fix bi* models on OrangePi (#2207)

Author: lvyufeng

mindtorch/_apis/npu.py

@@ -117,7 +117,7 @@ def add(input, other, alpha=1.0): # pylint: disable=unused-argument
     Returns:
         Tensor: The result of the addition.
     """
-    if use_pyboost():
+    if use_pyboost() and not ON_ORANGE_PI:
         return pyboost.add_ext_op(input, other, alpha)
     if alpha == 1.0:
         return legacy.add(input, other)
@@ -724,7 +724,7 @@ def less_equal(input, other):
 
 def select(condition, input, other):
     if ON_ORANGE_PI:
-        return add(mul(condition, input), mul(bitwise_not(condition), other))
+        return legacy.add(mul(condition, input), mul(bitwise_not(condition), other))
     if use_pyboost():
         return pyboost.select_op(condition, input, other)
     return legacy.select(condition, input, other)
@@ -975,8 +975,23 @@ def inplace_zero(input):
     return input
 
 def mse_loss(input, target, reduction):
-    if use_pyboost():
+    if use_pyboost() and not ON_ORANGE_PI:
         return pyboost.mse_loss_ext_op(input, target, reduction)
+    x = square(input - target)
+    average_flag = True
+    reduce_flag = True
+    if reduction == 'sum':
+        average_flag = False
+    if reduction == 'none':
+        reduce_flag = False
+
+    if reduce_flag and average_flag:
+        x = mean(x, tuple(range(x.ndim)), False, None)
+
+    if reduce_flag and not average_flag:
+        x = sum(x, tuple(range(x.ndim)), False, None)
+
+    return x
 
 def abs(input):
     if use_pyboost():
@@ -1126,7 +1141,7 @@ def pow_scalar_tensor(input, scalar):
     return legacy.pow(input, scalar)
 
 def adaptive_avg_pool2d(input, output_size):
-    if use_pyboost():
+    if use_pyboost() and not ON_ORANGE_PI:
         return pyboost.adaptive_avg_pool2d_ext_op(input, output_size)
     return legacy.adaptive_avg_pool2_d(input, output_size)
 
@@ -1362,6 +1377,21 @@ def _check_maxpool_padding(padding, nd):
         return (0,) * (3 - nd) + tuple(padding)
     return padding
 
+def _cal_dilation(dilation, nd):
+    """check the dilation"""
+    if isinstance(dilation, int):
+        return dilation
+    if isinstance(dilation, tuple):
+        if len(dilation) == 1:
+            return dilation[0]
+        if len(dilation) == nd:
+            return (3 - nd) * (1,) + dilation
+        if nd == 1:
+            raise ValueError(f"the length of 'dilation' must be 1, but got {len(dilation)}.")
+        raise ValueError(f"the length of 'dilation' must be 1 or {nd}, but got {len(dilation)}.")
+    raise ValueError(f"the 'dilation' must be int or tuple, but got {type(dilation)}.")
+
+
 def max_pool2d(input, kernel_size, stride=1, padding=0, dilation=1, ceil_mode=False, return_indices=False):
     # out, indices = legacy.max_pool_with_argmax_v2(input, kernel_size, stride, padding, dilation, ceil_mode)
     if not ON_ORANGE_PI:
@@ -1379,6 +1409,7 @@ def max_pool2d(input, kernel_size, stride=1, padding=0, dilation=1, ceil_mode=Fa
     elif isinstance(stride, int):
         stride = (1, stride, stride)
     padding = _check_maxpool_padding(padding, 2)
+    dilation = _cal_dilation(dilation, 2)
 
     input = expand_dims(input, 2)
     out, indices = legacy.max_pool3_d_with_argmax(input, kernel_size, stride, padding,
@@ -1550,9 +1581,9 @@ def outer(input, other):
     return legacy.outer(input, other)
 
 def addcmul(input, tensor1, tensor2, value=1.0):
-    if use_pyboost():
+    if use_pyboost() and not ON_ORANGE_PI:
         return pyboost.addcmul_op(input, tensor1, tensor2, value)
-    return legacy.addcmul(input, tensor1, tensor2, value)
+    return legacy.add(mul(mul(tensor1, tensor2), value), input)
 
 def prelu(input, weight):
     if use_pyboost():

mindtorch/_tensor.py

@@ -287,6 +287,8 @@ def __setitem__(self, slices, value):
     def __add__(self, other):
         # if 0 in self.shape:
         #     return self
+        if self.dtype == mindtorch.bool:
+            return ops.bitwise_or(self, other)
         return ops.add(self, other)
 
     def __iadd__(self, other):

mindtorch/nn/functional.py

@@ -9,7 +9,7 @@
 from mindtorch._C import default_generator
 from mindtorch.nn.modules.utils import _pair
 
-from ..configs import ON_A2, ON_A1, FLASH_ATTN_MASK_VALID
+from ..configs import ON_A2, ON_A1, ON_ORANGE_PI, FLASH_ATTN_MASK_VALID
 
 generator_step_ = 12
 
@@ -991,7 +991,7 @@ def group_norm(input, num_groups, weight=None, bias=None, eps=1e-5):
         weight = mindtorch.ones([input.shape[1]], dtype=input.dtype, device=input.device)
     if bias is None:
         bias = mindtorch.zeros([input.shape[1]], dtype=input.dtype, device=input.device)
-    if input.device.type == 'npu':
+    if input.device.type == 'npu' and not ON_ORANGE_PI:
         return execute('group_norm', input, num_groups, weight, bias, eps)[0]
 
     input_shape = input.shape