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_bk0/command.txt Normal file
View File

@@ -0,0 +1 @@
neuronx-cc compile --framework=XLA model.MODULE_5e3d0ce8963512f9ea2d+937cd7a2.hlo_module.pb --output model.MODULE_5e3d0ce8963512f9ea2d+937cd7a2.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 --enable-internal-neff-wrapper --verbose=35

1
token_generation_model/_tp0_bk0/compile_flags.MODULE_5e3d0ce8963512f9ea2d+937cd7a2.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_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]

590
token_generation_model/_tp0_bk0/global_metric_store.json Normal file
View File

@@ -0,0 +1,590 @@
{
"Average": {
"tensorizer": {
"StaticProfiler::AverageFractalPeUtilization": 96.97119140625,
"StaticProfiler::AveragePartitionUtilization": 88.53791809082031,
"StaticProfiler::AveragePeUtilization": 81.30671691894531,
"StaticProfiler::LocalizationEfficiency": 161.8649139404297,
"StaticProfiler::LocalizationEfficiencyIgnoreNonlocal": 167.2097930908203,
"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": 2.377948760986328,
"AffinePredicateResolution": 0.03525829315185547,
"AliasDependencyElimination": 0.0024094581604003906,
"AliasDependencyInduction": 0.2979590892791748,
"AliasDependencyReset": 0.30464816093444824,
"BFComputeCutting": 0.07316970825195313,
"BirCodeGenLoop": 2.482311487197876,
"CCOpFusion": 0.5325038433074951,
"CanonicalizeConv": 0.00023499999952036887,
"CanonicalizeDAGForPGTiling": 0.1513075828552246,
"CanonicalizeForTensorizer": 0.00026500000967644155,
"CanonicalizeIR": 0.04851794242858887,
"Canonicalizer": 0.004513999912887812,
"CoalesceCCOp": 0.16019058227539063,
"CommuteConcat": 0.024104833602905273,
"DMALocalityOpt": 0.037362098693847656,
"DMAProfiler": 0.07356810569763184,
"DMATilingProfiler": 0.09021782875061035,
"DataLocalityOpt": 3.034395694732666,
"DataStreaming": 0.1286299228668213,
"DeConcat": 0.04242873191833496,
"DeadCodeElimination": 0.02465653419494629,
"DeadStoreElimination": 0.8050529956817627,
"DelinearIndices": 0.46034955978393555,
"Delinearization": 0.11224937438964844,
"DelinearizeSPMD": 0.14354729652404785,
"DoNothing": 0.00033402442932128906,
"DramToDramTranspose": 0.2800898551940918,
"DumpGraphAndMetadata": 0.14458084106445313,
"EliminateDivs": 0.11128425598144531,
"ExpandBatchNorm": 0.04912686347961426,
"ExpandISAMacro": 0.07862186431884766,
"FactorizeBlkDims": 0.49788737297058105,
"FactorizeThreadAxesInFreeDims": 0.05206179618835449,
"FlattenMacroLoop": 0.08013129234313965,
"GenericAccessSimplifier": 0.022362232208251953,
"HoistCompute": 3.899999865097925e-05,
"IdentifyCrossPassTensors": 0.00022899999748915434,
"InferInitValue": 1.3044743537902832,
"InferIntrinsicOnCC": 0.26564502716064453,
"InferNeuronTensor": 1.4273626804351807,
"InferNonlocalTensors": 3.070617198944092,
"InferPSumTensor": 1.877610206604004,
"InferShardAxis": 4.046135902404785,
"InferSharedMemLoc": 0.10418868064880371,
"InlineNativeKernels": 0.046288251876831055,
"InsertCoreBarrier": 0.12113451957702637,
"InsertIOTransposes": 1.5809462070465088,
"InsertImplicitShardAxisBeforeISel": 0.37781310081481934,
"InsertLocalTransposes": 0.8543226718902588,
"InsertOffloadedTransposes": 0.0779869556427002,
"LICM": 0.10844612121582031,
"LateLegalizeInst": 0.1468517780303955,
"LateLegalizePostSplit": 0.08665323257446289,
"LateLowerReshapeOp": 0.02942204475402832,
"LateLowerTensorOp": 0.22415471076965332,
"LateNeuronInstComb": 0.9669840335845947,
"LayoutPreprocessing": 0.7032866477966309,
"LayoutPreprocessingAndAnalysis": 1.1142585277557373,
"LayoutRequirementAnalysis": 0.40787601470947266,
"LegalizeCCOpLayout": 0.052059173583984375,
"LegalizeOpLevelAlias": 0.022985219955444336,
"LegalizePartitionReduce": 0.036779165267944336,
"LegalizeSundaAccess": 0.9409115314483643,
"LegalizeSundaMacro": 0.7163646221160889,
"LegalizeType": 0.1376960277557373,
"LocalLayoutOpt": 0.5141463279724121,
"LoopFusion": 0.2369997501373291,
"LoopSplitting": 0.026223182678222656,
"LowerBroadcast": 0.04980111122131348,
"LowerCCOpBlockAxis": 0.17850804328918457,
"LowerComplexBroadcast": 0.06136465072631836,
"LowerIntrinsics": 0.9023723602294922,
"LowerShardAxis": 0.19546747207641602,
"LowerTensorOp": 0.38861823081970215,
"LowerToSendRecv": 0.1479358673095703,
"LowerTranspose": 0.41634607315063477,
"MacroGeneration": 1.907278060913086,
"MaskPropagation": 0.10989689826965332,
"MemcastMotion": 0.00011899999663000926,
"MemcpyElimination": 3.3127715587615967,
"MutateDataType": 0.031205177307128906,
"NeuronAliasDependencyInduction": 0.014397859573364258,
"NeuronAliasDependencyReset": 0.018595218658447266,
"NeuronInstComb": 0.29677605628967285,
"NeuronLICM": 0.2591841220855713,
"NeuronLoopFusion": 0.9968361854553223,
"NeuronLoopInterchange": 0.049851179122924805,
"NeuronSimplifier": 0.4482152462005615,
"NeuronSimplifyPredicates": 0.13344645500183105,
"NeuronValueNumbering": 0.10327720642089844,
"OptimizeAliasedCopyChain": 0.010796785354614258,
"OptimizeNKIKernels": 0.9786763191223145,
"PAGLayoutOpt": 11.095458984375,
"PComputeCutting": 0.28275370597839355,
"PGLayoutTilingPipeline": 27.67894172668457,
"PGTiling": 4.957591533660889,
"PadElimination": 0.009732246398925781,
"ParAxesAnnotation": 10.236833572387695,
"PartialLoopFusion": 1.0972201824188232,
"PartialSimdFusion": 0.529569149017334,
"PenguinizeFunctions": 0.00012799999967683107,
"PerfectLoopNest": 0.059952735900878906,
"PruneFunctions": 0.000391999987186864,
"RecognizeOpIdiom": 0.120391845703125,
"Recompute": 0.005687713623046875,
"RelaxPredicates": 0.09513998031616211,
"Rematerialization": 0.14099407196044922,
"RemoveOptimizationBarriers": 0.0004239999980200082,
"RemoveShardedPartitionAxes": 0.6145329475402832,
"ReshapeWeights": 0.0214080810546875,
"ResolveAccessConflict": 0.1656482219696045,
"ResolveComplicatePredicates": 0.03499007225036621,
"RewriteReplicationMatmul": 0.03796839714050293,
"RewriteWeights": 0.06430387496948242,
"SFKVectorizer": 5.783308029174805,
"ScatterMotion": 0.002827000105753541,
"ShardingPropagationAnalysis": 0.5885381698608398,
"SimpleAllReduceTiling": 0.05830836296081543,
"Simplifier": 0.07847452163696289,
"SimplifyMacroPredicates": 0.25547003746032715,
"SimplifyNeuronTensor": 0.3350214958190918,
"SimplifySlice": 0.02300572395324707,
"SimplifyTensor": 0.2370157241821289,
"SpillPSum": 0.4782223701477051,
"SplitAPUnionSets": 0.6796882152557373,
"SplitAccGrp": 0.03665971755981445,
"StaticProfiler": 0.13864398002624512,
"StaticTransposeLocalTensor": 0.2623753547668457,
"SundaISel": 1.3397884368896484,
"TCTransform": 0.025269269943237305,
"TensorInitialization": 0.16660237312316895,
"TensorOpSimplifier": 0.15259718894958496,
"TensorOpTransform": 0.8282039165496826,
"TensorizerLegalizationPass": 0.00011800000356743112,
"TileCCOps": 0.3066561222076416,
"TilingProfiler": 0.33243393898010254,
"TransformConvOp": 0.05726790428161621,
"TritiumFusion": 0.13151931762695313,
"ValueNumbering": 0.06717801094055176,
"VectorizeDMA": 0.5187788009643555,
"VectorizeMatMult": 0.04601097106933594,
"VerifySupportedOps": 0.000195999993593432,
"WeightCoalescing": 0.05271005630493164,
"ZeroSizeTensorElimination": 0.0003685951232910156,
"algsimp": 0.0011439999798312783,
"batchnorm_expander": 0.0005740000051446259,
"boundary-marker-removal": 0.00017699999443721026,
"call-inliner": 0.00013800000306218863,
"canonicalize-boundary-marker": 0.0003129999968223274,
"collective-stream-id-checker": 7.899999764049426e-05,
"comparison-expander": 0.000195999993593432,
"computation-deduplicator": 0.00031900001340545714,
"config-lowering": 0.00015100000018719584,
"constant_folding": 9.40000027185306e-05,
"cse": 0.00038400001358240843,
"dce": 2.4000000848900527e-05,
"dynamic-slice-transpose": 8.099999831756577e-05,
"eliminate-redundant-compare": 7.999999797903001e-05,
"emit-offloaded-dropout": 0.00013899999612476677,
"flatten-call-graph": 0.0001720000000204891,
"fuse-send-recv": 0.0008440000237897038,
"hilo-conditional-to-select": 5.2999999752501026e-05,
"hilo::LegalizeAlias": 0.0019950000569224358,
"hilo::NeuronInstCombine": 0.0008609999786131084,
"hilo::NeuronOpFusion": 0.0003239999932702631,
"hilo::ReplaceTokenTypeWithU8Pass": 0.00031400000443682075,
"hilo::ScheduleFusion": 2.9000000722589903e-05,
"hilo::SixtyFourHack": 0.00023200000578071922,
"hilo::VerifyAliasing": 5.0999999075429514e-05,
"hlo-mac-count": 0.006806999910622835,
"io-con-pipe-begin": 9.999999747378752e-06,
"io-con-pipe-end": 0.0,
"io-layout-normalization": 0.0014359999913722277,
"legalize-ccops-for-tensorizer": 1.4999999621068127e-05,
"legalize-compare": 0.000155999994603917,
"lower-argminmax-custom-call": 7.599999662488699e-05,
"map-inline": 0.0003760000108741224,
"metadata-naming": 0.0007949999999254942,
"mlir::detail::OpToOpPassAdaptor": 0.00014600000577047467,
"mlir::hlo::MhloToPyPenguin": 0.057739999145269394,
"mlir::mhlo::LowerComplexExtraPass": 0.0017610000213608146,
"mlir::mhlo::LowerComplexPass": 0.00240899994969368,
"native-to-custom-softmax": 0.00022000000171829015,
"native-to-custom-softmax-dx": 0.00023200000578071922,
"neuron-hlo-verifier": 0.01620600000023842,
"operand_upcaster": 0.0005629999795928597,
"post-par-pipe-begin": 9.999999974752427e-07,
"post-par-pipe-end": 0.0,
"post-partition-simplification": 0.03471200168132782,
"pre-hlo-begin": 3.000000106112566e-06,
"pre-hlo-end": 0.0,
"replace-minimum-constant": 0.00010599999950500205,
"reshape-mover": 4.5000000682193786e-05,
"simplify-concat": 0.0008580000139772892,
"simplify-while-loops": 3.300000025774352e-05,
"transform-variadic-reduce": 0.00025599999935366213,
"tuple-simplifier": 8.900000102585182e-05,
"unpack-nested-aws-ntwsr": 0.00014000000373926014,
"unroll-while-loop": 6.000000212225132e-06
},
"hilo": {
"HloMacCount": 1751965696.0,
"Traffic": 1252616320.0
},
"tensorizer": {
"DMATilingProfiler::TotalInstructionsAfterTiling": 41972,
"StaticProfiler::AifUb": 17.11549949645996,
"StaticProfiler::ArithmeticIntensityTensorizer": 27.70398712158203,
"StaticProfiler::AverageDmaLength": 2723.7119140625,
"StaticProfiler::DDRTransferBytes": 918042224,
"StaticProfiler::InternalTransferBytes": 174444896,
"StaticProfiler::LoadExpanded": 238040,
"StaticProfiler::StoreExpanded": 18609,
"StaticProfiler::TotalDMAExpanded": 256649,
"StaticProfiler::TotalDynamicInstancesCount": 57730,
"StaticProfiler::TotalDynamicInstancesWithMmPackedCount": 50535,
"StaticProfiler::TotalLNCComm": 0,
"StaticProfiler::TotalLNCCommTransfer": 0,
"TilingProfiler::BatchnormInstructionsAfterTiling": 0,
"TilingProfiler::DmaInstructionsAfterTiling": 0,
"TilingProfiler::GenericInstructionsAfterTiling": 179,
"TilingProfiler::MatMultInstructionsAfterTiling": 30304,
"TilingProfiler::NumPfTransposes": 348,
"TilingProfiler::NumPfTransposesForIo": 30,
"TilingProfiler::NumPfTransposesForLocal": 198,
"TilingProfiler::NumPfTransposesForNonlocal": 120,
"TilingProfiler::PfTransposeInstructions": 7835,
"TilingProfiler::PfTransposeInstructionsForIo": 5217,
"TilingProfiler::PfTransposeInstructionsForLocal": 508,
"TilingProfiler::PfTransposeInstructionsForNonlocal": 2110,
"TilingProfiler::ReduceInstructionsAfterTiling": 59,
"TilingProfiler::SimdInstructionsAfterTiling": 2266,
"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": 0.00023499999952036887,
"CanonicalizeForTensorizer": 0.00026500000967644155,
"Canonicalizer": 0.004513999912887812,
"HoistCompute": 3.899999865097925e-05,
"IdentifyCrossPassTensors": 0.00022899999748915434,
"MemcastMotion": 0.00011899999663000926,
"PenguinizeFunctions": 0.00012799999967683107,
"PruneFunctions": 0.000391999987186864,
"RemoveOptimizationBarriers": 0.0004239999980200082,
"ScatterMotion": 0.002827000105753541,
"TensorizerLegalizationPass": 0.00011800000356743112,
"VerifySupportedOps": 0.000195999993593432,
"algsimp": 0.0011439999798312783,
"batchnorm_expander": 0.0005740000051446259,
"boundary-marker-removal": 0.00017699999443721026,
"call-inliner": 0.00013800000306218863,
"canonicalize-boundary-marker": 0.0003129999968223274,
"collective-stream-id-checker": 7.899999764049426e-05,
"comparison-expander": 0.000195999993593432,
"computation-deduplicator": 0.00031900001340545714,
"config-lowering": 0.00015100000018719584,
"constant_folding": 9.40000027185306e-05,
"cse": 0.00038400001358240843,
"dce": 2.4000000848900527e-05,
"dynamic-slice-transpose": 8.099999831756577e-05,
"eliminate-redundant-compare": 7.999999797903001e-05,
"emit-offloaded-dropout": 0.00013899999612476677,
"flatten-call-graph": 0.0001720000000204891,
"fuse-send-recv": 0.0008440000237897038,
"hilo-conditional-to-select": 5.2999999752501026e-05,
"hilo::LegalizeAlias": 0.0019950000569224358,
"hilo::NeuronInstCombine": 0.0008609999786131084,
"hilo::NeuronOpFusion": 0.0003239999932702631,
"hilo::ReplaceTokenTypeWithU8Pass": 0.00031400000443682075,
"hilo::ScheduleFusion": 2.9000000722589903e-05,
"hilo::SixtyFourHack": 0.00023200000578071922,
"hilo::VerifyAliasing": 5.0999999075429514e-05,
"hlo-mac-count": 0.006806999910622835,
"io-con-pipe-begin": 9.999999747378752e-06,
"io-con-pipe-end": 0.0,
"io-layout-normalization": 0.0014359999913722277,
"legalize-ccops-for-tensorizer": 1.4999999621068127e-05,
"legalize-compare": 0.000155999994603917,
"lower-argminmax-custom-call": 7.599999662488699e-05,
"map-inline": 0.0003760000108741224,
"metadata-naming": 0.0007949999999254942,
"mlir::detail::OpToOpPassAdaptor": 0.00014600000577047467,
"mlir::hlo::MhloToPyPenguin": 0.057739999145269394,
"mlir::mhlo::LowerComplexExtraPass": 0.0017610000213608146,
"mlir::mhlo::LowerComplexPass": 0.00240899994969368,
"native-to-custom-softmax": 0.00022000000171829015,
"native-to-custom-softmax-dx": 0.00023200000578071922,
"neuron-hlo-verifier": 0.01620600000023842,
"operand_upcaster": 0.0005629999795928597,
"post-par-pipe-begin": 9.999999974752427e-07,
"post-par-pipe-end": 0.0,
"post-partition-simplification": 0.03471200168132782,
"pre-hlo-begin": 3.000000106112566e-06,
"pre-hlo-end": 0.0,
"replace-minimum-constant": 0.00010599999950500205,
"reshape-mover": 4.5000000682193786e-05,
"simplify-concat": 0.0008580000139772892,
"simplify-while-loops": 3.300000025774352e-05,
"transform-variadic-reduce": 0.00025599999935366213,
"tuple-simplifier": 8.900000102585182e-05,
"unpack-nested-aws-ntwsr": 0.00014000000373926014,
"unroll-while-loop": 6.000000212225132e-06
}
},
"cumsum": {
"compiletime": {
"CoalesceCCOp": 0.0002205371856689453,
"DMALocalityOpt": 0.00016808509826660156,
"DMAProfiler": 0.0006198883056640625,
"DataStreaming": 0.0002627372741699219,
"DoNothing": 0.00013875961303710938,
"ExpandISAMacro": 0.0005130767822265625,
"FactorizeBlkDims": 0.0005578994750976563,
"InferPSumTensor": 0.0005154609680175781,
"InferSharedMemLoc": 0.00028228759765625,
"InsertCoreBarrier": 0.00024247169494628906,
"LateLegalizeInst": 0.0003752708435058594,
"LateNeuronInstComb": 0.0005624294281005859,
"LegalizeSundaAccess": 0.0013544559478759766,
"LegalizeType": 0.00024437904357910156,
"LowerBroadcast": 0.00022649765014648438,
"LowerIntrinsics": 0.00021767616271972656,
"LowerTranspose": 0.0002415180206298828,
"NeuronInstComb": 0.0007607936859130859,
"NeuronLICM": 0.00041222572326660156,
"NeuronSimplifyPredicates": 0.002053976058959961,
"NeuronValueNumbering": 0.0004031658172607422,
"SFKVectorizer": 0.002403736114501953,
"SimpleAllReduceTiling": 0.0002071857452392578,
"SimplifyNeuronTensor": 0.0004944801330566406,
"SpillPSum": 0.0004985332489013672,
"WeightCoalescing": 0.000213623046875
}
},
"sg00": {
"hilo": {
"ArithmeticIntensity": 2.797290325164795,
"HloMacCount": 1751965696.0,
"Traffic": 1252616320.0
}
},
"sg0000": {
"compiletime": {
"AGOrderingAnalysisPass": 2.377948760986328,
"AffinePredicateResolution": 0.03525829315185547,
"AliasDependencyElimination": 0.0024094581604003906,
"AliasDependencyInduction": 0.2979590892791748,
"AliasDependencyReset": 0.30464816093444824,
"BFComputeCutting": 0.07316970825195313,
"BirCodeGenLoop": 2.482311487197876,
"CCOpFusion": 0.5325038433074951,
"CanonicalizeDAGForPGTiling": 0.1513075828552246,
"CanonicalizeIR": 0.04851794242858887,
"CoalesceCCOp": 0.1563706398010254,
"CommuteConcat": 0.024104833602905273,
"DMALocalityOpt": 0.03418087959289551,
"DMAProfiler": 0.06951689720153809,
"DMATilingProfiler": 0.09021782875061035,
"DataLocalityOpt": 3.034395694732666,
"DataStreaming": 0.12200498580932617,
"DeConcat": 0.04242873191833496,
"DeadCodeElimination": 0.02465653419494629,
"DeadStoreElimination": 0.8050529956817627,
"DelinearIndices": 0.46034955978393555,
"Delinearization": 0.11224937438964844,
"DelinearizeSPMD": 0.14354729652404785,
"DoNothing": 6.198883056640625e-05,
"DramToDramTranspose": 0.2800898551940918,
"DumpGraphAndMetadata": 0.14458084106445313,
"EliminateDivs": 0.11128425598144531,
"ExpandBatchNorm": 0.04912686347961426,
"ExpandISAMacro": 0.07448863983154297,
"FactorizeBlkDims": 0.48647499084472656,
"FactorizeThreadAxesInFreeDims": 0.05206179618835449,
"FlattenMacroLoop": 0.08013129234313965,
"GenericAccessSimplifier": 0.022362232208251953,
"InferInitValue": 1.3044743537902832,
"InferIntrinsicOnCC": 0.26564502716064453,
"InferNeuronTensor": 1.4273626804351807,
"InferNonlocalTensors": 3.070617198944092,
"InferPSumTensor": 1.8667116165161133,
"InferShardAxis": 4.046135902404785,
"InferSharedMemLoc": 0.10098028182983398,
"InlineNativeKernels": 0.046288251876831055,
"InsertCoreBarrier": 0.1175682544708252,
"InsertIOTransposes": 1.5809462070465088,
"InsertImplicitShardAxisBeforeISel": 0.37781310081481934,
"InsertLocalTransposes": 0.8543226718902588,
"InsertOffloadedTransposes": 0.0779869556427002,
"LICM": 0.10844612121582031,
"LateLegalizeInst": 0.13918232917785645,
"LateLegalizePostSplit": 0.08665323257446289,
"LateLowerReshapeOp": 0.02942204475402832,
"LateLowerTensorOp": 0.22415471076965332,
"LateNeuronInstComb": 0.9580295085906982,
"LayoutPreprocessing": 0.7032866477966309,
"LayoutPreprocessingAndAnalysis": 1.1142585277557373,
"LayoutRequirementAnalysis": 0.40787601470947266,
"LegalizeCCOpLayout": 0.052059173583984375,
"LegalizeOpLevelAlias": 0.022985219955444336,
"LegalizePartitionReduce": 0.036779165267944336,
"LegalizeSundaAccess": 0.9256985187530518,
"LegalizeSundaMacro": 0.7163646221160889,
"LegalizeType": 0.12823128700256348,
"LocalLayoutOpt": 0.5141463279724121,
"LoopFusion": 0.2369997501373291,
"LoopSplitting": 0.026223182678222656,
"LowerBroadcast": 0.04629969596862793,
"LowerCCOpBlockAxis": 0.17850804328918457,
"LowerComplexBroadcast": 0.06136465072631836,
"LowerIntrinsics": 0.898535966873169,
"LowerShardAxis": 0.19546747207641602,
"LowerTensorOp": 0.38861823081970215,
"LowerToSendRecv": 0.1479358673095703,
"LowerTranspose": 0.4128274917602539,
"MacroGeneration": 1.907278060913086,
"MaskPropagation": 0.10989689826965332,
"MemcpyElimination": 3.3127715587615967,
"MutateDataType": 0.031205177307128906,
"NeuronAliasDependencyInduction": 0.014397859573364258,
"NeuronAliasDependencyReset": 0.018595218658447266,
"NeuronInstComb": 0.28766393661499023,
"NeuronLICM": 0.24966120719909668,
"NeuronLoopFusion": 0.9968361854553223,
"NeuronLoopInterchange": 0.049851179122924805,
"NeuronSimplifier": 0.4482152462005615,
"NeuronSimplifyPredicates": 0.12775635719299316,
"NeuronValueNumbering": 0.09903788566589355,
"OptimizeAliasedCopyChain": 0.010796785354614258,
"OptimizeNKIKernels": 0.9786763191223145,
"PAGLayoutOpt": 11.095458984375,
"PComputeCutting": 0.28275370597839355,
"PGLayoutTilingPipeline": 27.67894172668457,
"PGTiling": 4.957591533660889,
"PadElimination": 0.009732246398925781,
"ParAxesAnnotation": 10.236833572387695,
"PartialLoopFusion": 1.0972201824188232,
"PartialSimdFusion": 0.529569149017334,
"PerfectLoopNest": 0.059952735900878906,
"RecognizeOpIdiom": 0.120391845703125,
"Recompute": 0.005687713623046875,
"RelaxPredicates": 0.09513998031616211,
"Rematerialization": 0.14099407196044922,
"RemoveShardedPartitionAxes": 0.6145329475402832,
"ReshapeWeights": 0.0214080810546875,
"ResolveAccessConflict": 0.1656482219696045,
"ResolveComplicatePredicates": 0.03499007225036621,
"RewriteReplicationMatmul": 0.03796839714050293,
"RewriteWeights": 0.06430387496948242,
"SFKVectorizer": 5.7498369216918945,
"ShardingPropagationAnalysis": 0.5885381698608398,
"SimpleAllReduceTiling": 0.05400824546813965,
"Simplifier": 0.07847452163696289,
"SimplifyMacroPredicates": 0.25547003746032715,
"SimplifyNeuronTensor": 0.2766604423522949,
"SimplifySlice": 0.02300572395324707,
"SimplifyTensor": 0.2370157241821289,
"SpillPSum": 0.4576537609100342,
"SplitAPUnionSets": 0.6796882152557373,
"SplitAccGrp": 0.03665971755981445,
"StaticProfiler": 0.13864398002624512,
"StaticTransposeLocalTensor": 0.2623753547668457,
"SundaISel": 1.3397884368896484,
"TCTransform": 0.025269269943237305,
"TensorInitialization": 0.16660237312316895,
"TensorOpSimplifier": 0.15259718894958496,
"TensorOpTransform": 0.8282039165496826,
"TileCCOps": 0.3066561222076416,
"TilingProfiler": 0.33243393898010254,
"TransformConvOp": 0.05726790428161621,
"TritiumFusion": 0.13151931762695313,
"ValueNumbering": 0.06717801094055176,
"VectorizeDMA": 0.5187788009643555,
"VectorizeMatMult": 0.04601097106933594,
"WeightCoalescing": 0.049105167388916016,
"ZeroSizeTensorElimination": 0.0003685951232910156
},
"tensorizer": {
"DMATilingProfiler::TotalInstructionsAfterTiling": 41972,
"StaticProfiler::AifUb": 17.11549949645996,
"StaticProfiler::ArithmeticIntensityTensorizer": 27.70398712158203,
"StaticProfiler::AverageDmaLength": 2723.7119140625,
"StaticProfiler::AverageFractalPeUtilization": 96.97119140625,
"StaticProfiler::AveragePartitionUtilization": 88.53791809082031,
"StaticProfiler::AveragePeUtilization": 81.30671691894531,
"StaticProfiler::DDRTransferBytes": 918042224,
"StaticProfiler::InternalTransferBytes": 174444896,
"StaticProfiler::LoadExpanded": 238040,
"StaticProfiler::LocalizationEfficiency": 161.8649139404297,
"StaticProfiler::LocalizationEfficiencyIgnoreNonlocal": 167.2097930908203,
"StaticProfiler::StoreExpanded": 18609,
"StaticProfiler::TotalDMAExpanded": 256649,
"StaticProfiler::TotalDynamicInstancesCount": 57730,
"StaticProfiler::TotalDynamicInstancesWithMmPackedCount": 50535,
"StaticProfiler::TotalLNCComm": 0,
"StaticProfiler::TotalLNCCommTransfer": 0,
"TilingProfiler::AveragePartitionUtilizationAfterTiling": 0,
"TilingProfiler::AveragePeUtilizationAfterTiling": 0,
"TilingProfiler::BatchnormInstructionsAfterTiling": 0,
"TilingProfiler::DmaInstructionsAfterTiling": 0,
"TilingProfiler::GenericInstructionsAfterTiling": 179,
"TilingProfiler::MatMultInstructionsAfterTiling": 30304,
"TilingProfiler::NumPfTransposes": 348,
"TilingProfiler::NumPfTransposesForIo": 30,
"TilingProfiler::NumPfTransposesForLocal": 198,
"TilingProfiler::NumPfTransposesForNonlocal": 120,
"TilingProfiler::PfTransposeInstructions": 7835,
"TilingProfiler::PfTransposeInstructionsForIo": 5217,
"TilingProfiler::PfTransposeInstructionsForLocal": 508,
"TilingProfiler::PfTransposeInstructionsForNonlocal": 2110,
"TilingProfiler::ReduceInstructionsAfterTiling": 59,
"TilingProfiler::SimdInstructionsAfterTiling": 2266,
"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": 0.003599405288696289,
"DMALocalityOpt": 0.003013134002685547,
"DMAProfiler": 0.0034313201904296875,
"DataStreaming": 0.006362199783325195,
"DoNothing": 0.00013327598571777344,
"ExpandISAMacro": 0.003620147705078125,
"FactorizeBlkDims": 0.010854482650756836,
"InferPSumTensor": 0.010383129119873047,
"InferSharedMemLoc": 0.0029261112213134766,
"InsertCoreBarrier": 0.003323793411254883,
"LateLegalizeInst": 0.007294178009033203,
"LateNeuronInstComb": 0.008392095565795898,
"LegalizeSundaAccess": 0.013858556747436523,
"LegalizeType": 0.009220361709594727,
"LowerBroadcast": 0.0032749176025390625,
"LowerIntrinsics": 0.0036187171936035156,
"LowerTranspose": 0.0032770633697509766,
"NeuronInstComb": 0.008351325988769531,
"NeuronLICM": 0.009110689163208008,
"NeuronSimplifyPredicates": 0.0036361217498779297,
"NeuronValueNumbering": 0.0038361549377441406,
"SFKVectorizer": 0.031067371368408203,
"SimpleAllReduceTiling": 0.0040929317474365234,
"SimplifyNeuronTensor": 0.057866573333740234,
"SpillPSum": 0.02007007598876953,
"WeightCoalescing": 0.003391265869140625
}
}
}

3
token_generation_model/_tp0_bk0/graph.neff Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:29d9d8ef7e5a0b735090051a93d177f1b4866fe07fc6988421cb47de73a1953f
size 3185664

4225
token_generation_model/_tp0_bk0/log-neuron-cc.txt Normal file
View File

File diff suppressed because it is too large Load Diff

3
token_generation_model/_tp0_bk0/metaneff.pb Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:2751bdd6298e760b3ed04356f31f6ffbb7d7c37e3002f4e686a4fba8c24c2240
size 2465286

3
token_generation_model/_tp0_bk0/model.MODULE_5e3d0ce8963512f9ea2d+937cd7a2.hlo_module.pb Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:166776f75e13c12167868e06c05edd9ecd84b1f5005be8ff59ffb090e4b60e00
size 2443667

3
token_generation_model/_tp0_bk0/model.MODULE_5e3d0ce8963512f9ea2d+937cd7a2.neff Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:29d9d8ef7e5a0b735090051a93d177f1b4866fe07fc6988421cb47de73a1953f
size 3185664

224
token_generation_model/_tp0_bk0/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": [
128
],
"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": [
128
],
"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
}

3
token_generation_model/_tp0_bk0/wrapped_neff.hlo Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:3d7d771432f2a5b315c36747141a382cd677d56bd52622fbc82564ea1e8d3890
size 3380710