sinapsis-huggingface 0.2.3__tar.gz → 0.2.6__tar.gz

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sinapsis-huggingface
-Version: 0.2.3
+Version: 0.2.6
 Summary: Package for HuggingFace-based templates
 Author-email: SinapsisAI <dev@sinapsis.tech>
 Project-URL: Homepage, https://sinapsis.tech

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sinapsis-huggingface
-Version: 0.2.3
+Version: 0.2.6
 Summary: Package for HuggingFace-based templates
 Author-email: SinapsisAI <dev@sinapsis.tech>
 Project-URL: Homepage, https://sinapsis.tech

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface.egg-info/SOURCES.txt

@@ -7,6 +7,8 @@ packages/sinapsis_huggingface.egg-info/dependency_links.txt
 packages/sinapsis_huggingface.egg-info/requires.txt
 packages/sinapsis_huggingface.egg-info/top_level.txt
 packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/__init__.py
+packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/helpers/__init__.py
+packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/helpers/tags.py
 packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/__init__.py
 packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/base_diffusers.py
 packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/image_to_image_diffusers.py
@@ -14,6 +16,8 @@ packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templ
 packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/inpainting_diffusers.py
 packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/text_to_image_diffusers.py
 packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/__init__.py
+packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/helpers/__init__.py
+packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/helpers/tags.py
 packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/templates/__init__.py
 packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/templates/hugging_face_embedding_extractor.py
 packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/templates/speaker_embedding_from_audio.py
@@ -21,11 +25,13 @@ packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/tem
 packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/__init__.py
 packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/helpers/__init__.py
 packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/helpers/grounding_dino_keys.py
+packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/helpers/tags.py
 packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/templates/__init__.py
 packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/templates/grounding_dino.py
 packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/templates/grounding_dino_classification.py
 packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/__init__.py
 packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/helpers/__init__.py
+packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/helpers/tags.py
 packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/helpers/text_to_sentences.py
 packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/__init__.py
 packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/base_transformers.py

sinapsis_huggingface-0.2.6/packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/helpers/tags.py

@@ -0,0 +1,21 @@
+# -*- coding: utf-8 -*-
+from enum import Enum
+
+
+class Tags(Enum):
+    BBOX = "bbox"
+    DIFFUSERS = "diffusers"
+    GENERATIVE = "generative"
+    HUGGINGFACE = "huggingface"
+    IMAGE = "image"
+    IMAGE_TO_IMAGE = "image_to_image"
+    IMAGE_TO_VIDEO = "image_to_video"
+    IMAGE_GENERATION = "image_generation"
+    INPAINTING = "inpainting"
+    MASK = "mask"
+    MODELS = "models"
+    PROMPTS = "prompts"
+    TEXT = "text"
+    TEXT_TO_IMAGE = "text_to_image"
+    VIDEO = "video"
+    VIDEO_GENERATION = "video_generation"

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/base_diffusers.py

@@ -1,6 +1,5 @@
 # -*- coding: utf-8 -*-
 
-import gc
 from abc import ABC, abstractmethod
 from typing import Any, Literal
 
@@ -18,6 +17,8 @@ from sinapsis_core.template_base.base_models import (
 )
 from sinapsis_core.utils.env_var_keys import SINAPSIS_CACHE_DIR
 
+from sinapsis_huggingface_diffusers.helpers.tags import Tags
+
 
 class BaseDiffusersAttributes(TemplateAttributes):
     """Configuration attributes for setting up a diffusion pipeline and generating images.
@@ -66,7 +67,11 @@ class BaseDiffusers(Template, ABC):
     """
 
     AttributesBaseModel = BaseDiffusersAttributes
-    UIProperties = UIPropertiesMetadata(category="HuggingFace", output_type=OutputTypes.IMAGE)
+    UIProperties = UIPropertiesMetadata(
+        category="HuggingFace",
+        output_type=OutputTypes.IMAGE,
+        tags=[Tags.HUGGINGFACE, Tags.DIFFUSERS, Tags.MODELS, Tags.GENERATIVE],
+    )
     TORCH_DTYPE: dict = TorchTypes().model_dump()
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
@@ -188,6 +193,9 @@ class BaseDiffusers(Template, ABC):
         This method performs garbage collection and clears GPU memory (if applicable) to prevent memory leaks
         and ensure efficient resource usage.
         """
-        gc.collect()
         if self.attributes.device == "cuda":
             torch.cuda.empty_cache()
+
+    def reset_state(self, template_name: str | None = None) -> None:
+        self._clear_memory()
+        super().reset_state(template_name)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/image_to_image_diffusers.py

@@ -10,8 +10,12 @@ from PIL import Image
 from sinapsis_core.data_containers.data_packet import DataContainer, ImagePacket
 from sinapsis_core.template_base.base_models import TemplateAttributeType
 
+from sinapsis_huggingface_diffusers.helpers.tags import Tags
 from sinapsis_huggingface_diffusers.templates.base_diffusers import BaseDiffusers
 
+ImageToImageDiffusersUIProperties = BaseDiffusers.UIProperties
+ImageToImageDiffusersUIProperties.tags.extend([Tags.IMAGE, Tags.IMAGE_GENERATION, Tags.IMAGE_TO_IMAGE])
+
 
 class ImageToImageDiffusers(BaseDiffusers):
     """Template for the image-to-image generative task using diffusers.
@@ -46,6 +50,8 @@ class ImageToImageDiffusers(BaseDiffusers):
 
     """
 
+    UIProperties = ImageToImageDiffusersUIProperties
+
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)
         self.requires_pil = False

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/image_to_video_gen_xl_diffusers.py

@@ -3,10 +3,14 @@
 from diffusers import I2VGenXLPipeline
 from sinapsis_core.template_base.base_models import TemplateAttributeType
 
+from sinapsis_huggingface_diffusers.helpers.tags import Tags
 from sinapsis_huggingface_diffusers.templates.image_to_image_diffusers import (
     ImageToImageDiffusers,
 )
 
+ImageToVideoGenXLDiffusersUIProperties = ImageToImageDiffusers.UIProperties
+ImageToVideoGenXLDiffusersUIProperties.tags.extend([Tags.VIDEO, Tags.IMAGE_TO_VIDEO])
+
 
 class ImageToVideoGenXLDiffusers(ImageToImageDiffusers):
     """This class implements a specific template for image-to-video generation using Hugging Face's
@@ -38,6 +42,7 @@ class ImageToVideoGenXLDiffusers(ImageToImageDiffusers):
 
     """
 
+    UIProperties = ImageToVideoGenXLDiffusersUIProperties
     DEFAULT_NUM_FRAMES = 16
 
     def __init__(self, attributes: TemplateAttributeType) -> None:

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/inpainting_diffusers.py

@@ -9,6 +9,7 @@ from diffusers import AutoPipelineForInpainting
 from sinapsis_core.data_containers.data_packet import ImageAnnotations, ImagePacket
 from sinapsis_core.template_base.base_models import TemplateAttributeType
 
+from sinapsis_huggingface_diffusers.helpers.tags import Tags
 from sinapsis_huggingface_diffusers.templates.base_diffusers import (
     BaseDiffusersAttributes,
 )
@@ -16,6 +17,9 @@ from sinapsis_huggingface_diffusers.templates.image_to_image_diffusers import (
     ImageToImageDiffusers,
 )
 
+InpaintingDiffusersUIProperties = ImageToImageDiffusers.UIProperties
+InpaintingDiffusersUIProperties.tags.extend([Tags.INPAINTING, Tags.BBOX, Tags.MASK])
+
 
 class InpaintingDiffusersAttributes(BaseDiffusersAttributes):
     """Attributes specific to the InpaintingDiffusers template.
@@ -68,6 +72,7 @@ class InpaintingDiffusers(ImageToImageDiffusers):
 
     """
 
+    UIProperties = InpaintingDiffusersUIProperties
     AttributesBaseModel = InpaintingDiffusersAttributes
 
     def __init__(self, attributes: TemplateAttributeType) -> None:

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_diffusers/src/sinapsis_huggingface_diffusers/templates/text_to_image_diffusers.py

@@ -3,8 +3,12 @@
 from diffusers import AutoPipelineForText2Image
 from sinapsis_core.data_containers.data_packet import DataContainer, ImagePacket
 
+from sinapsis_huggingface_diffusers.helpers.tags import Tags
 from sinapsis_huggingface_diffusers.templates.base_diffusers import BaseDiffusers
 
+TextToImageDiffusersUIProperties = BaseDiffusers.UIProperties
+TextToImageDiffusersUIProperties.tags.extend([Tags.TEXT, Tags.TEXT_TO_IMAGE, Tags.PROMPTS])
+
 
 class TextToImageDiffusers(BaseDiffusers):
     """This class implements a specific template for text-to-image generation using Hugging Face's
@@ -34,6 +38,8 @@ class TextToImageDiffusers(BaseDiffusers):
         generation_params: {}
     """
 
+    UIProperties = TextToImageDiffusersUIProperties
+
     @staticmethod
     def _pipeline_class() -> AutoPipelineForText2Image:
         """Returns the `AutoPipelineForText2Image` class to be used for text-to-image generation.

sinapsis_huggingface-0.2.6/packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/helpers/tags.py

@@ -0,0 +1,13 @@
+# -*- coding: utf-8 -*-
+from enum import Enum
+
+
+class Tags(Enum):
+    AUDIO = "audio"
+    DATASET = "dataset"
+    EMBEDDINGS = "embeddings"
+    EXTRACTOR = "extractor"
+    HUGGINGFACE = "huggingface"
+    MODELS = "models"
+    SPEAKER_EMBEDDING = "speaker_embedding"
+    TEXT = "text"

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/templates/__init__.py

@@ -5,10 +5,8 @@ from typing import Callable
 _root_lib_path = "sinapsis_huggingface_embeddings.templates"
 
 _template_lookup = {
-    "HuggingFaceEmbeddingNodeGenerator": f"{_root_lib_path}.embedding_node_generator",
     "HuggingFaceEmbeddingExtractor": f"{_root_lib_path}.hugging_face_embedding_extractor",
     "SpeakerEmbeddingFromAudio": f"{_root_lib_path}.speaker_embedding_from_audio",
-    "SpeakerEmbeddingFromDataset": f"{_root_lib_path}.speaker_embedding_from_dataset",
 }
 
 

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/templates/hugging_face_embedding_extractor.py

@@ -12,6 +12,8 @@ from sinapsis_core.template_base.base_models import (
     UIPropertiesMetadata,
 )
 
+from sinapsis_huggingface_embeddings.helpers.tags import Tags
+
 
 class HuggingFaceEmbeddingExtractorAttributes(TemplateAttributes):
     """
@@ -46,7 +48,11 @@ class HuggingFaceEmbeddingExtractor(Template):
     """
 
     AttributesBaseModel = HuggingFaceEmbeddingExtractorAttributes
-    UIProperties = UIPropertiesMetadata(category="HuggingFace", output_type=OutputTypes.TEXT)
+    UIProperties = UIPropertiesMetadata(
+        category="HuggingFace",
+        output_type=OutputTypes.TEXT,
+        tags=[Tags.TEXT, Tags.EXTRACTOR, Tags.TEXT, Tags.HUGGINGFACE, Tags.MODELS],
+    )
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_embeddings/src/sinapsis_huggingface_embeddings/templates/speaker_embedding_from_audio.py

@@ -9,10 +9,17 @@ import torch.nn.functional as F
 import torchaudio
 from sinapsis_core.data_containers.data_packet import AudioPacket, DataContainer
 from sinapsis_core.template_base import Template
-from sinapsis_core.template_base.base_models import TemplateAttributes, TemplateAttributeType
+from sinapsis_core.template_base.base_models import (
+    OutputTypes,
+    TemplateAttributes,
+    TemplateAttributeType,
+    UIPropertiesMetadata,
+)
 from sinapsis_core.utils.env_var_keys import SINAPSIS_CACHE_DIR
 from speechbrain.inference.speaker import EncoderClassifier
 
+from sinapsis_huggingface_embeddings.helpers.tags import Tags
+
 
 class SpeakerEmbeddingFromAudioAttributes(TemplateAttributes):
     """Attributes for SpeakerEmbeddingFromAudio template.
@@ -63,7 +70,11 @@ class SpeakerEmbeddingFromAudio(Template):
     """
 
     AttributesBaseModel = SpeakerEmbeddingFromAudioAttributes
-    CATEGORY = "Embeddings"
+    UIProperties = UIPropertiesMetadata(
+        category="HuggingFace",
+        output_type=OutputTypes.AUDIO,
+        tags=[Tags.EMBEDDINGS, Tags.HUGGINGFACE, Tags.MODELS, Tags.AUDIO, Tags.SPEAKER_EMBEDDING],
+    )
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)
@@ -134,8 +145,8 @@ class SpeakerEmbeddingFromAudio(Template):
                 packets.
         """
         if not container.audios:
-            raise ValueError("No audio packets found in the container.")
-
+            self.logger.debug("No audio packets found in the container.")
+            return container
         embeddings = [
             self._postprocess_speaker_embedding(
                 self.classifier.encode_batch(self._process_audio(audio_packet), normalize=self.attributes.normalize)

sinapsis_huggingface-0.2.6/packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/helpers/tags.py

@@ -0,0 +1,15 @@
+# -*- coding: utf-8 -*-
+from enum import Enum
+
+
+class Tags(Enum):
+    ANNOTATIONS = "annotations"
+    CLASSIFICATION = "classification"
+    CLASSIFIER = "classifier"
+    GROUNDING_DINO = "grounding_dino"
+    HUGGINGFACE = "huggingface"
+    IMAGE = "image"
+    INFERENCE = "inference"
+    MODELS = "models"
+    TRANSFORMERS = "transformers"
+    ZERO_SHOT = "zero_shot"

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/templates/grounding_dino.py

@@ -22,6 +22,7 @@ from transformers import (
 )
 
 from sinapsis_huggingface_grounding_dino.helpers.grounding_dino_keys import GroundingDINOKeys
+from sinapsis_huggingface_grounding_dino.helpers.tags import Tags
 
 
 class GroundingBaseAttributes(TemplateAttributes):
@@ -80,7 +81,19 @@ class GroundingDINO(Template):
 
     """
 
-    UIProperties = UIPropertiesMetadata(category="HuggingFace", output_type=OutputTypes.IMAGE)
+    UIProperties = UIPropertiesMetadata(
+        category="HuggingFace",
+        output_type=OutputTypes.IMAGE,
+        tags=[
+            Tags.ANNOTATIONS,
+            Tags.GROUNDING_DINO,
+            Tags.HUGGINGFACE,
+            Tags.IMAGE,
+            Tags.INFERENCE,
+            Tags.MODELS,
+            Tags.TRANSFORMERS,
+        ],
+    )
     KEYS = GroundingDINOKeys()
 
     class AttributesBaseModel(GroundingBaseAttributes):
@@ -337,3 +350,16 @@ class GroundingDINO(Template):
             self._process_image_packet(image_packet)
 
         return container
+
+    def _clear_memory(self) -> None:
+        """Clears memory to free up resources.
+
+        This method performs garbage collection and clears GPU memory (if applicable) to prevent memory leaks
+        and ensure efficient resource usage.
+        """
+        if self.attributes.device == "cuda":
+            torch.cuda.empty_cache()
+
+    def reset_state(self, template_name: str | None = None) -> None:
+        self._clear_memory()
+        super().reset_state(template_name)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_grounding_dino/src/sinapsis_huggingface_grounding_dino/templates/grounding_dino_classification.py

@@ -5,8 +5,12 @@ from typing import Any
 from sinapsis_core.data_containers.annotations import ImageAnnotations
 from sinapsis_core.data_containers.data_packet import ImagePacket
 
+from sinapsis_huggingface_grounding_dino.helpers.tags import Tags
 from sinapsis_huggingface_grounding_dino.templates.grounding_dino import GroundingBaseAttributes, GroundingDINO
 
+GroundingDINOClassificationUIProperties = GroundingDINO.UIProperties
+GroundingDINOClassificationUIProperties.tags.extend([Tags.CLASSIFICATION, Tags.CLASSIFIER, Tags.ZERO_SHOT])
+
 
 class GroundingDINOClassificationAttributes(GroundingBaseAttributes):
     """Attributes for GroundingDINO Classification.
@@ -56,6 +60,7 @@ class GroundingDINOClassification(GroundingDINO):
     """
 
     AttributesBaseModel = GroundingDINOClassificationAttributes
+    UIProperties = GroundingDINOClassificationUIProperties
 
     def validate_and_format_text_input(self, text_input: str) -> str:
         """Validates and formats the text input for consistency.

sinapsis_huggingface-0.2.6/packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/helpers/tags.py

@@ -0,0 +1,24 @@
+# -*- coding: utf-8 -*-
+from enum import Enum
+
+
+class Tags(Enum):
+    AUDIO = "audio"
+    CAPTION_GENERATION = "caption_generation"
+    HUGGINGFACE = "huggingface"
+    IMAGE = "image"
+    IMAGE_TO_TEXT = "image_to_text"
+    INFERENCE = "inference"
+    LANGUAGE = "language"
+    MODELS = "models"
+    OBJECT_DETECTION = "object_detection"
+    PALIGEMMA = "paligemma"
+    PROMPT = "prompt"
+    SPEECH = "speech"
+    SPEECH_RECOGNITION = "speech_recognition"
+    SPEECH_TO_TEXT = "speech_to_text"
+    SUMMARIZATION = "summarization"
+    TEXT = "text"
+    TEXT_TO_SPEECH = "text_to_speech"
+    TRANSFORMERS = "transformers"
+    TRANSLATION = "translation"

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/base_transformers.py

@@ -13,6 +13,8 @@ from sinapsis_core.utils.env_var_keys import SINAPSIS_CACHE_DIR
 from transformers import AutoProcessor, pipeline
 from transformers.pipelines import Pipeline
 
+from sinapsis_huggingface_transformers.helpers.tags import Tags
+
 
 class TransformersBaseAttributes(TemplateAttributes):
     """Attributes for configuring the TransformersPipelineTemplate.
@@ -53,7 +55,10 @@ class TransformersBase(Template):
     """
 
     AttributesBaseModel = TransformersBaseAttributes
-    UIProperties = UIPropertiesMetadata(category="Transformers")
+    UIProperties = UIPropertiesMetadata(
+        category="Transformers",
+        tags=[Tags.HUGGINGFACE, Tags.TRANSFORMERS, Tags.MODELS],
+    )
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)
@@ -140,3 +145,8 @@ class TransformersBase(Template):
         """
         transformed_data_container = self.transformation_method(container)
         return transformed_data_container
+
+    def reset_state(self, template_name: str | None = None) -> None:
+        if self.attributes.device == "cuda":
+            torch.cuda.empty_cache()
+        super().reset_state(template_name)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/image_to_text_transformers.py

@@ -3,10 +3,15 @@
 import numpy as np
 from PIL import Image
 from sinapsis_core.data_containers.data_packet import DataContainer, TextPacket
-from sinapsis_core.template_base.base_models import TemplateAttributeType
+from sinapsis_core.template_base.base_models import OutputTypes, TemplateAttributeType
 
+from sinapsis_huggingface_transformers.helpers.tags import Tags
 from sinapsis_huggingface_transformers.templates.base_transformers import TransformersBase
 
+ImageToTextTransformersUIProperties = TransformersBase.UIProperties
+ImageToTextTransformersUIProperties.output_type = OutputTypes.TEXT
+ImageToTextTransformersUIProperties.tags.extend([Tags.IMAGE, Tags.TEXT, Tags.IMAGE_TO_TEXT])
+
 
 class ImageToTextTransformers(TransformersBase):
     """
@@ -34,6 +39,7 @@ class ImageToTextTransformers(TransformersBase):
     """
 
     GENERATED_TEXT_KEY = "generated_text"
+    UIProperties = ImageToTextTransformersUIProperties
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/pali_gemma/pali_gemma_base.py

@@ -12,6 +12,7 @@ from sinapsis_core.template_base.base_models import (
     UIPropertiesMetadata,
 )
 from sinapsis_core.utils.env_var_keys import SINAPSIS_CACHE_DIR
+from sinapsis_huggingface_transformers.helpers.tags import Tags
 from transformers import AutoProcessor, PaliGemmaForConditionalGeneration
 
 
@@ -43,7 +44,11 @@ class PaliGemmaBase(Template):
     """Base class for PaliGemma implementations."""
 
     AttributesBaseModel = PaliGemmaBaseAttributes
-    UIProperties = UIPropertiesMetadata(category="HuggingFace", output_type=OutputTypes.IMAGE)
+    UIProperties = UIPropertiesMetadata(
+        category="HuggingFace",
+        output_type=OutputTypes.IMAGE,
+        tags=[Tags.HUGGINGFACE, Tags.IMAGE, Tags.PALIGEMMA, Tags.MODELS],
+    )
     _TORCH_DTYPE: ClassVar[dict[str, Any]] = {"float16": torch.float16, "float32": torch.float32}
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
@@ -97,3 +102,8 @@ class PaliGemmaBase(Template):
         Returns:
             DataContainer: The processed container with model outputs.
         """
+
+    def reset_state(self, template_name: str | None = None) -> None:
+        if self.attributes.device == "cuda":
+            torch.cuda.empty_cache()
+        super().reset_state(template_name)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/pali_gemma/pali_gemma_inference.py

@@ -5,12 +5,16 @@ from sinapsis_core.data_containers.annotations import BoundingBox, ImageAnnotati
 from sinapsis_core.data_containers.data_packet import DataContainer, ImagePacket
 from sinapsis_core.template_base.base_models import TemplateAttributeType
 from sinapsis_data_visualization.helpers.detection_utils import bbox_xyxy_to_xywh
+from sinapsis_huggingface_transformers.helpers.tags import Tags
 from sinapsis_huggingface_transformers.templates.pali_gemma.pali_gemma_base import (
     PaliGemmaBase,
     PaliGemmaBaseAttributes,
 )
 from transformers.generation.utils import GenerateOutput
 
+PaliGemmaInferenceUIProperties = PaliGemmaBase.UIProperties
+PaliGemmaInferenceUIProperties.tags.extend([Tags.CAPTION_GENERATION, Tags.OBJECT_DETECTION, Tags.INFERENCE])
+
 
 class PaliGemmaInferenceAttributes(PaliGemmaBaseAttributes):
     """Configuration attributes for PaliGemma inference.
@@ -61,6 +65,7 @@ class PaliGemmaInference(PaliGemmaBase):
 
     AttributesBaseModel = PaliGemmaInferenceAttributes
     INPUT_IDS = "input_ids"
+    UIProperties = PaliGemmaInferenceUIProperties
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/speech_to_text_transformers.py

@@ -2,10 +2,17 @@
 
 import numpy as np
 from sinapsis_core.data_containers.data_packet import DataContainer, TextPacket
-from sinapsis_core.template_base.base_models import TemplateAttributeType
+from sinapsis_core.template_base.base_models import OutputTypes, TemplateAttributeType
 
+from sinapsis_huggingface_transformers.helpers.tags import Tags
 from sinapsis_huggingface_transformers.templates.base_transformers import TransformersBase
 
+SpeechToTextTransformersUIProperties = TransformersBase.UIProperties
+SpeechToTextTransformersUIProperties.output_type = OutputTypes.TEXT
+SpeechToTextTransformersUIProperties.tags.extend(
+    [Tags.SPEECH, Tags.SPEECH_TO_TEXT, Tags.AUDIO, Tags.SPEECH_RECOGNITION, Tags.TEXT]
+)
+
 
 class SpeechToTextTransformers(TransformersBase):
     """
@@ -35,6 +42,7 @@ class SpeechToTextTransformers(TransformersBase):
     """
 
     TEXT_KEY = "text"
+    UIProperties = SpeechToTextTransformersUIProperties
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)
@@ -51,14 +59,13 @@ class SpeechToTextTransformers(TransformersBase):
             DataContainer: DataContainer including the transcribed audio.
         """
         for audio_packet in container.audios:
-            if self.instance_name not in audio_packet.modified_by_templates:
-                audio = audio_packet.content
-                audio = audio.astype(np.float32)
-                transcribed_text = self.pipeline(audio, **self.attributes.inference_kwargs)[self.TEXT_KEY]
-                transcribed_text_textpacket = TextPacket(
-                    content=transcribed_text,
-                    source=audio_packet.source,
-                )
-                self.logger.info(f"Speech-to-text transcription: {transcribed_text}")
-                container.texts.append(transcribed_text_textpacket)
+            audio = audio_packet.content
+            audio = audio.astype(np.float32)
+            transcribed_text = self.pipeline(audio, **self.attributes.inference_kwargs)[self.TEXT_KEY]
+            transcribed_text_textpacket = TextPacket(
+                content=transcribed_text,
+                source=audio_packet.source,
+            )
+            self.logger.info(f"Speech-to-text transcription: {transcribed_text}")
+            container.texts.append(transcribed_text_textpacket)
         return container

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/summarization_transformers.py

@@ -1,10 +1,15 @@
 # -*- coding: utf-8 -*-
 
 from sinapsis_core.data_containers.data_packet import DataContainer
-from sinapsis_core.template_base.base_models import TemplateAttributeType
+from sinapsis_core.template_base.base_models import OutputTypes, TemplateAttributeType
 
+from sinapsis_huggingface_transformers.helpers.tags import Tags
 from sinapsis_huggingface_transformers.templates.base_transformers import TransformersBase
 
+SummarizationTransformersUIProperties = TransformersBase.UIProperties
+SummarizationTransformersUIProperties.output_type = OutputTypes.TEXT
+SummarizationTransformersUIProperties.tags.extend([Tags.SUMMARIZATION, Tags.TEXT])
+
 
 class SummarizationTransformers(TransformersBase):
     """
@@ -34,6 +39,7 @@ class SummarizationTransformers(TransformersBase):
     """
 
     SUMMARY_TEXT_KEY = "summary_text"
+    UIProperties = SummarizationTransformersUIProperties
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/text_to_speech_transformers.py

@@ -3,14 +3,19 @@
 import numpy as np
 import torch
 from sinapsis_core.data_containers.data_packet import AudioPacket, DataContainer, TextPacket
-from sinapsis_core.template_base.base_models import TemplateAttributeType
+from sinapsis_core.template_base.base_models import OutputTypes, TemplateAttributeType
 
 from sinapsis_huggingface_transformers.helpers import sentences_to_n_words, split_text_into_sentences
+from sinapsis_huggingface_transformers.helpers.tags import Tags
 from sinapsis_huggingface_transformers.templates.base_transformers import (
     TransformersBase,
     TransformersBaseAttributes,
 )
 
+TextToSpeechTransformersUIProperties = TransformersBase.UIProperties
+TextToSpeechTransformersUIProperties.output_type = OutputTypes.AUDIO
+TextToSpeechTransformersUIProperties.tags.extend([Tags.AUDIO, Tags.TEXT, Tags.TEXT_TO_SPEECH])
+
 
 class TextToSpeechAttributes(TransformersBaseAttributes):
     """Attributes for the TextToSpeech template.
@@ -61,6 +66,7 @@ class TextToSpeechTransformers(TransformersBase):
     """
 
     AttributesBaseModel = TextToSpeechAttributes
+    UIProperties = TextToSpeechTransformersUIProperties
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/packages/sinapsis_huggingface_transformers/src/sinapsis_huggingface_transformers/templates/translation_transformers.py

@@ -2,13 +2,18 @@
 
 
 from sinapsis_core.data_containers.data_packet import DataContainer
-from sinapsis_core.template_base.base_models import TemplateAttributeType
+from sinapsis_core.template_base.base_models import OutputTypes, TemplateAttributeType
 
+from sinapsis_huggingface_transformers.helpers.tags import Tags
 from sinapsis_huggingface_transformers.templates.base_transformers import (
     TransformersBase,
     TransformersBaseAttributes,
 )
 
+TranslationTransformersUIProperties = TransformersBase.UIProperties
+TranslationTransformersUIProperties.output_type = OutputTypes.TEXT
+TranslationTransformersUIProperties.tags.extend([Tags.LANGUAGE, Tags.TRANSLATION])
+
 
 class TranslationTransformersAttributes(TransformersBaseAttributes):
     """Attributes for the transformers pipeline translation task.
@@ -52,6 +57,7 @@ class TranslationTransformers(TransformersBase):
 
     AttributesBaseModel = TranslationTransformersAttributes
     TRANSLATION_TEXT_KEY = "translation_text"
+    UIProperties = TranslationTransformersUIProperties
 
     def __init__(self, attributes: TemplateAttributeType) -> None:
         super().__init__(attributes)

{sinapsis_huggingface-0.2.3 → sinapsis_huggingface-0.2.6}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "sinapsis-huggingface"
-version = "0.2.3"
+version = "0.2.6"
 description = "Package for HuggingFace-based templates"
 authors = [
     {name = "SinapsisAI", email = "dev@sinapsis.tech"},