Braindecode 1.3.0.dev171854083__tar.gz → 1.3.0.dev172315966__tar.gz

{braindecode-1.3.0.dev171854083/braindecode.egg-info → braindecode-1.3.0.dev172315966}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: braindecode
-Version: 1.3.0.dev171854083
+Version: 1.3.0.dev172315966
 Summary: Deep learning software to decode EEG, ECG or MEG signals
 Author-email: Robin Tibor Schirrmeister <robintibor@gmail.com>, Bruno Aristimunha Pinto <b.aristimunha@gmail.com>, Alexandre Gramfort <agramfort@meta.com>
 Maintainer-email: Alexandre Gramfort <agramfort@meta.com>, Bruno Aristimunha Pinto <b.aristimunha@gmail.com>, Robin Tibor Schirrmeister <robintibor@gmail.com>

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/braindecode/models/bendr.py

@@ -95,6 +95,31 @@ class BENDR(EEGModuleMixin, nn.Module):
         **self-supervised pre-training** on large, unlabeled EEG datasets (like TUEG) followed
         by subsequent fine-tuning on the specific downstream classification task [bendr]_.
 
+    .. important::
+       **Pre-trained Weights Available**
+
+       This model has pre-trained weights available on the Hugging Face Hub.
+       You can load them using:
+
+       .. code-block:: python
+
+           from braindecode.models import BENDR
+
+           # Load pre-trained model from Hugging Face Hub
+           # you can specify `n_outputs` for your downstream task
+           model = BENDR.from_pretrained("braindecode/braindecode-bendr", n_outputs=2)
+
+       To push your own trained model to the Hub:
+
+       .. code-block:: python
+
+           # After training your model
+           model.push_to_hub(
+               repo_id="username/my-bendr-model", commit_message="Upload trained BENDR model"
+           )
+
+       Requires installing ``braindecode[hug]`` for Hub integration.
+
     Parameters
     ----------
     encoder_h : int, default=512

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/braindecode/models/biot.py

@@ -41,6 +41,35 @@ class BIOT(EEGModuleMixin, nn.Module):
     linear layer that takes the output of the `BIOTEncoder` and outputs
     the classification probabilities.
 
+    .. important::
+       **Pre-trained Weights Available**
+
+       This model has pre-trained weights available on the Hugging Face Hub.
+       You can load them using:
+
+       .. code-block:: python
+
+           from braindecode.models import BIOT
+
+           # Load the original pre-trained model from Hugging Face Hub
+           # For 16-channel models:
+           model = BIOT.from_pretrained("braindecode/biot-pretrained-prest-16chs")
+
+           # For 18-channel models:
+           model = BIOT.from_pretrained("braindecode/biot-pretrained-shhs-prest-18chs")
+           model = BIOT.from_pretrained("braindecode/biot-pretrained-six-datasets-18chs")
+
+       To push your own trained model to the Hub:
+
+       .. code-block:: python
+
+           # After training your model
+           model.push_to_hub(
+               repo_id="username/my-biot-model", commit_message="Upload trained BIOT model"
+           )
+
+       Requires installing ``braindecode[hug]`` for Hub integration.
+
     .. versionadded:: 0.9
 
     Parameters

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/braindecode/models/labram.py

@@ -64,6 +64,30 @@ class Labram(EEGModuleMixin, nn.Module):
             - LayerNorm: Apply layer normalization to the data;
             - Linear: An head linear layer to transformer the data into classes.
 
+    .. important::
+       **Pre-trained Weights Available**
+
+       This model has pre-trained weights available on the Hugging Face Hub.
+       You can load them using:
+
+       .. code-block:: python
+
+           from braindecode.models import Labram
+
+           # Load pre-trained model from Hugging Face Hub
+           model = Labram.from_pretrained("braindecode/labram-pretrained")
+
+       To push your own trained model to the Hub:
+
+       .. code-block:: python
+
+           # After training your model
+           model.push_to_hub(
+               repo_id="username/my-labram-model", commit_message="Upload trained Labram model"
+           )
+
+       Requires installing ``braindecode[hug]`` for Hub integration.
+
     .. versionadded:: 0.9
 
 

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/braindecode/models/luna.py

@@ -44,6 +44,61 @@ class LUNA(EEGModuleMixin, nn.Module):
     3. Patch-wise Temporal Encoder (RoPE-based transformer)
     4. Decoder Heads (classification or reconstruction)
 
+    .. important::
+       **Pre-trained Weights Available**
+
+       This model has pre-trained weights available on the Hugging Face Hub
+       at `thorir/LUNA <https://huggingface.co/thorir/LUNA>`_.
+
+       Available model variants:
+
+       - **LUNA_base.safetensors** - Base model (embed_dim=64, num_queries=4, depth=8)
+       - **LUNA_large.safetensors** - Large model (embed_dim=96, num_queries=6, depth=10)
+       - **LUNA_huge.safetensors** - Huge model (embed_dim=128, num_queries=8, depth=24)
+
+       Example loading for fine-tuning:
+
+       .. code-block:: python
+
+           from huggingface_hub import hf_hub_download
+           from safetensors.torch import load_file
+           from braindecode.models import LUNA
+
+           # Download pre-trained weights
+           model_path = hf_hub_download(
+               repo_id="thorir/LUNA",
+               filename="LUNA_base.safetensors",
+           )
+
+           # Create model for classification (fine-tuning)
+           model = LUNA(
+               n_outputs=2,  # Number of classes for your task
+               n_chans=22,
+               n_times=1000,
+               embed_dim=64,
+               num_queries=4,
+               depth=8,
+           )
+
+           # Load pre-trained encoder weights
+           state_dict = load_file(model_path)
+           # Apply key mapping for pretrained weights
+           mapping = model.mapping.copy()
+           mapping["cross_attn.temparature"] = "cross_attn.temperature"
+           mapped_state_dict = {mapping.get(k, k): v for k, v in state_dict.items()}
+           model.load_state_dict(mapped_state_dict, strict=False)
+
+       To push your own trained model to the Hub:
+
+       .. code-block:: python
+
+           # After training your model
+           model.push_to_hub(
+               repo_id="username/my-luna-model", commit_message="Upload trained LUNA model"
+           )
+
+       Requires installing ``braindecode[hug]`` for Hub integration.
+
     Parameters
     ----------
     patch_size : int

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/braindecode/models/reve.py

@@ -137,6 +137,30 @@ class REVE(EEGModuleMixin, nn.Module):
     - ``brain-bzh/reve-base``: 72M parameters, 512 embedding dim, 22 layers (~260 A100 GPU hours)
     - ``brain-bzh/reve-large``: ~400M parameters, 1250 embedding dim
 
+    .. important::
+       **Pre-trained Weights Available**
+
+       This model has pre-trained weights available on the Hugging Face Hub.
+       You can load them using:
+
+       .. code-block:: python
+
+           from braindecode.models import REVE
+
+           # Load pre-trained model from Hugging Face Hub
+           model = REVE.from_pretrained("brain-bzh/reve-base")
+
+       To push your own trained model to the Hub:
+
+       .. code-block:: python
+
+           # After training your model
+           model.push_to_hub(
+               repo_id="username/my-reve-model", commit_message="Upload trained REVE model"
+           )
+
+       Requires installing ``braindecode[hug]`` for Hub integration.
+
     .. rubric:: Usage
 
     .. code-block:: python

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/braindecode/models/signal_jepa.py

@@ -569,6 +569,33 @@ class SignalJEPA_PreLocal(_BaseSignalJEPA):
 
     .. versionadded:: 0.9
 
+    .. important::
+       **Pre-trained Weights Available**
+
+       This model has pre-trained weights available on the Hugging Face Hub.
+       You can load them using:
+
+       .. code-block:: python
+
+           from braindecode.models import SignalJEPA_PreLocal
+
+           # Load pre-trained model from Hugging Face Hub
+           model = SignalJEPA_PreLocal.from_pretrained(
+               "braindecode/SignalJEPA-PreLocal-pretrained"
+           )
+
+       To push your own trained model to the Hub:
+
+       .. code-block:: python
+
+           # After training your model
+           model.push_to_hub(
+               repo_id="username/my-sjepa-model",
+               commit_message="Upload trained SignalJEPA model",
+           )
+
+       Requires installing ``braindecode[hug]`` for Hub integration.
+
     Parameters
     ----------
     n_spat_filters : int

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/braindecode/samplers/ssl.py

@@ -190,7 +190,7 @@ class DistributedRelativePositioningSampler(DistributedRecordingSampler):
         self.tau_max = np.inf if tau_max is None else tau_max
         self.same_rec_neg = same_rec_neg
 
-        self.n_examples = n_examples // self.info.shape[0] * self.n_recordings
+        self.n_examples = n_examples * self.n_recordings // self.info.shape[0]
         warnings.warn(
             f"Rank {dist.get_rank()} - Number of datasets: {self.n_recordings}"
         )

braindecode-1.3.0.dev172315966/braindecode/version.py

@@ -0,0 +1 @@
+__version__ = "1.3.0.dev172315966"

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966/braindecode.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: braindecode
-Version: 1.3.0.dev171854083
+Version: 1.3.0.dev172315966
 Summary: Deep learning software to decode EEG, ECG or MEG signals
 Author-email: Robin Tibor Schirrmeister <robintibor@gmail.com>, Bruno Aristimunha Pinto <b.aristimunha@gmail.com>, Alexandre Gramfort <agramfort@meta.com>
 Maintainer-email: Alexandre Gramfort <agramfort@meta.com>, Bruno Aristimunha Pinto <b.aristimunha@gmail.com>, Robin Tibor Schirrmeister <robintibor@gmail.com>

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/docs/api.rst

@@ -2,9 +2,9 @@
 
 .. _api_reference:
 
-=======================
-Braindece API Reference
-=======================
+=========================
+Braindecode API Reference
+=========================
 
 .. automodule:: braindecode
    :no-members:
@@ -34,8 +34,60 @@ All the models are implemented as subclasses of :py:class:`EEGModuleMixin`, whic
 base class for all EEG models in Braindecode. The :class:`EEGModuleMixin` class
 provides a common interface for all EEG models and derivate variables names if necessary.
 
-Also, all models inherit from :class:`PyTorchModelHubMixin`, which provides functionality to
-save and load models from the Hugging Face Hub, if the ``braindecode[hug]`` package is installed.
+.. important::
+   **Hugging Face Hub Integration**
+
+   All models in braindecode naturally possess the capability to push and pull from the
+   Hugging Face Hub through inheritance from :class:`PyTorchModelHubMixin`. This allows you to:
+
+   * **Load pre-trained models** from the Hub using ``Model.from_pretrained("repo_id")``
+   * **Share your trained models** with the community using ``model.push_to_hub("repo_id")``
+   * **Version control your models** with git-like versioning on the Hub
+
+   To enable this functionality, install braindecode with Hub support::
+
+       pip install braindecode[hug]
+
+   **Available pre-trained models:**
+
+   Some models have pre-trained weights available on the Hugging Face BrainDecode organization:
+
+   * :class:`BIOT` - Foundation model with pre-trained weights
+   * :class:`Labram` - Large Brain Model with pre-trained weights
+   * :class:`REVE` - EEG foundation model with pre-trained weights
+   * :class:`LUNA` - Universal EEG embedding model with pre-trained weights
+   * :class:`BENDR` - Foundation model with pre-trained weights
+   * :class:`SignalJEPA` - Self-supervised learning model with pre-trained weights
+
+   **Example - Loading a pre-trained model:**
+
+   .. code-block:: python
+
+       from braindecode.models import BIOT
+
+       # Load pre-trained BIOT model from Hugging Face Hub
+       model = BIOT.from_pretrained("braindecode/biot-pretrained-prest-16chs")
+
+       # Use for your EEG classification task
+       # ... your code here ...
+
+   **Example - Pushing your trained model:**
+
+   .. code-block:: python
+
+       from braindecode.models import EEGNeX
+
+       # Train your model
+       model = EEGNeX(n_chans=22, n_outputs=4, n_times=1000)
+       # ... training code ...
+
+       # Push to the Hub (requires huggingface-cli login)
+       model.push_to_hub(
+           repo_id="username/my-eegnex-model", commit_message="Initial model upload"
+       )
+
+   For more details, see the :class:`EEGModuleMixin` documentation and the
+   :ref:`hub-integration` tutorial.
 
 :note: Auto-generated Pydantic configs are available when the optional
        ``braindecode[pydantic]`` extra (which installs ``pydantic`` and ``numpydantic``)

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/docs/conf.py

@@ -293,7 +293,7 @@ sphinx_gallery_conf = {
 #
 
 html_theme = "pydata_sphinx_theme"
-switcher_version_match = "dev" if release.endswith("dev0") else version
+switcher_version_match = "dev" if (release.endswith("dev0")) else version
 # Theme options are theme-specific and customize the look and feel of a theme
 # further.  For a list of options available for each theme, see the
 # documentation.

{braindecode-1.3.0.dev171854083 → braindecode-1.3.0.dev172315966}/docs/whats_new.rst

@@ -22,6 +22,7 @@ Current 1.3 (dev)
 
 Enhancements
 ~~~~~~~~~~~~
+- Adding Hugging Face Hub integration documentation with comprehensive examples for loading pre-trained models (:class:`braindecode.models.BIOT`, :class:`braindecode.models.Labram`, :class:`braindecode.models.REVE`, :class:`braindecode.models.LUNA`, :class:`braindecode.models.BENDR`, :class:`braindecode.models.SignalJEPA`) and pushing models to the Hub (:gh:`879` by `Bruno Aristimunha`_)
 - Adding :class:`braindecode.models.REVE` with comprehensive documentation covering architecture details, 4D positional encoding, and pretraining scale (:gh:`866` by `Jonathan Lys`_ )
 - Adding augmentation :class:`braindecode.augmentation.AmplitudeScale` (:gh:`848` by `Bruna Lopes`_)
 - Adding augmentation :class:`braindecode.augmentation.ChannelsReref` (:gh:`848` by `Bruna Lopes`_)

braindecode-1.3.0.dev171854083/braindecode/version.py

@@ -1 +0,0 @@
-__version__ = "1.3.0.dev171854083"