Installation
In this page, we will guide through you how to install FixML on your computer, and get started to using it for evaluating your code projects!
Prerequisites
FixML is written to support Python version 3.12 or later. Make sure that you have Python with version 3.12 or later installed in your system.
Note
Get Python on their official website.
Alternatively, you can use the ones that come with your Operating System, or come from a virtual environment managers such as virtualenv and/or conda.
Getting the package from PyPI
fixml package is hosted
on the Python Package Index (PyPI). You can visit the
project page on PyPI here.
To install this, enter this command in your terminal:
pip install fixml
# or
python3 -m pip install fixml
Configuring API Keys
As of version 0.1.0, the only supported connector to LLMs is the OpenAI API. Therefore, for any workload that involves calls to LLMs, an API key for accessing OpenAI API is required.
Getting the API Key
You can refer to OpenAI’s page on obtaining API keys.
Warning
The API keys are credentials that should be treated the same way you treat passwords. Refer to OpenAI’s page on best practices for keeping your API key safe.
Note
FixML only use the API key stored locally in your system for calls to LLMs, and will not save, transmit, or leak the API keys in any way.
Increasing Quota
Note
The free trial version of OpenAI API comes with many limits, such as a very low ceiling on the token per minute quota.
Although optimizations have been done to reduce the usage on the tokens, FixML still need to transmit a significant portion of the code base when conducting the analysis.
As such, for most code bases, the free trial version of the API is unsuitable for use and would often result in errors stating rate limit has been reached.
To prevent FixML from hitting rate limit error, user should upgrade their API to the paid version and thus raising the rate limit to remove this restriction.
Based on OpenAI’s documentation, we estimate you must be at least at Tier 1 to be able to use our tool without the frequent intermittent rate limit errors.
Refer to OpenAI’s documentations on rate limits and quotas and account limits.
Saving the API Key into your system
Once obtained the API key, the key needs to be stored in your system in order for FixML to be able to discover it and subsequently use it for calling the OpenAI service endpoints.
Currently, FixML will look for such key through the use of Environment Variables.
There are two ways to do this:
1. Saving it Directly as an Environment Variable
This way we will directly save the API key into the Operating System’s set of environment variables.
Note
Advantage: Since it is saved as an Environment Variable in your Operating
System, it is accessible by fixml from any directory you are working on.
Disadvantage: This setting is transient and will not persist after a system
reboot. To make this change permanent, you can add the export command below as
a part of start up script.
Unix-like systems (Linux, MacOS, etc)
Run this in your console/terminal emulator:
export OPENAI_API_KEY={your-openai-api-key}
After running the command, confirm that the variable has been saved into the system:
export | grep OPENAI_API_KEY
Windows
Run this in your console:
SET OPENAI_API_KEY={your-openai-api-key}After running the command, confirm that the variable has been saved into the system:
SET OPENAI_API_KEY
2. Saving it inside an .env File
This method will save your API key into a file named .env. This will not
be directly injecting your API key as an environment variable into your
computer. Rather, when FixML is being run, it will look for file named . env file in the current working directory and inject the content inside
as a set of temporary Environment Variables.
Note
Advantage: Persistent storage, and will survive reboots as this is a file.
Disadvantage: Since it depends on the .env file’s location, it is not
runnable outside the directory where there is no such .env file.
To do this, run the follow command:
echo OPENAI_API_KEY={your-openai-api-key} > .env