FilesLast commitLast update
12 days ago
1 month ago
3 months ago
22 days ago
21 days ago
18 days ago
18 days ago
18 days ago
22 days ago
1 month ago
5 days ago
README

DropOutV3

English | 简体中文

Product Support Status

Product Supported
Ascend 950PR&950DT products √
Atlas A3 products ×
Atlas A2 products ×
Atlas 200I/500 A2 inference products ×
Atlas inference products ×
Atlas training products ×

Function Description

  • Operator function: during training, randomly sets the elements in the input to zero with the probability p, and scales the output by a factor of 1/(1-p).
  • Computation formula:

    outi={0, with probability p11−pinputi, with probability 1−pout_i=\begin{cases}0,&\text { with probability }p \\\frac{1}{1-p}input_i,&\text { with probability }1-p\end{cases}

Parameter Description

Parameter Name Input/Output/Attribute Description Data Type Data Format
input Input Input elements. FLOAT, FLOAT16, BFLOAT16 ND
noise_shape Input Reserved parameter. Use a null pointer for this input parameter. INT64 ND
p Input Probability of setting elements to zero. The value range is [0, 1]. FLOAT, FLOAT16, BFLOAT16 ND
seed Input Seed of the random number, which affects the generated random number sequence. INT64, INT32 ND
offset Input Offset of the random number, which affects the position in the generated random number sequence. INT64 ND
y Output Output data. FLOAT, FLOAT16, BFLOAT16 ND
mask Output Mask data of the bit type stored in the UINT8 type. UINT8 ND

Constraint Description

None

Invocation Description

Invocation Method Sample Code Description
aclnn API test_aclnn_drop_out_v3 Invokes the DropOutV3 operator through the aclnn_drop_out_v3 API.
aclnn API test_aclnn_drop_out_v3_tensor Invokes the DropOutV3 operator through the aclnn_drop_out_v3_tensor API.
Graph-mode invocation test_geir_drop_out_v3 Invokes the DropOutV3 operator by building a graph through the operator IR.