aws-neuron/Qwen3-1.7B-TP4-BS4-SEQ2048
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: aws-neuron/Qwen3-1.7B-TP4-BS4-SEQ2048
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-08-30 21:56:23 +08:00
commit fd57cdfd17
114 changed files with 237628 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
token_generation_model/_tp0_bk2/command.txt Normal file
View File

@@ -0,0 +1 @@
neuronx-cc compile --framework=XLA model.MODULE_c60e20a111715e620aa2+6bb8acc9.hlo_module.pb --output model.MODULE_c60e20a111715e620aa2+6bb8acc9.neff --target=trn2 --auto-cast=none --model-type=transformer '--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=1 --vectorize-strided-dma ' --lnc=2 -O2 --internal-hlo2tensorizer-options=--verify-hlo=true --logfile=log-neuron-cc.txt --verbose=35

1
token_generation_model/_tp0_bk2/compile_flags.MODULE_c60e20a111715e620aa2+6bb8acc9.json Normal file
View File

@@ -0,0 +1 @@
["--target=trn2", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=1 --vectorize-strided-dma ", "--lnc=2", "-O2", "--internal-hlo2tensorizer-options=--verify-hlo=true", "--logfile=/models/Qwen3-1.7B-compiled/token_generation_model/_tp0_bk2/log-neuron-cc.txt"]

590
token_generation_model/_tp0_bk2/global_metric_store.json Normal file
View File

@@ -0,0 +1,590 @@
{
"Average": {
"tensorizer": {
"StaticProfiler::AverageFractalPeUtilization": 97.78141784667969,
"StaticProfiler::AveragePartitionUtilization": 90.33787536621094,
"StaticProfiler::AveragePeUtilization": 81.002685546875,
"StaticProfiler::LocalizationEfficiency": 155.71730041503906,
"StaticProfiler::LocalizationEfficiencyIgnoreNonlocal": 160.65768432617188,
"TilingProfiler::AveragePartitionUtilizationAfterTiling": 0,
"TilingProfiler::AveragePeUtilizationAfterTiling": 0
}
},
"Count": {
"tensorizer": {
"StaticProfiler::AverageFractalPeUtilization": 1,
"StaticProfiler::AveragePartitionUtilization": 1,
"StaticProfiler::AveragePeUtilization": 1,
"StaticProfiler::LocalizationEfficiency": 1,
"StaticProfiler::LocalizationEfficiencyIgnoreNonlocal": 1,
"TilingProfiler::AveragePartitionUtilizationAfterTiling": 1,
"TilingProfiler::AveragePeUtilizationAfterTiling": 1
}
},
"Sum": {
"compiletime": {
"AGOrderingAnalysisPass": 6.775559425354004,
"AffinePredicateResolution": 0.7124242782592773,
"AliasDependencyElimination": 0.0037508010864257813,
"AliasDependencyInduction": 17.88080596923828,
"AliasDependencyReset": 19.01345443725586,
"BFComputeCutting": 0.14321279525756836,
"BirCodeGenLoop": 4.049878120422363,
"CCOpFusion": 0.9675219058990479,
"CanonicalizeConv": 0.0,
"CanonicalizeDAGForPGTiling": 0.18202924728393555,
"CanonicalizeForTensorizer": 0.0005849999724887311,
"CanonicalizeIR": 0.8085732460021973,
"Canonicalizer": 0.015991000458598137,
"CoalesceCCOp": 0.25977158546447754,
"CommuteConcat": 0.027615070343017578,
"DMALocalityOpt": 0.048050880432128906,
"DMAProfiler": 0.09339690208435059,
"DMATilingProfiler": 0.09855914115905762,
"DataLocalityOpt": 2.8309335708618164,
"DataStreaming": 0.2996983528137207,
"DeConcat": 0.08213448524475098,
"DeadCodeElimination": 0.030428647994995117,
"DeadStoreElimination": 1.051145315170288,
"DelinearIndices": 0.5529322624206543,
"Delinearization": 0.14493751525878906,
"DelinearizeSPMD": 0.181105375289917,
"DoNothing": 0.0003535747528076172,
"DramToDramTranspose": 0.5176758766174316,
"DumpGraphAndMetadata": 0.1749565601348877,
"EliminateDivs": 1.6087908744812012,
"ExpandBatchNorm": 1.096595048904419,
"ExpandISAMacro": 0.09282922744750977,
"FactorizeBlkDims": 0.5121583938598633,
"FactorizeThreadAxesInFreeDims": 0.08890652656555176,
"FlattenMacroLoop": 0.08152222633361816,
"GenericAccessSimplifier": 0.02545022964477539,
"HoistCompute": 9.899999713525176e-05,
"IdentifyCrossPassTensors": 0.00011500000255182385,
"InferInitValue": 1.2978432178497314,
"InferIntrinsicOnCC": 0.7616860866546631,
"InferNeuronTensor": 2.6053829193115234,
"InferNonlocalTensors": 7.113053321838379,
"InferPSumTensor": 4.450908660888672,
"InferShardAxis": 12.849608421325684,
"InferSharedMemLoc": 0.19225406646728516,
"InlineNativeKernels": 0.08442234992980957,
"InsertCoreBarrier": 0.15249347686767578,
"InsertIOTransposes": 0.8695223331451416,
"InsertImplicitShardAxisBeforeISel": 0.49620485305786133,
"InsertLocalTransposes": 1.934779167175293,
"InsertOffloadedTransposes": 0.23423099517822266,
"LICM": 0.1213834285736084,
"LateLegalizeInst": 0.28924560546875,
"LateLegalizePostSplit": 0.22411513328552246,
"LateLowerReshapeOp": 0.033789873123168945,
"LateLowerTensorOp": 4.297260761260986,
"LateNeuronInstComb": 1.0230679512023926,
"LayoutPreprocessing": 1.8572075366973877,
"LayoutPreprocessingAndAnalysis": 2.2904410362243652,
"LayoutRequirementAnalysis": 0.4237644672393799,
"LegalizeCCOpLayout": 1.1935856342315674,
"LegalizeOpLevelAlias": 0.7793729305267334,
"LegalizePartitionReduce": 0.09432244300842285,
"LegalizeSundaAccess": 1.0004379749298096,
"LegalizeSundaMacro": 0.7198967933654785,
"LegalizeType": 0.1766369342803955,
"LocalLayoutOpt": 0.6666848659515381,
"LoopFusion": 0.29988932609558105,
"LoopSplitting": 0.05208778381347656,
"LowerBroadcast": 0.08215570449829102,
"LowerCCOpBlockAxis": 0.2862672805786133,
"LowerComplexBroadcast": 0.09356045722961426,
"LowerIntrinsics": 1.3606979846954346,
"LowerShardAxis": 0.39397311210632324,
"LowerTensorOp": 4.018380165100098,
"LowerToSendRecv": 0.27350616455078125,
"LowerTranspose": 0.5171041488647461,
"MacroGeneration": 2.0922586917877197,
"MaskPropagation": 0.12857651710510254,
"MemcastMotion": 0.0002789999998640269,
"MemcpyElimination": 30.280338287353516,
"MutateDataType": 0.03657245635986328,
"NeuronAliasDependencyInduction": 0.026292085647583008,
"NeuronAliasDependencyReset": 0.03253936767578125,
"NeuronInstComb": 0.36768054962158203,
"NeuronLICM": 0.3019218444824219,
"NeuronLoopFusion": 1.711080551147461,
"NeuronLoopInterchange": 0.08134770393371582,
"NeuronSimplifier": 0.5104148387908936,
"NeuronSimplifyPredicates": 0.24121522903442383,
"NeuronValueNumbering": 0.11496901512145996,
"OptimizeAliasedCopyChain": 0.40732264518737793,
"OptimizeNKIKernels": 1.0493371486663818,
"PAGLayoutOpt": 18.23680305480957,
"PComputeCutting": 0.5901892185211182,
"PGLayoutTilingPipeline": 54.30891036987305,
"PGTiling": 10.472368240356445,
"PadElimination": 0.0187380313873291,
"ParAxesAnnotation": 16.293506622314453,
"PartialLoopFusion": 1.5408625602722168,
"PartialSimdFusion": 0.8349573612213135,
"PenguinizeFunctions": 0.0005789999850094318,
"PerfectLoopNest": 0.07371997833251953,
"PruneFunctions": 0.00018899999849963933,
"RecognizeOpIdiom": 0.12681221961975098,
"Recompute": 0.007548809051513672,
"RelaxPredicates": 0.11768078804016113,
"Rematerialization": 0.15080475807189941,
"RemoveOptimizationBarriers": 0.0004290000069886446,
"RemoveShardedPartitionAxes": 1.1734652519226074,
"ReshapeWeights": 0.022508859634399414,
"ResolveAccessConflict": 0.20145153999328613,
"ResolveComplicatePredicates": 0.3191838264465332,
"RewriteReplicationMatmul": 0.039293527603149414,
"RewriteWeights": 0.0773766040802002,
"SFKVectorizer": 11.08578872680664,
"ScatterMotion": 0.005127000156790018,
"ShardingPropagationAnalysis": 0.665086030960083,
"SimpleAllReduceTiling": 0.07452917098999023,
"Simplifier": 0.09160900115966797,
"SimplifyMacroPredicates": 0.28479433059692383,
"SimplifyNeuronTensor": 0.5389096736907959,
"SimplifySlice": 0.02684783935546875,
"SimplifyTensor": 0.2644212245941162,
"SpillPSum": 1.0592920780181885,
"SplitAPUnionSets": 1.0491511821746826,
"SplitAccGrp": 0.055588722229003906,
"StaticProfiler": 0.39172840118408203,
"StaticTransposeLocalTensor": 0.7211995124816895,
"SundaISel": 1.5357580184936523,
"TCTransform": 0.029030561447143555,
"TensorInitialization": 0.20336008071899414,
"TensorOpSimplifier": 3.100177049636841,
"TensorOpTransform": 20.175758361816406,
"TensorizerLegalizationPass": 0.000455000001238659,
"TileCCOps": 0.17463970184326172,
"TilingProfiler": 0.7014575004577637,
"TransformConvOp": 0.9923229217529297,
"TritiumFusion": 0.33277034759521484,
"ValueNumbering": 0.08796954154968262,
"VectorizeDMA": 0.520815372467041,
"VectorizeMatMult": 0.08555078506469727,
"VerifySupportedOps": 0.0002469999890308827,
"WeightCoalescing": 0.09020423889160156,
"ZeroSizeTensorElimination": 0.0006709098815917969,
"algsimp": 0.0018210000125691295,
"batchnorm_expander": 0.0022700000554323196,
"boundary-marker-removal": 0.0009800000116229057,
"call-inliner": 0.0002410000015515834,
"canonicalize-boundary-marker": 0.0009430000209249556,
"collective-stream-id-checker": 0.0008960000122897327,
"comparison-expander": 0.000623999978415668,
"computation-deduplicator": 0.003625999903306365,
"config-lowering": 0.00017899999511428177,
"constant_folding": 0.0668639987707138,
"cse": 0.001184999942779541,
"dce": 6.399999983841553e-05,
"dynamic-slice-transpose": 0.0003530000103637576,
"eliminate-redundant-compare": 0.00014400000509340316,
"emit-offloaded-dropout": 0.0007820000173524022,
"flatten-call-graph": 0.0009519999730400741,
"fuse-send-recv": 0.07950299978256226,
"hilo-conditional-to-select": 0.00015700000221841037,
"hilo::LegalizeAlias": 0.003444999922066927,
"hilo::NeuronInstCombine": 0.0008500000112690032,
"hilo::NeuronOpFusion": 0.00020799999765586108,
"hilo::ReplaceTokenTypeWithU8Pass": 0.0007050000131130219,
"hilo::ScheduleFusion": 4.3000000005122274e-05,
"hilo::SixtyFourHack": 0.0005150000215508044,
"hilo::VerifyAliasing": 0.00035099999513477087,
"hlo-mac-count": 0.08947300165891647,
"io-con-pipe-begin": 6.199999916134402e-05,
"io-con-pipe-end": 9.999999974752427e-07,
"io-layout-normalization": 0.0706389993429184,
"legalize-ccops-for-tensorizer": 0.00013800000306218863,
"legalize-compare": 0.00027200000477023423,
"lower-argminmax-custom-call": 0.0004130000015720725,
"map-inline": 0.0010939999483525753,
"metadata-naming": 0.009581999853253365,
"mlir::detail::OpToOpPassAdaptor": 0.0004569999873638153,
"mlir::hlo::MhloToPyPenguin": 0.5751969814300537,
"mlir::mhlo::LowerComplexExtraPass": 0.00508299982175231,
"mlir::mhlo::LowerComplexPass": 0.0012469999492168427,
"native-to-custom-softmax": 0.0004349999944679439,
"native-to-custom-softmax-dx": 0.00044800000614486635,
"neuron-hlo-verifier": 0.3175100088119507,
"operand_upcaster": 0.0018560000462457538,
"post-par-pipe-begin": 9.999999974752427e-07,
"post-par-pipe-end": 0.0,
"post-partition-simplification": 0.48272499442100525,
"pre-hlo-begin": 1.4000000192027073e-05,
"pre-hlo-end": 1.9999999949504854e-06,
"replace-minimum-constant": 0.00022400000307243317,
"reshape-mover": 5.700000110664405e-05,
"simplify-concat": 0.002764000091701746,
"simplify-while-loops": 9.40000027185306e-05,
"transform-variadic-reduce": 0.001218999968841672,
"tuple-simplifier": 0.00048099999548867345,
"unpack-nested-aws-ntwsr": 0.0007040000054985285,
"unroll-while-loop": 1.2000000424450263e-05
},
"hilo": {
"HloMacCount": 1796005888.0,
"Traffic": 1252622464.0
},
"tensorizer": {
"DMATilingProfiler::TotalInstructionsAfterTiling": 45730,
"StaticProfiler::AifUb": 19.087068557739258,
"StaticProfiler::ArithmeticIntensityTensorizer": 29.72186851501465,
"StaticProfiler::AverageDmaLength": 2694.496826171875,
"StaticProfiler::DDRTransferBytes": 954289776,
"StaticProfiler::InternalTransferBytes": 196796000,
"StaticProfiler::LoadExpanded": 281500,
"StaticProfiler::StoreExpanded": 4273,
"StaticProfiler::TotalDMAExpanded": 285773,
"StaticProfiler::TotalDynamicInstancesCount": 62161,
"StaticProfiler::TotalDynamicInstancesWithMmPackedCount": 53675,
"StaticProfiler::TotalLNCComm": 0,
"StaticProfiler::TotalLNCCommTransfer": 0,
"TilingProfiler::BatchnormInstructionsAfterTiling": 0,
"TilingProfiler::DmaInstructionsAfterTiling": 0,
"TilingProfiler::GenericInstructionsAfterTiling": 123,
"TilingProfiler::MatMultInstructionsAfterTiling": 32320,
"TilingProfiler::NumPfTransposes": 292,
"TilingProfiler::NumPfTransposesForIo": 30,
"TilingProfiler::NumPfTransposesForLocal": 142,
"TilingProfiler::NumPfTransposesForNonlocal": 120,
"TilingProfiler::PfTransposeInstructions": 8762,
"TilingProfiler::PfTransposeInstructionsForIo": 6564,
"TilingProfiler::PfTransposeInstructionsForLocal": 536,
"TilingProfiler::PfTransposeInstructionsForNonlocal": 1662,
"TilingProfiler::ReduceInstructionsAfterTiling": 171,
"TilingProfiler::SimdInstructionsAfterTiling": 2465,
"TilingProfiler::TotalInstructionsAfterTiling": 0,
"TransformConvOp::Conv1d_depthwise_bf01_oi01_bf01": 0,
"TransformConvOp::Conv2d_dw_fb01_io01_01bf_rep_nhwc_Pcinh": 0,
"TransformConvOp::Conv2d_pbp_0f1b_0i1o_01fb_experimental_1": 0,
"TransformConvOp::Conv2d_pbp_fb01_io01_01bf_experimental_1": 0,
"TransformConvOp::conv2d_column_packing": 0,
"TransformConvOp::conv2d_column_packing_1": 0,
"TransformConvOp::conv2d_column_packing_io10": 0,
"TransformConvOp::conv2d_depthwise_f01b_o01i_bf01": 0
}
},
"all": {
"compiletime": {
"CanonicalizeConv": 0.0,
"CanonicalizeForTensorizer": 0.0005849999724887311,
"Canonicalizer": 0.015991000458598137,
"HoistCompute": 9.899999713525176e-05,
"IdentifyCrossPassTensors": 0.00011500000255182385,
"MemcastMotion": 0.0002789999998640269,
"PenguinizeFunctions": 0.0005789999850094318,
"PruneFunctions": 0.00018899999849963933,
"RemoveOptimizationBarriers": 0.0004290000069886446,
"ScatterMotion": 0.005127000156790018,
"TensorizerLegalizationPass": 0.000455000001238659,
"VerifySupportedOps": 0.0002469999890308827,
"algsimp": 0.0018210000125691295,
"batchnorm_expander": 0.0022700000554323196,
"boundary-marker-removal": 0.0009800000116229057,
"call-inliner": 0.0002410000015515834,
"canonicalize-boundary-marker": 0.0009430000209249556,
"collective-stream-id-checker": 0.0008960000122897327,
"comparison-expander": 0.000623999978415668,
"computation-deduplicator": 0.003625999903306365,
"config-lowering": 0.00017899999511428177,
"constant_folding": 0.0668639987707138,
"cse": 0.001184999942779541,
"dce": 6.399999983841553e-05,
"dynamic-slice-transpose": 0.0003530000103637576,
"eliminate-redundant-compare": 0.00014400000509340316,
"emit-offloaded-dropout": 0.0007820000173524022,
"flatten-call-graph": 0.0009519999730400741,
"fuse-send-recv": 0.07950299978256226,
"hilo-conditional-to-select": 0.00015700000221841037,
"hilo::LegalizeAlias": 0.003444999922066927,
"hilo::NeuronInstCombine": 0.0008500000112690032,
"hilo::NeuronOpFusion": 0.00020799999765586108,
"hilo::ReplaceTokenTypeWithU8Pass": 0.0007050000131130219,
"hilo::ScheduleFusion": 4.3000000005122274e-05,
"hilo::SixtyFourHack": 0.0005150000215508044,
"hilo::VerifyAliasing": 0.00035099999513477087,
"hlo-mac-count": 0.08947300165891647,
"io-con-pipe-begin": 6.199999916134402e-05,
"io-con-pipe-end": 9.999999974752427e-07,
"io-layout-normalization": 0.0706389993429184,
"legalize-ccops-for-tensorizer": 0.00013800000306218863,
"legalize-compare": 0.00027200000477023423,
"lower-argminmax-custom-call": 0.0004130000015720725,
"map-inline": 0.0010939999483525753,
"metadata-naming": 0.009581999853253365,
"mlir::detail::OpToOpPassAdaptor": 0.0004569999873638153,
"mlir::hlo::MhloToPyPenguin": 0.5751969814300537,
"mlir::mhlo::LowerComplexExtraPass": 0.00508299982175231,
"mlir::mhlo::LowerComplexPass": 0.0012469999492168427,
"native-to-custom-softmax": 0.0004349999944679439,
"native-to-custom-softmax-dx": 0.00044800000614486635,
"neuron-hlo-verifier": 0.3175100088119507,
"operand_upcaster": 0.0018560000462457538,
"post-par-pipe-begin": 9.999999974752427e-07,
"post-par-pipe-end": 0.0,
"post-partition-simplification": 0.48272499442100525,
"pre-hlo-begin": 1.4000000192027073e-05,
"pre-hlo-end": 1.9999999949504854e-06,
"replace-minimum-constant": 0.00022400000307243317,
"reshape-mover": 5.700000110664405e-05,
"simplify-concat": 0.002764000091701746,
"simplify-while-loops": 9.40000027185306e-05,
"transform-variadic-reduce": 0.001218999968841672,
"tuple-simplifier": 0.00048099999548867345,
"unpack-nested-aws-ntwsr": 0.0007040000054985285,
"unroll-while-loop": 1.2000000424450263e-05
}
},
"cumsum": {
"compiletime": {
"CoalesceCCOp": 0.0002155303955078125,
"DMALocalityOpt": 0.00016641616821289063,
"DMAProfiler": 0.0007414817810058594,
"DataStreaming": 0.0002970695495605469,
"DoNothing": 0.0001289844512939453,
"ExpandISAMacro": 0.0005061626434326172,
"FactorizeBlkDims": 0.0004284381866455078,
"InferPSumTensor": 0.0005972385406494141,
"InferSharedMemLoc": 0.0002605915069580078,
"InsertCoreBarrier": 0.000263214111328125,
"LateLegalizeInst": 0.00037741661071777344,
"LateNeuronInstComb": 0.0005505084991455078,
"LegalizeSundaAccess": 0.0013835430145263672,
"LegalizeType": 0.00022840499877929688,
"LowerBroadcast": 0.0002124309539794922,
"LowerIntrinsics": 0.0002105236053466797,
"LowerTranspose": 0.00021958351135253906,
"NeuronInstComb": 0.000583648681640625,
"NeuronLICM": 0.00035643577575683594,
"NeuronSimplifyPredicates": 0.002285003662109375,
"NeuronValueNumbering": 0.0004379749298095703,
"SFKVectorizer": 0.0024003982543945313,
"SimpleAllReduceTiling": 0.00020265579223632813,
"SimplifyNeuronTensor": 0.0004928112030029297,
"SpillPSum": 0.00045371055603027344,
"WeightCoalescing": 0.0002079010009765625
}
},
"sg00": {
"hilo": {
"ArithmeticIntensity": 2.867593288421631,
"HloMacCount": 1796005888.0,
"Traffic": 1252622464.0
}
},
"sg0000": {
"compiletime": {
"AGOrderingAnalysisPass": 6.775559425354004,
"AffinePredicateResolution": 0.7124242782592773,
"AliasDependencyElimination": 0.0037508010864257813,
"AliasDependencyInduction": 17.88080596923828,
"AliasDependencyReset": 19.01345443725586,
"BFComputeCutting": 0.14321279525756836,
"BirCodeGenLoop": 4.049878120422363,
"CCOpFusion": 0.9675219058990479,
"CanonicalizeDAGForPGTiling": 0.18202924728393555,
"CanonicalizeIR": 0.8085732460021973,
"CoalesceCCOp": 0.2558913230895996,
"CommuteConcat": 0.027615070343017578,
"DMALocalityOpt": 0.04517340660095215,
"DMAProfiler": 0.08888816833496094,
"DMATilingProfiler": 0.09855914115905762,
"DataLocalityOpt": 2.8309335708618164,
"DataStreaming": 0.29266810417175293,
"DeConcat": 0.08213448524475098,
"DeadCodeElimination": 0.030428647994995117,
"DeadStoreElimination": 1.051145315170288,
"DelinearIndices": 0.5529322624206543,
"Delinearization": 0.14493751525878906,
"DelinearizeSPMD": 0.181105375289917,
"DoNothing": 7.724761962890625e-05,
"DramToDramTranspose": 0.5176758766174316,
"DumpGraphAndMetadata": 0.1749565601348877,
"EliminateDivs": 1.6087908744812012,
"ExpandBatchNorm": 1.096595048904419,
"ExpandISAMacro": 0.08580923080444336,
"FactorizeBlkDims": 0.49906015396118164,
"FactorizeThreadAxesInFreeDims": 0.08890652656555176,
"FlattenMacroLoop": 0.08152222633361816,
"GenericAccessSimplifier": 0.02545022964477539,
"InferInitValue": 1.2978432178497314,
"InferIntrinsicOnCC": 0.7616860866546631,
"InferNeuronTensor": 2.6053829193115234,
"InferNonlocalTensors": 7.113053321838379,
"InferPSumTensor": 4.437821388244629,
"InferShardAxis": 12.849608421325684,
"InferSharedMemLoc": 0.18901801109313965,
"InlineNativeKernels": 0.08442234992980957,
"InsertCoreBarrier": 0.1489267349243164,
"InsertIOTransposes": 0.8695223331451416,
"InsertImplicitShardAxisBeforeISel": 0.49620485305786133,
"InsertLocalTransposes": 1.934779167175293,
"InsertOffloadedTransposes": 0.23423099517822266,
"LICM": 0.1213834285736084,
"LateLegalizeInst": 0.2814202308654785,
"LateLegalizePostSplit": 0.22411513328552246,
"LateLowerReshapeOp": 0.033789873123168945,
"LateLowerTensorOp": 4.297260761260986,
"LateNeuronInstComb": 1.014098882675171,
"LayoutPreprocessing": 1.8572075366973877,
"LayoutPreprocessingAndAnalysis": 2.2904410362243652,
"LayoutRequirementAnalysis": 0.4237644672393799,
"LegalizeCCOpLayout": 1.1935856342315674,
"LegalizeOpLevelAlias": 0.7793729305267334,
"LegalizePartitionReduce": 0.09432244300842285,
"LegalizeSundaAccess": 0.972252607345581,
"LegalizeSundaMacro": 0.7198967933654785,
"LegalizeType": 0.16599535942077637,
"LocalLayoutOpt": 0.6666848659515381,
"LoopFusion": 0.29988932609558105,
"LoopSplitting": 0.05208778381347656,
"LowerBroadcast": 0.07862257957458496,
"LowerCCOpBlockAxis": 0.2862672805786133,
"LowerComplexBroadcast": 0.09356045722961426,
"LowerIntrinsics": 1.3567829132080078,
"LowerShardAxis": 0.39397311210632324,
"LowerTensorOp": 4.018380165100098,
"LowerToSendRecv": 0.27350616455078125,
"LowerTranspose": 0.513293981552124,
"MacroGeneration": 2.0922586917877197,
"MaskPropagation": 0.12857651710510254,
"MemcpyElimination": 30.280338287353516,
"MutateDataType": 0.03657245635986328,
"NeuronAliasDependencyInduction": 0.026292085647583008,
"NeuronAliasDependencyReset": 0.03253936767578125,
"NeuronInstComb": 0.35875773429870605,
"NeuronLICM": 0.29233455657958984,
"NeuronLoopFusion": 1.711080551147461,
"NeuronLoopInterchange": 0.08134770393371582,
"NeuronSimplifier": 0.5104148387908936,
"NeuronSimplifyPredicates": 0.2326216697692871,
"NeuronValueNumbering": 0.11057472229003906,
"OptimizeAliasedCopyChain": 0.40732264518737793,
"OptimizeNKIKernels": 1.0493371486663818,
"PAGLayoutOpt": 18.23680305480957,
"PComputeCutting": 0.5901892185211182,
"PGLayoutTilingPipeline": 54.30891036987305,
"PGTiling": 10.472368240356445,
"PadElimination": 0.0187380313873291,
"ParAxesAnnotation": 16.293506622314453,
"PartialLoopFusion": 1.5408625602722168,
"PartialSimdFusion": 0.8349573612213135,
"PerfectLoopNest": 0.07371997833251953,
"RecognizeOpIdiom": 0.12681221961975098,
"Recompute": 0.007548809051513672,
"RelaxPredicates": 0.11768078804016113,
"Rematerialization": 0.15080475807189941,
"RemoveShardedPartitionAxes": 1.1734652519226074,
"ReshapeWeights": 0.022508859634399414,
"ResolveAccessConflict": 0.20145153999328613,
"ResolveComplicatePredicates": 0.3191838264465332,
"RewriteReplicationMatmul": 0.039293527603149414,
"RewriteWeights": 0.0773766040802002,
"SFKVectorizer": 11.047651290893555,
"ShardingPropagationAnalysis": 0.665086030960083,
"SimpleAllReduceTiling": 0.07064080238342285,
"Simplifier": 0.09160900115966797,
"SimplifyMacroPredicates": 0.28479433059692383,
"SimplifyNeuronTensor": 0.4839820861816406,
"SimplifySlice": 0.02684783935546875,
"SimplifyTensor": 0.2644212245941162,
"SpillPSum": 1.035245418548584,
"SplitAPUnionSets": 1.0491511821746826,
"SplitAccGrp": 0.055588722229003906,
"StaticProfiler": 0.39172840118408203,
"StaticTransposeLocalTensor": 0.7211995124816895,
"SundaISel": 1.5357580184936523,
"TCTransform": 0.029030561447143555,
"TensorInitialization": 0.20336008071899414,
"TensorOpSimplifier": 3.100177049636841,
"TensorOpTransform": 20.175758361816406,
"TileCCOps": 0.17463970184326172,
"TilingProfiler": 0.7014575004577637,
"TransformConvOp": 0.9923229217529297,
"TritiumFusion": 0.33277034759521484,
"ValueNumbering": 0.08796954154968262,
"VectorizeDMA": 0.520815372467041,
"VectorizeMatMult": 0.08555078506469727,
"WeightCoalescing": 0.08340644836425781,
"ZeroSizeTensorElimination": 0.0006709098815917969
},
"tensorizer": {
"DMATilingProfiler::TotalInstructionsAfterTiling": 45730,
"StaticProfiler::AifUb": 19.087068557739258,
"StaticProfiler::ArithmeticIntensityTensorizer": 29.72186851501465,
"StaticProfiler::AverageDmaLength": 2694.496826171875,
"StaticProfiler::AverageFractalPeUtilization": 97.78141784667969,
"StaticProfiler::AveragePartitionUtilization": 90.33787536621094,
"StaticProfiler::AveragePeUtilization": 81.002685546875,
"StaticProfiler::DDRTransferBytes": 954289776,
"StaticProfiler::InternalTransferBytes": 196796000,
"StaticProfiler::LoadExpanded": 281500,
"StaticProfiler::LocalizationEfficiency": 155.71730041503906,
"StaticProfiler::LocalizationEfficiencyIgnoreNonlocal": 160.65768432617188,
"StaticProfiler::StoreExpanded": 4273,
"StaticProfiler::TotalDMAExpanded": 285773,
"StaticProfiler::TotalDynamicInstancesCount": 62161,
"StaticProfiler::TotalDynamicInstancesWithMmPackedCount": 53675,
"StaticProfiler::TotalLNCComm": 0,
"StaticProfiler::TotalLNCCommTransfer": 0,
"TilingProfiler::AveragePartitionUtilizationAfterTiling": 0,
"TilingProfiler::AveragePeUtilizationAfterTiling": 0,
"TilingProfiler::BatchnormInstructionsAfterTiling": 0,
"TilingProfiler::DmaInstructionsAfterTiling": 0,
"TilingProfiler::GenericInstructionsAfterTiling": 123,
"TilingProfiler::MatMultInstructionsAfterTiling": 32320,
"TilingProfiler::NumPfTransposes": 292,
"TilingProfiler::NumPfTransposesForIo": 30,
"TilingProfiler::NumPfTransposesForLocal": 142,
"TilingProfiler::NumPfTransposesForNonlocal": 120,
"TilingProfiler::PfTransposeInstructions": 8762,
"TilingProfiler::PfTransposeInstructionsForIo": 6564,
"TilingProfiler::PfTransposeInstructionsForLocal": 536,
"TilingProfiler::PfTransposeInstructionsForNonlocal": 1662,
"TilingProfiler::ReduceInstructionsAfterTiling": 171,
"TilingProfiler::SimdInstructionsAfterTiling": 2465,
"TilingProfiler::TotalInstructionsAfterTiling": 0,
"TransformConvOp::Conv1d_depthwise_bf01_oi01_bf01": 0,
"TransformConvOp::Conv2d_dw_fb01_io01_01bf_rep_nhwc_Pcinh": 0,
"TransformConvOp::Conv2d_pbp_0f1b_0i1o_01fb_experimental_1": 0,
"TransformConvOp::Conv2d_pbp_fb01_io01_01bf_experimental_1": 0,
"TransformConvOp::conv2d_column_packing": 0,
"TransformConvOp::conv2d_column_packing_1": 0,
"TransformConvOp::conv2d_column_packing_io10": 0,
"TransformConvOp::conv2d_depthwise_f01b_o01i_bf01": 0
}
},
"topk": {
"compiletime": {
"CoalesceCCOp": 0.003664731979370117,
"DMALocalityOpt": 0.002711057662963867,
"DMAProfiler": 0.003767251968383789,
"DataStreaming": 0.0067331790924072266,
"DoNothing": 0.00014734268188476563,
"ExpandISAMacro": 0.006513833999633789,
"FactorizeBlkDims": 0.012669801712036133,
"InferPSumTensor": 0.012490510940551758,
"InferSharedMemLoc": 0.0029754638671875,
"InsertCoreBarrier": 0.00330352783203125,
"LateLegalizeInst": 0.007447957992553711,
"LateNeuronInstComb": 0.008418560028076172,
"LegalizeSundaAccess": 0.02680182456970215,
"LegalizeType": 0.010413169860839844,
"LowerBroadcast": 0.0033206939697265625,
"LowerIntrinsics": 0.003704547882080078,
"LowerTranspose": 0.0035905838012695313,
"NeuronInstComb": 0.008339166641235352,
"NeuronLICM": 0.009230852127075195,
"NeuronSimplifyPredicates": 0.006308555603027344,
"NeuronValueNumbering": 0.003956317901611328,
"SFKVectorizer": 0.03573727607727051,
"SimpleAllReduceTiling": 0.0036857128143310547,
"SimplifyNeuronTensor": 0.054434776306152344,
"SpillPSum": 0.02359294891357422,
"WeightCoalescing": 0.0065898895263671875
}
}
}

3
token_generation_model/_tp0_bk2/graph.neff Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:2e9104a1923d9a888389573352849a655d6f7005c5782380d0fa3306dab803f3
size 3390464

4468
token_generation_model/_tp0_bk2/log-neuron-cc.txt Normal file
View File

File diff suppressed because it is too large Load Diff

3
token_generation_model/_tp0_bk2/metaneff.pb Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:a4a34ac27ace8c75df52c6f7fb5e0c1736f327638e3ac42ff906c6b52e125a5e
size 2464555

3
token_generation_model/_tp0_bk2/model.MODULE_c60e20a111715e620aa2+6bb8acc9.hlo_module.pb Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:0baed4147c67e64b22e5d725b3e55d70e3e16ed2fe838758927f1cb592ebdc7a
size 2550451

3
token_generation_model/_tp0_bk2/model.MODULE_c60e20a111715e620aa2+6bb8acc9.neff Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:2e9104a1923d9a888389573352849a655d6f7005c5782380d0fa3306dab803f3
size 3390464

224
token_generation_model/_tp0_bk2/neuron_config.json Normal file
View File

@@ -0,0 +1,224 @@
{
"_attn_implementation_autoset": false,
"_name_or_path": "/models/Qwen3-1.7B/",
"add_cross_attention": false,
"architectures": [
"Qwen3ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"attribute_map": {},
"bad_words_ids": null,
"begin_suppress_tokens": null,
"bos_token_id": 151643,
"chunk_size_feed_forward": 0,
"cross_attention_hidden_size": null,
"decoder_start_token_id": null,
"diversity_penalty": 0.0,
"do_sample": false,
"early_stopping": false,
"encoder_no_repeat_ngram_size": 0,
"eos_token_id": 151645,
"exponential_decay_length_penalty": null,
"finetuning_task": null,
"forced_bos_token_id": null,
"forced_eos_token_id": null,
"fused_spec_config": null,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 2048,
"id2label": {
"0": "LABEL_0",
"1": "LABEL_1"
},
"initializer_range": 0.02,
"intermediate_size": 6144,
"is_decoder": false,
"is_encoder_decoder": false,
"label2id": {
"LABEL_0": 0,
"LABEL_1": 1
},
"length_penalty": 1.0,
"max_length": 20,
"max_position_embeddings": 40960,
"max_window_layers": 28,
"metadata": null,
"min_length": 0,
"model_type": "qwen3",
"neuron_config": {
"activation_quantization_type": null,
"allow_input_truncation": false,
"apply_seq_ids_mask": false,
"async_mode": false,
"attention_dp_degree": 1,
"attention_dtype": null,
"attn_block_cte_nki_kernel_enabled": false,
"attn_block_tkg_nki_kernel_cache_update": false,
"attn_block_tkg_nki_kernel_cascaded_attention": false,
"attn_block_tkg_nki_kernel_enabled": false,
"attn_cls": {
"__module__": "neuronx_distributed_inference.models.qwen3.modeling_qwen3",
"__name__": "NeuronQwen3Attention"
},
"attn_kernel_enabled": null,
"attn_tkg_builtin_kernel_enabled": false,
"attn_tkg_nki_kernel_enabled": false,
"batch_size": 4,
"bucket_n_active_tokens": false,
"buckets": [
512
],
"cast_type": "config",
"cc_pipeline_tiling_factor": 1,
"chunked_prefill_config": null,
"context_encoding_buckets": null,
"cp_degree": 1,
"ctx_batch_size": 1,
"disable_kv_cache_tiling": false,
"draft_model_modules_to_not_convert": null,
"enable_bucketing": true,
"enable_cte_modular_flow": false,
"enable_eagle_draft_input_norm": false,
"enable_eagle_speculation": false,
"enable_fused_speculation": false,
"enable_long_context_mode": false,
"enable_output_completion_notifications": false,
"enable_spill_reload_dge": false,
"enable_token_tree": false,
"ep_degree": 1,
"expert_mlp_nki_kernel_enabled": null,
"flash_decoding_enabled": false,
"fused_qkv": false,
"fused_rmsnorm_skip_gamma": false,
"is_block_kv_layout": null,
"is_chunked_prefill": false,
"is_continuous_batching": true,
"is_eagle_draft": false,
"is_medusa": false,
"is_prefill_stage": false,
"is_prefix_caching": false,
"k_cache_transposed": false,
"kv_cache_batch_size": 4,
"kv_cache_padding_size": 0,
"kv_cache_quant": false,
"kv_cache_tiling": false,
"layer_boundary_markers": false,
"lm_head_pad": true,
"lm_head_pad_alignment_size": 1,
"local_ranks_size": 4,
"logical_nc_config": 2,
"lora_config": null,
"max_batch_size": 4,
"max_context_length": 2048,
"max_length": 2048,
"max_new_tokens": null,
"medusa_speculation_length": 0,
"medusa_tree": null,
"mlp_kernel_enabled": false,
"mlp_kernel_fuse_residual_add": false,
"modules_to_not_convert": null,
"moe_fused_nki_kernel_enabled": null,
"n_active_tokens": 1,
"n_positions": 2048,
"num_medusa_heads": 0,
"on_cpu": false,
"on_device_sampling_config": {
"deterministic": false,
"do_sample": false,
"dynamic": true,
"global_topk": 256,
"on_device_sampling_config": true,
"temperature": 1.0,
"top_k": 1,
"top_k_kernel_enabled": false,
"top_p": 1.0
},
"output_logits": false,
"overrides_torch_dtype": true,
"pa_block_size": 2048,
"pa_num_blocks": 4,
"padding_side": "right",
"pp_degree": 1,
"prefix_buckets": null,
"qk_layernorm": false,
"qkv_kernel_enabled": false,
"qkv_kernel_fuse_residual_add": false,
"qkv_kernel_nbsd_layout": false,
"quantization_dtype": "int8",
"quantization_type": "per_tensor_symmetric",
"quantize_clamp_bound": Infinity,
"quantized": false,
"quantized_checkpoints_path": null,
"quantized_mlp_kernel_enabled": false,
"rmsnorm_quantize_kernel_enabled": false,
"router_topk_nki_kernel_enabled": null,
"rpl_reduce_dtype": null,
"save_sharded_checkpoint": true,
"scratchpad_page_size": null,
"seq_len": 2048,
"seq_len_threshold_for_cc_tiling": 16384,
"sequence_parallel_enabled": false,
"shared_mlp_nki_kernel_enabled": null,
"skip_sharding": false,
"skip_warmup": false,
"spec_batch_size": 4,
"speculation_length": 0,
"start_rank_id": 0,
"strided_context_parallel_kernel_enabled": false,
"target": null,
"tensor_capture_config": null,
"tile_cc": false,
"tkg_batch_size": 4,
"token_generation_buckets": [
512
],
"token_tree_config": null,
"torch_dtype": "bfloat16",
"tp_degree": 4,
"vocab_parallel": false,
"weight_gather_seq_len_threshold": 32768,
"weights_to_skip_layout_optimization": [],
"world_size": 4
},
"no_repeat_ngram_size": 0,
"num_attention_heads": 16,
"num_beam_groups": 1,
"num_beams": 1,
"num_cores_per_group": 1,
"num_hidden_layers": 28,
"num_key_value_heads": 8,
"num_return_sequences": 1,
"output_attentions": false,
"output_hidden_states": false,
"output_scores": false,
"pad_token_id": 0,
"prefix": null,
"problem_type": null,
"pruned_heads": {},
"remove_invalid_values": false,
"repetition_penalty": 1.0,
"return_dict": true,
"return_dict_in_generate": false,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rope_theta": 1000000,
"sep_token_id": null,
"sliding_window": null,
"suppress_tokens": null,
"task_specific_params": null,
"temperature": 1.0,
"tf_legacy_loss": false,
"tie_encoder_decoder": false,
"tie_word_embeddings": true,
"tokenizer_class": null,
"top_k": 50,
"top_p": 1.0,
"torchscript": false,
"transformers_version": "4.51.0",
"typical_p": 1.0,
"use_bfloat16": false,
"use_cache": true,
"use_sliding_window": false,
"vocab_size": 151936
}