huggingface
diff --git a/‎docs/source/en/_toctree.yml‎
Lines changed: 2 additions & 0 deletions b/‎docs/source/en/_toctree.yml‎
Lines changed: 2 additions & 0 deletions
diff --git a/‎docs/source/en/model_doc/apertus.md‎
Lines changed: 100 additions & 0 deletions b/‎docs/source/en/model_doc/apertus.md‎
Lines changed: 100 additions & 0 deletions
diff --git a/‎src/transformers/activations.py‎
Lines changed: 96 additions & 0 deletions b/‎src/transformers/activations.py‎
Lines changed: 96 additions & 0 deletions
diff --git a/‎src/transformers/models/apertus/__init__.py‎
Lines changed: 32 additions & 0 deletions b/‎src/transformers/models/apertus/__init__.py‎
Lines changed: 32 additions & 0 deletions
@@ -373,6 +373,8 @@
     - sections:
       - local: model_doc/albert
         title: ALBERT
+      - local: model_doc/apertus
+        title: Apertus
       - local: model_doc/arcee
         title: Arcee
       - local: model_doc/bamba
 
@@ -0,0 +1,100 @@
+<!--Copyright 2025 The HuggingFace Team and the Swiss AI Initiative. All rights reserved.
+
+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.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+        <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
+    </div>
+</div>
+
+# Apertus
+
+[Apertus](https://www.swiss-ai.org) is a family of large language models from the Swiss AI Initiative.
+
+> [!TIP]
+> Coming soon
+
+The example below demonstrates how to generate text with [`Pipeline`] or the [`AutoModel`], and from the command line.
+
+<hfoptions id="usage">
+<hfoption id="Pipeline">
+
+```py
+import torch
+from transformers import pipeline
+
+pipeline = pipeline(
+    task="text-generation",
+    model="swiss-ai/Apertus-8B",
+    dtype=torch.bfloat16,
+    device=0
+)
+pipeline("Plants create energy through a process known as")
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```py
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+tokenizer = AutoTokenizer.from_pretrained(
+    "swiss-ai/Apertus-8B",
+)
+model = AutoModelForCausalLM.from_pretrained(
+    "swiss-ai/Apertus-8B",
+    dtype=torch.bfloat16,
+    device_map="auto",
+    attn_implementation="sdpa"
+)
+input_ids = tokenizer("Plants create energy through a process known as", return_tensors="pt").to("cuda")
+
+output = model.generate(**input_ids)
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+
+</hfoption>
+<hfoption id="transformers CLI">
+
+```bash
+echo -e "Plants create energy through a process known as" | transformers run --task text-generation --model swiss-ai/Apertus-8B --device 0
+```
+
+</hfoption>
+</hfoptions>
+
+## ApertusConfig
+
+[[autodoc]] ApertusConfig
+
+## ApertusModel
+
+[[autodoc]] ApertusModel
+    - forward
+
+## ApertusForCausalLM
+
+[[autodoc]] ApertusForCausalLM
+    - forward
+
+## ApertusForTokenClassification
+
+[[autodoc]] ApertusForTokenClassification
+    - forward
@@ -19,6 +19,7 @@
 from torch import Tensor, nn
 
 from .utils import logging
+from .utils.import_utils import is_torchdynamo_compiling
 
 
 logger = logging.get_logger(__name__)
@@ -185,6 +186,100 @@ def __getitem__(self, key):
         return cls(**kwargs)
 
 
+class XIELUActivation(nn.Module):
+    """
+    Applies the xIELU activation function introduced in https://arxiv.org/abs/2411.13010
+
+    If the user has installed the nickjbrowning/XIELU wheel, we import xIELU CUDA
+    Otherwise, we emit a single warning and use xIELU Python
+    """
+
+    def __init__(
+        self,
+        alpha_p_init=0.8,
+        alpha_n_init=0.8,
+        beta=0.5,
+        eps=-1e-6,
+        dtype=torch.bfloat16,
+        with_vector_loads=False,
+    ):
+        super().__init__()
+        self.alpha_p = nn.Parameter(torch.log(torch.exp(torch.tensor(alpha_p_init, dtype=dtype)) - 1).unsqueeze(0))
+        self.alpha_n = nn.Parameter(
+            torch.log(torch.exp(torch.tensor(alpha_n_init - beta, dtype=dtype)) - 1).unsqueeze(0)
+        )
+        self.register_buffer("beta", torch.tensor(beta, dtype=dtype))
+        self.register_buffer("eps", torch.tensor(eps, dtype=dtype))
+        self.with_vector_loads = with_vector_loads
+        # Temporary until xIELU CUDA fully implemented
+        self._beta_scalar = float(self.beta.detach().cpu().float().item())
+        self._eps_scalar = float(self.eps.detach().cpu().float().item())
+
+        self._xielu_cuda_obj = None
+        try:
+            import xielu.ops  # noqa: F401
+
+            self._xielu_cuda_obj = torch.classes.xielu.XIELU()
+            msg = "Using experimental xIELU CUDA."
+            try:
+                from torch._dynamo import allow_in_graph
+
+                self._xielu_cuda_fn = allow_in_graph(self._xielu_cuda)
+                msg += " Enabled torch._dynamo for xIELU CUDA."
+            except Exception as err:
+                msg += f" Could not enable torch._dynamo for xIELU ({err}) - this may result in slower performance."
+                self._xielu_cuda_fn = self._xielu_cuda
+            logger.warning_once(msg)
+        except Exception as err:
+            logger.warning_once(
+                "CUDA-fused xIELU not available (%s) – falling back to a Python version.\n"
+                "For CUDA xIELU (experimental), `pip install git+https://github.com/nickjbrowning/XIELU`",
+                str(err),
+            )
+
+    def _xielu_python(self, x: Tensor) -> Tensor:
+        alpha_p = nn.functional.softplus(self.alpha_p)
+        alpha_n = self.beta + nn.functional.softplus(self.alpha_n)
+        return torch.where(
+            x > 0,
+            alpha_p * x * x + self.beta * x,
+            (torch.expm1(torch.min(x, self.eps)) - x) * alpha_n + self.beta * x,
+        )
+
+    def _xielu_cuda(self, x: Tensor) -> Tensor:
+        """Firewall function to prevent torch.compile from seeing .item() calls"""
+        original_shape = x.shape
+        # CUDA kernel expects 3D tensors, reshape if needed
+        while x.dim() < 3:
+            x = x.unsqueeze(0)
+        if x.dim() > 3:
+            x = x.view(-1, 1, x.size(-1))
+        if original_shape != x.shape:
+            logger.warning_once(
+                "Warning: xIELU input tensor expects 3 dimensions but got (shape: %s). Reshaping to (shape: %s).",
+                original_shape,
+                x.shape,
+            )
+        result = self._xielu_cuda_obj.forward(
+            x,
+            self.alpha_p,
+            self.alpha_n,
+            # Temporary until xIELU CUDA fully implemented -> self.{beta,eps}.item()
+            self._beta_scalar,
+            self._eps_scalar,
+            self.with_vector_loads,
+        )
+        return result.view(original_shape)
+
+    def forward(self, input: Tensor) -> Tensor:
+        if self._xielu_cuda_obj is not None and input.is_cuda:
+            if not is_torchdynamo_compiling():
+                return self._xielu_cuda_fn(input)
+            else:
+                logger.warning_once("torch._dynamo is compiling, using Python version of xIELU.")
+        return self._xielu_python(input)
+
+
 ACT2CLS = {
     "gelu": GELUActivation,
     "gelu_10": (ClippedGELUActivation, {"min": -10, "max": 10}),
@@ -206,6 +301,7 @@ def __getitem__(self, key):
     "swish": nn.SiLU,
     "tanh": nn.Tanh,
     "prelu": nn.PReLU,
+    "xielu": XIELUActivation,
 }
 ACT2FN = ClassInstantier(ACT2CLS)
 
 
@@ -0,0 +1,32 @@
+# coding=utf-8
+# Copyright 2025 The HuggingFace Inc. team and the Swiss AI Initiative. All rights reserved.
+#
+# This code is based on HuggingFace's LLaMA implementation in this library.
+# It has been modified from its original forms to accommodate the architectural
+# differences made by the Swiss AI Initiative that trained the model.
+#
+# 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.
+from typing import TYPE_CHECKING
+
+from ...utils import _LazyModule
+from ...utils.import_utils import define_import_structure
+
+
+if TYPE_CHECKING:
+    from .configuration_apertus import *
+    from .modeling_apertus import *
+else:
+    import sys
+
+    _file = globals()["__file__"]
+    sys.modules[__name__] = _LazyModule(__name__, _file, define_import_structure(_file), module_spec=__spec__)