ecomoptimizer/chatgpt-turbocli

Supports inputting data via files, features summarization of larger inputs, multi-line input, token usage stats. The program is flexible when initiating new chats and supports various named personas and preloading history in yaml files. Easily store jailbreak personas and stories.

ChatGPT TurboCLI

Key features

Assistant personas via assistants.yml

Few shot learning (context loading) through stories.yml

File input while in CLI

Feature	Benefits	Status
File Input for Analysis of Reports	Enables analysis of a wide range of text data including financial reports, news reports, source code, etc.	✅
Multiline Input for Quick Paste	Allows quick and easy pasting of large amounts of data or complex text data.	✅
Summarization for Large File Input	Summarizes large text files using natural language processing techniques to identify important trends and insights.	✅ (Using textract module - working on macOS & Linux, Windows not supported)
Token Usage for Cost Calculation	Calculates the number of tokens used and their associated cost in the application.	✅
Adjustable Completion Tokens for Customization	Adjusts how large of an output from model you want at most.	✅
Chat History for Context-Aware Communication	Keeps track of previous interactions with users, allowing the chatbot to provide more natural and intuitive responses.	✅
Transcripts to View Communication History	Provides historic questions and outputs, allowing for easier tracking of communication history and referencing of previous interactions with the chatbot.	✅
Logs for Debugging	Detailed records of all interactions with the application, including input and output data and any errors that occur during processing. Useful for debugging purposes in case of errors.	✅
Toggleable logging	Choose if you want logging at all	✅
Summarization on any input over x length	Summarize any input that are exceeding a given length to optimize token usage.	✅
Load assistant personas from file	Load assistant personas defined in assistants.yml file. This should enable the usage of e.g. jailbreak personas.	✅

Installation

1. Clone this repository
2. Install the required Python packages:

   pip install -r requirements.txt
   python3 -m spacy download en_core_web_sm
   

3. Create an API key for OpenAI at https://beta.openai.com/signup/
4. Run the script in the command line:

   python chatgpt-turbocli.py --api_key YOUR_API_KEY --log ERROR --temperature 1
   

   Replace YOUR_API_KEY with your OpenAI API key and MODEL_NAME with the name of the OpenAI model you would like to use.

Usage

Commands supported while in CLI:

QUIT_COMMAND = "quit()"
TOKEN_USAGE_COMMAND = "tokenusage"
FEW_SHOT_COMMAND = "fewshot"
NEW_CHAT_COMMAND = "newchat"
NEW_AI_COMMAND = "newai"
MULTI_LINE_COMMAND = "mlin"

• To input multiple lines at once, type mlin and then enter your message. Press ctrl + D to end the message.

usage: chatgpt-turbocli.py [-h] [--model MODEL] [--temperature TEMPERATURE] [--max_tokens MAX_TOKENS] [--summary_tokens SUMMARY_TOKENS]
                           [--completition_limit COMPLETITION_LIMIT] [--api_key API_KEY] [--log {DEBUG,INFO,WARNING,ERROR,CRITICAL}] [--transcript TRANSCRIPT]
                           [--logenabled LOGENABLED]

Chatbot CLI tool

options:
  -h, --help            show this help message and exit
  --model MODEL         The name of the model to be used
  --temperature TEMPERATURE
                        The temperature for the model
  --max_tokens MAX_TOKENS
                        The maximum amount of tokens
  --summary_tokens SUMMARY_TOKENS
                        The number of input tokens to start summarizing
  --completition_limit COMPLETITION_LIMIT
                        The max amount of tokens to be used for completition
  --api_key API_KEY     The OpenAI API key
  --log {DEBUG,INFO,WARNING,ERROR,CRITICAL}
                        Set the logging level
  --transcript TRANSCRIPT
                        Write a transcript on exit?
  --logenabled LOGENABLED
                        Logging enabled

License

This project is licensed under the Mozilla Public License 2.0.

References

Inspired by https://www.haihai.ai/chatgpt-api/