!13825 Fix crnn precision by remove manul float16 assignment

From: @c_34
Reviewed-by: @wuxuejian,@linqingke
Signed-off-by: @linqingke

model_zoo/official/cv/crnn/eval.py

@@ -22,7 +22,7 @@ from mindspore.train.serialization import load_checkpoint, load_param_into_net
 
 from src.loss import CTCLoss
 from src.dataset import create_dataset
-from src.crnn import CRNN
+from src.crnn import crnn
 from src.metric import CRNNAccuracy
 
 set_seed(1)
@@ -60,7 +60,7 @@ if __name__ == '__main__':
     loss = CTCLoss(max_sequence_length=config.num_step,
                    max_label_length=max_text_length,
                    batch_size=config.batch_size)
-    net = CRNN(config)
+    net = crnn(config)
     # load checkpoint
     param_dict = load_checkpoint(args_opt.checkpoint_path)
     load_param_into_net(net, param_dict)

model_zoo/official/cv/crnn/export.py

@@ -20,7 +20,7 @@ import numpy as np
 import mindspore as ms
 from mindspore import Tensor, context, load_checkpoint, export
 
-from src.crnn import CRNN
+from src.crnn import crnn
 from src.config import config1 as config
 
 parser = argparse.ArgumentParser(description="CRNN_export")
@@ -37,7 +37,7 @@ if args.device_target == "Ascend":
 
 if __name__ == "__main__":
     config.batch_size = 1
-    net = CRNN(config)
+    net = crnn(config)
 
     load_checkpoint(args.ckpt_file, net=net)
     net.set_train(False)

model_zoo/official/cv/crnn/src/crnn.py

@@ -96,28 +96,28 @@ class CRNN(nn.Cell):
         self.rnn2_bw = P.DynamicRNN(forget_bias=0.0)
 
         w1 = np.random.uniform(-k, k, (self.input_size + self.hidden_size, 4 * self.hidden_size))
-        self.w1 = Parameter(w1.astype(np.float16), name="w1")
+        self.w1 = Parameter(w1.astype(np.float32), name="w1")
         w2 = np.random.uniform(-k, k, (2 * self.hidden_size + self.hidden_size, 4 * self.hidden_size))
-        self.w2 = Parameter(w2.astype(np.float16), name="w2")
+        self.w2 = Parameter(w2.astype(np.float32), name="w2")
         w1_bw = np.random.uniform(-k, k, (self.input_size + self.hidden_size, 4 * self.hidden_size))
-        self.w1_bw = Parameter(w1_bw.astype(np.float16), name="w1_bw")
+        self.w1_bw = Parameter(w1_bw.astype(np.float32), name="w1_bw")
         w2_bw = np.random.uniform(-k, k, (2 * self.hidden_size + self.hidden_size, 4 * self.hidden_size))
-        self.w2_bw = Parameter(w2_bw.astype(np.float16), name="w2_bw")
+        self.w2_bw = Parameter(w2_bw.astype(np.float32), name="w2_bw")
 
-        self.b1 = Parameter(np.random.uniform(-k, k, (4 * self.hidden_size)).astype(np.float16), name="b1")
-        self.b2 = Parameter(np.random.uniform(-k, k, (4 * self.hidden_size)).astype(np.float16), name="b2")
-        self.b1_bw = Parameter(np.random.uniform(-k, k, (4 * self.hidden_size)).astype(np.float16), name="b1_bw")
-        self.b2_bw = Parameter(np.random.uniform(-k, k, (4 * self.hidden_size)).astype(np.float16), name="b2_bw")
+        self.b1 = Parameter(np.random.uniform(-k, k, (4 * self.hidden_size)).astype(np.float32), name="b1")
+        self.b2 = Parameter(np.random.uniform(-k, k, (4 * self.hidden_size)).astype(np.float32), name="b2")
+        self.b1_bw = Parameter(np.random.uniform(-k, k, (4 * self.hidden_size)).astype(np.float32), name="b1_bw")
+        self.b2_bw = Parameter(np.random.uniform(-k, k, (4 * self.hidden_size)).astype(np.float32), name="b2_bw")
 
-        self.h1 = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float16))
-        self.h2 = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float16))
-        self.h1_bw = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float16))
-        self.h2_bw = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float16))
+        self.h1 = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float32))
+        self.h2 = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float32))
+        self.h1_bw = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float32))
+        self.h2_bw = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float32))
 
-        self.c1 = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float16))
-        self.c2 = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float16))
-        self.c1_bw = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float16))
-        self.c2_bw = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float16))
+        self.c1 = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float32))
+        self.c2 = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float32))
+        self.c1_bw = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float32))
+        self.c2_bw = Tensor(np.zeros(shape=(1, self.batch_size, self.hidden_size)).astype(np.float32))
 
         self.fc_weight = np.random.random((self.num_classes, self.hidden_size)).astype(np.float32)
         self.fc_bias = np.random.random((self.num_classes)).astype(np.float32)
@@ -142,7 +142,6 @@ class CRNN(nn.Cell):
 
     def construct(self, x):
         x = self.vgg(x)
-        x = self.cast(x, mstype.float16)
 
         x = self.reshape(x, (self.batch_size, self.input_size, -1))
         x = self.transpose(x, (2, 0, 1))
@@ -169,3 +168,11 @@ class CRNN(nn.Cell):
             output += (y2_after_fc,)
         output = self.concat(output)
         return output
+
+
+def crnn(config, full_precision=False):
+    """Create a CRNN network with mixed_precision or full_precision"""
+    net = CRNN(config)
+    if not full_precision:
+        net = net.to_float(mstype.float16)
+    return net

model_zoo/official/cv/crnn/train.py

@@ -26,7 +26,7 @@ from mindspore.communication.management import init, get_group_size, get_rank
 
 from src.loss import CTCLoss
 from src.dataset import create_dataset
-from src.crnn import CRNN
+from src.crnn import crnn
 from src.crnn_for_train import TrainOneStepCellWithGradClip
 
 set_seed(1)
@@ -83,7 +83,7 @@ if __name__ == '__main__':
     loss = CTCLoss(max_sequence_length=config.num_step,
                    max_label_length=max_text_length,
                    batch_size=config.batch_size)
-    net = CRNN(config)
+    net = crnn(config)
     opt = nn.SGD(params=net.trainable_params(), learning_rate=lr, momentum=config.momentum, nesterov=config.nesterov)
 
     net = WithLossCell(net, loss)