Revert "feat: Add support for VoyageAI embeddings API"

.env.example

@@ -211,12 +211,6 @@ SMALL_ANTHROPIC_MODEL=          # Default: claude-3-haiku-20240307
 MEDIUM_ANTHROPIC_MODEL=         # Default: claude-3-5-sonnet-20241022
 LARGE_ANTHROPIC_MODEL=          # Default: claude-3-5-sonnet-20241022
 
-# VoyageAI Configuration
-VOYAGEAI_API_KEY=
-USE_VOYAGEAI_EMBEDDING=         # Set to TRUE for VoyageAI, leave blank for local
-VOYAGEAI_EMBEDDING_MODEL=       # Default: voyage-3-lite
-VOYAGEAI_EMBEDDING_DIMENSIONS=  # Default: 512
-
 # Heurist Configuration
 HEURIST_API_KEY=                # Get from https://heurist.ai/dev-access
 SMALL_HEURIST_MODEL=            # Default: meta-llama/llama-3-70b-instruct

packages/core/src/embedding.ts

@@ -1,9 +1,7 @@
-import path from "node:path";
+import { getEmbeddingModelSettings, getEndpoint } from "./models.ts";
+import { IAgentRuntime, ModelProviderName } from "./types.ts";
 import settings from "./settings.ts";
 import elizaLogger from "./logger.ts";
-import { getVoyageAIEmbeddingConfig } from "./voyageai.ts";
-import { models, getEmbeddingModelSettings, getEndpoint } from "./models.ts";
-import { IAgentRuntime, ModelProviderName } from "./types.ts";
 import LocalEmbeddingModelManager from "./localembeddingManager.ts";
 
 interface EmbeddingOptions {
@@ -22,93 +20,63 @@ export const EmbeddingProvider = {
     GaiaNet: "GaiaNet",
     Heurist: "Heurist",
     BGE: "BGE",
-    VoyageAI: "VoyageAI",
 } as const;
 
 export type EmbeddingProviderType =
     (typeof EmbeddingProvider)[keyof typeof EmbeddingProvider];
 
-export namespace EmbeddingProvider {
-    export type OpenAI = typeof EmbeddingProvider.OpenAI;
-    export type Ollama = typeof EmbeddingProvider.Ollama;
-    export type GaiaNet = typeof EmbeddingProvider.GaiaNet;
-    export type BGE = typeof EmbeddingProvider.BGE;
-    export type VoyageAI = typeof EmbeddingProvider.VoyageAI;
-}
-
 export type EmbeddingConfig = {
     readonly dimensions: number;
     readonly model: string;
-    readonly provider: EmbeddingProvider;
-    readonly endpoint?: string;
-    readonly apiKey?: string;
-    readonly maxInputTokens?: number;
+    readonly provider: EmbeddingProviderType;
 };
 
-// Get embedding config based on settings
-export function getEmbeddingConfig(): EmbeddingConfig {
-    if (settings.USE_OPENAI_EMBEDDING?.toLowerCase() === "true") {
-        return {
-            dimensions: 1536,
-            model: "text-embedding-3-small",
-            provider: "OpenAI",
-            endpoint: "https://api.openai.com/v1",
-            apiKey: settings.OPENAI_API_KEY,
-            maxInputTokens: 1000000,
-        };
-    }
-    if (settings.USE_OLLAMA_EMBEDDING?.toLowerCase() === "true") {
-        return {
-            dimensions: 1024,
-            model: settings.OLLAMA_EMBEDDING_MODEL || "mxbai-embed-large",
-            provider: "Ollama",
-            endpoint: "https://ollama.eliza.ai/",
-            apiKey: settings.OLLAMA_API_KEY,
-            maxInputTokens: 1000000,
-        };
-    }
-    if (settings.USE_GAIANET_EMBEDDING?.toLowerCase() === "true") {
-        return {
-            dimensions: 768,
-            model: settings.GAIANET_EMBEDDING_MODEL || "nomic-embed",
-            provider: "GaiaNet",
-            endpoint: settings.SMALL_GAIANET_SERVER_URL || settings.MEDIUM_GAIANET_SERVER_URL || settings.LARGE_GAIANET_SERVER_URL,
-            apiKey: settings.GAIANET_API_KEY,
-            maxInputTokens: 1000000,
-        };
-    }
-    if (settings.USE_VOYAGEAI_EMBEDDING?.toLowerCase() === "true") {
-        return getVoyageAIEmbeddingConfig();
-    }
-
-    // Fallback to local BGE
-    return {
-        dimensions: 384,
-        model: "BGE-small-en-v1.5",
-        provider: "BGE",
-        maxInputTokens: 1000000,
-    };
-};
+export const getEmbeddingConfig = (): EmbeddingConfig => ({
+    dimensions:
+        settings.USE_OPENAI_EMBEDDING?.toLowerCase() === "true"
+            ? getEmbeddingModelSettings(ModelProviderName.OPENAI).dimensions
+            : settings.USE_OLLAMA_EMBEDDING?.toLowerCase() === "true"
+              ? getEmbeddingModelSettings(ModelProviderName.OLLAMA).dimensions
+              : settings.USE_GAIANET_EMBEDDING?.toLowerCase() === "true"
+                ? getEmbeddingModelSettings(ModelProviderName.GAIANET)
+                      .dimensions
+                : settings.USE_HEURIST_EMBEDDING?.toLowerCase() === "true"
+                  ? getEmbeddingModelSettings(ModelProviderName.HEURIST)
+                        .dimensions
+                  : 384, // BGE
+    model:
+        settings.USE_OPENAI_EMBEDDING?.toLowerCase() === "true"
+            ? getEmbeddingModelSettings(ModelProviderName.OPENAI).name
+            : settings.USE_OLLAMA_EMBEDDING?.toLowerCase() === "true"
+              ? getEmbeddingModelSettings(ModelProviderName.OLLAMA).name
+              : settings.USE_GAIANET_EMBEDDING?.toLowerCase() === "true"
+                ? getEmbeddingModelSettings(ModelProviderName.GAIANET).name
+                : settings.USE_HEURIST_EMBEDDING?.toLowerCase() === "true"
+                  ? getEmbeddingModelSettings(ModelProviderName.HEURIST).name
+                  : "BGE-small-en-v1.5",
+    provider:
+        settings.USE_OPENAI_EMBEDDING?.toLowerCase() === "true"
+            ? "OpenAI"
+            : settings.USE_OLLAMA_EMBEDDING?.toLowerCase() === "true"
+              ? "Ollama"
+              : settings.USE_GAIANET_EMBEDDING?.toLowerCase() === "true"
+                ? "GaiaNet"
+                : settings.USE_HEURIST_EMBEDDING?.toLowerCase() === "true"
+                  ? "Heurist"
+                  : "BGE",
+});
 
 async function getRemoteEmbedding(
     input: string,
-    options: EmbeddingConfig
+    options: EmbeddingOptions
 ): Promise<number[]> {
-    elizaLogger.debug("Getting remote embedding using provider:", options.provider);
+    // Ensure endpoint ends with /v1 for OpenAI
+    const baseEndpoint = options.endpoint.endsWith("/v1")
+        ? options.endpoint
+        : `${options.endpoint}${options.isOllama ? "/v1" : ""}`;
 
     // Construct full URL
-    const fullUrl = `${options.endpoint}/embeddings`;
-
-    // jank. voyageai is the only one that doesn't use "dimensions".
-    const body = options.provider === "VoyageAI" ? {
-        input,
-        model: options.model,
-        output_dimension: options.dimensions,
-    } : {
-        input,
-        model: options.model,
-        dimensions: options.dimensions,
-    };
+    const fullUrl = `${baseEndpoint}/embeddings`;
 
     const requestOptions = {
         method: "POST",
@@ -120,7 +88,14 @@ async function getRemoteEmbedding(
                   }
                 : {}),
         },
-        body: JSON.stringify(body),
+        body: JSON.stringify({
+            input,
+            model: options.model,
+            dimensions:
+                options.dimensions ||
+                options.length ||
+                getEmbeddingConfig().dimensions, // Prefer dimensions, fallback to length
+        }),
     };
 
     try {
@@ -166,18 +141,44 @@ export function getEmbeddingType(runtime: IAgentRuntime): "local" | "remote" {
 }
 
 export function getEmbeddingZeroVector(): number[] {
-    // Default BGE dimension is 384
-    return Array(getEmbeddingConfig().dimensions).fill(0);
+    let embeddingDimension = 384; // Default BGE dimension
+
+    if (settings.USE_OPENAI_EMBEDDING?.toLowerCase() === "true") {
+        embeddingDimension = getEmbeddingModelSettings(
+            ModelProviderName.OPENAI
+        ).dimensions; // OpenAI dimension
+    } else if (settings.USE_OLLAMA_EMBEDDING?.toLowerCase() === "true") {
+        embeddingDimension = getEmbeddingModelSettings(
+            ModelProviderName.OLLAMA
+        ).dimensions; // Ollama mxbai-embed-large dimension
+    } else if (settings.USE_GAIANET_EMBEDDING?.toLowerCase() === "true") {
+        embeddingDimension = getEmbeddingModelSettings(
+            ModelProviderName.GAIANET
+        ).dimensions; // GaiaNet dimension
+    } else if (settings.USE_HEURIST_EMBEDDING?.toLowerCase() === "true") {
+        embeddingDimension = getEmbeddingModelSettings(
+            ModelProviderName.HEURIST
+        ).dimensions; // Heurist dimension
+    }
+
+    return Array(embeddingDimension).fill(0);
 }
 
 /**
  * Gets embeddings from a remote API endpoint.  Falls back to local BGE/384
  *
  * @param {string} input - The text to generate embeddings for
+ * @param {EmbeddingOptions} options - Configuration options including:
+ *   - model: The model name to use
+ *   - endpoint: Base API endpoint URL
+ *   - apiKey: Optional API key for authentication
+ *   - isOllama: Whether this is an Ollama endpoint
+ *   - dimensions: Desired embedding dimensions
  * @param {IAgentRuntime} runtime - The agent runtime context
  * @returns {Promise<number[]>} Array of embedding values
- * @throws {Error} If the API request fails or configuration is invalid
+ * @throws {Error} If the API request fails
  */
+
 export async function embed(runtime: IAgentRuntime, input: string) {
     elizaLogger.debug("Embedding request:", {
         modelProvider: runtime.character.modelProvider,
@@ -206,11 +207,6 @@ export async function embed(runtime: IAgentRuntime, input: string) {
     const config = getEmbeddingConfig();
     const isNode = typeof process !== "undefined" && process.versions?.node;
 
-    // Attempt remote embedding if it is configured.
-    if (config.provider !== EmbeddingProvider.BGE) {
-        return await getRemoteEmbedding(input, config);
-    }
-
     // Determine which embedding path to use
     if (config.provider === EmbeddingProvider.OpenAI) {
         return await getRemoteEmbedding(input, {
@@ -275,7 +271,6 @@ export async function embed(runtime: IAgentRuntime, input: string) {
             getEndpoint(runtime.character.modelProvider),
         apiKey: runtime.token,
         dimensions: config.dimensions,
-        provider: config.provider,
     });
 
     async function getLocalEmbedding(input: string): Promise<number[]> {