Update README.md

README.md

@@ -31,6 +31,8 @@ MolmoAct is trained on a subset of OXE and MolmoAct Dataset, a dataset with 10k
 
 This checkpoint is a **preview** of the MolmoAct release. All artifacts used in creating MolmoAct (data, training code, evaluations, intermediate checkpoints) will be made available at a later date, furthering our commitment to open-source AI development and reproducibility.
 
+**Update:** Checkpoints are now stored in FP32 (previously BF16). The model was trained in FP32, so publishing FP32 weights aligns with training and enables fine-tuning or continued training directly from this repo. For inference, you can still run BF16 by casting at load, which is what we did for evaluations. See more in the [instructions](#quick-start) below.
+
 Quick links:
 - 📂 [All Models](https://huggingface.co/collections/allenai/molmoact-689697591a3936fba38174d7)
 - 📂 [All Data](https://huggingface.co/collections/allenai/molmoact-data-mixture-6897e583e13b6c2cf3ea2b80)
@@ -64,7 +66,7 @@ ckpt = "allenai/MolmoAct-7B-D-Pretrain-RT-1-0812"
 processor = AutoProcessor.from_pretrained(
     ckpt,
     trust_remote_code=True,
-    torch_dtype="auto",
+    torch_dtype="bfloat16",
     device_map="auto",
     padding_side="left",
 )
@@ -73,7 +75,7 @@ processor = AutoProcessor.from_pretrained(
 model = AutoModelForImageTextToText.from_pretrained(
     ckpt,
     trust_remote_code=True,
-    torch_dtype="auto",
+    torch_dtype="bfloat16",
     device_map="auto",
 )