feat(policies): add Nvidia Gr00t N1.5 model (#2292)

* feat(policies): add Nvidia Gr00t N1.5 model

Co-authored-by: lbenhorin <lbenhorin@nvidia.com>
Co-authored-by: Aravindh <aravindhs@nvidia.com>
Co-authored-by: nv-sachdevkartik <ksachdev@nvidia.com>
Co-authored-by: youliangt <youliangt@nvidia.com>
Co-authored-by: Michel Aractingi <michel.aractingi@huggingface.co>
Co-authored-by: Pepijn <138571049+pkooij@users.noreply.github.com>
Co-authored-by: Jade Choghari <chogharijade@gmail.com>

* fix(docs): add groot to index

Co-authored-by: sachdevkartik <sachdev.kartik25@gmail.com>

---------

Co-authored-by: lbenhorin <lbenhorin@nvidia.com>
Co-authored-by: Aravindh <aravindhs@nvidia.com>
Co-authored-by: nv-sachdevkartik <ksachdev@nvidia.com>
Co-authored-by: youliangt <youliangt@nvidia.com>
Co-authored-by: Michel Aractingi <michel.aractingi@huggingface.co>
Co-authored-by: Pepijn <138571049+pkooij@users.noreply.github.com>
Co-authored-by: Jade Choghari <chogharijade@gmail.com>
Co-authored-by: sachdevkartik <sachdev.kartik25@gmail.com>

src/lerobot/policies/groot/action_head/action_encoder.py

@@ -0,0 +1,54 @@
+# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import torch
+import torch.nn as nn
+
+
+def swish(x):
+    return x * torch.sigmoid(x)
+
+
+class SinusoidalPositionalEncoding(nn.Module):
+    """
+    Produces a sinusoidal encoding of shape (B, T, w)
+    given timesteps of shape (B, T).
+    """
+
+    def __init__(self, embedding_dim):
+        super().__init__()
+        self.embedding_dim = embedding_dim
+
+    def forward(self, timesteps):
+        # timesteps: shape (B, T)
+        # We'll compute sin/cos frequencies across dim T
+        timesteps = timesteps.float()  # ensure float
+
+        b, t = timesteps.shape
+        device = timesteps.device
+
+        half_dim = self.embedding_dim // 2
+        # typical log space frequencies for sinusoidal encoding
+        exponent = -torch.arange(half_dim, dtype=torch.float, device=device) * (
+            torch.log(torch.tensor(10000.0)) / half_dim
+        )
+        # Expand timesteps to (B, T, 1) then multiply
+        freqs = timesteps.unsqueeze(-1) * exponent.exp()  # (B, T, half_dim)
+
+        sin = torch.sin(freqs)
+        cos = torch.cos(freqs)
+        enc = torch.cat([sin, cos], dim=-1)  # (B, T, w)
+
+        return enc