Searched refs:beta (Results 1 – 4 of 4) sorted by relevance
| /hardware/qcom/neuralnetworks/hvxservice/1.0/ |
| D | HexagonOperationsPrepare.cpp | 299 const hexagon_nn_input& beta = model->getTensor(ins[4]); in local_response_normalization() local
306 return model->addBasicOperation(OP_LRN_f, NN_PAD_NA, {input, window, bias, alpha, beta}, outs); in local_response_normalization()
462 const hexagon_nn_input& beta = model->getTensor(ins[1]); in softmax() local
465 return model->addBasicOperation(OP_Softmax_f, NN_PAD_NA, {input, beta}, outs); in softmax()
900 const hexagon_nn_input& beta = model->getTensor(ins[1]); in softmax() local
907 {input, input_min, input_max, beta}, outs); in softmax()
|
| /hardware/interfaces/neuralnetworks/1.0/ |
| D | types.hal | 732 * output = input / pow((bias + alpha * sqr_sum), beta)
750 * * 4: A scalar, specifying the exponent, beta.
751 * For input tensor of {@link OperandType::TENSOR_FLOAT32}, the beta
1296 * exp((input[batch, i] - max(input[batch, :])) * beta) /
1297 * sum_{k}{exp((input[batch, k] - max(input[batch, :])) * beta)}
1311 * beta. If input0 is of {@link OperandType::TENSOR_FLOAT32} or
|
| /hardware/interfaces/neuralnetworks/1.2/ |
| D | types.hal | 973 * output = input / pow((bias + alpha * sqr_sum), beta)
1000 * * 4: A scalar, specifying the exponent, beta.
1001 * For input tensor of {@link OperandType::TENSOR_FLOAT16}, the beta
1003 * For input tensor of {@link OperandType::TENSOR_FLOAT32}, the beta
1664 * exp((input[batch, i] - max(input[batch, :])) * beta) /
1665 * sum_{k}{exp((input[batch, k] - max(input[batch, :])) * beta)}
1682 * beta. If input0 is of {@link OperandType::TENSOR_FLOAT32} or
3311 * sqrt(var[b, c] + epsilon) + beta
3337 * * 2: A scalar, specifying beta, the offset applied to the normalized
3481 * output = logits * beta - log(reduce_sum(exp(logits * beta), axis))
[all …]
|
| /hardware/interfaces/neuralnetworks/1.3/ |
| D | types.hal | 967 * output = input / pow((bias + alpha * sqr_sum), beta)
994 * * 4: A scalar, specifying the exponent, beta.
995 * For input tensor of {@link OperandType::TENSOR_FLOAT16}, the beta
997 * For input tensor of {@link OperandType::TENSOR_FLOAT32}, the beta
1700 * exp((input[batch, i] - max(input[batch, :])) * beta) /
1701 * sum_{k}{exp((input[batch, k] - max(input[batch, :])) * beta)}
1719 * beta. If input0 is of {@link OperandType::TENSOR_FLOAT32},
3516 * sqrt(var[b, c] + epsilon) + beta
3542 * * 2: A scalar, specifying beta, the offset applied to the normalized
3688 * output = logits * beta - log(reduce_sum(exp(logits * beta), axis))
[all …]
|