Switch path to router.huggingface.co (#1188)

Author: julien-c

README.md

@@ -183,7 +183,7 @@ const { generated_text } = await gpt2.textGeneration({inputs: 'The answer to the
 
 // Chat Completion
 const llamaEndpoint = inference.endpoint(
- "https://api-inference.huggingface.co/models/meta-llama/Llama-3.1-8B-Instruct"
+  "https://router.huggingface.co/together/models/meta-llama/Llama-3.1-8B-Instruct"
 );
 const out = await llamaEndpoint.chatCompletion({
   model: "meta-llama/Llama-3.1-8B-Instruct",

packages/inference/README.md

@@ -117,7 +117,7 @@ for await (const output of hf.textGenerationStream({
 
 ### Text Generation (Chat Completion API Compatible)
 
-Using the `chatCompletion` method, you can generate text with models compatible with the OpenAI Chat Completion API. All models served by [TGI](https://api-inference.huggingface.co/framework/text-generation-inference) on Hugging Face support Messages API.
+Using the `chatCompletion` method, you can generate text with models compatible with the OpenAI Chat Completion API. All models served by [TGI](https://huggingface.co/docs/text-generation-inference/) on Hugging Face support Messages API.
 
 [Demo](https://huggingface.co/spaces/huggingfacejs/streaming-chat-completion)
 
@@ -611,7 +611,7 @@ const { generated_text } = await gpt2.textGeneration({inputs: 'The answer to the
 
 // Chat Completion Example
 const ep = hf.endpoint(
-  "https://api-inference.huggingface.co/models/meta-llama/Llama-3.1-8B-Instruct"
+  "https://router.huggingface.co/together/models/meta-llama/Llama-3.1-8B-Instruct"
 );
 const stream = ep.chatCompletionStream({
   model: "tgi",

packages/inference/src/config.ts

@@ -1 +1,2 @@
 export const HF_HUB_URL = "https://huggingface.co";
+export const HF_ROUTER_URL = "https://router.huggingface.co";

packages/inference/src/lib/makeRequestOptions.ts

@@ -1,4 +1,4 @@
-import { HF_HUB_URL } from "../config";
+import { HF_HUB_URL, HF_ROUTER_URL } from "../config";
 import { FAL_AI_API_BASE_URL } from "../providers/fal-ai";
 import { REPLICATE_API_BASE_URL } from "../providers/replicate";
 import { SAMBANOVA_API_BASE_URL } from "../providers/sambanova";
@@ -9,7 +9,7 @@ import { isUrl } from "./isUrl";
 import { version as packageVersion, name as packageName } from "../../package.json";
 import { getProviderModelId } from "./getProviderModelId";
 
-const HF_HUB_INFERENCE_PROXY_TEMPLATE = `${HF_HUB_URL}/api/inference-proxy/{{PROVIDER}}`;
+const HF_HUB_INFERENCE_PROXY_TEMPLATE = `${HF_ROUTER_URL}/{{PROVIDER}}`;
 
 /**
  * Lazy-loaded from huggingface.co/api/tasks when needed

packages/inference/test/HfInference.spec.ts

@@ -21,13 +21,14 @@ describe.concurrent("HfInference", () => {
 		"HF Inference",
 		() => {
 			const hf = new HfInference(env.HF_TOKEN);
+
 			it("throws error if model does not exist", () => {
 				expect(
 					hf.fillMask({
-						model: "this-model-does-not-exist-123",
+						model: "this-model/does-not-exist-123",
 						inputs: "[MASK] world!",
 					})
-				).rejects.toThrowError("Not Found: Model not found");
+				).rejects.toThrowError("Model this-model/does-not-exist-123 does not exist");
 			});
 
 			it("fillMask", async () => {
@@ -647,7 +648,7 @@ describe.concurrent("HfInference", () => {
 			});
 
 			it("endpoint - makes request to specified endpoint", async () => {
-				const ep = hf.endpoint("https://api-inference.huggingface.co/models/openai-community/gpt2");
+				const ep = hf.endpoint("https://router.huggingface.co/hf-inference/models/openai-community/gpt2");
 				const { generated_text } = await ep.textGeneration({
 					inputs: "one plus two equals",
 				});
@@ -685,7 +686,7 @@ describe.concurrent("HfInference", () => {
 				expect(out).toContain("2");
 			});
 
-			it("chatCompletionStream modelId Fail - OpenAI Specs", async () => {
+			it.skip("chatCompletionStream modelId Fail - OpenAI Specs", async () => {
 				expect(
 					hf
 						.chatCompletionStream({
@@ -702,7 +703,7 @@ describe.concurrent("HfInference", () => {
 			});
 
 			it("chatCompletion - OpenAI Specs", async () => {
-				const ep = hf.endpoint("https://api-inference.huggingface.co/models/mistralai/Mistral-7B-Instruct-v0.2");
+				const ep = hf.endpoint("https://router.huggingface.co/hf-inference/models/mistralai/Mistral-7B-Instruct-v0.2");
 				const res = await ep.chatCompletion({
 					model: "tgi",
 					messages: [{ role: "user", content: "Complete the this sentence with words one plus one is equal " }],
@@ -716,7 +717,7 @@ describe.concurrent("HfInference", () => {
 				}
 			});
 			it("chatCompletionStream - OpenAI Specs", async () => {
-				const ep = hf.endpoint("https://api-inference.huggingface.co/models/mistralai/Mistral-7B-Instruct-v0.2");
+				const ep = hf.endpoint("https://router.huggingface.co/hf-inference/models/mistralai/Mistral-7B-Instruct-v0.2");
 				const stream = ep.chatCompletionStream({
 					model: "tgi",
 					messages: [{ role: "user", content: "Complete the equation 1+1= ,just the answer" }],