Compared to the ChatGPT website, this tool is a bit more configurable, is only a few hundred lines of code in one module so is easy to understand/follow/modify, and it allows for use of GPT-4 and plug-in options (like inserting weblinks without paying the hefty monthly fee for ChatGPT-Plus). Note it does involve paying fees for the OpenAI API calls - but for one's individual use rather than public internet app use by many people, even at heavy usage the API fees total far less per month than the website. This app can be run either as a terminal/readline based CLI or as a browser based web-app:
Get an API key (requires
creating an OpenAI account if you don't have one already) and put that key
in your shell environment, e.g. export OPENAI_API_KEY=xyz123xyz123...
Note that any client outside of the one on OpenAI's website (which is free for ChatGPT3.5) requires an API key to access the API, and making calls to this API does cost money. However, it's quite cheap even at heavy usage if just using it oneself, as opposed to serving it out to zillions of users to use in a web app. In my initial two days I racked up a whole $0.20; now in heavy daily usage for coding templating, database queries, and other technical assistance in my work I've now gotten up to almost $7/month (that's with that latest gpt-4-1106-preview model).
As a new user of the OpenAI platform you may initially only have access to the gpt-3.5-turbo model, not gpt-4. (This might be better by the time you read this.) I found that I was able to get access to gpt-4 API use myself by first signing up onto its waitlist on the OpenAI website, then signing up for just one month of the $20/month ChatGPT-Plus program on the website, cancelling that after one month, and a week or so later they sent me an email saying my waitlist position had come up because they prioritized folks who'd paid for at least one month of ChatGPT-Plus. I guess it's no guarantee (they hadn't stated this on the waitlist website) but maybe info that other potential users here would want to know.
This app can be run either as a terminal/readline based ChatGPT CLI app or as a browser-based web-app:
Usage:
python3 gpt_client.py cli --model=gpt-4 --temperature=0.2 --prompt="Me: "
python3 gpt_client.py webapp --model=gpt-4 --temperature=0.2 . You must run this in an environment with the following Python packages (as installed from the requirements.txt file):
- openai (for the core Open API calls)
- beautifulsoup4 (for reading contents of urls)
- rich (for the markdown/syntax-highlighting formatting in CLI)
- darkdetect (for putting code syntax or webpage into light/dark mode in CLI)
- gradio (for the quickie webapp interface)
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
export OPENAI_API_KEY=xyz123xyz123...
python gpt_client.pyThe command line (CLI) client uses readline, so all the usual hot keys /
editing work including a command history via Ctrl-P and Ctrl-N. Quit via:
quit or exit or q or ctrl-D or ctrl-C.
The web-app is a quickie implemention via Gradio - after starting the app it'll output the local URL to paste into your web browser to access the app.
Lastly know that OpenAI's GPT API is an immensely popular service right now and that regardless of which client you use (one of these, or the ChatGPT website one, or any other), expect that there are definitely slow times. I've experienced response wait times of up to 20sec and once-in-a-while responses of "too busy right now". But usually the responses are within a few seconds. These waits/too-busy are not about the client apps but about the API service.
Model choices/explanations at:
https://learn.microsoft.com/en-us/azure/cognitive-services/openai/concepts/models
More detailed usage instructions for these models in terms of crafting queries:
Details about additional chatgpt parameters for tuning functionality:
https://platform.openai.com/docs/api-reference/chat/create
Thoughts for values of temperature and top_p for different use-cases:
Arbitrary example table below comes from a forum entry at: https://community.openai.com/t/cheat-sheet-mastering-temperature-and-top-p-in-chatgpt-api-a-few-tips-and-tricks-on-controlling-the-creativity-deterministic-output-of-prompt-responses/172683 Take these contents with a grain of salt - eg note OpenAI ref docs suggest only changing Temp and Top_p independently - but experiment and see what works!
| Use Case | Temp | Top_p | Description |
|---|---|---|---|
| Code Generation | 0.2 | 0.1 | Generates code that adheres to established patterns and conventions. Output is more deterministic and focused. Useful for generating syntactically correct code. |
| Creative Writing | 0.7 | 0.8 | Generates creative and diverse text for storytelling. Output is more exploratory and less constrained by patterns. |
| Chatbot Responses | 0.5 | 0.5 | Generates conversational responses that balance coherence and diversity. Output is more natural and engaging. |
| Code Comment Generation | 0.3 | 0.2 | Generates code comments that are more likely to be concise and relevant. Output is more deterministic and adheres to conventions. |
| Data Analysis Scripting | 0.2 | 0.1 | Generates data analysis scripts that are more likely to be correct and efficient. Output is more deterministic and focused. |
| Exploratory Code Writing | 0.6 | 0.7 | Generates code that explores alternative solutions and creative approaches. Output is less constrained by established patterns. |