p83571920
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fix some bug for DynamicGRUV2.

This commit is contained in:
liuxiao93 2020-12-22 15:10:49 +08:00
parent 3836665f29
commit 4817978d28
1 changed files with 18 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
mindspore/ops/operations/nn_ops.py
View File

@ -6491,7 +6491,7 @@ class DynamicRNN(PrimitiveWithInfer):
class DynamicGRUV2(PrimitiveWithInfer):
r"""
DynamicGRUV2 Operator.
Applies a single-layer gated recurrent unit (GRU) to an input sequence.
Args:
direction (str): A string identifying the direction in the op. Default: 'UNIDIRECTIONAL'.
@ -6532,19 +6532,19 @@ class DynamicGRUV2(PrimitiveWithInfer):
- **y** (Tensor) - A Tensor of shape :math:
if num_proj > 0 `(num_step, batch_size, min(hidden_size, num_proj)`,
if num_proj == 0 `(num_step, batch_size, hidden_size)`.
Has the same data type with input `bais_type`.
Has the same data type with input `bias_type`.
- **output_h** (Tensor) - A Tensor of shape :math:`(\text{num_step}, \text{batch_size}, \text{hidden_size})`.
Has the same data type with input `bais_type`.
Has the same data type with input `bias_type`.
- **update** (Tensor) - A Tensor of shape :math:`(\text{num_step}, \text{batch_size}, \text{hidden_size})`.
Has the same data type with input `bais_type`.
Has the same data type with input `bias_type`.
- **reset** (Tensor) - A Tensor of shape :math:`(\text{num_step}, \text{batch_size}, \text{hidden_size})`.
Has the same data type with input `bais_type`.
Has the same data type with input `bias_type`.
- **new** (Tensor) - A Tensor of shape :math:`(\text{num_step}, \text{batch_size}, \text{hidden_size})`.
Has the same data type with input `bais_type`.
Has the same data type with input `bias_type`.
- **hidden_new** (Tensor) - A Tensor of shape :math:`(\text{num_step}, \text{batch_size}, \text{hidden_size})`.
Has the same data type with input `bais_type`.
Has the same data type with input `bias_type`.
- If `bias_input` and `bias_hidden` both are `None`, `bias_type` is float32.
- If `bias_input` and `bias_hidden` both are `None`, `bias_type` is date type of `init_h`.
- If `bias_input` is not `None`, `bias_type` is the date type of `bias_input`.
- If `bias_input` is `None` and `bias_hidden` is not `None, `bias_type` is the date type of `bias_hidden`.
@ -6563,6 +6563,15 @@ class DynamicGRUV2(PrimitiveWithInfer):
>>> print(output[0].shape)
(2, 8, 16)
"""
__mindspore_signature__ = (
sig.make_sig('x', dtype=sig.sig_dtype.T1),
sig.make_sig('weight_input', dtype=sig.sig_dtype.T2),
sig.make_sig('weight_hidden', dtype=sig.sig_dtype.T3),
sig.make_sig('bias_input', dtype=sig.sig_dtype.T),
sig.make_sig('bias_hidden', dtype=sig.sig_dtype.T),
sig.make_sig('seq_length', dtype=sig.sig_dtype.T4),
sig.make_sig('init_h', dtype=sig.sig_dtype.T),
)
@prim_attr_register
def __init__(self,
@ -6631,7 +6640,7 @@ class DynamicGRUV2(PrimitiveWithInfer):
validator.check_tensor_dtype_valid("weight input dtype", winput_dtype, [mstype.float16], self.name)
validator.check_tensor_dtype_valid("weight hidden dtype", whidden_dtype, [mstype.float16], self.name)
validator.check_tensor_dtype_valid("init_h dtype", h_dtype, (mstype.float16, mstype.float32), self.name)
b_dtype = mstype.float32
b_dtype = h_dtype
if binput_dtype is not None:
validator.check_tensor_dtype_valid("bias input dtype", binput_dtype,
(mstype.float16, mstype.float32), self.name)