FEATURE: Add support for mxbai-embed-xsmall-v1 embedding model

config/settings.yml

@@ -308,6 +308,7 @@ discourse_ai:
       - text-embedding-3-small
       - text-embedding-3-large
       - multilingual-e5-large
+      - mxbai-embed-xsmall-v1
       - bge-large-en
       - gemini
       - bge-m3

lib/embeddings/vector_representations/base.rb

@@ -14,6 +14,7 @@ def find_representation(model_name)
               DiscourseAi::Embeddings::VectorRepresentations::BgeM3,
               DiscourseAi::Embeddings::VectorRepresentations::Gemini,
               DiscourseAi::Embeddings::VectorRepresentations::MultilingualE5Large,
+              DiscourseAi::Embeddings::VectorRepresentations::MxbaiEmbedXsmallV1,
               DiscourseAi::Embeddings::VectorRepresentations::TextEmbedding3Large,
               DiscourseAi::Embeddings::VectorRepresentations::TextEmbedding3Small,
               DiscourseAi::Embeddings::VectorRepresentations::TextEmbeddingAda002,

lib/embeddings/vector_representations/mxbai_embed_xsmall_v1.rb

@@ -0,0 +1,64 @@
+# frozen_string_literal: true
+
+module DiscourseAi
+  module Embeddings
+    module VectorRepresentations
+      class MxbaiEmbedXsmallV1 < Base
+        class << self
+          def name
+            "mxbai-embed-xsmall-v1"
+          end
+
+          def correctly_configured?
+            SiteSetting.ai_embeddings_discourse_service_api_endpoint_srv.present? ||
+              SiteSetting.ai_embeddings_discourse_service_api_endpoint.present?
+          end
+
+          def dependant_setting_names
+            %w[
+              ai_embeddings_discourse_service_api_key
+              ai_embeddings_discourse_service_api_endpoint_srv
+              ai_embeddings_discourse_service_api_endpoint
+            ]
+          end
+        end
+
+        def vector_from(text, asymetric: false)
+          inference_client.perform!(text)
+        end
+
+        def dimensions
+          384
+        end
+
+        def max_sequence_length
+          512
+        end
+
+        def id
+          1
+        end
+
+        def version
+          1
+        end
+
+        def pg_function
+          "<#>"
+        end
+
+        def pg_index_type
+          "halfvec_ip_ops"
+        end
+
+        def tokenizer
+          DiscourseAi::Tokenizer::MxbaiEmbedXsmallV1Tokenizer
+        end
+
+        def inference_client
+          DiscourseAi::Inference::DiscourseClassifier.instance(self.class.name)
+        end
+      end
+    end
+  end
+end

lib/tokenizer/mxbai_embed_xsmall_v1_tokenizer.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module DiscourseAi
+  module Tokenizer
+    class MxbaiEmbedXsmallV1Tokenizer < BasicTokenizer
+      def self.tokenizer
+        @@tokenizer ||=
+          Tokenizers.from_file("./plugins/discourse-ai/tokenizers/mxbai-embed-xsmall-v1.json")
+      end
+    end
+  end
+end