Skip to content

llm_chat

thml.llm_chat

This module contains the chatbot classes and methods to work with text generation models.

Modules:

post_api

This module contains the chatbot classes and methods to work with text generation models.

Modules:

chat_google

Build a chatbot using Google's Gemini API

REF: - Python API: https://ai.google.dev/tutorials/python_quickstart - Prompt examples: https://ai.google.dev/docs/prompt_best_practices - Palm vs Gemini: https://ai.google.dev/docs/migration_guide

Classes:

Google(api_key=None, **kwargs)

Class for chatbot using Google's Gemini API via google.generativeai package

Parameters:

  • api_key (str, default: None ) –

    The OpenAI API key.

Other Parameters:

  • model (str = 'gemini-pro') –

    The model to use for the chat client.

  • temperature (float = 0.7) –

    The temperature to use for the chat client.

  • top_p (float = 1) –

    An alternative to sampling with temperature.

  • max_tokens (int = 8096) –

    The maximum number of tokens to generate in the response.

Methods:

  • ask –

    Ask Google Gemini a question and return the answer.

Attributes:

avail_models = avail_models instance-attribute
params = kwargs instance-attribute
ask(prompt='hello my friend')

Ask Google Gemini a question and return the answer.

Parameters:

  • prompt (str, default: 'hello my friend' ) –

    The question or prompt to ask the chatbot.

Returns:

  • text ( str ) –

    The answer to the question.

chat_gpt4free

Classes:

  • FreeChat –

    Class for chatbot using reverse enginerring models.

  • FreeImage –

    Class for image generation using reverse-enginered models.

FreeChat(**kwargs: Any)

Bases: _Base

Class for chatbot using reverse enginerring models.

Other Parameters:

  • provider (str = None) –

    The provider of the model. If None, the best provider will be used.

  • api_key (str = None) –

    The API key for the provider.

  • model (str = 'gpt-4') –

    The model to use.

  • temperature (float = 0.7) –

    The temperature of the model.

  • top_p (float = 1) –

    The top_p of the model.

  • max_tokens (int = 8096) –

    The max tokens of the model.

  • system_prompt (str = '') –

    The system prompt of the model.

Methods:

  • ask –

    Ask the chatbot a question.

Attributes:

params = _params(**kwargs) instance-attribute
avail_models = avail_models instance-attribute
avail_providers = self._avail_providers() instance-attribute
ask(prompt: str) -> str

Ask the chatbot a question. Args: prompt (str): The input string for the chatbot.

Returns:

  • text ( str ) –

    The answer to the question.

FreeImage()

Class for image generation using reverse-enginered models.

chat_openai

Using chatGPT API to build a chatbot

Implementation following this repo: https://github.com/stancsz/chatgpt/blob/master/ChatGPT.py

Terms: - OpenAI's text generation models (often called generative pre-trained transformers or large language models) - The inputs to these models are also referred to as "prompts".

REF: - Refer to file: src_thatool\devtools\dev_chatGPT\chat_API/chat_Copilot2GPT.ipynb - openai docs: https://platform.openai.com/docs/guides/text-generation/text-generation-models - openai repo: https://github.com/openai/openai-python - prompt examples: https://github.com/f/awesome-chatgpt-prompts - openai examples: https://platform.openai.com/examples - inherent class in python: https://stackoverflow.com/questions/9575409/calling-parent-class-init-with-multiple-inheritance-whats-the-right-way - super() in python: https://stackoverflow.com/questions/34550425/how-to-initialize-subclass-paramseters-in-python-using-super - create chatbot using openai API: https://medium.com/data-professor/beginners-guide-to-openai-api-a0420bc58ee5 - OpenAI API tips: https://arize.com/blog-course/mastering-openai-api-tips-and-tricks/

Classes:

  • BaseChat –

    Base class for chatbot, to define common attributes and methods for chatbot

  • Openai –

    Class for chatbot using OpenAI API via openai package

  • Post –

    Class for chatbot using OpenAI API via POST request

BaseChat

Base class for chatbot, to define common attributes and methods for chatbot

Methods:

save_history(prompt, response)
export_history(filename='chat_history.txt')
load_history(filename='chat_history.txt')
Openai(service: str = 'openai', **kwargs: Any)

Bases: BaseChat

Class for chatbot using OpenAI API via openai package

Parameters:

  • service (str, default: 'openai' ) –

    The service to use for the chat client (preset of base_url). Available services are: openai, copilot, local_gpt4all

Other Parameters:

  • base_url (str) –

    The OpenAI API base URL. Presetted based on the service.

  • api_key (str) –

    The OpenAI API key. Presetted based on the service.

  • model (str = 'gpt-4') –

    The model to use for the chat client. All models can be found at the OpenAI site. Only 2 models 'gpt-4' and 'gpt-3.5-turbo' for copilot.

  • temperature (float = 0.7) –

    The temperature to use for the chat client. The temperature is a value between 0 and 1. Lower temperatures will cause the model to repeat itself more often, while higher temperatures will increase the model's diversity of responses. Use either temperature or top_p, but not both.

  • top_p (float = 1) –

    An alternative to sampling with temperature. The top_p is a value between 0 and 1. Use either temperature or top_p, but not both.

  • max_tokens (int = 8096) –

    The maximum number of tokens to generate in the response.

  • stream (bool = False) –

    Whether to stream the response or not.

  • system_prompt (str = '') –

    The prompt to use for the system.

Methods:

Attributes:

params = _params(service, **kwargs) instance-attribute
avail_models = avail_models instance-attribute
save_history(prompt, response)
export_history(filename='chat_history.txt')
load_history(filename='chat_history.txt')
ask(prompt='hello', **kwargs: Any) -> str

Ask GPT-4 a question and return the answer. Use new openai API

Parameters:

  • prompt (str, default: 'hello' ) –

    The question to ask GPT-4.

Other Parameters:

  • save_history (bool = False) –

    Whether to save the question and answer to the chat history.

  • use_history (bool = False) –

    Whether to use the chat history in the current request.

Returns:

  • text ( str ) –

    The answer to the question.

Post(service: str = 'copilot', **kwargs: Any)

Class for chatbot using OpenAI API via POST request

Parameters:

  • **kwargs (Any, default: {} ) –

    See Openai class for the arguments.

Refs

Methods:

Attributes:

params = _params(service, **kwargs) instance-attribute
avail_models = avail_models instance-attribute
ask(prompt='hello')

web_playwright

Modules:

chatgpt

Classes:

WebChatgpt(cookie_file: str = None, proxy: str = None, chat_id: str = 'temporary')

Bases: WebBase

Interacte with chatgpt web

Parameters:

  • cookie_file (str, default: None ) –

    path to cookie file

  • proxy (str, default: None ) –

    proxy server. e.g., "http://something.com:8080"

  • chat_id (str, default: 'temporary' ) –

    "id", "last", "temporary", "new". If "id", use chat_id. If chat_id is not be found on the web, fallback to chat_page="temporary". If "last", use last chat. If "temporary", use temporary chat (need to go wedsite to turn on this feature at the first time login of an account). If "new", start new chat.

Methods:

Attributes:

browser_kwargs = {} instance-attribute
context_kwargs = {} instance-attribute
device: str = None instance-attribute
page = None instance-attribute
send_count = 0 instance-attribute
base_url = 'https://chatgpt.com' instance-attribute
cookie_file = cookie_file instance-attribute
chat_id = chat_id instance-attribute
login = False instance-attribute
prompt_textarea = self.page.get_by_placeholder('Message ChatGPT') instance-attribute
send_button = self.page.locator('button.mb-1.me-1.h-8.w-8') instance-attribute
stop_button = self.page.locator('button.mb-1.me-1.h-8.w-8').locator('rect') instance-attribute
ask(prompt: str = 'Hello, are you gpt-4o?')

Ask and get reponse from web

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

Returns:

  • str –

    response from AI

get_chat_history()

alias of get_all_messages() method, but in sync mode

close()

Alias of _close_page() method, but in sync mode

send_prompt(prompt: str = 'Hello, are you gpt-4o?') async

Submit prompt text

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

get_last_ai_message() async

Get the last AI message

get_all_messages() -> list[dict] async

Get all messages by user and AI

get_all_chat_id() async

Get all conversation ids

get_last_chat_id() async

Get all conversation ids

claude

Classes:

WebClaude(cookie_file: str = None, proxy: str = None, chat_id: str = 'last')

Bases: WebBase

Interacte with chatgpt web

Parameters:

  • cookie_file (str, default: None ) –

    path to cookie file

  • proxy (str, default: None ) –

    proxy server. e.g., "http://something.com:8080"

  • chat_id (str, default: 'last' ) –

    "id", "last", "new". If "id", use chat_id. If chat_id is not be found on the web, fallback to chat_id="last". If "last", use last_id. If "new", start new chat.

Methods:

Attributes:

browser_kwargs = {} instance-attribute
context_kwargs = {} instance-attribute
device: str = None instance-attribute
page = None instance-attribute
send_count = 0 instance-attribute
base_url = 'https://claude.ai' instance-attribute
cookie_file = cookie_file instance-attribute
chat_id = chat_id instance-attribute
prompt_textarea = self.page.get_by_label('Write your prompt to Claude').locator('p') instance-attribute
send_button = self.page.get_by_role('button', name='Send Message') instance-attribute
stop_button = self.page.get_by_role('button', name='Stop Response') instance-attribute
ask(prompt: str = 'Hello, are you gpt-4o?')

Ask and get reponse from web

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

Returns:

  • str –

    response from AI

get_chat_history()

alias of get_all_messages() method, but in sync mode

close()

Alias of _close_page() method, but in sync mode

send_prompt(prompt: str = 'Hello, are you gpt-4o?') async

Submit prompt text

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

get_last_ai_message() async

Get the last AI message

get_all_chat_id() async

Get all conversation ids

get_last_chat_id() async

Get all conversation ids

copilot_playwright

Classes:

WebCopilot(cookie_file: str = None, proxy: str = None, chat_id: str = None, converstion_style: str = None)

Bases: WebBase

Interacte with chatgpt web

Parameters:

  • cookie_file (str, default: None ) –

    path to cookie file

  • proxy (str, default: None ) –

    proxy server. e.g., "http://something.com:8080"

Methods:

Attributes:

browser_kwargs = {} instance-attribute
context_kwargs = {} instance-attribute
device: str = None instance-attribute
page = None instance-attribute
send_count = 0 instance-attribute
base_url = 'https://copilot.microsoft.com/' instance-attribute
cookie_file = cookie_file instance-attribute
prompt_textarea = self.page.get_by_role('textbox', name='Ask me anything...') instance-attribute
send_button = self.page.get_by_role('button', name='Submit') instance-attribute
stop_button = self.page.get_by_role('button', name='Stop Responding') instance-attribute
upload_image_button = self.page.get_by_role('button', name='Add an image to search') instance-attribute
upload_file_button = self.page.get_by_role('button', name='Add a file') instance-attribute
ask(prompt: str = 'Hello, are you gpt-4o?')

Ask and get reponse from web

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

Returns:

  • str –

    response from AI

get_chat_history()

alias of get_all_messages() method, but in sync mode

close()

Alias of _close_page() method, but in sync mode

send_prompt(prompt: str = 'Hello, are you gpt-4o?') async

Submit prompt text

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

get_last_ai_message() async

Get the last AI message

get_last_ai_reference() async

Get the references in last AI message

get_all_messages() -> list[dict] async

Get all messages by user and AI

gemini

Classes:

WebGemini(cookie_file: str = None, proxy: str = None, conversation_id: str = None)

Bases: WebBase

Interacte with chatgpt web

Parameters:

  • cookie_file (str, default: None ) –

    path to cookie file

  • proxy (str, default: None ) –

    proxy server. e.g., "http://something.com:8080"

Methods:

Attributes:

browser_kwargs = {} instance-attribute
context_kwargs = {} instance-attribute
device: str = None instance-attribute
page = None instance-attribute
send_count = 0 instance-attribute
base_url = 'https://gemini.google.com/app' instance-attribute
cookie_file = cookie_file instance-attribute
prompt_textarea = self.page.get_by_role('textbox') instance-attribute
ask(prompt: str = 'Hello, are you gpt-4o?')

Ask and get reponse from web

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

Returns:

  • str –

    response from AI

get_chat_history()

alias of get_all_messages() method, but in sync mode

close()

Alias of _close_page() method, but in sync mode

send_prompt(prompt: str = 'Hello, are you gpt-4o?') async

Submit prompt text

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

get_last_message() async

Get the last AI message

get_all_messages() -> list[dict] async

Get all messages by user and AI

up_load_file() async

Get text

llama

Classes:

  • WebLlama –

    Interacte with chatgpt web

WebLlama(cookie_file: str = None, proxy: str = None, chat_id: str = 'last')

Bases: WebBase

Interacte with chatgpt web

Parameters:

  • cookie_file (str, default: None ) –

    path to cookie file

  • proxy (str, default: None ) –

    proxy server. e.g., "http://something.com:8080"

  • chat_id (str, default: 'last' ) –

    "id", "last", "new". If "id", use chat_id. If chat_id is not be found on the web, fallback to chat_id="last". If "last", use last_id. If "new", start new chat.

Methods:

Attributes:

browser_kwargs = {} instance-attribute
context_kwargs = {} instance-attribute
device: str = None instance-attribute
page = None instance-attribute
send_count = 0 instance-attribute
base_url = 'https://chatwithllama.com' instance-attribute
cookie_file = cookie_file instance-attribute
chat_id = chat_id instance-attribute
prompt_textarea = self.page.get_by_placeholder('Ask anything!') instance-attribute
send_button = self.page.get_by_role('button', name='Send question') instance-attribute
stop_button = self.page.get_by_role('button', name='Stop generation') instance-attribute
ask(prompt: str = 'Hello, are you gpt-4o?')

Ask and get reponse from web

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

Returns:

  • str –

    response from AI

get_chat_history()

alias of get_all_messages() method, but in sync mode

close()

Alias of _close_page() method, but in sync mode

send_prompt(prompt: str = 'Hello, are you gpt-4o?') async

Submit prompt text

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

get_last_ai_message() async

Get the last AI message

get_all_chat_id() async

Get all conversation ids

get_last_chat_id() async

Get all conversation ids

mistral

Classes:

WebMistral(cookie_file: str = None, proxy: str = None, chat_id: str = 'last')

Bases: WebBase

Interacte with chatgpt web

Parameters:

  • cookie_file (str, default: None ) –

    path to cookie file

  • proxy (str, default: None ) –

    proxy server. e.g., "http://something.com:8080"

  • chat_id (str, default: 'last' ) –

    "id", "last", "new". If "id", use chat_id. If chat_id is not be found on the web, fallback to chat_id="last". If "last", use last_id. If "new", start new chat.

Methods:

Attributes:

browser_kwargs = {} instance-attribute
context_kwargs = {} instance-attribute
device: str = None instance-attribute
page = None instance-attribute
send_count = 0 instance-attribute
base_url = 'https://chat.mistral.ai' instance-attribute
cookie_file = cookie_file instance-attribute
chat_id = chat_id instance-attribute
prompt_textarea = self.page.get_by_placeholder('Ask anything!') instance-attribute
send_button = self.page.get_by_role('button', name='Send question') instance-attribute
stop_button = self.page.get_by_role('button', name='Stop generation') instance-attribute
ask(prompt: str = 'Hello, are you gpt-4o?')

Ask and get reponse from web

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

Returns:

  • str –

    response from AI

get_chat_history()

alias of get_all_messages() method, but in sync mode

close()

Alias of _close_page() method, but in sync mode

send_prompt(prompt: str = 'Hello, are you gpt-4o?') async

Submit prompt text

Parameters:

  • prompt (str, default: 'Hello, are you gpt-4o?' ) –

    prompt text

get_last_ai_message() async

Get the last AI message

get_all_chat_id() async

Get all conversation ids

get_last_chat_id() async

Get all conversation ids

web_requests

Modules:

bing_copilot

Classes:

  • RWebCopilot –

    Reverse-engineered Bing/Edge Copilot via Web browser.

Functions:

RWebCopilot(cookie_file: str = None, conversation_style: Literal['creative', 'balanced', 'precise'] = 'precise')

Reverse-engineered Bing/Edge Copilot via Web browser.

Parameters:

  • cookie_file (str, default: None ) –

    The path to the cookie file.

  • conversation_style (str, default: 'precise' ) –

    The conversation style. Available options: 'creative', 'balanced', 'precise'

Methods:

  • ask –

    Ask the bot a question

Attributes:

cookie_file = cookie_file instance-attribute
bot = asyncio.run(Chatbot.create(cookies=cookies)) instance-attribute
conversation_style = style_map[conversation_style] instance-attribute
ask(prompt: str, attachment: str = None, return_refs: bool = False) -> Union[str, dict[str, list]]

Ask the bot a question

Parameters:

  • prompt (str) –

    The prompt to ask the bot

Returns:

  • Union[str, dict[str, list]] –

    dict[str, list[str]]: final_text, references

response_parser(response: dict) -> dict[str, list[str]]

Parse the response from the re_edge_gpt chatbot

Parameters:

  • response (dict) –

    response from the re_edge_gpt chatbot

Returns:

  • dict[str, list[str]] –

    dict[str, list[str]]: final_text, references