@huggingface/inference 3.5.0 → 3.5.2

package/dist/index.js

@@ -46,7 +46,7 @@ var HF_HUB_URL = "https://huggingface.co";
 var HF_ROUTER_URL = "https://router.huggingface.co";
 
 // src/providers/black-forest-labs.ts
-var BLACK_FOREST_LABS_AI_API_BASE_URL = "https://api.us1.bfl.ai/v1";
+var BLACK_FOREST_LABS_AI_API_BASE_URL = "https://api.us1.bfl.ai";
 var makeBody = (params) => {
   return params.args;
 };
@@ -58,7 +58,7 @@ var makeHeaders = (params) => {
   }
 };
 var makeUrl = (params) => {
-  return `${params.baseUrl}/${params.model}`;
+  return `${params.baseUrl}/v1/${params.model}`;
 };
 var BLACK_FOREST_LABS_CONFIG = {
   baseUrl: BLACK_FOREST_LABS_AI_API_BASE_URL,
@@ -130,7 +130,7 @@ var FAL_AI_CONFIG = {
 };
 
 // src/providers/fireworks-ai.ts
-var FIREWORKS_AI_API_BASE_URL = "https://api.fireworks.ai/inference";
+var FIREWORKS_AI_API_BASE_URL = "https://api.fireworks.ai";
 var makeBody5 = (params) => {
   return {
     ...params.args,
@@ -142,9 +142,9 @@ var makeHeaders5 = (params) => {
 };
 var makeUrl5 = (params) => {
   if (params.task === "text-generation" && params.chatCompletion) {
-    return `${params.baseUrl}/v1/chat/completions`;
+    return `${params.baseUrl}/inference/v1/chat/completions`;
   }
-  return params.baseUrl;
+  return `${params.baseUrl}/inference`;
 };
 var FIREWORKS_AI_CONFIG = {
   baseUrl: FIREWORKS_AI_API_BASE_URL,
@@ -269,7 +269,7 @@ var makeBody10 = (params) => {
   };
 };
 var makeHeaders10 = (params) => {
-  return { Authorization: `Bearer ${params.accessToken}` };
+  return { Authorization: `Bearer ${params.accessToken}`, Prefer: "wait" };
 };
 var makeUrl10 = (params) => {
   if (params.model.includes(":")) {
@@ -373,7 +373,7 @@ function isUrl(modelOrUrl) {
 
 // package.json
 var name = "@huggingface/inference";
-var version = "3.5.0";
+var version = "3.5.2";
 
 // src/providers/consts.ts
 var HARDCODED_MODEL_ID_MAPPING = {
@@ -1526,8 +1526,8 @@ async function tabularClassification(args, options) {
   return res;
 }
 
-// src/HfInference.ts
-var HfInference = class {
+// src/InferenceClient.ts
+var InferenceClient = class {
   accessToken;
   defaultOptions;
   constructor(accessToken = "", defaultOptions = {}) {
@@ -1544,13 +1544,13 @@ var HfInference = class {
     }
   }
   /**
-   * Returns copy of HfInference tied to a specified endpoint.
+   * Returns copy of InferenceClient tied to a specified endpoint.
    */
   endpoint(endpointUrl) {
-    return new HfInferenceEndpoint(endpointUrl, this.accessToken, this.defaultOptions);
+    return new InferenceClientEndpoint(endpointUrl, this.accessToken, this.defaultOptions);
   }
 };
-var HfInferenceEndpoint = class {
+var InferenceClientEndpoint = class {
   constructor(endpointUrl, accessToken = "", defaultOptions = {}) {
     accessToken;
     defaultOptions;
@@ -1565,6 +1565,8 @@ var HfInferenceEndpoint = class {
     }
   }
 };
+var HfInference = class extends InferenceClient {
+};
 
 // src/types.ts
 var INFERENCE_PROVIDERS = [
@@ -1722,18 +1724,7 @@ function getCurlInferenceSnippet(model, accessToken, provider, providerModelId,
 // src/snippets/python.ts
 var python_exports = {};
 __export(python_exports, {
-  getPythonInferenceSnippet: () => getPythonInferenceSnippet,
-  pythonSnippets: () => pythonSnippets,
-  snippetBasic: () => snippetBasic2,
-  snippetConversational: () => snippetConversational,
-  snippetDocumentQuestionAnswering: () => snippetDocumentQuestionAnswering,
-  snippetFile: () => snippetFile2,
-  snippetTabular: () => snippetTabular,
-  snippetTextToAudio: () => snippetTextToAudio,
-  snippetTextToImage: () => snippetTextToImage,
-  snippetTextToVideo: () => snippetTextToVideo,
-  snippetZeroShotClassification: () => snippetZeroShotClassification2,
-  snippetZeroShotImageClassification: () => snippetZeroShotImageClassification
+  getPythonInferenceSnippet: () => getPythonInferenceSnippet
 });
 import { openAIbaseUrl } from "@huggingface/tasks";
 import {
@@ -1773,8 +1764,8 @@ var HFH_INFERENCE_CLIENT_METHODS = {
 var snippetImportInferenceClient = (accessToken, provider) => `from huggingface_hub import InferenceClient
 
 client = InferenceClient(
-	provider="${provider}",
-	api_key="${accessToken || "{API_TOKEN}"}"
+    provider="${provider}",
+    api_key="${accessToken || "{API_TOKEN}"}",
 )`;
 var snippetConversational = (model, accessToken, provider, providerModelId, opts) => {
   const streaming = opts?.streaming ?? true;
@@ -1802,7 +1793,7 @@ stream = client.chat.completions.create(
 	model="${model.id}", 
 	messages=messages, 
 	${configStr}
-	stream=True
+	stream=True,
 )
 
 for chunk in stream:
@@ -1912,9 +1903,8 @@ var snippetBasic2 = (model, accessToken, provider) => {
         content: `${snippetImportInferenceClient(accessToken, provider)}
 
 result = client.${HFH_INFERENCE_CLIENT_METHODS[model.pipeline_tag]}(
-	model="${model.id}",
 	inputs=${getModelInputSnippet2(model)},
-	provider="${provider}",
+	model="${model.id}",
 )
 
 print(result)
@@ -1956,7 +1946,7 @@ var snippetTextToImage = (model, accessToken, provider, providerModelId) => {
 # output is a PIL.Image object
 image = client.text_to_image(
 	${getModelInputSnippet2(model)},
-	model="${model.id}"
+	model="${model.id}",
 )`
     },
     ...provider === "fal-ai" ? [
@@ -2001,7 +1991,7 @@ var snippetTextToVideo = (model, accessToken, provider) => {
 
 video = client.text_to_video(
 	${getModelInputSnippet2(model)},
-	model="${model.id}"
+	model="${model.id}",
 )`
     }
   ] : [];
@@ -2055,23 +2045,79 @@ Audio(audio, rate=sampling_rate)`
     ];
   }
 };
-var snippetDocumentQuestionAnswering = (model) => {
+var snippetAutomaticSpeechRecognition = (model, accessToken, provider) => {
   return [
+    {
+      client: "huggingface_hub",
+      content: `${snippetImportInferenceClient(accessToken, provider)}
+output = client.automatic_speech_recognition(${getModelInputSnippet2(model)}, model="${model.id}")`
+    },
+    snippetFile2(model)[0]
+  ];
+};
+var snippetDocumentQuestionAnswering = (model, accessToken, provider) => {
+  const inputsAsStr = getModelInputSnippet2(model);
+  const inputsAsObj = JSON.parse(inputsAsStr);
+  return [
+    {
+      client: "huggingface_hub",
+      content: `${snippetImportInferenceClient(accessToken, provider)}
+output = client.document_question_answering(
+    "${inputsAsObj.image}",
+	question="${inputsAsObj.question}",
+	model="${model.id}",
+)`
+    },
     {
       client: "requests",
       content: `def query(payload):
 	with open(payload["image"], "rb") as f:
 		img = f.read()
-		payload["image"] = base64.b64encode(img).decode("utf-8")  
+		payload["image"] = base64.b64encode(img).decode("utf-8")
 	response = requests.post(API_URL, headers=headers, json=payload)
 	return response.json()
 
 output = query({
-    "inputs": ${getModelInputSnippet2(model)},
+    "inputs": ${inputsAsStr},
 })`
     }
   ];
 };
+var snippetImageToImage = (model, accessToken, provider) => {
+  const inputsAsStr = getModelInputSnippet2(model);
+  const inputsAsObj = JSON.parse(inputsAsStr);
+  return [
+    {
+      client: "huggingface_hub",
+      content: `${snippetImportInferenceClient(accessToken, provider)}
+# output is a PIL.Image object
+image = client.image_to_image(
+    "${inputsAsObj.image}",
+    prompt="${inputsAsObj.prompt}",
+    model="${model.id}",
+)`
+    },
+    {
+      client: "requests",
+      content: `def query(payload):
+	with open(payload["inputs"], "rb") as f:
+		img = f.read()
+		payload["inputs"] = base64.b64encode(img).decode("utf-8")
+	response = requests.post(API_URL, headers=headers, json=payload)
+	return response.content
+
+image_bytes = query({
+	"inputs": "${inputsAsObj.image}",
+	"parameters": {"prompt": "${inputsAsObj.prompt}"},
+})
+
+# You can access the image with PIL.Image for example
+import io
+from PIL import Image
+image = Image.open(io.BytesIO(image_bytes))`
+    }
+  ];
+};
 var pythonSnippets = {
   // Same order as in tasks/src/pipelines.ts
   "text-classification": snippetBasic2,
@@ -2087,7 +2133,7 @@ var pythonSnippets = {
   "image-text-to-text": snippetConversational,
   "fill-mask": snippetBasic2,
   "sentence-similarity": snippetBasic2,
-  "automatic-speech-recognition": snippetFile2,
+  "automatic-speech-recognition": snippetAutomaticSpeechRecognition,
   "text-to-image": snippetTextToImage,
   "text-to-video": snippetTextToVideo,
   "text-to-speech": snippetTextToAudio,
@@ -2101,6 +2147,7 @@ var pythonSnippets = {
   "image-segmentation": snippetFile2,
   "document-question-answering": snippetDocumentQuestionAnswering,
   "image-to-text": snippetFile2,
+  "image-to-image": snippetImageToImage,
   "zero-shot-image-classification": snippetZeroShotImageClassification
 };
 function getPythonInferenceSnippet(model, accessToken, provider, providerModelId, opts) {
@@ -2111,23 +2158,33 @@ function getPythonInferenceSnippet(model, accessToken, provider, providerModelId
     return snippets.map((snippet) => {
       return {
         ...snippet,
-        content: snippet.client === "requests" ? `import requests
-
-API_URL = "${openAIbaseUrl(provider)}"
-headers = {"Authorization": ${accessToken ? `"Bearer ${accessToken}"` : `f"Bearer {API_TOKEN}"`}}
-
-${snippet.content}` : snippet.content
+        content: addImportsToSnippet(snippet.content, model, accessToken)
       };
     });
   }
 }
+var addImportsToSnippet = (snippet, model, accessToken) => {
+  if (snippet.includes("requests")) {
+    snippet = `import requests
+
+API_URL = "https://router.huggingface.co/hf-inference/models/${model.id}"
+headers = {"Authorization": ${accessToken ? `"Bearer ${accessToken}"` : `f"Bearer {API_TOKEN}"`}}
+
+${snippet}`;
+  }
+  if (snippet.includes("base64")) {
+    snippet = `import base64
+${snippet}`;
+  }
+  return snippet;
+};
 
 // src/snippets/js.ts
 var js_exports = {};
 __export(js_exports, {
   getJsInferenceSnippet: () => getJsInferenceSnippet,
   jsSnippets: () => jsSnippets,
-  snippetAutomaticSpeechRecognition: () => snippetAutomaticSpeechRecognition,
+  snippetAutomaticSpeechRecognition: () => snippetAutomaticSpeechRecognition2,
   snippetBasic: () => snippetBasic3,
   snippetFile: () => snippetFile3,
   snippetTextGeneration: () => snippetTextGeneration2,
@@ -2160,9 +2217,9 @@ var snippetBasic3 = (model, accessToken, provider) => {
     ...model.pipeline_tag && model.pipeline_tag in HFJS_METHODS ? [
       {
         client: "huggingface.js",
-        content: `import { HfInference } from "@huggingface/inference";
+        content: `import { InferenceClient } from "@huggingface/inference";
 
-const client = new HfInference("${accessToken || `{API_TOKEN}`}");
+const client = new InferenceClient("${accessToken || `{API_TOKEN}`}");
 
 const output = await client.${HFJS_METHODS[model.pipeline_tag]}({
 	model: "${model.id}",
@@ -2217,9 +2274,9 @@ var snippetTextGeneration2 = (model, accessToken, provider, providerModelId, opt
       return [
         {
           client: "huggingface.js",
-          content: `import { HfInference } from "@huggingface/inference";
+          content: `import { InferenceClient } from "@huggingface/inference";
 
-const client = new HfInference("${accessToken || `{API_TOKEN}`}");
+const client = new InferenceClient("${accessToken || `{API_TOKEN}`}");
 
 let out = "";
 
@@ -2269,9 +2326,9 @@ for await (const chunk of stream) {
       return [
         {
           client: "huggingface.js",
-          content: `import { HfInference } from "@huggingface/inference";
+          content: `import { InferenceClient } from "@huggingface/inference";
 
-const client = new HfInference("${accessToken || `{API_TOKEN}`}");
+const client = new InferenceClient("${accessToken || `{API_TOKEN}`}");
 
 const chatCompletion = await client.chatCompletion({
 	model: "${model.id}",
@@ -2339,9 +2396,9 @@ var snippetTextToImage2 = (model, accessToken, provider) => {
   return [
     {
       client: "huggingface.js",
-      content: `import { HfInference } from "@huggingface/inference";
+      content: `import { InferenceClient } from "@huggingface/inference";
 
-const client = new HfInference("${accessToken || `{API_TOKEN}`}");
+const client = new InferenceClient("${accessToken || `{API_TOKEN}`}");
 
 const image = await client.textToImage({
 	model: "${model.id}",
@@ -2381,9 +2438,9 @@ var snippetTextToVideo2 = (model, accessToken, provider) => {
   return ["fal-ai", "replicate"].includes(provider) ? [
     {
       client: "huggingface.js",
-      content: `import { HfInference } from "@huggingface/inference";
+      content: `import { InferenceClient } from "@huggingface/inference";
 
-const client = new HfInference("${accessToken || `{API_TOKEN}`}");
+const client = new InferenceClient("${accessToken || `{API_TOKEN}`}");
 
 const video = await client.textToVideo({
 	model: "${model.id}",
@@ -2441,13 +2498,13 @@ var snippetTextToAudio2 = (model, accessToken, provider) => {
     ];
   }
 };
-var snippetAutomaticSpeechRecognition = (model, accessToken, provider) => {
+var snippetAutomaticSpeechRecognition2 = (model, accessToken, provider) => {
   return [
     {
       client: "huggingface.js",
-      content: `import { HfInference } from "@huggingface/inference";
+      content: `import { InferenceClient } from "@huggingface/inference";
 
-const client = new HfInference("${accessToken || `{API_TOKEN}`}");
+const client = new InferenceClient("${accessToken || `{API_TOKEN}`}");
 
 const data = fs.readFileSync(${getModelInputSnippet3(model)});
 
@@ -2508,7 +2565,7 @@ var jsSnippets = {
   "text2text-generation": snippetBasic3,
   "fill-mask": snippetBasic3,
   "sentence-similarity": snippetBasic3,
-  "automatic-speech-recognition": snippetAutomaticSpeechRecognition,
+  "automatic-speech-recognition": snippetAutomaticSpeechRecognition2,
   "text-to-image": snippetTextToImage2,
   "text-to-video": snippetTextToVideo2,
   "text-to-speech": snippetTextToAudio2,
@@ -2525,8 +2582,9 @@ function getJsInferenceSnippet(model, accessToken, provider, providerModelId, op
 }
 export {
   HfInference,
-  HfInferenceEndpoint,
   INFERENCE_PROVIDERS,
+  InferenceClient,
+  InferenceClientEndpoint,
   InferenceOutputError,
   audioClassification,
   audioToAudio,

package/dist/src/{HfInference.d.ts → InferenceClient.d.ts}

@@ -8,21 +8,26 @@ type TaskWithNoAccessToken = {
 type TaskWithNoAccessTokenNoEndpointUrl = {
     [key in keyof Task]: (args: DistributiveOmit<Parameters<Task[key]>[0], "accessToken" | "endpointUrl">, options?: Parameters<Task[key]>[1]) => ReturnType<Task[key]>;
 };
-export declare class HfInference {
+export declare class InferenceClient {
     private readonly accessToken;
     private readonly defaultOptions;
     constructor(accessToken?: string, defaultOptions?: Options);
     /**
-     * Returns copy of HfInference tied to a specified endpoint.
+     * Returns copy of InferenceClient tied to a specified endpoint.
      */
-    endpoint(endpointUrl: string): HfInferenceEndpoint;
+    endpoint(endpointUrl: string): InferenceClientEndpoint;
 }
-export declare class HfInferenceEndpoint {
+export declare class InferenceClientEndpoint {
     constructor(endpointUrl: string, accessToken?: string, defaultOptions?: Options);
 }
-export interface HfInference extends TaskWithNoAccessToken {
+export interface InferenceClient extends TaskWithNoAccessToken {
 }
-export interface HfInferenceEndpoint extends TaskWithNoAccessTokenNoEndpointUrl {
+export interface InferenceClientEndpoint extends TaskWithNoAccessTokenNoEndpointUrl {
+}
+/**
+ * For backward compatibility only.
+ */
+export declare class HfInference extends InferenceClient {
 }
 export {};
-//# sourceMappingURL=HfInference.d.ts.map
+//# sourceMappingURL=InferenceClient.d.ts.map

package/dist/src/InferenceClient.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"InferenceClient.d.ts","sourceRoot":"","sources":["../../src/InferenceClient.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,KAAK,MAAM,SAAS,CAAC;AACjC,OAAO,KAAK,EAAE,OAAO,EAAe,MAAM,SAAS,CAAC;AACpD,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,2BAA2B,CAAC;AAKlE,KAAK,IAAI,GAAG,OAAO,KAAK,CAAC;AAEzB,KAAK,qBAAqB,GAAG;KAC3B,GAAG,IAAI,MAAM,IAAI,GAAG,CACpB,IAAI,EAAE,gBAAgB,CAAC,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,aAAa,CAAC,EAC/D,OAAO,CAAC,EAAE,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,KAC9B,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;CAC1B,CAAC;AAEF,KAAK,kCAAkC,GAAG;KACxC,GAAG,IAAI,MAAM,IAAI,GAAG,CACpB,IAAI,EAAE,gBAAgB,CAAC,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,aAAa,GAAG,aAAa,CAAC,EAC/E,OAAO,CAAC,EAAE,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,KAC9B,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;CAC1B,CAAC;AAEF,qBAAa,eAAe;IAC3B,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAS;IACrC,OAAO,CAAC,QAAQ,CAAC,cAAc,CAAU;gBAE7B,WAAW,SAAK,EAAE,cAAc,GAAE,OAAY;IAc1D;;OAEG;IACI,QAAQ,CAAC,WAAW,EAAE,MAAM,GAAG,uBAAuB;CAG7D;AAED,qBAAa,uBAAuB;gBACvB,WAAW,EAAE,MAAM,EAAE,WAAW,SAAK,EAAE,cAAc,GAAE,OAAY;CAa/E;AAED,MAAM,WAAW,eAAgB,SAAQ,qBAAqB;CAAG;AAEjE,MAAM,WAAW,uBAAwB,SAAQ,kCAAkC;CAAG;AAEtF;;GAEG;AACH,qBAAa,WAAY,SAAQ,eAAe;CAAG"}

package/dist/src/index.d.ts

@@ -1,4 +1,4 @@
-export { HfInference, HfInferenceEndpoint } from "./HfInference";
+export { InferenceClient, InferenceClientEndpoint, HfInference } from "./InferenceClient";
 export { InferenceOutputError } from "./lib/InferenceOutputError";
 export * from "./types";
 export * from "./tasks";

package/dist/src/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,mBAAmB,EAAE,MAAM,eAAe,CAAC;AACjE,OAAO,EAAE,oBAAoB,EAAE,MAAM,4BAA4B,CAAC;AAClE,cAAc,SAAS,CAAC;AACxB,cAAc,SAAS,CAAC;AAExB,OAAO,KAAK,QAAQ,MAAM,qBAAqB,CAAC;AAChD,OAAO,EAAE,QAAQ,EAAE,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,eAAe,EAAE,uBAAuB,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAC1F,OAAO,EAAE,oBAAoB,EAAE,MAAM,4BAA4B,CAAC;AAClE,cAAc,SAAS,CAAC;AACxB,cAAc,SAAS,CAAC;AAExB,OAAO,KAAK,QAAQ,MAAM,qBAAqB,CAAC;AAChD,OAAO,EAAE,QAAQ,EAAE,CAAC"}

package/dist/src/providers/cohere.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"cohere.d.ts","sourceRoot":"","sources":["../../../src/providers/cohere.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AACH,OAAO,KAAK,EAAE,cAAc,EAAuC,MAAM,UAAU,CAAC;AAoBpF,eAAO,MAAM,aAAa,EAAE,cAK3B,CAAC"}
+{"version":3,"file":"cohere.d.ts","sourceRoot":"","sources":["../../../src/providers/cohere.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AACH,OAAO,KAAK,EAAE,cAAc,EAAuC,MAAM,UAAU,CAAC;AAmBpF,eAAO,MAAM,aAAa,EAAE,cAK3B,CAAC"}

package/dist/src/snippets/python.d.ts

@@ -1,23 +1,4 @@
 import { type SnippetInferenceProvider } from "@huggingface/tasks";
-import type { PipelineType } from "@huggingface/tasks/src/pipelines.js";
-import type { ChatCompletionInputMessage, GenerationParameters } from "@huggingface/tasks/src/tasks/index.js";
 import { type InferenceSnippet, type ModelDataMinimal } from "@huggingface/tasks";
-export declare const snippetConversational: (model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider, providerModelId?: string, opts?: {
-    streaming?: boolean;
-    messages?: ChatCompletionInputMessage[];
-    temperature?: GenerationParameters["temperature"];
-    max_tokens?: GenerationParameters["max_tokens"];
-    top_p?: GenerationParameters["top_p"];
-}) => InferenceSnippet[];
-export declare const snippetZeroShotClassification: (model: ModelDataMinimal) => InferenceSnippet[];
-export declare const snippetZeroShotImageClassification: (model: ModelDataMinimal) => InferenceSnippet[];
-export declare const snippetBasic: (model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider) => InferenceSnippet[];
-export declare const snippetFile: (model: ModelDataMinimal) => InferenceSnippet[];
-export declare const snippetTextToImage: (model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider, providerModelId?: string) => InferenceSnippet[];
-export declare const snippetTextToVideo: (model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider) => InferenceSnippet[];
-export declare const snippetTabular: (model: ModelDataMinimal) => InferenceSnippet[];
-export declare const snippetTextToAudio: (model: ModelDataMinimal) => InferenceSnippet[];
-export declare const snippetDocumentQuestionAnswering: (model: ModelDataMinimal) => InferenceSnippet[];
-export declare const pythonSnippets: Partial<Record<PipelineType, (model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider, providerModelId?: string, opts?: Record<string, unknown>) => InferenceSnippet[]>>;
 export declare function getPythonInferenceSnippet(model: ModelDataMinimal, accessToken: string, provider: SnippetInferenceProvider, providerModelId?: string, opts?: Record<string, unknown>): InferenceSnippet[];
 //# sourceMappingURL=python.d.ts.map

package/dist/src/snippets/python.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"python.d.ts","sourceRoot":"","sources":["../../../src/snippets/python.ts"],"names":[],"mappings":"AAAA,OAAO,EAAiB,KAAK,wBAAwB,EAAE,MAAM,oBAAoB,CAAC;AAClF,OAAO,KAAK,EAAE,YAAY,EAAc,MAAM,qCAAqC,CAAC;AACpF,OAAO,KAAK,EAAE,0BAA0B,EAAE,oBAAoB,EAAE,MAAM,uCAAuC,CAAC;AAC9G,OAAO,EACN,KAAK,gBAAgB,EACrB,KAAK,gBAAgB,EAIrB,MAAM,oBAAoB,CAAC;AAyC5B,eAAO,MAAM,qBAAqB,UAC1B,gBAAgB,eACV,MAAM,YACT,wBAAwB,oBAChB,MAAM,SACjB;IACN,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,QAAQ,CAAC,EAAE,0BAA0B,EAAE,CAAC;IACxC,WAAW,CAAC,EAAE,oBAAoB,CAAC,aAAa,CAAC,CAAC;IAClD,UAAU,CAAC,EAAE,oBAAoB,CAAC,YAAY,CAAC,CAAC;IAChD,KAAK,CAAC,EAAE,oBAAoB,CAAC,OAAO,CAAC,CAAC;CACtC,KACC,gBAAgB,EAiGlB,CAAC;AAEF,eAAO,MAAM,6BAA6B,UAAW,gBAAgB,KAAG,gBAAgB,EAevF,CAAC;AAEF,eAAO,MAAM,kCAAkC,UAAW,gBAAgB,KAAG,gBAAgB,EAqB5F,CAAC;AAEF,eAAO,MAAM,YAAY,UACjB,gBAAgB,eACV,MAAM,YACT,wBAAwB,KAChC,gBAAgB,EAgClB,CAAC;AAEF,eAAO,MAAM,WAAW,UAAW,gBAAgB,KAAG,gBAAgB,EAcrE,CAAC;AAEF,eAAO,MAAM,kBAAkB,UACvB,gBAAgB,eACV,MAAM,YACT,wBAAwB,oBAChB,MAAM,KACtB,gBAAgB,EAoDlB,CAAC;AAEF,eAAO,MAAM,kBAAkB,UACvB,gBAAgB,eACV,MAAM,YACT,wBAAwB,KAChC,gBAAgB,EAelB,CAAC;AAEF,eAAO,MAAM,cAAc,UAAW,gBAAgB,KAAG,gBAAgB,EAcxE,CAAC;AAEF,eAAO,MAAM,kBAAkB,UAAW,gBAAgB,KAAG,gBAAgB,EAuC5E,CAAC;AAEF,eAAO,MAAM,gCAAgC,UAAW,gBAAgB,KAAG,gBAAgB,EAiB1F,CAAC;AAEF,eAAO,MAAM,cAAc,EAAE,OAAO,CACnC,MAAM,CACL,YAAY,EACZ,CACC,KAAK,EAAE,gBAAgB,EACvB,WAAW,EAAE,MAAM,EACnB,QAAQ,EAAE,wBAAwB,EAClC,eAAe,CAAC,EAAE,MAAM,EACxB,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,KAC1B,gBAAgB,EAAE,CACvB,CA+BD,CAAC;AAEF,wBAAgB,yBAAyB,CACxC,KAAK,EAAE,gBAAgB,EACvB,WAAW,EAAE,MAAM,EACnB,QAAQ,EAAE,wBAAwB,EAClC,eAAe,CAAC,EAAE,MAAM,EACxB,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAC5B,gBAAgB,EAAE,CA0BpB"}
+{"version":3,"file":"python.d.ts","sourceRoot":"","sources":["../../../src/snippets/python.ts"],"names":[],"mappings":"AAAA,OAAO,EAAiB,KAAK,wBAAwB,EAAE,MAAM,oBAAoB,CAAC;AAGlF,OAAO,EACN,KAAK,gBAAgB,EACrB,KAAK,gBAAgB,EAIrB,MAAM,oBAAoB,CAAC;AAkgB5B,wBAAgB,yBAAyB,CACxC,KAAK,EAAE,gBAAgB,EACvB,WAAW,EAAE,MAAM,EACnB,QAAQ,EAAE,wBAAwB,EAClC,eAAe,CAAC,EAAE,MAAM,EACxB,IAAI,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAC5B,gBAAgB,EAAE,CAiBpB"}

package/dist/test/InferenceClient.spec.d.ts

@@ -0,0 +1,2 @@
+import "./vcr";
+//# sourceMappingURL=InferenceClient.spec.d.ts.map

package/dist/test/InferenceClient.spec.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"InferenceClient.spec.d.ts","sourceRoot":"","sources":["../../test/InferenceClient.spec.ts"],"names":[],"mappings":"AAeA,OAAO,OAAO,CAAC"}

package/package.json

@@ -1,10 +1,10 @@
 {
   "name": "@huggingface/inference",
-  "version": "3.5.0",
+  "version": "3.5.2",
   "packageManager": "pnpm@8.10.5",
   "license": "MIT",
-  "author": "Tim Mikeladze <tim.mikeladze@gmail.com>",
-  "description": "Typescript wrapper for the Hugging Face Inference Endpoints & Inference API",
+  "author": "Hugging Face and Tim Mikeladze <tim.mikeladze@gmail.com>",
+  "description": "Typescript client for the Hugging Face Inference Providers and Inference Endpoints",
   "repository": {
     "type": "git",
     "url": "https://github.com/huggingface/huggingface.js.git"
@@ -39,7 +39,7 @@
   },
   "type": "module",
   "dependencies": {
-    "@huggingface/tasks": "^0.17.0"
+    "@huggingface/tasks": "^0.17.4"
   },
   "devDependencies": {
     "@types/node": "18.13.0"

package/src/{HfInference.ts → InferenceClient.ts}

@@ -21,7 +21,7 @@ type TaskWithNoAccessTokenNoEndpointUrl = {
 	) => ReturnType<Task[key]>;
 };
 
-export class HfInference {
+export class InferenceClient {
 	private readonly accessToken: string;
 	private readonly defaultOptions: Options;
 
@@ -40,14 +40,14 @@ export class HfInference {
 	}
 
 	/**
-	 * Returns copy of HfInference tied to a specified endpoint.
+	 * Returns copy of InferenceClient tied to a specified endpoint.
 	 */
-	public endpoint(endpointUrl: string): HfInferenceEndpoint {
-		return new HfInferenceEndpoint(endpointUrl, this.accessToken, this.defaultOptions);
+	public endpoint(endpointUrl: string): InferenceClientEndpoint {
+		return new InferenceClientEndpoint(endpointUrl, this.accessToken, this.defaultOptions);
 	}
 }
 
-export class HfInferenceEndpoint {
+export class InferenceClientEndpoint {
 	constructor(endpointUrl: string, accessToken = "", defaultOptions: Options = {}) {
 		accessToken;
 		defaultOptions;
@@ -63,6 +63,11 @@ export class HfInferenceEndpoint {
 	}
 }
 
-export interface HfInference extends TaskWithNoAccessToken {}
+export interface InferenceClient extends TaskWithNoAccessToken {}
 
-export interface HfInferenceEndpoint extends TaskWithNoAccessTokenNoEndpointUrl {}
+export interface InferenceClientEndpoint extends TaskWithNoAccessTokenNoEndpointUrl {}
+
+/**
+ * For backward compatibility only.
+ */
+export class HfInference extends InferenceClient {}

package/src/index.ts

@@ -1,4 +1,4 @@
-export { HfInference, HfInferenceEndpoint } from "./HfInference";
+export { InferenceClient, InferenceClientEndpoint, HfInference } from "./InferenceClient";
 export { InferenceOutputError } from "./lib/InferenceOutputError";
 export * from "./types";
 export * from "./tasks";

package/src/providers/black-forest-labs.ts

@@ -16,7 +16,7 @@
  */
 import type { ProviderConfig, UrlParams, HeaderParams, BodyParams } from "../types";
 
-const BLACK_FOREST_LABS_AI_API_BASE_URL = "https://api.us1.bfl.ai/v1";
+const BLACK_FOREST_LABS_AI_API_BASE_URL = "https://api.us1.bfl.ai";
 
 const makeBody = (params: BodyParams): Record<string, unknown> => {
 	return params.args;
@@ -31,7 +31,7 @@ const makeHeaders = (params: HeaderParams): Record<string, string> => {
 };
 
 const makeUrl = (params: UrlParams): string => {
-	return `${params.baseUrl}/${params.model}`;
+	return `${params.baseUrl}/v1/${params.model}`;
 };
 
 export const BLACK_FOREST_LABS_CONFIG: ProviderConfig = {

package/src/providers/cohere.ts

@@ -18,7 +18,6 @@ import type { ProviderConfig, UrlParams, HeaderParams, BodyParams } from "../typ
 
 const COHERE_API_BASE_URL = "https://api.cohere.com";
 
-
 const makeBody = (params: BodyParams): Record<string, unknown> => {
 	return {
 		...params.args,

package/src/providers/fireworks-ai.ts

@@ -16,7 +16,7 @@
  */
 import type { ProviderConfig, UrlParams, HeaderParams, BodyParams } from "../types";
 
-const FIREWORKS_AI_API_BASE_URL = "https://api.fireworks.ai/inference";
+const FIREWORKS_AI_API_BASE_URL = "https://api.fireworks.ai";
 
 const makeBody = (params: BodyParams): Record<string, unknown> => {
 	return {
@@ -31,9 +31,9 @@ const makeHeaders = (params: HeaderParams): Record<string, string> => {
 
 const makeUrl = (params: UrlParams): string => {
 	if (params.task === "text-generation" && params.chatCompletion) {
-		return `${params.baseUrl}/v1/chat/completions`;
+		return `${params.baseUrl}/inference/v1/chat/completions`;
 	}
-	return params.baseUrl;
+	return `${params.baseUrl}/inference`;
 };
 
 export const FIREWORKS_AI_CONFIG: ProviderConfig = {

package/src/providers/replicate.ts

@@ -26,7 +26,7 @@ const makeBody = (params: BodyParams): Record<string, unknown> => {
 };
 
 const makeHeaders = (params: HeaderParams): Record<string, string> => {
-	return { Authorization: `Bearer ${params.accessToken}` };
+	return { Authorization: `Bearer ${params.accessToken}`, Prefer: "wait" };
 };
 
 const makeUrl = (params: UrlParams): string => {