dssm train

recommend-model-produce/src/main/python/tools/inferv2.py

@@ -42,25 +42,34 @@ def main():
 
     # 根据 config 创建 predictor
     predictor = paddle_infer.create_predictor(config)
-
     # 获取输入的名称
     input_names = predictor.get_input_names()
     input_handle = predictor.get_input_handle(input_names[0])
+    output_names = predictor.get_output_names()
+    output_handle = predictor.get_output_handle(output_names[0])
 
-    # 设置输入
-    fake_input = np.abs(np.random.randn(1, 157).astype("float32"))
-    input_handle.reshape([1, 157])
-    input_handle.copy_from_cpu(fake_input)
 
-    # 运行predictor
-    predictor.run()
+    ret, out = hdfs_client._run_cmd("text /dw/recommend/model/56_dssm_i2i_itempredData/20241206/part-00016.gz")
+    input_data = {}
+    for line in out:
+        sample_values = line.rstrip('\n').split('\t')
+        vid, left_features_str = sample_values
+        left_features = [float(x) for x in left_features_str.split(',')]
+        input_data[vid] = left_features
+        
 
-    # 获取输出
-    output_names = predictor.get_output_names()
-    output_handle = predictor.get_output_handle(output_names[0])
-    output_data = output_handle.copy_to_cpu() # numpy.ndarray类型
-    print("Output data size is {}".format(output_data))
-    print("Output data shape is {}".format(fake_input))
+    input_data2 = {k: input_data[k] for k in list(input_data)[:5]}
+    # 设置输入
+    
+
+    for k,v in input_data2:
+        input_handle.copy_from_cpu(v)
+        # 运行predictor
+        predictor.run()
+        # 获取输出
+        output_data = output_handle.copy_to_cpu() # numpy.ndarray类型
+        print("input k:{} v:{}".format(k, fake_input))
+        print("Output {}".format(output_data))
 
 if __name__ == "__main__":
     main()