dearwolf/aider
1
0
Fork 0
mirror of https://github.com/Aider-AI/aider.git synced 2025-06-02 10:45:00 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 5

copy

Browse source
This commit is contained in:
Paul Gauthier 2023-06-14 06:21:00 -07:00
parent b7d229ac03
commit 91fed81c31
1 changed files with 13 additions and 10 deletions
Download patch file Download diff file Expand all files Collapse all files

23
README.md
View file

@ -1,7 +1,7 @@
# aider is GPT powered coding in your terminal # aider is GPT powered coding in your terminal
`aider` is a command-line chat tool that allows you to write and edit `aider` is a command-line chat tool that allows you to write and edit
code with gpt-4 or gpt-3.5-turbo. You can ask GPT to help you start code with OpenAI's GPT models. You can ask GPT to help you start
a new project, or modify code in your existing git repo. a new project, or modify code in your existing git repo.
Aider makes it easy to git commit, diff & undo changes proposed by GPT without copy/pasting. Aider makes it easy to git commit, diff & undo changes proposed by GPT without copy/pasting.
It also has features that [help GPT-4 understand and modify larger codebases](https://aider.chat/docs/ctags.html). It also has features that [help GPT-4 understand and modify larger codebases](https://aider.chat/docs/ctags.html).
@ -60,17 +60,20 @@ You can find more chat transcripts on the [examples page](https://aider.chat/exa
Aider supports Aider supports
`gpt-4`, `gpt-4`,
`gpt-4-32k` `gpt-4-32k`,
`gpt-3.5-turbo`, `gpt-3.5-turbo`
and the the brand new `gpt-3.5-turbo-16k` model. and the the brand new `gpt-3.5-turbo-16k` model.
You will probably get the best results with GPT-4, because of its large context window and You will probably get the best results with GPT-4,
greater competance at code editing. because of its large context window and
greater competance at coding
which will allow you to edit larger codebases.
The GPT-3.5 models are less able to follow instructions for The GPT-3.5 models are limited to editing somewhat smaller codebases.
returning code edits in a diff-like format. They are less able to follow instructions for
So aider needs to send GPT-3.5 the original code returning code edits in a compact format,
and ask it to return a full copy of the modified code. so aider has
Both the original and modified copies of the code count towards the context window token limit. and ask GPT-3.5 to return a full copy of any code that needs to be edited.
This rapidly uses up the context window.
In practice, this means you can only use `gpt-3.5-turbo` to edit files that are In practice, this means you can only use `gpt-3.5-turbo` to edit files that are
smaller than about 2k tokens (8k bytes). smaller than about 2k tokens (8k bytes).
The new `gpt-3.5-turbo-16k` model should be able to edit code up to 8k tokens (32k bytes). The new `gpt-3.5-turbo-16k` model should be able to edit code up to 8k tokens (32k bytes).