@huggingface/tasks 0.2.2 → 0.3.1

package/src/tasks/fill-mask/spec/output.json

@@ -19,7 +19,7 @@
 				"type": "integer",
 				"description": "The predicted token id (to replace the masked one)."
 			},
-			"tokenStr": {
+			"token_str": {
 				"type": "string",
 				"description": "The predicted token (to replace the masked one)."
 			}

package/src/tasks/image-classification/inference.ts

@@ -10,7 +10,7 @@ export interface ImageClassificationInput {
 	/**
 	 * The input image data
 	 */
-	data: unknown;
+	inputs: unknown;
 	/**
 	 * Additional inference parameters
 	 */
@@ -23,11 +23,11 @@ export interface ImageClassificationInput {
  * Additional inference parameters for Image Classification
  */
 export interface ImageClassificationParameters {
-	functionToApply?: ClassificationOutputTransform;
+	function_to_apply?: ClassificationOutputTransform;
 	/**
 	 * When specified, limits the output to the top K most probable classes.
 	 */
-	topK?: number;
+	top_k?: number;
 	[property: string]: unknown;
 }
 /**

package/src/tasks/image-classification/spec/input.json

@@ -5,7 +5,7 @@
 	"title": "ImageClassificationInput",
 	"type": "object",
 	"properties": {
-		"data": {
+		"inputs": {
 			"description": "The input image data"
 		},
 		"parameters": {
@@ -19,16 +19,16 @@
 			"description": "Additional inference parameters for Image Classification",
 			"type": "object",
 			"properties": {
-				"functionToApply": {
+				"function_to_apply": {
 					"title": "ImageClassificationOutputTransform",
 					"$ref": "/inference/schemas/common-definitions.json#/definitions/ClassificationOutputTransform"
 				},
-				"topK": {
+				"top_k": {
 					"type": "integer",
 					"description": "When specified, limits the output to the top K most probable classes."
 				}
 			}
 		}
 	},
-	"required": ["data"]
+	"required": ["inputs"]
 }

package/src/tasks/image-segmentation/inference.ts

@@ -10,7 +10,7 @@ export interface ImageSegmentationInput {
 	/**
 	 * The input image data
 	 */
-	data: unknown;
+	inputs: unknown;
 	/**
 	 * Additional inference parameters
 	 */
@@ -26,11 +26,11 @@ export interface ImageSegmentationParameters {
 	/**
 	 * Threshold to use when turning the predicted masks into binary values.
 	 */
-	maskThreshold?: number;
+	mask_threshold?: number;
 	/**
 	 * Mask overlap threshold to eliminate small, disconnected segments.
 	 */
-	overlapMaskAreaThreshold?: number;
+	overlap_mask_area_threshold?: number;
 	/**
 	 * Segmentation task to be performed, depending on model capabilities.
 	 */

package/src/tasks/image-segmentation/spec/input.json

@@ -5,7 +5,7 @@
 	"title": "ImageSegmentationInput",
 	"type": "object",
 	"properties": {
-		"data": {
+		"inputs": {
 			"description": "The input image data"
 		},
 		"parameters": {
@@ -19,11 +19,11 @@
 			"description": "Additional inference parameters for Image Segmentation",
 			"type": "object",
 			"properties": {
-				"maskThreshold": {
+				"mask_threshold": {
 					"type": "number",
 					"description": "Threshold to use when turning the predicted masks into binary values."
 				},
-				"overlapMaskAreaThreshold": {
+				"overlap_mask_area_threshold": {
 					"type": "number",
 					"description": "Mask overlap threshold to eliminate small, disconnected segments."
 				},
@@ -50,5 +50,5 @@
 			}
 		}
 	},
-	"required": ["data"]
+	"required": ["inputs"]
 }

package/src/tasks/image-to-image/inference.ts

@@ -11,7 +11,7 @@ export interface ImageToImageInput {
 	/**
 	 * The input image data
 	 */
-	data: unknown;
+	inputs: unknown;
 	/**
 	 * Additional inference parameters
 	 */
@@ -29,20 +29,20 @@ export interface ImageToImageParameters {
 	 * For diffusion models. A higher guidance scale value encourages the model to generate
 	 * images closely linked to the text prompt at the expense of lower image quality.
 	 */
-	guidanceScale?: number;
+	guidance_scale?: number;
 	/**
 	 * One or several prompt to guide what NOT to include in image generation.
 	 */
-	negativePrompt?: string[];
+	negative_prompt?: string[];
 	/**
 	 * For diffusion models. The number of denoising steps. More denoising steps usually lead to
 	 * a higher quality image at the expense of slower inference.
 	 */
-	numInferenceSteps?: number;
+	num_inference_steps?: number;
 	/**
 	 * The size in pixel of the output image
 	 */
-	targetSize?: TargetSize;
+	target_size?: TargetSize;
 	[property: string]: unknown;
 }
 

package/src/tasks/image-to-image/spec/input.json

@@ -5,7 +5,7 @@
 	"title": "ImageToImageInput",
 	"type": "object",
 	"properties": {
-		"data": {
+		"inputs": {
 			"description": "The input image data"
 		},
 		"parameters": {
@@ -19,20 +19,22 @@
 			"description": "Additional inference parameters for Image To Image",
 			"type": "object",
 			"properties": {
-				"guidanceScale": {
+				"guidance_scale": {
 					"type": "number",
 					"description": "For diffusion models. A higher guidance scale value encourages the model to generate images closely linked to the text prompt at the expense of lower image quality."
 				},
-				"negativePrompt": {
+				"negative_prompt": {
 					"type": "array",
-					"items": { "type": "string" },
+					"items": {
+						"type": "string"
+					},
 					"description": "One or several prompt to guide what NOT to include in image generation."
 				},
-				"numInferenceSteps": {
+				"num_inference_steps": {
 					"type": "integer",
 					"description": "For diffusion models. The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference."
 				},
-				"targetSize": {
+				"target_size": {
 					"type": "object",
 					"description": "The size in pixel of the output image",
 					"properties": {
@@ -48,5 +50,5 @@
 			}
 		}
 	},
-	"required": ["data"]
+	"required": ["inputs"]
 }

package/src/tasks/image-to-text/inference.ts

@@ -3,6 +3,7 @@
  *
  * Using src/scripts/inference-codegen
  */
+
 /**
  * Inputs for Image To Text inference
  */
@@ -10,13 +11,14 @@ export interface ImageToTextInput {
 	/**
 	 * The input image data
 	 */
-	data: unknown;
+	inputs: unknown;
 	/**
 	 * Additional inference parameters
 	 */
 	parameters?: ImageToTextParameters;
 	[property: string]: unknown;
 }
+
 /**
  * Additional inference parameters
  *
@@ -30,9 +32,10 @@ export interface ImageToTextParameters {
 	/**
 	 * The amount of maximum tokens to generate.
 	 */
-	maxNewTokens?: number;
+	max_new_tokens?: number;
 	[property: string]: unknown;
 }
+
 /**
  * Parametrization of the text generation process
  *
@@ -42,18 +45,18 @@ export interface GenerationParameters {
 	/**
 	 * Whether to use sampling instead of greedy decoding when generating new tokens.
 	 */
-	doSample?: boolean;
+	do_sample?: boolean;
 	/**
 	 * Controls the stopping condition for beam-based methods.
 	 */
-	earlyStopping?: EarlyStoppingUnion;
+	early_stopping?: EarlyStoppingUnion;
 	/**
 	 * If set to float strictly between 0 and 1, only tokens with a conditional probability
 	 * greater than epsilon_cutoff will be sampled. In the paper, suggested values range from
 	 * 3e-4 to 9e-4, depending on the size of the model. See [Truncation Sampling as Language
 	 * Model Desmoothing](https://hf.co/papers/2210.15191) for more details.
 	 */
-	epsilonCutoff?: number;
+	epsilon_cutoff?: number;
 	/**
 	 * Eta sampling is a hybrid of locally typical sampling and epsilon sampling. If set to
 	 * float strictly between 0 and 1, a token is only considered if it is greater than either
@@ -63,37 +66,37 @@ export interface GenerationParameters {
 	 * See [Truncation Sampling as Language Model Desmoothing](https://hf.co/papers/2210.15191)
 	 * for more details.
 	 */
-	etaCutoff?: number;
+	eta_cutoff?: number;
 	/**
 	 * The maximum length (in tokens) of the generated text, including the input.
 	 */
-	maxLength?: number;
+	max_length?: number;
 	/**
 	 * The maximum number of tokens to generate. Takes precedence over maxLength.
 	 */
-	maxNewTokens?: number;
+	max_new_tokens?: number;
 	/**
 	 * The minimum length (in tokens) of the generated text, including the input.
 	 */
-	minLength?: number;
+	min_length?: number;
 	/**
 	 * The minimum number of tokens to generate. Takes precedence over maxLength.
 	 */
-	minNewTokens?: number;
+	min_new_tokens?: number;
 	/**
 	 * Number of groups to divide num_beams into in order to ensure diversity among different
 	 * groups of beams. See [this paper](https://hf.co/papers/1610.02424) for more details.
 	 */
-	numBeamGroups?: number;
+	num_beam_groups?: number;
 	/**
 	 * Number of beams to use for beam search.
 	 */
-	numBeams?: number;
+	num_beams?: number;
 	/**
 	 * The value balances the model confidence and the degeneration penalty in contrastive
 	 * search decoding.
 	 */
-	penaltyAlpha?: number;
+	penalty_alpha?: number;
 	/**
 	 * The value used to modulate the next token probabilities.
 	 */
@@ -101,12 +104,12 @@ export interface GenerationParameters {
 	/**
 	 * The number of highest probability vocabulary tokens to keep for top-k-filtering.
 	 */
-	topK?: number;
+	top_k?: number;
 	/**
 	 * If set to float < 1, only the smallest set of most probable tokens with probabilities
 	 * that add up to top_p or higher are kept for generation.
 	 */
-	topP?: number;
+	top_p?: number;
 	/**
 	 * Local typicality measures how similar the conditional probability of predicting a target
 	 * token next is to the expected conditional probability of predicting a random token next,
@@ -114,25 +117,27 @@ export interface GenerationParameters {
 	 * most locally typical tokens with probabilities that add up to typical_p or higher are
 	 * kept for generation. See [this paper](https://hf.co/papers/2202.00666) for more details.
 	 */
-	typicalP?: number;
+	typical_p?: number;
 	/**
 	 * Whether the model should use the past last key/values attentions to speed up decoding
 	 */
-	useCache?: boolean;
+	use_cache?: boolean;
 	[property: string]: unknown;
 }
+
 /**
  * Controls the stopping condition for beam-based methods.
  */
 export type EarlyStoppingUnion = boolean | "never";
-export type ImageToTextOutput = ImageToTextOutputElement[];
+
 /**
  * Outputs of inference for the Image To Text task
  */
-export interface ImageToTextOutputElement {
+export interface ImageToTextOutput {
+	generatedText: unknown;
 	/**
 	 * The generated text.
 	 */
-	generatedText: string;
+	generated_text?: string;
 	[property: string]: unknown;
 }

package/src/tasks/image-to-text/spec/input.json

@@ -5,7 +5,7 @@
 	"title": "ImageToTextInput",
 	"type": "object",
 	"properties": {
-		"data": {
+		"inputs": {
 			"description": "The input image data"
 		},
 		"parameters": {
@@ -19,7 +19,7 @@
 			"description": "Additional inference parameters for Image To Text",
 			"type": "object",
 			"properties": {
-				"maxNewTokens": {
+				"max_new_tokens": {
 					"type": "integer",
 					"description": "The amount of maximum tokens to generate."
 				},
@@ -30,5 +30,5 @@
 			}
 		}
 	},
-	"required": ["data"]
+	"required": ["inputs"]
 }

package/src/tasks/image-to-text/spec/output.json

@@ -3,15 +3,12 @@
 	"$schema": "http://json-schema.org/draft-06/schema#",
 	"description": "Outputs of inference for the Image To Text task",
 	"title": "ImageToTextOutput",
-	"type": "array",
-	"items": {
-		"type": "object",
-		"properties": {
-			"generatedText": {
-				"type": "string",
-				"description": "The generated text."
-			}
-		},
-		"required": ["generatedText"]
-	}
+	"type": "object",
+	"properties": {
+		"generated_text": {
+			"type": "string",
+			"description": "The generated text."
+		}
+	},
+	"required": ["generatedText"]
 }

package/src/tasks/index.ts

@@ -53,6 +53,7 @@ export const TASKS_MODEL_LIBRARIES: Record<PipelineType, ModelLibraryKey[]> = {
 	"graph-ml": ["transformers"],
 	"image-classification": ["keras", "timm", "transformers", "transformers.js"],
 	"image-segmentation": ["transformers", "transformers.js"],
+	"image-text-to-text": ["transformers"],
 	"image-to-image": ["diffusers", "transformers", "transformers.js"],
 	"image-to-text": ["transformers", "transformers.js"],
 	"image-to-video": ["diffusers"],
@@ -130,6 +131,7 @@ export const TASKS_DATA: Record<PipelineType, TaskData | undefined> = {
 	"graph-ml": undefined,
 	"image-classification": getData("image-classification", imageClassification),
 	"image-segmentation": getData("image-segmentation", imageSegmentation),
+	"image-text-to-text": undefined,
 	"image-to-image": getData("image-to-image", imageToImage),
 	"image-to-text": getData("image-to-text", imageToText),
 	"image-to-video": undefined,

package/src/tasks/object-detection/inference.ts

@@ -10,7 +10,7 @@ export interface ObjectDetectionInput {
 	/**
 	 * The input image data
 	 */
-	data: unknown;
+	inputs: unknown;
 	/**
 	 * Additional inference parameters
 	 */

package/src/tasks/object-detection/spec/input.json

@@ -5,7 +5,7 @@
 	"title": "ObjectDetectionInput",
 	"type": "object",
 	"properties": {
-		"data": {
+		"inputs": {
 			"description": "The input image data"
 		},
 		"parameters": {
@@ -26,5 +26,5 @@
 			}
 		}
 	},
-	"required": ["data"]
+	"required": ["inputs"]
 }

package/src/tasks/placeholder/spec/input.json

@@ -5,7 +5,7 @@
 	"title": "PlaceholderInput",
 	"type": "object",
 	"properties": {
-		"data": {
+		"inputs": {
 			"description": "TODO: describe the input here. This must be model & framework agnostic.",
 			"type": "string"
 		},
@@ -20,16 +20,16 @@
 			"description": "TODO: describe additional parameters here.",
 			"type": "object",
 			"properties": {
-				"dummyParameterName": {
+				"dummy_parameter_name": {
 					"type": "boolean",
 					"description": "TODO: describe the parameter here"
 				},
-				"dummyParameterName2": {
+				"dummy_parameter_name2": {
 					"type": "integer",
 					"description": "TODO: describe the parameter here"
 				}
 			}
 		}
 	},
-	"required": ["data"]
+	"required": ["inputs"]
 }

package/src/tasks/placeholder/spec/output.json

@@ -7,7 +7,7 @@
 	"items": {
 		"type": "object",
 		"properties": {
-			"meaningfulOutputName": {
+			"meaningful_output_name": {
 				"type": "string",
 				"description": "TODO: Describe what is outputed by the inference here"
 			}

package/src/tasks/question-answering/inference.ts

@@ -10,7 +10,7 @@ export interface QuestionAnsweringInput {
 	/**
 	 * One (context, question) pair to answer
 	 */
-	data: QuestionAnsweringInputData;
+	inputs: QuestionAnsweringInputData;
 	/**
 	 * Additional inference parameters
 	 */
@@ -41,37 +41,37 @@ export interface QuestionAnsweringParameters {
 	 * Attempts to align the answer to real words. Improves quality on space separated
 	 * languages. Might hurt on non-space-separated languages (like Japanese or Chinese)
 	 */
-	alignToWords?: boolean;
+	align_to_words?: boolean;
 	/**
 	 * If the context is too long to fit with the question for the model, it will be split in
 	 * several chunks with some overlap. This argument controls the size of that overlap.
 	 */
-	docStride?: number;
+	doc_stride?: number;
 	/**
 	 * Whether to accept impossible as an answer.
 	 */
-	handleImpossibleAnswer?: boolean;
+	handle_impossible_answer?: boolean;
 	/**
 	 * The maximum length of predicted answers (e.g., only answers with a shorter length are
 	 * considered).
 	 */
-	maxAnswerLen?: number;
+	max_answer_len?: number;
 	/**
 	 * The maximum length of the question after tokenization. It will be truncated if needed.
 	 */
-	maxQuestionLen?: number;
+	max_question_len?: number;
 	/**
 	 * The maximum length of the total sentence (context + question) in tokens of each chunk
 	 * passed to the model. The context will be split in several chunks (using docStride as
 	 * overlap) if needed.
 	 */
-	maxSeqLen?: number;
+	max_seq_len?: number;
 	/**
 	 * The number of answers to return (will be chosen by order of likelihood). Note that we
 	 * return less than topk answers if there are not enough options available within the
 	 * context.
 	 */
-	topK?: number;
+	top_k?: number;
 	[property: string]: unknown;
 }
 export type QuestionAnsweringOutput = QuestionAnsweringOutputElement[];

package/src/tasks/question-answering/spec/input.json

@@ -5,7 +5,7 @@
 	"title": "QuestionAnsweringInput",
 	"type": "object",
 	"properties": {
-		"data": {
+		"inputs": {
 			"title": "QuestionAnsweringInputData",
 			"description": "One (context, question) pair to answer",
 			"type": "object",
@@ -32,36 +32,36 @@
 			"description": "Additional inference parameters for Question Answering",
 			"type": "object",
 			"properties": {
-				"topK": {
+				"top_k": {
 					"type": "integer",
 					"description": "The number of answers to return (will be chosen by order of likelihood). Note that we return less than topk answers if there are not enough options available within the context."
 				},
-				"docStride": {
+				"doc_stride": {
 					"type": "integer",
 					"description": "If the context is too long to fit with the question for the model, it will be split in several chunks with some overlap. This argument controls the size of that overlap."
 				},
-				"maxAnswerLen": {
+				"max_answer_len": {
 					"type": "integer",
 					"description": "The maximum length of predicted answers (e.g., only answers with a shorter length are considered)."
 				},
-				"maxSeqLen": {
+				"max_seq_len": {
 					"type": "integer",
 					"description": "The maximum length of the total sentence (context + question) in tokens of each chunk passed to the model. The context will be split in several chunks (using docStride as overlap) if needed."
 				},
-				"maxQuestionLen": {
+				"max_question_len": {
 					"type": "integer",
 					"description": "The maximum length of the question after tokenization. It will be truncated if needed."
 				},
-				"handleImpossibleAnswer": {
+				"handle_impossible_answer": {
 					"type": "boolean",
 					"description": "Whether to accept impossible as an answer."
 				},
-				"alignToWords": {
+				"align_to_words": {
 					"type": "boolean",
 					"description": "Attempts to align the answer to real words. Improves quality on space separated languages. Might hurt on non-space-separated languages (like Japanese or Chinese)"
 				}
 			}
 		}
 	},
-	"required": ["data"]
+	"required": ["inputs"]
 }

package/src/tasks/sentence-similarity/inference.ts

@@ -10,7 +10,7 @@ export type SentenceSimilarityOutput = number[];
  * Inputs for Sentence similarity inference
  */
 export interface SentenceSimilarityInput {
-	data: SentenceSimilarityInputData;
+	inputs: SentenceSimilarityInputData;
 	/**
 	 * Additional inference parameters
 	 */

package/src/tasks/sentence-similarity/spec/input.json

@@ -5,7 +5,7 @@
 	"title": "SentenceSimilarityInput",
 	"type": "object",
 	"properties": {
-		"data": {
+		"inputs": {
 			"title": "SentenceSimilarityInputData",
 			"type": "object",
 			"properties": {
@@ -36,5 +36,5 @@
 			"properties": {}
 		}
 	},
-	"required": ["data"]
+	"required": ["inputs"]
 }

package/src/tasks/summarization/inference.ts

@@ -13,7 +13,7 @@ export interface SummarizationInput {
 	/**
 	 * The input text data
 	 */
-	data: string;
+	inputs: string;
 	/**
 	 * Additional inference parameters
 	 */
@@ -30,11 +30,11 @@ export interface Text2TextGenerationParameters {
 	/**
 	 * Whether to clean up the potential extra spaces in the text output.
 	 */
-	cleanUpTokenizationSpaces?: boolean;
+	clean_up_tokenization_spaces?: boolean;
 	/**
 	 * Additional parametrization of the text generation algorithm
 	 */
-	generateParameters?: { [key: string]: unknown };
+	generate_parameters?: { [key: string]: unknown };
 	/**
 	 * The truncation strategy to use
 	 */
@@ -50,9 +50,10 @@ export type Text2TextGenerationTruncationStrategy = "do_not_truncate" | "longest
  * Outputs of inference for the Text2text Generation task
  */
 export interface SummarizationOutput {
+	generatedText: unknown;
 	/**
 	 * The generated text.
 	 */
-	generatedText: string;
+	generated_text?: string;
 	[property: string]: unknown;
 }

package/src/tasks/table-question-answering/inference.ts

@@ -10,7 +10,7 @@ export interface TableQuestionAnsweringInput {
 	/**
 	 * One (table, question) pair to answer
 	 */
-	data: TableQuestionAnsweringInputData;
+	inputs: TableQuestionAnsweringInputData;
 	/**
 	 * Additional inference parameters
 	 */