@huggingface/tasks 0.15.8 → 0.15.9

package/src/snippets/python.ts

@@ -1,11 +1,45 @@
-import { HF_HUB_INFERENCE_PROXY_TEMPLATE, openAIbaseUrl, type InferenceProvider } from "../inference-providers.js";
-import type { PipelineType } from "../pipelines.js";
+import {
+	HF_HUB_INFERENCE_PROXY_TEMPLATE,
+	openAIbaseUrl,
+	type SnippetInferenceProvider,
+} from "../inference-providers.js";
+import type { PipelineType, WidgetType } from "../pipelines.js";
 import type { ChatCompletionInputMessage, GenerationParameters } from "../tasks/index.js";
 import { stringifyGenerationConfig, stringifyMessages } from "./common.js";
 import { getModelInputSnippet } from "./inputs.js";
 import type { InferenceSnippet, ModelDataMinimal } from "./types.js";
 
-const snippetImportInferenceClient = (accessToken: string, provider: InferenceProvider): string =>
+const HFH_INFERENCE_CLIENT_METHODS: Partial<Record<WidgetType, string>> = {
+	"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: string, provider: SnippetInferenceProvider): string =>
 	`\
 from huggingface_hub import InferenceClient
 
@@ -17,7 +51,7 @@ client = InferenceClient(
 export const snippetConversational = (
 	model: ModelDataMinimal,
 	accessToken: string,
-	provider: InferenceProvider,
+	provider: SnippetInferenceProvider,
 	opts?: {
 		streaming?: boolean;
 		messages?: ChatCompletionInputMessage[];
@@ -53,7 +87,7 @@ messages = ${messagesStr}
 stream = client.chat.completions.create(
 	model="${model.id}", 
 	messages=messages, 
-	${configStr},
+	${configStr}
 	stream=True
 )
 
@@ -75,7 +109,7 @@ messages = ${messagesStr}
 stream = client.chat.completions.create(
     model="${model.id}", 
 	messages=messages, 
-	${configStr},
+	${configStr}
 	stream=True
 )
 
@@ -164,8 +198,30 @@ output = query({
 	];
 };
 
-export const snippetBasic = (model: ModelDataMinimal): InferenceSnippet[] => {
+export const snippetBasic = (
+	model: ModelDataMinimal,
+	accessToken: string,
+	provider: SnippetInferenceProvider
+): InferenceSnippet[] => {
 	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="${provider}",
+)
+
+print(result)
+`,
+					},
+			  ]
+			: []),
 		{
 			client: "requests",
 			content: `\
@@ -199,7 +255,7 @@ output = query(${getModelInputSnippet(model)})`,
 export const snippetTextToImage = (
 	model: ModelDataMinimal,
 	accessToken: string,
-	provider: InferenceProvider
+	provider: SnippetInferenceProvider
 ): InferenceSnippet[] => {
 	return [
 		{
@@ -337,7 +393,7 @@ export const pythonSnippets: Partial<
 		(
 			model: ModelDataMinimal,
 			accessToken: string,
-			provider: InferenceProvider,
+			provider: SnippetInferenceProvider,
 			opts?: Record<string, unknown>
 		) => InferenceSnippet[]
 	>
@@ -375,7 +431,7 @@ export const pythonSnippets: Partial<
 export function getPythonInferenceSnippet(
 	model: ModelDataMinimal,
 	accessToken: string,
-	provider: InferenceProvider,
+	provider: SnippetInferenceProvider,
 	opts?: Record<string, unknown>
 ): InferenceSnippet[] {
 	if (model.tags.includes("conversational")) {
@@ -387,11 +443,6 @@ export function getPythonInferenceSnippet(
 				? pythonSnippets[model.pipeline_tag]?.(model, accessToken, provider) ?? []
 				: [];
 
-		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,
@@ -400,7 +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}`

package/src/tasks/audio-classification/about.md

@@ -33,7 +33,7 @@ import json
 import requests
 
 headers = {"Authorization": f"Bearer {API_TOKEN}"}
-API_URL = "https://api-inference.huggingface.co/models/superb/hubert-large-superb-er"
+API_URL = "https://router.huggingface.co/hf-inference/models/superb/hubert-large-superb-er"
 
 def query(filename):
     with open(filename, "rb") as f:

package/src/tasks/audio-to-audio/about.md

@@ -19,7 +19,7 @@ import json
 import requests
 
 headers = {"Authorization": f"Bearer {API_TOKEN}"}
-API_URL = "https://api-inference.huggingface.co/models/speechbrain/mtl-mimic-voicebank"
+API_URL = "https://router.huggingface.co/hf-inference/models/speechbrain/mtl-mimic-voicebank"
 
 def query(filename):
     with open(filename, "rb") as f:

package/src/tasks/automatic-speech-recognition/about.md

@@ -25,7 +25,7 @@ import json
 import requests
 
 headers = {"Authorization": f"Bearer {API_TOKEN}"}
-API_URL = "https://api-inference.huggingface.co/models/openai/whisper-large-v3"
+API_URL = "https://router.huggingface.co/hf-inference/models/openai/whisper-large-v3"
 
 def query(filename):
     with open(filename, "rb") as f:

package/src/tasks/feature-extraction/spec/input.json

@@ -8,11 +8,18 @@
 	"properties": {
 		"inputs": {
 			"title": "FeatureExtractionInputs",
+			"description": "The text or list of texts to embed.",
 			"oneOf": [
-				{ "type": "string" },
-				{ "type": "array", "items": { "type": "string" } }
-			],
-			"description": "The text or list of texts to embed."
+				{
+					"type": "string"
+				},
+				{
+					"type": "array",
+					"items": {
+						"type": "string"
+					}
+				}
+			]
 		},
 		"normalize": {
 			"type": "boolean",

package/src/tasks/image-text-to-text/about.md

@@ -76,7 +76,7 @@ outputs[0]["generated_text"]
 You can also use the Inference API to test image-text-to-text models. You need to use a [Hugging Face token](https://huggingface.co/settings/tokens) for authentication.
 
 ```bash
-curl https://api-inference.huggingface.co/models/meta-llama/Llama-3.2-11B-Vision-Instruct \
+curl https://router.huggingface.co/hf-inference/models/meta-llama/Llama-3.2-11B-Vision-Instruct \
 	-X POST \
 	-d '{"messages": [{"role": "user","content": [{"type": "image"}, {"type": "text", "text": "Can you describe the image?"}]}]}' \
 	-H "Content-Type: application/json" \

package/src/tasks/index.ts

@@ -174,6 +174,7 @@ export const TASKS_MODEL_LIBRARIES: Record<PipelineType, ModelLibraryKey[]> = {
 	"text-to-3d": ["diffusers"],
 	"image-to-3d": ["diffusers"],
 	"any-to-any": ["transformers"],
+	"visual-document-retrieval": ["transformers"],
 };
 
 /**
@@ -202,6 +203,7 @@ export const TASKS_DATA: Record<PipelineType, TaskData | undefined> = {
 	"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/src/tasks/sentence-similarity/about.md

@@ -22,7 +22,7 @@ You can infer with Passage Ranking models using [Inference Endpoints](https://hu
 import json
 import requests
 
-API_URL = "https://api-inference.huggingface.co/models/sentence-transformers/msmarco-distilbert-base-tas-b"
+API_URL = "https://router.huggingface.co/hf-inference/models/sentence-transformers/msmarco-distilbert-base-tas-b"
 headers = {"Authorization": f"Bearer {api_token}"}
 
 def query(payload):
@@ -51,7 +51,7 @@ Semantic Textual Similarity is the task of evaluating how similar two texts are
 import json
 import requests
 
-API_URL = "https://api-inference.huggingface.co/models/sentence-transformers/all-MiniLM-L6-v2"
+API_URL = "https://router.huggingface.co/hf-inference/models/sentence-transformers/all-MiniLM-L6-v2"
 headers = {"Authorization": f"Bearer {api_token}"}
 
 def query(payload):

package/src/tasks/text-generation/data.ts

@@ -61,9 +61,7 @@ const taskData: TaskDataCustom = {
 		},
 	],
 	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/src/tasks/text-to-speech/about.md

@@ -19,7 +19,7 @@ import json
 import requests
 
 headers = {"Authorization": f"Bearer {API_TOKEN}"}
-API_URL = "https://api-inference.huggingface.co/models/microsoft/speecht5_tts"
+API_URL = "https://router.huggingface.co/hf-inference/models/microsoft/speecht5_tts"
 
 def query(payload):
 	response = requests.post(API_URL, headers=headers, json=payload)