Discourse/discourse
Discourse/discourse
0
0
Fork 0
mirror of https://gh.wpcy.net/https://github.com/discourse/discourse.git synced 2026-05-24 18:43:51 +08:00
Code Packages Activity Actions
discourse/plugins/discourse-ai
History
Exact
Roman Rizzi a27e20c300
PERF: Speed up embedding text preparation. (#33791) 
When collecting text for vectorizing a topic, we iterate over as many
posts as possible within the context window, parsing their cooked
attribute using Nokogiri. We noticed this method doesn't scale well when
working with larger contexts.

Instead, we'll collect as much unparsed cooked text as we can, then
parse it all in a single Nokogiri call.

I ran this a hundred times in a benchmark, and the perf gains are
significant:

```
user     system total        real
prepare_target_text:           114.887620   3.731693 118.619313 (118.952465)
prepare_target_text_bis:        10.264950   0.186204  10.451154 ( 10.465957)
```

Tried running it 1k times, but the old method took too long.
2025-07-23 13:52:48 -03:00
..
admin/assets/javascripts/discourse DEV: Reapply gjs-codemod in d-ai (#33758) 2025-07-23 12:05:40 +02:00
app
assets DEV: Reapply gjs-codemod in d-ai (#33758) 2025-07-23 12:05:40 +02:00
config Update translations (#33788) 2025-07-23 12:39:43 +02:00
db DEV: Update migration to add index to run first (#33782) 2025-07-23 15:44:56 +08:00
discourse_automation
evals
lib PERF: Speed up embedding text preparation. (#33791) 2025-07-23 13:52:48 -03:00
public/ai-share
spec DEV: Reapply gjs-codemod in d-ai (#33758) 2025-07-23 12:05:40 +02:00
svg-icons
test/javascripts
.prettierignore
about.json
plugin.rb
README.md

README.md

Discourse AI Plugin

Plugin Summary

For more information, please see: https://meta.discourse.org/t/discourse-ai/259214?u=falco

Evals

The directory evals contains AI evals for the Discourse AI plugin. You may create a local config by copying config/eval-llms.yml to config/eval-llms.local.yml and modifying the values.

To run them use:

cd evals ./run --help

Usage: evals/run [options]
    -e, --eval NAME                  Name of the evaluation to run
        --list-models                List models
    -m, --model NAME                 Model to evaluate (will eval all models if not specified)
    -l, --list                       List evals

To run evals you will need to configure API keys in your environment:

OPENAI_API_KEY=your_openai_api_key ANTHROPIC_API_KEY=your_anthropic_api_key GEMINI_API_KEY=your_gemini_api_key