Add FlexOlmo model (#40921)

* transformers add-new-model-like

* Add FlexOlmo implementation

* Update FlexOlmo docs

* Set default tokenization for flex olmo

* Update FlexOlmo tests

* Update attention comment

* Remove unneeded use of `sliding_window`

Author: 2015aroras

docs/source/en/_toctree.yml

@@ -485,6 +485,8 @@
         title: FLAN-UL2
       - local: model_doc/flaubert
         title: FlauBERT
+      - local: model_doc/flex_olmo
+        title: FlexOlmo
       - local: model_doc/fnet
         title: FNet
       - local: model_doc/fsmt

docs/source/en/model_doc/flex_olmo.md

@@ -0,0 +1,139 @@
+<!--Copyright 2025 the HuggingFace Team. 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.
+
+-->
+*This model was released on 2025-07-09 and added to Hugging Face Transformers on 2025-09-15.*
+<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">
+    </div>
+</div>
+
+# FlexOlmo
+
+[FlexOlmo](https://huggingface.co/papers/2507.07024) is a new class of language models (LMs) that supports (1) distributed training without data sharing, where different model parameters are independently trained on closed datasets, and (2) data-flexible inference, where these parameters along with their associated data can be flexibly included or excluded from model inferences with no further training. FlexOlmo employs a mixture-of-experts (MoE) architecture where each expert is trained independently on closed datasets and later integrated through a new domain-informed routing without any joint training. FlexOlmo is trained on FlexMix, a corpus we curate comprising publicly available datasets alongside seven domain-specific sets, representing realistic approximations of closed sets. 
+
+You can find all the original FlexOlmo checkpoints under the [FlexOlmo](https://huggingface.co/collections/allenai/flexolmo-68471177a386b6e20a54c55f) collection.
+
+> [!TIP]
+> Click on the FlexOlmo models in the right sidebar for more examples of how to apply FlexOlmo to different language tasks.
+
+The example below demonstrates how to generate text with [`Pipeline`], [`AutoModel`] and from the command line.
+
+<hfoptions id="usage">
+<hfoption id="Pipeline">
+
+```py
+import torch
+from transformers import pipeline
+
+pipe = pipeline(
+    task="text-generation",
+    model="allenai/FlexOlmo-7x7B-1T",
+    dtype=torch.bfloat16,
+    device=0,
+)
+    
+result = pipe("Plants create energy through a process known as")
+print(result)
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```py
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+tokenizer = AutoTokenizer.from_pretrained(
+    "allenai/FlexOlmo-7x7B-1T"
+)
+
+model = AutoModelForCausalLM.from_pretrained(
+    "allenai/FlexOlmo-7x7B-1T",
+    dtype=torch.bfloat16,
+    device_map="auto",
+    attn_implementation="sdpa"
+)
+input_ids = tokenizer("Plants create energy through a process known as", return_tensors="pt").to(model.device)
+
+output = model.generate(**input_ids, max_length=50, cache_implementation="static")
+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-cli run --task text-generation --model allenai/FlexOlmo-7x7B-1T --device 0
+```
+
+</hfoption>
+</hfoptions>
+
+Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the [Quantization](../quantization/overview) overview for more available quantization backends.
+
+The example below uses [torchao](../quantization/torchao) to only quantize the weights to 4-bits.
+```py
+
+#pip install torchao
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, TorchAoConfig
+
+torchao_config = TorchAoConfig(
+    "int4_weight_only",
+    group_size=128
+)
+
+tokenizer = AutoTokenizer.from_pretrained(
+    "allenai/FlexOlmo-7x7B-1T"
+)
+
+model = AutoModelForCausalLM.from_pretrained(
+    "allenai/FlexOlmo-7x7B-1T",
+    quantization_config=torchao_config,
+    dtype=torch.bfloat16,
+    device_map="auto",
+    attn_implementation="sdpa"
+)
+input_ids = tokenizer("Plants create energy through a process known as", return_tensors="pt").to(model.device)
+
+output = model.generate(**input_ids, max_length=50, cache_implementation="static")
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+
+```
+
+
+## FlexOlmoConfig
+
+[[autodoc]] FlexOlmoConfig
+
+## FlexOlmoForCausalLM
+
+[[autodoc]] FlexOlmoForCausalLM
+
+## FlexOlmoModel
+
+[[autodoc]] FlexOlmoModel
+    - forward
+
+## FlexOlmoPreTrainedModel
+
+[[autodoc]] FlexOlmoPreTrainedModel
+    - forward

src/transformers/models/__init__.py

@@ -123,6 +123,7 @@
     from .fastspeech2_conformer import *
     from .flaubert import *
     from .flava import *
+    from .flex_olmo import *
     from .florence2 import *
     from .fnet import *
     from .focalnet import *

src/transformers/models/auto/configuration_auto.py

@@ -148,6 +148,7 @@
         ("fastspeech2_conformer_with_hifigan", "FastSpeech2ConformerWithHifiGanConfig"),
         ("flaubert", "FlaubertConfig"),
         ("flava", "FlavaConfig"),
+        ("flex_olmo", "FlexOlmoConfig"),
         ("florence2", "Florence2Config"),
         ("fnet", "FNetConfig"),
         ("focalnet", "FocalNetConfig"),
@@ -580,6 +581,7 @@
         ("flan-ul2", "FLAN-UL2"),
         ("flaubert", "FlauBERT"),
         ("flava", "FLAVA"),
+        ("flex_olmo", "FlexOlmo"),
         ("florence2", "Florence2"),
         ("fnet", "FNet"),
         ("focalnet", "FocalNet"),

src/transformers/models/auto/modeling_auto.py

@@ -150,6 +150,7 @@ class _BaseModelWithGenerate(PreTrainedModel, GenerationMixin):
         ("fastspeech2_conformer_with_hifigan", "FastSpeech2ConformerWithHifiGan"),
         ("flaubert", "FlaubertModel"),
         ("flava", "FlavaModel"),
+        ("flex_olmo", "FlexOlmoModel"),
         ("florence2", "Florence2Model"),
         ("fnet", "FNetModel"),
         ("focalnet", "FocalNetModel"),
@@ -653,6 +654,7 @@ class _BaseModelWithGenerate(PreTrainedModel, GenerationMixin):
         ("falcon", "FalconForCausalLM"),
         ("falcon_h1", "FalconH1ForCausalLM"),
         ("falcon_mamba", "FalconMambaForCausalLM"),
+        ("flex_olmo", "FlexOlmoForCausalLM"),
         ("fuyu", "FuyuForCausalLM"),
         ("gemma", "GemmaForCausalLM"),
         ("gemma2", "Gemma2ForCausalLM"),

src/transformers/models/auto/tokenization_auto.py

@@ -245,6 +245,7 @@
             ("FastSpeech2ConformerTokenizer" if is_g2p_en_available() else None, None),
         ),
         ("flaubert", ("FlaubertTokenizer", None)),
+        ("flex_olmo", (None, "GPT2TokenizerFast" if is_tokenizers_available() else None)),
         ("fnet", ("FNetTokenizer", "FNetTokenizerFast" if is_tokenizers_available() else None)),
         ("fsmt", ("FSMTTokenizer", None)),
         ("funnel", ("FunnelTokenizer", "FunnelTokenizerFast" if is_tokenizers_available() else None)),

src/transformers/models/flex_olmo/__init__.py

@@ -0,0 +1,29 @@
+# coding=utf-8
+# Copyright 2025 the HuggingFace Team. 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.
+
+from typing import TYPE_CHECKING
+
+from ...utils import _LazyModule
+from ...utils.import_utils import define_import_structure
+
+
+if TYPE_CHECKING:
+    from .configuration_flex_olmo import *
+    from .modeling_flex_olmo import *
+else:
+    import sys
+
+    _file = globals()["__file__"]
+    sys.modules[__name__] = _LazyModule(__name__, _file, define_import_structure(_file), module_spec=__spec__)