def main():
inshape = (1, 3, 368, 368)
net = loadNet(inplace=False, modelpath="../TestData/pose_iter_440000.hdf")
outshape = net.dataShapeFrom(inshape)
pzlEngine = buildRTEngine(net,
inshape=inshape,
dtype=DataType.float32,
savepath="../TestData")
caffeEngine = buildRTEngineFromCaffe(
("../TestData/pose_deploy_linevec.prototxt",
"../TestData/pose_iter_440000.caffemodel"),
inshape=inshape,
outshape=outshape,
outlayers=["net_output"],
dtype=DataType.float32,
savepath="../TestData")
data = gpuarray.to_gpu(np.random.randn(*inshape).astype(np.float32))
pzlData = pzlEngine(data)
caffeData = caffeEngine(data)
assert np.allclose(pzlData.get(), caffeData.get(), atol=1e-7)
benchModels(pzlEngine, caffeEngine, data, lognames=("puzzle", "caffe "))
def main(): batchsize, insize = 16, 1000 inNode = Linear(insize, 1000, name="linear1").node() node = Activation(relu, name="relu1").node(inNode) node1 = Linear(1000, 800, name="linear2").node(node) node1 = Activation(relu, name="relu2").node(node1) node2 = Linear(1000, 800, name="linear3").node(node) node2 = Activation(relu, name="relu3").node(node2) outNode = Add(name="add").node(node1, node2) graph = Graph(inputs=inNode, outputs=outNode, name="graph") data = gpuarray.to_gpu(np.random.randn(batchsize, insize).astype(np.float32)) engine = buildRTEngine(graph, (batchsize, insize), savepath="../TestData", dtype=DataType.float32) outdata = graph(data) enginedata = engine(data) assert np.allclose(outdata.get(), enginedata.get(), atol=1e-6) benchModels(graph, engine, data)
def main(): net = loadResNet(modelpath="../../TestData/ResNet-50-model.hdf", layers="50") data = gpuarray.to_gpu(loadResNetSample(net, "../../TestData/tarantula.jpg")) labels = loadLabels(synpath="../../TestData/synsets.txt", wordpath="../../TestData/synset_words.txt") engine = buildRTEngine(net, inshape=data.shape, savepath="../TestData", dtype=DataType.float32) scoreModels(net, engine, data, labels) benchModels(net, engine, data)
def main():
net = loadUNet(None)
data = gpuarray.to_gpu(np.random.randn(1, 1, 256, 256).astype(np.float32))
engine = buildRTEngine(net,
inshape=data.shape,
savepath="../TestData",
dtype=DataType.float32)
net.evalMode()
outdata = net(data)
enginedata = engine(data)
assert np.allclose(outdata.get(), enginedata.get())
benchModels(net, engine, data)
def main():
inmaps = 161
net = loadW2L(None, inmaps, nlabels=29)
data = gpuarray.to_gpu(np.random.randn(1, inmaps, 200).astype(np.float32))
engine = buildRTEngine(net,
inshape=data.shape,
savepath="../TestData",
dtype=DataType.float32)
net.evalMode()
outdata = net(data)
enginedata = engine(data)
assert np.allclose(outdata.get(), enginedata.get())
benchModels(net, engine, data)
def main():
mnist = MnistLoader()
data, labels = mnist.load(path="../TestData/")
data, labels = data[:], labels[:]
print("Loaded mnist")
np.random.seed(1234)
net = buildNet()
trainNet(net, data, labels, 15)
calibrator = DataCalibrator(data[:60000])
net.evalMode()
engine = buildRTEngine(
net, inshape=data[:1].shape, savepath="../TestData", dtype=DataType.int8, calibrator=calibrator
)
benchModels(net, engine, gpuarray.to_gpu(data[:1]))
print("Net accuracy: %s" % validate(net, data, labels))
print("Engine accuracy: %s" % validate(engine, data, labels, batchsize=1))
def main():
inshape = (1, 3, 368, 368)
net = loadNet("../TestData/pose_iter_116000.hdf")
net.optimizeForShape(inshape)
outshape = net.dataShapeFrom(inshape)
engine = buildRTEngineFromCaffe(
("../TestData/pose_deploy.prototxt",
"../TestData/pose_iter_116000.caffemodel"),
inshape=inshape,
outshape=outshape,
outlayers=["net_output"],
dtype=DataType.float32,
savepath="../TestData")
data = gpuarray.to_gpu(np.random.randn(*inshape).astype(np.float32))
netData = net(data).get()
engineData = engine(data).get()
assert np.allclose(netData, engineData)
benchModels(net, engine, data)