aix 0.0.4

Artificial Intelligence eXtensions

aix

Artificial Intelligence eXtensions

Fast access to your favorite A.I. tools.

To install: pip install aix

AI chat

aix.chat_funcs gathers the models you have access to in your environment (depends on what you have installed (e.g. google.generativeai, oa (itself a facade to openai functionality), etc.)).

from aix import chat_funcs

list(chat_funcs)

['gemini-1.5-flash',
 'gpt-4',
 'gpt-4-32k',
 'gpt-4-turbo',
 'gpt-3.5-turbo',
 'o1-preview',
 'o1-mini',
 'gpt-4o',
 'gpt-4o-mini']

chat_funcs is a dictionary whos keys are the names of the available models, and values are a chat function with the model set to that model name.

chat_funcs['o1-mini']

functools.partial(<function chat at 0x1355f68c0>, model='o1-mini')

Note that different providers have different interfaces, but the functions that chat_funcs provides all have prompt as their first argument.

from inspect import signature

signature(chat_funcs['o1-mini'])

<Sig (prompt=None, *, model='o1-mini', messages=None, frequency_penalty: 'Optional[float] | NotGiven' = NOT_GIVEN, function_call: 'completion_create_params.FunctionCall | NotGiven' = NOT_GIVEN, functions: 'Iterable[completion_create_params.Function] | NotGiven' = NOT_GIVEN, logit_bias: 'Optional[Dict[str, int]] | NotGiven' = NOT_GIVEN, logprobs: 'Optional[bool] | NotGiven' = NOT_GIVEN, max_tokens: 'Optional[int] | NotGiven' = NOT_GIVEN, n: 'Optional[int] | NotGiven' = NOT_GIVEN, parallel_tool_calls: 'bool | NotGiven' = NOT_GIVEN, presence_penalty: 'Optional[float] | NotGiven' = NOT_GIVEN, response_format: 'completion_create_params.ResponseFormat | NotGiven' = NOT_GIVEN, seed: 'Optional[int] | NotGiven' = NOT_GIVEN, service_tier: "Optional[Literal['auto', 'default']] | NotGiven" = NOT_GIVEN, stop: 'Union[Optional[str], List[str]] | NotGiven' = NOT_GIVEN, stream: 'Optional[Literal[False]] | Literal[True] | NotGiven' = NOT_GIVEN, stream_options: 'Optional[ChatCompletionStreamOptionsParam] | NotGiven' = NOT_GIVEN, temperature: 'Optional[float] | NotGiven' = NOT_GIVEN, tool_choice: 'ChatCompletionToolChoiceOptionParam | NotGiven' = NOT_GIVEN, tools: 'Iterable[ChatCompletionToolParam] | NotGiven' = NOT_GIVEN, top_logprobs: 'Optional[int] | NotGiven' = NOT_GIVEN, top_p: 'Optional[float] | NotGiven' = NOT_GIVEN, user: 'str | NotGiven' = NOT_GIVEN, extra_headers: 'Headers | None' = None, extra_query: 'Query | None' = None, extra_body: 'Body | None' = None, timeout: 'float | httpx.Timeout | None | NotGiven' = NOT_GIVEN)>

signature(chat_funcs['gemini-1.5-flash'])

<Sig (prompt: str, *, model='gemini-1.5-flash', generation_config: 'generation_types.GenerationConfigType | None' = None, safety_settings: 'safety_types.SafetySettingOptions | None' = None, stream: 'bool' = False, tools: 'content_types.FunctionLibraryType | None' = None, tool_config: 'content_types.ToolConfigType | None' = None, request_options: 'helper_types.RequestOptionsType | None' = None)>

For tab-completion convenience, the (python identifier version of the) models were placed as attributes of chat_funcs, so you can access them directly there.

print(chat_funcs.gemini_1_5_flash('What is the capital of France?'))

The capital of France is **Paris**. 

print(chat_funcs.gpt_3_5_turbo('What is the capital of France?'))

The capital of France is Paris.

There's also a dictionary called chat_models that contains the same keys:

from aix import chat_models

list(chat_models)

['gemini-1.5-flash',
 'gpt-4',
 'gpt-4-32k',
 'gpt-4-turbo',
 'gpt-3.5-turbo',
 'o1-preview',
 'o1-mini',
 'gpt-4o',
 'gpt-4o-mini']

But here the values are some useful metadatas on the model, like pricing...

chat_models['gpt-4o']

{'price_per_million_tokens': 5.0,
 'pages_per_dollar': 804,
 'performance_on_eval': 'Efficiency-optimized version of GPT-4 for better performance on reasoning tasks',
 'max_input': 8192,
 'provider': 'openai'}

The corresponding attributes are only the name of the model (the key itself):

chat_models.gpt_4

'gpt-4'

This is for the convenience of entering a model name in a different context, with less errors than if you were typing the name as a string. For example, you can enter it yourself in the general chat function:

from aix import chat, chat_models

chat('How many Rs in "Strawberry"?', model=chat_models.gpt_4o, frequency_penalty=0.5)  

'The word "Strawberry" contains two instances of the letter \'R\'.'

Extras (older version -- might deprecate or change interface)

Want all your faves at your fingertips?

Never remember where to import that learner from?

Say LinearDiscriminantAnalysis?

... was it from sklearn?

... was it from sklearn.linear_model?

... ah no! It was from sklearn.discriminant_analysis import LinearDiscriminantAnalysis.

Sure, you can do that. Or you can simply type from aix.Lin... click tab, and; there it is! Select, enter, and moving on with real work.

Note: This is meant to get off the ground quickly -- once your code is stable, you should probably import your stuff directly from it's origin

Coming up

Now that the AI revolution is on its way, we'll add the ability to find, and one day, use the right AI tool -- until the day that AI will do even that for us...