[ENH]: Added ability to use a diffferent API endpoint for OpenAI Embedding Function

clients/new-js/packages/ai-embeddings/openai/src/index.ts

@@ -22,6 +22,7 @@ export interface OpenAIArgs {
   organizationId?: string;
   dimensions?: number;
   apiKey?: string;
+  apiBase?: string;
 }
 
 export class OpenAIEmbeddingFunction implements EmbeddingFunction {
@@ -30,6 +31,7 @@ export class OpenAIEmbeddingFunction implements EmbeddingFunction {
   private readonly modelName: string;
   private readonly dimensions: number | undefined;
   private readonly organizationId: string | undefined;
+  private readonly apiBase: string | undefined;
   private client: OpenAI;
 
   constructor(args: OpenAIArgs) {
@@ -38,6 +40,7 @@ export class OpenAIEmbeddingFunction implements EmbeddingFunction {
       modelName,
       dimensions,
       organizationId,
+      apiBase = "",
     } = args;
 
     const apiKey = args.apiKey || process.env[apiKeyEnvVar];
@@ -51,8 +54,9 @@ export class OpenAIEmbeddingFunction implements EmbeddingFunction {
     this.organizationId = organizationId;
     this.apiKeyEnvVar = apiKeyEnvVar;
     this.dimensions = dimensions;
+    this.apiBase = apiBase;
 
-    this.client = new OpenAI({ apiKey, organization: this.organizationId });
+    this.client = new OpenAI({ apiKey, organization: this.organizationId, baseURL: this.apiBase });
   }
 
   public async generate(texts: string[]): Promise<number[][]> {

docs/docs.trychroma.com/markdoc/content/integrations/embedding-models/openai.md

@@ -76,6 +76,18 @@ collection = await client.getCollection({
 });
 ```
 
+To use the OpenAI embedding models on other platforms such as GitHub AI Models, you can use the `apiBase` parameter:
+```typescript
+import { OpenAIEmbeddingFunction } from "@chroma-core/openai";
+
+const embeddingFunction = new OpenAIEmbeddingFunction({
+  apiKey: "apiKey",
+  apiBase: "https://models.github.ai/inference",
+  modelName: "text-embedding-3-small",
+});
+```
+
+
 {% /Tab %}
 
 {% /Tabs %}