mirror of
https://github.com/discourse/discourse-ai.git
synced 2025-10-25 19:48:42 +00:00
534b0df391
In a previous refactor, we moved the responsibility of querying and storing embeddings into the `Schema` class. Now, it's time for embedding generation. The motivation behind these changes is to isolate vector characteristics in simple objects to later replace them with a DB-backed version, similar to what we did with LLM configs.
41 lines
1.4 KiB
Ruby
41 lines
1.4 KiB
Ruby
# frozen_string_literal: true
|
|
|
|
RSpec.describe DiscourseAi::AiHelper::SemanticCategorizer do
|
|
fab!(:user)
|
|
fab!(:muted_category) { Fabricate(:category) }
|
|
fab!(:category_mute) do
|
|
CategoryUser.create!(
|
|
user: user,
|
|
category: muted_category,
|
|
notification_level: CategoryUser.notification_levels[:muted],
|
|
)
|
|
end
|
|
fab!(:muted_topic) { Fabricate(:topic, category: muted_category) }
|
|
fab!(:category)
|
|
fab!(:topic) { Fabricate(:topic, category: category) }
|
|
|
|
let(:vector) { DiscourseAi::Embeddings::Vector.instance }
|
|
let(:categorizer) { DiscourseAi::AiHelper::SemanticCategorizer.new({ text: "hello" }, user) }
|
|
let(:expected_embedding) { [0.0038493] * vector.vdef.dimensions }
|
|
|
|
before do
|
|
SiteSetting.ai_embeddings_enabled = true
|
|
SiteSetting.ai_embeddings_discourse_service_api_endpoint = "http://test.com"
|
|
SiteSetting.ai_embeddings_model = "bge-large-en"
|
|
|
|
WebMock.stub_request(
|
|
:post,
|
|
"#{SiteSetting.ai_embeddings_discourse_service_api_endpoint}/api/v1/classify",
|
|
).to_return(status: 200, body: JSON.dump(expected_embedding))
|
|
|
|
vector.generate_representation_from(topic)
|
|
vector.generate_representation_from(muted_topic)
|
|
end
|
|
|
|
it "respects user muted categories when making suggestions" do
|
|
category_ids = categorizer.categories.map { |c| c[:id] }
|
|
expect(category_ids).not_to include(muted_category.id)
|
|
expect(category_ids).to include(category.id)
|
|
end
|
|
end
|