Add ST-EEGFormer largeV2 (ported from ST-EEGFormer-largeV2)

README.md

@@ -0,0 +1,49 @@
+---
+library_name: braindecode
+tags:
+- braindecode
+- STEEGFormer
+- eeg
+- foundation-model
+license: mit
+---
+
+# STEEGFormer (largeV2)
+
+ViT-MAE EEG foundation model — braindecode port of **ST-EEGFormer** (largeV2 variant).
+
+## Provenance
+
+- **Weights ported from:** [LiuyinYang1101/STEEGFormer](https://github.com/LiuyinYang1101/STEEGFormer),
+  release [`ST-EEGFormer-largeV2`](https://github.com/LiuyinYang1101/STEEGFormer/releases/tag/ST-EEGFormer-largeV2)
+  (asset `large_weights_only_210.pth`).
+- **Upstream license:** MIT. The braindecode wrapper code is BSD-3-Clause.
+- The pre-trained encoder is loaded faithfully (numerical equivalence verified:
+  pre-encoder bit-exact, post-encoder relative error ~4e-6). The MAE decoder is
+  dropped and the classification head is re-initialised.
+
+## Architecture
+
+| | embed_dim | depth | num_heads | patch_size | channel vocab |
+|---|---|---|---|---|---|
+| largeV2 | 1024 | 24 | 16 | 16 | 256 |
+
+This variant uses a **256-slot** channel vocabulary (further pre-trained on HBN for the EEG 2025 Foundation Challenge). Name-based mapping via `chs_info` works for standard electrodes; pass `chan_pos_idx` explicitly for the HBN montage / non-standard channels.
+
+## Usage
+
+```python
+from braindecode.models import STEEGFormer
+
+model = STEEGFormer.from_pretrained(
+    "braindecode/STEEGFormer-largeV2",
+    n_outputs=4, n_chans=22, n_times=1000, chs_info=chs_info,
+)
+# Encoder features:  out = model(x, return_features=True); out["features"]
+```
+
+## Citation
+
+Yang, L., Sun, Q., Li, A. & Van Hulle, M. M. (2026). *Are EEG foundation models
+worth it? Comparative evaluation with traditional decoders in diverse BCI tasks.*
+ICLR 2026. https://openreview.net/forum?id=5Xwm8e6vbh

config.json

@@ -0,0 +1,20 @@
+{
+  "n_outputs": 4,
+  "n_chans": 22,
+  "chs_info": null,
+  "n_times": 1000,
+  "input_window_seconds": null,
+  "sfreq": null,
+  "patch_size": 16,
+  "embed_dim": 1024,
+  "depth": 24,
+  "num_heads": 16,
+  "mlp_ratio": 4.0,
+  "drop_prob": 0.0,
+  "drop_path": 0.0,
+  "activation": "torch.nn.modules.activation.GELU",
+  "global_pool": "avg",
+  "n_chans_pos": 256,
+  "chan_pos_idx": null,
+  "braindecode_version": "1.6.1dev0"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0051f40a5b3ebfa6ca226d204220c04ccfeb12544b621907ba52791ddf6011ff
+size 1210423512

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82409f7cee1be6ff0be5498cf040d229dfb68b6479aace3e90c095644c84a5eb
+size 1210530499