Docs/update further

autointent/custom_types/__init__.py

@@ -1,3 +1,5 @@
+"""Types used throughout AutoIntent library."""
+
 from ._types import (
     FloatFromZeroToOne,
     LabelType,

autointent/generation/__init__.py

@@ -0,0 +1 @@
+"""Some generative methods for enriching training datasets."""

autointent/generation/chat_templates/__init__.py

@@ -0,0 +1,38 @@
+"""Chat templates used throughout :py:mod:`autointent.generation` module."""
+
+from ._abstract import AbstractEvolution
+from ._base_evolver import EvolutionChatTemplate
+from ._base_synthesizer import BaseSynthesizerTemplate
+from ._concrete import ConcreteEvolution
+from ._evolution_templates_schemas import Message, Role
+from ._formal import FormalEvolution
+from ._funny import FunnyEvolution
+from ._goofy import GoofyEvolution
+from ._informal import InformalEvolution
+from ._intent_descriptions import PromptDescription
+from ._reasoning import ReasoningEvolution
+from ._synthesizer_en import EnglishSynthesizerTemplate
+from ._synthesizer_ru import RussianSynthesizerTemplate
+
+EVOLUTION_NAMES = [evolution.name for evolution in EvolutionChatTemplate.__subclasses__()]
+
+EVOLUTION_MAPPING = {evolution.name: evolution() for evolution in EvolutionChatTemplate.__subclasses__()}
+
+__all__ = [
+    "EVOLUTION_MAPPING",
+    "EVOLUTION_NAMES",
+    "AbstractEvolution",
+    "BaseSynthesizerTemplate",
+    "ConcreteEvolution",
+    "EnglishSynthesizerTemplate",
+    "EvolutionChatTemplate",
+    "FormalEvolution",
+    "FunnyEvolution",
+    "GoofyEvolution",
+    "InformalEvolution",
+    "Message",
+    "PromptDescription",
+    "ReasoningEvolution",
+    "Role",
+    "RussianSynthesizerTemplate",
+]

autointent/generation/utterances/evolution/chat_templates/abstract.py → autointent/generation/chat_templates/_abstract.py

@@ -2,9 +2,8 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
-
-from .base import EvolutionChatTemplate
+from ._base_evolver import EvolutionChatTemplate
+from ._evolution_templates_schemas import Message, Role
 
 
 class AbstractEvolution(EvolutionChatTemplate):

autointent/generation/utterances/evolution/chat_templates/base.py → autointent/generation/chat_templates/_base_evolver.py

@@ -2,9 +2,10 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
 from autointent.schemas import Intent
 
+from ._evolution_templates_schemas import Message, Role
+
 
 class EvolutionChatTemplate:
     """Base class for chat templates for evolution augmentation."""

autointent/generation/utterances/basic/chat_templates/_base.py → autointent/generation/chat_templates/_base_synthesizer.py

@@ -6,9 +6,10 @@
 from typing import ClassVar
 
 from autointent import Dataset
-from autointent.generation.utterances.schemas import Message, Role
 from autointent.schemas import Intent
 
+from ._evolution_templates_schemas import Message, Role
+
 
 class BaseChatTemplate(ABC):
     """Base class."""

autointent/generation/utterances/evolution/chat_templates/concrete.py → autointent/generation/chat_templates/_concrete.py

@@ -2,9 +2,8 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
-
-from .base import EvolutionChatTemplate
+from ._base_evolver import EvolutionChatTemplate
+from ._evolution_templates_schemas import Message, Role
 
 
 class ConcreteEvolution(EvolutionChatTemplate):

autointent/generation/utterances/evolution/chat_templates/formal.py → autointent/generation/chat_templates/_formal.py

@@ -2,9 +2,8 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
-
-from .base import EvolutionChatTemplate
+from ._base_evolver import EvolutionChatTemplate
+from ._evolution_templates_schemas import Message, Role
 
 
 class FormalEvolution(EvolutionChatTemplate):

autointent/generation/utterances/evolution/chat_templates/funny.py → autointent/generation/chat_templates/_funny.py

@@ -2,9 +2,8 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
-
-from .base import EvolutionChatTemplate
+from ._base_evolver import EvolutionChatTemplate
+from ._evolution_templates_schemas import Message, Role
 
 
 class FunnyEvolution(EvolutionChatTemplate):

autointent/generation/utterances/evolution/chat_templates/goofy.py → autointent/generation/chat_templates/_goofy.py

@@ -2,9 +2,8 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
-
-from .base import EvolutionChatTemplate
+from ._base_evolver import EvolutionChatTemplate
+from ._evolution_templates_schemas import Message, Role
 
 
 class GoofyEvolution(EvolutionChatTemplate):

autointent/generation/utterances/evolution/chat_templates/informal.py → autointent/generation/chat_templates/_informal.py

@@ -2,9 +2,8 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
-
-from .base import EvolutionChatTemplate
+from ._base_evolver import EvolutionChatTemplate
+from ._evolution_templates_schemas import Message, Role
 
 
 class InformalEvolution(EvolutionChatTemplate):

autointent/generation/intents/prompts.py → autointent/generation/chat_templates/_intent_descriptions.py

@@ -1,4 +1,6 @@
-"""Prompt description."""
+"""Prompt description configuration."""
+
+from pydantic import BaseModel, field_validator
 
 PROMPT_DESCRIPTION = """
 Your task is to write a description of the intent.
@@ -55,3 +57,35 @@
 description:
 
 """
+
+
+class PromptDescription(BaseModel):
+    """Prompt description configuration."""
+
+    text: str = PROMPT_DESCRIPTION
+    """
+    The template for the prompt to generate descriptions for intents.
+    Should include placeholders for {intent_name} and {user_utterances}.
+    - `{intent_name}` will be replaced with the name of the intent.
+    - `{user_utterances}` will be replaced with the user utterances related to the intent.
+    - (optionally) `{regex_patterns}` will be replaced with the regular expressions that match user utterances.
+    """
+
+    @classmethod
+    @field_validator("text")
+    def check_valid_prompt(cls, value: str) -> str:
+        """Validate the prompt description template.
+
+        Args:
+            value: The prompt description template.
+
+        Returns:
+            The validated prompt description template.
+        """
+        if value.find("{intent_name}") == -1 or value.find("{user_utterances}") == -1:
+            text_error = (
+                "The 'prompt_description' template must properly "
+                "include {intent_name} and {user_utterances} placeholders."
+            )
+            raise ValueError(text_error)
+        return value

autointent/generation/utterances/evolution/chat_templates/reasoning.py → autointent/generation/chat_templates/_reasoning.py

@@ -2,9 +2,8 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
-
-from .base import EvolutionChatTemplate
+from ._base_evolver import EvolutionChatTemplate
+from ._evolution_templates_schemas import Message, Role
 
 
 class ReasoningEvolution(EvolutionChatTemplate):

autointent/generation/utterances/basic/chat_templates/_synthesizer_en.py → autointent/generation/chat_templates/_synthesizer_en.py

@@ -2,9 +2,8 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
-
-from ._base import BaseSynthesizerTemplate
+from ._base_synthesizer import BaseSynthesizerTemplate
+from ._evolution_templates_schemas import Message, Role
 
 
 class EnglishSynthesizerTemplate(BaseSynthesizerTemplate):

autointent/generation/utterances/basic/chat_templates/_synthesizer_ru.py → autointent/generation/chat_templates/_synthesizer_ru.py

@@ -2,9 +2,8 @@
 
 from typing import ClassVar
 
-from autointent.generation.utterances.schemas import Message, Role
-
-from ._base import BaseSynthesizerTemplate
+from ._base_synthesizer import BaseSynthesizerTemplate
+from ._evolution_templates_schemas import Message, Role
 
 
 class RussianSynthesizerTemplate(BaseSynthesizerTemplate):

autointent/generation/intents/__init__.py

@@ -0,0 +1,5 @@
+"""Generative methods for enriching intents' metadata."""
+
+from ._description_generation import generate_descriptions
+
+__all__ = ["generate_descriptions"]

autointent/generation/intents/description_generation.py → autointent/generation/intents/_description_generation.py

@@ -12,7 +12,7 @@
 from openai import AsyncOpenAI
 
 from autointent import Dataset
-from autointent.generation.intents.prompt_scheme import PromptDescription
+from autointent.generation.chat_templates import PromptDescription
 from autointent.schemas import Intent, Sample
 
 
@@ -59,9 +59,6 @@ async def create_intent_description(
                user_utterances, and regex_patterns.
         model_name: Identifier of the OpenAI model to use.
 
-    Returns:
-        Generated description of the intent.
-
     Raises:
         TypeError: If the model response is not a string.
     """
@@ -103,9 +100,6 @@ async def generate_intent_descriptions(
         prompt: Template for model prompt with placeholders for intent_name,
                user_utterances, and regex_patterns.
         model_name: Name of the OpenAI model to use.
-
-    Returns:
-        List of intents with updated descriptions.
     """
     tasks = []
     for intent in intents:
@@ -131,28 +125,27 @@ async def generate_intent_descriptions(
     return intents
 
 
-def enhance_dataset_with_descriptions(
+def generate_descriptions(
     dataset: Dataset,
     client: AsyncOpenAI,
-    prompt: PromptDescription,
-    model_name: str = "gpt-4o-mini",
+    model_name: str,
+    prompt: PromptDescription | None = None,
 ) -> Dataset:
-    """Enhance a dataset by adding generated descriptions to its intents.
+    """Add LLM-generated text descriptions to dataset's intents.
 
     Args:
         dataset: Dataset containing utterances and intents needing descriptions.
         client: OpenAI client for generating descriptions.
         prompt: Template for model prompt with placeholders for intent_name,
                user_utterances, and regex_patterns.
         model_name: OpenAI model identifier for generating descriptions.
-
-    Returns:
-        Dataset with enhanced intent descriptions.
     """
     samples = []
     for split in dataset.values():
         samples.extend([Sample(**sample) for sample in split.to_list()])
     intent_utterances = group_utterances_by_label(samples)
+    if prompt is None:
+        prompt = PromptDescription()
     dataset.intents = asyncio.run(
         generate_intent_descriptions(client, intent_utterances, dataset.intents, prompt, model_name),
     )

autointent/generation/utterances/__init__.py

@@ -1,31 +1,17 @@
+"""Generative methods for enriching dataset with synthetic samples."""
+
 from .balancer import DatasetBalancer
-from .basic import EnglishSynthesizerTemplate, RussianSynthesizerTemplate, UtteranceGenerator
+from .basic import UtteranceGenerator
 from .evolution import (
-    AbstractEvolution,
-    ConcreteEvolution,
-    EvolutionChatTemplate,
-    FormalEvolution,
-    FunnyEvolution,
-    GoofyEvolution,
     IncrementalUtteranceEvolver,
-    InformalEvolution,
-    ReasoningEvolution,
     UtteranceEvolver,
 )
 from .generator import Generator
 
 __all__ = [
-    "AbstractEvolution",
-    "ConcreteEvolution",
     "DatasetBalancer",
-    "EvolutionChatTemplate",
-    "FormalEvolution",
-    "FunnyEvolution",
     "Generator",
-    "GoofyEvolution",
     "IncrementalUtteranceEvolver",
-    "InformalEvolution",
-    "ReasoningEvolution",
     "UtteranceEvolver",
     "UtteranceGenerator",
 ]