ai-simple-engine-diffusion 0.0.3__tar.gz → 0.0.6__tar.gz

{ai_simple_engine_diffusion-0.0.3 → ai_simple_engine_diffusion-0.0.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ai-simple-engine-diffusion
-Version: 0.0.3
+Version: 0.0.6
 Summary: AI Simple Engine Diffusion Module
 License-File: LICENSE
 Author: danialcala94
@@ -9,7 +9,7 @@ Requires-Python: >=3.10,<3.12
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
-Requires-Dist: ai-simple-engine (>=0.2.8,<1.0.0)
+Requires-Dist: ai-simple-engine (>=0.2.9,<1.0.0)
 Requires-Dist: ai-simple-engine-common (>=0.0.3,<1.0.0)
 Description-Content-Type: text/markdown
 

{ai_simple_engine_diffusion-0.0.3 → ai_simple_engine_diffusion-0.0.6}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ai-simple-engine-diffusion"
-version = "0.0.3"
+version = "0.0.6"
 description = "AI Simple Engine Diffusion Module"
 authors = [
     {name = "danialcala94", email = "danielalcalavalera@gmail.com"}
@@ -10,7 +10,7 @@ requires-python = ">=3.10,<3.12"
 
 [tool.poetry.dependencies]
 # Mandatory
-ai-simple-engine = { version = ">=0.2.8,<1.0.0", optional = false }
+ai-simple-engine = { version = ">=0.2.9,<1.0.0", optional = false }
 ai-simple-engine-common = { version = ">=0.0.3,<1.0.0", optional = false }
 # Optional
 

{ai_simple_engine_diffusion-0.0.3 → ai_simple_engine_diffusion-0.0.6}/src/ai_simple_engine_diffusion/graph/operation/generate_image_from_latents.py

@@ -3,6 +3,7 @@ from ai_simple_engine.graph.operation.abstract.atomic_operation import AtomicOpe
 from ai_simple_engine.models.executor.registry.family_model_executor_registry import FamilyModelExecutorRegistry
 from ai_simple_engine.graph.input import Input
 from ai_simple_engine.graph.output import Output
+from ai_simple_engine.types.image import Image
 from ai_simple_engine.types.data_type.base import LOADED_MODEL, STRING, INT, FLOAT, IMAGE
 
 
@@ -59,5 +60,5 @@ class GenerateImageFromLatents(
         )
 
         return {
-            'image': image
+            'image': Image(image)
         }

{ai_simple_engine_diffusion-0.0.3 → ai_simple_engine_diffusion-0.0.6}/src/ai_simple_engine_diffusion/graph/operation/sample_latents.py

@@ -52,22 +52,15 @@ class SampleLatents(
             height = self.height
         )
 
-        shape = (
-            space.batch_size,
-            space.channels,
-            space.height,
-            space.width
-        )
-
         runtime = self.model.instance
 
         # Dynamically detect 'device' and 'dtype'
         device = next(runtime.unet.parameters()).device
         dtype = next(runtime.unet.parameters()).dtype
 
-        latents_torch = noise_generator.generate(
+        latents_torch = await noise_generator.generate(
             # space = space,
-            shape = shape,
+            shape = space.to_shape(),
             device = device,
             dtype = dtype,
             seed = self.seed

{ai_simple_engine_diffusion-0.0.3 → ai_simple_engine_diffusion-0.0.6}/src/ai_simple_engine_diffusion/model/executor/abstract/latent_diffusion_model_executor_abstract.py

@@ -88,7 +88,7 @@ class LatentDiffusionModelExecutorAbstract(
         negative_prompt: str,
         steps: int,
         guidance_scale: float
-    ):
+    ) -> 'torch.Tensor':
         """
         Generate an image with the given `prompt`,
         `negative_prompt` and `model`.

{ai_simple_engine_diffusion-0.0.3 → ai_simple_engine_diffusion-0.0.6}/src/ai_simple_engine_diffusion/model/info/diffusion_model_info.py

@@ -9,6 +9,11 @@ from dataclasses import dataclass
 class DiffusionModelInfo(
     ModelInfo
 ):
+    """
+    Information about the diffusion model to
+    be able to calculate the latent spaces,
+    the scheduler that must be used, etc.
+    """
 
     latent_channels: int
     vae_scale_factor: int
@@ -24,6 +29,11 @@ class DiffusionModelInfo(
         width: int,
         height: int
     ) -> LatentSpace:
+        """
+        Get a `LatentSpace` instance of the given `width`
+        and `height`, according to the model vae scale
+        factor and latent channels.
+        """
         return LatentSpace(
             batch_size = 1,
             width = width // self.vae_scale_factor,

{ai_simple_engine_diffusion-0.0.3 → ai_simple_engine_diffusion-0.0.6}/src/ai_simple_engine_diffusion/noise/noise_generator_abstract.py

@@ -6,6 +6,11 @@ from abc import ABC, abstractmethod
 class NoiseGenerator(
     ABC
 ):
+    """
+    Class to inject random statistical static into
+    data during training to be the starting point
+    to creating new images during generation.
+    """
 
     @abstractmethod
     async def generate(
@@ -15,7 +20,9 @@ class NoiseGenerator(
         seed: int
     ) -> Latents:
         """
-        Generate the initial latent noise.
+        Generate the initial latent noise for the
+        given latent `space` and using the `seed`
+        provided.
         """
         ...
 

ai_simple_engine_diffusion-0.0.6/src/ai_simple_engine_diffusion/prompt_embeddings/abstract.py

@@ -0,0 +1,9 @@
+class PromptEmbeddings:
+    """
+    The numerical, high-dimensional vectors that
+    represent text (the prompt). An AI model 
+    transforms words into tokens and then maps
+    those tokens to these vectors.
+    """
+    
+    pass

ai_simple_engine_diffusion-0.0.6/src/ai_simple_engine_diffusion/types/embeddings.py

@@ -0,0 +1,35 @@
+from dataclasses import dataclass
+
+
+@dataclass(frozen = True)
+class Embeddings:
+    """
+    Vectors of numbers that represent text in a
+    high-dimensional space. Usually used outside the
+    model for search, clustering, retrieval, etc.
+
+    When an AI converts the word 'King' into a list
+    of 1,536 numbers, or a photo into a numerical
+    array, those vectors are embeddings. They are
+    the data points used for semantic search,
+    retrieval, and classification.
+
+    These are the POINTS we plot in our latent
+    space.
+    """
+
+    instance: object
+
+
+"""
+I don't use 'torch.Tensor' here because this
+one is general and the implementation could
+be 'torch.Tensor' or could be not.
+
+Having a specific 'Latents' class allows us
+to avoid receiving 'Embeddings' in a method
+that is not expected, even if the 'Latents'
+and the 'Embeddings' have a 'torch' value
+inside. They are distinct concepts. same with
+'NoisePrediction'.
+"""

ai_simple_engine_diffusion-0.0.6/src/ai_simple_engine_diffusion/types/latent_space.py

@@ -0,0 +1,20 @@
+from dataclasses import dataclass
+
+
+@dataclass(frozen = True)
+class LatentSpace:
+    """
+    The space in which all the internal vectors
+    (embeddings) live, where they are plotted and
+    compared. The internal geometry created by the
+    model's parameters.
+
+    This is the CANVAS (space) in which we plot the
+    embeddings, and is only affected when tunning
+    parameters or training the model.
+    """
+
+    channels: int
+    width: int
+    height: int
+    batch_size: int = 1

{ai_simple_engine_diffusion-0.0.3 → ai_simple_engine_diffusion-0.0.6}/src/ai_simple_engine_diffusion/types/latents.py

@@ -3,6 +3,14 @@ from dataclasses import dataclass
 
 @dataclass(frozen = True)
 class Latents:
+    """
+    Latents are the raw, often compressed
+    representations of data actively flowing
+    through the neural network during encoding
+    and decoding. They represent a more general,
+    transient, and hidden layer of thought the
+    model uses to generate outputs.
+    """
     
     instance: object
 

{ai_simple_engine_diffusion-0.0.3 → ai_simple_engine_diffusion-0.0.6}/src/ai_simple_engine_diffusion/types/noise_prediction.py

@@ -3,6 +3,11 @@ from dataclasses import dataclass
 
 @dataclass(frozen = True)
 class NoisePrediction:
+    """
+    The estimation of data distortion that will
+    be used in machine learning to provide a
+    different result.
+    """
 
     instance: object
 

ai_simple_engine_diffusion-0.0.3/src/ai_simple_engine_diffusion/prompt_embeddings/abstract.py

@@ -1,3 +0,0 @@
-class PromptEmbeddings:
-    
-    pass

ai_simple_engine_diffusion-0.0.3/src/ai_simple_engine_diffusion/scheduler/abstract.py

@@ -1,57 +0,0 @@
-from ai_simple_engine_diffusion.scheduler.abstract import Scheduler
-
-
-class DiffusersScheduler(
-    Scheduler
-):
-
-    def __init__(
-        self,
-        scheduler
-    ):
-        self._scheduler = scheduler
-
-    @property
-    def init_noise_sigma(
-        self
-    ):
-        """
-        The `gamma` that must be used when starting
-        to create the noise.
-        """
-        return self._scheduler.init_noise_sigma
-
-    @property
-    def timesteps(
-        self
-    ):
-        return self._scheduler.timesteps
-
-    def set_timesteps(
-        self,
-        steps: int
-    ):
-        self._scheduler.set_timesteps(steps)
-
-    def scale_model_input(
-        self,
-        latents,
-        timestep
-    ):
-        return self._scheduler.scale_model_input(
-            latents,
-            timestep
-        )
-    
-    def step(
-        self,
-        *,
-        model_output,
-        timestep,
-        sample
-    ):
-        return self._scheduler.step(
-            model_output = model_output,
-            timestep = timestep,
-            sample = sample
-        )

ai_simple_engine_diffusion-0.0.3/src/ai_simple_engine_diffusion/types/embeddings.py

@@ -1,21 +0,0 @@
-from dataclasses import dataclass
-
-
-@dataclass(frozen = True)
-class Embeddings:
-
-    instance: object
-
-
-"""
-I don't use 'torch.Tensor' here because this
-one is general and the implementation could
-be 'torch.Tensor' or could be not.
-
-Having a specific 'Latents' class allows us
-to avoid receiving 'Embeddings' in a method
-that is not expected, even if the 'Latents'
-and the 'Embeddings' have a 'torch' value
-inside. They are distinct concepts. same with
-'NoisePrediction'.
-"""

ai_simple_engine_diffusion-0.0.3/src/ai_simple_engine_diffusion/types/latent_space.py

@@ -1,10 +0,0 @@
-from dataclasses import dataclass
-
-
-@dataclass(frozen = True)
-class LatentSpace:
-
-    channels: int
-    width: int
-    height: int
-    batch_size: int = 1