This commit is contained in:
Paul Gauthier 2024-06-17 11:44:44 -07:00
parent 3dc3ebe25f
commit b8bb33fe25
4 changed files with 10 additions and 10 deletions

View file

@ -46,7 +46,7 @@
- dirname: 2024-06-17-14-45-54--deepseek-coder2-whole - dirname: 2024-06-17-14-45-54--deepseek-coder2-whole
test_cases: 133 test_cases: 133
model: Deepseek Coder V2 model: DeepSeek Coder V2
edit_format: whole edit_format: whole
commit_hash: ca8672b commit_hash: ca8672b
pass_rate_1: 63.9 pass_rate_1: 63.9

View file

@ -16,13 +16,13 @@ While [aider can connect to almost any LLM](/docs/llms.html),
it works best with models that score well on the benchmarks. it works best with models that score well on the benchmarks.
## Deepseek Coder V2 beats GPT-4o, Opus ## DeepSeek Coder V2 beats GPT-4o, Opus
The new The new
[Deepseek Coder V2](https://aider.chat/docs/llms/deepseek.html) [DeepSeek Coder V2](https://aider.chat/docs/llms/deepseek.html)
model is now atop aider's code editing leaderboard! model is now atop aider's code editing leaderboard!
It's worth noting that Deepseek Coder V2 is only capable of using aider's "whole" edit format. It's worth noting that DeepSeek Coder V2 is only capable of using aider's "whole" edit format.
This means it returns a modified full copy of each file when it makes changes. This means it returns a modified full copy of each file when it makes changes.
Most other strong models are able to use aider's "diff" editing format, Most other strong models are able to use aider's "diff" editing format,
which allows them to return diffs of edits -- saving time and token costs. which allows them to return diffs of edits -- saving time and token costs.

View file

@ -23,7 +23,7 @@ as they are the very best models for editing code.
Aider works with a number of **free** API providers: Aider works with a number of **free** API providers:
- The [Deepseek Coder V2](/docs/llms/deepseek.html) model gets the top score on aider's code editing benchmark. Deepseek currently offers 5M free tokens of API usage. - The [DeepSeek Coder V2](/docs/llms/deepseek.html) model gets the top score on aider's code editing benchmark. DeepSeek currently offers 5M free tokens of API usage.
- Google's [Gemini 1.5 Pro](/docs/llms/gemini.html) is the most capable free model to use with aider, with - Google's [Gemini 1.5 Pro](/docs/llms/gemini.html) is the most capable free model to use with aider, with
code editing capabilities similar to GPT-3.5. code editing capabilities similar to GPT-3.5.
- You can use [Llama 3 70B on Groq](/docs/llms/groq.html) which is comparable to GPT-3.5 in code editing performance. - You can use [Llama 3 70B on Groq](/docs/llms/groq.html) which is comparable to GPT-3.5 in code editing performance.

View file

@ -3,11 +3,11 @@ parent: Connecting to LLMs
nav_order: 500 nav_order: 500
--- ---
# Deepseek # DeepSeek
Aider can connect to the Deepseek.com API. Aider can connect to the DeepSeek.com API.
The Deepseek Coder V2 model gets the top score on aider's code editing benchmark. The DeepSeek Coder V2 model gets the top score on aider's code editing benchmark.
Deepseek appears to grant 5M tokens of free API usage to new accounts. DeepSeek appears to grant 5M tokens of free API usage to new accounts.
``` ```
pip install aider-chat pip install aider-chat
@ -15,7 +15,7 @@ pip install aider-chat
export DEEPSEEK_API_KEY=<key> # Mac/Linux export DEEPSEEK_API_KEY=<key> # Mac/Linux
setx DEEPSEEK_API_KEY <key> # Windows setx DEEPSEEK_API_KEY <key> # Windows
# Use Deepseek Coder V2 # Use DeepSeek Coder V2
aider --model deepseek/deepseek-coder aider --model deepseek/deepseek-coder
``` ```