not-again-ai 0.14.0__py3-none-any.whl → 0.16.0__py3-none-any.whl

not_again_ai/llm/prompting/providers/openai_tiktoken.py

@@ -0,0 +1,122 @@
+from collections.abc import Collection, Set
+from typing import Literal
+
+from loguru import logger
+import tiktoken
+
+from not_again_ai.llm.chat_completion.types import MessageT
+from not_again_ai.llm.prompting.types import BaseTokenizer
+
+
+class TokenizerOpenAI(BaseTokenizer):
+    def __init__(
+        self,
+        model: str,
+        provider: str = "openai",
+        allowed_special: Literal["all"] | Set[str] | None = None,
+        disallowed_special: Literal["all"] | Collection[str] | None = None,
+    ):
+        self.model = model
+        self.provider = provider
+        self.allowed_special = allowed_special
+        self.disallowed_special = disallowed_special
+
+        self.init_tokenizer(model, provider, allowed_special, disallowed_special)
+
+    def init_tokenizer(
+        self,
+        model: str,
+        provider: str = "openai",
+        allowed_special: Literal["all"] | Set[str] | None = None,
+        disallowed_special: Literal["all"] | Collection[str] | None = None,
+    ) -> None:
+        try:
+            self.encoding = tiktoken.encoding_for_model(model)
+        except KeyError:
+            default_encoding = "o200k_base"
+            logger.warning(f"Model {model} not found. Using {default_encoding} encoding.")
+            self.encoding = tiktoken.get_encoding(default_encoding)
+
+        # Set defaults if not provided
+        if not allowed_special:
+            self.allowed_special = set()
+        if not disallowed_special:
+            self.disallowed_special = ()
+
+    def truncate_str(self, text: str, max_len: int) -> str:
+        tokens = self.encoding.encode(
+            text,
+            allowed_special=self.allowed_special if self.allowed_special is not None else set(),
+            disallowed_special=self.disallowed_special if self.disallowed_special is not None else (),
+        )
+        if len(tokens) > max_len:
+            tokens = tokens[:max_len]
+            truncated_text = self.encoding.decode(tokens)
+            return truncated_text
+        else:
+            return text
+
+    def num_tokens_in_str(self, text: str) -> int:
+        return len(
+            self.encoding.encode(
+                text,
+                allowed_special=self.allowed_special if self.allowed_special is not None else set(),
+                disallowed_special=self.disallowed_special if self.disallowed_special is not None else (),
+            )
+        )
+
+    def num_tokens_in_messages(self, messages: list[MessageT]) -> int:
+        if self.model in {
+            "gpt-3.5-turbo-0613",
+            "gpt-3.5-turbo-16k-0613",
+            "gpt-3.5-turbo-1106",
+            "gpt-3.5-turbo-0125",
+            "gpt-4-0314",
+            "gpt-4-32k-0314",
+            "gpt-4-0613",
+            "gpt-4-32k-0613",
+            "gpt-4-1106-preview",
+            "gpt-4-turbo-preview",
+            "gpt-4-0125-preview",
+            "gpt-4-turbo",
+            "gpt-4-turbo-2024-04-09",
+            "gpt-4o",
+            "gpt-4o-2024-05-13",
+            "gpt-4o-2024-08-06",
+            "gpt-4o-2024-11-20",
+            "gpt-4o-mini",
+            "gpt-4o-mini-2024-07-18",
+            "o1",
+            "o1-2024-12-17",
+            "o1-mini",
+            "o1-mini-2024-09-12",
+            "o1-preview",
+            "o1-preview-2024-09-12",
+        }:
+            tokens_per_message = 3  # every message follows <|start|>{role/name}\n{content}<|end|>\n
+            tokens_per_name = 1  # if there's a name, the role is omitted
+        elif self.model == "gpt-3.5-turbo-0301":
+            tokens_per_message = 4
+            tokens_per_name = -1
+        else:
+            logger.warning(f"Model {self.model} not supported. Assuming gpt-4o encoding.")
+            tokens_per_message = 3
+            tokens_per_name = 1
+
+        num_tokens = 0
+        for message in messages:
+            num_tokens += tokens_per_message
+            message_dict = message.model_dump(exclude_none=True)
+            for key, value in message_dict.items():
+                if isinstance(value, str):
+                    num_tokens += len(
+                        self.encoding.encode(
+                            value,
+                            allowed_special=self.allowed_special if self.allowed_special is not None else set(),
+                            disallowed_special=self.disallowed_special if self.disallowed_special is not None else (),
+                        )
+                    )
+                    if key == "name":
+                        num_tokens += tokens_per_name
+        num_tokens += 3
+        return num_tokens

not_again_ai/llm/prompting/types.py

@@ -0,0 +1,43 @@
+from abc import ABC, abstractmethod
+from collections.abc import Collection, Set
+from typing import Literal
+
+from not_again_ai.llm.chat_completion.types import MessageT
+
+
+class BaseTokenizer(ABC):
+    def __init__(
+        self,
+        model: str,
+        provider: str,
+        allowed_special: Literal["all"] | Set[str] | None = None,
+        disallowed_special: Literal["all"] | Collection[str] | None = None,
+    ):
+        self.model = model
+        self.provider = provider
+        self.allowed_special = allowed_special
+        self.disallowed_special = disallowed_special
+
+        self.init_tokenizer(model, provider, allowed_special, disallowed_special)
+
+    @abstractmethod
+    def init_tokenizer(
+        self,
+        model: str,
+        provider: str,
+        allowed_special: Literal["all"] | Set[str] | None = None,
+        disallowed_special: Literal["all"] | Collection[str] | None = None,
+    ) -> None:
+        pass
+
+    @abstractmethod
+    def truncate_str(self, text: str, max_len: int) -> str:
+        pass
+
+    @abstractmethod
+    def num_tokens_in_str(self, text: str) -> int:
+        pass
+
+    @abstractmethod
+    def num_tokens_in_messages(self, messages: list[MessageT]) -> int:
+        pass

{not_again_ai-0.14.0.dist-info → not_again_ai-0.16.0.dist-info}/METADATA

@@ -1,12 +1,11 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.3
 Name: not-again-ai
-Version: 0.14.0
+Version: 0.16.0
 Summary: Designed to once and for all collect all the little things that come up over and over again in AI projects and put them in one place.
-Home-page: https://github.com/DaveCoDev/not-again-ai
 License: MIT
 Author: DaveCoDev
 Author-email: dave.co.dev@gmail.com
-Requires-Python: >=3.11,<4.0
+Requires-Python: >=3.11, <3.13
 Classifier: Development Status :: 3 - Alpha
 Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Science/Research
@@ -19,26 +18,25 @@ Classifier: Programming Language :: Python :: 3.12
 Classifier: Typing :: Typed
 Provides-Extra: data
 Provides-Extra: llm
-Provides-Extra: local-llm
 Provides-Extra: statistics
 Provides-Extra: viz
-Requires-Dist: azure-ai-inference (==1.0.0b5) ; extra == "llm"
-Requires-Dist: azure-identity (>=1.19,<2.0) ; extra == "llm"
-Requires-Dist: jinja2 (>=3.1,<4.0) ; extra == "local-llm"
-Requires-Dist: loguru (>=0.7,<0.8)
-Requires-Dist: numpy (>=2.1,<3.0) ; extra == "statistics" or extra == "viz"
-Requires-Dist: ollama (>=0.3,<0.4) ; extra == "local-llm"
-Requires-Dist: openai (>=1.52,<2.0) ; extra == "llm"
-Requires-Dist: pandas (>=2.2,<3.0) ; extra == "viz"
-Requires-Dist: pydantic (>=2.9,<3.0)
-Requires-Dist: pytest-playwright (>=0.5,<0.6) ; extra == "data"
-Requires-Dist: python-liquid (>=1.12,<2.0) ; extra == "llm"
-Requires-Dist: scikit-learn (>=1.5,<2.0) ; extra == "statistics"
-Requires-Dist: scipy (>=1.14,<2.0) ; extra == "statistics"
-Requires-Dist: seaborn (>=0.13,<0.14) ; extra == "viz"
-Requires-Dist: tiktoken (>=0.8,<0.9) ; extra == "llm"
-Requires-Dist: transformers (>=4.45,<5.0) ; extra == "local-llm"
-Project-URL: Documentation, https://github.com/DaveCoDev/not-again-ai
+Requires-Dist: azure-identity (>=1.19) ; extra == "llm"
+Requires-Dist: loguru (>=0.7)
+Requires-Dist: numpy (>=2.2) ; extra == "statistics"
+Requires-Dist: numpy (>=2.2) ; extra == "viz"
+Requires-Dist: ollama (>=0.4) ; extra == "llm"
+Requires-Dist: openai (>=1) ; extra == "llm"
+Requires-Dist: pandas (>=2.2) ; extra == "viz"
+Requires-Dist: playwright (>=1.49) ; extra == "data"
+Requires-Dist: pydantic (>=2.10)
+Requires-Dist: pytest-playwright (>=0.7) ; extra == "data"
+Requires-Dist: python-liquid (>=1.12) ; extra == "llm"
+Requires-Dist: scikit-learn (>=1.6) ; extra == "statistics"
+Requires-Dist: scipy (>=1.15) ; extra == "statistics"
+Requires-Dist: seaborn (>=0.13) ; extra == "viz"
+Requires-Dist: tiktoken (>=0.8) ; extra == "llm"
+Project-URL: Documentation, https://davecodev.github.io/not-again-ai/
+Project-URL: Homepage, https://github.com/DaveCoDev/not-again-ai
 Project-URL: Repository, https://github.com/DaveCoDev/not-again-ai
 Description-Content-Type: text/markdown
 
@@ -68,11 +66,9 @@ Requires: Python 3.11, or 3.12
 Install the entire package from [PyPI](https://pypi.org/project/not-again-ai/) with: 
 
 ```bash
-$ pip install not_again_ai[llm,local_llm,statistics,viz]
+$ pip install not_again_ai[data,llm,statistics,viz]
 ```
 
-Note that local LLM requires separate installations and will not work out of the box due to how hardware dependent it is. Be sure to check the [notebooks](notebooks/local_llm/) for more details.
-
 The package is split into subpackages, so you can install only the parts you need.
 
 ### Base
@@ -93,16 +89,7 @@ The package is split into subpackages, so you can install only the parts you nee
    1. Using AOAI requires using Entra ID authentication. See https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/managed-identity for how to set this up for your AOAI deployment.
       * Requires the correct role assigned to your user account and being signed into the Azure CLI.
    1. (Optional) Set the `AZURE_OPENAI_ENDPOINT` environment variable.
-1. Setup GitHub Models
-   1. Get a Personal Access Token from https://github.com/settings/tokens and set the `GITHUB_TOKEN` environment variable. The token does not need any permissions.
-   1. Check the [Github Marketplace](https://github.com/marketplace/models) to see which models are available.
-
-
-### Local LLM
- 1. `pip install not_again_ai[llm,local_llm]`
- 1. Some HuggingFace transformers tokenizers are gated behind access requests. If you wish to use these, you will need to request access from HuggingFace on the model card.
-    * Then set the `HF_TOKEN` environment variable to your HuggingFace API token which can be found here: https://huggingface.co/settings/tokens
- 1. If you wish to use Ollama:
+1. If you wish to use Ollama:
      1. Follow the instructions at https://github.com/ollama/ollama to install Ollama for your system. 
      1. (Optional) [Add Ollama as a startup service (recommended)](https://github.com/ollama/ollama/blob/main/docs/linux.md#adding-ollama-as-a-startup-service-recommended)
      1. (Optional) To make the Ollama service accessible on your local network from a Linux server, add the following to the `/etc/systemd/system/ollama.service` file which will make Ollama available at `http://<local_address>:11434`:
@@ -112,7 +99,6 @@ The package is split into subpackages, so you can install only the parts you nee
          Environment="OLLAMA_HOST=0.0.0.0"
          ```
      1. It is recommended to always have the latest version of Ollama. To update Ollama check the [docs](https://github.com/ollama/ollama/blob/main/docs/). The command for Linux is: `curl -fsSL https://ollama.com/install.sh | sh`
- 1. HuggingFace transformers and other requirements are hardware dependent so for providers other than Ollama, this only installs some generic dependencies. Check the [notebooks](notebooks/local_llm/) for more details on what is available and how to install it.
 
 
 ### Statistics
@@ -156,10 +142,8 @@ $ poetry update
 
 To install all dependencies (with all extra dependencies) into an isolated virtual environment:
 
-> Append `--sync` to uninstall dependencies that are no longer in use from the virtual environment.
-
 ```bash
-$ poetry install --all-extras
+$ poetry sync --all-extras
 ```
 
 To [activate](https://python-poetry.org/docs/basic-usage#activating-the-virtual-environment) the
@@ -215,7 +199,7 @@ Automated code quality checks are performed using
 environments and run commands based on [`noxfile.py`](./noxfile.py) for unit testing, PEP 8 style
 guide checking, type checking and documentation generation.
 
-> Note: `nox` is installed into the virtual environment automatically by the `poetry install`
+> Note: `nox` is installed into the virtual environment automatically by the `poetry sync`
 > command above. Run `poetry shell` to activate the virtual environment.
 
 To run all default sessions:

not_again_ai-0.16.0.dist-info/RECORD

@@ -0,0 +1,38 @@
+not_again_ai/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+not_again_ai/base/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+not_again_ai/base/file_system.py,sha256=KNQmacO4Q__CQuq2oPzWrg3rQO48n3evglc9bNiP7KM,949
+not_again_ai/base/parallel.py,sha256=fcYhKBYBWvob84iKp3O93wvFFdXeidljZsShgBLTNGA,3448
+not_again_ai/data/__init__.py,sha256=1jF6mwvtB2PT7IEc3xpbRtZm3g3Lyf8zUqH4AEE4qlQ,244
+not_again_ai/data/web.py,sha256=wjx9cc33jcoJBGonYCIpwygPBFOwz7F-dx_ominmbnI,1838
+not_again_ai/llm/__init__.py,sha256=_wNUL6FDaT369Z8W48FsaC_NkcOZ-ib2MMUvnaLOS-0,451
+not_again_ai/llm/chat_completion/__init__.py,sha256=a2qmmmrXjMKyHGZDjt_xdqYbSrEOBea_VvZArzMboe0,200
+not_again_ai/llm/chat_completion/interface.py,sha256=FCyE-1gLdhwuS0Lv8iTbZvraa4iZjnKB8qb31WF53uk,1204
+not_again_ai/llm/chat_completion/providers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+not_again_ai/llm/chat_completion/providers/ollama_api.py,sha256=iBTMyF8edo8uxxrorNPtShzmCXG7m0RlEBunWLSO4Mo,7999
+not_again_ai/llm/chat_completion/providers/openai_api.py,sha256=S7TZhDIQ_xpp3JakRVcd3Gpw2UjeHCETdA9MfRKUjCU,12294
+not_again_ai/llm/chat_completion/types.py,sha256=q8APUWWzwCKL0Rs_zEFfph9uBcwh5nAT0f0rp4crvk0,4039
+not_again_ai/llm/embedding/__init__.py,sha256=wscUfROukvw0M0vYccfaVTdXV0P-eICAT5mqM0LaHHc,182
+not_again_ai/llm/embedding/interface.py,sha256=Hj3UiktXEeCUeMwpIDtRkwBfKgaJSnJvclLNyjwUAtE,1144
+not_again_ai/llm/embedding/providers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+not_again_ai/llm/embedding/providers/ollama_api.py,sha256=m-OCis9WAUT2baGsGVPzejlive40eSNyO6tHmPh6joM,3201
+not_again_ai/llm/embedding/providers/openai_api.py,sha256=JFFqbq0O5snIEnr9VESdp5xehikQBPbs7nwyE6acFsY,5441
+not_again_ai/llm/embedding/types.py,sha256=J4FFLx35Aow2kOaafDReeY9cUNqhWMjaAk5gXkX7SVk,506
+not_again_ai/llm/prompting/__init__.py,sha256=7YnHro1yH01FLGnao27WyqQDFjNYf9npE5UxoR9YrUU,84
+not_again_ai/llm/prompting/compile_prompt.py,sha256=lnbTOoTc7PumyP_GhfHaLZHp3UUpSB7VAeWOilS1wpI,4703
+not_again_ai/llm/prompting/interface.py,sha256=SMKYabmu3zTWbEDukU6aLU_JQ88apeBWWOF_qZ0s3ww,1783
+not_again_ai/llm/prompting/providers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+not_again_ai/llm/prompting/providers/openai_tiktoken.py,sha256=8YrEiK3ZHyKVGiVsJ_Rd6eVdISIvcub7ooj-HB7Prsc,4536
+not_again_ai/llm/prompting/types.py,sha256=xz70dnawL9rji7Zr1_mOekY-uUlvKJJf7k9nXJsOXc4,1219
+not_again_ai/py.typed,sha256=UaCuPFa3H8UAakbt-5G8SPacldTOGvJv18pPjUJ5gDY,93
+not_again_ai/statistics/__init__.py,sha256=gA8r9JQFbFSN0ykrHy4G1IQgcky4f2eM5Oo24oVI5Ik,466
+not_again_ai/statistics/dependence.py,sha256=4xaniMkLlTjdXcNVXdwepEAiZ-WaaGYfR9haJC1lU2Q,4434
+not_again_ai/viz/__init__.py,sha256=MeaWae_QRbDEHJ4MWYoY1-Ad6S0FhSDaRhQncS2cpSc,447
+not_again_ai/viz/barplots.py,sha256=rr_2phZgDaqcF5Ve7mBZrVvNXVzEt84RQPIyyeJxsMo,3384
+not_again_ai/viz/distributions.py,sha256=OyWwJaNI6lMRm_iSrhq-CORLNvXfeuLSgDtVo3umnzU,4354
+not_again_ai/viz/scatterplot.py,sha256=5CUOWeknbBOaZPeX9oPin5sBkRKEwk8qeFH45R-9LlY,2292
+not_again_ai/viz/time_series.py,sha256=pOGZqXp_2nd6nKo-PUQNCtmMh__69jxQ6bQibTGLwZA,5212
+not_again_ai/viz/utils.py,sha256=hN7gwxtBt3U6jQni2K8j5m5pCXpaJDoNzGhBBikEU28,238
+not_again_ai-0.16.0.dist-info/LICENSE,sha256=btjOgNGpp-ux5xOo1Gx1MddxeWtT9sof3s3Nui29QfA,1071
+not_again_ai-0.16.0.dist-info/METADATA,sha256=kvwxTcEi-elRl-LuHyh2QtFLrpYHd-U6HjyuAkHYvWQ,15035
+not_again_ai-0.16.0.dist-info/WHEEL,sha256=IYZQI976HJqqOpQU6PHkJ8fb3tMNBFjg-Cn-pwAbaFM,88
+not_again_ai-0.16.0.dist-info/RECORD,,

{not_again_ai-0.14.0.dist-info → not_again_ai-0.16.0.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: poetry-core 1.9.0
+Generator: poetry-core 2.0.1
 Root-Is-Purelib: true
 Tag: py3-none-any

not_again_ai/llm/gh_models/azure_ai_client.py

@@ -1,20 +0,0 @@
-import os
-
-from azure.ai.inference import ChatCompletionsClient
-from azure.core.credentials import AzureKeyCredential
-
-
-def azure_ai_client(
-    token: str | None = None,
-    endpoint: str = "https://models.inference.ai.azure.com",
-) -> ChatCompletionsClient:
-    if not token:
-        token = os.getenv("GITHUB_TOKEN")
-        if not token:
-            raise ValueError("Token must be provided or GITHUB_TOKEN environment variable must be set")
-
-    client = ChatCompletionsClient(
-        endpoint=endpoint,
-        credential=AzureKeyCredential(token),
-    )
-    return client

not_again_ai/llm/gh_models/chat_completion.py

@@ -1,81 +0,0 @@
-import contextlib
-import json
-import time
-from typing import Any
-
-from azure.ai.inference import ChatCompletionsClient
-from azure.ai.inference.models import ChatCompletionsToolDefinition, ChatRequestMessage
-
-
-def chat_completion(
-    messages: list[ChatRequestMessage],
-    model: str,
-    client: ChatCompletionsClient,
-    tools: list[ChatCompletionsToolDefinition] | None = None,
-    max_tokens: int | None = None,
-    temperature: float | None = None,
-    json_mode: bool = False,
-    seed: int | None = None,
-) -> dict[str, Any]:
-    """Gets a response from GitHub Models using the Azure AI Inference SDK.
-    See the available models at https://github.com/marketplace/models
-    Full documentation of the SDK is at: https://learn.microsoft.com/en-us/azure/ai-studio/reference/reference-model-inference-chat-completions
-    And samples at: https://github.com/Azure/azure-sdk-for-python/tree/main/sdk/ai/azure-ai-inference/samples
-
-    Returns:
-        dict[str, Any]: A dictionary with the following keys
-            message (str | dict): The content of the generated assistant message.
-                If json_mode is True, this will be a dictionary.
-            tool_names (list[str], optional): The names of the tools called by the model.
-                If the model does not support tools, a ResponseError is raised.
-            tool_args_list (list[dict], optional): The arguments of the tools called by the model.
-            prompt_tokens (int): The number of tokens in the messages sent to the model.
-            completion_tokens (int): The number of tokens used by the model to generate the completion.
-            response_duration (float): The time, in seconds, taken to generate the response by using the model.
-            system_fingerprint (str, optional): If seed is set, a unique identifier for the model used to generate the response.
-    """
-    response_format = {"type": "json_object"} if json_mode else None
-    start_time = time.time()
-    response = client.complete(  # type: ignore
-        messages=messages,
-        model=model,
-        response_format=response_format,  # type: ignore
-        max_tokens=max_tokens,
-        temperature=temperature,
-        tools=tools,
-        seed=seed,
-    )
-    end_time = time.time()
-    response_duration = end_time - start_time
-
-    response_data = {}
-    finish_reason = response.choices[0].finish_reason
-    response_data["finish_reason"] = finish_reason.value  # type: ignore
-
-    message = response.choices[0].message.content
-    if message and json_mode:
-        with contextlib.suppress(json.JSONDecodeError):
-            message = json.loads(message)
-    response_data["message"] = message
-
-    # Check for tool calls because even if the finish_reason is stop, the model may have called a tool
-    tool_calls = response.choices[0].message.tool_calls
-    if tool_calls:
-        tool_names = []
-        tool_args_list = []
-        for tool_call in tool_calls:
-            tool_names.append(tool_call.function.name)
-            tool_args_list.append(json.loads(tool_call.function.arguments))
-        response_data["tool_names"] = tool_names
-        response_data["tool_args_list"] = tool_args_list
-
-    if seed is not None and hasattr(response, "system_fingerprint"):
-        response_data["system_fingerprint"] = response.system_fingerprint
-
-    usage = response.usage
-    if usage is not None:
-        response_data["completion_tokens"] = usage.completion_tokens
-        response_data["prompt_tokens"] = usage.prompt_tokens
-    response_data["response_duration"] = round(response_duration, 4)
-
-    return response_data