bash-gpt is a bash extension that rewrites a natural language command into an actual command when you press a keyboard shortcut
bash-gpt only depends on sed, awk and curl
note: sorry! now fixed as of 2023/8/7 -- openai added whitespaces to their JSON which broke the regexp for parsing it
sudo ./install.sh /usr/local ~/.bashrc " " <YOUR_OPENAI_API_KEY>
or
./install.sh ~ ~/.bashrc " " <YOUR_OPENAI_API_KEY>
./uninstall.sh [BASHRC]
if BASHRC is not specified, it will attempt to uninstall using the BASHRC in your BASHGPT_BASHRC environment variable
once installed, open a new shell and type a natural language command
$ find all executable files in /usr created within the last day
press Ctrl-[space]
$ find /usr -type f -executable -mtime -1
/usr/local/bash-gpt/bin/init
/usr/local/bash-gpt/bin/gpt
| natural language | generated command |
|---|---|
| replace every instance of foo with baz in my_file.txt | sed -i 's/foo/baz/g' my_file.txt |
| flip my_video.mp4 vertically, scale it to 720p and save it as my_output.mp4 | ffmpeg -i my_video.mp4 -vf "vflip,scale=720:-1" my_output.mp4 |
Usage: ./install.sh PREFIX BASHRC SHORTCUT_KEY OPENAI_KEY [MODEL] [TEMPERATURE] [MAX_TOKENS] [TEMPLATE]
| usage | description | example/default value |
|---|---|---|
| PREFIX | The path to install bash-gpt | /usr/local |
| BASHRC | The path to your .bashrc file |
~/.bashrc |
| SHORTCUT_KEY | The key to trigger bash-gpt (CTRL-KEY) | " " |
| OPENAI_KEY | The key to access OpenAI API | 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 |
| MODEL | The model to use | text-davinci-003 |
| TEMPERATURE | The temperature to use | 0 |
| MAX_TOKENS | The maximum number of tokens to generate (smaller is faster) | 100 |
| TEMPLATE | The prompt template to use ({{TEXT}} is replaced with the natural language query) |
here is the bash command in a code block: {{TEXT}} |
to make changes to the default MODEL, TEMPERATURE, MAX_TOKENS or TEMPLATE you can run the install.sh again and it will update your .bashrc
you can make temporary changes to these parameters like so
$ export BASHGPT_MODEL=<MODEL>
$ export BASHGPT_TEMPERATURE=<TEMPERATURE>
$ export BASHGPT_MAX_TOKENS=<MAX_TOKENS>
$ export BASHGPT_TEMPLATE=<TEMPLATE>
| template | description | max tokens | notes |
|---|---|---|---|
| text-davinci-003 | Can do any language task with better quality, longer output, and consistent instruction-following than the curie, babbage, or ada models. Also supports inserting completions within text. | 4,097 tokens | default, fastest |
| text-davinci-002 | Similar capabilities to text-davinci-003 but trained with supervised fine-tuning instead of reinforcement learning | 4,097 tokens | not great for code |
| gpt-3.5-turbo | Most capable GPT-3.5 model and optimized for chat at 1/10th the cost of text-davinci-003. Will be updated with our latest model iteration. | 4,096 tokens | good model, but slow |
| gpt-4 | More capable than any GPT-3.5 model, able to do more complex tasks, and optimized for chat. Will be updated with our latest model iteration. | 8,192 tokens | slow, not available to all users |
| gpt-4-32k | Same capabilities as the base gpt-4 mode but with 4x the context length. Will be updated with our latest model iteration. | 32,768 tokens | slow, not available to all users |
| code-cushman-001 | Almost as capable as Davinci Codex, but slightly faster. This speed advantage may make it preferable for real-time applications. | 2,048 tokens | fastest model, but deprecated as of march 23rd 2023 |
| code-davinci-002 | Most capable Codex model. Particularly good at translating natural language to code. In addition to completing code, also supports inserting completions within code. | 8,001 tokens | best code model, but deprecated as of march 23rd 2023 |