braindecode 1.7.0.dev179223210__tar.gz → 1.7.0.dev182244820__tar.gz

{braindecode-1.7.0.dev179223210/braindecode.egg-info → braindecode-1.7.0.dev182244820}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: braindecode
-Version: 1.7.0.dev179223210
+Version: 1.7.0.dev182244820
 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-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/braindecode/classifier.py

@@ -3,6 +3,7 @@
 #          Lukas Gemein <l.gemein@gmail.com>
 #          Bruno Aristimunha <b.aristimunha@gmail.com>
 #          Pierre Guetschel <pierre.guetschel@gmail.com>
+#          Sarthak Tayal <sarthaktayal2@gmail.com>
 #
 # License: BSD (3-clause)
 
@@ -195,9 +196,7 @@ class EEGClassifier(_EEGNeuralNet, NeuralNetClassifier):
         return self.predict_proba(X).argmax(1)
 
     def predict_trials(self, X, return_targets=True):
-        """Create trialwise predictions and optionally also return trialwise.
-
-        labels from cropped dataset.
+        """Create trialwise predictions from a cropped dataset.
 
         Parameters
         ----------
@@ -212,10 +211,11 @@ class EEGClassifier(_EEGNeuralNet, NeuralNetClassifier):
             3-dimensional array (n_trials x n_classes x n_predictions), where
             the number of predictions depend on the chosen window size and the
             receptive field of the network.
-        trial_labels : np.ndarray
-            2-dimensional array (n_trials x n_targets) where the number of
-            targets depends on the decoding paradigm and can be either a single
-            value, multiple values, or a sequence.
+        trial_targets : np.ndarray
+            Ground-truth targets from the dataset in a 2-dimensional array
+            (n_trials x n_targets). Only returned when ``return_targets=True``.
+            The number of targets depends on the decoding paradigm and can be
+            either a single value, multiple values, or a sequence.
         """
         if not self.cropped:
             warnings.warn(

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/braindecode/functional/functions.py

@@ -1,5 +1,6 @@
 # Authors: Robin Schirrmeister <robintibor@gmail.com>
 #          Bruno Aristimunha <b.aristimunha@gmail.com>
+#          Sarthak Tayal <sarthaktayal2@gmail.com>
 #
 # License: BSD (3-clause)
 
@@ -109,7 +110,7 @@ def _get_gaussian_kernel1d(kernel_size: int, sigma: float) -> torch.Tensor:
     return kernel1d
 
 
-def hilbert_freq(x, forward_fourier=True):
+def hilbert_freq(x: torch.Tensor, forward_fourier: bool = True) -> torch.Tensor:
     r"""
     Compute the Hilbert transform using PyTorch, separating the real and
     imaginary parts.
@@ -178,7 +179,9 @@ def hilbert_freq(x, forward_fourier=True):
     return x
 
 
-def plv_time(x, forward_fourier=True, epsilon: float = 1e-6):
+def plv_time(
+    x: torch.Tensor, forward_fourier: bool = True, epsilon: float = 1e-6
+) -> torch.Tensor:
     """Compute the Phase Locking Value (PLV) metric in the time domain.
 
     The Phase Locking Value (PLV) is a measure of the synchronization between

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/braindecode/models/eegminer.py

@@ -5,10 +5,9 @@
 * https://www.ipo.gov.uk/p-ipsum/Case/ApplicationNumber/GB2113420.0
 """
 
-from functools import partial
+# Authors: Sarthak Tayal <sarthaktayal2@gmail.com>
 
 import torch
-from einops.layers.torch import Rearrange
 from torch import nn
 
 import braindecode.functional as F
@@ -19,6 +18,36 @@ from braindecode.modules import GeneralizedGaussianFilter
 _eeg_miner_methods = ["mag", "corr", "plv"]
 
 
+class _EEGMinerFeatures(nn.Module):
+    def __init__(self, method, n_chans, n_filters, n_times):
+        super().__init__()
+        self.method = method
+        self.n_chans = n_chans
+        self.n_filters = n_filters
+        self.n_times = n_times
+        triu = torch.triu_indices(n_chans, n_chans, 1)
+        self.register_buffer("triu_row", triu[0], persistent=False)
+        self.register_buffer("triu_col", triu[1], persistent=False)
+
+    def forward(self, x):
+        if self.method == "mag":
+            return torch.sqrt((x * x).mean(dim=-1))
+        if self.method == "corr":
+            x = x.reshape(
+                x.shape[0], self.n_chans, self.n_filters, self.n_times
+            ).transpose(-3, -2)
+            x = (x - x.mean(dim=-1, keepdim=True)) / torch.sqrt(
+                x.var(dim=-1, keepdim=True) + 1e-6
+            )
+            x = torch.matmul(x, x.transpose(-2, -1)) / x.shape[-1]
+            x = x.permute(0, 2, 3, 1).abs()
+        else:
+            x = x.transpose(-4, -3)
+            x = F.plv_time(x, forward_fourier=False)
+            x = x.permute(0, 2, 3, 1)
+        return x[:, self.triu_row, self.triu_col, :]
+
+
 class EEGMiner(EEGModuleMixin, nn.Module):
     r"""EEGMiner from Ludwig et al (2024) [eegminer]_.
 
@@ -168,24 +197,18 @@ class EEGMiner(EEGModuleMixin, nn.Module):
         )
 
         # Forward method
+        self.ensure_dim = nn.Identity()
         if self.method == "mag":
-            self.method_forward = self._apply_mag_forward
             self.n_features = self.n_chans * self.n_filters
-            self.ensure_dim = nn.Identity()
         elif self.method == "corr":
-            self.method_forward = partial(
-                self._apply_corr_forward,
-                n_chans=self.n_chans,
-                n_filters=self.n_filters,
-                n_times=self.n_times,
-            )
             self.n_features = self.n_filters * self.n_chans * (self.n_chans - 1) // 2
-            self.ensure_dim = nn.Identity()
         elif self.method == "plv":
-            self.method_forward = partial(self._apply_plv, n_chans=self.n_chans)
-            self.ensure_dim = Rearrange("... d -> ... 1 d")
+            self.ensure_dim = nn.Unflatten(-1, (1, self.n_times))
             self.n_features = (self.n_filters * self.n_chans * (self.n_chans - 1)) // 2
 
+        self.feature_layer = _EEGMinerFeatures(
+            self.method, self.n_chans, self.n_filters, self.n_times
+        )
         self.flatten_layer = nn.Flatten()
         # Classifier
         self.batch_layer = nn.BatchNorm1d(self.n_features, affine=False)
@@ -195,64 +218,17 @@ class EEGMiner(EEGModuleMixin, nn.Module):
     @_disable_batch_norm_training_if_batch_size_one
     def forward(self, x):
         """x: (batch, electrodes, time)"""
-        batch = x.shape[0]
         x = self.ensure_dim(x)
         # Apply Gaussian filters in frequency domain
         # x -> (batch, electrodes * filters, time)
         x = self.filter(x)
 
-        x = self.method_forward(x=x, batch=batch)
+        x = self.feature_layer(x)
         # Classifier
         # Note that the order of dimensions before flattening the feature vector is important
         # for attributing feature weights during interpretation.
-        x = x.reshape(batch, self.n_features)
+        x = x.reshape(x.shape[0], self.n_features)
         x = self.batch_layer(x)
         x = self.final_layer(x)
 
         return x
-
-    @staticmethod
-    def _apply_mag_forward(x, batch=None):
-        # Signal magnitude
-        x = x * x
-        x = x.mean(dim=-1)
-        x = torch.sqrt(x)
-        return x
-
-    @staticmethod
-    def _apply_corr_forward(
-        x, batch, n_chans, n_filters, n_times, epilson: float = 1e-6
-    ):
-        x = x.reshape(batch, n_chans, n_filters, n_times).transpose(-3, -2)
-        x = (x - x.mean(dim=-1, keepdim=True)) / torch.sqrt(
-            x.var(dim=-1, keepdim=True) + epilson
-        )
-        x = torch.matmul(x, x.transpose(-2, -1)) / x.shape[-1]
-        # Original tensor shape: [batch, n_filters, chans, chans]
-        x = x.permute(0, 2, 3, 1)
-        # New tensor shape: [batch, chans, chans, n_filters]
-        # move filter channels to the end
-        x = x.abs()
-
-        # Get upper triu of symmetric connectivity matrix
-        triu = torch.triu_indices(n_chans, n_chans, 1)
-        x = x[:, triu[0], triu[1], :]
-
-        return x
-
-    @staticmethod
-    def _apply_plv(x, n_chans, batch=None):
-        # Compute PLV connectivity
-        # x -> (batch, electrodes, electrodes, filters)
-        x = x.transpose(-4, -3)  # swap electrodes and filters
-        # adjusting to compute the plv
-        x = F.plv_time(x, forward_fourier=False)
-        # batch, number of filters, connectivity matrix
-        # [batch, n_filters, chans, chans]
-        x = x.permute(0, 2, 3, 1)
-        # [batch, chans, chans, n_filters]
-
-        # Get upper triu of symmetric connectivity matrix
-        triu = torch.triu_indices(n_chans, n_chans, 1)
-        x = x[:, triu[0], triu[1], :]
-        return x

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/braindecode/models/reve.py

@@ -8,13 +8,13 @@ import json
 import logging
 import math
 import os
-from pathlib import Path
 from typing import Optional, Union
 
-import requests
+import pooch
 import torch
 import torch.nn.functional as F
 from einops import rearrange
+from mne.datasets.utils import _get_path
 from torch import nn
 
 # Safe import for older PyTorch versions (Support for Intel-based Macs)
@@ -785,10 +785,13 @@ class FourierEmb4D(nn.Module):
 class RevePositionBank(torch.nn.Module):
     """Position bank for REVE model that maps standard EEG channel names to 3D coordinates.
 
-    The position bank is cached locally in the library root to avoid repeated downloads.
-
     The coordinates come from the 92 datasets used during REVE pretraining.
 
+    The bank is cached in the MNE data directory after the first download, so it
+    works offline afterwards. On restricted nodes (e.g. HPC compute nodes without
+    a proxy), drop a prefetched ``reve_positions.json`` in that directory and no
+    download is attempted.
+
     Parameters
     ----------
     url : str, optional
@@ -796,7 +799,9 @@ class RevePositionBank(torch.nn.Module):
     timeout : int, optional
         Timeout in seconds for the HTTP request. Default is 5 seconds.
     cache_dir : str, optional
-        Directory to cache the position bank. Default is the models folder within the library.
+        Directory to cache the position bank. If ``None``, resolved like other
+        braindecode datasets via the ``REVE_POSITIONS_PATH`` config key, falling
+        back to the MNE data directory (usually ``~/mne_data``).
     """
 
     def __init__(
@@ -807,38 +812,32 @@ class RevePositionBank(torch.nn.Module):
     ):
         super().__init__()
 
-        if cache_dir is None:
-            # Use the model root directory
-            cache_dir = str(Path(__file__).parent)
-
-        cache_file = os.path.join(cache_dir, ".cache", "reve_positions.json")
-        os.makedirs(os.path.dirname(cache_file), exist_ok=True)
-
-        config = None
+        cache_dir = _get_path(cache_dir, "REVE_POSITIONS_PATH", "REVE positions")
+        cache_file = os.path.join(cache_dir, "reve_positions.json")
 
-        # Try to load from cache first
-        if os.path.exists(cache_file):
+        # Offline path: reuse the cached/prefetched file. Download only on a miss
+        # or if the cached copy is unreadable (e.g. an interrupted download).
+        try:
+            with open(cache_file) as f:
+                config = json.load(f)
+        except (OSError, json.JSONDecodeError):
             try:
-                with open(cache_file, "r") as f:
+                if os.path.exists(cache_file):
+                    os.remove(cache_file)  # corrupt/partial; force a fresh copy
+                # pooch ships with mne and writes straight into the cache dir.
+                pooch.retrieve(
+                    url,
+                    known_hash=None,
+                    fname="reve_positions.json",
+                    path=cache_dir,
+                    downloader=pooch.HTTPDownloader(timeout=timeout),
+                )
+                with open(cache_file) as f:
                     config = json.load(f)
-                logger.info(f"Loaded position bank from cache: {cache_file}")
-            except (json.JSONDecodeError, IOError) as e:
-                logger.warning(f"Failed to load cache, downloading: {e}")
-
-        # Download if cache miss or failed to load
-        if config is None:
-            try:
-                response = requests.get(url, timeout=timeout)
-                response.raise_for_status()
-                config = json.loads(response.text)
-
-                # Save to cache
-                with open(cache_file, "w") as f:
-                    json.dump(config, f)
-                logger.info(f"Downloaded and cached position bank to: {cache_file}")
-            except (requests.RequestException, json.JSONDecodeError) as e:
+            except Exception as e:
                 raise RuntimeError(
-                    f"Failed to download or parse the position bank from {url}: {e}"
+                    f"Failed to download the position bank from {url}. On an "
+                    f"offline node, prefetch it to {cache_file} first: {e}"
                 ) from e
 
         try:

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/braindecode/modules/filter.py

@@ -1,3 +1,7 @@
+# Authors: Sarthak Tayal <sarthaktayal2@gmail.com>
+#
+# License: BSD-3
+
 from __future__ import annotations
 
 from typing import Optional
@@ -637,20 +641,22 @@ class GeneralizedGaussianFilter(nn.Module):
             The constructed filters with shape `(out_channels, freq_bins, 2)`.
 
         """
-        # Clamp parameters
-        self.f_mean.data = torch.clamp(
-            self.f_mean.data,
-            min=self.clamp_f_mean[0] / (self.sample_rate / 2),
-            max=self.clamp_f_mean[1] / (self.sample_rate / 2),
-        )
-        self.bandwidth.data = torch.clamp(
-            self.bandwidth.data, min=1.0 / (self.sample_rate / 2), max=1.0
+        min_f_mean = self.clamp_f_mean[0] / (self.sample_rate / 2)
+        max_f_mean = self.clamp_f_mean[1] / (self.sample_rate / 2)
+        f_mean = torch.clamp(self.f_mean, min=min_f_mean, max=max_f_mean)
+        bandwidth = torch.clamp(
+            self.bandwidth, min=1.0 / (self.sample_rate / 2), max=1.0
         )
-        self.shape.data = torch.clamp(self.shape.data, min=2.0, max=3.0)
+        shape = torch.clamp(self.shape, min=2.0, max=3.0)
+
+        if not torch.jit.is_scripting():
+            self.f_mean.data = f_mean
+            self.bandwidth.data = bandwidth
+            self.shape.data = shape
 
         # Create magnitude response with gain=1 -> (channels, freqs)
         mag_response = self.exponential_power(
-            self.n_range, self.f_mean, self.bandwidth, self.shape * 8 - 14
+            self.n_range, f_mean, bandwidth, shape * 8 - 14
         )
         mag_response = mag_response / mag_response.max(dim=-1, keepdim=True)[0]
 

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/braindecode/regressor.py

@@ -3,6 +3,7 @@
 #          Lukas Gemein <l.gemein@gmail.com>
 #          Bruno Aristimunha <b.aristimunha@gmail.com>
 #          Pierre Guetschel <pierre.guetschel@gmail.com>
+#          Sarthak Tayal <sarthaktayal2@gmail.com>
 #
 # License: BSD (3-clause)
 
@@ -139,9 +140,7 @@ class EEGRegressor(_EEGNeuralNet, NeuralNetRegressor):
             return y_pred
 
     def predict_trials(self, X, return_targets=True):
-        """Create trialwise predictions and optionally also return trialwise.
-
-        labels from cropped dataset.
+        """Create trialwise predictions from a cropped dataset.
 
         Parameters
         ----------
@@ -156,10 +155,11 @@ class EEGRegressor(_EEGNeuralNet, NeuralNetRegressor):
             3-dimensional array (n_trials x n_classes x n_predictions), where
             the number of predictions depend on the chosen window size and the
             receptive field of the network.
-        trial_labels : np.ndarray
-            2-dimensional array (n_trials x n_targets) where the number of
-            targets depends on the decoding paradigm and can be either a single
-            value, multiple values, or a sequence.
+        trial_targets : np.ndarray
+            Ground-truth targets from the dataset in a 2-dimensional array
+            (n_trials x n_targets). Only returned when ``return_targets=True``.
+            The number of targets depends on the decoding paradigm and can be
+            either a single value, multiple values, or a sequence.
         """
         if not self.cropped:
             warnings.warn(

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/braindecode/training/scoring.py

@@ -3,6 +3,7 @@
 #          Alexandre Gramfort <alexandre.gramfort@inria.fr>
 #          Lukas Gemein <l.gemein@gmail.com>
 #          Mohammed Fattouh <mo.fattouh@gmail.com>
+#          Sarthak Tayal <sarthaktayal2@gmail.com>
 #
 # License: BSD-3
 
@@ -401,7 +402,7 @@ class PostEpochTrainScoring(EpochScoring):
 
 def predict_trials(module, dataset, return_targets=True, batch_size=1, num_workers=0):
     """Create trialwise predictions and optionally also return trialwise
-    labels from cropped dataset given module.
+    targets from a cropped dataset given a module.
 
     Parameters
     ----------
@@ -422,10 +423,11 @@ def predict_trials(module, dataset, return_targets=True, batch_size=1, num_worke
             3-dimensional array (n_trials x n_classes x n_predictions), where
             the number of predictions depend on the chosen window size and the
             receptive field of the network.
-        trial_labels: np.ndarray
-            2-dimensional array (n_trials x n_targets) where the number of
-            targets depends on the decoding paradigm and can be either a single
-            value, multiple values, or a sequence.
+        trial_targets: np.ndarray
+            Ground-truth targets from the dataset in a 2-dimensional array
+            (n_trials x n_targets). Only returned when ``return_targets=True``.
+            The number of targets depends on the decoding paradigm and can be
+            either a single value, multiple values, or a sequence.
     """
     # Ensure the model is in evaluation mode
     module.eval()

braindecode-1.7.0.dev182244820/braindecode/version.py

@@ -0,0 +1 @@
+__version__ = "1.7.0.dev182244820"

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820/braindecode.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: braindecode
-Version: 1.7.0.dev179223210
+Version: 1.7.0.dev182244820
 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-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/docs/index.rst

@@ -161,7 +161,7 @@
       <div class="container bd-facts-inner">
         <span class="bd-facts-label">By the numbers</span>
         <div class="bd-facts-row">
-          <a href="models/models.html"><b>65+</b> models</a>
+          <a href="models/models_table.html"><b>65+</b> models</a>
           <a href="api.html#augmentation"><b>20+</b> augmentations</a>
           <a href="https://moabb.neurotechx.com/" target="_blank" rel="noopener"><b>150+</b> datasets <span style="color:var(--bd-muted)">via MOABB</span></a>
           <a href="https://eegdash.org/" target="_blank" rel="noopener"><b>700+</b> datasets <span style="color:var(--bd-muted)">via EEGDash</span></a>
@@ -224,7 +224,7 @@
           <p class="section-sub">Built to <strong>plug into the EEG ecosystem you already use</strong>: every <a href="https://moabb.neurotechx.com/" target="_blank" rel="noopener">MOABB</a> dataset, every <a href="https://mne.tools/" target="_blank" rel="noopener">MNE-Python</a> preprocessing function, every <a href="https://scikit-learn.org/" target="_blank" rel="noopener">scikit-learn</a> training loop, plus 700+ BIDS datasets via <a href="https://eegdash.org/" target="_blank" rel="noopener">EEGDash</a>. One library, no lock-in.</p>
         </div>
         <div class="feature-grid">
-          <a href="models/models.html" class="feature-card">
+          <a href="models/models_table.html" class="feature-card">
             <span class="feature-icon"><svg width="26" height="26" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round"><path d="M2 12c2-4 4-4 6 0s4 4 6 0 4-4 6 0 4 4 2 0"/></svg></span>
             <h3>Decode raw electrophysiology</h3>
             <p>End-to-end models go straight from raw EEG/ECoG/MEG to predictions. No hand-crafted features required.</p>
@@ -248,7 +248,7 @@
             <p>Fully compatible with every <a class="feature-inline-link" href="https://mne.tools/" target="_blank" rel="noopener">MNE</a> preprocessing function and with <a class="feature-inline-link" href="https://eegprep.org/" target="_blank" rel="noopener">EEGPrep</a>, plus exponential standardization and 20+ EEG augmentations.</p>
             <a href="api.html#augmentation" class="feature-link">Augmentation <svg width="12" height="12" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.7" stroke-linecap="round" stroke-linejoin="round"><path d="M5 12h14M13 5l7 7-7 7"/></svg></a>
           </div>
-          <a href="models/models.html" class="feature-card">
+          <a href="models/models_table.html" class="feature-card">
             <span class="feature-icon"><svg width="26" height="26" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round"><circle cx="6" cy="6" r="2"/><circle cx="6" cy="18" r="2"/><circle cx="18" cy="12" r="2"/><path d="M8 6h2a3 3 0 0 1 3 3v0a3 3 0 0 0 3 3M8 18h2a3 3 0 0 0 3-3v0a3 3 0 0 1 3-3"/></svg></span>
             <h3>Curated model zoo</h3>
             <p>EEGNeX, ConvNets, ATCNet, EEGConformer, foundation models. 60+ architectures reproduced from the original papers.</p>
@@ -359,7 +359,7 @@
         <div class="zoo-grid"><!-- populated by landing.js --></div>
         <div class="zoo-foot">
           <span class="zoo-foot-count">Loading…</span>
-          <a href="models/models.html">Browse the full zoo on the Models page <svg width="12" height="12" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.7" stroke-linecap="round" stroke-linejoin="round"><path d="M5 12h14M13 5l7 7-7 7"/></svg></a>
+          <a href="models/models_table.html">Browse the full zoo on the Models table <svg width="12" height="12" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.7" stroke-linecap="round" stroke-linejoin="round"><path d="M5 12h14M13 5l7 7-7 7"/></svg></a>
         </div>
       </div>
     </section>
@@ -521,7 +521,7 @@
               <span class="pipeline-col-name">Modeling</span>
             </div>
             <div class="pipeline-col-cards">
-              <a class="pipeline-card" href="models/models.html">
+              <a class="pipeline-card" href="models/models_table.html">
                 <span class="pipeline-card-icon"><svg width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.7" stroke-linecap="round" stroke-linejoin="round"><circle cx="6" cy="6" r="2"/><circle cx="6" cy="18" r="2"/><circle cx="18" cy="12" r="2"/><path d="M8 6h2a3 3 0 0 1 3 3v0a3 3 0 0 0 3 3M8 18h2a3 3 0 0 0 3-3v0a3 3 0 0 1 3-3"/></svg></span>
                 <span class="pipeline-card-text">
                   <span class="pipeline-card-name">braindecode.models</span>
@@ -679,7 +679,7 @@
             <ul>
               <li><a href="install/install.html">Install</a></li>
               <li><a href="auto_examples/index.html">Tutorials</a></li>
-              <li><a href="models/models.html">Models</a></li>
+              <li><a href="models/models_table.html">Models</a></li>
               <li><a href="api.html">API reference</a></li>
             </ul>
           </div>

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/docs/models/models.rst

@@ -68,12 +68,12 @@ With this, in this model's sub-pages, we provide:
     - 4. :doc:`A visualization of the common important information from the models
          <models_visualization>`.
 
-.. button-ref:: models_categorization
+.. button-ref:: models_table
     :ref-type: doc
     :color: primary
     :expand:
 
-    Next: Models categorization →
+    Browse the models table →
 
 .. rubric:: References
 

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/docs/models/models_table.rst

@@ -191,6 +191,15 @@ Braindecode.
       // Initial color pass
       applyTagPalette(document);
 
+      function applyInitialModelFilter() {
+        var model = new URLSearchParams(window.location.search).get('model');
+        if (!model) return;
+        table.search(model).draw();
+        $(table.table().container()).find('div.dataTables_filter input').val(model);
+      }
+
+      applyInitialModelFilter();
+
       /* --- Bonus UX: click a column header to open panes focused there ----- */
       $('.sortable thead th').each(function (i) {
         if (!FILTER_COLS.includes(i)) return;

{braindecode-1.7.0.dev179223210 → braindecode-1.7.0.dev182244820}/docs/whats_new.rst

@@ -49,7 +49,22 @@ Requirements
 Bug fixes
 ==========
 
-- None yet
+- Clarify that :func:`braindecode.training.scoring.predict_trials`,
+  :meth:`braindecode.EEGClassifier.predict_trials`, and
+  :meth:`braindecode.EEGRegressor.predict_trials` return ground-truth dataset
+  targets alongside predictions. By `Sarthak Tayal`_.
+
+- Make :class:`braindecode.models.EEGMiner` compatible with TorchScript across
+  magnitude, correlation, and phase-locking-value feature modes by replacing
+  runtime callable dispatch with a scriptable feature module. (:gh:`1101` by
+  `Sarthak Tayal`_)
+
+- Make the :class:`braindecode.models.REVE` position bank robust on offline /
+  limited-network nodes: it is now cached in the writable MNE data directory
+  (resolved via the ``REVE_POSITIONS_PATH`` config key, defaulting under
+  ``~/mne_data``) instead of the package folder, so a prefetched
+  ``reve_positions.json`` there is used without any download (:gh:`1098` by
+  `Bruno Aristimunha`_)
 
 Code health
 ============

braindecode-1.7.0.dev179223210/braindecode/version.py

@@ -1 +0,0 @@
-__version__ = "1.7.0.dev179223210"