Deprecate unused OpenLlama architecture (#24922)

* Resolve typo in check_repo.py

* Specify encoding when opening modeling files

* Deprecate the OpenLlama architecture

* Add disclaimer pointing to Llama

I'm open to different wordings here

* Match the capitalisation of LLaMA

docs/source/en/model_doc/open-llama.md

@@ -16,6 +16,21 @@ rendered properly in your Markdown viewer.
 
 # Open-Llama
 
+<Tip warning={true}>
+
+This model is in maintenance mode only, so we won't accept any new PRs changing its code.
+
+If you run into any issues running this model, please reinstall the last version that supported this model: v4.31.0.
+You can do so by running the following command: `pip install -U transformers==4.31.0`.
+
+</Tip>
+
+<Tip warning={true}>
+
+This model differs from the [OpenLLaMA models](https://huggingface.co/models?search=openllama) on the Hugging Face Hub, which primarily use the [LLaMA](llama) architecture.
+
+</Tip>
+
 ## Overview
 
 The Open-Llama model was proposed in [Open-Llama project](https://github.com/s-JoL/Open-Llama) by community developer s-JoL.

src/transformers/__init__.py

@@ -278,6 +278,7 @@
         "MCTCTProcessor",
     ],
     "models.deprecated.mmbt": ["MMBTConfig"],
+    "models.deprecated.open_llama": ["OPEN_LLAMA_PRETRAINED_CONFIG_ARCHIVE_MAP", "OpenLlamaConfig"],
     "models.deprecated.retribert": [
         "RETRIBERT_PRETRAINED_CONFIG_ARCHIVE_MAP",
         "RetriBertConfig",
@@ -445,7 +446,6 @@
         "NystromformerConfig",
     ],
     "models.oneformer": ["ONEFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP", "OneFormerConfig", "OneFormerProcessor"],
-    "models.open_llama": ["OPEN_LLAMA_PRETRAINED_CONFIG_ARCHIVE_MAP", "OpenLlamaConfig"],
     "models.openai": ["OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP", "OpenAIGPTConfig", "OpenAIGPTTokenizer"],
     "models.opt": ["OPTConfig"],
     "models.owlvit": [
@@ -1536,6 +1536,9 @@
         ]
     )
     _import_structure["models.deprecated.mmbt"].extend(["MMBTForClassification", "MMBTModel", "ModalEmbeddings"])
+    _import_structure["models.deprecated.open_llama"].extend(
+        ["OpenLlamaForCausalLM", "OpenLlamaForSequenceClassification", "OpenLlamaModel", "OpenLlamaPreTrainedModel"]
+    )
     _import_structure["models.deprecated.retribert"].extend(
         ["RETRIBERT_PRETRAINED_MODEL_ARCHIVE_LIST", "RetriBertModel", "RetriBertPreTrainedModel"]
     )
@@ -2300,9 +2303,6 @@
             "OneFormerPreTrainedModel",
         ]
     )
-    _import_structure["models.open_llama"].extend(
-        ["OpenLlamaForCausalLM", "OpenLlamaForSequenceClassification", "OpenLlamaModel", "OpenLlamaPreTrainedModel"]
-    )
     _import_structure["models.openai"].extend(
         [
             "OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST",
@@ -4239,6 +4239,7 @@
         MCTCTProcessor,
     )
     from .models.deprecated.mmbt import MMBTConfig
+    from .models.deprecated.open_llama import OPEN_LLAMA_PRETRAINED_CONFIG_ARCHIVE_MAP, OpenLlamaConfig
     from .models.deprecated.retribert import (
         RETRIBERT_PRETRAINED_CONFIG_ARCHIVE_MAP,
         RetriBertConfig,
@@ -4390,7 +4391,6 @@
     from .models.nllb_moe import NLLB_MOE_PRETRAINED_CONFIG_ARCHIVE_MAP, NllbMoeConfig
     from .models.nystromformer import NYSTROMFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, NystromformerConfig
     from .models.oneformer import ONEFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, OneFormerConfig, OneFormerProcessor
-    from .models.open_llama import OPEN_LLAMA_PRETRAINED_CONFIG_ARCHIVE_MAP, OpenLlamaConfig
     from .models.openai import OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP, OpenAIGPTConfig, OpenAIGPTTokenizer
     from .models.opt import OPTConfig
     from .models.owlvit import (
@@ -5334,6 +5334,12 @@
             MCTCTPreTrainedModel,
         )
         from .models.deprecated.mmbt import MMBTForClassification, MMBTModel, ModalEmbeddings
+        from .models.deprecated.open_llama import (
+            OpenLlamaForCausalLM,
+            OpenLlamaForSequenceClassification,
+            OpenLlamaModel,
+            OpenLlamaPreTrainedModel,
+        )
         from .models.deprecated.retribert import (
             RETRIBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
             RetriBertModel,
@@ -5954,12 +5960,6 @@
             OneFormerModel,
             OneFormerPreTrainedModel,
         )
-        from .models.open_llama import (
-            OpenLlamaForCausalLM,
-            OpenLlamaForSequenceClassification,
-            OpenLlamaModel,
-            OpenLlamaPreTrainedModel,
-        )
         from .models.openai import (
             OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST,
             OpenAIGPTDoubleHeadsModel,

src/transformers/models/__init__.py

@@ -145,7 +145,6 @@
     nllb_moe,
     nystromformer,
     oneformer,
-    open_llama,
     openai,
     opt,
     owlvit,

src/transformers/models/auto/configuration_auto.py

@@ -652,6 +652,7 @@
     "bort",
     "mctct",
     "mmbt",
+    "open_llama",
     "retribert",
     "tapex",
     "trajectory_transformer",

src/transformers/models/open_llama/__init__.py → src/transformers/models/deprecated/open_llama/__init__.py

@@ -13,7 +13,7 @@
 # limitations under the License.
 from typing import TYPE_CHECKING
 
-from ...utils import (
+from ....utils import (
     OptionalDependencyNotAvailable,
     _LazyModule,
     is_sentencepiece_available,

src/transformers/models/open_llama/configuration_open_llama.py → src/transformers/models/deprecated/open_llama/configuration_open_llama.py

@@ -19,8 +19,8 @@
 # limitations under the License.
 """ Open-Llama model configuration"""
 
-from ...configuration_utils import PretrainedConfig
-from ...utils import logging
+from ....configuration_utils import PretrainedConfig
+from ....utils import logging
 
 
 logger = logging.get_logger(__name__)

src/transformers/models/open_llama/modeling_open_llama.py → src/transformers/models/deprecated/open_llama/modeling_open_llama.py

@@ -26,10 +26,10 @@
 from torch import nn
 from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
 
-from ...activations import ACT2FN
-from ...modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast, SequenceClassifierOutputWithPast
-from ...modeling_utils import PreTrainedModel
-from ...utils import add_start_docstrings, add_start_docstrings_to_model_forward, logging, replace_return_docstrings
+from ....activations import ACT2FN
+from ....modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast, SequenceClassifierOutputWithPast
+from ....modeling_utils import PreTrainedModel
+from ....utils import add_start_docstrings, add_start_docstrings_to_model_forward, logging, replace_return_docstrings
 from .configuration_open_llama import OpenLlamaConfig
 
 

src/transformers/utils/dummy_pt_objects.py

@@ -2396,6 +2396,34 @@ def __init__(self, *args, **kwargs):
         requires_backends(self, ["torch"])
 
 
+class OpenLlamaForCausalLM(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class OpenLlamaForSequenceClassification(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class OpenLlamaModel(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
+class OpenLlamaPreTrainedModel(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+
 RETRIBERT_PRETRAINED_MODEL_ARCHIVE_LIST = None
 
 
@@ -5461,34 +5489,6 @@ def __init__(self, *args, **kwargs):
         requires_backends(self, ["torch"])
 
 
-class OpenLlamaForCausalLM(metaclass=DummyObject):
-    _backends = ["torch"]
-
-    def __init__(self, *args, **kwargs):
-        requires_backends(self, ["torch"])
-
-
-class OpenLlamaForSequenceClassification(metaclass=DummyObject):
-    _backends = ["torch"]
-
-    def __init__(self, *args, **kwargs):
-        requires_backends(self, ["torch"])
-
-
-class OpenLlamaModel(metaclass=DummyObject):
-    _backends = ["torch"]
-
-    def __init__(self, *args, **kwargs):
-        requires_backends(self, ["torch"])
-
-
-class OpenLlamaPreTrainedModel(metaclass=DummyObject):
-    _backends = ["torch"]
-
-    def __init__(self, *args, **kwargs):
-        requires_backends(self, ["torch"])
-
-
 OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST = None