fix b class models on OrangePi (#2210)

Author: lvyufeng

mindtorch/_apis/npu.py

@@ -62,10 +62,10 @@ def inplace_copy(self, value):
     Args:
         value (Tensor): The tensor from which to copy the data.
     """
-    if use_pyboost:
+    if use_pyboost():
         return pyboost.inplace_copy_op(self, value)
     else:
-        self.assign_value(value)
+        legacy.assign(self, value)
         return self
 
 def slice(input, dim, start, end, step):
@@ -85,7 +85,15 @@ def slice(input, dim, start, end, step):
     if use_pyboost():
         return pyboost.slice_ext_op(input, dim, start, end, step)
     else:
-        return legacy.slice(input, dim, start, end, step)
+        ndim = input.ndim
+        begins = [0] * ndim
+        ends = [i for i in input.shape]
+        strides = [1] * ndim
+        begins[dim] = start
+        ends[dim] = end
+        strides[dim] = step
+        return legacy.strided_slice(input, tuple(begins), tuple(ends), tuple(strides), 0, 0, 0, 0, 0)
+
 
 def embedding(input, weight, padding_idx, max_norm, norm_type, scale_grad_by_freq):
     """
@@ -829,7 +837,7 @@ def bmm(input, other):
     return legacy.batch_mat_mul(input, other, False, False)
 
 def topk(input, k, dim, largest, sorted):
-    if use_pyboost():
+    if use_pyboost() and not ON_ORANGE_PI:
         return pyboost.topk_ext_op(input, k, dim, largest, sorted)
 
     if not largest:
@@ -1296,9 +1304,9 @@ def remainder_tensor_scalar(input, other):
     return out
 
 def baddbmm(input, batch1, batch2, alpha=1, beta=1):
-    if use_pyboost():
+    if use_pyboost() and not ON_ORANGE_PI:
         return pyboost.baddbmm_op(input, batch1, batch2, alpha, beta)
-    return legacy.baddbmm(input, batch1, batch2, alpha, beta)
+    return add(mul(input, beta), mul(bmm(batch1, batch2), alpha))
 
 def floor(input):
     if use_pyboost():
@@ -1844,4 +1852,16 @@ def cumprod(input, dim, dtype):
     out = legacy.cum_prod(input, dim, False, False)
     if dtype is not None:
         out = cast(out, dtype)
-    return out
+    return out
+
+def scatter_nd_update(input, indices, updates):
+    return legacy.scatter_nd_update(input, indices, updates, True)
+
+def assign(input, value):
+    return inplace_copy(input, value)
+
+def strided_slice(input, begin, end, strides, begin_mask=0, end_mask=0, ellipsis_mask=0, new_axis_mask=0, shrink_axis_mask=0):
+    return legacy.strided_slice(input, tuple(begin), tuple(end), tuple(strides), begin_mask, end_mask, ellipsis_mask, new_axis_mask, shrink_axis_mask)
+
+def tensor_scatter_update(input, indices, updates):
+    return legacy.tensor_scatter_update(input, indices, updates)

mindtorch/_tensor.py

@@ -23,7 +23,7 @@ class StubTensor: pass
 from ._bind import get_device_in_context, device_, get_default_dtype
 from ._utils import _rebuild_tensor_v2
 from ._C.size import Size
-from .configs import DEVICE_TARGET, cpu_use_numpy
+from .configs import DEVICE_TARGET, cpu_use_numpy, ON_ORANGE_PI
 
 device_map = {
     'cpu': 'CPU',
@@ -282,7 +282,7 @@ def __setitem__(self, slices, value):
         if value.device != self.device:
             value._device = self.device
 
-        if self.device.type == 'npu':
+        if self.device.type == 'npu' and not ON_ORANGE_PI:
             if value.device != self.device:
                 value._device = self.device
             out = ops.tensor_setitem(self, slices, value)
@@ -301,7 +301,7 @@ def __iadd__(self, other):
         return self.copy_(ops.add(self, other))
 
     def __radd__(self, other):
-        return Tensor.__add__(other, self)
+        return ops.add(other, self)
 
     def __div__(self, other):
         # if 0 in self.shape:

mindtorch/nn/modules/activation.py

@@ -1804,7 +1804,7 @@ def forward(self, input: Tensor) -> Tensor:
         """
         Runs the forward pass.
         """
-        return F.softmax(input, self.dim, _stacklevel=5)
+        return F.softmax(input, self.dim)
 
     def extra_repr(self) -> str:
         """

mindtorch/ops/array.py

@@ -634,6 +634,7 @@ def _process_multi_dim_index(self, indexes, remain_indexes, indexed_dims):
                 raise TypeError(f"Index {index} contain unsupported elements")
         self_viewed, dim, remain_indexes, self_viewed_shape = _process_dim_in_multi_dim_index(
             self_viewed, self, index, dim, indexed_dims, i, remain_indexes, self_viewed_shape)
+
     return self_viewed, remain_indexes
 
 
@@ -1162,7 +1163,7 @@ def strided_slice_update(x, begin, end, strides, updates,
     # for i in range(ndim-1, -1, -1):
     #     if (shrink_axis_mask >> i) & 1:
     #         x_updated = mindtorch.squeeze(x_updated, dim=i)
-    
+
     return x_updated