@huggingface/tasks 0.13.17 → 0.14.0

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

@@ -3,7 +3,6 @@
  *
  * Using src/scripts/inference-codegen
  */
-
 /**
  * Inputs for Image To Image inference
  */
@@ -12,14 +11,13 @@ export interface ImageToImageInput {
 	 * The input image data as a base64-encoded string. If no `parameters` are provided, you can
 	 * also provide the image data as a raw bytes payload.
 	 */
-	inputs: string;
+	inputs: Blob;
 	/**
 	 * Additional inference parameters for Image To Image
 	 */
 	parameters?: ImageToImageParameters;
 	[property: string]: unknown;
 }
-
 /**
  * Additional inference parameters for Image To Image
  */
@@ -30,9 +28,9 @@ export interface ImageToImageParameters {
 	 */
 	guidance_scale?: number;
 	/**
-	 * One or several prompt to guide what NOT to include in image generation.
+	 * One prompt to guide what NOT to include in image generation.
 	 */
-	negative_prompt?: 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.
@@ -44,7 +42,6 @@ export interface ImageToImageParameters {
 	target_size?: TargetSize;
 	[property: string]: unknown;
 }
-
 /**
  * The size in pixel of the output image.
  */
@@ -53,7 +50,6 @@ export interface TargetSize {
 	width: number;
 	[property: string]: unknown;
 }
-
 /**
  * Outputs of inference for the Image To Image task
  */

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

@@ -7,7 +7,8 @@
 	"properties": {
 		"inputs": {
 			"type": "string",
-			"description": "The input image data as a base64-encoded string. If no `parameters` are provided, you can also provide the image data as a raw bytes payload."
+			"description": "The input image data as a base64-encoded string. If no `parameters` are provided, you can also provide the image data as a raw bytes payload.",
+			"comment": "type=binary"
 		},
 		"parameters": {
 			"description": "Additional inference parameters for Image To Image",
@@ -24,11 +25,8 @@
 					"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."
 				},
 				"negative_prompt": {
-					"type": "array",
-					"items": {
-						"type": "string"
-					},
-					"description": "One or several prompt to guide what NOT to include in image generation."
+					"type": "string",
+					"description": "One prompt to guide what NOT to include in image generation."
 				},
 				"num_inference_steps": {
 					"type": "integer",

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

@@ -3,7 +3,6 @@
  *
  * Using src/scripts/inference-codegen
  */
-
 /**
  * Inputs for Image To Text inference
  */
@@ -11,14 +10,13 @@ export interface ImageToTextInput {
 	/**
 	 * The input image data
 	 */
-	inputs: unknown;
+	inputs: Blob;
 	/**
 	 * Additional inference parameters for Image To Text
 	 */
 	parameters?: ImageToTextParameters;
 	[property: string]: unknown;
 }
-
 /**
  * Additional inference parameters for Image To Text
  */
@@ -33,7 +31,6 @@ export interface ImageToTextParameters {
 	max_new_tokens?: number;
 	[property: string]: unknown;
 }
-
 /**
  * Parametrization of the text generation process
  */
@@ -120,12 +117,10 @@ export interface GenerationParameters {
 	use_cache?: boolean;
 	[property: string]: unknown;
 }
-
 /**
  * Controls the stopping condition for beam-based methods.
  */
 export type EarlyStoppingUnion = boolean | "never";
-
 /**
  * Outputs of inference for the Image To Text task
  */

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

@@ -6,7 +6,8 @@
 	"type": "object",
 	"properties": {
 		"inputs": {
-			"description": "The input image data"
+			"description": "The input image data",
+			"comment": "type=binary"
 		},
 		"parameters": {
 			"description": "Additional inference parameters for Image To Text",

package/src/tasks/index.ts

@@ -73,6 +73,7 @@ export type * from "./sentence-similarity/inference.js";
 export type * from "./summarization/inference.js";
 export type * from "./table-question-answering/inference.js";
 export type { TextToImageInput, TextToImageOutput, TextToImageParameters } from "./text-to-image/inference.js";
+export type { TextToVideoParameters, TextToVideoOutput, TextToVideoInput } from "./text-to-video/inference.js";
 export type { TextToSpeechParameters, TextToSpeechInput, TextToSpeechOutput } from "./text-to-speech/inference.js";
 export type * from "./token-classification/inference.js";
 export type { TranslationInput, TranslationOutput } from "./translation/inference.js";

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

@@ -11,7 +11,7 @@ export interface ObjectDetectionInput {
 	 * The input image data as a base64-encoded string. If no `parameters` are provided, you can
 	 * also provide the image data as a raw bytes payload.
 	 */
-	inputs: string;
+	inputs: Blob;
 	/**
 	 * Additional inference parameters for Object Detection
 	 */

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

@@ -7,7 +7,8 @@
 	"properties": {
 		"inputs": {
 			"type": "string",
-			"description": "The input image data as a base64-encoded string. If no `parameters` are provided, you can also provide the image data as a raw bytes payload."
+			"description": "The input image data as a base64-encoded string. If no `parameters` are provided, you can also provide the image data as a raw bytes payload.",
+			"comment": "type=binary"
 		},
 		"parameters": {
 			"description": "Additional inference parameters for Object Detection",

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

@@ -3,9 +3,7 @@
  *
  * Using src/scripts/inference-codegen
  */
-
 export type SentenceSimilarityOutput = number[];
-
 /**
  * Inputs for Sentence similarity inference
  */
@@ -14,10 +12,11 @@ export interface SentenceSimilarityInput {
 	/**
 	 * Additional inference parameters for Sentence Similarity
 	 */
-	parameters?: { [key: string]: unknown };
+	parameters?: {
+		[key: string]: unknown;
+	};
 	[property: string]: unknown;
 }
-
 export interface SentenceSimilarityInputData {
 	/**
 	 * A list of strings which will be compared against the source_sentence.

package/src/tasks/summarization/inference.ts

@@ -3,7 +3,6 @@
  *
  * Using src/scripts/inference-codegen
  */
-
 /**
  * Inputs for Summarization inference
  */
@@ -18,7 +17,6 @@ export interface SummarizationInput {
 	parameters?: SummarizationParameters;
 	[property: string]: unknown;
 }
-
 /**
  * Additional inference parameters for summarization.
  */
@@ -30,19 +28,19 @@ export interface SummarizationParameters {
 	/**
 	 * Additional parametrization of the text generation algorithm.
 	 */
-	generate_parameters?: { [key: string]: unknown };
+	generate_parameters?: {
+		[key: string]: unknown;
+	};
 	/**
 	 * The truncation strategy to use.
 	 */
 	truncation?: SummarizationTruncationStrategy;
 	[property: string]: unknown;
 }
-
 /**
  * The truncation strategy to use.
  */
 export type SummarizationTruncationStrategy = "do_not_truncate" | "longest_first" | "only_first" | "only_second";
-
 /**
  * Outputs of inference for the Summarization task
  */

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

@@ -3,7 +3,6 @@
  *
  * Using src/scripts/inference-codegen
  */
-
 /**
  * Text Generation Input.
  *
@@ -17,7 +16,6 @@ export interface TextGenerationInput {
 	stream?: boolean;
 	[property: string]: unknown;
 }
-
 export interface TextGenerationInputGenerateParameters {
 	/**
 	 * Lora adapter id
@@ -100,7 +98,6 @@ export interface TextGenerationInputGenerateParameters {
 	watermark?: boolean;
 	[property: string]: unknown;
 }
-
 export interface TextGenerationInputGrammarType {
 	type: Type;
 	/**
@@ -112,9 +109,7 @@ export interface TextGenerationInputGrammarType {
 	value: unknown;
 	[property: string]: unknown;
 }
-
 export type Type = "json" | "regex";
-
 /**
  * Text Generation Output.
  *
@@ -127,7 +122,6 @@ export interface TextGenerationOutput {
 	generated_text: string;
 	[property: string]: unknown;
 }
-
 export interface TextGenerationOutputDetails {
 	best_of_sequences?: TextGenerationOutputBestOfSequence[];
 	finish_reason: TextGenerationOutputFinishReason;
@@ -138,7 +132,6 @@ export interface TextGenerationOutputDetails {
 	top_tokens?: Array<TextGenerationOutputToken[]>;
 	[property: string]: unknown;
 }
-
 export interface TextGenerationOutputBestOfSequence {
 	finish_reason: TextGenerationOutputFinishReason;
 	generated_text: string;
@@ -149,16 +142,13 @@ export interface TextGenerationOutputBestOfSequence {
 	top_tokens?: Array<TextGenerationOutputToken[]>;
 	[property: string]: unknown;
 }
-
 export type TextGenerationOutputFinishReason = "length" | "eos_token" | "stop_sequence";
-
 export interface TextGenerationOutputPrefillToken {
 	id: number;
 	logprob: number;
 	text: string;
 	[property: string]: unknown;
 }
-
 export interface TextGenerationOutputToken {
 	id: number;
 	logprob: number;
@@ -166,7 +156,6 @@ export interface TextGenerationOutputToken {
 	text: string;
 	[property: string]: unknown;
 }
-
 /**
  * Text Generation Stream Output.
  *
@@ -182,7 +171,6 @@ export interface TextGenerationStreamOutput {
 	top_tokens?: TextGenerationStreamOutputToken[];
 	[property: string]: unknown;
 }
-
 export interface TextGenerationStreamOutputStreamDetails {
 	finish_reason: TextGenerationOutputFinishReason;
 	generated_tokens: number;
@@ -190,7 +178,6 @@ export interface TextGenerationStreamOutputStreamDetails {
 	seed?: number;
 	[property: string]: unknown;
 }
-
 export interface TextGenerationStreamOutputToken {
 	id: number;
 	logprob: number;

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

@@ -1,9 +1,22 @@
+/**
+ * Outputs of inference for the Text To Audio task
+ */
+export interface TextToAudioOutput {
+	/**
+	 * The generated audio waveform.
+	 */
+	audio: Blob;
+	/**
+	 * The sampling rate of the generated audio waveform.
+	 */
+	sampling_rate: number;
+	[property: string]: unknown;
+}
 /**
  * Inference code generated from the JSON schema spec in ./spec
  *
  * Using src/scripts/inference-codegen
  */
-
 /**
  * Inputs for Text To Audio inference
  */
@@ -18,7 +31,6 @@ export interface TextToAudioInput {
 	parameters?: TextToAudioParameters;
 	[property: string]: unknown;
 }
-
 /**
  * Additional inference parameters for Text To Audio
  */
@@ -29,7 +41,6 @@ export interface TextToAudioParameters {
 	generation_parameters?: GenerationParameters;
 	[property: string]: unknown;
 }
-
 /**
  * Parametrization of the text generation process
  */
@@ -116,24 +127,7 @@ export interface GenerationParameters {
 	use_cache?: boolean;
 	[property: string]: unknown;
 }
-
 /**
  * Controls the stopping condition for beam-based methods.
  */
 export type EarlyStoppingUnion = boolean | "never";
-
-/**
- * Outputs of inference for the Text To Audio task
- */
-export interface TextToAudioOutput {
-	/**
-	 * The generated audio waveform.
-	 */
-	audio: unknown;
-	samplingRate: unknown;
-	/**
-	 * The sampling rate of the generated audio waveform.
-	 */
-	sampling_rate?: number;
-	[property: string]: unknown;
-}

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

@@ -6,12 +6,13 @@
 	"type": "object",
 	"properties": {
 		"audio": {
-			"description": "The generated audio waveform."
+			"description": "The generated audio waveform.",
+			"comment": "type=binary"
 		},
 		"sampling_rate": {
 			"type": "number",
 			"description": "The sampling rate of the generated audio waveform."
 		}
 	},
-	"required": ["audio", "samplingRate"]
+	"required": ["audio", "sampling_rate"]
 }

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

@@ -3,7 +3,6 @@
  *
  * Using src/scripts/inference-codegen
  */
-
 /**
  * Inputs for Text To Image inference
  */
@@ -18,7 +17,6 @@ export interface TextToImageInput {
 	parameters?: TextToImageParameters;
 	[property: string]: unknown;
 }
-
 /**
  * Additional inference parameters for Text To Image
  */
@@ -29,9 +27,9 @@ export interface TextToImageParameters {
 	 */
 	guidance_scale?: number;
 	/**
-	 * One or several prompt to guide what NOT to include in image generation.
+	 * One prompt to guide what NOT to include in image generation.
 	 */
-	negative_prompt?: string[];
+	negative_prompt?: string;
 	/**
 	 * The number of denoising steps. More denoising steps usually lead to a higher quality
 	 * image at the expense of slower inference.
@@ -51,7 +49,6 @@ export interface TextToImageParameters {
 	target_size?: TargetSize;
 	[property: string]: unknown;
 }
-
 /**
  * The size in pixel of the output image
  */
@@ -60,7 +57,6 @@ export interface TargetSize {
 	width: number;
 	[property: string]: unknown;
 }
-
 /**
  * Outputs of inference for the Text To Image task
  */

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

@@ -24,11 +24,8 @@
 					"description": "A higher guidance scale value encourages the model to generate images closely linked to the text prompt, but values too high may cause saturation and other artifacts."
 				},
 				"negative_prompt": {
-					"type": "array",
-					"items": {
-						"type": "string"
-					},
-					"description": "One or several prompt to guide what NOT to include in image generation."
+					"type": "string",
+					"description": "One prompt to guide what NOT to include in image generation."
 				},
 				"num_inference_steps": {
 					"type": "integer",

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

@@ -1,9 +1,22 @@
+/**
+ * Outputs of inference for the Text To Speech task
+ */
+export interface TextToSpeechOutput {
+	/**
+	 * The generated audio
+	 */
+	audio: Blob;
+	/**
+	 * The sampling rate of the generated audio waveform.
+	 */
+	sampling_rate?: number;
+	[property: string]: unknown;
+}
 /**
  * Inference code generated from the JSON schema spec in ./spec
  *
  * Using src/scripts/inference-codegen
  */
-
 /**
  * Inputs for Text To Speech inference
  */
@@ -18,7 +31,6 @@ export interface TextToSpeechInput {
 	parameters?: TextToSpeechParameters;
 	[property: string]: unknown;
 }
-
 /**
  * Additional inference parameters for Text To Speech
  */
@@ -29,7 +41,6 @@ export interface TextToSpeechParameters {
 	generation_parameters?: GenerationParameters;
 	[property: string]: unknown;
 }
-
 /**
  * Parametrization of the text generation process
  */
@@ -116,26 +127,7 @@ export interface GenerationParameters {
 	use_cache?: boolean;
 	[property: string]: unknown;
 }
-
 /**
  * Controls the stopping condition for beam-based methods.
  */
 export type EarlyStoppingUnion = boolean | "never";
-
-/**
- * Outputs for Text to Speech inference
- *
- * Outputs of inference for the Text To Audio task
- */
-export interface TextToSpeechOutput {
-	/**
-	 * The generated audio waveform.
-	 */
-	audio: unknown;
-	samplingRate: unknown;
-	/**
-	 * The sampling rate of the generated audio waveform.
-	 */
-	sampling_rate?: number;
-	[property: string]: unknown;
-}

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

@@ -1,7 +1,18 @@
 {
-	"$ref": "/inference/schemas/text-to-audio/output.json",
 	"$id": "/inference/schemas/text-to-speech/output.json",
 	"$schema": "http://json-schema.org/draft-06/schema#",
+	"description": "Outputs of inference for the Text To Speech task",
 	"title": "TextToSpeechOutput",
-	"description": "Outputs for Text to Speech inference"
+	"type": "object",
+	"properties": {
+		"audio": {
+			"description": "The generated audio",
+			"comment": "type=binary"
+		},
+		"sampling_rate": {
+			"type": "number",
+			"description": "The sampling rate of the generated audio waveform."
+		}
+	},
+	"required": ["audio"]
 }

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

@@ -0,0 +1,57 @@
+/**
+ * Inference code generated from the JSON schema spec in ./spec
+ *
+ * Using src/scripts/inference-codegen
+ */
+/**
+ * Inputs for Text To Video inference
+ */
+export interface TextToVideoInput {
+	/**
+	 * The input text data (sometimes called "prompt")
+	 */
+	inputs: string;
+	/**
+	 * Additional inference parameters for Text To Video
+	 */
+	parameters?: TextToVideoParameters;
+	[property: string]: unknown;
+}
+/**
+ * Additional inference parameters for Text To Video
+ */
+export interface TextToVideoParameters {
+	/**
+	 * A higher guidance scale value encourages the model to generate images closely linked to
+	 * the text prompt, but values too high may cause saturation and other artifacts.
+	 */
+	guidance_scale?: number;
+	/**
+	 * One or several prompt to guide what NOT to include in image generation.
+	 */
+	negative_prompt?: string[];
+	/**
+	 * The num_frames parameter determines how many video frames are generated.
+	 */
+	num_frames?: number;
+	/**
+	 * The number of denoising steps. More denoising steps usually lead to a higher quality
+	 * image at the expense of slower inference.
+	 */
+	num_inference_steps?: number;
+	/**
+	 * Seed for the random number generator.
+	 */
+	seed?: number;
+	[property: string]: unknown;
+}
+/**
+ * Outputs of inference for the Text To Video task
+ */
+export interface TextToVideoOutput {
+	/**
+	 * The generated video returned as raw bytes in the payload.
+	 */
+	video: unknown;
+	[property: string]: unknown;
+}

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

@@ -0,0 +1,49 @@
+{
+	"$id": "/inference/schemas/text-to-video/input.json",
+	"$schema": "http://json-schema.org/draft-06/schema#",
+	"description": "Inputs for Text To Video inference",
+	"title": "TextToVideoInput",
+	"type": "object",
+	"properties": {
+		"inputs": {
+			"description": "The input text data (sometimes called \"prompt\")",
+			"type": "string"
+		},
+		"parameters": {
+			"description": "Additional inference parameters for Text To Video",
+			"$ref": "#/$defs/TextToVideoParameters"
+		}
+	},
+	"$defs": {
+		"TextToVideoParameters": {
+			"title": "TextToVideoParameters",
+			"type": "object",
+			"properties": {
+				"num_frames": {
+					"type": "number",
+					"description": "The num_frames parameter determines how many video frames are generated."
+				},
+				"guidance_scale": {
+					"type": "number",
+					"description": "A higher guidance scale value encourages the model to generate images closely linked to the text prompt, but values too high may cause saturation and other artifacts."
+				},
+				"negative_prompt": {
+					"type": "array",
+					"items": {
+						"type": "string"
+					},
+					"description": "One or several prompt to guide what NOT to include in image generation."
+				},
+				"num_inference_steps": {
+					"type": "integer",
+					"description": "The number of denoising steps. More denoising steps usually lead to a higher quality image at the expense of slower inference."
+				},
+				"seed": {
+					"type": "integer",
+					"description": "Seed for the random number generator."
+				}
+			}
+		}
+	},
+	"required": ["inputs"]
+}

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

@@ -0,0 +1,13 @@
+{
+	"$id": "/inference/schemas/text-to-video/output.json",
+	"$schema": "http://json-schema.org/draft-06/schema#",
+	"description": "Outputs of inference for the Text To Video task",
+	"title": "TextToVideoOutput",
+	"type": "object",
+	"properties": {
+		"video": {
+			"description": "The generated video returned as raw bytes in the payload."
+		}
+	},
+	"required": ["video"]
+}