@huggingface/tasks 0.15.8 → 0.16.0

package/dist/esm/snippets/js.js

@@ -31,7 +31,7 @@ const output = await client.${HFJS_METHODS[model.pipeline_tag]}({
 	provider: "${provider}",
 });
 
-console.log(output)
+console.log(output);
 `,
                 },
             ]
@@ -41,7 +41,7 @@ console.log(output)
             content: `\
 async function query(data) {
 	const response = await fetch(
-		"https://api-inference.huggingface.co/models/${model.id}",
+		"https://router.huggingface.co/hf-inference/models/${model.id}",
 		{
 			headers: {
 				Authorization: "Bearer ${accessToken || `{API_TOKEN}`}",
@@ -61,7 +61,7 @@ query({"inputs": ${getModelInputSnippet(model)}}).then((response) => {
         },
     ];
 };
-export const snippetTextGeneration = (model, accessToken, provider, opts) => {
+export const snippetTextGeneration = (model, accessToken, provider, providerModelId, opts) => {
     if (model.tags.includes("conversational")) {
         // Conversational model detected, so we display a code snippet that features the Messages API
         const streaming = opts?.streaming ?? true;
@@ -114,9 +114,9 @@ const client = new OpenAI({
 let out = "";
 
 const stream = await client.chat.completions.create({
-	model: "${model.id}",
+	model: "${providerModelId ?? model.id}",
 	messages: ${messagesStr},
-	${configStr},
+	${configStr}
 	stream: true,
 });
 
@@ -145,7 +145,8 @@ const chatCompletion = await client.chatCompletion({
 	${configStr}
 });
 
-console.log(chatCompletion.choices[0].message);`,
+console.log(chatCompletion.choices[0].message);
+`,
                 },
                 {
                     client: "openai",
@@ -157,12 +158,13 @@ const client = new OpenAI({
 });
 
 const chatCompletion = await client.chat.completions.create({
-	model: "${model.id}",
+	model: "${providerModelId ?? model.id}",
 	messages: ${messagesStr},
 	${configStr}
 });
 
-console.log(chatCompletion.choices[0].message);`,
+console.log(chatCompletion.choices[0].message);
+`,
                 },
             ];
         }
@@ -177,7 +179,7 @@ export const snippetZeroShotClassification = (model, accessToken) => {
             client: "fetch",
             content: `async function query(data) {
 			const response = await fetch(
-				"https://api-inference.huggingface.co/models/${model.id}",
+				"https://router.huggingface.co/hf-inference/models/${model.id}",
 				{
 					headers: {
 						Authorization: "Bearer ${accessToken || `{API_TOKEN}`}",
@@ -221,7 +223,7 @@ const image = await client.textToImage({
                     client: "fetch",
                     content: `async function query(data) {
 	const response = await fetch(
-		"https://api-inference.huggingface.co/models/${model.id}",
+		"https://router.huggingface.co/hf-inference/models/${model.id}",
 		{
 			headers: {
 				Authorization: "Bearer ${accessToken || `{API_TOKEN}`}",
@@ -248,7 +250,7 @@ export const snippetTextToAudio = (model, accessToken, provider) => {
     }
     const commonSnippet = `async function query(data) {
 		const response = await fetch(
-			"https://api-inference.huggingface.co/models/${model.id}",
+			"https://router.huggingface.co/hf-inference/models/${model.id}",
 			{
 				headers: {
 					Authorization: "Bearer ${accessToken || `{API_TOKEN}`}",
@@ -323,7 +325,7 @@ export const snippetFile = (model, accessToken, provider) => {
             content: `async function query(filename) {
 	const data = fs.readFileSync(filename);
 	const response = await fetch(
-		"https://api-inference.huggingface.co/models/${model.id}",
+		"https://router.huggingface.co/hf-inference/models/${model.id}",
 		{
 			headers: {
 				Authorization: "Bearer ${accessToken || `{API_TOKEN}`}",
@@ -369,8 +371,8 @@ export const jsSnippets = {
     "object-detection": snippetFile,
     "image-segmentation": snippetFile,
 };
-export function getJsInferenceSnippet(model, accessToken, provider, opts) {
+export function getJsInferenceSnippet(model, accessToken, provider, providerModelId, opts) {
     return model.pipeline_tag && model.pipeline_tag in jsSnippets
-        ? jsSnippets[model.pipeline_tag]?.(model, accessToken, provider, opts) ?? []
+        ? jsSnippets[model.pipeline_tag]?.(model, accessToken, provider, providerModelId, opts) ?? []
         : [];
 }

package/dist/esm/snippets/python.d.ts

@@ -1,8 +1,8 @@
-import { type InferenceProvider } from "../inference-providers.js";
+import { type SnippetInferenceProvider } from "../inference-providers.js";
 import type { PipelineType } from "../pipelines.js";
 import type { ChatCompletionInputMessage, GenerationParameters } from "../tasks/index.js";
 import type { InferenceSnippet, ModelDataMinimal } from "./types.js";
-export declare const snippetConversational: (model: ModelDataMinimal, accessToken: string, provider: InferenceProvider, opts?: {
+export declare const snippetConversational: (model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider, providerModelId?: string, opts?: {
     streaming?: boolean;
     messages?: ChatCompletionInputMessage[];
     temperature?: GenerationParameters["temperature"];
@@ -11,12 +11,12 @@ export declare const snippetConversational: (model: ModelDataMinimal, accessToke
 }) => InferenceSnippet[];
 export declare const snippetZeroShotClassification: (model: ModelDataMinimal) => InferenceSnippet[];
 export declare const snippetZeroShotImageClassification: (model: ModelDataMinimal) => InferenceSnippet[];
-export declare const snippetBasic: (model: ModelDataMinimal) => InferenceSnippet[];
+export declare const snippetBasic: (model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider) => InferenceSnippet[];
 export declare const snippetFile: (model: ModelDataMinimal) => InferenceSnippet[];
-export declare const snippetTextToImage: (model: ModelDataMinimal, accessToken: string, provider: InferenceProvider) => InferenceSnippet[];
+export declare const snippetTextToImage: (model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider, providerModelId?: string) => InferenceSnippet[];
 export declare const snippetTabular: (model: ModelDataMinimal) => InferenceSnippet[];
 export declare const snippetTextToAudio: (model: ModelDataMinimal) => InferenceSnippet[];
 export declare const snippetDocumentQuestionAnswering: (model: ModelDataMinimal) => InferenceSnippet[];
-export declare const pythonSnippets: Partial<Record<PipelineType, (model: ModelDataMinimal, accessToken: string, provider: InferenceProvider, opts?: Record<string, unknown>) => InferenceSnippet[]>>;
-export declare function getPythonInferenceSnippet(model: ModelDataMinimal, accessToken: string, provider: InferenceProvider, opts?: Record<string, unknown>): InferenceSnippet[];
+export declare const pythonSnippets: Partial<Record<PipelineType, (model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider, providerModelId?: string, opts?: Record<string, unknown>) => InferenceSnippet[]>>;
+export declare function getPythonInferenceSnippet(model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider, providerModelId?: string, opts?: Record<string, unknown>): InferenceSnippet[];
 //# sourceMappingURL=python.d.ts.map

package/dist/esm/snippets/python.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"python.d.ts","sourceRoot":"","sources":["../../../src/snippets/python.ts"],"names":[],"mappings":"AAAA,OAAO,EAAkD,KAAK,iBAAiB,EAAE,MAAM,2BAA2B,CAAC;AACnH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AACpD,OAAO,KAAK,EAAE,0BAA0B,EAAE,oBAAoB,EAAE,MAAM,mBAAmB,CAAC;AAG1F,OAAO,KAAK,EAAE,gBAAgB,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AAWrE,eAAO,MAAM,qBAAqB,UAC1B,gBAAgB,eACV,MAAM,YACT,iBAAiB,SACpB;IACN,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,QAAQ,CAAC,EAAE,0BAA0B,EAAE,CAAC;IACxC,WAAW,CAAC,EAAE,oBAAoB,CAAC,aAAa,CAAC,CAAC;IAClD,UAAU,CAAC,EAAE,oBAAoB,CAAC,YAAY,CAAC,CAAC;IAChD,KAAK,CAAC,EAAE,oBAAoB,CAAC,OAAO,CAAC,CAAC;CACtC,KACC,gBAAgB,EAiGlB,CAAC;AAEF,eAAO,MAAM,6BAA6B,UAAW,gBAAgB,KAAG,gBAAgB,EAevF,CAAC;AAEF,eAAO,MAAM,kCAAkC,UAAW,gBAAgB,KAAG,gBAAgB,EAqB5F,CAAC;AAEF,eAAO,MAAM,YAAY,UAAW,gBAAgB,KAAG,gBAAgB,EActE,CAAC;AAEF,eAAO,MAAM,WAAW,UAAW,gBAAgB,KAAG,gBAAgB,EAcrE,CAAC;AAEF,eAAO,MAAM,kBAAkB,UACvB,gBAAgB,eACV,MAAM,YACT,iBAAiB,KACzB,gBAAgB,EAqDlB,CAAC;AAEF,eAAO,MAAM,cAAc,UAAW,gBAAgB,KAAG,gBAAgB,EAcxE,CAAC;AAEF,eAAO,MAAM,kBAAkB,UAAW,gBAAgB,KAAG,gBAAgB,EAuC5E,CAAC;AAEF,eAAO,MAAM,gCAAgC,UAAW,gBAAgB,KAAG,gBAAgB,EAiB1F,CAAC;AAEF,eAAO,MAAM,cAAc,EAAE,OAAO,CACnC,MAAM,CACL,YAAY,EACZ,CACC,KAAK,EAAE,gBAAgB,EACvB,WAAW,EAAE,MAAM,EACnB,QAAQ,EAAE,iBAAiB,EAC3B,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,KAC1B,gBAAgB,EAAE,CACvB,CA8BD,CAAC;AAEF,wBAAgB,yBAAyB,CACxC,KAAK,EAAE,gBAAgB,EACvB,WAAW,EAAE,MAAM,EACnB,QAAQ,EAAE,iBAAiB,EAC3B,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAC5B,gBAAgB,EAAE,CA+BpB"}
+{"version":3,"file":"python.d.ts","sourceRoot":"","sources":["../../../src/snippets/python.ts"],"names":[],"mappings":"AAAA,OAAO,EAEN,KAAK,wBAAwB,EAC7B,MAAM,2BAA2B,CAAC;AACnC,OAAO,KAAK,EAAE,YAAY,EAAc,MAAM,iBAAiB,CAAC;AAChE,OAAO,KAAK,EAAE,0BAA0B,EAAE,oBAAoB,EAAE,MAAM,mBAAmB,CAAC;AAG1F,OAAO,KAAK,EAAE,gBAAgB,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AAyCrE,eAAO,MAAM,qBAAqB,UAC1B,gBAAgB,eACV,MAAM,YACT,wBAAwB,oBAChB,MAAM,SACjB;IACN,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,QAAQ,CAAC,EAAE,0BAA0B,EAAE,CAAC;IACxC,WAAW,CAAC,EAAE,oBAAoB,CAAC,aAAa,CAAC,CAAC;IAClD,UAAU,CAAC,EAAE,oBAAoB,CAAC,YAAY,CAAC,CAAC;IAChD,KAAK,CAAC,EAAE,oBAAoB,CAAC,OAAO,CAAC,CAAC;CACtC,KACC,gBAAgB,EAiGlB,CAAC;AAEF,eAAO,MAAM,6BAA6B,UAAW,gBAAgB,KAAG,gBAAgB,EAevF,CAAC;AAEF,eAAO,MAAM,kCAAkC,UAAW,gBAAgB,KAAG,gBAAgB,EAqB5F,CAAC;AAEF,eAAO,MAAM,YAAY,UACjB,gBAAgB,eACV,MAAM,YACT,wBAAwB,KAChC,gBAAgB,EAgClB,CAAC;AAEF,eAAO,MAAM,WAAW,UAAW,gBAAgB,KAAG,gBAAgB,EAcrE,CAAC;AAEF,eAAO,MAAM,kBAAkB,UACvB,gBAAgB,eACV,MAAM,YACT,wBAAwB,oBAChB,MAAM,KACtB,gBAAgB,EAoDlB,CAAC;AAEF,eAAO,MAAM,cAAc,UAAW,gBAAgB,KAAG,gBAAgB,EAcxE,CAAC;AAEF,eAAO,MAAM,kBAAkB,UAAW,gBAAgB,KAAG,gBAAgB,EAuC5E,CAAC;AAEF,eAAO,MAAM,gCAAgC,UAAW,gBAAgB,KAAG,gBAAgB,EAiB1F,CAAC;AAEF,eAAO,MAAM,cAAc,EAAE,OAAO,CACnC,MAAM,CACL,YAAY,EACZ,CACC,KAAK,EAAE,gBAAgB,EACvB,WAAW,EAAE,MAAM,EACnB,QAAQ,EAAE,wBAAwB,EAClC,eAAe,CAAC,EAAE,MAAM,EACxB,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,KAC1B,gBAAgB,EAAE,CACvB,CA8BD,CAAC;AAEF,wBAAgB,yBAAyB,CACxC,KAAK,EAAE,gBAAgB,EACvB,WAAW,EAAE,MAAM,EACnB,QAAQ,EAAE,wBAAwB,EAClC,eAAe,CAAC,EAAE,MAAM,EACxB,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAC5B,gBAAgB,EAAE,CA0BpB"}

package/dist/esm/snippets/python.js

@@ -1,6 +1,35 @@
-import { HF_HUB_INFERENCE_PROXY_TEMPLATE, openAIbaseUrl } from "../inference-providers.js";
+import { openAIbaseUrl, } from "../inference-providers.js";
 import { stringifyGenerationConfig, stringifyMessages } from "./common.js";
 import { getModelInputSnippet } from "./inputs.js";
+const HFH_INFERENCE_CLIENT_METHODS = {
+    "audio-classification": "audio_classification",
+    "audio-to-audio": "audio_to_audio",
+    "automatic-speech-recognition": "automatic_speech_recognition",
+    "text-to-speech": "text_to_speech",
+    "image-classification": "image_classification",
+    "image-segmentation": "image_segmentation",
+    "image-to-image": "image_to_image",
+    "image-to-text": "image_to_text",
+    "object-detection": "object_detection",
+    "text-to-image": "text_to_image",
+    "text-to-video": "text_to_video",
+    "zero-shot-image-classification": "zero_shot_image_classification",
+    "document-question-answering": "document_question_answering",
+    "visual-question-answering": "visual_question_answering",
+    "feature-extraction": "feature_extraction",
+    "fill-mask": "fill_mask",
+    "question-answering": "question_answering",
+    "sentence-similarity": "sentence_similarity",
+    summarization: "summarization",
+    "table-question-answering": "table_question_answering",
+    "text-classification": "text_classification",
+    "text-generation": "text_generation",
+    "token-classification": "token_classification",
+    translation: "translation",
+    "zero-shot-classification": "zero_shot_classification",
+    "tabular-classification": "tabular_classification",
+    "tabular-regression": "tabular_regression",
+};
 const snippetImportInferenceClient = (accessToken, provider) => `\
 from huggingface_hub import InferenceClient
 
@@ -8,7 +37,7 @@ client = InferenceClient(
 	provider="${provider}",
 	api_key="${accessToken || "{API_TOKEN}"}"
 )`;
-export const snippetConversational = (model, accessToken, provider, opts) => {
+export const snippetConversational = (model, accessToken, provider, providerModelId, opts) => {
     const streaming = opts?.streaming ?? true;
     const exampleMessages = getModelInputSnippet(model);
     const messages = opts?.messages ?? exampleMessages;
@@ -34,7 +63,7 @@ messages = ${messagesStr}
 stream = client.chat.completions.create(
 	model="${model.id}", 
 	messages=messages, 
-	${configStr},
+	${configStr}
 	stream=True
 )
 
@@ -54,9 +83,9 @@ client = OpenAI(
 messages = ${messagesStr}
 
 stream = client.chat.completions.create(
-    model="${model.id}", 
+    model="${providerModelId ?? model.id}", 
 	messages=messages, 
-	${configStr},
+	${configStr}
 	stream=True
 )
 
@@ -95,7 +124,7 @@ client = OpenAI(
 messages = ${messagesStr}
 
 completion = client.chat.completions.create(
-	model="${model.id}", 
+	model="${providerModelId ?? model.id}", 
 	messages=messages, 
 	${configStr}
 )
@@ -143,8 +172,26 @@ output = query({
         },
     ];
 };
-export const snippetBasic = (model) => {
+export const snippetBasic = (model, accessToken, provider) => {
     return [
+        ...(model.pipeline_tag && model.pipeline_tag in HFH_INFERENCE_CLIENT_METHODS
+            ? [
+                {
+                    client: "huggingface_hub",
+                    content: `\
+${snippetImportInferenceClient(accessToken, provider)}
+
+result = client.${HFH_INFERENCE_CLIENT_METHODS[model.pipeline_tag]}(
+	model="${model.id}",
+	inputs=${getModelInputSnippet(model)},
+	provider="${provider}",
+)
+
+print(result)
+`,
+                },
+            ]
+            : []),
         {
             client: "requests",
             content: `\
@@ -173,7 +220,7 @@ output = query(${getModelInputSnippet(model)})`,
         },
     ];
 };
-export const snippetTextToImage = (model, accessToken, provider) => {
+export const snippetTextToImage = (model, accessToken, provider, providerModelId) => {
     return [
         {
             client: "huggingface_hub",
@@ -194,8 +241,7 @@ image = client.text_to_image(
 import fal_client
 
 result = fal_client.subscribe(
-	# replace with correct id from fal.ai
-	"fal-ai/${model.id}",
+	"${providerModelId ?? model.id}",
 	arguments={
 		"prompt": ${getModelInputSnippet(model)},
 	},
@@ -331,18 +377,15 @@ export const pythonSnippets = {
     "image-to-text": snippetFile,
     "zero-shot-image-classification": snippetZeroShotImageClassification,
 };
-export function getPythonInferenceSnippet(model, accessToken, provider, opts) {
+export function getPythonInferenceSnippet(model, accessToken, provider, providerModelId, opts) {
     if (model.tags.includes("conversational")) {
         // Conversational model detected, so we display a code snippet that features the Messages API
-        return snippetConversational(model, accessToken, provider, opts);
+        return snippetConversational(model, accessToken, provider, providerModelId, opts);
     }
     else {
         const snippets = model.pipeline_tag && model.pipeline_tag in pythonSnippets
-            ? pythonSnippets[model.pipeline_tag]?.(model, accessToken, provider) ?? []
+            ? pythonSnippets[model.pipeline_tag]?.(model, accessToken, provider, providerModelId) ?? []
             : [];
-        const baseUrl = provider === "hf-inference"
-            ? `https://api-inference.huggingface.co/models/${model.id}`
-            : HF_HUB_INFERENCE_PROXY_TEMPLATE.replace("{{PROVIDER}}", provider);
         return snippets.map((snippet) => {
             return {
                 ...snippet,
@@ -350,7 +393,7 @@ export function getPythonInferenceSnippet(model, accessToken, provider, opts) {
                     ? `\
 import requests
 
-API_URL = "${baseUrl}"
+API_URL = "${openAIbaseUrl(provider)}"
 headers = {"Authorization": ${accessToken ? `"Bearer ${accessToken}"` : `f"Bearer {API_TOKEN}"`}}
 
 ${snippet.content}`

package/dist/esm/tasks/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/tasks/index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AA4CpD,mBAAmB,qCAAqC,CAAC;AACzD,mBAAmB,6CAA6C,CAAC;AACjE,YAAY,EACX,mBAAmB,EACnB,0BAA0B,EAC1B,oBAAoB,EACpB,4BAA4B,EAC5B,2BAA2B,EAC3B,0BAA0B,EAC1B,gCAAgC,EAChC,+BAA+B,GAC/B,MAAM,gCAAgC,CAAC;AACxC,mBAAmB,4CAA4C,CAAC;AAChE,mBAAmB,mCAAmC,CAAC;AACvD,mBAAmB,0BAA0B,CAAC;AAC9C,YAAY,EACX,wBAAwB,EACxB,yBAAyB,EACzB,gCAAgC,EAChC,6BAA6B,GAC7B,MAAM,qCAAqC,CAAC;AAC7C,mBAAmB,+BAA+B,CAAC;AACnD,YAAY,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,qBAAqB,EAAE,MAAM,8BAA8B,CAAC;AAC/G,mBAAmB,mCAAmC,CAAC;AACvD,mBAAmB,iCAAiC,CAAC;AACrD,mBAAmB,iCAAiC,CAAC;AACrD,mBAAmB,mCAAmC,CAAC;AACvD,mBAAmB,oCAAoC,CAAC;AACxD,mBAAmB,8BAA8B,CAAC;AAClD,mBAAmB,yCAAyC,CAAC;AAC7D,YAAY,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,qBAAqB,EAAE,MAAM,8BAA8B,CAAC;AAC/G,YAAY,EAAE,qBAAqB,EAAE,iBAAiB,EAAE,gBAAgB,EAAE,MAAM,8BAA8B,CAAC;AAC/G,YAAY,EAAE,sBAAsB,EAAE,iBAAiB,EAAE,kBAAkB,EAAE,MAAM,+BAA+B,CAAC;AACnH,mBAAmB,qCAAqC,CAAC;AACzD,YAAY,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAC;AACtF,YAAY,EACX,6BAA6B,EAC7B,uBAAuB,EACvB,wBAAwB,EACxB,+BAA+B,EAC/B,4BAA4B,GAC5B,MAAM,oCAAoC,CAAC;AAC5C,YAAY,EACX,gCAAgC,EAChC,gCAAgC,EAChC,mBAAmB,EACnB,oBAAoB,EACpB,2BAA2B,EAC3B,qCAAqC,EACrC,kCAAkC,EAClC,yBAAyB,EACzB,uCAAuC,EACvC,0BAA0B,GAC1B,MAAM,gCAAgC,CAAC;AACxC,mBAAmB,qCAAqC,CAAC;AACzD,mBAAmB,0CAA0C,CAAC;AAC9D,mBAAmB,yCAAyC,CAAC;AAC7D,mBAAmB,+CAA+C,CAAC;AACnE,YAAY,EACX,WAAW,EACX,4BAA4B,EAC5B,6BAA6B,EAC7B,oCAAoC,GACpC,MAAM,2CAA2C,CAAC;AAEnD,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,uBAAuB,CAAC;AAE7D;;GAEG;AACH,eAAO,MAAM,qBAAqB,EAAE,MAAM,CAAC,YAAY,EAAE,eAAe,EAAE,CA8DzE,CAAC;AAoBF,eAAO,MAAM,UAAU,EAAE,MAAM,CAAC,YAAY,EAAE,QAAQ,GAAG,SAAS,CAsDxD,CAAC;AAEX,MAAM,WAAW,WAAW;IAC3B,WAAW,EAAE,MAAM,CAAC;IACpB,EAAE,EAAE,MAAM,CAAC;CACX;AAED,MAAM,MAAM,aAAa,GACtB;IACA,QAAQ,EAAE,MAAM,CAAC;IACjB,IAAI,EAAE,OAAO,CAAC;CACb,GACD;IACA,IAAI,EAAE,KAAK,CAAC;QACX,KAAK,EAAE,MAAM,CAAC;QACd,KAAK,EAAE,MAAM,CAAC;KACd,CAAC,CAAC;IACH,IAAI,EAAE,OAAO,CAAC;CACb,GACD;IACA,QAAQ,EAAE,MAAM,CAAC;IACjB,IAAI,EAAE,KAAK,CAAC;CACX,GACD;IACA,KAAK,EAAE,MAAM,EAAE,EAAE,CAAC;IAClB,IAAI,EAAE,SAAS,CAAC;CACf,GACD;IACA,OAAO,EAAE,MAAM,CAAC;IAChB,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,MAAM,CAAC;CACZ,GACD;IACA,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,KAAK,CAAC;QACb,GAAG,EAAE,MAAM,CAAC;QACZ,KAAK,EAAE,MAAM,CAAC;QACd,IAAI,EAAE,MAAM,CAAC;KACb,CAAC,CAAC;IACH,IAAI,EAAE,kBAAkB,CAAC;CACxB,CAAC;AAEL,MAAM,WAAW,QAAQ;IACxB,MAAM,EAAE,aAAa,EAAE,CAAC;IACxB,OAAO,EAAE,aAAa,EAAE,CAAC;CACzB;AAED,MAAM,WAAW,QAAQ;IACxB,QAAQ,EAAE,WAAW,EAAE,CAAC;IACxB,IAAI,EAAE,QAAQ,CAAC;IACf,EAAE,EAAE,YAAY,CAAC;IACjB,WAAW,CAAC,EAAE,YAAY,CAAC;IAC3B,aAAa,CAAC,EAAE,OAAO,CAAC;IACxB,KAAK,EAAE,MAAM,CAAC;IACd,SAAS,EAAE,eAAe,EAAE,CAAC;IAC7B,OAAO,EAAE,WAAW,EAAE,CAAC;IACvB,MAAM,EAAE,WAAW,EAAE,CAAC;IACtB,MAAM,EAAE,WAAW,EAAE,CAAC;IACtB,OAAO,EAAE,MAAM,CAAC;IAChB,YAAY,EAAE,MAAM,EAAE,CAAC;IACvB,SAAS,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,MAAM,MAAM,cAAc,GAAG,IAAI,CAAC,QAAQ,EAAE,IAAI,GAAG,OAAO,GAAG,WAAW,CAAC,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/tasks/index.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,iBAAiB,CAAC;AA4CpD,mBAAmB,qCAAqC,CAAC;AACzD,mBAAmB,6CAA6C,CAAC;AACjE,YAAY,EACX,mBAAmB,EACnB,0BAA0B,EAC1B,oBAAoB,EACpB,4BAA4B,EAC5B,2BAA2B,EAC3B,0BAA0B,EAC1B,gCAAgC,EAChC,+BAA+B,GAC/B,MAAM,gCAAgC,CAAC;AACxC,mBAAmB,4CAA4C,CAAC;AAChE,mBAAmB,mCAAmC,CAAC;AACvD,mBAAmB,0BAA0B,CAAC;AAC9C,YAAY,EACX,wBAAwB,EACxB,yBAAyB,EACzB,gCAAgC,EAChC,6BAA6B,GAC7B,MAAM,qCAAqC,CAAC;AAC7C,mBAAmB,+BAA+B,CAAC;AACnD,YAAY,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,qBAAqB,EAAE,MAAM,8BAA8B,CAAC;AAC/G,mBAAmB,mCAAmC,CAAC;AACvD,mBAAmB,iCAAiC,CAAC;AACrD,mBAAmB,iCAAiC,CAAC;AACrD,mBAAmB,mCAAmC,CAAC;AACvD,mBAAmB,oCAAoC,CAAC;AACxD,mBAAmB,8BAA8B,CAAC;AAClD,mBAAmB,yCAAyC,CAAC;AAC7D,YAAY,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,qBAAqB,EAAE,MAAM,8BAA8B,CAAC;AAC/G,YAAY,EAAE,qBAAqB,EAAE,iBAAiB,EAAE,gBAAgB,EAAE,MAAM,8BAA8B,CAAC;AAC/G,YAAY,EAAE,sBAAsB,EAAE,iBAAiB,EAAE,kBAAkB,EAAE,MAAM,+BAA+B,CAAC;AACnH,mBAAmB,qCAAqC,CAAC;AACzD,YAAY,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAC;AACtF,YAAY,EACX,6BAA6B,EAC7B,uBAAuB,EACvB,wBAAwB,EACxB,+BAA+B,EAC/B,4BAA4B,GAC5B,MAAM,oCAAoC,CAAC;AAC5C,YAAY,EACX,gCAAgC,EAChC,gCAAgC,EAChC,mBAAmB,EACnB,oBAAoB,EACpB,2BAA2B,EAC3B,qCAAqC,EACrC,kCAAkC,EAClC,yBAAyB,EACzB,uCAAuC,EACvC,0BAA0B,GAC1B,MAAM,gCAAgC,CAAC;AACxC,mBAAmB,qCAAqC,CAAC;AACzD,mBAAmB,0CAA0C,CAAC;AAC9D,mBAAmB,yCAAyC,CAAC;AAC7D,mBAAmB,+CAA+C,CAAC;AACnE,YAAY,EACX,WAAW,EACX,4BAA4B,EAC5B,6BAA6B,EAC7B,oCAAoC,GACpC,MAAM,2CAA2C,CAAC;AAEnD,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,uBAAuB,CAAC;AAE7D;;GAEG;AACH,eAAO,MAAM,qBAAqB,EAAE,MAAM,CAAC,YAAY,EAAE,eAAe,EAAE,CA+DzE,CAAC;AAoBF,eAAO,MAAM,UAAU,EAAE,MAAM,CAAC,YAAY,EAAE,QAAQ,GAAG,SAAS,CAuDxD,CAAC;AAEX,MAAM,WAAW,WAAW;IAC3B,WAAW,EAAE,MAAM,CAAC;IACpB,EAAE,EAAE,MAAM,CAAC;CACX;AAED,MAAM,MAAM,aAAa,GACtB;IACA,QAAQ,EAAE,MAAM,CAAC;IACjB,IAAI,EAAE,OAAO,CAAC;CACb,GACD;IACA,IAAI,EAAE,KAAK,CAAC;QACX,KAAK,EAAE,MAAM,CAAC;QACd,KAAK,EAAE,MAAM,CAAC;KACd,CAAC,CAAC;IACH,IAAI,EAAE,OAAO,CAAC;CACb,GACD;IACA,QAAQ,EAAE,MAAM,CAAC;IACjB,IAAI,EAAE,KAAK,CAAC;CACX,GACD;IACA,KAAK,EAAE,MAAM,EAAE,EAAE,CAAC;IAClB,IAAI,EAAE,SAAS,CAAC;CACf,GACD;IACA,OAAO,EAAE,MAAM,CAAC;IAChB,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,MAAM,CAAC;CACZ,GACD;IACA,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,KAAK,CAAC;QACb,GAAG,EAAE,MAAM,CAAC;QACZ,KAAK,EAAE,MAAM,CAAC;QACd,IAAI,EAAE,MAAM,CAAC;KACb,CAAC,CAAC;IACH,IAAI,EAAE,kBAAkB,CAAC;CACxB,CAAC;AAEL,MAAM,WAAW,QAAQ;IACxB,MAAM,EAAE,aAAa,EAAE,CAAC;IACxB,OAAO,EAAE,aAAa,EAAE,CAAC;CACzB;AAED,MAAM,WAAW,QAAQ;IACxB,QAAQ,EAAE,WAAW,EAAE,CAAC;IACxB,IAAI,EAAE,QAAQ,CAAC;IACf,EAAE,EAAE,YAAY,CAAC;IACjB,WAAW,CAAC,EAAE,YAAY,CAAC;IAC3B,aAAa,CAAC,EAAE,OAAO,CAAC;IACxB,KAAK,EAAE,MAAM,CAAC;IACd,SAAS,EAAE,eAAe,EAAE,CAAC;IAC7B,OAAO,EAAE,WAAW,EAAE,CAAC;IACvB,MAAM,EAAE,WAAW,EAAE,CAAC;IACtB,MAAM,EAAE,WAAW,EAAE,CAAC;IACtB,OAAO,EAAE,MAAM,CAAC;IAChB,YAAY,EAAE,MAAM,EAAE,CAAC;IACvB,SAAS,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,MAAM,MAAM,cAAc,GAAG,IAAI,CAAC,QAAQ,EAAE,IAAI,GAAG,OAAO,GAAG,WAAW,CAAC,CAAC"}

package/dist/esm/tasks/index.js

@@ -104,6 +104,7 @@ export const TASKS_MODEL_LIBRARIES = {
     "text-to-3d": ["diffusers"],
     "image-to-3d": ["diffusers"],
     "any-to-any": ["transformers"],
+    "visual-document-retrieval": ["transformers"],
 };
 /**
  * Return the whole TaskData object for a certain task.
@@ -130,6 +131,7 @@ export const TASKS_DATA = {
     "automatic-speech-recognition": getData("automatic-speech-recognition", automaticSpeechRecognition),
     "depth-estimation": getData("depth-estimation", depthEstimation),
     "document-question-answering": getData("document-question-answering", documentQuestionAnswering),
+    "visual-document-retrieval": getData("visual-document-retrieval", placeholder),
     "feature-extraction": getData("feature-extraction", featureExtraction),
     "fill-mask": getData("fill-mask", fillMask),
     "graph-ml": undefined,

package/dist/esm/tasks/text-generation/data.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"data.d.ts","sourceRoot":"","sources":["../../../../src/tasks/text-generation/data.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAElD,QAAA,MAAM,QAAQ,EAAE,cA2Hf,CAAC;AAEF,eAAe,QAAQ,CAAC"}
+{"version":3,"file":"data.d.ts","sourceRoot":"","sources":["../../../../src/tasks/text-generation/data.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAElD,QAAA,MAAM,QAAQ,EAAE,cAyHf,CAAC;AAEF,eAAe,QAAQ,CAAC"}

package/dist/esm/tasks/text-generation/data.js

@@ -56,9 +56,7 @@ const taskData = {
         },
     ],
     models: [
-        { description: "A text-generation model trained to follow instructions.",
-            id: "google/gemma-2-2b-it",
-        },
+        { description: "A text-generation model trained to follow instructions.", id: "google/gemma-2-2b-it" },
         {
             description: "Smaller variant of one of the most powerful models.",
             id: "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B",

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@huggingface/tasks",
   "packageManager": "pnpm@8.10.5",
-  "version": "0.15.8",
+  "version": "0.16.0",
   "description": "List of ML tasks for huggingface.co/tasks",
   "repository": "https://github.com/huggingface/huggingface.js.git",
   "publishConfig": {

package/src/inference-providers.ts

@@ -1,16 +1,25 @@
-export const INFERENCE_PROVIDERS = ["hf-inference", "fal-ai", "replicate", "sambanova", "together"] as const;
+/// This list is for illustration purposes only.
+/// in the `tasks` sub-package, we do not need actual strong typing of the inference providers.
+const INFERENCE_PROVIDERS = [
+	"fal-ai",
+	"fireworks-ai",
+	"hf-inference",
+	"hyperbolic",
+	"replicate",
+	"sambanova",
+	"together",
+] as const;
 
-export type InferenceProvider = (typeof INFERENCE_PROVIDERS)[number];
+export type SnippetInferenceProvider = (typeof INFERENCE_PROVIDERS)[number] | string;
 
-export const HF_HUB_INFERENCE_PROXY_TEMPLATE = `https://huggingface.co/api/inference-proxy/{{PROVIDER}}`;
+export const HF_HUB_INFERENCE_PROXY_TEMPLATE = `https://router.huggingface.co/{{PROVIDER}}`;
 
 /**
  * URL to set as baseUrl in the OpenAI SDK.
  *
  * TODO(Expose this from HfInference in the future?)
  */
-export function openAIbaseUrl(provider: InferenceProvider): string {
-	return provider === "hf-inference"
-		? "https://api-inference.huggingface.co/v1/"
-		: HF_HUB_INFERENCE_PROXY_TEMPLATE.replace("{{PROVIDER}}", provider);
+export function openAIbaseUrl(provider: SnippetInferenceProvider): string {
+	const url = HF_HUB_INFERENCE_PROXY_TEMPLATE.replace("{{PROVIDER}}", provider);
+	return provider === "hf-inference" ? `${url}/v1` : url;
 }

package/src/library-to-tasks.ts

@@ -3,7 +3,7 @@ import type { PipelineType } from "./pipelines.js";
 
 /**
  * Mapping from library name to its supported tasks.
- * Inference API (serverless) should be disabled for all other (library, task) pairs beyond this mapping.
+ * HF-Inference API (serverless) should be disabled for all other (library, task) pairs beyond this mapping.
  * This mapping is partially generated automatically by "python-api-export-tasks" action in
  * huggingface/api-inference-community repo upon merge. For transformers, the mapping is manually
  * based on api-inference (hf_types.rs).

package/src/model-libraries-snippets.ts

@@ -83,6 +83,20 @@ function get_prompt_from_diffusers_model(model: ModelData): string | undefined {
 	}
 }
 
+export const ben2 = (model: ModelData): string[] => [
+	`import requests
+from PIL import Image
+from ben2 import AutoModel
+
+url = "https://huggingface.co/datasets/mishig/sample_images/resolve/main/teapot.jpg"
+image = Image.open(requests.get(url, stream=True).raw)
+
+model = AutoModel.from_pretrained("${model.id}")
+model.to("cuda").eval()
+foreground = model.inference(image)
+`,
+];
+
 export const bertopic = (model: ModelData): string[] => [
 	`from bertopic import BERTopic
 

package/src/model-libraries.ts

@@ -109,6 +109,13 @@ export const MODEL_LIBRARIES_UI_ELEMENTS = {
 		countDownloads: `path_extension:"pth"`,
 		snippets: snippets.audioseal,
 	},
+	ben2: {
+		prettyLabel: "BEN2",
+		repoName: "BEN2",
+		repoUrl: "https://github.com/PramaLLC/BEN2",
+		snippets: snippets.ben2,
+		filter: false,
+	},
 	bertopic: {
 		prettyLabel: "BERTopic",
 		repoName: "BERTopic",

package/src/pipelines.ts

@@ -676,6 +676,12 @@ export const PIPELINE_DATA = {
 		color: "red",
 		hideInDatasets: true,
 	},
+	"visual-document-retrieval": {
+		name: "Visual Document Retrieval",
+		modality: "multimodal",
+		color: "yellow",
+		hideInDatasets: true,
+	},
 	"any-to-any": {
 		name: "Any-to-Any",
 		modality: "multimodal",

package/src/snippets/common.ts

@@ -34,6 +34,6 @@ export function stringifyGenerationConfig(
 	const quote = opts.attributeKeyQuotes ? `"` : "";
 
 	return Object.entries(config)
-		.map(([key, val]) => `${quote}${key}${quote}${opts.attributeValueConnector}${val}`)
-		.join(`,${opts.indent}`);
+		.map(([key, val]) => `${quote}${key}${quote}${opts.attributeValueConnector}${val},`)
+		.join(`${opts.indent}`);
 }

package/src/snippets/curl.ts

@@ -1,4 +1,4 @@
-import { HF_HUB_INFERENCE_PROXY_TEMPLATE, type InferenceProvider } from "../inference-providers.js";
+import { HF_HUB_INFERENCE_PROXY_TEMPLATE, type SnippetInferenceProvider } from "../inference-providers.js";
 import type { PipelineType } from "../pipelines.js";
 import type { ChatCompletionInputMessage, GenerationParameters } from "../tasks/index.js";
 import { stringifyGenerationConfig, stringifyMessages } from "./common.js";
@@ -8,7 +8,7 @@ import type { InferenceSnippet, ModelDataMinimal } from "./types.js";
 export const snippetBasic = (
 	model: ModelDataMinimal,
 	accessToken: string,
-	provider: InferenceProvider
+	provider: SnippetInferenceProvider
 ): InferenceSnippet[] => {
 	if (provider !== "hf-inference") {
 		return [];
@@ -17,7 +17,7 @@ export const snippetBasic = (
 		{
 			client: "curl",
 			content: `\
-curl https://api-inference.huggingface.co/models/${model.id} \\
+curl https://router.huggingface.co/hf-inference/models/${model.id} \\
 	-X POST \\
 	-d '{"inputs": ${getModelInputSnippet(model, true)}}' \\
 	-H 'Content-Type: application/json' \\
@@ -29,7 +29,8 @@ curl https://api-inference.huggingface.co/models/${model.id} \\
 export const snippetTextGeneration = (
 	model: ModelDataMinimal,
 	accessToken: string,
-	provider: InferenceProvider,
+	provider: SnippetInferenceProvider,
+	providerModelId?: string,
 	opts?: {
 		streaming?: boolean;
 		messages?: ChatCompletionInputMessage[];
@@ -41,8 +42,9 @@ export const snippetTextGeneration = (
 	if (model.tags.includes("conversational")) {
 		const baseUrl =
 			provider === "hf-inference"
-				? `https://api-inference.huggingface.co/models/${model.id}/v1/chat/completions`
+				? `https://router.huggingface.co/hf-inference/models/${model.id}/v1/chat/completions`
 				: HF_HUB_INFERENCE_PROXY_TEMPLATE.replace("{{PROVIDER}}", provider) + "/v1/chat/completions";
+		const modelId = providerModelId ?? model.id;
 
 		// Conversational model detected, so we display a code snippet that features the Messages API
 		const streaming = opts?.streaming ?? true;
@@ -61,7 +63,7 @@ export const snippetTextGeneration = (
 -H 'Authorization: Bearer ${accessToken || `{API_TOKEN}`}' \\
 -H 'Content-Type: application/json' \\
 --data '{
-    "model": "${model.id}",
+    "model": "${modelId}",
     "messages": ${stringifyMessages(messages, {
 			indent: "\t",
 			attributeKeyQuotes: true,
@@ -71,7 +73,7 @@ export const snippetTextGeneration = (
 			indent: "\n    ",
 			attributeKeyQuotes: true,
 			attributeValueConnector: ": ",
-		})},
+		})}
     "stream": ${!!streaming}
 }'`,
 			},
@@ -84,7 +86,7 @@ export const snippetTextGeneration = (
 export const snippetZeroShotClassification = (
 	model: ModelDataMinimal,
 	accessToken: string,
-	provider: InferenceProvider
+	provider: SnippetInferenceProvider
 ): InferenceSnippet[] => {
 	if (provider !== "hf-inference") {
 		return [];
@@ -92,7 +94,7 @@ export const snippetZeroShotClassification = (
 	return [
 		{
 			client: "curl",
-			content: `curl https://api-inference.huggingface.co/models/${model.id} \\
+			content: `curl https://router.huggingface.co/hf-inference/models/${model.id} \\
 	-X POST \\
 	-d '{"inputs": ${getModelInputSnippet(model, true)}, "parameters": {"candidate_labels": ["refund", "legal", "faq"]}}' \\
 	-H 'Content-Type: application/json' \\
@@ -104,7 +106,7 @@ export const snippetZeroShotClassification = (
 export const snippetFile = (
 	model: ModelDataMinimal,
 	accessToken: string,
-	provider: InferenceProvider
+	provider: SnippetInferenceProvider
 ): InferenceSnippet[] => {
 	if (provider !== "hf-inference") {
 		return [];
@@ -112,7 +114,7 @@ export const snippetFile = (
 	return [
 		{
 			client: "curl",
-			content: `curl https://api-inference.huggingface.co/models/${model.id} \\
+			content: `curl https://router.huggingface.co/hf-inference/models/${model.id} \\
 	-X POST \\
 	--data-binary '@${getModelInputSnippet(model, true, true)}' \\
 	-H 'Authorization: Bearer ${accessToken || `{API_TOKEN}`}'`,
@@ -126,7 +128,8 @@ export const curlSnippets: Partial<
 		(
 			model: ModelDataMinimal,
 			accessToken: string,
-			provider: InferenceProvider,
+			provider: SnippetInferenceProvider,
+			providerModelId?: string,
 			opts?: Record<string, unknown>
 		) => InferenceSnippet[]
 	>
@@ -160,10 +163,11 @@ export const curlSnippets: Partial<
 export function getCurlInferenceSnippet(
 	model: ModelDataMinimal,
 	accessToken: string,
-	provider: InferenceProvider,
+	provider: SnippetInferenceProvider,
+	providerModelId?: string,
 	opts?: Record<string, unknown>
 ): InferenceSnippet[] {
 	return model.pipeline_tag && model.pipeline_tag in curlSnippets
-		? curlSnippets[model.pipeline_tag]?.(model, accessToken, provider, opts) ?? []
+		? curlSnippets[model.pipeline_tag]?.(model, accessToken, provider, providerModelId, opts) ?? []
 		: [];
 }