@huggingface/tasks 0.10.20 → 0.10.22

package/src/model-libraries.ts

@@ -90,6 +90,7 @@ export const MODEL_LIBRARIES_UI_ELEMENTS = {
 		repoUrl: "https://github.com/facebookresearch/audiocraft",
 		snippets: snippets.audiocraft,
 		filter: false,
+		countDownloads: `path:"state_dict.bin"`,
 	},
 	audioseal: {
 		prettyLabel: "AudioSeal",
@@ -241,6 +242,16 @@ export const MODEL_LIBRARIES_UI_ELEMENTS = {
 		docsUrl: "https://huggingface.co/docs/hub/keras",
 		snippets: snippets.keras,
 		filter: true,
+		countDownloads: `path:"config.json" OR path_extension:"keras"`,
+	},
+	"tf-keras": {
+		// Legacy "Keras 2" library (tensorflow-only)
+		prettyLabel: "TF-Keras",
+		repoName: "TF-Keras",
+		repoUrl: "https://github.com/keras-team/tf-keras",
+		docsUrl: "https://huggingface.co/docs/hub/tf-keras",
+		snippets: snippets.tf_keras,
+		filter: true,
 		countDownloads: `path:"saved_model.pb"`,
 	},
 	"keras-nlp": {
@@ -260,6 +271,14 @@ export const MODEL_LIBRARIES_UI_ELEMENTS = {
 		repoName: "mindspore",
 		repoUrl: "https://github.com/mindspore-ai/mindspore",
 	},
+	"mars5-tts": {
+		prettyLabel: "MARS5-TTS",
+		repoName: "MARS5-TTS",
+		repoUrl: "https://github.com/Camb-ai/MARS5-TTS",
+		filter: false,
+		countDownloads: `path:"mars5_ar.safetensors"`,
+		snippets: snippets.mars5_tts,
+	},
 	"ml-agents": {
 		prettyLabel: "ml-agents",
 		repoName: "ml-agents",
@@ -495,6 +514,8 @@ export const MODEL_LIBRARIES_UI_ELEMENTS = {
 		prettyLabel: "WhisperKit",
 		repoName: "WhisperKit",
 		repoUrl: "https://github.com/argmaxinc/WhisperKit",
+		docsUrl: "https://github.com/argmaxinc/WhisperKit?tab=readme-ov-file#homebrew",
+		snippets: snippets.whisperkit,
 		countDownloads: `path_filename:"model" AND path_extension:"mil" AND _exists_:"path_prefix"`,
 	},
 } satisfies Record<string, LibraryUiElement>;

package/src/snippets/curl.ts

@@ -10,6 +10,24 @@ export const snippetBasic = (model: ModelDataMinimal, accessToken: string): stri
 	-H "Authorization: Bearer ${accessToken || `{API_TOKEN}`}"
 `;
 
+export const snippetTextGeneration = (model: ModelDataMinimal, accessToken: string): string => {
+	if (model.config?.tokenizer_config?.chat_template) {
+		// Conversational model detected, so we display a code snippet that features the Messages API
+		return `curl 'https://api-inference.huggingface.co/models/${model.id}/v1/chat/completions' \\
+-H "Authorization: Bearer ${accessToken || `{API_TOKEN}`}" \\
+-H 'Content-Type: application/json' \\
+-d '{
+	"model": "${model.id}",
+	"messages": [{"role": "user", "content": "What is the capital of France?"}],
+	"max_tokens": 500,
+	"stream": false
+}'
+`;
+	} else {
+		return snippetBasic(model, accessToken);
+	}
+};
+
 export const snippetZeroShotClassification = (model: ModelDataMinimal, accessToken: string): string =>
 	`curl https://api-inference.huggingface.co/models/${model.id} \\
 	-X POST \\
@@ -35,7 +53,7 @@ export const curlSnippets: Partial<Record<PipelineType, (model: ModelDataMinimal
 	translation: snippetBasic,
 	summarization: snippetBasic,
 	"feature-extraction": snippetBasic,
-	"text-generation": snippetBasic,
+	"text-generation": snippetTextGeneration,
 	"text2text-generation": snippetBasic,
 	"fill-mask": snippetBasic,
 	"sentence-similarity": snippetBasic,

package/src/snippets/inputs.ts

@@ -11,30 +11,30 @@ const inputsSummarization = () =>
 
 const inputsTableQuestionAnswering = () =>
 	`{
-		"query": "How many stars does the transformers repository have?",
-		"table": {
-			"Repository": ["Transformers", "Datasets", "Tokenizers"],
-			"Stars": ["36542", "4512", "3934"],
-			"Contributors": ["651", "77", "34"],
-			"Programming language": [
-				"Python",
-				"Python",
-				"Rust, Python and NodeJS"
-			]
-		}
-	}`;
+	"query": "How many stars does the transformers repository have?",
+	"table": {
+		"Repository": ["Transformers", "Datasets", "Tokenizers"],
+		"Stars": ["36542", "4512", "3934"],
+		"Contributors": ["651", "77", "34"],
+		"Programming language": [
+			"Python",
+			"Python",
+			"Rust, Python and NodeJS"
+		]
+	}
+}`;
 
 const inputsVisualQuestionAnswering = () =>
 	`{
-		"image": "cat.png",
-		"question": "What is in this image?"
-	}`;
+	"image": "cat.png",
+	"question": "What is in this image?"
+}`;
 
 const inputsQuestionAnswering = () =>
 	`{
-		"question": "What is my name?",
-		"context": "My name is Clara and I live in Berkeley."
-	}`;
+	"question": "What is my name?",
+	"context": "My name is Clara and I live in Berkeley."
+}`;
 
 const inputsTextClassification = () => `"I like you. I love you"`;
 
@@ -48,13 +48,13 @@ const inputsFillMask = (model: ModelDataMinimal) => `"The answer to the universe
 
 const inputsSentenceSimilarity = () =>
 	`{
-		"source_sentence": "That is a happy person",
-		"sentences": [
-			"That is a happy dog",
-			"That is a very happy person",
-			"Today is a sunny day"
-		]
-	}`;
+	"source_sentence": "That is a happy person",
+	"sentences": [
+		"That is a happy dog",
+		"That is a very happy person",
+		"Today is a sunny day"
+	]
+}`;
 
 const inputsFeatureExtraction = () => `"Today is a sunny day and I will get some ice cream."`;
 

package/src/snippets/js.ts

@@ -7,7 +7,10 @@ export const snippetBasic = (model: ModelDataMinimal, accessToken: string): stri
 	const response = await fetch(
 		"https://api-inference.huggingface.co/models/${model.id}",
 		{
-			headers: { Authorization: "Bearer ${accessToken || `{API_TOKEN}`}" },
+			headers: {
+				Authorization: "Bearer ${accessToken || `{API_TOKEN}`}"
+				"Content-Type": "application/json",
+			},
 			method: "POST",
 			body: JSON.stringify(data),
 		}
@@ -20,12 +23,34 @@ query({"inputs": ${getModelInputSnippet(model)}}).then((response) => {
 	console.log(JSON.stringify(response));
 });`;
 
+export const snippetTextGeneration = (model: ModelDataMinimal, accessToken: string): string => {
+	if (model.config?.tokenizer_config?.chat_template) {
+		// Conversational model detected, so we display a code snippet that features the Messages API
+		return `import { HfInference } from "@huggingface/inference";
+
+const inference = new HfInference("${accessToken || `{API_TOKEN}`}");
+
+for await (const chunk of inference.chatCompletionStream({
+	model: "${model.id}",
+	messages: [{ role: "user", content: "What is the capital of France?" }],
+	max_tokens: 500,
+})) {
+	process.stdout.write(chunk.choices[0]?.delta?.content || "");
+}
+`;
+	} else {
+		return snippetBasic(model, accessToken);
+	}
+};
 export const snippetZeroShotClassification = (model: ModelDataMinimal, accessToken: string): string =>
 	`async function query(data) {
 	const response = await fetch(
 		"https://api-inference.huggingface.co/models/${model.id}",
 		{
-			headers: { Authorization: "Bearer ${accessToken || `{API_TOKEN}`}" },
+			headers: {
+				Authorization: "Bearer ${accessToken || `{API_TOKEN}`}"
+				"Content-Type": "application/json",
+			},
 			method: "POST",
 			body: JSON.stringify(data),
 		}
@@ -45,7 +70,10 @@ export const snippetTextToImage = (model: ModelDataMinimal, accessToken: string)
 	const response = await fetch(
 		"https://api-inference.huggingface.co/models/${model.id}",
 		{
-			headers: { Authorization: "Bearer ${accessToken || `{API_TOKEN}`}" },
+			headers: {
+				Authorization: "Bearer ${accessToken || `{API_TOKEN}`}"
+				"Content-Type": "application/json",
+			},
 			method: "POST",
 			body: JSON.stringify(data),
 		}
@@ -62,7 +90,10 @@ export const snippetTextToAudio = (model: ModelDataMinimal, accessToken: string)
 		const response = await fetch(
 			"https://api-inference.huggingface.co/models/${model.id}",
 			{
-				headers: { Authorization: "Bearer ${accessToken || `{API_TOKEN}`}" },
+				headers: {
+					Authorization: "Bearer ${accessToken || `{API_TOKEN}`}"
+					"Content-Type": "application/json",
+				},
 				method: "POST",
 				body: JSON.stringify(data),
 			}
@@ -99,7 +130,10 @@ export const snippetFile = (model: ModelDataMinimal, accessToken: string): strin
 	const response = await fetch(
 		"https://api-inference.huggingface.co/models/${model.id}",
 		{
-			headers: { Authorization: "Bearer ${accessToken || `{API_TOKEN}`}" },
+			headers: {
+				Authorization: "Bearer ${accessToken || `{API_TOKEN}`}"
+				"Content-Type": "application/json",
+			},
 			method: "POST",
 			body: data,
 		}
@@ -122,7 +156,7 @@ export const jsSnippets: Partial<Record<PipelineType, (model: ModelDataMinimal,
 	translation: snippetBasic,
 	summarization: snippetBasic,
 	"feature-extraction": snippetBasic,
-	"text-generation": snippetBasic,
+	"text-generation": snippetTextGeneration,
 	"text2text-generation": snippetBasic,
 	"fill-mask": snippetBasic,
 	"sentence-similarity": snippetBasic,

package/src/snippets/python.ts

@@ -2,6 +2,22 @@ import type { PipelineType } from "../pipelines.js";
 import { getModelInputSnippet } from "./inputs.js";
 import type { ModelDataMinimal } from "./types.js";
 
+export const snippetConversational = (model: ModelDataMinimal, accessToken: string): string =>
+	`from huggingface_hub import InferenceClient
+
+client = InferenceClient(
+    "${model.id}",
+    token="${accessToken || "{API_TOKEN}"}",
+)
+
+for message in client.chat_completion(
+	messages=[{"role": "user", "content": "What is the capital of France?"}],
+	max_tokens=500,
+	stream=True,
+):
+    print(message.choices[0].delta.content, end="")
+`;
+
 export const snippetZeroShotClassification = (model: ModelDataMinimal): string =>
 	`def query(payload):
 	response = requests.post(API_URL, headers=headers, json=payload)
@@ -107,7 +123,7 @@ output = query({
     "inputs": ${getModelInputSnippet(model)},
 })`;
 
-export const pythonSnippets: Partial<Record<PipelineType, (model: ModelDataMinimal) => string>> = {
+export const pythonSnippets: Partial<Record<PipelineType, (model: ModelDataMinimal, accessToken: string) => string>> = {
 	// Same order as in tasks/src/pipelines.ts
 	"text-classification": snippetBasic,
 	"token-classification": snippetBasic,
@@ -138,15 +154,22 @@ export const pythonSnippets: Partial<Record<PipelineType, (model: ModelDataMinim
 };
 
 export function getPythonInferenceSnippet(model: ModelDataMinimal, accessToken: string): string {
-	const body =
-		model.pipeline_tag && model.pipeline_tag in pythonSnippets ? pythonSnippets[model.pipeline_tag]?.(model) ?? "" : "";
+	if (model.pipeline_tag === "text-generation" && model.config?.tokenizer_config?.chat_template) {
+		// Conversational model detected, so we display a code snippet that features the Messages API
+		return snippetConversational(model, accessToken);
+	} else {
+		const body =
+			model.pipeline_tag && model.pipeline_tag in pythonSnippets
+				? pythonSnippets[model.pipeline_tag]?.(model, accessToken) ?? ""
+				: "";
 
-	return `import requests
+		return `import requests
 
 API_URL = "https://api-inference.huggingface.co/models/${model.id}"
 headers = {"Authorization": ${accessToken ? `"Bearer ${accessToken}"` : `f"Bearer {API_TOKEN}"`}}
 
 ${body}`;
+	}
 }
 
 export function hasPythonInferenceSnippet(model: ModelDataMinimal): boolean {

package/src/snippets/types.ts

@@ -5,4 +5,4 @@ import type { ModelData } from "../model-data";
  *
  * Add more fields as needed.
  */
-export type ModelDataMinimal = Pick<ModelData, "id" | "pipeline_tag" | "mask_token" | "library_name">;
+export type ModelDataMinimal = Pick<ModelData, "id" | "pipeline_tag" | "mask_token" | "library_name" | "config">;