Thanks for sending an issue! Please fill in the following template to help quickly solve your problem.

Describe the current behavior / 问题描述 (Mandatory / 必填)

1. [repos/ops-nn/activation/swiglu_group_grad/op_api/aclnn_swiglu_group_grad.cpp:198] aclnnSwigluGroupGradGetWorkspaceSize
   • 宏 APIName：aclnnSwigluGroupGrad
   • 接口算子参数：gradY, x, weightOptional, yOriginOptional, groupIndexOptional, clampLimit, gradXOut, gradWeightOutOptional
   • DFX_IN：gradY, x, weightOptional, yOriginOptional, groupIndexOptional
   • DFX_OUT：gradXOut, gradWeightOutOptional
   • ⚠ 问题：缺失: clampLimit
2. [repos/ops-nn/experimental/activation/fused_bias_leaky_relu/op_api/aclnn_fused_bias_leaky_relu.cpp:110] aclnnFusedBiasLeakyReluGetWorkspaceSize
   • 宏 APIName：aclnnFusedBiasLeakyRelu
   • 接口算子参数：x, bias, negativeSlope, scale, out
   • DFX_IN：x, bias
   • DFX_OUT：out
   • ⚠ 问题：缺失: negativeSlope, scale
3. [repos/ops-nn/experimental/activation/mish_grad_v2/op_host/op_api/aclnn_mish_backward_v2.cpp:112] aclnnMishBackwardGetWorkspaceSize
   • 宏 APIName：aclnnMishBackward
   • 接口算子参数：gradOutput, self, gradInput
   • DFX_IN：grad, x
   • DFX_OUT：xGrad
   • ⚠ 问题：缺失: gradOutput, self, gradInput；多余: grad, x, xGrad
4. [repos/ops-nn/experimental/activation/swiglu_clamp/op_host/op_api/aclnn_swiglu_clamp.cpp:91] aclnnSwigluClampGetWorkspaceSize
   • 宏 APIName：aclnnSwigluClamp
   • 接口算子参数：x, limit, y
   • DFX_IN：x
   • DFX_OUT：y
   • ⚠ 问题：缺失: limit
5. [repos/ops-nn/experimental/optim/apply_adadelta/op_api/aclnn_apply_adadelta.cpp:151] aclnnApplyAdadeltaGetWorkspaceSize
   • 宏 APIName：aclnnApplyAdadelta
   • 接口算子参数：var, accum, accumUpdate, lr, rho, epsilon, grad, varOut, accumOut, accumUpdateOut
   • DFX_IN：var, accum, accumUpdate, grad
   • DFX_OUT：varOut, accumOut, accumUpdateOut
   • ⚠ 问题：缺失: lr, rho, epsilon
6. [repos/ops-nn/experimental/optim/apply_adamax/op_api/aclnn_apply_adamax.cpp:149] aclnnApplyAdamaxGetWorkspaceSize
   • 宏 APIName：aclnnApplyAdamax
   • 接口算子参数：var, m, v, beta1Power, lr, beta1, beta2, epsilon, grad, varOut, mOut, vOut
   • DFX_IN：var, m, v, grad
   • DFX_OUT：varOut, mOut, vOut
   • ⚠ 问题：缺失: beta1Power, lr, beta1, beta2, epsilon
7. [repos/ops-nn/experimental/optim/apply_adam_d/op_api/aclnn_apply_adam_d.cpp:164] aclnnApplyAdamDGetWorkspaceSize
   • 宏 APIName：aclnnApplyAdamD
   • 接口算子参数：var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad, useLocking, useNesterov, varOut, mOut, vOut
   • DFX_IN：var, m, v, beta1Power, beta2Power, lr, beta1, beta2, epsilon, grad
   • DFX_OUT：varOut, mOut, vOut
   • ⚠ 问题：缺失: useLocking, useNesterov
8. [repos/ops-nn/experimental/pooling/max_pool_v3/op_host/op_api/aclnn_max_pool_v3.cpp:60] aclnnMaxPoolV3GetWorkspaceSize
   • 宏 APIName：aclnnMaxPoolV3
   • 接口算子参数：x, ksize, strides, pads, ceilMode, out
   • DFX_IN：x, ksize, strides, pads
   • DFX_OUT：out
   • ⚠ 问题：缺失: ceilMode
9. [repos/ops-nn/matmul/dual_level_quant_batch_matmul/op_host/op_api/aclnn_dual_level_quant_matmul.cpp:534] aclnnDualLevelQuantMatmulWeightNzGetWorkspaceSize
   • 宏 APIName：aclnnDualLevelQuantMatmulWeightNz
   • 接口算子参数：x1, x2, x1Level0Scale, x2Level0Scale, x1Level1Scale, x2Level1Scale, optionalBias, transposeX1, transposeX2, level0GroupSize, level1GroupSize, out
   • DFX_IN：x1, x2, x1Level0Scale, x2Level0Scale, x1Level1Scale, x2Level1Scale, optionalBias
   • DFX_OUT：out
   • ⚠ 问题：缺失: transposeX1, transposeX2, level0GroupSize, level1GroupSize
10. [repos/ops-nn/matmul/matmul_v2_compress_dequant/op_host/op_api/aclnn_matmul_compress_dequant.cpp:233] aclnnMatmulCompressDequantGetWorkspaceSize
    • 宏 APIName：aclnnMatmulCompressDequant
    • 接口算子参数：x1, x2, compressIndex, bias, deqScale, offsetW, offsetX, compressInfo, out
    • DFX_IN：x1, x2, compressIndex, bias, deqScale, offsetX, compressInfo
    • DFX_OUT：out
    • ⚠ 问题：缺失: offsetW
11. [repos/ops-nn/matmul/quant_batch_matmul_v3/op_api/aclnn_quant_matmul_v4.cpp:42] aclnnQuantMatmulV3GetWorkspaceSize
    • 宏 APIName：aclnnQuantMatmulV3
    • 接口算子参数：x1, x2, scale, offset, bias, transposeX1, transposeX2, out
    • DFX_IN：x1, x2, scale, offset, bias
    • DFX_OUT：out
    • ⚠ 问题：缺失: transposeX1, transposeX2
12. [repos/ops-nn/matmul/quant_batch_matmul_v3/op_api/aclnn_quant_matmul_v4.cpp:66] aclnnQuantMatmulV4GetWorkspaceSize
    • 宏 APIName：aclnnQuantMatmulV4
    • 接口算子参数：x1, x2, scale, offset, pertokenScaleOptional, bias, transposeX1, transposeX2, out
    • DFX_IN：x1, x2, scale, offset, pertokenScaleOptional, bias
    • DFX_OUT：out
    • ⚠ 问题：缺失: transposeX1, transposeX2
13. [repos/ops-nn/matmul/quant_matmul_reduce_sum/op_host/op_api/aclnn_quant_matmul_reduce_sum.cpp:334] aclnnQuantMatmulReduceSumWeightNzGetWorkspaceSize
    • 宏 APIName：aclnnQuantMatmulReduceSumWeightNz
    • 接口算子参数：x1, x2, x1Scale, x2Scale, yScale, x1Offset, x2Offset, yOffset, bias, transposeX1, transposeX2, groupSize, dims, keepDims, out
    • DFX_IN：x1, x2, x1Scale, x2Scale
    • DFX_OUT：out
    • ⚠ 问题：缺失: yScale, x1Offset, x2Offset, yOffset, bias, transposeX1, transposeX2, groupSize, dims, keepDims
14. [repos/ops-nn/matmul/rotate_quant/op_api/aclnn_rotate_quant.cpp:488] aclnnRotateQuantGetWorkspaceSize
    • 宏 APIName：aclnnRotateQuant
    • 接口算子参数：x, rotation, alpha, axis, roundMode, scaleAlg, dstTypeMax, trans, yOut, scaleOut
    • DFX_IN：x, rotation, alpha
    • DFX_OUT：yOut, scaleOut
    • ⚠ 问题：缺失: axis, roundMode, scaleAlg, dstTypeMax, trans
15. [repos/ops-nn/matmul/weight_quant_batch_matmul/op_host/op_api/aclnn_weight_quant_batch_matmul.cpp:357] aclnnWeightQuantBatchMatmulGetWorkspaceSize
    • 宏 APIName：aclnnWeightQuantBatchMatmul
    • 接口算子参数：x1, x2, diagonalMatrix, deqOffset, deqScale, addOffset, mulScale, bias, transposeX1, transposeX2, antiquantScale, antiquantOffset, out
    • DFX_IN：x1, x2, bias, addOffset, mulScale, diagonalMatrix, deqOffset, deqScale, transposeX1, transposeX2, antiquantScale, antiquantOffset
    • DFX_OUT：out
    • ⚠ 问题：IN 顺序不符
16. [repos/ops-nn/matmul/weight_quant_batch_matmul_v2/op_host/op_api/aclnn_weight_quant_batch_matmul_v2.cpp:1739] aclnnWeightQuantBatchMatmulV2GetWorkspaceSize
    • 宏 APIName：aclnnWeightQuantBatchMatmulV2
    • 接口算子参数：x, weight, antiquantScale, antiquantOffsetOptional, quantScaleOptional, quantOffsetOptional, biasOptional, antiquantGroupSize, y
    • DFX_IN：x, weight, antiquantScale, antiquantOffsetOptional, quantScaleOptional, quantOffsetOptional, biasOptional
    • DFX_OUT：y
    • ⚠ 问题：缺失: antiquantGroupSize
17. [repos/ops-nn/matmul/weight_quant_batch_matmul_v2/op_host/op_api/aclnn_weight_quant_batch_matmul_v2.cpp:1789] aclnnWeightQuantBatchMatmulV3GetWorkspaceSize
    • 宏 APIName：aclnnWeightQuantBatchMatmulV3
    • 接口算子参数：x, weight, antiquantScale, antiquantOffsetOptional, quantScaleOptional, quantOffsetOptional, biasOptional, antiquantGroupSize, innerPrecise, y
    • DFX_IN：x, weight, antiquantScale, antiquantOffsetOptional, quantScaleOptional, quantOffsetOptional, biasOptional
    • DFX_OUT：y
    • ⚠ 问题：缺失: antiquantGroupSize, innerPrecise
18. [repos/ops-nn/matmul/weight_quant_batch_matmul_v2/op_host/op_api/aclnn_weight_quant_batch_matmul_v2.cpp:1840] aclnnWeightQuantBatchMatmulNzGetWorkspaceSize
    • 宏 APIName：aclnnWeightQuantBatchMatmulNz
    • 接口算子参数：x, weight, antiquantScale, antiquantOffsetOptional, quantScaleOptional, quantOffsetOptional, biasOptional, antiquantGroupSize, y
    • DFX_IN：x, weight, antiquantScale, antiquantOffsetOptional, quantScaleOptional, quantOffsetOptional, biasOptional
    • DFX_OUT：y
    • ⚠ 问题：缺失: antiquantGroupSize
19. [repos/ops-nn/quant/ascend_quant_v2/op_api/aclnn_ascend_quant.cpp:320] aclnnAscendQuantGetWorkspaceSize
    • 宏 APIName：aclnnAscendQuant
    • 接口算子参数：x, scale, offset, sqrtMode, roundMode, dstType, y
    • DFX_IN：x, scale, offset, roundMode, sqrtMode, dstType
    • DFX_OUT：y
    • ⚠ 问题：IN 顺序不符
20. [repos/ops-nn/quant/ascend_quant_v2/op_api/aclnn_ascend_quant_v3.cpp:372] aclnnAscendQuantV3GetWorkspaceSize
    • 宏 APIName：aclnnAscendQuantV3
    • 接口算子参数：x, scale, offset, sqrtMode, roundMode, dstType, axis, y
    • DFX_IN：x, scale, offset, roundMode, sqrtMode, dstType, axis
    • DFX_OUT：y
    • ⚠ 问题：IN 顺序不符
21. [repos/ops-nn/quant/quant_max/op_api/aclnn_quant_max.cpp:220] aclnnQuantMaxGetWorkspaceSize
    • 宏 APIName：aclnnQuantMax
    • 接口算子参数：x, scale, roundMode, dstType, y, amax
    • DFX_IN：x, scale
    • DFX_OUT：y, amax
    • ⚠ 问题：缺失: roundMode, dstType
22. [repos/ops-nn/rnn/single_layer_lstm_grad/op_api/aclnn_lstm_backward.cpp:1624] aclnnLstmBackwardGetWorkspaceSize
    • 宏 APIName：aclnnLstmBackward
    • 接口算子参数：input, hx, params, dy, dh, dc, i, g, f, o, h, c, tanhc, batchSizesOptional, hasBias, numLayers, dropout, train, bidirectional, batchFirst, outputMask, dxOut, dhPrevOut, dcPrevOut, dparamsOut
    • DFX_IN：input, hx, params, dy, dh, dc, i, g, f, o, h, c, tanhc, batchSizesOptional, hasBias, numLayers, dropout, train, bidirectional, batchFirst
    • DFX_OUT：dxOut, dhPrevOut, dcPrevOut, dparamsOut
    • ⚠ 问题：缺失: outputMask

Environment / 环境信息 (Mandatory / 必填)

不涉及

Steps to reproduce the issue / 重现步骤 (Mandatory / 必填)

不涉及

Describe the expected behavior / 预期结果 (Mandatory / 必填)

L2_DFX_PHASE_1宏使用说明：
https://www.hiascend.com/document/detail/zh/CANNCommunityEdition/latest/API/aolapi/atlasascendc_api_07_1052.html

Related log / screenshot / 日志 / 截图 (Mandatory / 必填)

无

Special notes for this issue/备注 (Optional / 选填)