Improve repr for torch.iinfo & torch.finfo (#40488)

Summary: - fix pytorch/pytorch#39991 - Include directly `min`/`max`/`eps`/`tiny` values in repr of `torch.iinfo` & `torch.finfo` for inspection - Use `torch.float16` / `torch.int16` instead of uncorrespond names `Half` / `Short` - The improved repr is shown just like: ``` >>> torch.iinfo(torch.int8) iinfo(type=torch.int8, max=127, min=-128) >>> torch.iinfo(torch.int16) iinfo(type=torch.int16, max=32767, min=-32768) >>> torch.iinfo(torch.int32) iinfo(type=torch.int32, max=2.14748e+09, min=-2.14748e+09) >>> torch.iinfo(torch.int64) iinfo(type=torch.int64, max=9.22337e+18, min=-9.22337e+18) >>> torch.finfo(torch.float16) finfo(type=torch.float16, eps=0.000976563, max=65504, min=-65504, tiny=6.10352e-05) >>> torch.finfo(torch.float32) finfo(type=torch.float32, eps=1.19209e-07, max=3.40282e+38, min=-3.40282e+38, tiny=1.17549e-38) >>> torch.finfo(torch.float64) finfo(type=torch.float64, eps=2.22045e-16, max=1.79769e+308, min=-1.79769e+308, tiny=2.22507e-308) ``` Pull Request resolved: pytorch/pytorch#40488 Differential Revision: D22445301 Pulled By: mruberry fbshipit-source-id: 552af9904c423006084b45d6c4adfb4b5689db54
Sconcer · Jul 10, 2020 · 0651887 · 0651887
1 parent cb6c352
commit 0651887
Show file tree

Hide file tree

Showing 6 changed files with 91 additions and 27 deletions.
diff --git a/docs/source/type_info.rst b/docs/source/type_info.rst
@@ -27,6 +27,7 @@ eps         float   The smallest representable number such that ``1.0 + eps != 1
 max         float   The largest representable number.
 min         float   The smallest representable number (typically ``-max``).
 tiny        float   The smallest positive representable number.
+resolution  float   The approximate decimal resolution of this type, i.e., ``10**-precision``.
 =========   =====   ========================================
 
 .. note::

diff --git a/test/test_type_info.py b/test/test_type_info.py
@@ -14,29 +14,30 @@
 class TestDTypeInfo(TestCase):
 
     def test_invalid_input(self):
-        for dtype in [torch.float32, torch.float64]:
+        for dtype in [torch.float16, torch.float32, torch.float64, torch.bfloat16, torch.complex64, torch.complex128, torch.bool]:
             with self.assertRaises(TypeError):
                 _ = torch.iinfo(dtype)
 
-        for dtype in [torch.int64, torch.int32, torch.int16, torch.uint8]:
+        for dtype in [torch.int64, torch.int32, torch.int16, torch.int8, torch.uint8, torch.bool]:
             with self.assertRaises(TypeError):
                 _ = torch.finfo(dtype)
 
     @unittest.skipIf(not TEST_NUMPY, "Numpy not found")
     def test_iinfo(self):
-        for dtype in [torch.int64, torch.int32, torch.int16, torch.uint8]:
+        for dtype in [torch.int64, torch.int32, torch.int16, torch.int8, torch.uint8]:
             x = torch.zeros((2, 2), dtype=dtype)
             xinfo = torch.iinfo(x.dtype)
             xn = x.cpu().numpy()
             xninfo = np.iinfo(xn.dtype)
             self.assertEqual(xinfo.bits, xninfo.bits)
             self.assertEqual(xinfo.max, xninfo.max)
             self.assertEqual(xinfo.min, xninfo.min)
+            self.assertEqual(xinfo.dtype, xninfo.dtype)
 
     @unittest.skipIf(not TEST_NUMPY, "Numpy not found")
     def test_finfo(self):
         initial_default_type = torch.get_default_dtype()
-        for dtype in [torch.float32, torch.float64]:
+        for dtype in [torch.float16, torch.float32, torch.float64, torch.complex64, torch.complex128]:
             x = torch.zeros((2, 2), dtype=dtype)
             xinfo = torch.finfo(x.dtype)
             xn = x.cpu().numpy()
@@ -46,8 +47,25 @@ def test_finfo(self):
             self.assertEqual(xinfo.min, xninfo.min)
             self.assertEqual(xinfo.eps, xninfo.eps)
             self.assertEqual(xinfo.tiny, xninfo.tiny)
-            torch.set_default_dtype(dtype)
-            self.assertEqual(torch.finfo(dtype), torch.finfo())
+            self.assertEqual(xinfo.resolution, xninfo.resolution)
+            self.assertEqual(xinfo.dtype, xninfo.dtype)
+            if not dtype.is_complex:
+                torch.set_default_dtype(dtype)
+                self.assertEqual(torch.finfo(dtype), torch.finfo())
+
+        # Special test case for BFloat16 type
+        x = torch.zeros((2, 2), dtype=torch.bfloat16)
+        xinfo = torch.finfo(x.dtype)
+        self.assertEqual(xinfo.bits, 16)
+        self.assertEqual(xinfo.max, 3.38953e+38)
+        self.assertEqual(xinfo.min, -3.38953e+38)
+        self.assertEqual(xinfo.eps, 0.0078125)
+        self.assertEqual(xinfo.tiny, 1.17549e-38)
+        self.assertEqual(xinfo.resolution, 0.01)
+        self.assertEqual(xinfo.dtype, "bfloat16")
+        torch.set_default_dtype(x.dtype)
+        self.assertEqual(torch.finfo(x.dtype), torch.finfo())
+
         # Restore the default type to ensure that the test has no side effect
         torch.set_default_dtype(initial_default_type)
 

diff --git a/torch/_C/__init__.pyi.in b/torch/_C/__init__.pyi.in
@@ -59,6 +59,7 @@ class iinfo:
     bits: _int
     min: _int
     max: _int
+    dtype: str
 
     def __init__(self, dtype: _dtype) -> None: ...
 
@@ -68,6 +69,8 @@ class finfo:
     max: _float
     eps: _float
     tiny: _float
+    resolution: _float
+    dtype: str
 
     @overload
     def __init__(self, dtype: _dtype) -> None: ...

diff --git a/torch/csrc/TypeInfo.cpp b/torch/csrc/TypeInfo.cpp
@@ -6,6 +6,7 @@
 #include <torch/csrc/utils/python_arg_parser.h>
 #include <torch/csrc/utils/python_numbers.h>
 #include <torch/csrc/utils/python_strings.h>
+#include <torch/csrc/utils/tensor_dtypes.h>
 
 #include <c10/util/Exception.h>
 
@@ -20,7 +21,7 @@ PyObject* THPFInfo_New(const at::ScalarType& type) {
   if (!self)
     throw python_error();
   auto self_ = reinterpret_cast<THPDTypeInfo*>(self.get());
-  self_->type = type;
+  self_->type = c10::toValueType(type);
   return self.release();
 }
 
@@ -34,18 +35,6 @@ PyObject* THPIInfo_New(const at::ScalarType& type) {
   return self.release();
 }
 
-PyObject* THPFInfo_str(THPFInfo* self) {
-  std::ostringstream oss;
-  oss << "finfo(type=" << self->type << ")";
-  return THPUtils_packString(oss.str().c_str());
-}
-
-PyObject* THPIInfo_str(THPIInfo* self) {
-  std::ostringstream oss;
-  oss << "iinfo(type=" << self->type << ")";
-  return THPUtils_packString(oss.str().c_str());
-}
-
 PyObject* THPFInfo_pynew(PyTypeObject* type, PyObject* args, PyObject* kwargs) {
   HANDLE_TH_ERRORS
   static torch::PythonArgParser parser({
@@ -63,7 +52,7 @@ PyObject* THPFInfo_pynew(PyTypeObject* type, PyObject* args, PyObject* kwargs) {
     AT_ASSERT(at::isFloatingType(scalar_type));
   } else {
     scalar_type = r.scalartype(0);
-    if (!at::isFloatingType(scalar_type)) {
+    if (!at::isFloatingType(scalar_type) && !at::isComplexType(scalar_type)) {
       return PyErr_Format(
           PyExc_TypeError,
           "torch.finfo() requires a floating point input type. Use torch.iinfo to handle '%s'",
@@ -123,7 +112,7 @@ static PyObject* THPDTypeInfo_bits(THPDTypeInfo* self, void*) {
 }
 
 static PyObject* THPFInfo_eps(THPFInfo* self, void*) {
-  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND1(at::kHalf,
+  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND2(at::kHalf, at::ScalarType::BFloat16,
       self->type, "epsilon", [] {
         return PyFloat_FromDouble(
             std::numeric_limits<
@@ -132,20 +121,20 @@ static PyObject* THPFInfo_eps(THPFInfo* self, void*) {
 }
 
 static PyObject* THPFInfo_max(THPFInfo* self, void*) {
-  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND1(at::kHalf, self->type, "max", [] {
+  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND2(at::kHalf, at::ScalarType::BFloat16, self->type, "max", [] {
     return PyFloat_FromDouble(
         std::numeric_limits<at::scalar_value_type<scalar_t>::type>::max());
   });
 }
 
 static PyObject* THPFInfo_min(THPFInfo* self, void*) {
-  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND1(at::kHalf, self->type, "min", [] {
+  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND2(at::kHalf, at::ScalarType::BFloat16, self->type, "lowest", [] {
     return PyFloat_FromDouble(
         std::numeric_limits<at::scalar_value_type<scalar_t>::type>::lowest());
   });
 }
 
-static PyObject* THPIInfo_max(THPFInfo* self, void*) {
+static PyObject* THPIInfo_max(THPIInfo* self, void*) {
   if (at::isIntegralType(self->type, /*includeBool=*/false)) {
     return AT_DISPATCH_INTEGRAL_TYPES(self->type, "max", [] {
       return THPUtils_packInt64(std::numeric_limits<scalar_t>::max());
@@ -157,7 +146,7 @@ static PyObject* THPIInfo_max(THPFInfo* self, void*) {
   });
 }
 
-static PyObject* THPIInfo_min(THPFInfo* self, void*) {
+static PyObject* THPIInfo_min(THPIInfo* self, void*) {
   if (at::isIntegralType(self->type, /*includeBool=*/false)) {
     return AT_DISPATCH_INTEGRAL_TYPES(self->type, "min", [] {
       return THPUtils_packInt64(std::numeric_limits<scalar_t>::lowest());
@@ -169,19 +158,69 @@ static PyObject* THPIInfo_min(THPFInfo* self, void*) {
   });
 }
 
+static PyObject* THPIInfo_dtype(THPIInfo* self, void*) {
+  std::string primary_name, legacy_name;
+  std::tie(primary_name, legacy_name) = torch::utils::getDtypeNames(self->type);
+  return AT_DISPATCH_INTEGRAL_TYPES(self->type, "dtype", [primary_name] {
+    return PyUnicode_FromString((char*)primary_name.data());
+  });
+}
+
 static PyObject* THPFInfo_tiny(THPFInfo* self, void*) {
-  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND1(at::kHalf, self->type, "min", [] {
+  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND2(at::kHalf, at::ScalarType::BFloat16, self->type, "min", [] {
     return PyFloat_FromDouble(
         std::numeric_limits<at::scalar_value_type<scalar_t>::type>::min());
   });
 }
 
+static PyObject* THPFInfo_resolution(THPFInfo* self, void*) {
+  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND2(at::kHalf, at::ScalarType::BFloat16, self->type, "digits10", [] {
+    return PyFloat_FromDouble(
+        std::pow(10, -std::numeric_limits<at::scalar_value_type<scalar_t>::type>::digits10));
+  });
+}
+
+static PyObject* THPFInfo_dtype(THPFInfo* self, void*) {
+  std::string primary_name, legacy_name;
+  std::tie(primary_name, legacy_name) = torch::utils::getDtypeNames(self->type);
+  return AT_DISPATCH_FLOATING_AND_COMPLEX_TYPES_AND2(at::kHalf, at::ScalarType::BFloat16, self->type, "dtype", [primary_name] {
+    return PyUnicode_FromString((char*)primary_name.data());
+  });
+}
+
+PyObject* THPFInfo_str(THPFInfo* self) {
+  std::ostringstream oss;
+  oss << "finfo(resolution=" << PyFloat_AsDouble(THPFInfo_resolution(self, nullptr));
+  oss << ", min=" << PyFloat_AsDouble(THPFInfo_min(self, nullptr));
+  oss << ", max=" << PyFloat_AsDouble(THPFInfo_max(self, nullptr));
+  oss << ", eps=" << PyFloat_AsDouble(THPFInfo_eps(self, nullptr));
+  oss << ", tiny=" << PyFloat_AsDouble(THPFInfo_tiny(self, nullptr));
+  oss << ", dtype=" << PyUnicode_AsUTF8(THPFInfo_dtype(self, nullptr)) << ")";
+
+  return THPUtils_packString(oss.str().c_str());
+}
+
+PyObject* THPIInfo_str(THPIInfo* self) {
+  auto type = self->type;
+  std::string primary_name, legacy_name;
+  std::tie(primary_name, legacy_name) = torch::utils::getDtypeNames(type);
+  std::ostringstream oss;
+
+  oss << "iinfo(min=" << PyFloat_AsDouble(THPIInfo_min(self, nullptr));
+  oss << ", max=" << PyFloat_AsDouble(THPIInfo_max(self, nullptr));
+  oss << ", dtype=" << PyUnicode_AsUTF8(THPIInfo_dtype(self, nullptr)) << ")";
+
+  return THPUtils_packString(oss.str().c_str());
+}
+
 static struct PyGetSetDef THPFInfo_properties[] = {
     {"bits", (getter)THPDTypeInfo_bits, nullptr, nullptr, nullptr},
     {"eps", (getter)THPFInfo_eps, nullptr, nullptr, nullptr},
     {"max", (getter)THPFInfo_max, nullptr, nullptr, nullptr},
     {"min", (getter)THPFInfo_min, nullptr, nullptr, nullptr},
     {"tiny", (getter)THPFInfo_tiny, nullptr, nullptr, nullptr},
+    {"resolution", (getter)THPFInfo_resolution, nullptr, nullptr, nullptr},
+    {"dtype", (getter)THPFInfo_dtype, nullptr, nullptr, nullptr},
     {nullptr}};
 
 static PyMethodDef THPFInfo_methods[] = {
@@ -232,6 +271,7 @@ static struct PyGetSetDef THPIInfo_properties[] = {
     {"bits", (getter)THPDTypeInfo_bits, nullptr, nullptr, nullptr},
     {"max", (getter)THPIInfo_max, nullptr, nullptr, nullptr},
     {"min", (getter)THPIInfo_min, nullptr, nullptr, nullptr},
+    {"dtype", (getter)THPIInfo_dtype, nullptr, nullptr, nullptr},
     {nullptr}};
 
 static PyMethodDef THPIInfo_methods[] = {

diff --git a/torch/csrc/utils/tensor_dtypes.cpp b/torch/csrc/utils/tensor_dtypes.cpp
@@ -10,7 +10,7 @@
 namespace torch {
 namespace utils {
 
-static std::pair<std::string, std::string> getDtypeNames(
+std::pair<std::string, std::string> getDtypeNames(
     at::ScalarType scalarType) {
   switch (scalarType) {
     case at::ScalarType::Byte:

diff --git a/torch/csrc/utils/tensor_dtypes.h b/torch/csrc/utils/tensor_dtypes.h
@@ -6,6 +6,8 @@
 
 namespace torch { namespace utils {
 
+std::pair<std::string, std::string> getDtypeNames(at::ScalarType scalarType);
+
 void initializeDtypes();
 
 }} // namespace torch::utils