@@ -141,3 +141,25 @@ def test_run_onnxdl_model(self):
141
141
con .modelrun ("onnx_model" , ["input" ], ["output" ])
142
142
outtensor = con .tensorget ("output" )
143
143
self .assertTrue (np .allclose (outtensor , [4.0 ]))
144
+
145
+ def test_run_pytorch_model (self ):
146
+ model_path = os .path .join (MODEL_DIR , 'pt-minimal.pt' )
147
+ ptmodel = load_model (model_path )
148
+ con = self .get_client ()
149
+ con .modelset ("pt_model" , Backend .torch , Device .cpu , ptmodel )
150
+ con .tensorset ('a' , [2 , 3 , 2 , 3 ], shape = (2 , 2 ), dtype = DType .float )
151
+ con .tensorset ('b' , [2 , 3 , 2 , 3 ], shape = (2 , 2 ), dtype = DType .float )
152
+ con .modelrun ("pt_model" , ["a" , "b" ], "output" )
153
+ output = con .tensorget ('output' , as_numpy = False )
154
+ self .assertTrue (np .allclose (output .value , [4 , 6 , 4 , 6 ]))
155
+
156
+ def test_run_tflite_model (self ):
157
+ model_path = os .path .join (MODEL_DIR , 'mnist_model_quant.tflite' )
158
+ tflmodel = load_model (model_path )
159
+ con = self .get_client ()
160
+ con .modelset ("tfl_model" , Backend .tflite , Device .cpu , tflmodel )
161
+ img = np .random .random ((1 , 1 , 28 , 28 )).astype (np .float )
162
+ con .tensorset ('img' , img )
163
+ con .modelrun ("tfl_model" , "img" , ["output1" , "output2" ])
164
+ output = con .tensorget ('output1' )
165
+ self .assertTrue (np .allclose (output , [8 ]))
0 commit comments