apply reviewer feedback (keep it as spec definition only and add snippet to tasks snippets)

Author: apolinario

packages/inference/src/lib/getProviderHelper.ts

@@ -24,7 +24,6 @@ import type {
 	ImageSegmentationTaskHelper,
 	ImageToImageTaskHelper,
 	ImageToTextTaskHelper,
-	ImageToVideoTaskHelper,
 	ObjectDetectionTaskHelper,
 	QuestionAnsweringTaskHelper,
 	SentenceSimilarityTaskHelper,
@@ -220,10 +219,6 @@ export function getProviderHelper(
 	provider: InferenceProviderOrPolicy,
 	task: "image-to-text"
 ): ImageToTextTaskHelper & TaskProviderHelper;
-export function getProviderHelper(
-	provider: InferenceProviderOrPolicy,
-	task: "image-to-video"
-): ImageToVideoTaskHelper & TaskProviderHelper;
 export function getProviderHelper(
 	provider: InferenceProviderOrPolicy,
 	task: "object-detection"

packages/inference/src/providers/providerHelper.ts

@@ -34,7 +34,6 @@ import type {
 	TextToImageInput,
 	TextToSpeechInput,
 	TextToVideoInput,
-	ImageToVideoInput,
 	TokenClassificationInput,
 	TokenClassificationOutput,
 	TranslationInput,
@@ -148,11 +147,6 @@ export interface ImageToImageTaskHelper {
 	preparePayloadAsync(args: ImageToImageArgs): Promise<RequestArgs>;
 }
 
-export interface ImageToVideoTaskHelper {
-	getResponse(response: unknown, url?: string, headers?: Record<string, string>): Promise<Blob>;
-	preparePayload(params: BodyParams<ImageToVideoInput & BaseArgs>): Record<string, unknown>;
-}
-
 export interface ImageSegmentationTaskHelper {
 	getResponse(response: unknown, url?: string, headers?: HeadersInit): Promise<ImageSegmentationOutput>;
 	preparePayload(params: BodyParams<ImageSegmentationInput & BaseArgs>): Record<string, unknown> | BodyInit;

packages/inference/src/snippets/getInferenceSnippets.ts

@@ -83,7 +83,6 @@ const HF_PYTHON_METHODS: Partial<Record<WidgetType, string>> = {
 	"image-segmentation": "image_segmentation",
 	"image-to-image": "image_to_image",
 	"image-to-text": "image_to_text",
-	"image-to-video": "image_to_video",
 	"object-detection": "object_detection",
 	"question-answering": "question_answering",
 	"sentence-similarity": "sentence_similarity",
@@ -110,7 +109,6 @@ const HF_JS_METHODS: Partial<Record<WidgetType, string>> = {
 	"feature-extraction": "featureExtraction",
 	"fill-mask": "fillMask",
 	"image-classification": "imageClassification",
-	"image-to-video": "imageToVideo",
 	"question-answering": "questionAnswering",
 	"sentence-similarity": "sentenceSimilarity",
 	summarization: "summarization",
@@ -268,11 +266,6 @@ const prepareImageToImageInput = (model: ModelDataMinimal): object => {
 	return { inputs: data.image, parameters: { prompt: data.prompt } };
 };
 
-const prepareImageToVideoInput = (model: ModelDataMinimal): object => {
-	const data = JSON.parse(getModelInputSnippet(model) as string);
-	return { inputs: data.image, parameters: { prompt: data.prompt } };
-};
-
 const prepareConversationalInput = (
 	model: ModelDataMinimal,
 	opts?: {
@@ -323,7 +316,6 @@ const snippets: Partial<
 	"image-segmentation": snippetGenerator("basicImage"),
 	"image-text-to-text": snippetGenerator("conversational"),
 	"image-to-image": snippetGenerator("imageToImage", prepareImageToImageInput),
-	"image-to-video": snippetGenerator("imageToVideo", prepareImageToVideoInput),
 	"image-to-text": snippetGenerator("basicImage"),
 	"object-detection": snippetGenerator("basicImage"),
 	"question-answering": snippetGenerator("questionAnswering", prepareQuestionAnsweringInput),

packages/inference/src/tasks/cv/imageToVideo.ts

@@ -16,7 +16,6 @@ export * from "./cv/imageToText";
 export * from "./cv/objectDetection";
 export * from "./cv/textToImage";
 export * from "./cv/textToVideo";
-export * from "./cv/imageToVideo";
 export * from "./cv/zeroShotImageClassification";
 
 // Natural Language Processing tasks

packages/inference/src/tasks/index.ts

@@ -91,6 +91,11 @@ const inputsImageToImage = () => `{
     "prompt": "Turn the cat into a tiger."
 }`;
 
+const inputsImageToVideo = () => `{
+    "image": "cat.png",
+    "prompt": "The cat starts to dance"
+}`;
+
 const inputsImageSegmentation = () => `"cats.jpg"`;
 
 const inputsObjectDetection = () => `"cats.jpg"`;
@@ -126,6 +131,7 @@ const modelInputSnippets: {
 	"image-classification": inputsImageClassification,
 	"image-to-text": inputsImageToText,
 	"image-to-image": inputsImageToImage,
+	"image-to-video": inputsImageToVideo,
 	"image-segmentation": inputsImageSegmentation,
 	"object-detection": inputsObjectDetection,
 	"question-answering": inputsQuestionAnswering,