[Feature] Add known dimension

langchain4j-embeddings-all-minilm-l6-v2-q/src/main/java/dev/langchain4j/model/embedding/AllMiniLmL6V2QuantizedEmbeddingModel.java

@@ -25,4 +25,9 @@ public class AllMiniLmL6V2QuantizedEmbeddingModel extends AbstractInProcessEmbed
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 384;
+    }
 }

langchain4j-embeddings-all-minilm-l6-v2-q/src/test/java/dev/langchain4j/model/embedding/AllMiniLmL6V2QuantizedEmbeddingModelIT.java

@@ -115,4 +115,12 @@ void should_embed_concurrently() throws Exception {
             assertThat(embedding).isEqualTo(referenceEmbedding);
         }
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new AllMiniLmL6V2QuantizedEmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(384);
+    }
 }

langchain4j-embeddings-all-minilm-l6-v2/src/main/java/dev/langchain4j/model/embedding/AllMiniLmL6V2EmbeddingModel.java

@@ -25,4 +25,9 @@ public class AllMiniLmL6V2EmbeddingModel extends AbstractInProcessEmbeddingModel
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 384;
+    }
 }

langchain4j-embeddings-all-minilm-l6-v2/src/test/java/dev/langchain4j/model/embedding/AllMiniLmL6V2EmbeddingModelIT.java

@@ -96,4 +96,12 @@ void should_return_token_usage() {
 
         assertThat(model.embed("hi, how are you doing?").tokenUsage().inputTokenCount()).isEqualTo(7);
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new AllMiniLmL6V2EmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(384);
+    }
 }

langchain4j-embeddings-bge-small-en-q/src/main/java/dev/langchain4j/model/embedding/BgeSmallEnQuantizedEmbeddingModel.java

@@ -25,4 +25,9 @@ public class BgeSmallEnQuantizedEmbeddingModel extends AbstractInProcessEmbeddin
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 384;
+    }
 }

langchain4j-embeddings-bge-small-en-q/src/test/java/dev/langchain4j/model/embedding/BgeSmallEnQuantizedEmbeddingModelIT.java

@@ -68,4 +68,12 @@ void should_produce_normalized_vectors() {
         assertThat(magnitudeOf(model.embed(repeat(oneToken, 999)).content()))
                 .isCloseTo(1, withPercentage(0.01));
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new BgeSmallEnQuantizedEmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(384);
+    }
 }

langchain4j-embeddings-bge-small-en-v15-q/src/main/java/dev/langchain4j/model/embedding/bge/small/en/v15/BgeSmallEnV15QuantizedEmbeddingModel.java

@@ -29,4 +29,9 @@ public class BgeSmallEnV15QuantizedEmbeddingModel extends AbstractInProcessEmbed
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 384;
+    }
 }

langchain4j-embeddings-bge-small-en-v15-q/src/test/java/dev/langchain4j/model/embedding/bge/small/en/v15/BgeSmallEnV15QuantizedEmbeddingModelTest.java

@@ -69,4 +69,12 @@ void should_produce_normalized_vectors() {
         assertThat(magnitudeOf(model.embed(repeat(oneToken, 999)).content()))
                 .isCloseTo(1, withPercentage(0.01));
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new BgeSmallEnV15QuantizedEmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(384);
+    }
 }

langchain4j-embeddings-bge-small-en-v15/src/main/java/dev/langchain4j/model/embedding/bge/small/en/v15/BgeSmallEnV15EmbeddingModel.java

@@ -29,4 +29,9 @@ public class BgeSmallEnV15EmbeddingModel extends AbstractInProcessEmbeddingModel
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 384;
+    }
 }

langchain4j-embeddings-bge-small-en-v15/src/test/java/dev/langchain4j/model/embedding/bge/small/en/v15/BgeSmallEnV15EmbeddingModelTest.java

@@ -82,4 +82,12 @@ void should_produce_normalized_vectors() {
         assertThat(magnitudeOf(model.embed(repeat(oneToken, 999)).content()))
                 .isCloseTo(1, withPercentage(0.01));
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new BgeSmallEnV15EmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(384);
+    }
 }

langchain4j-embeddings-bge-small-en/src/main/java/dev/langchain4j/model/embedding/BgeSmallEnEmbeddingModel.java

@@ -25,4 +25,9 @@ public class BgeSmallEnEmbeddingModel extends AbstractInProcessEmbeddingModel {
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 384;
+    }
 }

langchain4j-embeddings-bge-small-en/src/test/java/dev/langchain4j/model/embedding/BgeSmallEnEmbeddingModelIT.java

@@ -81,4 +81,12 @@ void should_produce_normalized_vectors() {
         assertThat(magnitudeOf(model.embed(repeat(oneToken, 999)).content()))
                 .isCloseTo(1, withPercentage(0.01));
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new BgeSmallEnEmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(384);
+    }
 }

langchain4j-embeddings-bge-small-zh-q/src/main/java/dev/langchain4j/model/embedding/BgeSmallZhQuantizedEmbeddingModel.java

@@ -25,4 +25,9 @@ public class BgeSmallZhQuantizedEmbeddingModel extends AbstractInProcessEmbeddin
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 512;
+    }
 }

langchain4j-embeddings-bge-small-zh-q/src/test/java/dev/langchain4j/model/embedding/BgeSmallZhQuantizedEmbeddingModelIT.java

@@ -68,4 +68,12 @@ void should_produce_normalized_vectors() {
         assertThat(magnitudeOf(model.embed(repeat(oneToken, 999)).content()))
                 .isCloseTo(1, withPercentage(0.01));
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new BgeSmallZhQuantizedEmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(512);
+    }
 }

langchain4j-embeddings-bge-small-zh-v15-q/src/main/java/dev/langchain4j/model/embedding/bge/small/zh/v15/BgeSmallZhV15QuantizedEmbeddingModel.java

@@ -29,4 +29,9 @@ public class BgeSmallZhV15QuantizedEmbeddingModel extends AbstractInProcessEmbed
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 512;
+    }
 }

langchain4j-embeddings-bge-small-zh-v15/src/main/java/dev/langchain4j/model/embedding/bge/small/zh/v15/BgeSmallZhV15EmbeddingModel.java

@@ -29,4 +29,9 @@ public class BgeSmallZhV15EmbeddingModel extends AbstractInProcessEmbeddingModel
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 512;
+    }
 }

langchain4j-embeddings-bge-small-zh-v15/src/test/java/dev/langchain4j/model/embedding/bge/small/zh/v15/BgeSmallZhV15EmbeddingModelIT.java

@@ -82,4 +82,12 @@ void should_produce_normalized_vectors() {
         assertThat(magnitudeOf(model.embed(repeat(oneToken, 999)).content()))
                 .isCloseTo(1, withPercentage(0.01));
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new BgeSmallZhV15EmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(512);
+    }
 }

langchain4j-embeddings-bge-small-zh/src/main/java/dev/langchain4j/model/embedding/BgeSmallZhEmbeddingModel.java

@@ -25,4 +25,9 @@ public class BgeSmallZhEmbeddingModel extends AbstractInProcessEmbeddingModel {
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 512;
+    }
 }

langchain4j-embeddings-bge-small-zh/src/test/java/dev/langchain4j/model/embedding/BgeSmallZhEmbeddingModelIT.java

@@ -81,4 +81,12 @@ void should_produce_normalized_vectors() {
         assertThat(magnitudeOf(model.embed(repeat(oneToken, 999)).content()))
                 .isCloseTo(1, withPercentage(0.01));
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new BgeSmallZhEmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(512);
+    }
 }

langchain4j-embeddings-e5-small-v2-q/src/main/java/dev/langchain4j/model/embedding/E5SmallV2QuantizedEmbeddingModel.java

@@ -25,4 +25,9 @@ public class E5SmallV2QuantizedEmbeddingModel extends AbstractInProcessEmbedding
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 384;
+    }
 }

langchain4j-embeddings-e5-small-v2-q/src/test/java/dev/langchain4j/model/embedding/E5SmallV2QuantizedEmbeddingModelIT.java

@@ -81,4 +81,12 @@ void should_produce_normalized_vectors() {
         assertThat(magnitudeOf(model.embed(repeat(oneToken, 999)).content()))
                 .isCloseTo(1, withPercentage(0.01));
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new E5SmallV2QuantizedEmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(384);
+    }
 }

langchain4j-embeddings-e5-small-v2/src/main/java/dev/langchain4j/model/embedding/E5SmallV2EmbeddingModel.java

@@ -25,4 +25,9 @@ public class E5SmallV2EmbeddingModel extends AbstractInProcessEmbeddingModel {
     protected OnnxBertBiEncoder model() {
         return MODEL;
     }
+
+    @Override
+    protected Integer knownDimension() {
+        return 384;
+    }
 }

langchain4j-embeddings-e5-small-v2/src/test/java/dev/langchain4j/model/embedding/E5SmallV2EmbeddingModelIT.java

@@ -81,4 +81,12 @@ void should_produce_normalized_vectors() {
         assertThat(magnitudeOf(model.embed(repeat(oneToken, 999)).content()))
                 .isCloseTo(1, withPercentage(0.01));
     }
+
+    @Test
+    void should_return_correct_dimension() {
+
+        EmbeddingModel model = new E5SmallV2EmbeddingModel();
+
+        assertThat(model.dimension()).isEqualTo(384);
+    }
 }

langchain4j-embeddings/src/main/java/dev/langchain4j/model/embedding/AbstractInProcessEmbeddingModel.java

@@ -14,7 +14,7 @@
 
 import static java.nio.file.Files.newInputStream;
 
-public abstract class AbstractInProcessEmbeddingModel implements EmbeddingModel, TokenCountEstimator {
+public abstract class AbstractInProcessEmbeddingModel extends DimensionAwareEmbeddingModel implements TokenCountEstimator {
 
     protected static OnnxBertBiEncoder loadFromJar(String modelFileName, String tokenizerFileName, PoolingMode poolingMode) {
         InputStream model = Thread.currentThread().getContextClassLoader().getResourceAsStream(modelFileName);