImpl->getReshapeDimensions();
3053 mImpl->setSecondTranspose(permutation);
3065 return mImpl->getSecondTranspose();
3081 return mImpl->setZeroIsPlaceholder(zeroIsPlaceholder);
3094 return mImpl->getZeroIsPlaceholder();
3128 static constexpr int32_t kVALUE = 5;
3207 mImpl->setStart(start);
3222 return mImpl->getStart();
3236 return mImpl->setSize(size);
3251 return mImpl->getSize();
3265 mImpl->setStride(stride);
3280 return mImpl->getStride();
3290 mImpl->setMode(mode);
3300 return mImpl->getMode();
3343 mImpl->setAxes(axes);
3358 return mImpl->getAxes();
3408 static constexpr int32_t kVALUE = 2;
3432 mImpl->setOperation(op);
3442 return mImpl->getOperation();
3470 return mImpl->getK();
3480 mImpl->setReduceAxes(reduceAxes);
3490 return mImpl->getReduceAxes();
3521 return mImpl->setIndicesType(type);
3533 return mImpl->getIndicesType();
3580 static constexpr int32_t kVALUE = 3;
3621 mImpl->setOperation(index, op);
3633 return mImpl->getOperation(index);
3679 return mImpl->setIndicesType(type);
3691 return mImpl->getIndicesType();
3794 mImpl->setToType(toType);
3805 return mImpl->getToType();
3836 mImpl->setWeights(weights);
3846 return mImpl->getWeights();
3858 mImpl->setDimensions(dimensions);
3870 return mImpl->getDimensions();
3918 static constexpr int32_t kVALUE = 3;
3969 static constexpr int32_t kVALUE = 3;
3996 static constexpr int32_t kVALUE = 2;
4029 static constexpr int32_t kVALUE = 4;
4091 return mImpl->setOutputDimensions(dimensions);
4101 return mImpl->getOutputDimensions();
4129 void setScales(
float const* scales, int32_t nbScales)
noexcept
4131 mImpl->setScales(scales, nbScales);
4148 int32_t
getScales(int32_t size,
float* scales)
const noexcept
4150 return mImpl->getScales(size, scales);
4162 mImpl->setResizeMode(interpolationMode);
4172 return mImpl->getResizeMode();
4207 mImpl->setCoordinateTransformation(coordTransform);
4217 return mImpl->getCoordinateTransformation();
4232 mImpl->setSelectorForSinglePixel(selector);
4242 return mImpl->getSelectorForSinglePixel();
4256 mImpl->setNearestRounding(value);
4266 return mImpl->getNearestRounding();
4288 mImpl->setCubicCoeff(A);
4298 return mImpl->getCubicCoeff();
4311 mImpl->setExcludeOutside(excludeFlag);
4321 return mImpl->getExcludeOutside();
4356 static constexpr int32_t kVALUE = 3;
4379 static constexpr int32_t kVALUE = 2;
4405 return mBoundary->getLoop();
4410 apiv::VLoopBoundaryLayer* mBoundary;
4430 return mBoundary->getConditional();
4435 apiv::VConditionalBoundaryLayer* mBoundary;
4527 return mImpl->setCondition(condition);
4545 return mImpl->addOutput(trueSubgraphOutput, falseSubgraphOutput);
4557 return mImpl->addInput(input);
4572 mImpl->setName(name);
4582 return mImpl->getName();
4656 return mImpl->getLoopOutput();
4673 mImpl->setAxis(axis);
4681 return mImpl->getAxis();
4732 return mImpl->getTripLimit();
4760 mImpl->setAxis(axis);
4768 return mImpl->getAxis();
4782 mImpl->setReverse(reverse);
4792 return mImpl->getReverse();
4823 return mImpl->addRecurrence(initialValue);
4844 return mImpl->addTripLimit(tensor, limit);
4857 return mImpl->addIterator(tensor, axis, reverse);
4870 return mImpl->addLoopOutput(tensor, outputKind, axis);
4885 mImpl->setName(name);
4895 return mImpl->getName();
4954 mImpl->setMessage(message);
4964 return mImpl->getMessage();
5019 static constexpr int32_t kVALUE = 3;
5071 mImpl->setDimensions(dimensions);
5086 return mImpl->getDimensions();
5096 mImpl->setOperation(op);
5106 return mImpl->getOperation();
5125 mImpl->setAlpha(alpha);
5140 return mImpl->getAlpha();
5159 mImpl->setBeta(beta);
5174 return mImpl->getBeta();
5235 mImpl->setAlphaInt64(alpha);
5250 return mImpl->getAlphaInt64();
5269 mImpl->setBetaInt64(beta);
5284 return mImpl->getBetaInt64();
5292 return mImpl->isAlphaBetaInt64();
5310 mImpl->setToType(toType);
5322 return mImpl->getToType();
5419 return mImpl->getAxis();
5430 mImpl->setAxis(axis);
5443 return mImpl->setBlockShape(blockShape);
5454 return mImpl->getBlockShape();
5470 mImpl->setToType(toType);
5482 return mImpl->getToType();
5573 return mImpl->getAxis();
5584 mImpl->setAxis(axis);
5601 return mImpl->setBlockShape(blockShape);
5612 return mImpl->getBlockShape();
5628 mImpl->setToType(toType);
5640 return mImpl->getToType();
5697 mImpl->setToType(toType);
5710 return mImpl->getToType();
5723 mImpl->setScaleType(scaleType);
5736 return mImpl->getScaleType();
5749 mImpl->setAxis(axis);
5759 return mImpl->getAxis();
5772 mImpl->setBlockSize(size);
5782 return mImpl->getBlockSize();
5795 mImpl->setBlockShape(blockShape);
5807 return mImpl->getBlockShape();
5865 return mImpl->setEquation(equation);
5875 return mImpl->getEquation();
5906 static constexpr int32_t kVALUE = 2;
5976 mImpl->setMode(mode);
5986 return mImpl->getMode();
5996 mImpl->setAxis(axis);
6004 return mImpl->getAxis();
6051 mImpl->setAxis(axis);
6059 return mImpl->getAxis();
6090 mImpl->setInterpolationMode(mode);
6102 return mImpl->getInterpolationMode();
6112 mImpl->setAlignCorners(alignCorners);
6124 return mImpl->getAlignCorners();
6136 return mImpl->setSampleMode(mode);
6148 return mImpl->getSampleMode();
6181 static constexpr int32_t kVALUE = 2;
6248 mImpl->setBoundingBoxFormat(fmt);
6260 return mImpl->getBoundingBoxFormat();
6274 mImpl->setTopKBoxLimit(limit);
6284 return mImpl->getTopKBoxLimit();
6319 return mImpl->setIndicesType(type);
6331 return mImpl->getIndicesType();
6366 mImpl->setBatchAxis(batchAxis);
6376 return mImpl->getBatchAxis();
6389 mImpl->setSequenceAxis(sequenceAxis);
6399 return mImpl->getSequenceAxis();
6439 return mImpl->setEpsilon(eps);
6449 return mImpl->getEpsilon();
6459 return mImpl->setAxes(axesMask);
6469 return mImpl->getAxes();
6490 return mImpl->setNbGroups(nbGroups);
6500 return mImpl->getNbGroups();
6511 return mImpl->isV2();
6611 static constexpr int32_t kVALUE = 1;
6655 return mImpl->setOperation(op);
6667 return mImpl->getOperation();
6679 mImpl->setExclusive(exclusive);
6691 return mImpl->getExclusive();
6703 mImpl->setReverse(reverse);
6715 return mImpl->getReverse();
6745 static constexpr int32_t kVALUE = 2;
6782 static constexpr int32_t kVALUE = 3;
6808 static constexpr int32_t kVALUE = 2;
6828 return mBoundary->getAttention();
6833 apiv::VAttentionBoundaryLayer* mBoundary;
6969 return mImpl->setNormalizationOperation(op);
6981 return mImpl->getNormalizationOperation();
6998 return mImpl->setMask(mask);
7010 return mImpl->getMask();
7028 return mImpl->setCausal(isCausal);
7042 return mImpl->getCausal();
7062 return mImpl->setCausalKind(kind);
7074 return mImpl->getCausalKind();
7086 return mImpl->setDecomposable(decomposable);
7099 return mImpl->getDecomposable();
7118 return mImpl->setInput(index, input);
7127 return mImpl->getNbInputs();
7139 return mImpl->getInput(index);
7147 return mImpl->getNbOutputs();
7159 return mImpl->getOutput(index);
7176 return mImpl->setName(name);
7188 return mImpl->getName();
7204 return mImpl->setNormalizationQuantizeScale(tensor);
7215 return mImpl->getNormalizationQuantizeScale();
7228 return mImpl->setNormalizationQuantizeToType(type);
7240 return mImpl->getNormalizationQuantizeToType();
7260 return mImpl->setMetadata(metadata);
7273 return mImpl->getMetadata();
7289 return mImpl->setNbRanks(nbRanks);
7301 return mImpl->getNbRanks();
7318 return mImpl->setQueryForm(form);
7331 return mImpl->getQueryForm();
7348 return mImpl->setKeyValueForm(form);
7361 return mImpl->getKeyValueForm();
7385 return mImpl->setQueryLengths(lengths);
7397 return mImpl->getQueryLengths();
7424 return mImpl->setKeyValueLengths(lengths);
7436 return mImpl->getKeyValueLengths();
7462 mImpl->setInterleaved(interleaved);
7473 return mImpl->getInterleaved();
7484 return mImpl->setRotaryEmbeddingDim(rotaryEmbeddingDim);
7495 return mImpl->getRotaryEmbeddingDim();
7539 static constexpr int32_t kVALUE = 1;
7589 return mImpl->setCacheMode(cacheMode);
7599 return mImpl->getCacheMode();
7617 return mImpl->setUpdateForm(form);
7630 return mImpl->getUpdateForm();
7652 return mImpl->setUpdateLengths(lengths);
7664 return mImpl->getUpdateLengths();
7693 static constexpr int32_t kVALUE = 2;
7828 mImpl->setGatedWeights(fcGateWeights, fcUpWeights, fcDownWeights, activationType);
7840 mImpl->setGatedBiases(fcGateBiases, fcUpBiases, fcDownBiases);
7852 mImpl->setActivationType(activationType);
7864 return mImpl->getActivationType();
7890 mImpl->setQuantizationStatic(fcDownActivationScale, dataType);
7923 mImpl->setQuantizationDynamicDblQ(fcDownActivationDblQScale, dataType, blockShape, dynQOutputScaleType);
7938 mImpl->setQuantizationToType(type);
7950 return mImpl->getQuantizationToType();
7966 mImpl->setQuantizationBlockShape(blockShape);
7978 return mImpl->getQuantizationBlockShape();
7990 mImpl->setDynQOutputScaleType(type);
8002 return mImpl->getDynQOutputScaleType();
8023 mImpl->setSwigluParams(limit, alpha, beta);
8037 mImpl->setSwigluParamLimit(limit);
8049 return mImpl->getSwigluParamLimit();
8063 mImpl->setSwigluParamAlpha(alpha);
8075 return mImpl->getSwigluParamAlpha();
8089 mImpl->setSwigluParamBeta(beta);
8101 return mImpl->getSwigluParamBeta();
8118 mImpl->setInput(index, tensor);
8205 return mImpl->addInput(name, type, dimensions);
8219 mImpl->markOutput(tensor);
8237 return mImpl->markDebug(tensor);
8253 return mImpl->unmarkDebug(tensor);
8263 return mImpl->isDebugTensor(tensor);
8285 return mImpl->markUnfusedTensorsAsDebugTensors();
8299 return mImpl->unmarkUnfusedTensorsAsDebugTensors();
8319 return mImpl->addActivation(input, type);
8338 return mImpl->addLRN(input, window, alpha, beta, k);
8364 return mImpl->addScale(input, mode, shift, scale, power);
8377 return mImpl->addSoftMax(input);
8394 return mImpl->addConcatenation(inputs, nbInputs);
8421 return mImpl->addElementWise(input1, input2, op);
8443 return mImpl->addUnary(input, operation);
8457 return mImpl->addShuffle(input);
8474 return mImpl->addOneHot(indices, values, depth, axis);
8486 return mImpl->getNbLayers();
8500 return mImpl->getLayer(index);
8512 return mImpl->getNbInputs();
8528 return mImpl->getInput(index);
8542 return mImpl->getNbOutputs();
8558 return mImpl->getOutput(index);
8585 return mImpl->addReduce(input, operation, reduceAxes, keepDimensions);
8620 return mImpl->addTopK(input, op, k, reduceAxes);
8653 return mImpl->addTopKV2(input, op, k, reduceAxes, indicesType);
8669 return mImpl->addGather(data, indices, axis);
8685 return mImpl->addGatherV2(data, indices, mode);
8704 return mImpl->addRaggedSoftMax(input, bounds);
8726 return mImpl->addMatrixMultiply(input0, op0, input1, op1);
8744 return mImpl->addNonZero(input);
8760 return mImpl->addNonZeroV2(input, indicesType);
8784 return mImpl->addConstant(dimensions, weights);
8798 return mImpl->addIdentity(input);
8813 return mImpl->addCast(input, toType);
8828 mImpl->removeTensor(tensor);
8840 mImpl->unmarkOutput(tensor);
8861 return mImpl->addPluginV2(inputs, nbInputs, plugin);
8878 int32_t nbShapeInputs,
IPluginV3& plugin)
noexcept
8880 return mImpl->addPluginV3(inputs, nbInputs, shapeInputs, nbShapeInputs, plugin);
8899 return mImpl->addSlice(input, start, size, stride);
8923 mImpl->setName(name);
8937 return mImpl->getName();
8953 return mImpl->addShape(input);
8967 return mImpl->hasImplicitBatchDimension();
8977 return mImpl->getFlags();
8989 return mImpl->getFlag(networkDefinitionCreationFlag);
9006 return mImpl->markOutputForShapes(tensor);
9018 return mImpl->unmarkOutputForShapes(tensor);
9036 return mImpl->addParametricReLU(input, slope);
9059 return mImpl->addConvolutionNd(input, nbOutputMaps, kernelSize, kernelWeights, biasWeights);
9078 return mImpl->addPoolingNd(input, type, windowSize);
9101 return mImpl->addDeconvolutionNd(input, nbOutputMaps, kernelSize, kernelWeights, biasWeights);
9138 return mImpl->addScaleNd(input, mode, shift, scale, power, channelAxis);
9154 return mImpl->addResize(input);
9168 return mImpl->addLoop();
9183 return mImpl->addIfConditional();
9222 return mImpl->addSelect(condition, thenInput, elseInput);
9239 return mImpl->addAssertion(condition, message);
9265 return mImpl->addFillV2(dimensions, op, outputType);
9281 return mImpl->addPaddingNd(input, prePadding, postPadding);
9305 return mImpl->setWeightsName(weights, name);
9324 mImpl->setErrorRecorder(recorder);
9339 return mImpl->getErrorRecorder();
9362 return mImpl->addDequantizeV2(input, scale, outputType);
9382 return mImpl->addScatter(data, indices, updates, mode);
9406 return mImpl->addQuantizeV2(input, scale, outputType);
9434 return mImpl->addDynamicQuantize(input, axis, blockSize, outputType, scaleType);
9458 return mImpl->addDynamicQuantizeV2(input, blockShape, outputType, scaleType);
9473 return mImpl->addEinsum(inputs, nbInputs, equation);
9491 return mImpl->addGridSample(input, grid);
9513 return mImpl->addNMS(boxes, scores, maxOutputBoxesPerClass);
9533 return mImpl->addNMSV2(boxes, scores, maxOutputBoxesPerClass, indicesType);
9550 return mImpl->addReverseSequence(input, sequenceLens);
9582 return mImpl->addNormalization(input, scale, bias, axesMask);
9604 return mImpl->addCumulative(input, axis, operation, exclusive, reverse);
9635 return mImpl->addAttention(query, key, value, normOp, causal);
9665 return mImpl->addAttentionV2(query, key, value, normOp, causalKind);
9689 return mImpl->addRotaryEmbedding(input, cosCache, sinCache, interleaved, rotaryEmbeddingDim);
9724 return mImpl->addKVCacheUpdate(cache, update, writeIndices, cacheMode);
9745 return mImpl->addMoE(hiddenStates, selectedExpertsForTokens, scoresForSelectedExperts);
9776 ReduceOperation reduceOp, int64_t root, int64_t* groups, int64_t groupSize)
noexcept
9778 return mImpl->addDistCollective(input, distCollectiveOp, reduceOp, root, groups, groupSize);
9789 return mImpl->getBuilder();
9802 return mImpl->markWeightsRefittable(name);
9814 return mImpl->unmarkWeightsRefittable(name);
9827 return mImpl->areWeightsMarkedRefittable(name);
9846 return mImpl->addSqueeze(input, axes);
9867 return mImpl->addUnsqueeze(input, axes);
9893 return mImpl->addNormalizationV2(input, scale, bias, axesMask);
9940 static constexpr int32_t kVALUE = 2;
10111 static constexpr int32_t kVALUE = 28;
10147 static constexpr uint64_t kINVALID_TACTIC_HASH = UINT64_MAX;
10180 return mImpl->serialize();
10204 return mImpl->combine(inputCache, ignoreMismatch);
10214 return mImpl->reset();
10231 int64_t
queryKeys(TimingCacheKey* keyBuffer, int64_t capacity)
const noexcept
10233 return mImpl->queryKeys(keyBuffer, capacity);
10248 TimingCacheValue
query(TimingCacheKey
const& key)
const noexcept
10250 return mImpl->query(key);
10270 bool update(TimingCacheKey
const& key, TimingCacheValue
const& value)
noexcept
10272 return mImpl->update(key, value);
10351 static constexpr int32_t kVALUE = 6;
10385 static constexpr int32_t kVALUE = 2;
10434 static constexpr int32_t kVALUE = 3;
10471 static constexpr int32_t kVALUE = 4;
10509 virtual void phaseStart(
char const* phaseName,
char const* parentPhase, int32_t nbSteps)
noexcept = 0;
10582 virtual
void setAvgTimingIterations(int32_t avgTiming) noexcept
10584 mImpl->setAvgTimingIterations(avgTiming);
10596 return mImpl->getAvgTimingIterations();
10609 mImpl->setEngineCapability(capability);
10621 return mImpl->getEngineCapability();
10638 mImpl->setFlags(builderFlags);
10650 return mImpl->getFlags();
10662 mImpl->clearFlag(builderFlag);
10674 mImpl->setFlag(builderFlag);
10686 return mImpl->getFlag(builderFlag);
10703 mImpl->setDeviceType(layer, deviceType);
10713 return mImpl->getDeviceType(layer);
10725 return mImpl->isDeviceTypeSet(layer);
10735 mImpl->resetDeviceType(layer);
10745 return mImpl->canRunOnDLA(layer);
10761 mImpl->setDLACore(dlaCore);
10771 return mImpl->getDLACore();
10782 mImpl->setDefaultDeviceType(deviceType);
10792 return mImpl->getDefaultDeviceType();
10814 return mImpl->setProfileStream(stream);
10826 return mImpl->getProfileStream();
10843 return mImpl->addOptimizationProfile(profile);
10856 return mImpl->getNbOptimizationProfiles();
10868 mImpl->setProfilingVerbosity(verbosity);
10881 return mImpl->getProfilingVerbosity();
10903 return mImpl->setTacticSources(tacticSources);
10918 return mImpl->getTacticSources();
10938 return mImpl->createTimingCache(blob, size);
10961 return mImpl->setTimingCache(cache, ignoreMismatch);
10971 return mImpl->getTimingCache();
11003 mImpl->setMemoryPoolLimit(pool, poolSize);
11022 return mImpl->getMemoryPoolLimit(pool);
11040 mImpl->setPreviewFeature(feature, enable);
11054 return mImpl->getPreviewFeature(feature);
11087 mImpl->setBuilderOptimizationLevel(level);
11099 return mImpl->getBuilderOptimizationLevel();
11116 mImpl->setHardwareCompatibilityLevel(hardwareCompatibilityLevel);
11129 return mImpl->getHardwareCompatibilityLevel();
11142 mImpl->setPluginsToSerialize(paths, nbPaths);
11155 return mImpl->getPluginToSerialize(index);
11165 return mImpl->getNbPluginsToSerialize();
11196 return mImpl->setMaxAuxStreams(nbStreams);
11206 return mImpl->getMaxAuxStreams();
11222 return mImpl->setProgressMonitor(monitor);
11232 return mImpl->getProgressMonitor();
11248 mImpl->setRuntimePlatform(runtimePlatform);
11260 return mImpl->getRuntimePlatform();
11272 mImpl->setMaxNbTactics(maxNbTactics);
11284 return mImpl->getMaxNbTactics();
11300 return mImpl->setTilingOptimizationLevel(level);
11312 return mImpl->getTilingOptimizationLevel();
11328 return mImpl->setL2LimitForTiling(size);
11340 return mImpl->getL2LimitForTiling();
11354 return mImpl->setRemoteAutoTuningConfig(config);
11364 return mImpl->getRemoteAutoTuningConfig();
11419 static constexpr int32_t kVALUE = 3;
11441 int32_t getMaxDLABatchSize() const noexcept
11443 return mImpl->getMaxDLABatchSize();
11451 return mImpl->getNbDLACores();
11469 mImpl->setGpuAllocator(allocator);
11483 return mImpl->createBuilderConfig();
11509 return mImpl->createNetworkV2(flags);
11524 return mImpl->createOptimizationProfile();
11543 mImpl->setErrorRecorder(recorder);
11558 return mImpl->getErrorRecorder();
11585 return mImpl->buildSerializedNetwork(network, config);
11607 return mImpl->buildSerializedNetworkToStream(network, config, writer);
11631 return mImpl->buildSerializedNetworkWithKernelText(network, config, kernelText);
11651 return mImpl->buildEngineWithConfig(network, config);
11677 return mImpl->isNetworkSupported(network, config);
11687 return mImpl->getLogger();
11703 return mImpl->setMaxThreads(maxThreads);
11717 return mImpl->getMaxThreads();
11727 return mImpl->getPluginRegistry();
11742extern "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:222
static constexpr int32_t MAX_DIMS
The maximum rank (number of dimensions) supported for a tensor.
Definition: NvInferRuntimeBase.h:225
An Activation layer in a network definition.
Definition: NvInfer.h: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:4942
void setMessage(char const *message) noexcept
Set the message to print if the assertion fails.
Definition: NvInfer.h:4952
char const * getMessage() const noexcept
Return the assertion message.
Definition: NvInfer.h:4962
virtual ~IAssertionLayer() noexcept=0
This is a base class for Attention boundary layers.
Definition: NvInfer.h:6821
IAttention * getAttention() const noexcept
Get a pointer to the IAttention associated with this boundary layer.
Definition: NvInfer.h:6826
virtual ~IAttentionBoundaryLayer() noexcept=0
Helper for constructing an attention that consumes query, key and value tensors.
Definition: NvInfer.h:6958
ITensor * getMask() noexcept
Get the optional mask in attention.
Definition: NvInfer.h:7008
bool setMetadata(char const *metadata) noexcept
Set the metadata for IAttention.
Definition: NvInfer.h:7258
TRT_NODISCARD bool setQueryLengths(ITensor *lengths) noexcept
Set the query lengths tensor.
Definition: NvInfer.h:7383
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:7084
bool setName(char const *name) noexcept
Set the name of the attention.
Definition: NvInfer.h:7174
bool getDecomposable() const noexcept
Get whether the attention can be decomposed to use multiple kernels if no fused kernel support found.
Definition: NvInfer.h:7097
ITensor * getInput(int32_t index) const noexcept
Get the IAttention input corresponding to the given index.
Definition: NvInfer.h:7137
CausalMaskKind getCausalKind() const noexcept
Get the causal mask alignment orientation for the attention.
Definition: NvInfer.h:7072
ITensor * getOutput(int32_t index) const noexcept
Get the IAttention output corresponding to the given index. IAttention has only one output.
Definition: NvInfer.h:7157
int32_t getNbOutputs() const noexcept
Get the number of outputs of a layer. IAttention has one output.
Definition: NvInfer.h:7145
bool setNbRanks(int32_t nbRanks) noexcept
Set the number of ranks for multi-device attention execution.
Definition: NvInfer.h:7287
int32_t getNbInputs() const noexcept
Get the number of inputs of IAttention. IAttention has three inputs.
Definition: NvInfer.h:7125
TRT_NODISCARD bool setKeyValueLengths(ITensor *lengths) noexcept
Set the key-value lengths tensor.
Definition: NvInfer.h:7422
TRT_NODISCARD ITensor * getKeyValueLengths() const noexcept
Get the key-value lengths tensor.
Definition: NvInfer.h:7434
bool setNormalizationOperation(AttentionNormalizationOp op) noexcept
Set the normalization operation for the attention.
Definition: NvInfer.h:6967
TRT_NODISCARD AttentionIOForm getKeyValueForm() const noexcept
Get the key-value form.
Definition: NvInfer.h:7359
char const * getName() const noexcept
Return the name of the attention.
Definition: NvInfer.h:7186
bool setNormalizationQuantizeToType(DataType type) noexcept
Set the datatype the attention normalization is quantized to.
Definition: NvInfer.h:7226
int32_t getNbRanks() const noexcept
Get the number of ranks for multi-device execution.
Definition: NvInfer.h:7299
TRT_DEPRECATED bool getCausal() const noexcept
Get whether the attention will run a causal inference.
Definition: NvInfer.h:7040
AttentionNormalizationOp getNormalizationOperation() const noexcept
Get the normalization operation for the attention.
Definition: NvInfer.h:6979
bool setNormalizationQuantizeScale(ITensor &tensor) noexcept
Set the quantization scale for the attention normalization output.
Definition: NvInfer.h:7202
bool setCausalKind(CausalMaskKind kind) noexcept
Set the causal mask alignment orientation for the attention.
Definition: NvInfer.h:7060
TRT_NODISCARD AttentionIOForm getQueryForm() const noexcept
Get the query form.
Definition: NvInfer.h:7329
char const * getMetadata() const noexcept
Get the metadata of IAttention.
Definition: NvInfer.h:7271
DataType getNormalizationQuantizeToType() const noexcept
Get the datatype the attention normalization is quantized to.
Definition: NvInfer.h:7238
TRT_NODISCARD bool setQueryForm(AttentionIOForm form) noexcept
Set the query form.
Definition: NvInfer.h:7316
virtual ~IAttention() noexcept=0
ITensor * getNormalizationQuantizeScale() const noexcept
Get the quantization scale for the attention normalization output.
Definition: NvInfer.h:7213
bool setInput(int32_t index, ITensor &input) noexcept
Append or replace an input of this layer with a specific tensor.
Definition: NvInfer.h:7116
TRT_NODISCARD ITensor * getQueryLengths() const noexcept
Get the query lengths tensor.
Definition: NvInfer.h:7395
bool setMask(ITensor &mask) noexcept
Set whether a mask will be used for the normalization operation.
Definition: NvInfer.h:6996
TRT_NODISCARD bool setKeyValueForm(AttentionIOForm form) noexcept
Set the key-value form.
Definition: NvInfer.h:7346
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:7026
apiv::VAttention * mImpl
Definition: NvInfer.h:7440
This layer represents an output of an IAttention.
Definition: NvInfer.h:6888
virtual ~IAttentionOutputLayer() noexcept=0
Holds properties for configuring a builder to produce an engine.
Definition: NvInfer.h:10570
void setMemoryPoolLimit(MemoryPoolType pool, std::size_t poolSize) noexcept
Set the memory size for the memory pool.
Definition: NvInfer.h:11001
nvinfer1::ITimingCache * createTimingCache(void const *blob, std::size_t size) const noexcept
Create timing cache.
Definition: NvInfer.h:10936
bool setMaxAuxStreams(int32_t nbStreams) noexcept
Set the maximum number of auxiliary streams that TRT is allowed to use.
Definition: NvInfer.h:11194
void setPreviewFeature(PreviewFeature feature, bool enable) noexcept
Enable or disable a specific preview feature.
Definition: NvInfer.h:11038
bool getPreviewFeature(PreviewFeature feature) const noexcept
Get status of preview feature.
Definition: NvInfer.h:11052
int32_t getBuilderOptimizationLevel() noexcept
Get builder optimization level.
Definition: NvInfer.h:11097
bool setTacticSources(TacticSources tacticSources) noexcept
Set tactic sources.
Definition: NvInfer.h:10901
void setPluginsToSerialize(char const *const *paths, int32_t nbPaths) noexcept
Set the plugin libraries to be serialized with version-compatible engines.
Definition: NvInfer.h:11140
bool setTilingOptimizationLevel(TilingOptimizationLevel level) noexcept
Set the Tiling optimization level.
Definition: NvInfer.h:11298
bool setL2LimitForTiling(int64_t size) noexcept
Set the L2 cache usage limit for Tiling optimization.
Definition: NvInfer.h:11326
std::size_t getMemoryPoolLimit(MemoryPoolType pool) const noexcept
Get the memory size limit of the memory pool.
Definition: NvInfer.h:11020
int32_t getDLACore() const noexcept
Get the DLA core that the engine executes on.
Definition: NvInfer.h:10769
int32_t getNbPluginsToSerialize() const noexcept
Get the number of plugin library paths to be serialized with version-compatible engines.
Definition: NvInfer.h:11163
void setDeviceType(ILayer const *layer, DeviceType deviceType) noexcept
Set the device that this layer must execute on.
Definition: NvInfer.h:10701
void setEngineCapability(EngineCapability capability) noexcept
Configure the builder to target specified EngineCapability flow.
Definition: NvInfer.h:10607
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:11204
bool getFlag(BuilderFlag builderFlag) const noexcept
Returns true if the build mode flag is set.
Definition: NvInfer.h:10684
void setMaxNbTactics(int32_t maxNbTactics) noexcept
Set the maximum number of tactics to time when there is a choice of tactics.
Definition: NvInfer.h:11270
int64_t getL2LimitForTiling() const noexcept
Get the L2 cache usage limit for tiling optimization.
Definition: NvInfer.h:11338
bool setRemoteAutoTuningConfig(char const *config) noexcept
Set a config string for remote auto tuning.
Definition: NvInfer.h:11352
void setProgressMonitor(IProgressMonitor *monitor) noexcept
Sets the progress monitor for building a network.
Definition: NvInfer.h:11220
void setProfilingVerbosity(ProfilingVerbosity verbosity) noexcept
Set verbosity level of layer information exposed in NVTX annotations and IEngineInspector.
Definition: NvInfer.h:10866
int32_t getNbOptimizationProfiles() const noexcept
Get number of optimization profiles.
Definition: NvInfer.h:10854
nvinfer1::ITimingCache const * getTimingCache() const noexcept
Get the pointer to the timing cache from current IBuilderConfig.
Definition: NvInfer.h:10969
void reset() noexcept
Resets the builder configuration to defaults.
Definition: NvInfer.h:10800
bool setTimingCache(ITimingCache const &cache, bool ignoreMismatch) noexcept
Attach a timing cache to IBuilderConfig.
Definition: NvInfer.h:10959
char const * getPluginToSerialize(int32_t index) const noexcept
Get the plugin library path to be serialized with version-compatible engines.
Definition: NvInfer.h:11153
EngineCapability getEngineCapability() const noexcept
Query EngineCapability flow configured for the builder.
Definition: NvInfer.h:10619
RuntimePlatform getRuntimePlatform() const noexcept
Get the target platform for runtime execution.
Definition: NvInfer.h:11258
DeviceType getDefaultDeviceType() const noexcept
Get the default DeviceType which was set by setDefaultDeviceType.
Definition: NvInfer.h:10790
void setRuntimePlatform(RuntimePlatform runtimePlatform) noexcept
Set the target platform for runtime execution.
Definition: NvInfer.h:11246
int32_t getMaxNbTactics() const noexcept
Query the maximum number of tactics timed when there is a choice.
Definition: NvInfer.h:11282
BuilderFlags getFlags() const noexcept
Get the build mode flags for this builder config. Defaults to 0.
Definition: NvInfer.h:10648
void setFlags(BuilderFlags builderFlags) noexcept
Set the build mode flags to turn on builder options for this network.
Definition: NvInfer.h:10636
TacticSources getTacticSources() const noexcept
Get tactic sources.
Definition: NvInfer.h:10916
void resetDeviceType(ILayer const *layer) noexcept
reset the DeviceType for this layer
Definition: NvInfer.h:10733
void setDLACore(int32_t dlaCore) noexcept
Sets the DLA core used by the network. Defaults to -1.
Definition: NvInfer.h:10759
HardwareCompatibilityLevel getHardwareCompatibilityLevel() const noexcept
Get the hardware compatibility level.
Definition: NvInfer.h:11127
char const * getRemoteAutoTuningConfig() const noexcept
Get a config string for remote auto tuning.
Definition: NvInfer.h:11362
void clearFlag(BuilderFlag builderFlag) noexcept
clear a single build mode flag.
Definition: NvInfer.h:10660
int32_t addOptimizationProfile(IOptimizationProfile const *profile) noexcept
Add an optimization profile.
Definition: NvInfer.h:10841
IProgressMonitor * getProgressMonitor() const noexcept
Definition: NvInfer.h:11230
apiv::VBuilderConfig * mImpl
Definition: NvInfer.h:11368
int32_t getAvgTimingIterations() const noexcept
Query the number of averaging iterations.
Definition: NvInfer.h:10594
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:10780
void setFlag(BuilderFlag builderFlag) noexcept
Set a single build mode flag.
Definition: NvInfer.h:10672
DeviceType getDeviceType(ILayer const *layer) const noexcept
Get the device that this layer executes on.
Definition: NvInfer.h:10711
bool canRunOnDLA(ILayer const *layer) const noexcept
Checks if a layer can run on DLA.
Definition: NvInfer.h:10743
cudaStream_t getProfileStream() const noexcept
Get the CUDA stream that is used to profile this network.
Definition: NvInfer.h:10824
void setHardwareCompatibilityLevel(HardwareCompatibilityLevel hardwareCompatibilityLevel) noexcept
Set the hardware compatibility level.
Definition: NvInfer.h:11114
TilingOptimizationLevel getTilingOptimizationLevel() const noexcept
Get the Tiling optimization level.
Definition: NvInfer.h:11310
ProfilingVerbosity getProfilingVerbosity() const noexcept
Get verbosity level of layer information exposed in NVTX annotations and IEngineInspector.
Definition: NvInfer.h:10879
bool isDeviceTypeSet(ILayer const *layer) const noexcept
whether the DeviceType has been explicitly set for this layer
Definition: NvInfer.h:10723
void setBuilderOptimizationLevel(int32_t level) noexcept
Set builder optimization level.
Definition: NvInfer.h:11085
void setProfileStream(const cudaStream_t stream) noexcept
Set the CUDA stream that is used to profile this network.
Definition: NvInfer.h:10812
Builds an engine from a network definition.
Definition: NvInfer.h:11430
int32_t getNbDLACores() const noexcept
Return the number of DLA engines available to this builder.
Definition: NvInfer.h:11449
virtual ~IBuilder() noexcept=0
IErrorRecorder * getErrorRecorder() const noexcept
get the ErrorRecorder assigned to this interface.
Definition: NvInfer.h:11556
apiv::VBuilder * mImpl
Definition: NvInfer.h:11731
ILogger * getLogger() const noexcept
get the logger with which the builder was created
Definition: NvInfer.h:11685
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:11675
int32_t getMaxThreads() const noexcept
get the maximum number of threads that can be used by the builder.
Definition: NvInfer.h:11715
IPluginRegistry & getPluginRegistry() noexcept
get the local plugin registry that can be used by the builder.
Definition: NvInfer.h:11725
nvinfer1::IOptimizationProfile * createOptimizationProfile() noexcept
Create a new optimization profile.
Definition: NvInfer.h:11522
void setGpuAllocator(IGpuAllocator *allocator) noexcept
Set the GPU allocator.
Definition: NvInfer.h:11467
nvinfer1::INetworkDefinition * createNetworkV2(NetworkDefinitionCreationFlags flags) noexcept
Create a network definition object.
Definition: NvInfer.h:11507
nvinfer1::IBuilderConfig * createBuilderConfig() noexcept
Create a builder configuration object.
Definition: NvInfer.h:11481
void reset() noexcept
Resets the builder state to default values.
Definition: NvInfer.h:11564
bool setMaxThreads(int32_t maxThreads) noexcept
Set the maximum number of threads.
Definition: NvInfer.h:11701
void setErrorRecorder(IErrorRecorder *recorder) noexcept
Set the ErrorRecorder for this interface.
Definition: NvInfer.h:11541
nvinfer1::IHostMemory * buildSerializedNetwork(INetworkDefinition &network, IBuilderConfig &config) noexcept
Builds and serializes a network for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:11583
bool buildSerializedNetworkToStream(INetworkDefinition &network, IBuilderConfig &config, IStreamWriter &writer) noexcept
Builds and serializes a network into stream for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:11604
nvinfer1::ICudaEngine * buildEngineWithConfig(INetworkDefinition &network, IBuilderConfig &config) noexcept
Builds a network for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:11649
nvinfer1::IHostMemory * buildSerializedNetwork(INetworkDefinition &network, IBuilderConfig &config, IHostMemory *&kernelText) noexcept
Extended form of buildSerializedNetwork that optionally permits getting the kernelText.
Definition: NvInfer.h:11628
A cast layer in a network.
Definition: NvInfer.h:3783
apiv::VCastLayer * mImpl
Definition: NvInfer.h:3809
DataType getToType() const noexcept
Return cast layer output type.
Definition: NvInfer.h:3803
void setToType(DataType toType) noexcept
Set cast layer output type.
Definition: NvInfer.h:3792
virtual ~ICastLayer() noexcept=0
A concatenation layer in a network definition.
Definition: NvInfer.h:1896
void setAxis(int32_t axis) noexcept
Set the axis along which concatenation occurs.
Definition: NvInfer.h:1909
int32_t getAxis() const noexcept
Get the axis along which concatenation occurs.
Definition: NvInfer.h:1919
virtual ~IConcatenationLayer() noexcept=0
This layer represents a condition input to an IIfConditional.
Definition: NvInfer.h:4446
virtual ~IConditionLayer() noexcept=0
Layer that represents a constant value.
Definition: NvInfer.h:3824
void setWeights(Weights weights) noexcept
Set the weights for the layer.
Definition: NvInfer.h:3834
Weights getWeights() const noexcept
Get the weights for the layer.
Definition: NvInfer.h:3844
virtual ~IConstantLayer() noexcept=0
void setDimensions(Dims const &dimensions) noexcept
Set the dimensions for the layer.
Definition: NvInfer.h:3856
apiv::VConstantLayer * mImpl
Definition: NvInfer.h:3874
Dims getDimensions() const noexcept
Get the dimensions for the layer.
Definition: NvInfer.h:3868
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:3013
Layer that represents a cumulative operation across a tensor.
Definition: NvInfer.h:6642
bool setOperation(CumulativeOperation op) noexcept
Set the cumulative operation for the layer.
Definition: NvInfer.h:6653
void setReverse(bool reverse) noexcept
Specify whether the cumulative operation should be applied backward.
Definition: NvInfer.h:6701
apiv::VCumulativeLayer * mImpl
Definition: NvInfer.h:6719
virtual ~ICumulativeLayer() noexcept=0
bool getExclusive() const noexcept
Get whether it is exclusive accumulation or inclusive accumulation.
Definition: NvInfer.h:6689
bool getReverse() const noexcept
Get the boolean that specifies whether the cumulative operation should be applied backward.
Definition: NvInfer.h:6713
void setExclusive(bool exclusive) noexcept
Set whether it is an exclusive accumulation or inclusive accumulation.
Definition: NvInfer.h:6677
CumulativeOperation getOperation() const noexcept
Get the cumulative operation for the layer.
Definition: NvInfer.h:6665
A deconvolution layer in a network definition.
Definition: NvInfer.h:1939
void setBiasWeights(Weights weights) noexcept
Set the bias weights for the deconvolution.
Definition: NvInfer.h:2027
int64_t getNbGroups() const noexcept
Get the number of groups for a deconvolution.
Definition: NvInfer.h:1988
Weights getKernelWeights() const noexcept
Get the kernel weights for the deconvolution.
Definition: NvInfer.h:2012
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding of the deconvolution.
Definition: NvInfer.h:2054
Dims getStrideNd() const noexcept
Get the multi-dimension stride of the deconvolution.
Definition: NvInfer.h:2169
Dims getDilationNd() const noexcept
Get the multi-dimension dilation of the deconvolution.
Definition: NvInfer.h:2235
virtual ~IDeconvolutionLayer() noexcept=0
Weights getBiasWeights() const noexcept
Get the bias weights for the deconvolution.
Definition: NvInfer.h:2037
void setKernelWeights(Weights weights) noexcept
Set the kernel weights for the deconvolution.
Definition: NvInfer.h:2002
int64_t getNbOutputMaps() const noexcept
Get the number of output feature maps for the deconvolution.
Definition: NvInfer.h:1958
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride of the deconvolution.
Definition: NvInfer.h:2159
Dims getPostPadding() const noexcept
Get the padding.
Definition: NvInfer.h:2091
Dims getKernelSizeNd() const noexcept
Get the multi-dimension kernel size of the deconvolution.
Definition: NvInfer.h:2142
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding of the deconvolution.
Definition: NvInfer.h:2081
void setKernelSizeNd(Dims const &kernelSize) noexcept
Set the multi-dimension kernel size of the deconvolution.
Definition: NvInfer.h:2132
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding of the deconvolution.
Definition: NvInfer.h:2187
void setNbOutputMaps(int64_t nbOutputMaps) noexcept
Set the number of output feature maps for the deconvolution.
Definition: NvInfer.h:1948
Dims getPaddingNd() const noexcept
Get the multi-dimension padding of the deconvolution.
Definition: NvInfer.h:2199
void setDilationNd(Dims const &dilation) noexcept
Set the multi-dimension dilation of the deconvolution.
Definition: NvInfer.h:2225
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:2105
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups for a deconvolution.
Definition: NvInfer.h:1978
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:2064
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:2117
A Dequantize layer in a network definition.
Definition: NvInfer.h:5561
TRT_NODISCARD Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5610
void setToType(DataType toType) noexcept
Set the Dequantize layer output type.
Definition: NvInfer.h:5626
int32_t getAxis() const noexcept
Get the quantization axis.
Definition: NvInfer.h:5571
bool setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5599
virtual ~IDequantizeLayer() noexcept=0
DataType getToType() const noexcept
Return the Dequantize layer output type.
Definition: NvInfer.h:5638
void setAxis(int32_t axis) noexcept
Set the quantization axis.
Definition: NvInfer.h:5582
Definition: NvInfer.h:8138
virtual ~IDistCollectiveLayer() noexcept=0
A network layer to perform dynamic quantization.
Definition: NvInfer.h:5668
virtual ~IDynamicQuantizeLayer() noexcept=0
DataType getScaleType() const noexcept
Return the scale factors data type.
Definition: NvInfer.h:5734
TRT_DEPRECATED void setAxis(int32_t axis) noexcept
Set the axis along which block quantization occurs.
Definition: NvInfer.h:5747
TRT_DEPRECATED void setBlockSize(int32_t size) noexcept
Set the size of the quantization block.
Definition: NvInfer.h:5770
Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5805
void setScaleType(DataType scaleType) noexcept
Set the data type of the scale factors used to quantize the data.
Definition: NvInfer.h:5721
DataType getToType() const noexcept
Return DynamicQuantizeLayer's quantized output type.
Definition: NvInfer.h:5708
TRT_DEPRECATED int32_t getAxis() const noexcept
Get the axis along which blocking occurs.
Definition: NvInfer.h:5757
void setToType(DataType toType) noexcept
Set DynamicQuantizeLayer's quantized output type.
Definition: NvInfer.h:5695
void setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5793
TRT_DEPRECATED int32_t getBlockSize() const noexcept
Get the size of the quantization block.
Definition: NvInfer.h:5780
An Einsum layer in a network.
Definition: NvInfer.h:5852
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:5863
virtual ~IEinsumLayer() noexcept=0
char const * getEquation() const noexcept
Return the equation.
Definition: NvInfer.h:5873
A elementwise layer in a network definition.
Definition: NvInfer.h:2308
apiv::VElementWiseLayer * mImpl
Definition: NvInfer.h:2337
ElementWiseOperation getOperation() const noexcept
Get the binary operation for the layer.
Definition: NvInfer.h:2331
void setOperation(ElementWiseOperation op) noexcept
Set the binary operation for the layer.
Definition: NvInfer.h:2319
virtual ~IElementWiseLayer() noexcept=0
Generate a tensor according to a specified mode.
Definition: NvInfer.h:5058
bool isAlphaBetaInt64() const noexcept
Return true if alpha/beta have type int64, false if they have type double.
Definition: NvInfer.h:5290
FillOperation getOperation() const noexcept
Get the fill operation for the layer.
Definition: NvInfer.h:5104
void setOperation(FillOperation op) noexcept
Set the fill operation for the layer.
Definition: NvInfer.h:5094
DataType getToType() const noexcept
Get the fill layer output type.
Definition: NvInfer.h:5320
void setAlphaInt64(int64_t alpha) noexcept
Set the alpha parameter with int64 datatype.
Definition: NvInfer.h:5233
void setBetaInt64(int64_t beta) noexcept
Set the beta parameter with int64 datatype.
Definition: NvInfer.h:5267
virtual ~IFillLayer() noexcept=0
void setBeta(double beta) noexcept
Set the beta parameter.
Definition: NvInfer.h:5157
int64_t getAlphaInt64() const noexcept
Get the value of alpha parameter with int64 datatype.
Definition: NvInfer.h:5248
int64_t getBetaInt64() const noexcept
Get the value of beta parameter with int64 datatype.
Definition: NvInfer.h:5282
double getAlpha() const noexcept
Get the value of alpha parameter.
Definition: NvInfer.h:5138
void setDimensions(Dims const &dimensions) noexcept
Set the output tensor's dimensions.
Definition: NvInfer.h:5069
void setAlpha(double alpha) noexcept
Set the alpha parameter.
Definition: NvInfer.h:5123
void setToType(DataType toType) noexcept
Set the fill layer output type.
Definition: NvInfer.h:5308
Dims getDimensions() const noexcept
Get the output tensor's dimensions.
Definition: NvInfer.h:5084
double getBeta() const noexcept
Get the value of beta parameter.
Definition: NvInfer.h:5172
A Gather layer in a network definition. Supports several kinds of gathering.
Definition: NvInfer.h:2443
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:2454
void setNbElementWiseDims(int32_t elementWiseDims) noexcept
Set the number of leading dimensions of indices tensor to be handled elementwise.
Definition: NvInfer.h:2489
apiv::VGatherLayer * mImpl
Definition: NvInfer.h:2525
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:2499
void setMode(GatherMode mode) noexcept
Set the gather mode.
Definition: NvInfer.h:2509
int32_t getGatherAxis() const noexcept
Get the axis to gather on.
Definition: NvInfer.h:2466
GatherMode getMode() const noexcept
Get the gather mode.
Definition: NvInfer.h:2519
A GridSample layer in a network definition.
Definition: NvInfer.h:6081
void setInterpolationMode(InterpolationMode mode) noexcept
Set the grid sample interpolation mode.
Definition: NvInfer.h:6088
bool setSampleMode(SampleMode mode) noexcept
Set the sample mode.
Definition: NvInfer.h:6134
void setAlignCorners(bool alignCorners) noexcept
Set the align corners mode.
Definition: NvInfer.h:6110
apiv::VGridSampleLayer * mImpl
Definition: NvInfer.h:6152
SampleMode getSampleMode() const noexcept
Get the sample mode.
Definition: NvInfer.h:6146
virtual ~IGridSampleLayer() noexcept=0
InterpolationMode getInterpolationMode() const noexcept
Get the grid sample interpolation mode.
Definition: NvInfer.h:6100
bool getAlignCorners() const noexcept
Get the align corners mode.
Definition: NvInfer.h:6122
Class to handle library allocated memory that is accessible to the user.
Definition: NvInferRuntime.h:139
A layer that represents the identity function.
Definition: NvInfer.h:3768
virtual ~IIdentityLayer() noexcept=0
apiv::VIdentityLayer * mImpl
Definition: NvInfer.h:3770
This is a base class for Conditional boundary layers.
Definition: NvInfer.h:4423
IIfConditional * getConditional() const noexcept
Get a pointer to the IIfConditional associated with this boundary layer.
Definition: NvInfer.h:4428
virtual ~IIfConditionalBoundaryLayer() noexcept=0
Helper for constructing conditionally-executed subgraphs.
Definition: NvInfer.h:4514
IIfConditionalInputLayer * addInput(ITensor &input) noexcept
Add an If-conditional input.
Definition: NvInfer.h:4555
char const * getName() const noexcept
Return the name of the conditional.
Definition: NvInfer.h:4580
IConditionLayer * setCondition(ITensor &condition) noexcept
Set the condition tensor for this If-Conditional construct.
Definition: NvInfer.h:4525
virtual ~IIfConditional() noexcept=0
IIfConditionalOutputLayer * addOutput(ITensor &trueSubgraphOutput, ITensor &falseSubgraphOutput) noexcept
Add an If-conditional output.
Definition: NvInfer.h:4543
void setName(char const *name) noexcept
Set the name of the conditional.
Definition: NvInfer.h:4570
This layer represents an output of an IIfConditional.
Definition: NvInfer.h:4465
virtual ~IIfConditionalOutputLayer() noexcept=0
A layer to do iterations.
Definition: NvInfer.h:4753
void setReverse(bool reverse) noexcept
Set iteration order to be reverse.
Definition: NvInfer.h:4780
virtual ~IIteratorLayer() noexcept=0
bool getReverse() const noexcept
Check if the iteration order is reverse.
Definition: NvInfer.h:4790
int32_t getAxis() const noexcept
Get axis being iterated over.
Definition: NvInfer.h:4766
void setAxis(int32_t axis) noexcept
Set axis to iterate over.
Definition: NvInfer.h:4758
Layer that represents a KVCacheUpdate operation.
Definition: NvInfer.h:7562
bool setCacheMode(KVCacheMode cacheMode) noexcept
Set the mode of the KVCacheUpdate layer.
Definition: NvInfer.h:7587
TRT_NODISCARD ITensor * getUpdateLengths() const noexcept
Get the update lengths tensor.
Definition: NvInfer.h:7662
virtual ~IKVCacheUpdateLayer() noexcept=0
TRT_NODISCARD AttentionIOForm getUpdateForm() const noexcept
Get the update form.
Definition: NvInfer.h:7628
TRT_NODISCARD bool setUpdateLengths(ITensor *lengths) noexcept
Set the update lengths tensor.
Definition: NvInfer.h:7650
TRT_NODISCARD bool setUpdateForm(AttentionIOForm form) noexcept
Set the update form.
Definition: NvInfer.h:7615
KVCacheMode getCacheMode() const noexcept
Get the mode of the KVCacheUpdate layer.
Definition: NvInfer.h:7597
apiv::VKVCacheUpdateLayer * mImpl
Definition: NvInfer.h:7668
A LRN layer in a network definition.
Definition: NvInfer.h:1545
int64_t getWindowSize() const noexcept
Get the LRN window size.
Definition: NvInfer.h:1566
virtual ~ILRNLayer() noexcept=0
float getAlpha() const noexcept
Get the LRN alpha value.
Definition: NvInfer.h:1588
void setWindowSize(int64_t windowSize) noexcept
Set the LRN window size.
Definition: NvInfer.h:1556
void setK(float k) noexcept
Set the LRN K value.
Definition: NvInfer.h:1622
void setAlpha(float alpha) noexcept
Set the LRN alpha value.
Definition: NvInfer.h:1578
void setBeta(float beta) noexcept
Set the LRN beta value.
Definition: NvInfer.h:1600
float getBeta() const noexcept
Get the LRN beta value.
Definition: NvInfer.h:1610
float getK() const noexcept
Get the LRN K value.
Definition: NvInfer.h:1632
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:4398
virtual ~ILoopBoundaryLayer() noexcept=0
ILoop * getLoop() const noexcept
Get a pointer to ILoop associated with this boundary layer.
Definition: NvInfer.h:4403
Helper for creating a recurrent subgraph.
Definition: NvInfer.h:4813
void setName(char const *name) noexcept
Set the name of the loop.
Definition: NvInfer.h:4883
ITripLimitLayer * addTripLimit(ITensor &tensor, TripLimit limit) noexcept
Add a trip-count limiter, based on the given tensor.
Definition: NvInfer.h:4842
IIteratorLayer * addIterator(ITensor &tensor, int32_t axis=0, bool reverse=false) noexcept
Return layer that subscripts tensor by loop iteration.
Definition: NvInfer.h:4855
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:4868
virtual ~ILoop() noexcept=0
char const * getName() const noexcept
Return the name of the loop.
Definition: NvInfer.h:4893
IRecurrenceLayer * addRecurrence(ITensor &initialValue) noexcept
Create a recurrence layer for this loop with initialValue as its first input.
Definition: NvInfer.h:4821
An ILoopOutputLayer is the sole way to get output from a loop.
Definition: NvInfer.h:4649
int32_t getAxis() const noexcept
Get axis being concatenated over.
Definition: NvInfer.h:4679
LoopOutput getLoopOutput() const noexcept
Get which kind a loop output has.
Definition: NvInfer.h:4654
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:4671
Layer that represents a Matrix Multiplication.
Definition: NvInfer.h:3609
apiv::VMatrixMultiplyLayer * mImpl
Definition: NvInfer.h:3637
virtual ~IMatrixMultiplyLayer() noexcept=0
MatrixOperation getOperation(int32_t index) const noexcept
Get the operation for an input tensor.
Definition: NvInfer.h:3631
void setOperation(int32_t index, MatrixOperation op) noexcept
Set the operation for an input tensor.
Definition: NvInfer.h:3619
A MoE layer in a network definition. Mixture of Experts (MoE) is a collection of experts with each ex...
Definition: NvInfer.h:7813
void setSwigluParamLimit(float limit) noexcept
Set the SwiGLU parameter limit.
Definition: NvInfer.h:8035
void setDynQOutputScaleType(DataType type) noexcept
Set the dynamic quantization output scale type.
Definition: NvInfer.h:7988
MoEActType getActivationType() const noexcept
Get the activation type for the MoE layer.
Definition: NvInfer.h:7862
void setQuantizationToType(DataType type) noexcept
Set the data type the mul output is quantized to.
Definition: NvInfer.h:7936
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:7921
void setQuantizationStatic(ITensor &fcDownActivationScale, DataType dataType) noexcept
Configure static quantization after the mul op.
Definition: NvInfer.h:7888
float getSwigluParamLimit() const noexcept
Get the SwiGLU parameter limit.
Definition: NvInfer.h:8047
DataType getQuantizationToType() const noexcept
Get the data type the mul in MoE layer is quantized to.
Definition: NvInfer.h:7948
DataType getDynQOutputScaleType() const noexcept
Get the dynamic quantization output scale type.
Definition: NvInfer.h:8000
virtual ~IMoELayer() noexcept=0
void setActivationType(MoEActType activationType) noexcept
Set the activation type for the MoE layer.
Definition: NvInfer.h:7850
Dims getQuantizationBlockShape() const noexcept
Get the block shape for the quantization of the Mul output.
Definition: NvInfer.h:7976
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:7826
float getSwigluParamBeta() const noexcept
Get the SwiGLU parameter beta.
Definition: NvInfer.h:8099
void setSwigluParamBeta(float beta) noexcept
Set the SwiGLU parameter beta.
Definition: NvInfer.h:8087
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:7838
void setSwigluParams(float limit, float alpha, float beta) noexcept
Set the SwiGLU parameters.
Definition: NvInfer.h:8021
void setQuantizationBlockShape(Dims const &blockShape) noexcept
Set the block shape for the quantization of the Mul output.
Definition: NvInfer.h:7964
void setInput(int32_t index, ITensor &tensor) noexcept
Set the input of the MoE layer.
Definition: NvInfer.h:8116
float getSwigluParamAlpha() const noexcept
Get the SwiGLU parameter alpha.
Definition: NvInfer.h:8073
void setSwigluParamAlpha(float alpha) noexcept
Set the SwiGLU parameter alpha.
Definition: NvInfer.h:8061
A non-maximum suppression layer in a network definition.
Definition: NvInfer.h:6235
void setTopKBoxLimit(int32_t limit) noexcept
Set the TopK box limit parameter for the layer.
Definition: NvInfer.h:6272
void setBoundingBoxFormat(BoundingBoxFormat fmt) noexcept
Set the bounding box format parameter for the layer.
Definition: NvInfer.h:6246
BoundingBoxFormat getBoundingBoxFormat() const noexcept
Get the bounding box format parameter for the layer.
Definition: NvInfer.h:6258
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:6317
apiv::VNMSLayer * mImpl
Definition: NvInfer.h:6335
int32_t getTopKBoxLimit() const noexcept
Get the TopK box limit parameter for the layer.
Definition: NvInfer.h:6282
DataType getIndicesType() const noexcept
Return the NMS layer indices type.
Definition: NvInfer.h:6329
virtual ~INMSLayer() noexcept=0
A network definition for input to the builder.
Definition: NvInfer.h:8164
IConcatenationLayer * addConcatenation(ITensor *const *inputs, int32_t nbInputs) noexcept
Add a concatenation layer to the network.
Definition: NvInfer.h:8392
IShuffleLayer * addShuffle(ITensor &input) noexcept
Add a shuffle layer to the network.
Definition: NvInfer.h:8455
void setName(char const *name) noexcept
Sets the name of the network.
Definition: NvInfer.h:8921
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:8651
bool markDebug(ITensor &tensor) noexcept
Mark a tensor as a debug tensor.
Definition: NvInfer.h:8235
ILRNLayer * addLRN(ITensor &input, int64_t window, float alpha, float beta, float k) noexcept
Add a LRN layer to the network.
Definition: NvInfer.h:8336
ICumulativeLayer * addCumulative(ITensor &input, ITensor &axis, CumulativeOperation operation, bool exclusive, bool reverse) noexcept
Add a cumulative layer to the network.
Definition: NvInfer.h:9602
IAssertionLayer * addAssertion(ITensor &condition, char const *message) noexcept
Add an assertion layer to the network.
Definition: NvInfer.h:9237
TRT_DEPRECATED INonZeroLayer * addNonZero(ITensor &input) noexcept
Add a nonzero layer to the network.
Definition: NvInfer.h:8742
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:9056
ICastLayer * addCast(ITensor &input, DataType toType) noexcept
Add a cast layer.
Definition: NvInfer.h:8811
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:9135
char const * getName() const noexcept
Returns the name associated with the network.
Definition: NvInfer.h:8935
IParametricReLULayer * addParametricReLU(ITensor &input, ITensor &slope) noexcept
Add a parametric ReLU layer to the network.
Definition: NvInfer.h:9034
ITensor * getOutput(int32_t index) const noexcept
Get the output tensor specified by the given index.
Definition: NvInfer.h:8556
ITensor * getInput(int32_t index) const noexcept
Get the input tensor specified by the given index.
Definition: NvInfer.h:8526
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:8618
IDequantizeLayer * addDequantize(ITensor &input, ITensor &scale, DataType outputType) noexcept
Add a dequantization layer to the network.
Definition: NvInfer.h:9360
bool unmarkOutputForShapes(ITensor &tensor) noexcept
Undo markOutputForShapes.
Definition: NvInfer.h:9016
IFillLayer * addFill(Dims const &dimensions, FillOperation op, DataType outputType) noexcept
Add a fill layer to the network.
Definition: NvInfer.h:9263
ILoop * addLoop() noexcept
Add a loop to the network.
Definition: NvInfer.h:9166
bool markUnfusedTensorsAsDebugTensors() noexcept
Mark unfused tensors as debug tensors.
Definition: NvInfer.h:8283
TRT_NODISCARD INormalizationLayer * addNormalizationV2(ITensor &input, ITensor &scale, ITensor &bias, uint32_t axesMask) noexcept
Add a normalization layer to the network.
Definition: NvInfer.h:9891
IActivationLayer * addActivation(ITensor &input, ActivationType type) noexcept
Add an activation layer to the network.
Definition: NvInfer.h:8317
ISliceLayer * addSlice(ITensor &input, Dims const &start, Dims const &size, Dims const &stride) noexcept
Add a slice layer to the network.
Definition: NvInfer.h:8897
virtual IBuilder & getBuilder() const noexcept
Return the builder from which this INetworkDefinition was created.
Definition: NvInfer.h:9787
ILayer * getLayer(int32_t index) const noexcept
Get the layer specified by the given index.
Definition: NvInfer.h:8498
bool isDebugTensor(ITensor const &tensor) const noexcept
Check if a tensor is marked as debug tensor.
Definition: NvInfer.h:8261
bool getFlag(NetworkDefinitionCreationFlag networkDefinitionCreationFlag) const noexcept
Returns true if the network definition creation flag is set.
Definition: NvInfer.h:8987
IIfConditional * addIfConditional() noexcept
Add an if-then-else to the network.
Definition: NvInfer.h:9181
IErrorRecorder * getErrorRecorder() const noexcept
get the ErrorRecorder assigned to this interface.
Definition: NvInfer.h:9337
ISqueezeLayer * addSqueeze(ITensor &input, ITensor &axes) noexcept
Add a squeeze layer to the network.
Definition: NvInfer.h:9844
TRT_DEPRECATED INMSLayer * addNMS(ITensor &boxes, ITensor &scores, ITensor &maxOutputBoxesPerClass) noexcept
Add a non-maximum suppression layer to the network.
Definition: NvInfer.h:9511
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:9662
IReverseSequenceLayer * addReverseSequence(ITensor &input, ITensor &sequenceLens) noexcept
Add a ReverseSequence layer to the network.
Definition: NvInfer.h:9548
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:9431
int32_t getNbInputs() const noexcept
Get the number of inputs in the network.
Definition: NvInfer.h:8510
NetworkDefinitionCreationFlags getFlags() const noexcept
Get the network definition creation flags for this network definition object. Defaults to 0.
Definition: NvInfer.h:8975
IQuantizeLayer * addQuantize(ITensor &input, ITensor &scale, DataType outputType) noexcept
Add a quantization layer to the network.
Definition: NvInfer.h:9404
IDynamicQuantizeLayer * addDynamicQuantizeV2(ITensor &input, Dims const &blockShape, DataType outputType, DataType scaleType) noexcept
Add a dynamic quantization layer to the network.
Definition: NvInfer.h:9455
IReduceLayer * addReduce(ITensor &input, ReduceOperation operation, uint32_t reduceAxes, bool keepDimensions) noexcept
Add a reduce layer to the network.
Definition: NvInfer.h:8582
IUnaryLayer * addUnary(ITensor &input, UnaryOperation operation) noexcept
Add a unary layer to the network.
Definition: NvInfer.h:8441
IGridSampleLayer * addGridSample(ITensor &input, ITensor &grid) noexcept
Add a GridSample layer to the network.
Definition: NvInfer.h:9489
void removeTensor(ITensor &tensor) noexcept
remove a tensor from the network definition.
Definition: NvInfer.h:8826
bool areWeightsMarkedRefittable(char const *name) const noexcept
Whether the weight has been marked as refittable.
Definition: NvInfer.h:9825
ISelectLayer * addSelect(ITensor &condition, ITensor &thenInput, ITensor &elseInput) noexcept
Add a select layer to the network.
Definition: NvInfer.h:9220
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:9380
TRT_DEPRECATED INormalizationLayer * addNormalization(ITensor &input, ITensor &scale, ITensor &bias, uint32_t axesMask) noexcept
Add a normalization layer to the network.
Definition: NvInfer.h:9580
int32_t getNbLayers() const noexcept
Get the number of layers in the network.
Definition: NvInfer.h:8484
TRT_DEPRECATED bool hasImplicitBatchDimension() const noexcept
Query whether the network was created with an implicit batch dimension.
Definition: NvInfer.h:8965
apiv::VNetworkDefinition * mImpl
Definition: NvInfer.h:9897
IKVCacheUpdateLayer * addKVCacheUpdate(ITensor &cache, ITensor &update, ITensor &writeIndices, KVCacheMode cacheMode) noexcept
Add a KVCacheUpdate layer to the network.
Definition: NvInfer.h:9721
bool markOutputForShapes(ITensor &tensor) noexcept
Enable tensor's value to be computed by IExecutionContext::getShapeBinding.
Definition: NvInfer.h:9004
IOneHotLayer * addOneHot(ITensor &indices, ITensor &values, ITensor &depth, int32_t axis) noexcept
Add a OneHot layer to the network.
Definition: NvInfer.h:8472
IScaleLayer * addScale(ITensor &input, ScaleMode mode, Weights shift, Weights scale, Weights power) noexcept
Add a Scale layer to the network.
Definition: NvInfer.h:8362
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:8877
void unmarkOutput(ITensor &tensor) noexcept
unmark a tensor as a network output.
Definition: NvInfer.h:8838
IIdentityLayer * addIdentity(ITensor &input) noexcept
Add an identity layer.
Definition: NvInfer.h:8796
IGatherLayer * addGatherV2(ITensor &data, ITensor &indices, GatherMode mode) noexcept
Add gather with specified mode, axis=0 and nbElementWiseDims=0.
Definition: NvInfer.h:8683
INonZeroLayer * addNonZero(ITensor &input, DataType indicesType) noexcept
Add a nonzero layer to the network.
Definition: NvInfer.h:8758
IElementWiseLayer * addElementWise(ITensor &input1, ITensor &input2, ElementWiseOperation op) noexcept
Add an elementwise layer to the network.
Definition: NvInfer.h:8419
IConstantLayer * addConstant(Dims const &dimensions, Weights weights) noexcept
Add a constant layer to the network.
Definition: NvInfer.h:8782
void setErrorRecorder(IErrorRecorder *recorder) noexcept
Set the ErrorRecorder for this interface.
Definition: NvInfer.h:9322
IPoolingLayer * addPoolingNd(ITensor &input, PoolingType type, Dims const &windowSize) noexcept
Add a multi-dimension pooling layer to the network.
Definition: NvInfer.h:9076
INMSLayer * addNMS(ITensor &boxes, ITensor &scores, ITensor &maxOutputBoxesPerClass, DataType indicesType) noexcept
Add a non-maximum suppression layer to the network.
Definition: NvInfer.h:9531
IRaggedSoftMaxLayer * addRaggedSoftMax(ITensor &input, ITensor &bounds) noexcept
Add a RaggedSoftMax layer to the network.
Definition: NvInfer.h:8702
IShapeLayer * addShape(ITensor &input) noexcept
Add a shape layer to the network.
Definition: NvInfer.h:8951
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:8667
bool unmarkWeightsRefittable(char const *name) noexcept
Unmark weights as refittable when the builder flag kREFIT_INDIVIDUAL is set.
Definition: NvInfer.h:9812
bool markWeightsRefittable(char const *name) noexcept
Mark weights as refittable when the builder flag kREFIT_INDIVIDUAL is set.
Definition: NvInfer.h:9800
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:9687
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:9098
IResizeLayer * addResize(ITensor &input) noexcept
Add a resize layer to the network.
Definition: NvInfer.h:9152
IUnsqueezeLayer * addUnsqueeze(ITensor &input, ITensor &axes) noexcept
Add an unsqueeze layer to the network.
Definition: NvInfer.h:9865
IMatrixMultiplyLayer * addMatrixMultiply(ITensor &input0, MatrixOperation op0, ITensor &input1, MatrixOperation op1) noexcept
Add a MatrixMultiply layer to the network.
Definition: NvInfer.h:8723
ISoftMaxLayer * addSoftMax(ITensor &input) noexcept
Add a SoftMax layer to the network.
Definition: NvInfer.h:8375
bool unmarkDebug(ITensor &tensor) noexcept
Unmark a tensor as a debug tensor.
Definition: NvInfer.h:8251
TRT_DEPRECATED IAttention * addAttention(ITensor &query, ITensor &key, ITensor &value, AttentionNormalizationOp normOp, bool causal) noexcept
Add an attention to the network.
Definition: NvInfer.h:9632
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:9471
void markOutput(ITensor &tensor) noexcept
Mark a tensor as a network output.
Definition: NvInfer.h:8217
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:8859
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:9279
int32_t getNbOutputs() const noexcept
Get the number of outputs in the network.
Definition: NvInfer.h:8540
bool setWeightsName(Weights weights, char const *name) noexcept
Associate a name with all current uses of the given weights.
Definition: NvInfer.h:9303
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:9775
IMoELayer * addMoE(ITensor &hiddenStates, ITensor &selectedExpertsForTokens, ITensor &scoresForSelectedExperts) noexcept
Add a MoE (Mixture of Experts) layer to the network.
Definition: NvInfer.h:9743
bool unmarkUnfusedTensorsAsDebugTensors() noexcept
Undo the marking of unfused tensors as debug tensors.
Definition: NvInfer.h:8297
Forward declaration of IEngineInspector for use by other interfaces.
Definition: NvInferRuntime.h:51
Definition: NvInfer.h:3665
DataType getIndicesType() const noexcept
Return the NonZero layer indices type.
Definition: NvInfer.h:3689
virtual ~INonZeroLayer() noexcept=0
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:3677
A normalization layer in a network definition.
Definition: NvInfer.h:6428
float getEpsilon() const noexcept
Get the epsilon value used for the normalization calculation.
Definition: NvInfer.h:6447
uint32_t getAxes() const noexcept
Get the axes value used for the normalization calculation.
Definition: NvInfer.h:6467
void setEpsilon(float eps) noexcept
Set the epsilon value used for the normalization calculation.
Definition: NvInfer.h:6437
virtual ~INormalizationLayer() noexcept=0
TRT_NODISCARD bool isV2() const noexcept
Returns true if this layer was created through addNormalizationV2().
Definition: NvInfer.h:6509
apiv::VNormalizationLayer * mImpl
Definition: NvInfer.h:6515
int64_t getNbGroups() const noexcept
Get the number of groups used to split the channels for the normalization calculation.
Definition: NvInfer.h:6498
void setAxes(uint32_t axesMask) noexcept
Set the reduction axes for the normalization calculation.
Definition: NvInfer.h:6457
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups used to split the channels in the normalization calculation.
Definition: NvInfer.h:6488
A OneHot layer in a network definition.
Definition: NvInfer.h:6042
apiv::VOneHotLayer * mImpl
Definition: NvInfer.h:6063
void setAxis(int32_t axis) noexcept
Set the axis parameter.
Definition: NvInfer.h:6049
virtual ~IOneHotLayer() noexcept=0
int32_t getAxis() const noexcept
Get the value of the axis parameter.
Definition: NvInfer.h:6057
Optimization profile for dynamic input dimensions and shape tensors.
Definition: NvInferRuntime.h:2575
Layer that represents a padding operation.
Definition: NvInfer.h:2850
Dims getPostPaddingNd() const noexcept
Get the padding that is applied at the end of the tensor.
Definition: NvInfer.h:2899
void setPrePaddingNd(Dims const &padding) noexcept
Set the padding that is applied at the start of the tensor.
Definition: NvInfer.h:2861
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:2887
Dims getPrePaddingNd() const noexcept
Get the padding that is applied at the start of the tensor.
Definition: NvInfer.h:2873
apiv::VPaddingLayer * mImpl
Definition: NvInfer.h:2905
Layer that represents a parametric ReLU operation.
Definition: NvInfer.h:3890
apiv::VParametricReLULayer * mImpl
Definition: NvInfer.h:3892
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:2543
apiv::VPluginV2Layer * mImpl
Definition: NvInfer.h:2556
IPluginV2 & getPlugin() noexcept
Get the plugin for the layer.
Definition: NvInfer.h:2550
virtual ~IPluginV2Layer() noexcept=0
Layer type for V3 plugins.
Definition: NvInfer.h:2572
virtual ~IPluginV3Layer() noexcept=0
IPluginV3 & getPlugin() noexcept
Get the plugin for the layer.
Definition: NvInfer.h:2579
apiv::VPluginV3Layer * mImpl
Definition: NvInfer.h:2585
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:1511
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:1482
Dims getStrideNd() const noexcept
Get the multi-dimension stride for pooling.
Definition: NvInfer.h:1492
Dims getPaddingNd() const noexcept
Get the multi-dimension padding for pooling.
Definition: NvInfer.h:1523
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:5407
void setToType(DataType toType) noexcept
Set the Quantize layer output type.
Definition: NvInfer.h:5468
bool setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5441
void setAxis(int32_t axis) noexcept
Set the quantization axis.
Definition: NvInfer.h:5428
virtual ~IQuantizeLayer() noexcept=0
TRT_NODISCARD Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5452
int32_t getAxis() const noexcept
Get the quantization axis.
Definition: NvInfer.h:5417
DataType getToType() const noexcept
Return the Quantize layer output type.
Definition: NvInfer.h:5480
A RaggedSoftmax layer in a network definition.
Definition: NvInfer.h:3716
apiv::VRaggedSoftMaxLayer * mImpl
Definition: NvInfer.h:3718
virtual ~IRaggedSoftMaxLayer() noexcept=0
A recurrence layer in a network definition.
Definition: NvInfer.h:4600
virtual ~IRecurrenceLayer() noexcept=0
Layer that represents a reduction across a non-bool tensor.
Definition: NvInfer.h:2768
void setKeepDimensions(bool keepDimensions) noexcept
Set the boolean that specifies whether or not to keep the reduced dimensions for the layer.
Definition: NvInfer.h:2815
virtual ~IReduceLayer() noexcept=0
void setOperation(ReduceOperation op) noexcept
Set the reduce operation for the layer.
Definition: NvInfer.h:2775
ReduceOperation getOperation() const noexcept
Get the reduce operation for the layer.
Definition: NvInfer.h:2785
uint32_t getReduceAxes() const noexcept
Get the axes over which to reduce for the layer.
Definition: NvInfer.h:2805
void setReduceAxes(uint32_t reduceAxes) noexcept
Set the axes over which to reduce.
Definition: NvInfer.h:2795
apiv::VReduceLayer * mImpl
Definition: NvInfer.h:2831
bool getKeepDimensions() const noexcept
Get the boolean that specifies whether or not to keep the reduced dimensions for the layer.
Definition: NvInfer.h:2825
A resize layer in a network definition.
Definition: NvInfer.h:4069
void setSelectorForSinglePixel(ResizeSelector selector) noexcept
Set coordinate selector function when resized to single pixel.
Definition: NvInfer.h:4230
void setNearestRounding(ResizeRoundMode value) noexcept
Set rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4254
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:4148
void setOutputDimensions(Dims const &dimensions) noexcept
Set the output dimensions.
Definition: NvInfer.h:4089
void setCubicCoeff(float A) noexcept
Set the coefficient 'A' used in cubic interpolation.
Definition: NvInfer.h:4286
void setScales(float const *scales, int32_t nbScales) noexcept
Set the resize scales.
Definition: NvInfer.h:4129
virtual ~IResizeLayer() noexcept=0
float getCubicCoeff() const noexcept
Get the coefficient 'A' used in cubic interpolation.
Definition: NvInfer.h:4296
ResizeSelector getSelectorForSinglePixel() const noexcept
Get the coordinate selector function when resized to single pixel.
Definition: NvInfer.h:4240
InterpolationMode getResizeMode() const noexcept
Get resize mode for an input tensor.
Definition: NvInfer.h:4170
void setCoordinateTransformation(ResizeCoordinateTransformation coordTransform) noexcept
Set coordinate transformation function.
Definition: NvInfer.h:4205
void setExcludeOutside(bool excludeFlag) noexcept
Set the state for excluding outside pixels.
Definition: NvInfer.h:4309
void setResizeMode(InterpolationMode interpolationMode) noexcept
Set resize mode for an input tensor.
Definition: NvInfer.h:4160
Dims getOutputDimensions() const noexcept
Get the output dimensions.
Definition: NvInfer.h:4099
ResizeRoundMode getNearestRounding() const noexcept
Get rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4264
bool getExcludeOutside() const noexcept
Get the state for excluding outside pixels.
Definition: NvInfer.h:4319
ResizeCoordinateTransformation getCoordinateTransformation() const noexcept
Get coordinate transformation function.
Definition: NvInfer.h:4215
A ReverseSequence layer in a network definition.
Definition: NvInfer.h:6354
void setSequenceAxis(int32_t sequenceAxis) noexcept
Set the sequence axis. Default is 0.
Definition: NvInfer.h:6387
int32_t getBatchAxis() const noexcept
Return the batch axis. Return 1 if no batch axis was set.
Definition: NvInfer.h:6374
apiv::VReverseSequenceLayer * mImpl
Definition: NvInfer.h:6403
int32_t getSequenceAxis() const noexcept
Return the sequence axis. Return 0 if no sequence axis was set.
Definition: NvInfer.h:6397
void setBatchAxis(int32_t batchAxis) noexcept
Set the batch axis. Default is 1.
Definition: NvInfer.h:6364
virtual ~IReverseSequenceLayer() noexcept=0
Layer that implements Rotary Position Embedding (RoPE) (https://arxiv.org/abs/2104....
Definition: NvInfer.h:7453
TRT_NODISCARD int32_t getRotaryEmbeddingDim() const noexcept
Get the number of hidden dimensions participating in RoPE. The default value is 0,...
Definition: NvInfer.h:7493
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:7460
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:7482
apiv::VRotaryEmbeddingLayer * mImpl
Definition: NvInfer.h:7515
TRT_NODISCARD bool getInterleaved() const noexcept
Get whether the input is in interleaved format. The default value is false.
Definition: NvInfer.h:7471
A Scale layer in a network definition.
Definition: NvInfer.h:1693
Weights getScale() const noexcept
Get the scale value.
Definition: NvInfer.h:1750
Weights getPower() const noexcept
Get the power value.
Definition: NvInfer.h:1770
void setScale(Weights scale) noexcept
Set the scale value.
Definition: NvInfer.h:1740
void setPower(Weights power) noexcept
Set the power value.
Definition: NvInfer.h:1760
ScaleMode getMode() const noexcept
Get the scale mode.
Definition: NvInfer.h:1710
void setShift(Weights shift) noexcept
Set the shift value.
Definition: NvInfer.h:1720
void setChannelAxis(int32_t channelAxis) noexcept
Set the channel axis.
Definition: NvInfer.h:1806
virtual ~IScaleLayer() noexcept=0
Weights getShift() const noexcept
Get the shift value.
Definition: NvInfer.h:1730
void setMode(ScaleMode mode) noexcept
Set the scale mode.
Definition: NvInfer.h:1700
int32_t getChannelAxis() const noexcept
Get the channel axis.
Definition: NvInfer.h:1785
A scatter layer in a network definition. Supports several kinds of scattering.
Definition: NvInfer.h:5967
void setMode(ScatterMode mode) noexcept
Set the scatter mode.
Definition: NvInfer.h:5974
apiv::VScatterLayer * mImpl
Definition: NvInfer.h:6008
void setAxis(int32_t axis) noexcept
Set the axis used by ScatterMode::kELEMENTS.
Definition: NvInfer.h:5994
int32_t getAxis() const noexcept
Get the axis.
Definition: NvInfer.h:6002
ScatterMode getMode() const noexcept
Get the scatter mode.
Definition: NvInfer.h:5984
virtual ~IScatterLayer() noexcept=0
Select elements from two data tensors based on a condition tensor.
Definition: NvInfer.h:4918
virtual ~ISelectLayer() noexcept=0
Layer type for getting shape of a tensor.
Definition: NvInfer.h:3381
virtual ~IShapeLayer() noexcept=0
apiv::VShapeLayer * mImpl
Definition: NvInfer.h:3383
Layer type for shuffling data.
Definition: NvInfer.h:2940
apiv::VShuffleLayer * mImpl
Definition: NvInfer.h:3098
virtual ~IShuffleLayer() noexcept=0
void setFirstTranspose(Permutation permutation) noexcept
Set the permutation applied by the first transpose operation.
Definition: NvInfer.h:2951
void setSecondTranspose(Permutation permutation) noexcept
Set the permutation applied by the second transpose operation.
Definition: NvInfer.h:3051
Dims getReshapeDimensions() const noexcept
Get the reshaped dimensions.
Definition: NvInfer.h:3004
void setReshapeDimensions(Dims const &dimensions) noexcept
Set the reshaped dimensions.
Definition: NvInfer.h:2991
Permutation getFirstTranspose() const noexcept
Get the permutation applied by the first transpose operation.
Definition: NvInfer.h:2963
Permutation getSecondTranspose() const noexcept
Get the permutation applied by the second transpose operation.
Definition: NvInfer.h:3063
bool getZeroIsPlaceholder() const noexcept
Get meaning of 0 in reshape dimensions.
Definition: NvInfer.h:3092
void setZeroIsPlaceholder(bool zeroIsPlaceholder) noexcept
Set meaning of 0 in reshape dimensions.
Definition: NvInfer.h:3079
Slices an input tensor into an output tensor based on the offset and strides.
Definition: NvInfer.h:3194
void setStride(Dims const &stride) noexcept
Set the stride for computing the output slice data.
Definition: NvInfer.h:3263
apiv::VSliceLayer * mImpl
Definition: NvInfer.h:3362
virtual ~ISliceLayer() noexcept=0
void setSize(Dims const &size) noexcept
Set the dimensions of the output slice.
Definition: NvInfer.h:3234
void setAxes(Dims const &axes) noexcept
Set the axes for this ISliceLayer.
Definition: NvInfer.h:3341
void setStart(Dims const &start) noexcept
Set the start offset that the slice layer uses to create the output slice.
Definition: NvInfer.h:3205
Dims getStart() const noexcept
Get the start offset for the slice layer.
Definition: NvInfer.h:3220
void setMode(SampleMode mode) noexcept
Set the slice mode.
Definition: NvInfer.h:3288
Dims getSize() const noexcept
Get dimensions of the output slice.
Definition: NvInfer.h:3249
SampleMode getMode() const noexcept
Get the slice mode.
Definition: NvInfer.h:3298
Dims getStride() const noexcept
Get the stride for the output slice.
Definition: NvInfer.h:3278
Dims getAxes() const noexcept
Get the axes for this ISliceLayer.
Definition: NvInfer.h:3356
A Softmax layer in a network definition.
Definition: NvInfer.h:1839
void setAxes(uint32_t axes) noexcept
Set the axis along which softmax is computed. Currently, only one axis can be set.
Definition: NvInfer.h:1861
virtual ~ISoftMaxLayer() noexcept=0
uint32_t getAxes() const noexcept
Get the axis along which softmax occurs.
Definition: NvInfer.h:1871
Layer that represents a squeeze operation, removing unit dimensions of the first input tensor on a se...
Definition: NvInfer.h:6531
apiv::VSqueezeLayer * mImpl
Definition: NvInfer.h:6548
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:10165
int64_t queryKeys(TimingCacheKey *keyBuffer, int64_t capacity) const noexcept
Query cache keys from Timing Cache.
Definition: NvInfer.h:10231
virtual ~ITimingCache() noexcept=0
bool combine(ITimingCache const &inputCache, bool ignoreMismatch) noexcept
Combine input timing cache into local instance.
Definition: NvInfer.h:10202
TimingCacheValue query(TimingCacheKey const &key) const noexcept
Query value in a cache entry.
Definition: NvInfer.h:10248
bool update(TimingCacheKey const &key, TimingCacheValue const &value) noexcept
Update values in a cache entry.
Definition: NvInfer.h:10270
apiv::VTimingCache * mImpl
Definition: NvInfer.h:10276
bool reset() noexcept
Empty the timing cache.
Definition: NvInfer.h:10212
Layer that represents a TopK reduction.
Definition: NvInfer.h:3423
void setK(int32_t k) noexcept
Set the static k value for the layer.
Definition: NvInfer.h:3454
void setReduceAxes(uint32_t reduceAxes) noexcept
Set which axes to reduce for the layer.
Definition: NvInfer.h:3478
TopKOperation getOperation() const noexcept
Get the operation for the layer.
Definition: NvInfer.h:3440
apiv::VTopKLayer * mImpl
Definition: NvInfer.h:3537
void setOperation(TopKOperation op) noexcept
Set the operation for the layer.
Definition: NvInfer.h:3430
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:3519
virtual ~ITopKLayer() noexcept=0
int32_t getK() const noexcept
Get the k value for the layer.
Definition: NvInfer.h:3468
uint32_t getReduceAxes() const noexcept
Get the axes to reduce for the layer.
Definition: NvInfer.h:3488
DataType getIndicesType() const noexcept
Return the TopK layer indices type.
Definition: NvInfer.h:3531
A layer that represents a trip-count limiter.
Definition: NvInfer.h:4725
virtual ~ITripLimitLayer() noexcept=0
TripLimit getTripLimit() const noexcept
Get a trip limiter type.
Definition: NvInfer.h:4730
Layer that represents an unary operation.
Definition: NvInfer.h:2655
void setOperation(UnaryOperation op) noexcept
Set the unary operation for the layer.
Definition: NvInfer.h:2664
apiv::VUnaryLayer * mImpl
Definition: NvInfer.h:2680
UnaryOperation getOperation() const noexcept
Get the unary operation for the layer.
Definition: NvInfer.h:2674
virtual ~IUnaryLayer() noexcept=0
Layer that represents an unsqueeze operation, which reshapes the first input tensor by inserting unit...
Definition: NvInfer.h:6563
virtual ~IUnsqueezeLayer() noexcept=0
apiv::VUnsqueezeLayer * mImpl
Definition: NvInfer.h:6581
An Interface class for version control.
Definition: NvInferRuntimeBase.h:282
Version information associated with a TRT interface.
Definition: NvInferRuntimeBase.h:247
An array of weights used as a layer parameter.
Definition: NvInferRuntime.h:121
Definition: NvInferRuntimeBase.h:419
Definition: NvInferRuntime.h:1652
Application-implemented logging interface for the builder, refitter and runtime.
Definition: NvInferRuntime.h:1575
Definition: NvInferPluginBase.h:206
Definition: NvInfer.h:10477
virtual bool stepComplete(char const *phaseName, int32_t step) noexcept=0
Signal that a step of an optimizer phase has finished.
virtual ~IProgressMonitor() noexcept=default
IProgressMonitor()=default
virtual void phaseFinish(char const *phaseName) noexcept=0
Signal that a phase of the optimizer has finished.
virtual void phaseStart(char const *phaseName, char const *parentPhase, int32_t nbSteps) noexcept=0
Signal that a phase of the optimizer has started.
Definition: NvInferRuntime.h:643
IBuilder * createInferBuilder(ILogger &logger) noexcept
Create an instance of an IBuilder class.
Definition: NvInfer.h:11756
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:2780
ResizeSelector
The coordinate selector when resize to single pixel output.
Definition: NvInfer.h:3980
@ 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:10289
AttentionIOForm
Enumerates the layout of the input/output tensors in an Attention layer.
Definition: NvInfer.h:6791
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:1650
@ 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:9920
HardwareCompatibilityLevel
Describes requirements of compatibility with GPU architectures other than that of the GPU on which th...
Definition: NvInfer.h:10397
@ kSAME_COMPUTE_CAPABILITY
CumulativeOperation
Enumerates the cumulative operations that may be performed by a Cumulative layer.
Definition: NvInfer.h:6599
BoundingBoxFormat
Representation of bounding box data used for the Boxes input tensor in INMSLayer.
Definition: NvInfer.h:6166
@ 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:2608
@ 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:4983
ResizeRoundMode
The rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4007
@ 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:6763
@ 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:4365
@ 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:11380
PreviewFeature
Define preview features.
Definition: NvInfer.h:10363
@ kRUNTIME_ACTIVATION_RESIZE_10_10
@ kALIASED_PLUGIN_IO_10_03
TilingOptimizationLevel
Define the optimization levels for Tiling.
Definition: NvInfer.h:10447
@ kFAST
Use a fast algorithm and heuristic based strategy. Slightly increases engine build time.
@ kFULL
Increase search space even wider. Significantly increases engine build time.
DataType
The type of weights and tensors. The datatypes other than kBOOL, kINT32, and kINT64 are "activation d...
Definition: NvInferRuntimeBase.h:149
uint32_t BuilderFlags
Represents one or more BuilderFlag values using binary OR operations, e.g., 1U << BuilderFlag::kDEBUG...
Definition: NvInfer.h:9949
DeviceType
The device that this layer/network will execute on.
Definition: NvInferRuntime.h:1341
LayerType
The type values of layer classes.
Definition: NvInfer.h:58
@ kGRID_SAMPLE
Grid sample layer.
@ kRAGGED_SOFTMAX
Ragged softmax layer.
@ kDECONVOLUTION
Deconvolution layer.
@ kASSERTION
Assertion layer.
@ kATTENTION_INPUT
Attention Input.
@ kMATRIX_MULTIPLY
Matrix multiply layer.
@ kCONDITION
Condition layer.
@ kCUMULATIVE
Cumulative layer.
@ kCONDITIONAL_INPUT
Conditional Input layer.
@ kIDENTITY
Identity layer.
@ kNORMALIZATION
Normalization layer.
@ kQUANTIZE
Quantize layer.
@ kCONVOLUTION
Convolution layer.
@ kPARAMETRIC_RELU
Parametric ReLU layer.
@ kATTENTION_OUTPUT
Attention Output.
@ kUNSQUEEZE
Unsqueeze Layer.
@ kCONCATENATION
Concatenation layer.
@ kREVERSE_SEQUENCE
Reverse sequence layer.
@ kROTARY_EMBEDDING
Rotary Embedding layer.
@ kRECURRENCE
Loop Recurrence layer.
@ kDEQUANTIZE
Dequantize layer.
@ kPLUGIN_V3
PluginV3 layer.
@ kITERATOR
Loop Iterator layer.
@ kTRIP_LIMIT
Loop Trip limit layer.
@ kDYNAMIC_QUANTIZE
Dynamic Quantize layer.
@ kUNARY
UnaryOp operation Layer.
@ kACTIVATION
Activation layer.
@ kELEMENTWISE
Elementwise layer.
@ kPLUGIN_V2
PluginV2 layer.
@ kLOOP_OUTPUT
Loop output layer.
@ kCONDITIONAL_OUTPUT
Conditional Output layer.
@ kCONSTANT
Constant layer.
@ kNON_ZERO
NonZero layer.
@ kKVCACHE_UPDATE
KV Cache Update layer.
@ kDIST_COLLECTIVE
DistCollective layer.
SampleMode
Controls how ISliceLayer and IGridSample handle out-of-bounds coordinates.
Definition: NvInfer.h:3110
@ 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:2349
@ 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:7680
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:2792
NetworkDefinitionCreationFlag
List of immutable network properties expressed at network creation time. NetworkDefinitionCreationFla...
Definition: NvInfer.h:11391
@ kPREFER_JIT_PYTHON_PLUGINS
@ kPREFER_AOT_PYTHON_PLUGINS
ElementWiseOperation
Enumerates the binary operations that may be performed by an ElementWise layer.
Definition: NvInfer.h:2260
@ 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:2738
@ kALL_TO_ALL
All-to-all exchange.
@ kREDUCE_SCATTER
Reduce scatter.
InterpolationMode
Enumerates various modes of interpolation.
Definition: NvInfer.h:3904
@ kNEAREST
ND (0 < N <= 8) nearest neighbor resizing.
@ kCUBIC
Supports bicubic (2D) interpolation.
@ kLINEAR
Supports linear (1D), bilinear (2D), and trilinear (3D) interpolation.
BuilderFlag
List of valid modes that the builder can enable when creating an engine from a network definition.
Definition: NvInfer.h:9959
@ 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
@ kSTRIP_PLAN
Strip the refittable weights from the engine plan file.
@ kMONITOR_MEMORY
Enable memory monitor during build time.
@ kDISABLE_TIMING_CACHE
Disable reuse of timing information across identical layers.
@ kDISABLE_COMPILATION_CACHE
@ kREFIT
Enable building a refittable engine.
TENSORRTAPI nvinfer1::IPluginRegistry * getBuilderPluginRegistry(nvinfer1::EngineCapability capability) noexcept
Return the plugin registry for building a Standard engine, or nullptr if no registry exists.
TopKOperation
Enumerates the operations that may be performed by a TopK layer.
Definition: NvInfer.h:3395
ReduceOperation
Enumerates the reduce operations that may be performed by a Reduce layer.
Definition: NvInfer.h:2710
@ 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:5893
MatrixOperation
Enumerates the operations that may be performed on a tensor by IMatrixMultiplyLayer before multiplica...
Definition: NvInfer.h:3550
@ kTRANSPOSE
Like kNONE, but transpose the matrix dimensions.
ResizeCoordinateTransformation
The resize coordinate transformation function.
Definition: NvInfer.h:3929
LoopOutput
Enum that describes kinds of loop outputs.
Definition: NvInfer.h:4337
@ 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:7527
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:10560
TensorLocation
The location for tensor data storage, device or host.
Definition: NvInferRuntime.h:203
AttentionNormalizationOp
Enumerates the operations that may be performed by the normalization in the attention subgraph.
Definition: NvInfer.h:6731
Represents a permutation of dimensions.
Definition: NvInfer.h:2917
Declaration of EnumMaxImpl struct to store the exclusive upper bound of an enumeration type.
Definition: NvInferRuntimeBase.h:131
The key to retrieve timing cache entries.
Definition: NvInfer.h:10129
Definition: NvInfer.h:10141
uint64_t tacticHash
Hash of the selected tactic.
Definition: NvInfer.h:10143
float timingMSec
Timing of this tactic in milliseconds. Negative numbers and NaN are invalid values.
Definition: NvInfer.h:10145