[XLA] log a WARNING for out of range F64 constants instead of a check…

… failure.

PiperOrigin-RevId: 336091777
Change-Id: I4344ac47d23102e6092d72e7fe665c5d8505f02a

tensorflow/compiler/xla/util.cc

@@ -374,7 +374,9 @@ std::pair<float, float> SplitF64ToF32(double x) {
 
   // Only values within the range of F32 are supported, unless it is infinity.
   // Small values with large negative exponents would be rounded to zero.
-  CHECK(std::isfinite(x_f32)) << x;
+  if (!std::isfinite(x_f32)) {
+    LOG(WARNING) << "Out of range F64 constant detected: " << x;
+  }
 
   // The high float is simply the double rounded to the nearest float. Because
   // we are rounding to nearest with ties to even, the error introduced in