[dreambooth] low precision guard (#1916)

* [dreambooth] low precision guard

* fix

* add docs to cli args

* Update examples/dreambooth/train_dreambooth.py
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* style
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

examples/dreambooth/train_dreambooth.py

@@ -70,7 +70,10 @@ def parse_args(input_args=None):
         type=str,
         default=None,
         required=False,
-        help="Revision of pretrained model identifier from huggingface.co/models.",
+        help=(
+            "Revision of pretrained model identifier from huggingface.co/models. Trainable model components should be"
+            " float32 precision."
+        ),
     )
     parser.add_argument(
         "--tokenizer_name",
@@ -140,7 +143,11 @@ def parse_args(input_args=None):
     parser.add_argument(
         "--center_crop", action="store_true", help="Whether to center crop images before resizing to resolution"
     )
-    parser.add_argument("--train_text_encoder", action="store_true", help="Whether to train the text encoder")
+    parser.add_argument(
+        "--train_text_encoder",
+        action="store_true",
+        help="Whether to train the text encoder. If set, the text encoder should be float32 precision.",
+    )
     parser.add_argument(
         "--train_batch_size", type=int, default=4, help="Batch size (per device) for the training dataloader."
     )
@@ -671,6 +678,17 @@ def main(args):
     if not args.train_text_encoder:
         text_encoder.to(accelerator.device, dtype=weight_dtype)
 
+    low_precision_error_string = (
+        "Please make sure to always have all model weights in full float32 precision when starting training - even if"
+        " doing mixed precision training. copy of the weights should still be float32."
+    )
+
+    if unet.dtype != torch.float32:
+        raise ValueError(f"Unet loaded as datatype {unet.dtype}. {low_precision_error_string}")
+
+    if args.train_text_encoder and text_encoder.dtype != torch.float32:
+        raise ValueError(f"Text encoder loaded as datatype {text_encoder.dtype}. {low_precision_error_string}")
+
     # We need to recalculate our total training steps as the size of the training dataloader may have changed.
     num_update_steps_per_epoch = math.ceil(len(train_dataloader) / args.gradient_accumulation_steps)
     if overrode_max_train_steps: