README: drop checkpoint epoch/step numbers

README.md

@@ -19,7 +19,8 @@ reconstruction (chunk_size=4, left_context=32).
 
 ## Checkpoint
 
-`final.pt` = phase-C (cosine) epoch 13. Dict keys:
+`final.pt` is the released tokenizer, taken from the end of the cosine
+phase. Dict keys:
 `{epoch, hubert_state_dict, optimizer_state_dict, train_loss}`.
 Load `hubert_state_dict` into
 `models/model_tokenizer.py::Data2VecSemanticAcousticModel`