unique_toolkit 0.6.1__tar.gz → 0.6.4__tar.gz

{unique_toolkit-0.6.1 → unique_toolkit-0.6.4}/CHANGELOG.md

@@ -5,6 +5,19 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), 
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+
+## [0.6.3] - 2025-02-27
+- Simplified imports for services. `from unique_toolkit.language_model import LanguageModelService` -> `from unique_toolkit import LanguageModelService` to reduce number of import lines.
+
+## [0.6.3] - 2025-02-26
+- Add `builder` method to `LanguageModelMessages` class
+
+## [0.6.2] - 2025-02-25
+- Deprecate `LanguageModel` in favor of `LanguageModelInfo`
+- `LanguageModelTokenLimits` properties become mandatory, initialization allows 
+  - init with `token_limit` and `fraction_input` or `input_token_limit` and `output_token_limit`
+  - only `input_token_limit` and `output_token_limit` are members of model
+
 ## [0.6.1] - 2025-02-25
 - [BREAKING] `LanguageModelService.stream_complete` and `LanguageModelService.stream_complete_async` are now moved to `ChatService.stream_complete` and `ChatService.stream_complete_async`. Correspondingly `assistant_message_id` and `user_message_id` are removed from `LanguageModelService`.
 - Add `create_user_message` and `create_user_message_async` to `ChatService` (similar to `create_assistant_message` and `create_assistant_message_async`)

{unique_toolkit-0.6.1 → unique_toolkit-0.6.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: unique_toolkit
-Version: 0.6.1
+Version: 0.6.4
 Summary: 
 License: Proprietary
 Author: Martin Fadler
@@ -111,6 +111,19 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), 
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+
+## [0.6.3] - 2025-02-27
+- Simplified imports for services. `from unique_toolkit.language_model import LanguageModelService` -> `from unique_toolkit import LanguageModelService` to reduce number of import lines.
+
+## [0.6.3] - 2025-02-26
+- Add `builder` method to `LanguageModelMessages` class
+
+## [0.6.2] - 2025-02-25
+- Deprecate `LanguageModel` in favor of `LanguageModelInfo`
+- `LanguageModelTokenLimits` properties become mandatory, initialization allows 
+  - init with `token_limit` and `fraction_input` or `input_token_limit` and `output_token_limit`
+  - only `input_token_limit` and `output_token_limit` are members of model
+
 ## [0.6.1] - 2025-02-25
 - [BREAKING] `LanguageModelService.stream_complete` and `LanguageModelService.stream_complete_async` are now moved to `ChatService.stream_complete` and `ChatService.stream_complete_async`. Correspondingly `assistant_message_id` and `user_message_id` are removed from `LanguageModelService`.
 - Add `create_user_message` and `create_user_message_async` to `ChatService` (similar to `create_assistant_message` and `create_assistant_message_async`)

{unique_toolkit-0.6.1 → unique_toolkit-0.6.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "unique_toolkit"
-version = "0.6.1"
+version = "0.6.4"
 description = ""
 authors = [
     "Martin Fadler <martin.fadler@unique.ch>",

unique_toolkit-0.6.4/unique_toolkit/__init__.py

@@ -0,0 +1,17 @@
+# Re-export commonly used classes for easier imports
+from unique_toolkit.chat import ChatService
+from unique_toolkit.content import ContentService
+from unique_toolkit.embedding import EmbeddingService
+from unique_toolkit.language_model import LanguageModelMessages, LanguageModelService
+from unique_toolkit.short_term_memory import ShortTermMemoryService
+
+# You can add other classes you frequently use here as well
+
+__all__ = [
+    "LanguageModelService",
+    "LanguageModelMessages",
+    "ChatService",
+    "ContentService",
+    "EmbeddingService",
+    "ShortTermMemoryService",
+]

{unique_toolkit-0.6.1 → unique_toolkit-0.6.4}/unique_toolkit/evaluators/config.py

@@ -25,7 +25,7 @@ class EvaluationMetricConfig(BaseModel):
     enabled: bool = False
     name: EvaluationMetricName
     language_model: LanguageModel = LanguageModel(
-        LanguageModelName.AZURE_GPT_35_TURBO_0613
+        LanguageModelName.AZURE_GPT_35_TURBO_0125
     )
     custom_prompts: dict[str, str] = {}
     score_to_emoji: dict[str, str] = {}

{unique_toolkit-0.6.1 → unique_toolkit-0.6.4}/unique_toolkit/evaluators/context_relevancy/constants.py

@@ -23,7 +23,7 @@ context_relevancy_required_input_fields = [
 default_config = EvaluationMetricConfig(
     enabled=False,
     name=EvaluationMetricName.CONTEXT_RELEVANCY,
-    language_model=LanguageModel(LanguageModelName.AZURE_GPT_35_TURBO_0613),
+    language_model=LanguageModel(LanguageModelName.AZURE_GPT_35_TURBO_0125),
     score_to_emoji={"LOW": "🟢", "MEDIUM": "🟡", "HIGH": "🔴"},
     custom_prompts={
         SYSTEM_MSG_KEY: CONTEXT_RELEVANCY_METRIC_SYSTEM_MSG,

{unique_toolkit-0.6.1 → unique_toolkit-0.6.4}/unique_toolkit/language_model/builder.py

@@ -1,3 +1,5 @@
+from typing_extensions import Self
+
 from unique_toolkit.language_model import (
     LanguageModelAssistantMessage,
     LanguageModelMessage,
@@ -12,25 +14,38 @@ class MessagesBuilder:
     def __init__(self):
         self.messages: list[LanguageModelMessage] = []
 
-    def system_message_append(self, content):
-        """Appends a user message to the messages list."""
+    def system_message_append(self, content: str) -> Self:
+        """Appends a system message to the messages list."""
         message = LanguageModelSystemMessage(content=content)
         self.messages.append(message)
         return self  # Return self to allow method chaining
 
-    def user_message_append(self, content):
+    def user_message_append(self, content: str) -> Self:
         """Appends a user message to the messages list."""
         message = LanguageModelUserMessage(content=content)
         self.messages.append(message)
         return self  # Return self to allow method chaining
 
-    def assistant_message_append(self, content):
+    def image_message_append(self, content: str, images: list[str]) -> Self:
+        message = LanguageModelUserMessage(
+            content=[
+                {"type": "text", "text": content},
+                *[
+                    {"type": "image_url", "imageUrl": {"url": image}}
+                    for image in images
+                ],
+            ]
+        )
+        self.messages.append(message)
+        return self
+
+    def assistant_message_append(self, content: str) -> Self:
         """Appends an assistant message to the messages list."""
         message = LanguageModelAssistantMessage(content=content)
         self.messages.append(message)
         return self  # Return self to allow method chaining
 
-    def tool_message_append(self, name: str, tool_call_id: str, content: str):
+    def tool_message_append(self, name: str, tool_call_id: str, content: str) -> Self:
         """Appends a tool message to the messages list."""
         message = LanguageModelToolMessage(
             name=name, tool_call_id=tool_call_id, content=content

unique_toolkit-0.6.4/unique_toolkit/language_model/infos.py

@@ -0,0 +1,352 @@
+from datetime import date
+from enum import StrEnum
+from typing import ClassVar, Optional
+
+from pydantic import BaseModel
+from typing_extensions import deprecated
+
+from unique_toolkit.language_model.schemas import LanguageModelTokenLimits
+
+
+class LanguageModelName(StrEnum):
+    AZURE_GPT_35_TURBO_0125 = "AZURE_GPT_35_TURBO_0125"
+    AZURE_GPT_4_0613 = "AZURE_GPT_4_0613"
+    AZURE_GPT_4_TURBO_1106 = "AZURE_GPT_4_TURBO_1106"
+    AZURE_GPT_4_VISION_PREVIEW = "AZURE_GPT_4_VISION_PREVIEW"
+    AZURE_GPT_4_32K_0613 = "AZURE_GPT_4_32K_0613"
+    AZURE_GPT_4_TURBO_2024_0409 = "AZURE_GPT_4_TURBO_2024_0409"
+    AZURE_GPT_4o_2024_0513 = "AZURE_GPT_4o_2024_0513"
+    AZURE_GPT_4o_2024_0806 = "AZURE_GPT_4o_2024_0806"
+    AZURE_GPT_4o_MINI_2024_0718 = "AZURE_GPT_4o_MINI_2024_0718"
+
+
+class EncoderName(StrEnum):
+    O200K_BASE = "o200k_base"
+    CL100K_BASE = "cl100k_base"
+
+
+def get_encoder_name(model_name: LanguageModelName) -> Optional[EncoderName]:
+    LMN = LanguageModelName
+    match model_name:
+        case LMN.AZURE_GPT_35_TURBO_0125:
+            return EncoderName.CL100K_BASE
+        case (
+            LMN.AZURE_GPT_4_0613
+            | LMN.AZURE_GPT_4_TURBO_1106
+            | LMN.AZURE_GPT_4_VISION_PREVIEW
+            | LMN.AZURE_GPT_4_32K_0613
+            | LMN.AZURE_GPT_4_TURBO_2024_0409
+        ):
+            return EncoderName.CL100K_BASE
+        case (
+            LMN.AZURE_GPT_4o_2024_0513
+            | LMN.AZURE_GPT_4o_2024_0806
+            | LMN.AZURE_GPT_4o_MINI_2024_0718
+        ):
+            return EncoderName.O200K_BASE
+        case _:
+            print(f"{model_name} is not supported. Please add encoder information.")
+            return None
+
+
+class LanguageModelProvider(StrEnum):
+    AZURE = "AZURE"
+    CUSTOM = "CUSTOM"
+
+
+class LanguageModelInfo(BaseModel):
+    name: LanguageModelName | str
+    version: str
+    provider: LanguageModelProvider
+
+    encoder_name: Optional[EncoderName] = None
+    token_limits: Optional[LanguageModelTokenLimits] = None
+
+    info_cutoff_at: Optional[date] = None
+    published_at: Optional[date] = None
+    retirement_at: Optional[date] = None
+
+    deprecated_at: Optional[date] = None
+    retirement_text: Optional[str] = None
+
+    @classmethod
+    def from_name(cls, model_name: LanguageModelName):
+        match model_name:
+            case LanguageModelName.AZURE_GPT_35_TURBO_0125:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.AZURE,
+                    version="0125",
+                    encoder_name=get_encoder_name(model_name),
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=16385, token_limit_output=4096
+                    ),
+                    info_cutoff_at=date(2021, 9, 1),
+                    published_at=date(2023, 1, 25),
+                    retirement_at=date(5, 3, 31),
+                )
+            case LanguageModelName.AZURE_GPT_4_0613:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.AZURE,
+                    version="0613",
+                    encoder_name=get_encoder_name(model_name),
+                    token_limits=LanguageModelTokenLimits(token_limit=8192),
+                    info_cutoff_at=date(2021, 9, 1),
+                    published_at=date(2023, 6, 13),
+                    deprecated_at=date(2024, 10, 1),
+                    retirement_at=date(2025, 6, 6),
+                )
+            case LanguageModelName.AZURE_GPT_4_TURBO_1106:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.AZURE,
+                    version="1106-preview",
+                    encoder_name=get_encoder_name(model_name),
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=128000, token_limit_output=4096
+                    ),
+                    info_cutoff_at=date(2023, 4, 1),
+                    published_at=date(2023, 11, 6),
+                )
+            case LanguageModelName.AZURE_GPT_4_VISION_PREVIEW:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.AZURE,
+                    version="vision-preview",
+                    encoder_name=get_encoder_name(model_name),
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=128000, token_limit_output=4096
+                    ),
+                    info_cutoff_at=date(2023, 4, 1),
+                    published_at=date(2023, 11, 6),
+                )
+            case LanguageModelName.AZURE_GPT_4_32K_0613:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.AZURE,
+                    version="1106-preview",
+                    encoder_name=get_encoder_name(model_name),
+                    token_limits=LanguageModelTokenLimits(token_limit=32768),
+                    info_cutoff_at=date(2021, 9, 1),
+                    published_at=date(2023, 6, 13),
+                    deprecated_at=date(2024, 10, 1),
+                    retirement_at=date(2025, 6, 6),
+                )
+            case LanguageModelName.AZURE_GPT_4_TURBO_2024_0409:
+                return cls(
+                    name=model_name,
+                    encoder_name=get_encoder_name(model_name),
+                    provider=LanguageModelProvider.AZURE,
+                    version="turbo-2024-04-09",
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=128000, token_limit_output=4096
+                    ),
+                    info_cutoff_at=date(2023, 12, 1),
+                    published_at=date(2024, 4, 9),
+                )
+            case LanguageModelName.AZURE_GPT_4o_2024_0513:
+                return cls(
+                    name=model_name,
+                    encoder_name=get_encoder_name(model_name),
+                    provider=LanguageModelProvider.AZURE,
+                    version="2024-05-13",
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=128000, token_limit_output=4096
+                    ),
+                    info_cutoff_at=date(2023, 10, 1),
+                    published_at=date(2024, 5, 13),
+                )
+            case LanguageModelName.AZURE_GPT_4o_2024_0806:
+                return cls(
+                    name=model_name,
+                    encoder_name=get_encoder_name(model_name),
+                    provider=LanguageModelProvider.AZURE,
+                    version="2024-08-06",
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=128000, token_limit_output=16384
+                    ),
+                    info_cutoff_at=date(2023, 10, 1),
+                    published_at=date(2024, 8, 6),
+                )
+            case LanguageModelName.AZURE_GPT_4o_MINI_2024_0718:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.AZURE,
+                    version="2024-07-18",
+                    encoder_name=get_encoder_name(model_name),
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=128000, token_limit_output=16384
+                    ),
+                    info_cutoff_at=date(2023, 10, 1),
+                    published_at=date(2024, 7, 18),
+                )
+            case _:
+                if isinstance(model_name, LanguageModelName):
+                    raise ValueError(
+                        f"{model_name} is not supported. Please add model information in toolkit."
+                    )
+
+                return cls(
+                    name=model_name,
+                    version="custom",
+                    provider=LanguageModelProvider.CUSTOM,
+                )
+
+    @property
+    def display_name(self) -> str:
+        """
+        Returns the name of the model as a string.
+        """
+
+        if isinstance(self.name, LanguageModelName):
+            return self.name.value
+        else:
+            return self.name
+
+
+@deprecated(
+    """
+Use `LanguageModelInfo` instead of `LanguageModel`
+"""
+)
+class LanguageModel:
+    _info: ClassVar[LanguageModelInfo]
+
+    def __init__(self, model_name: LanguageModelName | str):
+        self._model_info = self.get_model_info(model_name)
+
+    @property
+    def info(self) -> LanguageModelInfo:
+        """
+        Returns all infos about the model:
+        - name
+        - version
+        - provider
+        - encoder_name
+        - token_limits
+        - info_cutoff_at
+        - published_at
+        - retirement_at
+        - deprecated_at
+        - retirement_text
+        """
+        return self._model_info
+
+    @property
+    def name(self) -> LanguageModelName | str:
+        """
+        Returns the LanguageModelName of the model or the name string when it is a custom / not defined model.
+        """
+        return self._model_info.name
+
+    @property
+    def display_name(self) -> str:
+        """
+        Returns the name of the model as a string.
+        """
+        if isinstance(self._model_info.name, LanguageModelName):
+            return self._model_info.name.name
+        else:
+            return self._model_info.name
+
+    @property
+    def version(self) -> Optional[str]:
+        """
+        Returns the version of the model.
+        """
+        return self._model_info.version
+
+    @property
+    def encoder_name(self) -> Optional[EncoderName]:
+        """
+        Returns the encoder_name used for the model.
+        """
+        return self._model_info.encoder_name
+
+    @property
+    def token_limit(self) -> Optional[int]:
+        """
+        Returns the maximum number of tokens for the model.
+        """
+        if self._model_info.token_limits:
+            return self._model_info.token_limits.token_limit
+
+    @property
+    def token_limit_input(self) -> Optional[int]:
+        """
+        Returns the maximum number of input tokens for the model.
+        """
+        if self._model_info.token_limits:
+            return self._model_info.token_limits.token_limit_input
+
+    @property
+    def token_limit_output(self) -> Optional[int]:
+        """
+        Returns the maximum number of output tokens for the model.
+        """
+        if self._model_info.token_limits:
+            return self._model_info.token_limits.token_limit_output
+
+    @property
+    def info_cutoff_at(self) -> Optional[date]:
+        """
+        Returns the date the model was last updated.
+        """
+        return self._model_info.info_cutoff_at
+
+    @property
+    def published_at(self) -> Optional[date]:
+        """
+        Returns the date the model was published.
+        """
+        return self._model_info.published_at
+
+    @property
+    def retirement_at(self) -> Optional[date]:
+        """
+        Returns the date the model will be retired.
+        """
+        return self._model_info.retirement_at
+
+    @property
+    def deprecated_at(self) -> Optional[date]:
+        """
+        Returns the date the model was deprecated.
+        """
+        return self._model_info.deprecated_at
+
+    @property
+    def retirement_text(self) -> Optional[str]:
+        """
+        Returns the text that will be displayed when the model is retired.
+        """
+        return self._model_info.retirement_text
+
+    @property
+    def provider(self) -> LanguageModelProvider:
+        """
+        Returns the provider of the model.
+        """
+        return self._model_info.provider
+
+    @classmethod
+    def get_model_info(cls, model_name: LanguageModelName | str) -> LanguageModelInfo:
+        if isinstance(model_name, LanguageModelName):
+            return LanguageModelInfo.from_name(model_name)
+
+        return LanguageModelInfo(
+            name=model_name,
+            version="custom",
+            provider=LanguageModelProvider.CUSTOM,
+        )
+
+    @classmethod
+    def list_models(cls) -> list[LanguageModelInfo]:
+        """
+        Returns a list of the infos of all available models.
+        """
+
+        return [
+            LanguageModelInfo.from_name(model_name=name) for name in LanguageModelName
+        ]

{unique_toolkit-0.6.1 → unique_toolkit-0.6.4}/unique_toolkit/language_model/schemas.py

@@ -9,11 +9,13 @@ from pydantic import (
     BaseModel,
     ConfigDict,
     Field,
+    PrivateAttr,
     RootModel,
     field_validator,
     model_serializer,
     model_validator,
 )
+from typing_extensions import deprecated
 
 from unique_toolkit.language_model.utils import format_message
 
@@ -163,6 +165,14 @@ class LanguageModelMessages(RootModel):
     def __getitem__(self, item):
         return self.root[item]
 
+    def builder(self):
+        """Returns a MessagesBuilder instance pre-populated with existing messages."""
+        from unique_toolkit.language_model.builder import MessagesBuilder
+
+        builder = MessagesBuilder()
+        builder.messages = self.root.copy()  # Start with existing messages
+        return builder
+
 
 class LanguageModelCompletionChoice(BaseModel):
     model_config = model_config
@@ -205,30 +215,50 @@ class LanguageModelStreamResponse(BaseModel):
 
 
 class LanguageModelTokenLimits(BaseModel):
-    token_limit: Optional[int] = None
-    token_limit_input: Optional[int] = None
-    token_limit_output: Optional[int] = None
-
-    fraction_input: float = Field(default=0.4, le=1, ge=0)
-
-    @model_validator(mode="after")
-    def check_required_fields(self):
-        # Best case input and output is determined
-        if self.token_limit_input and self.token_limit_output:
-            self.token_limit = self.token_limit_input + self.token_limit_output
-            self.fraction_input = self.token_limit_input / self.token_limit
-            return self
-
-        # Deal with case where only token_limit and optional fraction_input is given
-        if self.token_limit:
-            if not self.fraction_input:
-                self.fraction_input = 0.4
-
-            self.token_limit_input = math.floor(self.fraction_input * self.token_limit)
-            self.token_limit_output = math.floor(
-                (1 - self.fraction_input) * self.token_limit
-            )
-            return self
+    token_limit_input: int
+    token_limit_output: int
+
+    _fraction_adaptable = PrivateAttr(default=False)
+
+    @property
+    @deprecated("""
+    Deprecated: Use the more specific `token_limit_input` and `token_limit_output` instead.
+    """)
+    def token_limit(self):
+        return self.token_limit_input + self.token_limit_output
+
+    @token_limit.setter
+    @deprecated("""
+    Deprecated: Token limit can only be reduced
+    """)
+    def token_limit(self, token_limit, fraction_input=0.4):
+        if self.token_limit > token_limit:
+            self.token_limit_input = math.floor(fraction_input * token_limit)
+            self.token_limit_output = math.floor((1 - fraction_input) * token_limit)
+            self._fraction_adaptable = True
+
+    def adapt_fraction(self, fraction_input: float):
+        if self._fraction_adaptable:
+            token_limit = self.token_limit_input + self.token_limit_output
+            self.token_limit_input = math.floor(fraction_input * token_limit)
+            self.token_limit_output = math.floor((1 - fraction_input) * token_limit)
+
+    @model_validator(mode="before")
+    def check_required_fields(cls, data):
+        if isinstance(data, dict):
+            if {"token_limit_input", "token_limit_output"}.issubset(data.keys()):
+                return data
+
+            if {"token_limit"}.issubset(data.keys()):
+                token_limit = data.get("token_limit")
+                fraction_input = data.get("fraction_input", 0.4)
+
+                data["token_limit_input"] = math.floor(fraction_input * token_limit)
+                data["token_limit_output"] = math.floor(
+                    (1 - fraction_input) * token_limit
+                )
+                data["_fraction_adaptpable"] = True
+            return data
 
         raise ValueError(
             'Either "token_limit_input" and "token_limit_output" must be provided together, or "token_limit" must be provided.'

unique_toolkit-0.6.1/unique_toolkit/language_model/infos.py

@@ -1,413 +0,0 @@
-from datetime import date
-from enum import StrEnum
-from typing import ClassVar, Optional, Type, cast
-
-from pydantic import BaseModel
-
-from unique_toolkit.language_model.schemas import LanguageModelTokenLimits
-
-
-class LanguageModelName(StrEnum):
-    AZURE_GPT_35_TURBO_0613 = "AZURE_GPT_35_TURBO_0613"
-    AZURE_GPT_35_TURBO = "AZURE_GPT_35_TURBO"
-    AZURE_GPT_35_TURBO_16K = "AZURE_GPT_35_TURBO_16K"
-    AZURE_GPT_35_TURBO_0125 = "AZURE_GPT_35_TURBO_0125"
-    AZURE_GPT_4_0613 = "AZURE_GPT_4_0613"
-    AZURE_GPT_4_TURBO_1106 = "AZURE_GPT_4_TURBO_1106"
-    AZURE_GPT_4_VISION_PREVIEW = "AZURE_GPT_4_VISION_PREVIEW"
-    AZURE_GPT_4_32K_0613 = "AZURE_GPT_4_32K_0613"
-    AZURE_GPT_4_TURBO_2024_0409 = "AZURE_GPT_4_TURBO_2024_0409"
-    AZURE_GPT_4o_2024_0513 = "AZURE_GPT_4o_2024_0513"
-    AZURE_GPT_4o_2024_0806 = "AZURE_GPT_4o_2024_0806"
-    AZURE_GPT_4o_MINI_2024_0718 = "AZURE_GPT_4o_MINI_2024_0718"
-
-
-class EncoderName(StrEnum):
-    O200K_BASE = "o200k_base"
-    CL100K_BASE = "cl100k_base"
-
-
-def get_encoder_name(model_name: LanguageModelName) -> Optional[EncoderName]:
-    LMN = LanguageModelName
-    match model_name:
-        case (
-            LMN.AZURE_GPT_35_TURBO
-            | LMN.AZURE_GPT_35_TURBO_16K
-            | LMN.AZURE_GPT_35_TURBO_0613
-            | LMN.AZURE_GPT_35_TURBO_0125
-        ):
-            return EncoderName.CL100K_BASE
-        case (
-            LMN.AZURE_GPT_4_0613
-            | LMN.AZURE_GPT_4_TURBO_1106
-            | LMN.AZURE_GPT_4_VISION_PREVIEW
-            | LMN.AZURE_GPT_4_32K_0613
-            | LMN.AZURE_GPT_4_TURBO_2024_0409
-        ):
-            return EncoderName.CL100K_BASE
-        case (
-            LMN.AZURE_GPT_4o_2024_0513
-            | LMN.AZURE_GPT_4o_2024_0806
-            | LMN.AZURE_GPT_4o_MINI_2024_0718
-        ):
-            return EncoderName.O200K_BASE
-        case _:
-            print(f"{model_name} is not supported. Please add encoder information.")
-            return None
-
-
-class LanguageModelProvider(StrEnum):
-    AZURE = "AZURE"
-    CUSTOM = "CUSTOM"
-
-
-class LanguageModelInfo(BaseModel):
-    name: LanguageModelName | str
-    version: str
-    provider: LanguageModelProvider
-
-    encoder_name: Optional[EncoderName] = None
-    token_limits: Optional[LanguageModelTokenLimits] = None
-
-    info_cutoff_at: Optional[date] = None
-    published_at: Optional[date] = None
-    retirement_at: Optional[date] = None
-
-    deprecated_at: Optional[date] = None
-    retirement_text: Optional[str] = None
-
-
-class LanguageModel:
-    _info: ClassVar[LanguageModelInfo]
-
-    def __init__(self, model_name: LanguageModelName | str):
-        self._model_info = self.get_model_info(model_name)
-
-    @property
-    def info(self) -> LanguageModelInfo:
-        """
-        Returns all infos about the model:
-        - name
-        - version
-        - provider
-        - encoder_name
-        - token_limits
-        - info_cutoff_at
-        - published_at
-        - retirement_at
-        - deprecated_at
-        - retirement_text
-        """
-        return self._model_info
-
-    @property
-    def name(self) -> LanguageModelName | str:
-        """
-        Returns the LanguageModelName of the model or the name string when it is a custom / not defined model.
-        """
-        return self._model_info.name
-
-    @property
-    def display_name(self) -> str:
-        """
-        Returns the name of the model as a string.
-        """
-        if isinstance(self._model_info.name, LanguageModelName):
-            return self._model_info.name.name
-        else:
-            return self._model_info.name
-
-    @property
-    def version(self) -> Optional[str]:
-        """
-        Returns the version of the model.
-        """
-        return self._model_info.version
-
-    @property
-    def encoder_name(self) -> Optional[EncoderName]:
-        """
-        Returns the encoder_name used for the model.
-        """
-        return self._model_info.encoder_name
-
-    @property
-    def token_limit(self) -> Optional[int]:
-        """
-        Returns the maximum number of tokens for the model.
-        """
-        if self._model_info.token_limits:
-            return self._model_info.token_limits.token_limit
-
-    @property
-    def token_limit_input(self) -> Optional[int]:
-        """
-        Returns the maximum number of input tokens for the model.
-        """
-        if self._model_info.token_limits:
-            return self._model_info.token_limits.token_limit_input
-
-    @property
-    def token_limit_output(self) -> Optional[int]:
-        """
-        Returns the maximum number of output tokens for the model.
-        """
-        if self._model_info.token_limits:
-            return self._model_info.token_limits.token_limit_output
-
-    @property
-    def info_cutoff_at(self) -> Optional[date]:
-        """
-        Returns the date the model was last updated.
-        """
-        return self._model_info.info_cutoff_at
-
-    @property
-    def published_at(self) -> Optional[date]:
-        """
-        Returns the date the model was published.
-        """
-        return self._model_info.published_at
-
-    @property
-    def retirement_at(self) -> Optional[date]:
-        """
-        Returns the date the model will be retired.
-        """
-        return self._model_info.retirement_at
-
-    @property
-    def deprecated_at(self) -> Optional[date]:
-        """
-        Returns the date the model was deprecated.
-        """
-        return self._model_info.deprecated_at
-
-    @property
-    def retirement_text(self) -> Optional[str]:
-        """
-        Returns the text that will be displayed when the model is retired.
-        """
-        return self._model_info.retirement_text
-
-    @property
-    def provider(self) -> LanguageModelProvider:
-        """
-        Returns the provider of the model.
-        """
-        return self._model_info.provider
-
-    @classmethod
-    def get_model_info(cls, model_name: LanguageModelName | str) -> LanguageModelInfo:
-        if not model_name:
-            raise ValueError("Model name must be provided to get the model info.")
-
-        for subclass in cls.__subclasses__():
-            if hasattr(subclass, "info") and subclass._info.name == model_name:
-                # TODO find alternative solution for warning
-                # if subclass._info.retirement_at:
-                #     warning_text = f"WARNING: {subclass._info.name} will be retired on {subclass._info.retirement_at.isoformat()} and from then on not accessible anymore. {subclass._info.retirement_text}"
-                #     print(warning_text)
-                #     warnings.warn(warning_text, DeprecationWarning, stacklevel=2)
-                return subclass._info
-
-        return LanguageModelInfo(
-            name=model_name,
-            version="custom",
-            provider=LanguageModelProvider.CUSTOM,
-        )
-
-    @classmethod
-    def list_models(cls) -> list[LanguageModelInfo]:
-        """
-        Returns a list of the infos of all available models.
-        """
-
-        return [
-            cast(LanguageModelInfo, subclass._info)
-            for subclass in cls.__subclasses__()
-            if hasattr(subclass, "_info")
-        ]
-
-
-def create_language_model(
-    name: LanguageModelName,
-    version: str,
-    provider: LanguageModelProvider,
-    info_cutoff_at: date,
-    published_at: date,
-    encoder_name: Optional[EncoderName] = None,
-    token_limit: Optional[int] = None,
-    token_limit_input: Optional[int] = None,
-    token_limit_output: Optional[int] = None,
-    retirement_at: Optional[date] = None,
-    deprecated_at: Optional[date] = None,
-    retirement_text: Optional[str] = None,
-) -> Type[LanguageModel]:
-    info = LanguageModelInfo(
-        name=name,
-        version=version,
-        provider=provider,
-        encoder_name=encoder_name,
-        token_limits=LanguageModelTokenLimits(
-            token_limit=token_limit,
-            token_limit_input=token_limit_input,
-            token_limit_output=token_limit_output,
-        ),
-        info_cutoff_at=info_cutoff_at,
-        published_at=published_at,
-        retirement_at=retirement_at,
-        deprecated_at=deprecated_at,
-        retirement_text=retirement_text,
-    )
-
-    class Model(LanguageModel):
-        _info = info
-
-    return Model
-
-
-############################################################################################################
-# Define the models here
-############################################################################################################
-
-
-AzureGpt35Turbo0613 = create_language_model(
-    name=LanguageModelName.AZURE_GPT_35_TURBO_0613,
-    provider=LanguageModelProvider.AZURE,
-    version="0613",
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_35_TURBO_0613),
-    token_limit=8192,
-    info_cutoff_at=date(2021, 9, 1),
-    published_at=date(2023, 6, 13),
-    retirement_at=date(2025, 2, 13),
-)
-
-AzureGpt35Turbo = create_language_model(
-    name=LanguageModelName.AZURE_GPT_35_TURBO,
-    provider=LanguageModelProvider.AZURE,
-    version="0301",
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_35_TURBO),
-    token_limit=4096,
-    info_cutoff_at=date(2021, 9, 1),
-    published_at=date(2023, 3, 1),
-    retirement_at=date(2025, 2, 13),
-)
-
-
-AzureGpt35Turbo16k = create_language_model(
-    name=LanguageModelName.AZURE_GPT_35_TURBO_16K,
-    provider=LanguageModelProvider.AZURE,
-    version="0613",
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_35_TURBO_16K),
-    token_limit=16382,
-    info_cutoff_at=date(2021, 9, 1),
-    published_at=date(2023, 6, 13),
-    retirement_at=date(2025, 2, 13),
-)
-
-
-AzureGpt35Turbo0125 = create_language_model(
-    name=LanguageModelName.AZURE_GPT_35_TURBO_0125,
-    provider=LanguageModelProvider.AZURE,
-    version="0125",
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_35_TURBO_0125),
-    token_limit_input=16385,
-    token_limit_output=4096,
-    info_cutoff_at=date(2021, 9, 1),
-    published_at=date(2023, 1, 25),
-    retirement_at=date(5, 3, 31),
-)
-
-
-AzureGpt40613 = create_language_model(
-    name=LanguageModelName.AZURE_GPT_4_0613,
-    provider=LanguageModelProvider.AZURE,
-    version="0613",
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_4_0613),
-    token_limit=8192,
-    info_cutoff_at=date(2021, 9, 1),
-    published_at=date(2023, 6, 13),
-    deprecated_at=date(2024, 10, 1),
-    retirement_at=date(2025, 6, 6),
-)
-
-
-AzureGpt4Turbo1106 = create_language_model(
-    name=LanguageModelName.AZURE_GPT_4_TURBO_1106,
-    provider=LanguageModelProvider.AZURE,
-    version="1106-preview",
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_4_TURBO_1106),
-    token_limit_input=128000,
-    token_limit_output=4096,
-    info_cutoff_at=date(2023, 4, 1),
-    published_at=date(2023, 11, 6),
-)
-
-
-AzureGpt4VisionPreview = create_language_model(
-    name=LanguageModelName.AZURE_GPT_4_VISION_PREVIEW,
-    provider=LanguageModelProvider.AZURE,
-    version="vision-preview",
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_4_VISION_PREVIEW),
-    token_limit_input=128000,
-    token_limit_output=4096,
-    info_cutoff_at=date(2023, 4, 1),
-    published_at=date(2023, 11, 6),
-)
-
-AzureGpt432k0613 = create_language_model(
-    name=LanguageModelName.AZURE_GPT_4_32K_0613,
-    provider=LanguageModelProvider.AZURE,
-    version="1106-preview",
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_4_32K_0613),
-    token_limit=32768,
-    info_cutoff_at=date(2021, 9, 1),
-    published_at=date(2023, 6, 13),
-    deprecated_at=date(2024, 10, 1),
-    retirement_at=date(2025, 6, 6),
-)
-
-AzureGpt4Turbo20240409 = create_language_model(
-    name=LanguageModelName.AZURE_GPT_4_TURBO_2024_0409,
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_4_TURBO_2024_0409),
-    provider=LanguageModelProvider.AZURE,
-    version="turbo-2024-04-09",
-    token_limit_input=128000,
-    token_limit_output=4096,
-    info_cutoff_at=date(2023, 12, 1),
-    published_at=date(2024, 4, 9),
-)
-
-AzureGpt4o20240513 = create_language_model(
-    name=LanguageModelName.AZURE_GPT_4o_2024_0513,
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_4o_2024_0513),
-    provider=LanguageModelProvider.AZURE,
-    version="2024-05-13",
-    token_limit_input=128000,
-    token_limit_output=4096,
-    info_cutoff_at=date(2023, 10, 1),
-    published_at=date(2024, 5, 13),
-)
-
-AzureGpt4o20240806 = create_language_model(
-    name=LanguageModelName.AZURE_GPT_4o_2024_0806,
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_4o_2024_0806),
-    provider=LanguageModelProvider.AZURE,
-    version="2024-08-06",
-    token_limit_input=128000,
-    token_limit_output=16384,
-    info_cutoff_at=date(2023, 10, 1),
-    published_at=date(2024, 8, 6),
-)
-
-AzureGpt4oMini20240718 = create_language_model(
-    name=LanguageModelName.AZURE_GPT_4o_MINI_2024_0718,
-    provider=LanguageModelProvider.AZURE,
-    version="2024-07-18",
-    encoder_name=get_encoder_name(LanguageModelName.AZURE_GPT_4o_MINI_2024_0718),
-    token_limit_input=128000,
-    token_limit_output=16384,
-    info_cutoff_at=date(2023, 10, 1),
-    published_at=date(2024, 7, 18),
-)