Tired of paying for OPENAI, PINECONE, GOOGLESEARCH APIs to try out the latest developments in the AI field? Perfect, this is the repository for you! 🎁
For any problem open an ISSUE 🚬, the project is very simple so any help is welcome💸.
Are you bored reading😴? Do you want to try our project now⏳? Open the notebook on Colab everything is ready!
By the way, thank you so much for 
Hello everyone 🥰 ,
I wanted to start by talking about how important it is to democratize AI. Unfortunately, most new applications or discoveries in this field end up enriching some big companies, leaving behind small businesses or simple projects. One striking example of this is Autogpt, an autonomous AI agent capable of performing tasks.
Autogpt and similar projects like BabyAGI only work with paid APIs, which is not fair. That's why I tried to recreate a simpler but very interesting and, above all, open-source version of Autogpt that does not require any API and does not need any particular hardware.
I believe that by providing free and open-source AI tools, we can give small businesses and individuals the opportunity to create new and innovative projects without the need for significant financial investment. This will allow for more equitable and diverse access to AI technology, which is essential for advancing society as a whole.
-
HUGGINGFACE🤗 : Visit this simple official guide
-
Now Running also with HuggingCHAT
-
(OPTIONAL BUT BETTER RESULT) CHATGPT🖥 :
- Go to https://chat.openai.com/chat and open the developer tools by
F12. - Find the
__Secure-next-auth.session-tokencookie inApplication>Storage>Cookies>https://chat.openai.com. - Copy the value in the
Cookie Valuefield.
- Go to https://chat.openai.com/chat and open the developer tools by
Or use Locally :
- Dowload the repository FREE AUTOGPT REPOSITORY
- pip3 install -r requirements.txt
- streamlit run Camel.py
Camel.mp4
Or use Locally :
- Dowload the repository FREE AUTOGPT REPOSITORY
- pip3 install -r requirements.txt
- Usage: python BABYAGI.py
BAbyAGI.mp4
Or use Locally :
- Dowload the repository FREE AUTOGPT REPOSITORY
- pip3 install -r requirements.txt
- Usage: python AUTOGPT.py
AUTOGPT.mp4
RUN NOW ON COLAB😮 
Or use Locally :
- Dowload the repository FREE AUTOGPT REPOSITORY
- pip3 install -r requirements.txt
- cd OtherAgent/
- Choose or develop your agent [ csvAgent.py ; pythonAgent.py ; customAgent.py ]
- Usage: python YourAgent.py
CustomAgent.mp4
To create an open-source version of Autogpt that does not require paid APIs or specific hardware, we performed a reverse engineering process on ChatGPT, a language model developed by OpenAI. By doing so, we were able to use the agents and new technologies of langchain for free.
We then created a custom LLM wrapper with langchain, which can be used as a plug-and-play solution with any langchain function or tool 💡.
from FreeLLM import ChatGPTAPI
# Instantiate a ChatGPT object with your token
llm = ChatGPTAPI.ChatGPT((token="YOURTOKEN") #for start new chat
# or if if u would to start from an existing chat
# llm = ChatGPT(token = "YOUR-TOKEN", conversation = "Add-XXXX-XXXX-Convesation-ID")
# Generate a response based on the given prompt
response = llm("Hello, how are you?")
# Print the response
print(response)from FreeLLM import HuggingChatAPI
# Instantiate a ChatGPT object with your token
llm = HuggingChatAPI.HuggingChat() #for start new chat
# Generate a response based on the given prompt
response = llm("Hello, how are you?")
# Print the response
print(response)
The code snippet provided above shows how to use our custom ChatGPT LLM class to interact with the language model. It requires a token from the ChatGPT API, which can be obtained from https://chat.openai.com/api/auth/session.
Please note that there is a limit of 50 requests per hour for each account on the ChatGPT API 💣. Therefore, we implemented a call counter in our ChatGPT class to prevent exceeding this limit.
We believe that our open-source version of Autogpt will promote equitable and diverse access to AI technology and empower individuals and small businesses to create innovative AI projects without significant financial investment.
This is an example of CUSTOM agent, in less of 60 line of code and totally for free, with:
- Internet access
- Python code execution
- Wikipedia knowledge
#pip install langchain==0.0.150
from langchain.agents import initialize_agent #use for create new agent
from langchain.agents import initialize_agent, Tool
from langchain.utilities import PythonREPL #tool for execute python script
from langchain.utilities import WikipediaAPIWrapper #tool get wiki info
from langchain.tools import DuckDuckGoSearchTool #tool get interet live info
from FreeLLM import ChatGPTAPI # FREE CHATGPT API
#or
from FreeLLM import HuggingChatAPI
# Instantiate a ChatGPT object with your token
llm = ChatGPTAPI.ChatGPT((token="YOURTOKEN")
# or use HuggingChatAPI if u dont have CHATGPT account
# llm = HuggingChatAPI.HuggingChat()
# Define the tools
wikipedia = WikipediaAPIWrapper()
python_repl = PythonREPL()
search = DuckDuckGoSearchTool()
tools = [
Tool(
name = "python repl",
func=python_repl.run,
description="useful for when you need to use python to answer a question. You should input python code"
)
]
wikipedia_tool = Tool(
name='wikipedia',
func= wikipedia.run,
description="Useful for when you need to look up a topic, country or person on wikipedia"
)
duckduckgo_tool = Tool(
name='DuckDuckGo Search',
func= search.run,
description="Useful for when you need to do a search on the internet to find information that another tool can't find. be specific with your input."
)
tools.append(duckduckgo_tool)
tools.append(wikipedia_tool)
#Create the Agent
iteration = (int(input("Enter the number of iterations: ")) if input("Do you want to set the number of iterations? (y/n): ") == "y" else 3)
zero_shot_agent = initialize_agent(
agent="zero-shot-react-description",
tools=tools,
llm=llm,
verbose=True,
max_iterations=iteration,
)
# Start your Custom Agent in loop
print(">> STRAT CUSTOM AGENT")
print("> Digit 'exit' for exit or 'your task or question' for start\n\n")
prompt = input("(Enter your task or question) >> ")
while prompt.toLowerCase() != "exit":
zero_shot_agent.run(prompt)
prompt = input("(Enter your task or question) >> ")
By the way, thank you so much for
-
Create free LLM langchain wrapper based on Reverse Engineered ChatGPT API by OpenAI
-
Create a simple versione of AUTOGPT based on Camel theory
-
Find a way to replace OpenAIEmbeddings()
-
Create a simple version of AUTOGPT based on Baby AGI
-
Add web search agent
-
Add file writer agent
-
Finally AUTOGPT without paids API
-
Add other free Custom LLM wrapper Add this
-
Add long term memory
-
Find a way to replace PINECONE api
-
Find a way to replace official Google API
- https://github.com/hwchase17/langchain
- https://github.com/terry3041/pyChatGPT
- https://github.com/Soulter/hugging-chat-api
- https://github.com/Significant-Gravitas/Auto-GPT
RUN NOW ON COLAB😮