今天遇到分类问题
TensorRT softmax层
如果直接在fc后面接softmax 则会进行全局softmax
需要进行reshape 比如2分类 需要转为[1, 1, 2]这种维度 然后在维度2上进行softmax
百度后 先使用 IShuffleLayer 进行维度变换
IShuffleLayer *shuffleLayer = network->addShuffle(input);
assert(shuffleLayer);
shuffleLayer->setReshapeDimensions(Dims3(1, -1, c));
再进行softmax层
ISoftMaxLayer *softmax = network->addSoftMax(*shuffleLayer1->getOutput(0));
assert(softmax);
softmax->setAxes(1<<2);
特别注意这里的 softmax->setAxes(1<<2);
如果按照pytorch思路 直接 softmax->setAxes(2); 则结果就是全部1 结果不是我我们预期的
看下官方代码注释
//!
//! \brief Set the axis along which softmax is computed. Currently, only one axis can be set.
//!
//! The axis is specified by setting the bit corresponding to the axis to 1.
//! Let's say we have an NCHW tensor as input (three non-batch dimensions).
//!
//! In implicit mode :
//! Bit 0 corresponds to the C dimension boolean.
//! Bit 1 corresponds to the H dimension boolean.
//! Bit 2 corresponds to the W dimension boolean.
//! By default, softmax is performed on the axis which is the number of axes minus three. It is 0 if
//! there are fewer than 3 non-batch axes. For example, if the input is NCHW, the default axis is C. If the input
//! is NHW, then the default axis is H.
//!
//! In explicit mode :
//! Bit 0 corresponds to the N dimension boolean.
//! Bit 1 corresponds to the C dimension boolean.
//! Bit 2 corresponds to the H dimension boolean.
//! Bit 3 corresponds to the W dimension boolean.
//! By default, softmax is performed on the axis which is the number of axes minus three. It is 0 if
//! there are fewer than 3 axes. For example, if the input is NCHW, the default axis is C. If the input
//! is NHW, then the default axis is N.
//!
//! For example, to perform softmax on axis R of a NPQRCHW input, set bit 2 with implicit batch mode,
//! set bit 3 with explicit batch mode.
//!
//! \param axes The axis along which softmax is computed.
//! Here axes is a bitmap. For example, when doing softmax along axis 0, bit 0 is set to 1, axes = 1 << axis = 1.
//!
按照网上的解释 :
例如以NCHW而言,如果想要对H所在维度进行softmax, mask为0010 对于bitmap表示法:0100 转为bit移位操作 (1<<2)
对于C维度操作,mask为0100 bit表示法则为0010 转为bit移位操作(1<<1)
对于我的代码 我的输入softmax维度为(1, 1, 2, 0)
对于维度2进行操作 mask 0010 bit操作则 0100 移位操作(1<<2)
完整代码:
// softmax layer
ILayer* reshapeSoftmax(INetworkDefinition *network, ITensor &input, int c) {
IShuffleLayer *shuffleLayer1 = network->addShuffle(input);
assert(shuffleLayer1);
shuffleLayer1->setReshapeDimensions(Dims3(1, -1, c));
Dims dim0 = shuffleLayer1->getOutput(0)->getDimensions();
cout << "softmax output dims " << dim0.d[0] << " " << dim0.d[1] << " " << dim0.d[2] << " " << dim0.d[3] << endl;
ISoftMaxLayer *softmax = network->addSoftMax(*shuffleLayer1->getOutput(0));
assert(softmax);
softmax->setAxes(1<<2);
// 再变为一维数组
Dims dim_{};
dim_.nbDims = 1;
dim_.d[0] = -1;
IShuffleLayer *shuffleLayer2 = network->addShuffle(*softmax->getOutput(0));
assert(shuffleLayer2);
shuffleLayer2->setReshapeDimensions(dim_);
return shuffleLayer2;
}
可以参考下