WTF

packages/tasks-gen/scripts/generate-snippets-fixtures.ts

@@ -133,8 +133,7 @@ function generateInferenceSnippet(
 	provider: SnippetInferenceProvider,
 	opts?: Record<string, unknown>
 ): InferenceSnippet[] {
-	const generatedSnippets = GET_SNIPPET_FN[language](model, "api_token", provider, opts);
-	return Array.isArray(generatedSnippets) ? generatedSnippets : [generatedSnippets];
+	return GET_SNIPPET_FN[language](model, "api_token", provider, opts);
 }
 
 async function getExpectedInferenceSnippet(

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/0.huggingface.js.hf-inference.js

@@ -11,7 +11,7 @@ const chatCompletion = await client.chatCompletion({
 		}
 	],
 	provider: "hf-inference",
-	max_tokens: 500
+	max_tokens: 500,
 });
 
-console.log(chatCompletion.choices[0].message);
+console.log(chatCompletion.choices[0].message);

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/0.huggingface.js.together.js

@@ -11,7 +11,7 @@ const chatCompletion = await client.chatCompletion({
 		}
 	],
 	provider: "together",
-	max_tokens: 500
+	max_tokens: 500,
 });
 
-console.log(chatCompletion.choices[0].message);
+console.log(chatCompletion.choices[0].message);

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/0.huggingface_hub.hf-inference.py

@@ -15,7 +15,7 @@
 completion = client.chat.completions.create(
     model="meta-llama/Llama-3.1-8B-Instruct", 
 	messages=messages, 
-	max_tokens=500
+	max_tokens=500,
 )
 
 print(completion.choices[0].message)

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/0.huggingface_hub.together.py

@@ -15,7 +15,7 @@
 completion = client.chat.completions.create(
     model="meta-llama/Llama-3.1-8B-Instruct", 
 	messages=messages, 
-	max_tokens=500
+	max_tokens=500,
 )
 
 print(completion.choices[0].message)

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/1.openai.hf-inference.js

@@ -1,7 +1,7 @@
 import { OpenAI } from "openai";
 
 const client = new OpenAI({
-	baseURL: "https://router.huggingface.co/hf-inference",
+	baseURL: "https://router.huggingface.co/hf-inference/v1",
 	apiKey: "api_token"
 });
 
@@ -13,7 +13,7 @@ const chatCompletion = await client.chat.completions.create({
 			content: "What is the capital of France?"
 		}
 	],
-	max_tokens: 500
+	max_tokens: 500,
 });
 
-console.log(chatCompletion.choices[0].message);
+console.log(chatCompletion.choices[0].message);

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/1.openai.hf-inference.py

@@ -1,7 +1,7 @@
 from openai import OpenAI
 
 client = OpenAI(
-	base_url="https://router.huggingface.co/hf-inference",
+	base_url="https://router.huggingface.co/hf-inference/v1",
 	api_key="api_token"
 )
 
@@ -15,7 +15,7 @@
 completion = client.chat.completions.create(
 	model="meta-llama/Llama-3.1-8B-Instruct", 
 	messages=messages, 
-	max_tokens=500
+	max_tokens=500,
 )
 
 print(completion.choices[0].message)

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/1.openai.together.js

@@ -13,7 +13,7 @@ const chatCompletion = await client.chat.completions.create({
 			content: "What is the capital of France?"
 		}
 	],
-	max_tokens: 500
+	max_tokens: 500,
 });
 
-console.log(chatCompletion.choices[0].message);
+console.log(chatCompletion.choices[0].message);

packages/tasks-gen/snippets-fixtures/conversational-llm-non-stream/1.openai.together.py

@@ -15,7 +15,7 @@
 completion = client.chat.completions.create(
 	model="meta-llama/Llama-3.1-8B-Instruct", 
 	messages=messages, 
-	max_tokens=500
+	max_tokens=500,
 )
 
 print(completion.choices[0].message)

packages/tasks-gen/snippets-fixtures/conversational-llm-stream/0.huggingface.js.hf-inference.js

@@ -13,7 +13,7 @@ const stream = client.chatCompletionStream({
 		}
 	],
 	provider: "hf-inference",
-	max_tokens: 500
+	max_tokens: 500,
 });
 
 for await (const chunk of stream) {

packages/tasks-gen/snippets-fixtures/conversational-llm-stream/1.openai.hf-inference.js

@@ -1,7 +1,7 @@
 import { OpenAI } from "openai";
 
 const client = new OpenAI({
-	baseURL: "https://router.huggingface.co/hf-inference",
+	baseURL: "https://router.huggingface.co/hf-inference/v1",
 	apiKey: "api_token"
 });
 

packages/tasks-gen/snippets-fixtures/conversational-llm-stream/1.openai.hf-inference.py

@@ -1,7 +1,7 @@
 from openai import OpenAI
 
 client = OpenAI(
-	base_url="https://router.huggingface.co/hf-inference",
+	base_url="https://router.huggingface.co/hf-inference/v1",
 	api_key="api_token"
 )
 

packages/tasks-gen/snippets-fixtures/conversational-vlm-non-stream/0.huggingface.js.hf-inference.js

@@ -22,7 +22,7 @@ const chatCompletion = await client.chatCompletion({
 		}
 	],
 	provider: "hf-inference",
-	max_tokens: 500
+	max_tokens: 500,
 });
 
-console.log(chatCompletion.choices[0].message);
+console.log(chatCompletion.choices[0].message);

packages/tasks-gen/snippets-fixtures/conversational-vlm-non-stream/0.huggingface_hub.hf-inference.py

@@ -26,7 +26,7 @@
 completion = client.chat.completions.create(
     model="meta-llama/Llama-3.2-11B-Vision-Instruct", 
 	messages=messages, 
-	max_tokens=500
+	max_tokens=500,
 )
 
 print(completion.choices[0].message)

packages/tasks-gen/snippets-fixtures/conversational-vlm-non-stream/1.openai.hf-inference.js

@@ -1,7 +1,7 @@
 import { OpenAI } from "openai";
 
 const client = new OpenAI({
-	baseURL: "https://router.huggingface.co/hf-inference",
+	baseURL: "https://router.huggingface.co/hf-inference/v1",
 	apiKey: "api_token"
 });
 
@@ -24,7 +24,7 @@ const chatCompletion = await client.chat.completions.create({
 			]
 		}
 	],
-	max_tokens: 500
+	max_tokens: 500,
 });
 
-console.log(chatCompletion.choices[0].message);
+console.log(chatCompletion.choices[0].message);

packages/tasks-gen/snippets-fixtures/conversational-vlm-non-stream/1.openai.hf-inference.py

@@ -1,7 +1,7 @@
 from openai import OpenAI
 
 client = OpenAI(
-	base_url="https://router.huggingface.co/hf-inference",
+	base_url="https://router.huggingface.co/hf-inference/v1",
 	api_key="api_token"
 )
 
@@ -26,7 +26,7 @@
 completion = client.chat.completions.create(
 	model="meta-llama/Llama-3.2-11B-Vision-Instruct", 
 	messages=messages, 
-	max_tokens=500
+	max_tokens=500,
 )
 
 print(completion.choices[0].message)

packages/tasks-gen/snippets-fixtures/conversational-vlm-stream/0.huggingface.js.hf-inference.js

@@ -24,7 +24,7 @@ const stream = client.chatCompletionStream({
 		}
 	],
 	provider: "hf-inference",
-	max_tokens: 500
+	max_tokens: 500,
 });
 
 for await (const chunk of stream) {

packages/tasks-gen/snippets-fixtures/conversational-vlm-stream/1.openai.hf-inference.js

@@ -1,7 +1,7 @@
 import { OpenAI } from "openai";
 
 const client = new OpenAI({
-	baseURL: "https://router.huggingface.co/hf-inference",
+	baseURL: "https://router.huggingface.co/hf-inference/v1",
 	apiKey: "api_token"
 });
 

packages/tasks-gen/snippets-fixtures/conversational-vlm-stream/1.openai.hf-inference.py

@@ -1,7 +1,7 @@
 from openai import OpenAI
 
 client = OpenAI(
-	base_url="https://router.huggingface.co/hf-inference",
+	base_url="https://router.huggingface.co/hf-inference/v1",
 	api_key="api_token"
 )
 

packages/tasks-gen/snippets-fixtures/text-classification/1.requests.hf-inference.py

@@ -1,6 +1,6 @@
 import requests
 
-API_URL = "https://router.huggingface.co/hf-inference"
+API_URL = "https://router.huggingface.co/hf-inference/v1"
 headers = {"Authorization": "Bearer api_token"}
 
 def query(payload):

packages/tasks-gen/snippets-fixtures/text-to-image/1.requests.hf-inference.py

@@ -1,6 +1,6 @@
 import requests
 
-API_URL = "https://router.huggingface.co/hf-inference"
+API_URL = "https://router.huggingface.co/hf-inference/v1"
 headers = {"Authorization": "Bearer api_token"}
 
 def query(payload):

packages/tasks/src/snippets/js.ts

@@ -27,9 +27,9 @@ export const snippetBasic = (
 	return [
 		...(model.pipeline_tag && model.pipeline_tag in HFJS_METHODS
 			? [
-				{
-					client: "huggingface.js",
-					content: `\
+					{
+						client: "huggingface.js",
+						content: `\
 import { HfInference } from "@huggingface/inference";
 
 const client = new HfInference("${accessToken || `{API_TOKEN}`}");
@@ -42,8 +42,8 @@ const output = await client.${HFJS_METHODS[model.pipeline_tag]}({
 
 console.log(output);
 `,
-				},
-			]
+					},
+			  ]
 			: []),
 		{
 			client: "fetch",
@@ -216,8 +216,8 @@ export const snippetZeroShotClassification = (model: ModelDataMinimal, accessTok
 		}
 		
 		query({"inputs": ${getModelInputSnippet(
-				model
-			)}, "parameters": {"candidate_labels": ["refund", "legal", "faq"]}}).then((response) => {
+			model
+		)}, "parameters": {"candidate_labels": ["refund", "legal", "faq"]}}).then((response) => {
 			console.log(JSON.stringify(response));
 		});`,
 		},
@@ -248,9 +248,9 @@ const image = await client.textToImage({
 		},
 		...(provider === "hf-inference"
 			? [
-				{
-					client: "fetch",
-					content: `async function query(data) {
+					{
+						client: "fetch",
+						content: `async function query(data) {
 	const response = await fetch(
 		"https://router.huggingface.co/hf-inference/models/${model.id}",
 		{
@@ -268,8 +268,8 @@ const image = await client.textToImage({
 query({"inputs": ${getModelInputSnippet(model)}}).then((response) => {
 	// Use image
 });`,
-				},
-			]
+					},
+			  ]
 			: []),
 	];
 };

packages/tasks/src/snippets/python.ts

@@ -271,9 +271,9 @@ image = client.text_to_image(
 		},
 		...(provider === "fal-ai"
 			? [
-				{
-					client: "fal-client",
-					content: `\
+					{
+						client: "fal-client",
+						content: `\
 import fal_client
 
 result = fal_client.subscribe(
@@ -285,14 +285,14 @@ result = fal_client.subscribe(
 )
 print(result)
 `,
-				},
-			]
+					},
+			  ]
 			: []),
 		...(provider === "hf-inference"
 			? [
-				{
-					client: "requests",
-					content: `\
+					{
+						client: "requests",
+						content: `\
 def query(payload):
 	response = requests.post(API_URL, headers=headers, json=payload)
 	return response.content
@@ -305,8 +305,8 @@ image_bytes = query({
 import io
 from PIL import Image
 image = Image.open(io.BytesIO(image_bytes))`,
-				},
-			]
+					},
+			  ]
 			: []),
 	];
 };
@@ -443,8 +443,6 @@ export function getPythonInferenceSnippet(
 				? pythonSnippets[model.pipeline_tag]?.(model, accessToken, provider) ?? []
 				: [];
 
-		const baseUrl = HF_HUB_INFERENCE_PROXY_TEMPLATE.replace("{{PROVIDER}}", provider);
-
 		return snippets.map((snippet) => {
 			return {
 				...snippet,
@@ -453,7 +451,7 @@ export function getPythonInferenceSnippet(
 						? `\
 import requests
 
-API_URL = "${baseUrl}"
+API_URL = "${openAIbaseUrl(provider)}"
 headers = {"Authorization": ${accessToken ? `"Bearer ${accessToken}"` : `f"Bearer {API_TOKEN}"`}}
 
 ${snippet.content}`