✨ Support for Featherless.ai as inference provider.

packages/inference/src/lib/getProviderHelper.ts

@@ -4,7 +4,7 @@ import * as Cohere from "../providers/cohere";
 import * as FalAI from "../providers/fal-ai";
 import * as Fireworks from "../providers/fireworks-ai";
 import * as HFInference from "../providers/hf-inference";
-
+import * as FeatherlessAI from "../providers/featherless-ai";
 import * as Hyperbolic from "../providers/hyperbolic";
 import * as Nebius from "../providers/nebius";
 import * as Novita from "../providers/novita";
@@ -62,6 +62,10 @@ export const PROVIDERS: Record<InferenceProvider, Partial<Record<InferenceTask,
 		"text-to-video": new FalAI.FalAITextToVideoTask(),
 		"automatic-speech-recognition": new FalAI.FalAIAutomaticSpeechRecognitionTask(),
 	},
+	"featherless-ai": {
+		conversational: new FeatherlessAI.FeatherlessAIConversationalTask(),
+		"text-generation": new FeatherlessAI.FeatherlessAITextGenerationTask(),
+	},
 	"hf-inference": {
 		"text-to-image": new HFInference.HFInferenceTextToImageTask(),
 		conversational: new HFInference.HFInferenceConversationalTask(),

packages/inference/src/providers/featherless-ai.ts

@@ -0,0 +1,51 @@
+import { BaseConversationalTask, BaseTextGenerationTask } from "./providerHelper";
+import type { ChatCompletionOutput, TextGenerationOutputFinishReason } from "@huggingface/tasks";
+import { InferenceOutputError } from "../lib/InferenceOutputError";
+
+interface FeatherlessAITextCompletionOutput extends Omit<ChatCompletionOutput, "choices"> {
+  choices: Array<{
+    text: string;
+    finish_reason: TextGenerationOutputFinishReason;
+    seed: number;
+    logprobs: unknown;
+    index: number;
+  }>;
+}
+
+const FEATHERLESS_API_BASE_URL = "https://api.featherless.ai";
+
+export class FeatherlessAIConversationalTask extends BaseConversationalTask {
+  constructor() {
+    super("featherless-ai", FEATHERLESS_API_BASE_URL);
+  }
+}
+
+export class FeatherlessAITextGenerationTask extends BaseTextGenerationTask {
+  constructor() {
+    super("featherless-ai", FEATHERLESS_API_BASE_URL);
+  }
+
+  override preparePayload(params: BodyParams): Record<string, unknown> {
+    return {
+      model: params.model,
+      ...params.args,
+      ...params.args.parameters,
+      prompt: params.args.inputs,
+    };
+  }
+
+  override async getResponse(response: FeatherlessAITextCompletionOutput): Promise<TextGenerationOutput> {
+    if (
+      typeof response === "object" &&
+      "choices" in response &&
+      Array.isArray(response?.choices) &&
+      typeof response?.model === "string"
+    ) {
+      const completion = response.choices[0];
+      return {
+        generated_text: completion.text,
+      };
+    }
+    throw new InferenceOutputError("Expected Together text generation response format");
+  }
+}

packages/inference/src/types.ts

@@ -41,6 +41,7 @@ export const INFERENCE_PROVIDERS = [
 	"cerebras",
 	"cohere",
 	"fal-ai",
+	"featherless-ai",
 	"fireworks-ai",
 	"hf-inference",
 	"hyperbolic",

packages/inference/test/InferenceClient.spec.ts

@@ -826,6 +826,69 @@ describe.concurrent("InferenceClient", () => {
 		TIMEOUT
 	);
 
+	describe.concurrent(
+		"Featherless",
+		() => {
+			HARDCODED_MODEL_ID_MAPPING['featherless-ai'] = {
+				"meta-llama/Llama-3.1-8B": "meta-llama/Meta-Llama-3.1-8B",
+				"meta-llama/Llama-3.1-8B-Instruct": "meta-llama/Meta-Llama-3.1-8B-Instruct",
+			};
+
+			it("chatCompletion", async () => {
+				const res = await chatCompletion({
+					accessToken: env.HF_FEATHERLESS_KEY ?? "dummy",
+					model: "meta-llama/Llama-3.1-8B-Instruct",
+					provider: "featherless-ai",
+					messages: [{ role: "user", content: "Complete this sentence with words, one plus one is equal " }],
+					temperature: 0.1,
+				});
+
+				expect(res).toBeDefined();
+				expect(res.choices).toBeDefined();
+				expect(res.choices?.length).toBeGreaterThan(0);
+
+				if (res.choices && res.choices.length > 0) {
+					const completion = res.choices[0].message?.content;
+					expect(completion).toBeDefined();
+					expect(typeof completion).toBe("string");
+					expect(completion).toContain("two");
+				}
+			});
+
+			it("chatCompletion stream", async () => {
+				const stream = chatCompletionStream({
+					accessToken: env.HF_FEATHERLESS_KEY ?? "dummy",
+					model: "meta-llama/Llama-3.1-8B-Instruct",
+					provider: "featherless-ai",
+					messages: [{ role: "user", content: "Complete the equation 1 + 1 = , just the answer" }],
+				}) as AsyncGenerator<ChatCompletionStreamOutput>;
+				let out = "";
+				for await (const chunk of stream) {
+					if (chunk.choices && chunk.choices.length > 0) {
+						out += chunk.choices[0].delta.content;
+					}
+				}
+				expect(out).toContain("2");
+			});
+
+			it("textGeneration", async () => {
+				const res = await textGeneration({
+					accessToken: env.HF_FEATHERLESS_KEY ?? "dummy",
+					model: "meta-llama/Llama-3.1-8B",
+					provider: "featherless-ai",
+					inputs: "Paris is a city of ",
+					parameters: {
+						temperature: 0,
+						top_p: 0.01,
+						max_tokens: 10,
+					},
+				});
+				expect(res).toMatchObject({ generated_text: "2.2 million people, and it is the" });
+			});
+		},
+		TIMEOUT
+	);
+
 	describe.concurrent(
 		"Replicate",
 		() => {