level 1 · 154/264 correct
trajectories
264
finished
264
correct
154
compiled
160
avg speedup (correct)
291.19x
sweep-wide eval metricsaveraged over correct kernels
outcomescorrect 154 · error 103 · incorrect 6 · compile_fail 1
avg speedup291.19x (over 154 correct)
avg memory ratio (kernel / ref)0.88x (154 samples)
avg fusion ratio (ref launches / kernel launches)774.02x (154 samples)
avg energy ratio (ref / kernel)1.56x (98 samples)
avg SOL score (0–1)0.536 (153 samples)
avg DRAM util %46.6%
avg occupancy %70.0%
avg max abs error5.72e-03
original · 115/200 correct
P1 · Kimi-K2.61_Square_matrix_multiplication_.
correct7.39x0.4μs
P2 · Kimi-K2.62_Standard_matrix_multiplication
error—
P3 · Kimi-K2.63_Batched_matrix_multiplication.
error—
P4 · Kimi-K2.64_Matrix_vector_multiplication_.
error—
P5 · Kimi-K2.65_Matrix_scalar_multiplication.p
error—
P6 · Kimi-K2.66_Matmul_with_large_K_dimension_
error—
P7 · Kimi-K2.67_Matmul_with_small_K_dimension_
error—
P8 · Kimi-K2.68_Matmul_with_irregular_shapes_.
error—
P9 · Kimi-K2.69_Tall_skinny_matrix_multiplicat
correct0.43x6.1μs
P10 · Kimi-K2.610_3D_tensor_matrix_multiplicati
correct1.00x1.0μs
P11 · Kimi-K2.611_4D_tensor_matrix_multiplicati
error—
P12 · Kimi-K2.612_Matmul_with_diagonal_matrices
correct1.27x0.1μs
P13 · Kimi-K2.613_Matmul_for_symmetric_matrices
error—
P14 · Kimi-K2.614_Matmul_for_upper_triangular_m
correct1.02x2.6μs
P15 · Kimi-K2.615_Matmul_for_lower_triangular_m
correct1.02x2.6μs
P16 · Kimi-K2.616_Matmul_with_transposed_A.py
correct5.05x0.5μs
P17 · Kimi-K2.617_Matmul_with_transposed_B.py
correct1.00x2.7μs
P18 · Kimi-K2.618_Matmul_with_transposed_both.p
error—
P19 · Kimi-K2.619_ReLU.py
correct0.93x4.6μs
P20 · Kimi-K2.620_LeakyReLU.py
correct0.92x4.7μs
P21 · Kimi-K2.621_Sigmoid.py
correct0.92x4.7μs
P22 · Kimi-K2.622_Tanh.py
correct0.93x4.6μs
P23 · Kimi-K2.623_Softmax.py
error—
P24 · Kimi-K2.624_LogSoftmax.py
error—
P25 · Kimi-K2.625_Swish.py
error—
P26 · Kimi-K2.626_GELU_.py
correct0.59x7.3μs
P27 · Kimi-K2.627_SELU_.py
correct0.99x4.3μs
P28 · Kimi-K2.628_HardSigmoid.py
correct0.98x4.4μs
P29 · Kimi-K2.629_Softplus.py
error—
P30 · Kimi-K2.630_Softsign.py
correct3.22x4.6μs
P31 · Kimi-K2.631_ELU.py
correct0.93x4.6μs
P32 · Kimi-K2.632_HardTanh.py
correct0.93x4.6μs
P33 · Kimi-K2.633_BatchNorm.py
correct0.30x29.2μs
P34 · Kimi-K2.634_InstanceNorm.py
error—
P35 · Kimi-K2.635_GroupNorm_.py
error—
P36 · Kimi-K2.636_RMSNorm_.py
correct1.29x11.0μs
P37 · Kimi-K2.637_FrobeniusNorm_.py
error—
P38 · Kimi-K2.638_L1Norm_.py
incorrect—
P39 · Kimi-K2.639_L2Norm_.py
incorrect—
P40 · Kimi-K2.640_LayerNorm.py
correct19.64x0.4μs
P41 · Kimi-K2.641_Max_Pooling_1D.py
correct4.01x3.9μs
P42 · Kimi-K2.642_Max_Pooling_2D.py
correct1.55x7.0μs
P43 · Kimi-K2.643_Max_Pooling_3D.py
correct1.49x2.6μs
P44 · Kimi-K2.644_Average_Pooling_1D.py
correct3.64x1.5μs
P45 · Kimi-K2.645_Average_Pooling_2D.py
error—
P46 · Kimi-K2.646_Average_Pooling_3D.py
correct1.51x5.7μs
P47 · Kimi-K2.647_Sum_reduction_over_a_dimensio
correct0.86x3.5μs
P48 · Kimi-K2.648_Mean_reduction_over_a_dimensi
correct0.90x3.4μs
P49 · Kimi-K2.649_Max_reduction_over_a_dimensio
error—
P50 · Kimi-K2.650_conv_standard_2D__square_inpu
error—
P51 · Kimi-K2.651_Argmax_over_a_dimension.py
correct0.86x3.8μs
P52 · Kimi-K2.652_Argmin_over_a_dimension.py
correct0.91x3.6μs
P53 · Kimi-K2.653_Min_reduction_over_a_dimensio
correct1.62x3.6μs
P54 · Kimi-K2.654_conv_standard_3D__square_inpu
correct0.24x7.8μs
P55 · Kimi-K2.655_conv_standard_2D__asymmetric_
error—
P56 · Kimi-K2.656_conv_standard_2D__asymmetric_
error—
P57 · Kimi-K2.657_conv_transposed_2D__square_in
error—
P58 · Kimi-K2.658_conv_transposed_3D__asymmetri
error—
P59 · Kimi-K2.659_conv_standard_3D__asymmetric_
error—
P60 · Kimi-K2.660_conv_standard_3D__square_inpu
correct0.18x55.0μs
P61 · Kimi-K2.661_conv_transposed_3D__square_in
error—
P62 · Kimi-K2.662_conv_standard_2D__square_inpu
error—
P63 · Kimi-K2.663_conv_standard_2D__square_inpu
incorrect—
P64 · Kimi-K2.664_conv_transposed_1D.py
error—
P65 · Kimi-K2.665_conv_transposed_2D__square_in
error—
P66 · Kimi-K2.666_conv_standard_3D__asymmetric_
error—
P67 · Kimi-K2.667_conv_standard_1D.py
error—
P68 · Kimi-K2.668_conv_transposed_3D__square_in
error—
P69 · Kimi-K2.669_conv_transposed_2D__asymmetri
correct0.97x3.0μs
P70 · Kimi-K2.670_conv_transposed_3D__asymmetri
error—
P71 · Kimi-K2.671_conv_transposed_2D__asymmetri
error—
P72 · Kimi-K2.672_conv_transposed_3D_asymmetric
error—
P73 · Kimi-K2.673_conv_transposed_3D_asymmetric
correct0.99x2.1μs
P74 · Kimi-K2.674_conv_transposed_1D_dilated.py
error—
P75 · Kimi-K2.675_conv_transposed_2D_asymmetric
error—
P76 · Kimi-K2.676_conv_standard_1D_dilated_stri
error—
P77 · Kimi-K2.677_conv_transposed_3D_square_inp
error—
P78 · Kimi-K2.678_conv_transposed_2D_asymmetric
error—
P79 · Kimi-K2.679_conv_transposed_1D_asymmetric
error—
P80 · Kimi-K2.680_conv_standard_2D_square_input
error—
P81 · Kimi-K2.681_conv_transposed_2D_asymmetric
error—
P82 · Kimi-K2.682_conv_depthwise_2D_square_inpu
correct2.58x1.0μs
P83 · Kimi-K2.683_conv_depthwise_2D_square_inpu
correct3.89x0.4μs
P84 · Kimi-K2.684_conv_depthwise_2D_asymmetric_
error—
P85 · Kimi-K2.685_conv_depthwise_2D_asymmetric_
correct1.93x1.2μs
P86 · Kimi-K2.686_conv_depthwise_separable_2D.p
error—
P87 · Kimi-K2.687_conv_pointwise_2D.py
error—
P88 · Kimi-K2.688_MinGPTNewGelu.py
error—
P89 · Kimi-K2.689_cumsum.py
correct0.06x78.8μs
P90 · Kimi-K2.690_cumprod.py
correct0.20x23.0μs
P91 · Kimi-K2.691_cumsum_reverse.py
correct1.07x10.7μs
P92 · Kimi-K2.692_cumsum_exclusive.py
error—
P93 · Kimi-K2.693_masked_cumsum.py
error—
P94 · Kimi-K2.694_MSELoss.py
correct2.96x2.9μs
P95 · Kimi-K2.695_CrossEntropyLoss.py
correct2.03x0.2μs
P96 · Kimi-K2.696_HuberLoss.py
error—
P97 · Kimi-K2.697_ScaledDotProductAttention.py
error—
P98 · Kimi-K2.698_KLDivLoss.py
correct5.20x0.8μs
P99 · Kimi-K2.699_TripletMarginLoss.py
error—
P100 · Kimi-K2.6100_HingeLoss.py
correct7.16x1.5μs
P1 · gpt-5.5-priority1_Square_matrix_multiplication_.
correct1.00x2.6μs
P2 · gpt-5.5-priority2_Standard_matrix_multiplication
correct8.19x0.3μs
P3 · gpt-5.5-priority3_Batched_matrix_multiplication.
correct5.76x0.9μs
P4 · gpt-5.5-priority4_Matrix_vector_multiplication_.
correct0.97x3.0μs
P5 · gpt-5.5-priority5_Matrix_scalar_multiplication.p
correct0.66x4.3μs
P6 · gpt-5.5-priority6_Matmul_with_large_K_dimension_
error—
P7 · gpt-5.5-priority7_Matmul_with_small_K_dimension_
correct0.84x5.0μs
P8 · gpt-5.5-priority8_Matmul_with_irregular_shapes_.
error—
P9 · gpt-5.5-priority9_Tall_skinny_matrix_multiplicat
correct0.42x11.5μs
P10 · gpt-5.5-priority10_3D_tensor_matrix_multiplicati
correct1.00x1.0μs
P11 · gpt-5.5-priority11_4D_tensor_matrix_multiplicati
correct1.00x11.0μs
P12 · gpt-5.5-priority12_Matmul_with_diagonal_matrices
correct1.23x0.1μs
P13 · gpt-5.5-priority13_Matmul_for_symmetric_matrices
correct1.00x2.7μs
P14 · gpt-5.5-priority14_Matmul_for_upper_triangular_m
correct3.23x0.8μs
P15 · gpt-5.5-priority15_Matmul_for_lower_triangular_m
correct1.82x1.5μs
P16 · gpt-5.5-priority16_Matmul_with_transposed_A.py
correct0.18x14.4μs
P17 · gpt-5.5-priority17_Matmul_with_transposed_B.py
error—
P18 · gpt-5.5-priority18_Matmul_with_transposed_both.p
correct1.00x2.8μs
P19 · gpt-5.5-priority19_ReLU.py
error—
P20 · gpt-5.5-priority20_LeakyReLU.py
correct833.33x0.0μs
P21 · gpt-5.5-priority21_Sigmoid.py
correct0.92x4.7μs
P22 · gpt-5.5-priority22_Tanh.py
correct0.95x4.5μs
P23 · gpt-5.5-priority23_Softmax.py
error—
P24 · gpt-5.5-priority24_LogSoftmax.py
correct0.99x8.8μs
P25 · gpt-5.5-priority25_Swish.py
correct2.26x4.7μs
P26 · gpt-5.5-priority26_GELU_.py
correct0.92x4.7μs
P27 · gpt-5.5-priority27_SELU_.py
error—
P28 · gpt-5.5-priority28_HardSigmoid.py
correct0.93x4.6μs
P29 · gpt-5.5-priority29_Softplus.py
correct0.92x4.7μs
P30 · gpt-5.5-priority30_Softsign.py
correct3.28x4.5μs
P31 · gpt-5.5-priority31_ELU.py
correct1303.03x0.0μs
P32 · gpt-5.5-priority32_HardTanh.py
error—
P33 · gpt-5.5-priority33_BatchNorm.py
correct1.95x4.6μs
P34 · gpt-5.5-priority34_InstanceNorm.py
correct0.89x10.9μs
P35 · gpt-5.5-priority35_GroupNorm_.py
error—
P36 · gpt-5.5-priority36_RMSNorm_.py
correct1.69x8.4μs
P37 · gpt-5.5-priority37_FrobeniusNorm_.py
error—
P38 · gpt-5.5-priority38_L1Norm_.py
correct1.39x11.3μs
P39 · gpt-5.5-priority39_L2Norm_.py
correct0.78x12.8μs
P40 · gpt-5.5-priority40_LayerNorm.py
correct28.87x0.3μs
P41 · gpt-5.5-priority41_Max_Pooling_1D.py
correct4.63x2.3μs
P42 · gpt-5.5-priority42_Max_Pooling_2D.py
correct2.53x4.3μs
P43 · gpt-5.5-priority43_Max_Pooling_3D.py
correct1.24x3.2μs
P44 · gpt-5.5-priority44_Average_Pooling_1D.py
correct3.10x1.8μs
P45 · gpt-5.5-priority45_Average_Pooling_2D.py
correct1.15x5.7μs
P46 · gpt-5.5-priority46_Average_Pooling_3D.py
correct1.00x8.6μs
P47 · gpt-5.5-priority47_Sum_reduction_over_a_dimensio
correct0.88x3.4μs
P48 · gpt-5.5-priority48_Mean_reduction_over_a_dimensi
error—
P49 · gpt-5.5-priority49_Max_reduction_over_a_dimensio
correct0.90x4.1μs
P50 · gpt-5.5-priority50_conv_standard_2D__square_inpu
error—
P51 · gpt-5.5-priority51_Argmax_over_a_dimension.py
correct0.92x3.5μs
P52 · gpt-5.5-priority52_Argmin_over_a_dimension.py
error—
P53 · gpt-5.5-priority53_Min_reduction_over_a_dimensio
error—
P54 · gpt-5.5-priority54_conv_standard_3D__square_inpu
correct0.28x6.7μs
P55 · gpt-5.5-priority55_conv_standard_2D__asymmetric_
error—
P56 · gpt-5.5-priority56_conv_standard_2D__asymmetric_
error—
P57 · gpt-5.5-priority57_conv_transposed_2D__square_in
correct0.03x191.0μs
P58 · gpt-5.5-priority58_conv_transposed_3D__asymmetri
error—
P59 · gpt-5.5-priority59_conv_standard_3D__asymmetric_
correct0.42x6.6μs
P60 · gpt-5.5-priority60_conv_standard_3D__square_inpu
correct0.49x20.3μs
P61 · gpt-5.5-priority61_conv_transposed_3D__square_in
correct1.00x9.1μs
P62 · gpt-5.5-priority62_conv_standard_2D__square_inpu
error—
P63 · gpt-5.5-priority63_conv_standard_2D__square_inpu
error—
P64 · gpt-5.5-priority64_conv_transposed_1D.py
incorrect—
P65 · gpt-5.5-priority65_conv_transposed_2D__square_in
correct0.05x51.5μs
P66 · gpt-5.5-priority66_conv_standard_3D__asymmetric_
correct1.04x4.7μs
P67 · gpt-5.5-priority67_conv_standard_1D.py
error—
P68 · gpt-5.5-priority68_conv_transposed_3D__square_in
correct1.00x36.9μs
P69 · gpt-5.5-priority69_conv_transposed_2D__asymmetri
correct0.03x101.0μs
P70 · gpt-5.5-priority70_conv_transposed_3D__asymmetri
correct0.17x169.0μs
P71 · gpt-5.5-priority71_conv_transposed_2D__asymmetri
error—
P72 · gpt-5.5-priority72_conv_transposed_3D_asymmetric
error—
P73 · gpt-5.5-priority73_conv_transposed_3D_asymmetric
error—
P74 · gpt-5.5-priority74_conv_transposed_1D_dilated.py
correct0.06x37.5μs
P75 · gpt-5.5-priority75_conv_transposed_2D_asymmetric
error—
P76 · gpt-5.5-priority76_conv_standard_1D_dilated_stri
correct0.11x115.0μs
P77 · gpt-5.5-priority77_conv_transposed_3D_square_inp
error—
P78 · gpt-5.5-priority78_conv_transposed_2D_asymmetric
correct0.78x14.7μs
P79 · gpt-5.5-priority79_conv_transposed_1D_asymmetric
compile_fail—
P80 · gpt-5.5-priority80_conv_standard_2D_square_input
correct0.10x170.0μs
P81 · gpt-5.5-priority81_conv_transposed_2D_asymmetric
incorrect—
P82 · gpt-5.5-priority82_conv_depthwise_2D_square_inpu
correct2.73x0.9μs
P83 · gpt-5.5-priority83_conv_depthwise_2D_square_inpu
correct3.75x0.4μs
P84 · gpt-5.5-priority84_conv_depthwise_2D_asymmetric_
correct2.60x3.9μs
P85 · gpt-5.5-priority85_conv_depthwise_2D_asymmetric_
error—
P86 · gpt-5.5-priority86_conv_depthwise_separable_2D.p
correct1.30x2.8μs
P87 · gpt-5.5-priority87_conv_pointwise_2D.py
correct0.58x8.1μs
P88 · gpt-5.5-priority88_MinGPTNewGelu.py
correct8.80x0.2μs
P89 · gpt-5.5-priority89_cumsum.py
error—
P90 · gpt-5.5-priority90_cumprod.py
correct3.30x1.4μs
P91 · gpt-5.5-priority91_cumsum_reverse.py
correct3.35x3.4μs
P92 · gpt-5.5-priority92_cumsum_exclusive.py
correct3.73x3.5μs
P93 · gpt-5.5-priority93_masked_cumsum.py
correct1.87x4.8μs
P94 · gpt-5.5-priority94_MSELoss.py
correct2.94x2.9μs
P95 · gpt-5.5-priority95_CrossEntropyLoss.py
incorrect—
P96 · gpt-5.5-priority96_HuberLoss.py
correct1.94x2.9μs
P97 · gpt-5.5-priority97_ScaledDotProductAttention.py
correct0.14x185.0μs
P98 · gpt-5.5-priority98_KLDivLoss.py
correct5.41x0.7μs
P99 · gpt-5.5-priority99_TripletMarginLoss.py
correct4.02x1.1μs
P100 · gpt-5.5-priority100_HingeLoss.py
correct6.93x1.5μs
popcorn · 39/64 correct
P1 · Kimi-K2.61_Conv1d_Depthwise.py
correct1.27x1.0μs
P2 · Kimi-K2.62_GraphEdgeSoftmaxCSR.py
error—
P3 · Kimi-K2.63_CSRSpMMMessagePassing.py
error—
P4 · Kimi-K2.64_GJKRobotEnvironmentIntersectio
correct679.75x0.0μs
P5 · Kimi-K2.65_Conv1d_Causal.py
error—
P6 · Kimi-K2.66_EdgeSoftmaxMultiHeadCSR.py
correct1053.33x0.0μs
P7 · Kimi-K2.67_Conv1d_FFT.py
error—
P8 · Kimi-K2.68_SegmentTopKCSR.py
error—
P9 · Kimi-K2.69_AssociativeScan.py
error—
P10 · Kimi-K2.610_SampledDenseDenseMatmulEdges.
error—
P11 · Kimi-K2.611_DegreeNormalizedAggregation.p
correct2978.45x0.0μs
P12 · Kimi-K2.612_DeepSeekMoEDispatchPermute.py
correct1121.02x0.1μs
P13 · Kimi-K2.613_FarthestPointSampling3D.py
error—
P14 · Kimi-K2.614_GaussianProcessRBFKernel.py
error—
P15 · Kimi-K2.615_COOScatterAddNodeFeatures.py
error—
P16 · Kimi-K2.616_DeepSeekMoECombineScatter.py
error—
P17 · Kimi-K2.617_CSRMaxAggregation.py
correct1802.03x0.0μs
P18 · Kimi-K2.618_CSRMultiHeadSpMM.py
error—
P19 · Kimi-K2.619_PairwiseDistanceMatrix.py
correct7.51x0.0μs
P20 · Kimi-K2.620_RotaryPositionEmbeddingBio.py
correct6.90x0.0μs
P21 · Kimi-K2.621_RadialBasisFunctionExpansion.
error—
P22 · Kimi-K2.622_SE3InvariantLinear.py
correct4.88x0.0μs
P23 · Kimi-K2.623_virtual_all_reduce_mean.py
correct1.18x0.0μs
P24 · Kimi-K2.624_VoxelGridPooling.py
correct276.88x0.0μs
P25 · Kimi-K2.625_virtual_reduce_scatter_masked
correct2.12x0.0μs
P26 · Kimi-K2.626_ImportanceSampling.py
error—
P27 · Kimi-K2.627_permutation_all_to_all.py
correct4.90x0.0μs
P28 · Kimi-K2.628_DirichletMultinomialLogLikeli
error—
P29 · Kimi-K2.629_logsumexp_across_virtual_rank
correct5.23x0.0μs
P30 · Kimi-K2.630_broadcast_masked_from_row0.py
correct1.18x0.0μs
P31 · Kimi-K2.631_bitmask_or_popcount.py
correct82.19x0.0μs
P32 · Kimi-K2.632_argmax_tiebreak_smallest_rank
error—
P1 · gpt-5.5-priority1_Conv1d_Depthwise.py
correct3.29x0.4μs
P2 · gpt-5.5-priority2_GraphEdgeSoftmaxCSR.py
correct2657.25x0.0μs
P3 · gpt-5.5-priority3_CSRSpMMMessagePassing.py
correct4284.55x0.0μs
P4 · gpt-5.5-priority4_GJKRobotEnvironmentIntersectio
correct155.34x0.0μs
P5 · gpt-5.5-priority5_Conv1d_Causal.py
error—
P6 · gpt-5.5-priority6_EdgeSoftmaxMultiHeadCSR.py
correct2169.68x0.0μs
P7 · gpt-5.5-priority7_Conv1d_FFT.py
correct1.02x1.2μs
P8 · gpt-5.5-priority8_SegmentTopKCSR.py
correct2853.54x0.0μs
P9 · gpt-5.5-priority9_AssociativeScan.py
correct284.79x0.3μs
P10 · gpt-5.5-priority10_SampledDenseDenseMatmulEdges.
correct6.32x0.0μs
P11 · gpt-5.5-priority11_DegreeNormalizedAggregation.p
correct9629.63x0.0μs
P12 · gpt-5.5-priority12_DeepSeekMoEDispatchPermute.py
error—
P13 · gpt-5.5-priority13_FarthestPointSampling3D.py
correct98.04x0.1μs
P14 · gpt-5.5-priority14_GaussianProcessRBFKernel.py
correct10.00x0.0μs
P15 · gpt-5.5-priority15_COOScatterAddNodeFeatures.py
correct0.66x0.0μs
P16 · gpt-5.5-priority16_DeepSeekMoECombineScatter.py
correct11503.50x0.0μs
P17 · gpt-5.5-priority17_CSRMaxAggregation.py
error—
P18 · gpt-5.5-priority18_CSRMultiHeadSpMM.py
error—
P19 · gpt-5.5-priority19_PairwiseDistanceMatrix.py
error—
P20 · gpt-5.5-priority20_RotaryPositionEmbeddingBio.py
correct7.42x0.0μs
P21 · gpt-5.5-priority21_RadialBasisFunctionExpansion.
error—
P22 · gpt-5.5-priority22_SE3InvariantLinear.py
correct4.48x0.0μs
P23 · gpt-5.5-priority23_virtual_all_reduce_mean.py
correct1.15x0.0μs
P24 · gpt-5.5-priority24_VoxelGridPooling.py
correct550.93x0.0μs
P25 · gpt-5.5-priority25_virtual_reduce_scatter_masked
error—
P26 · gpt-5.5-priority26_ImportanceSampling.py
correct11.26x0.0μs
P27 · gpt-5.5-priority27_permutation_all_to_all.py
correct4.65x0.0μs
P28 · gpt-5.5-priority28_DirichletMultinomialLogLikeli
correct8.57x0.0μs
P29 · gpt-5.5-priority29_logsumexp_across_virtual_rank
error—
P30 · gpt-5.5-priority30_broadcast_masked_from_row0.py
error—
P31 · gpt-5.5-priority31_bitmask_or_popcount.py
correct145.37x0.0μs
P32 · gpt-5.5-priority32_argmax_tiebreak_smallest_rank
correct38.28x0.0μs