Describe bfloat16 support in README

README.md

@@ -18,9 +18,8 @@ OpenFold is built to support inference with AlphaFold's original JAX weights.
 Try it out with our [Colab notebook](https://colab.research.google.com/github/aqlaboratory/openfold/blob/main/notebooks/OpenFold.ipynb).
 
 Unlike DeepMind's public code, OpenFold is also trainable. It can be trained 
-with [DeepSpeed](https://github.com/microsoft/deepspeed) and with mixed 
-precision. `bfloat16` training is not currently supported, but will be in the 
-future.
+with [DeepSpeed](https://github.com/microsoft/deepspeed) and with both `fp16`
+and `bfloat16` half-precision.
 
 ## Installation (Linux)
 
@@ -208,6 +207,10 @@ and supports the full range of training options that entails, including
 multi-node distributed training. For more information, consult PyTorch 
 Lightning documentation and the `--help` flag of the training script.
 
+Hardware permitting, you can train with `bfloat16` half-precision by passing
+`bf16` as the `--precision` option. If you're using DeepSpeed, make sure to
+enable `bfloat16` in the DeepSpeed config as well.
+
 Note that the data directory can also contain PDB files previously output by
 the model. These are treated as members of the self-distillation set and are
 subjected to distillation-set-only preprocessing steps.