deploy: braindecode/braindecode@ad83587

Author: bruAristimunha

dev/_downloads/07fcc19ba03226cd3d83d4e40ec44385/auto_examples_python.zip

@@ -94,7 +94,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "## Loading a pre-trained foundation model\n\n### Load Pre-trained Weights from the Hub\n\nWe load the pre-trained SignalJEPA downstream model from the Hugging Face\nHub using ``from_pretrained``.  The ``SignalJEPA_PreLocal`` checkpoint\nalready bundles the SSL backbone together with the downstream\nclassification layers, so a single call is all that is needed.\n\nFor other foundation models (BENDR, BIOT, Labram, etc.) the same\none-line pattern applies \u2014 see `load-pretrained-models`.\n\n\n"
+        "## Loading a pre-trained foundation model\n\n### Load Pre-trained Weights from the Hub\n\nWe load the pre-trained SignalJEPA downstream model from the Hugging Face\nHub using ``from_pretrained``.  The ``SignalJEPA_PreLocal`` checkpoint\nalready bundles the SSL backbone together with the downstream\nclassification layers, so a single call is all that is needed.\n\nFor other foundation models (BENDR, BIOT, Labram, STEEGFormer, etc.) the\nsame one-line pattern applies \u2014 see `load-pretrained-models`. For\nexample, the braindecode re-host of the STEEGFormer small checkpoint can be\nloaded as ``STEEGFormer.from_pretrained(\"braindecode/STEEGFormer-small\",\nn_outputs=len(classes), chs_info=chs_info)`` -- pass ``n_outputs`` to size the\nclassification head and ``chs_info`` to align your montage with the channel\nvocabulary.\n\n\n"
       ]
     },
     {