optimum-rbln 0.9.3.post1__py3-none-any.whl

optimum/rbln/transformers/__init__.py

@@ -0,0 +1,340 @@
+# Copyright 2025 Rebellions Inc. 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 transformers.utils import _LazyModule
+
+
+_import_structure = {
+    "models": [
+        "RBLNASTForAudioClassification",
+        "RBLNASTForAudioClassificationConfig",
+        "RBLNAutoModel",
+        "RBLNAutoModelForAudioClassification",
+        "RBLNAutoModelForCausalLM",
+        "RBLNAutoModelForCTC",
+        "RBLNAutoModelForDepthEstimation",
+        "RBLNAutoModelForImageClassification",
+        "RBLNAutoModelForImageTextToText",
+        "RBLNAutoModelForMaskedLM",
+        "RBLNAutoModelForQuestionAnswering",
+        "RBLNAutoModelForSeq2SeqLM",
+        "RBLNAutoModelForSequenceClassification",
+        "RBLNAutoModelForSpeechSeq2Seq",
+        "RBLNAutoModelForVision2Seq",
+        "RBLNAutoModelForTextEncoding",
+        "RBLNAutoModelForZeroShotObjectDetection",
+        "RBLNBartForConditionalGeneration",
+        "RBLNBartForConditionalGenerationConfig",
+        "RBLNBartModel",
+        "RBLNBartModelConfig",
+        "RBLNBertForMaskedLM",
+        "RBLNBertForMaskedLMConfig",
+        "RBLNBertForQuestionAnswering",
+        "RBLNBertForQuestionAnsweringConfig",
+        "RBLNBertModel",
+        "RBLNBertModelConfig",
+        "RBLNBlip2ForConditionalGeneration",
+        "RBLNBlip2ForConditionalGenerationConfig",
+        "RBLNBlip2QFormerModel",
+        "RBLNBlip2QFormerModelConfig",
+        "RBLNBlip2VisionModel",
+        "RBLNBlip2VisionModelConfig",
+        "RBLNColPaliForRetrieval",
+        "RBLNColPaliForRetrievalConfig",
+        "RBLNColQwen2ForRetrieval",
+        "RBLNColQwen2ForRetrievalConfig",
+        "RBLNCLIPTextModel",
+        "RBLNCLIPTextModelConfig",
+        "RBLNCLIPTextModelWithProjection",
+        "RBLNCLIPTextModelWithProjectionConfig",
+        "RBLNCLIPVisionModel",
+        "RBLNCLIPVisionModelConfig",
+        "RBLNCLIPVisionModelWithProjection",
+        "RBLNCLIPVisionModelWithProjectionConfig",
+        "RBLNDecoderOnlyModelForCausalLM",
+        "RBLNDecoderOnlyModelForCausalLMConfig",
+        "RBLNDecoderOnlyModelConfig",
+        "RBLNDecoderOnlyModel",
+        "RBLNDistilBertForQuestionAnswering",
+        "RBLNDistilBertForQuestionAnsweringConfig",
+        "RBLNDPTForDepthEstimation",
+        "RBLNDPTForDepthEstimationConfig",
+        "RBLNDepthAnythingForDepthEstimation",
+        "RBLNDepthAnythingForDepthEstimationConfig",
+        "RBLNExaoneForCausalLM",
+        "RBLNExaoneForCausalLMConfig",
+        "RBLNGemmaModel",
+        "RBLNGemmaModelConfig",
+        "RBLNGemma3ForCausalLM",
+        "RBLNGemma3ForCausalLMConfig",
+        "RBLNGemma3ForConditionalGeneration",
+        "RBLNGemma3ForConditionalGenerationConfig",
+        "RBLNGemmaForCausalLM",
+        "RBLNGemmaForCausalLMConfig",
+        "RBLNGPT2LMHeadModel",
+        "RBLNGPT2LMHeadModelConfig",
+        "RBLNGPT2Model",
+        "RBLNGPT2ModelConfig",
+        "RBLNGroundingDinoDecoder",
+        "RBLNGroundingDinoDecoderConfig",
+        "RBLNGroundingDinoForObjectDetection",
+        "RBLNGroundingDinoForObjectDetectionConfig",
+        "RBLNGroundingDinoEncoder",
+        "RBLNGroundingDinoEncoderConfig",
+        "RBLNIdefics3ForConditionalGeneration",
+        "RBLNIdefics3ForConditionalGenerationConfig",
+        "RBLNIdefics3VisionTransformer",
+        "RBLNIdefics3VisionTransformerConfig",
+        "RBLNLlamaForCausalLM",
+        "RBLNLlamaForCausalLMConfig",
+        "RBLNLlavaForConditionalGeneration",
+        "RBLNLlavaForConditionalGenerationConfig",
+        "RBLNLlamaModel",
+        "RBLNLlamaModelConfig",
+        "RBLNOPTForCausalLM",
+        "RBLNOPTForCausalLMConfig",
+        "RBLNPegasusForConditionalGeneration",
+        "RBLNPegasusForConditionalGenerationConfig",
+        "RBLNPegasusModel",
+        "RBLNPegasusModelConfig",
+        "RBLNLlavaNextForConditionalGeneration",
+        "RBLNLlavaNextForConditionalGenerationConfig",
+        "RBLNLoRAAdapterConfig",
+        "RBLNLoRAConfig",
+        "RBLNMidmLMHeadModel",
+        "RBLNMidmLMHeadModelConfig",
+        "RBLNMistralForCausalLM",
+        "RBLNMistralForCausalLMConfig",
+        "RBLNMistralModel",
+        "RBLNMistralModelConfig",
+        "RBLNOPTForCausalLM",
+        "RBLNOPTForCausalLMConfig",
+        "RBLNOPTModel",
+        "RBLNOPTModelConfig",
+        "RBLNPhiForCausalLM",
+        "RBLNPhiForCausalLMConfig",
+        "RBLNPixtralVisionModelConfig",
+        "RBLNPixtralVisionModel",
+        "RBLNPhiModel",
+        "RBLNPhiModelConfig",
+        "RBLNQwen2_5_VisionTransformerPretrainedModel",
+        "RBLNQwen2_5_VisionTransformerPretrainedModelConfig",
+        "RBLNQwen2_5_VLForConditionalGeneration",
+        "RBLNQwen2_5_VLForConditionalGenerationConfig",
+        "RBLNQwen2VisionTransformerPretrainedModel",
+        "RBLNQwen2VisionTransformerPretrainedModelConfig",
+        "RBLNQwen2VLForConditionalGeneration",
+        "RBLNQwen2VLForConditionalGenerationConfig",
+        "RBLNQwen2Model",
+        "RBLNQwen2ModelConfig",
+        "RBLNQwen2ForCausalLM",
+        "RBLNQwen2ForCausalLMConfig",
+        "RBLNQwen3ForCausalLM",
+        "RBLNQwen3ForCausalLMConfig",
+        "RBLNQwen3Model",
+        "RBLNQwen3ModelConfig",
+        "RBLNResNetForImageClassification",
+        "RBLNResNetForImageClassificationConfig",
+        "RBLNRobertaForMaskedLM",
+        "RBLNRobertaForMaskedLMConfig",
+        "RBLNRobertaForSequenceClassification",
+        "RBLNRobertaForSequenceClassificationConfig",
+        "RBLNSiglipVisionModel",
+        "RBLNSiglipVisionModelConfig",
+        "RBLNSwinBackbone",
+        "RBLNSwinBackboneConfig",
+        "RBLNT5EncoderModel",
+        "RBLNT5EncoderModelConfig",
+        "RBLNT5ForConditionalGeneration",
+        "RBLNT5ForConditionalGenerationConfig",
+        "RBLNTimeSeriesTransformerForPrediction",
+        "RBLNTimeSeriesTransformerForPredictionConfig",
+        "RBLNViTForImageClassification",
+        "RBLNViTForImageClassificationConfig",
+        "RBLNWav2Vec2ForCTC",
+        "RBLNWav2Vec2ForCTCConfig",
+        "RBLNWhisperForConditionalGeneration",
+        "RBLNWhisperForConditionalGenerationConfig",
+        "RBLNXLMRobertaForSequenceClassification",
+        "RBLNXLMRobertaForSequenceClassificationConfig",
+        "RBLNXLMRobertaModel",
+        "RBLNXLMRobertaModelConfig",
+    ],
+}
+
+if TYPE_CHECKING:
+    from .models import (
+        RBLNASTForAudioClassification,
+        RBLNASTForAudioClassificationConfig,
+        RBLNAutoModel,
+        RBLNAutoModelForAudioClassification,
+        RBLNAutoModelForCausalLM,
+        RBLNAutoModelForCTC,
+        RBLNAutoModelForDepthEstimation,
+        RBLNAutoModelForImageClassification,
+        RBLNAutoModelForImageTextToText,
+        RBLNAutoModelForMaskedLM,
+        RBLNAutoModelForQuestionAnswering,
+        RBLNAutoModelForSeq2SeqLM,
+        RBLNAutoModelForSequenceClassification,
+        RBLNAutoModelForSpeechSeq2Seq,
+        RBLNAutoModelForTextEncoding,
+        RBLNAutoModelForVision2Seq,
+        RBLNAutoModelForZeroShotObjectDetection,
+        RBLNBartForConditionalGeneration,
+        RBLNBartForConditionalGenerationConfig,
+        RBLNBartModel,
+        RBLNBartModelConfig,
+        RBLNBertForMaskedLM,
+        RBLNBertForMaskedLMConfig,
+        RBLNBertForQuestionAnswering,
+        RBLNBertForQuestionAnsweringConfig,
+        RBLNBertModel,
+        RBLNBertModelConfig,
+        RBLNBlip2ForConditionalGeneration,
+        RBLNBlip2ForConditionalGenerationConfig,
+        RBLNBlip2QFormerModel,
+        RBLNBlip2QFormerModelConfig,
+        RBLNBlip2VisionModel,
+        RBLNBlip2VisionModelConfig,
+        RBLNCLIPTextModel,
+        RBLNCLIPTextModelConfig,
+        RBLNCLIPTextModelWithProjection,
+        RBLNCLIPTextModelWithProjectionConfig,
+        RBLNCLIPVisionModel,
+        RBLNCLIPVisionModelConfig,
+        RBLNCLIPVisionModelWithProjection,
+        RBLNCLIPVisionModelWithProjectionConfig,
+        RBLNColPaliForRetrieval,
+        RBLNColPaliForRetrievalConfig,
+        RBLNColQwen2ForRetrieval,
+        RBLNColQwen2ForRetrievalConfig,
+        RBLNDecoderOnlyModel,
+        RBLNDecoderOnlyModelConfig,
+        RBLNDecoderOnlyModelForCausalLM,
+        RBLNDecoderOnlyModelForCausalLMConfig,
+        RBLNDepthAnythingForDepthEstimation,
+        RBLNDepthAnythingForDepthEstimationConfig,
+        RBLNDistilBertForQuestionAnswering,
+        RBLNDistilBertForQuestionAnsweringConfig,
+        RBLNDPTForDepthEstimation,
+        RBLNDPTForDepthEstimationConfig,
+        RBLNExaoneForCausalLM,
+        RBLNExaoneForCausalLMConfig,
+        RBLNGemma3ForCausalLM,
+        RBLNGemma3ForCausalLMConfig,
+        RBLNGemma3ForConditionalGeneration,
+        RBLNGemma3ForConditionalGenerationConfig,
+        RBLNGemmaForCausalLM,
+        RBLNGemmaForCausalLMConfig,
+        RBLNGemmaModel,
+        RBLNGemmaModelConfig,
+        RBLNGPT2LMHeadModel,
+        RBLNGPT2LMHeadModelConfig,
+        RBLNGPT2Model,
+        RBLNGPT2ModelConfig,
+        RBLNGroundingDinoDecoder,
+        RBLNGroundingDinoDecoderConfig,
+        RBLNGroundingDinoEncoder,
+        RBLNGroundingDinoEncoderConfig,
+        RBLNGroundingDinoForObjectDetection,
+        RBLNGroundingDinoForObjectDetectionConfig,
+        RBLNIdefics3ForConditionalGeneration,
+        RBLNIdefics3ForConditionalGenerationConfig,
+        RBLNIdefics3VisionTransformer,
+        RBLNIdefics3VisionTransformerConfig,
+        RBLNLlamaForCausalLM,
+        RBLNLlamaForCausalLMConfig,
+        RBLNLlamaModel,
+        RBLNLlamaModelConfig,
+        RBLNLlavaForConditionalGeneration,
+        RBLNLlavaForConditionalGenerationConfig,
+        RBLNLlavaNextForConditionalGeneration,
+        RBLNLlavaNextForConditionalGenerationConfig,
+        RBLNLoRAAdapterConfig,
+        RBLNLoRAConfig,
+        RBLNMidmLMHeadModel,
+        RBLNMidmLMHeadModelConfig,
+        RBLNMistralForCausalLM,
+        RBLNMistralForCausalLMConfig,
+        RBLNMistralModel,
+        RBLNMistralModelConfig,
+        RBLNOPTForCausalLM,
+        RBLNOPTForCausalLMConfig,
+        RBLNOPTModel,
+        RBLNOPTModelConfig,
+        RBLNPegasusForConditionalGeneration,
+        RBLNPegasusForConditionalGenerationConfig,
+        RBLNPegasusModel,
+        RBLNPegasusModelConfig,
+        RBLNPhiForCausalLM,
+        RBLNPhiForCausalLMConfig,
+        RBLNPhiModel,
+        RBLNPhiModelConfig,
+        RBLNPixtralVisionModel,
+        RBLNPixtralVisionModelConfig,
+        RBLNQwen2_5_VisionTransformerPretrainedModel,
+        RBLNQwen2_5_VisionTransformerPretrainedModelConfig,
+        RBLNQwen2_5_VLForConditionalGeneration,
+        RBLNQwen2_5_VLForConditionalGenerationConfig,
+        RBLNQwen2ForCausalLM,
+        RBLNQwen2ForCausalLMConfig,
+        RBLNQwen2Model,
+        RBLNQwen2ModelConfig,
+        RBLNQwen2VisionTransformerPretrainedModel,
+        RBLNQwen2VisionTransformerPretrainedModelConfig,
+        RBLNQwen2VLForConditionalGeneration,
+        RBLNQwen2VLForConditionalGenerationConfig,
+        RBLNQwen3ForCausalLM,
+        RBLNQwen3ForCausalLMConfig,
+        RBLNQwen3Model,
+        RBLNQwen3ModelConfig,
+        RBLNResNetForImageClassification,
+        RBLNResNetForImageClassificationConfig,
+        RBLNRobertaForMaskedLM,
+        RBLNRobertaForMaskedLMConfig,
+        RBLNRobertaForSequenceClassification,
+        RBLNRobertaForSequenceClassificationConfig,
+        RBLNSiglipVisionModel,
+        RBLNSiglipVisionModelConfig,
+        RBLNSwinBackbone,
+        RBLNSwinBackboneConfig,
+        RBLNT5EncoderModel,
+        RBLNT5EncoderModelConfig,
+        RBLNT5ForConditionalGeneration,
+        RBLNT5ForConditionalGenerationConfig,
+        RBLNTimeSeriesTransformerForPrediction,
+        RBLNTimeSeriesTransformerForPredictionConfig,
+        RBLNViTForImageClassification,
+        RBLNViTForImageClassificationConfig,
+        RBLNWav2Vec2ForCTC,
+        RBLNWav2Vec2ForCTCConfig,
+        RBLNWhisperForConditionalGeneration,
+        RBLNWhisperForConditionalGenerationConfig,
+        RBLNXLMRobertaForSequenceClassification,
+        RBLNXLMRobertaForSequenceClassificationConfig,
+        RBLNXLMRobertaModel,
+        RBLNXLMRobertaModelConfig,
+    )
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(
+        __name__,
+        globals()["__file__"],
+        _import_structure,
+        module_spec=__spec__,
+    )

optimum/rbln/transformers/configuration_generic.py

@@ -0,0 +1,120 @@
+# Copyright 2025 Rebellions Inc. 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 Any, List, Optional, Tuple, Union
+
+from ..configuration_utils import RBLNModelConfig
+
+
+class RBLNTransformerEncoderConfig(RBLNModelConfig):
+    rbln_model_input_names: Optional[List[str]] = None
+
+    def __init__(
+        self,
+        max_seq_len: Optional[int] = None,
+        batch_size: Optional[int] = None,
+        model_input_names: Optional[List[str]] = None,
+        model_input_shapes: Optional[List[Tuple[int, int]]] = None,
+        **kwargs: Any,
+    ):
+        """
+        Args:
+            max_seq_len (Optional[int]): Maximum sequence length supported by the model.
+            batch_size (Optional[int]): The batch size for inference. Defaults to 1.
+            model_input_names (Optional[List[str]]): Names of the input tensors for the model.
+                Defaults to class-specific rbln_model_input_names if not provided.
+            kwargs: Additional arguments passed to the parent RBLNModelConfig.
+
+        Raises:
+            ValueError: If batch_size is not a positive integer.
+        """
+        super().__init__(**kwargs)
+        self.max_seq_len = max_seq_len
+        self.batch_size = batch_size or 1
+        if not isinstance(self.batch_size, int) or self.batch_size < 0:
+            raise ValueError(f"batch_size must be a positive integer, got {self.batch_size}")
+
+        self.model_input_names = model_input_names or self.rbln_model_input_names
+        self.model_input_shapes = model_input_shapes
+
+
+class RBLNImageModelConfig(RBLNModelConfig):
+    def __init__(
+        self,
+        image_size: Optional[Union[int, Tuple[int, int]]] = None,
+        batch_size: Optional[int] = None,
+        **kwargs: Any,
+    ):
+        """
+        Args:
+            image_size (Optional[Union[int, Tuple[int, int]]]): The size of input images.
+                Can be an integer for square images or a tuple (height, width).
+            batch_size (Optional[int]): The batch size for inference. Defaults to 1.
+            kwargs: Additional arguments passed to the parent RBLNModelConfig.
+
+        Raises:
+            ValueError: If batch_size is not a positive integer.
+        """
+        super().__init__(**kwargs)
+        self.image_size = image_size
+        self.batch_size = batch_size or 1
+        if not isinstance(self.batch_size, int) or self.batch_size < 0:
+            raise ValueError(f"batch_size must be a positive integer, got {self.batch_size}")
+
+    @property
+    def image_width(self):
+        if isinstance(self.image_size, int):
+            return self.image_size
+        elif isinstance(self.image_size, (list, tuple)):
+            return self.image_size[1]
+        else:
+            return self.image_size["width"]
+
+    @property
+    def image_height(self):
+        if isinstance(self.image_size, int):
+            return self.image_size
+        elif isinstance(self.image_size, (list, tuple)):
+            return self.image_size[0]
+        else:
+            return self.image_size["height"]
+
+
+class RBLNModelForQuestionAnsweringConfig(RBLNTransformerEncoderConfig):
+    pass
+
+
+class RBLNModelForSequenceClassificationConfig(RBLNTransformerEncoderConfig):
+    pass
+
+
+class RBLNModelForMaskedLMConfig(RBLNTransformerEncoderConfig):
+    pass
+
+
+class RBLNModelForTextEncodingConfig(RBLNTransformerEncoderConfig):
+    pass
+
+
+# FIXME : Appropriate name ?
+class RBLNTransformerEncoderForFeatureExtractionConfig(RBLNTransformerEncoderConfig):
+    pass
+
+
+class RBLNModelForImageClassificationConfig(RBLNImageModelConfig):
+    pass
+
+
+class RBLNModelForDepthEstimationConfig(RBLNImageModelConfig):
+    pass