@huggingface/tasks 0.6.1 → 0.7.1

package/dist/index.cjs

@@ -23,7 +23,7 @@ __export(src_exports, {
   ALL_DISPLAY_MODEL_LIBRARY_KEYS: () => ALL_DISPLAY_MODEL_LIBRARY_KEYS,
   ALL_MODEL_LIBRARY_KEYS: () => ALL_MODEL_LIBRARY_KEYS,
   InferenceDisplayability: () => InferenceDisplayability,
-  LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS: () => LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS,
+  LIBRARY_TASK_MAPPING: () => LIBRARY_TASK_MAPPING,
   MAPPING_DEFAULT_WIDGET: () => MAPPING_DEFAULT_WIDGET,
   MODALITIES: () => MODALITIES,
   MODALITY_LABELS: () => MODALITY_LABELS,
@@ -40,7 +40,7 @@ __export(src_exports, {
 module.exports = __toCommonJS(src_exports);
 
 // src/library-to-tasks.ts
-var LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS = {
+var LIBRARY_TASK_MAPPING = {
   "adapter-transformers": ["question-answering", "text-classification", "token-classification"],
   allennlp: ["question-answering"],
   asteroid: [
@@ -76,6 +76,25 @@ var LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS = {
   ],
   stanza: ["token-classification"],
   timm: ["image-classification"],
+  transformers: [
+    "audio-classification",
+    "automatic-speech-recognition",
+    "depth-estimation",
+    "document-question-answering",
+    "feature-extraction",
+    "fill-mask",
+    "image-classification",
+    "image-segmentation",
+    "image-to-text",
+    "image-to-image",
+    "object-detection",
+    "question-answering",
+    "text-generation",
+    "text2text-generation",
+    "visual-question-answering",
+    "zero-shot-classification",
+    "zero-shot-image-classification"
+  ],
   mindspore: ["image-classification"]
 };
 
@@ -3847,7 +3866,7 @@ var data_default34 = taskData34;
 // src/tasks/index.ts
 var TASKS_MODEL_LIBRARIES = {
   "audio-classification": ["speechbrain", "transformers", "transformers.js"],
-  "audio-to-audio": ["asteroid", "speechbrain"],
+  "audio-to-audio": ["asteroid", "fairseq", "speechbrain"],
   "automatic-speech-recognition": ["espnet", "nemo", "speechbrain", "transformers", "transformers.js"],
   "depth-estimation": ["transformers", "transformers.js"],
   "document-question-answering": ["transformers", "transformers.js"],
@@ -5324,7 +5343,7 @@ function hasJsInferenceSnippet(model) {
   ALL_DISPLAY_MODEL_LIBRARY_KEYS,
   ALL_MODEL_LIBRARY_KEYS,
   InferenceDisplayability,
-  LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS,
+  LIBRARY_TASK_MAPPING,
   MAPPING_DEFAULT_WIDGET,
   MODALITIES,
   MODALITY_LABELS,

package/dist/index.d.ts

@@ -507,7 +507,10 @@ type SpecialTokensMap = {
  */
 interface TokenizerConfig extends SpecialTokensMap {
     use_default_system_prompt?: boolean;
-    chat_template?: string;
+    chat_template?: string | Array<{
+        name: string;
+        template: string;
+    }>;
 }
 
 declare enum InferenceDisplayability {
@@ -1145,13 +1148,13 @@ declare const ALL_MODEL_LIBRARY_KEYS: ("sklearn" | "adapter-transformers" | "all
 declare const ALL_DISPLAY_MODEL_LIBRARY_KEYS: ("sklearn" | "adapter-transformers" | "allennlp" | "asteroid" | "audiocraft" | "bertopic" | "diffusers" | "doctr" | "espnet" | "fairseq" | "fastai" | "fasttext" | "flair" | "gliner" | "grok" | "keras" | "k2" | "mindspore" | "ml-agents" | "mlx" | "mlx-image" | "nemo" | "open_clip" | "paddlenlp" | "peft" | "pyannote-audio" | "pythae" | "sample-factory" | "sentence-transformers" | "setfit" | "spacy" | "span-marker" | "speechbrain" | "stable-baselines3" | "stanza" | "tensorflowtts" | "timm" | "transformers" | "transformers.js" | "unity-sentis" | "whisperkit")[];
 
 /**
- * Mapping from library name (excluding Transformers) to its supported tasks.
+ * Mapping from library name to its supported tasks.
  * Inference API (serverless) should be disabled for all other (library, task) pairs beyond this mapping.
- * As an exception, we assume Transformers supports all inference tasks.
- * This mapping is generated automatically by "python-api-export-tasks" action in huggingface/api-inference-community repo upon merge.
- * Ref: https://github.com/huggingface/api-inference-community/pull/158
+ * 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.
  */
-declare const LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS: Partial<Record<ModelLibraryKey, PipelineType[]>>;
+declare const LIBRARY_TASK_MAPPING: Partial<Record<ModelLibraryKey, PipelineType[]>>;
 
 type PerLanguageMapping = Map<WidgetType, string[] | WidgetExample[]>;
 declare const MAPPING_DEFAULT_WIDGET: Map<string, PerLanguageMapping>;
@@ -2977,6 +2980,46 @@ interface TextGenerationOutputToken {
     text: string;
     [property: string]: unknown;
 }
+/**
+ * Text Generation Stream Output
+ */
+interface TextGenerationStreamOutput {
+    /**
+     * Generation details. Only available when the generation is finished.
+     */
+    details?: TextGenerationStreamDetails;
+    /**
+     * The complete generated text. Only available when the generation is finished.
+     */
+    generated_text?: string;
+    /**
+     * The token index within the stream. Optional to support older clients that omit it.
+     */
+    index?: number;
+    /**
+     * Generated token.
+     */
+    token: TextGenerationOutputToken;
+    [property: string]: unknown;
+}
+/**
+ * Generation details. Only available when the generation is finished.
+ */
+interface TextGenerationStreamDetails {
+    /**
+     * The reason why the generation was stopped.
+     */
+    finish_reason: TextGenerationFinishReason;
+    /**
+     * The number of generated tokens
+     */
+    generated_tokens: number;
+    /**
+     * The random seed used for generation
+     */
+    seed: number;
+    [property: string]: unknown;
+}
 
 /**
  * Inference code generated from the JSON schema spec in ./spec
@@ -3471,4 +3514,4 @@ declare namespace index {
   };
 }
 
-export { ALL_DISPLAY_MODEL_LIBRARY_KEYS, ALL_MODEL_LIBRARY_KEYS, AddedToken, AudioClassificationInput, AudioClassificationOutput, AudioClassificationOutputElement, AudioClassificationParameters, AutomaticSpeechRecognitionInput, AutomaticSpeechRecognitionOutput, AutomaticSpeechRecognitionOutputChunk, AutomaticSpeechRecognitionParameters, BoundingBox, ChatCompletionFinishReason, ChatCompletionInput, ChatCompletionInputMessage, ChatCompletionOutput, ChatCompletionOutputChoice, ChatCompletionOutputChoiceMessage, ChatCompletionStreamOutput, ChatCompletionStreamOutputChoice, ChatCompletionStreamOutputDelta, ChatMessage, ClassificationOutputTransform$1 as ClassificationOutputTransform, DepthEstimationInput, DepthEstimationOutput, DocumentQuestionAnsweringInput, DocumentQuestionAnsweringInputData, DocumentQuestionAnsweringOutput, DocumentQuestionAnsweringOutputElement, DocumentQuestionAnsweringParameters, EarlyStoppingUnion$2 as EarlyStoppingUnion, ExampleRepo, FeatureExtractionInput, FeatureExtractionOutput, FillMaskInput, FillMaskOutput, FillMaskOutputElement, FillMaskParameters, GenerationParameters$2 as GenerationParameters, ImageClassificationInput, ImageClassificationOutput, ImageClassificationOutputElement, ImageClassificationParameters, ImageSegmentationInput, ImageSegmentationOutput, ImageSegmentationOutputElement, ImageSegmentationParameters, ImageSegmentationSubtask, ImageToImageInput, ImageToImageOutput, ImageToImageParameters, ImageToTextInput, ImageToTextOutput, ImageToTextParameters, InferenceDisplayability, LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS, LibraryUiElement, MAPPING_DEFAULT_WIDGET, MODALITIES, MODALITY_LABELS, MODEL_LIBRARIES_UI_ELEMENTS, Modality, ModelData, ModelLibraryKey, ObjectDetectionInput, ObjectDetectionOutput, ObjectDetectionOutputElement, ObjectDetectionParameters, PIPELINE_DATA, PIPELINE_TYPES, PIPELINE_TYPES_SET, PipelineData, PipelineType, QuestionAnsweringInput, QuestionAnsweringInputData, QuestionAnsweringOutput, QuestionAnsweringOutputElement, QuestionAnsweringParameters, SPECIAL_TOKENS_ATTRIBUTES, SUBTASK_TYPES, SentenceSimilarityInput, SentenceSimilarityInputData, SentenceSimilarityOutput, SpecialTokensMap, SummarizationInput, SummarizationOutput, TASKS_DATA, TASKS_MODEL_LIBRARIES, TableQuestionAnsweringInput, TableQuestionAnsweringInputData, TableQuestionAnsweringOutput, TableQuestionAnsweringOutputElement, TargetSize$1 as TargetSize, TaskData, TaskDataCustom, TaskDemo, TaskDemoEntry, Text2TextGenerationParameters, Text2TextGenerationTruncationStrategy, TextClassificationInput, TextClassificationOutput, TextClassificationOutputElement, TextClassificationParameters, TextGenerationFinishReason, TextGenerationInput, TextGenerationOutput, TextGenerationOutputDetails, TextGenerationOutputSequenceDetails, TextGenerationOutputToken, TextGenerationParameters, TextGenerationPrefillToken, TextToAudioParameters, TextToImageInput, TextToImageOutput, TextToImageParameters, TextToSpeechInput, TextToSpeechOutput, TokenClassificationAggregationStrategy, TokenClassificationInput, TokenClassificationOutput, TokenClassificationOutputElement, TokenClassificationParameters, TokenizerConfig, TransformersInfo, TranslationInput, TranslationOutput, VideoClassificationInput, VideoClassificationOutput, VideoClassificationOutputElement, VideoClassificationParameters, VisualQuestionAnsweringInput, VisualQuestionAnsweringInputData, VisualQuestionAnsweringOutput, VisualQuestionAnsweringOutputElement, VisualQuestionAnsweringParameters, WidgetExample, WidgetExampleAssetAndPromptInput, WidgetExampleAssetAndTextInput, WidgetExampleAssetAndZeroShotInput, WidgetExampleAssetInput, WidgetExampleAttribute, WidgetExampleChatInput, WidgetExampleOutput, WidgetExampleOutputAnswerScore, WidgetExampleOutputLabels, WidgetExampleOutputText, WidgetExampleOutputUrl, WidgetExampleSentenceSimilarityInput, WidgetExampleStructuredDataInput, WidgetExampleTableDataInput, WidgetExampleTextAndContextInput, WidgetExampleTextAndTableInput, WidgetExampleTextInput, WidgetExampleZeroShotTextInput, WidgetType, WordBox, ZeroShotClassificationInput, ZeroShotClassificationInputData, ZeroShotClassificationOutput, ZeroShotClassificationOutputElement, ZeroShotClassificationParameters, ZeroShotImageClassificationInput, ZeroShotImageClassificationInputData, ZeroShotImageClassificationOutput, ZeroShotImageClassificationOutputElement, ZeroShotImageClassificationParameters, ZeroShotObjectDetectionInput, ZeroShotObjectDetectionInputData, ZeroShotObjectDetectionOutput, ZeroShotObjectDetectionOutputElement, index as snippets };
+export { ALL_DISPLAY_MODEL_LIBRARY_KEYS, ALL_MODEL_LIBRARY_KEYS, AddedToken, AudioClassificationInput, AudioClassificationOutput, AudioClassificationOutputElement, AudioClassificationParameters, AutomaticSpeechRecognitionInput, AutomaticSpeechRecognitionOutput, AutomaticSpeechRecognitionOutputChunk, AutomaticSpeechRecognitionParameters, BoundingBox, ChatCompletionFinishReason, ChatCompletionInput, ChatCompletionInputMessage, ChatCompletionOutput, ChatCompletionOutputChoice, ChatCompletionOutputChoiceMessage, ChatCompletionStreamOutput, ChatCompletionStreamOutputChoice, ChatCompletionStreamOutputDelta, ChatMessage, ClassificationOutputTransform$1 as ClassificationOutputTransform, DepthEstimationInput, DepthEstimationOutput, DocumentQuestionAnsweringInput, DocumentQuestionAnsweringInputData, DocumentQuestionAnsweringOutput, DocumentQuestionAnsweringOutputElement, DocumentQuestionAnsweringParameters, EarlyStoppingUnion$2 as EarlyStoppingUnion, ExampleRepo, FeatureExtractionInput, FeatureExtractionOutput, FillMaskInput, FillMaskOutput, FillMaskOutputElement, FillMaskParameters, GenerationParameters$2 as GenerationParameters, ImageClassificationInput, ImageClassificationOutput, ImageClassificationOutputElement, ImageClassificationParameters, ImageSegmentationInput, ImageSegmentationOutput, ImageSegmentationOutputElement, ImageSegmentationParameters, ImageSegmentationSubtask, ImageToImageInput, ImageToImageOutput, ImageToImageParameters, ImageToTextInput, ImageToTextOutput, ImageToTextParameters, InferenceDisplayability, LIBRARY_TASK_MAPPING, LibraryUiElement, MAPPING_DEFAULT_WIDGET, MODALITIES, MODALITY_LABELS, MODEL_LIBRARIES_UI_ELEMENTS, Modality, ModelData, ModelLibraryKey, ObjectDetectionInput, ObjectDetectionOutput, ObjectDetectionOutputElement, ObjectDetectionParameters, PIPELINE_DATA, PIPELINE_TYPES, PIPELINE_TYPES_SET, PipelineData, PipelineType, QuestionAnsweringInput, QuestionAnsweringInputData, QuestionAnsweringOutput, QuestionAnsweringOutputElement, QuestionAnsweringParameters, SPECIAL_TOKENS_ATTRIBUTES, SUBTASK_TYPES, SentenceSimilarityInput, SentenceSimilarityInputData, SentenceSimilarityOutput, SpecialTokensMap, SummarizationInput, SummarizationOutput, TASKS_DATA, TASKS_MODEL_LIBRARIES, TableQuestionAnsweringInput, TableQuestionAnsweringInputData, TableQuestionAnsweringOutput, TableQuestionAnsweringOutputElement, TargetSize$1 as TargetSize, TaskData, TaskDataCustom, TaskDemo, TaskDemoEntry, Text2TextGenerationParameters, Text2TextGenerationTruncationStrategy, TextClassificationInput, TextClassificationOutput, TextClassificationOutputElement, TextClassificationParameters, TextGenerationFinishReason, TextGenerationInput, TextGenerationOutput, TextGenerationOutputDetails, TextGenerationOutputSequenceDetails, TextGenerationOutputToken, TextGenerationParameters, TextGenerationPrefillToken, TextGenerationStreamDetails, TextGenerationStreamOutput, TextToAudioParameters, TextToImageInput, TextToImageOutput, TextToImageParameters, TextToSpeechInput, TextToSpeechOutput, TokenClassificationAggregationStrategy, TokenClassificationInput, TokenClassificationOutput, TokenClassificationOutputElement, TokenClassificationParameters, TokenizerConfig, TransformersInfo, TranslationInput, TranslationOutput, VideoClassificationInput, VideoClassificationOutput, VideoClassificationOutputElement, VideoClassificationParameters, VisualQuestionAnsweringInput, VisualQuestionAnsweringInputData, VisualQuestionAnsweringOutput, VisualQuestionAnsweringOutputElement, VisualQuestionAnsweringParameters, WidgetExample, WidgetExampleAssetAndPromptInput, WidgetExampleAssetAndTextInput, WidgetExampleAssetAndZeroShotInput, WidgetExampleAssetInput, WidgetExampleAttribute, WidgetExampleChatInput, WidgetExampleOutput, WidgetExampleOutputAnswerScore, WidgetExampleOutputLabels, WidgetExampleOutputText, WidgetExampleOutputUrl, WidgetExampleSentenceSimilarityInput, WidgetExampleStructuredDataInput, WidgetExampleTableDataInput, WidgetExampleTextAndContextInput, WidgetExampleTextAndTableInput, WidgetExampleTextInput, WidgetExampleZeroShotTextInput, WidgetType, WordBox, ZeroShotClassificationInput, ZeroShotClassificationInputData, ZeroShotClassificationOutput, ZeroShotClassificationOutputElement, ZeroShotClassificationParameters, ZeroShotImageClassificationInput, ZeroShotImageClassificationInputData, ZeroShotImageClassificationOutput, ZeroShotImageClassificationOutputElement, ZeroShotImageClassificationParameters, ZeroShotObjectDetectionInput, ZeroShotObjectDetectionInputData, ZeroShotObjectDetectionOutput, ZeroShotObjectDetectionOutputElement, index as snippets };

package/dist/index.js

@@ -5,7 +5,7 @@ var __export = (target, all) => {
 };
 
 // src/library-to-tasks.ts
-var LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS = {
+var LIBRARY_TASK_MAPPING = {
   "adapter-transformers": ["question-answering", "text-classification", "token-classification"],
   allennlp: ["question-answering"],
   asteroid: [
@@ -41,6 +41,25 @@ var LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS = {
   ],
   stanza: ["token-classification"],
   timm: ["image-classification"],
+  transformers: [
+    "audio-classification",
+    "automatic-speech-recognition",
+    "depth-estimation",
+    "document-question-answering",
+    "feature-extraction",
+    "fill-mask",
+    "image-classification",
+    "image-segmentation",
+    "image-to-text",
+    "image-to-image",
+    "object-detection",
+    "question-answering",
+    "text-generation",
+    "text2text-generation",
+    "visual-question-answering",
+    "zero-shot-classification",
+    "zero-shot-image-classification"
+  ],
   mindspore: ["image-classification"]
 };
 
@@ -3812,7 +3831,7 @@ var data_default34 = taskData34;
 // src/tasks/index.ts
 var TASKS_MODEL_LIBRARIES = {
   "audio-classification": ["speechbrain", "transformers", "transformers.js"],
-  "audio-to-audio": ["asteroid", "speechbrain"],
+  "audio-to-audio": ["asteroid", "fairseq", "speechbrain"],
   "automatic-speech-recognition": ["espnet", "nemo", "speechbrain", "transformers", "transformers.js"],
   "depth-estimation": ["transformers", "transformers.js"],
   "document-question-answering": ["transformers", "transformers.js"],
@@ -5288,7 +5307,7 @@ export {
   ALL_DISPLAY_MODEL_LIBRARY_KEYS,
   ALL_MODEL_LIBRARY_KEYS,
   InferenceDisplayability,
-  LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS,
+  LIBRARY_TASK_MAPPING,
   MAPPING_DEFAULT_WIDGET,
   MODALITIES,
   MODALITY_LABELS,

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@huggingface/tasks",
   "packageManager": "pnpm@8.10.5",
-  "version": "0.6.1",
+  "version": "0.7.1",
   "description": "List of ML tasks for huggingface.co/tasks",
   "repository": "https://github.com/huggingface/huggingface.js.git",
   "publishConfig": {

package/src/index.ts

@@ -1,4 +1,4 @@
-export { LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS } from "./library-to-tasks";
+export { LIBRARY_TASK_MAPPING } from "./library-to-tasks";
 export { MAPPING_DEFAULT_WIDGET } from "./default-widget-inputs";
 export type { TaskData, TaskDemo, TaskDemoEntry, ExampleRepo } from "./tasks";
 export * from "./tasks";

package/src/library-to-tasks.ts

@@ -2,13 +2,13 @@ import type { ModelLibraryKey } from "./model-libraries";
 import type { PipelineType } from "./pipelines";
 
 /**
- * Mapping from library name (excluding Transformers) to its supported tasks.
+ * Mapping from library name to its supported tasks.
  * Inference API (serverless) should be disabled for all other (library, task) pairs beyond this mapping.
- * As an exception, we assume Transformers supports all inference tasks.
- * This mapping is generated automatically by "python-api-export-tasks" action in huggingface/api-inference-community repo upon merge.
- * Ref: https://github.com/huggingface/api-inference-community/pull/158
+ * 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.
  */
-export const LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS: Partial<Record<ModelLibraryKey, PipelineType[]>> = {
+export const LIBRARY_TASK_MAPPING: Partial<Record<ModelLibraryKey, PipelineType[]>> = {
 	"adapter-transformers": ["question-answering", "text-classification", "token-classification"],
 	allennlp: ["question-answering"],
 	asteroid: [
@@ -44,5 +44,24 @@ export const LIBRARY_TASK_MAPPING_EXCLUDING_TRANSFORMERS: Partial<Record<ModelLi
 	],
 	stanza: ["token-classification"],
 	timm: ["image-classification"],
+	transformers: [
+		"audio-classification",
+		"automatic-speech-recognition",
+		"depth-estimation",
+		"document-question-answering",
+		"feature-extraction",
+		"fill-mask",
+		"image-classification",
+		"image-segmentation",
+		"image-to-text",
+		"image-to-image",
+		"object-detection",
+		"question-answering",
+		"text-generation",
+		"text2text-generation",
+		"visual-question-answering",
+		"zero-shot-classification",
+		"zero-shot-image-classification",
+	],
 	mindspore: ["image-classification"],
 };

package/src/tasks/index.ts

@@ -92,6 +92,8 @@ export type {
 	TextGenerationParameters,
 	TextGenerationOutputSequenceDetails,
 	TextGenerationOutputToken,
+	TextGenerationStreamDetails,
+	TextGenerationStreamOutput,
 } from "./text-generation/inference";
 export type * from "./video-classification/inference";
 export type * from "./visual-question-answering/inference";
@@ -112,7 +114,7 @@ import type { ModelLibraryKey } from "../model-libraries";
  */
 export const TASKS_MODEL_LIBRARIES: Record<PipelineType, ModelLibraryKey[]> = {
 	"audio-classification": ["speechbrain", "transformers", "transformers.js"],
-	"audio-to-audio": ["asteroid", "speechbrain"],
+	"audio-to-audio": ["asteroid", "fairseq", "speechbrain"],
 	"automatic-speech-recognition": ["espnet", "nemo", "speechbrain", "transformers", "transformers.js"],
 	"depth-estimation": ["transformers", "transformers.js"],
 	"document-question-answering": ["transformers", "transformers.js"],

package/src/tokenizer-data.ts

@@ -28,5 +28,5 @@ export type SpecialTokensMap = {
  */
 export interface TokenizerConfig extends SpecialTokensMap {
 	use_default_system_prompt?: boolean;
-	chat_template?: string;
+	chat_template?: string | Array<{ name: string; template: string }>;
 }