在paddle.tensor中加入float，int和complex等数据转换接口 #73351
Conversation
modified: python/paddle/base/dygraph/math_op_patch.py new file: test/legacy_test/test_tensor_float16_bfloat16.py
| @@ -104,6 +104,26 @@ def astype(self: Tensor, dtype: DTypeLike) -> Tensor: | |||
|
|
|||
| return _C_ops.cast(self, dtype) | |||
|
|
|||
| def float16(self: Tensor) -> Tensor: | |||
There was a problem hiding this comment.
这个是不是得整体加，包括fp32、fp64、int32等所有类型，而不是给某两个类型增加一个函数。
实现方式建议通过一个公共函数，批量注册
|
Sorry to inform you that 701fe26's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually. |
modified: python/paddle/base/dygraph/math_op_patch.py modified: test/legacy_test/test_tensor_float16_bfloat16.py
renamed: test/legacy_test/test_tensor_float16_bfloat16.py -> test/legacy_test/test_tensor_type_convert_api.py
Qin-sx
changed the title
Qin-sx
changed the title
modified: python/paddle/pir/math_op_patch.py modified: test/legacy_test/test_tensor_type_convert_api.py
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## develop #73351 +/- ##
===========================================
Coverage ? 100.00%
===========================================
Files ? 2
Lines ? 30
Branches ? 0
===========================================
Hits ? 30
Misses ? 0
Partials ? 0 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
| 'int32': 'int32', | ||
| 'int64': 'int64', | ||
| # other | ||
| 'bool': 'bool', |
There was a problem hiding this comment.
这个应该还有一些其他的用法吧，或者别名用法，也需要配置一下：
x.long、x.double、x.cfloat、x.uint8
这个要配全
There was a problem hiding this comment.
嗯，收到，已修改，之前是按照上面的文档的tensor type写的
| @@ -37,6 +37,23 @@ | |||
| DataType.INT64, | |||
| ] | |||
|
|
|||
| _supported_dtype_conversions = { | |||
| Unit tests for all supported tensor dtype conversion methods. | ||
| """ | ||
|
|
||
| _supported_dtype_conversions = { |
modified: python/paddle/base/dygraph/math_op_patch.py modified: python/paddle/pir/math_op_patch.py modified: test/legacy_test/test_tensor_type_convert_api.py
|
Sorry to inform you that 5dc2126's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually. |
PR Category
User Experience
PR Types
Improvements
Description
当前问题
paddle中没有float16和bfloat16等数据转换接口，需要用astype转换
解决方案
注册float16和bfloat16等数据转换接口