Merge pull request #227 from weaviate/modules/add-support-for-multi2vec-cohere

Add `configure.multi2VecCohere` factory method and types for the module

Author: tsmith023

src/collections/config/types/vectorizer.ts

@@ -20,6 +20,7 @@ type Text2VecPalmVectorizer = 'text2vec-palm';
 export type Vectorizer =
   | 'img2vec-neural'
   | 'multi2vec-clip'
+  | 'multi2vec-cohere'
   | 'multi2vec-bind'
   | Multi2VecPalmVectorizer
   | 'multi2vec-google'
@@ -81,6 +82,33 @@ export type Multi2VecClipConfig = {
   };
 };
 
+/**
+ * The configuration for multi-media vectorization using the Cohere module.
+ *
+ * See the [documentation](https://weaviate.io/developers/weaviate/model-providers/cohere/embeddings-multimodal) for detailed usage.
+ */
+export type Multi2VecCohereConfig = {
+  /** The base URL to use where API requests should go. */
+  baseURL?: string;
+  /** The image fields used when vectorizing. */
+  imageFields?: string[];
+  /** The specific model to use. */
+  model?: string;
+  /** The text fields used when vectorizing. */
+  textFields?: string[];
+  /** The truncation strategy to use. */
+  truncate?: string;
+  /** Whether the collection name is vectorized. */
+  vectorizeCollectionName?: boolean;
+  /** The weights of the fields used for vectorization. */
+  weights?: {
+    /** The weights of the image fields. */
+    imageFields?: number[];
+    /** The weights of the text fields. */
+    textFields?: number[];
+  };
+};
+
 /** The configuration for multi-media vectorization using the Bind module.
  *
  * See the [documentation](https://weaviate.io/developers/weaviate/model-providers/imagebind/embeddings-multimodal) for detailed usage.
@@ -421,6 +449,8 @@ export type VectorizerConfigType<V> = V extends 'img2vec-neural'
   ? Img2VecNeuralConfig | undefined
   : V extends 'multi2vec-clip'
   ? Multi2VecClipConfig | undefined
+  : V extends 'multi2vec-cohere'
+  ? Multi2VecCohereConfig | undefined
   : V extends 'multi2vec-bind'
   ? Multi2VecBindConfig | undefined
   : V extends 'multi2vec-google'

src/collections/configure/types/vectorizer.ts

@@ -111,6 +111,22 @@ export type Multi2VecBindConfigCreate = {
   vectorizeCollectionName?: boolean;
 };
 
+/** The configuration for the `multi2vec-cohere` vectorizer. */
+export type Multi2VecCohereConfigCreate = {
+  /** The base URL to use where API requests should go. */
+  baseURL?: string;
+  /** The image fields to use in vectorization. Can be string of `Multi2VecField` type. If string, weight 0 will be assumed. */
+  imageFields?: string[] | Multi2VecField[];
+  /** The specific model to use. */
+  model?: string;
+  /** The text fields to use in vectorization. Can be string of `Multi2VecField` type. If string, weight 0 will be assumed. */
+  textFields?: string[] | Multi2VecField[];
+  /** The truncation strategy to use. */
+  truncate?: string;
+  /** Whether to vectorize the collection name. */
+  vectorizeCollectionName?: boolean;
+};
+
 /** @deprecated Use `Multi2VecGoogleConfigCreate` instead.*/
 export type Multi2VecPalmConfigCreate = Multi2VecGoogleConfigCreate;
 
@@ -173,6 +189,8 @@ export type VectorizerConfigCreateType<V> = V extends 'img2vec-neural'
   ? Img2VecNeuralConfigCreate | undefined
   : V extends 'multi2vec-clip'
   ? Multi2VecClipConfigCreate | undefined
+  : V extends 'multi2vec-cohere'
+  ? Multi2VecCohereConfigCreate | undefined
   : V extends 'multi2vec-bind'
   ? Multi2VecBindConfigCreate | undefined
   : V extends 'multi2vec-palm'

src/collections/configure/unit.test.ts

@@ -259,6 +259,79 @@ describe('Unit testing of the vectorizer factory class', () => {
     });
   });
 
+  it('should create the correct Multi2VecCohereConfig type with defaults', () => {
+    const config = configure.vectorizer.multi2VecCohere();
+    expect(config).toEqual<VectorConfigCreate<never, undefined, 'hnsw', 'multi2vec-cohere'>>({
+      name: undefined,
+      vectorIndex: {
+        name: 'hnsw',
+        config: undefined,
+      },
+      vectorizer: {
+        name: 'multi2vec-cohere',
+        config: undefined,
+      },
+    });
+  });
+
+  it('should create the correct Multi2VecCohereConfig type with all values', () => {
+    const config = configure.vectorizer.multi2VecCohere({
+      name: 'test',
+      model: 'model',
+      vectorizeCollectionName: true,
+    });
+    expect(config).toEqual<VectorConfigCreate<never, 'test', 'hnsw', 'multi2vec-cohere'>>({
+      name: 'test',
+      vectorIndex: {
+        name: 'hnsw',
+        config: undefined,
+      },
+      vectorizer: {
+        name: 'multi2vec-cohere',
+        config: {
+          model: 'model',
+          vectorizeCollectionName: true,
+        },
+      },
+    });
+  });
+
+  it('should create the correct Multi2VecCohereConfig type with all values and weights', () => {
+    const config = configure.vectorizer.multi2VecCohere({
+      name: 'test',
+      model: 'model',
+      imageFields: [
+        { name: 'field1', weight: 0.1 },
+        { name: 'field2', weight: 0.2 },
+      ],
+      textFields: [
+        { name: 'field3', weight: 0.3 },
+        { name: 'field4', weight: 0.4 },
+      ],
+      vectorizeCollectionName: true,
+    });
+    expect(config).toEqual<VectorConfigCreate<never, 'test', 'hnsw', 'multi2vec-cohere'>>({
+      name: 'test',
+      vectorIndex: {
+        name: 'hnsw',
+        config: undefined,
+      },
+      vectorizer: {
+        name: 'multi2vec-cohere',
+        config: {
+          model: 'model',
+          imageFields: ['field1', 'field2'],
+          textFields: ['field3', 'field4'],
+          vectorizeCollectionName: true,
+          weights: {
+            imageFields: [0.1, 0.2],
+            textFields: [0.3, 0.4],
+          },
+        },
+      },
+    });
+  });
+
   it('should create the correct Multi2VecClipConfig type with defaults', () => {
     const config = configure.vectorizer.multi2VecClip();
     expect(config).toEqual<VectorConfigCreate<never, undefined, 'hnsw', 'multi2vec-clip'>>({

src/collections/configure/vectorizer.ts

@@ -129,6 +129,39 @@ export const vectorizer = {
       },
     });
   },
+  /**
+   * Create a `VectorConfigCreate` object with the vectorizer set to `'multi2vec-cohere'`.
+   *
+   * See the [documentation](https://weaviate.io/developers/weaviate/model-providers/cohere/embeddings) for detailed usage.
+   *
+   * @param {ConfigureNonTextVectorizerOptions<N, I, 'multi2vec-cohere'>} [opts] The configuration options for the `multi2vec-cohere` vectorizer.
+   * @returns {VectorConfigCreate<PrimitiveKeys<T>[], N, I, 'multi2vec-cohere'>} The configuration object.
+   */
+  multi2VecCohere: <N extends string | undefined = undefined, I extends VectorIndexType = 'hnsw'>(
+    opts?: ConfigureNonTextVectorizerOptions<N, I, 'multi2vec-cohere'>
+  ): VectorConfigCreate<never, N, I, 'multi2vec-cohere'> => {
+    const { name, vectorIndexConfig, ...config } = opts || {};
+    const imageFields = config.imageFields?.map(mapMulti2VecField);
+    const textFields = config.textFields?.map(mapMulti2VecField);
+    let weights: Multi2VecBindConfig['weights'] = {};
+    weights = formatMulti2VecFields(weights, 'imageFields', imageFields);
+    weights = formatMulti2VecFields(weights, 'textFields', textFields);
+    return makeVectorizer(name, {
+      vectorIndexConfig,
+      vectorizerConfig: {
+        name: 'multi2vec-cohere',
+        config:
+          Object.keys(config).length === 0
+            ? undefined
+            : {
+                ...config,
+                imageFields: imageFields?.map((f) => f.name),
+                textFields: textFields?.map((f) => f.name),
+                weights: Object.keys(weights).length === 0 ? undefined : weights,
+              },
+      },
+    });
+  },
   /**
    * Create a `VectorConfigCreate` object with the vectorizer set to `'multi2vec-clip'`.
    *