# Training with Half Precision

**URL:** <https://discuss.pytorch.org/t/training-with-half-precision/11815>\
**Category:** vision\
**Created:** [January 3, 2018, 2:27pm UTC](https://discuss.pytorch.org/t/training-with-half-precision/11815 "2018-01-03T14:27:08Z")\
**Posts on this page:** 1\
**Showing post:** 17

<div class="post-metadata">

**Author:** ![ptrblck](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/ptrblck/32/1823_2.png) [@ptrblck](https://discuss.pytorch.org/u/ptrblck)\
**Post date:** [September 20, 2020, 11:08pm UTC](https://discuss.pytorch.org/t/training-with-half-precision/11815/17 "2020-09-20T23:08:28Z")

</div>

I would generally recommend to use the automatic mixed precision package (via [`torch.cuda.amp`](https://pytorch.org/docs/stable/amp.html)), which uses casts the input to the appropriate `dtype` for each method.

---

_[View the full topic](https://discuss.pytorch.org/t/training-with-half-precision/11815)._