Merge branch 'main' into v2ark/add_centml

V2arK · web-flow · commit d73d6040c3f2 · 2025-04-30T01:08:23.000-06:00
diff --git a/.github/workflows/agents-publish.yml b/.github/workflows/agents-publish.yml
@@ -55,7 +55,7 @@ jobs:
       - run: pnpm --filter agents... build && pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/blob-publish.yml b/.github/workflows/blob-publish.yml
@@ -53,7 +53,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/dduf-publish.yml b/.github/workflows/dduf-publish.yml
@@ -56,7 +56,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/gguf-publish.yml b/.github/workflows/gguf-publish.yml
@@ -56,7 +56,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/hub-publish.yml b/.github/workflows/hub-publish.yml
@@ -59,7 +59,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/inference-publish.yml b/.github/workflows/inference-publish.yml
@@ -59,7 +59,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/jinja-publish.yml b/.github/workflows/jinja-publish.yml
@@ -52,7 +52,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/languages-publish.yml b/.github/workflows/languages-publish.yml
@@ -52,7 +52,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/lint.yml b/.github/workflows/lint.yml
@@ -47,7 +47,7 @@ jobs:
         run: |
           # Could also use pnpm --filter [${{ steps.since.outputs.SINCE }}] format:check
           # But this way we can see the diff
-          pnpm --filter [${{ steps.since.outputs.SINCE }}] format
+          pnpm -r format
           git diff | head -n 100
           git diff --name-only --exit-code
 
diff --git a/.github/workflows/mcp-client-publish.yml b/.github/workflows/mcp-client-publish.yml
@@ -55,7 +55,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/ollama-utils-publish.yml b/.github/workflows/ollama-utils-publish.yml
@@ -56,7 +56,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/space-header-publish.yml b/.github/workflows/space-header-publish.yml
@@ -52,7 +52,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/.github/workflows/tasks-publish.yml b/.github/workflows/tasks-publish.yml
@@ -53,7 +53,7 @@ jobs:
       - run: pnpm publish --no-git-checks .
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
-      - run: git pull --rebase && git push --follow-tags
+      - run: (git pull --rebase && git push --follow-tags) || (git pull --rebase && git push --follow-tags)
       # hack - reuse actions/setup-node@v3 just to set a new registry
       - uses: actions/setup-node@v3
         with:
diff --git a/README.md b/README.md
@@ -97,7 +97,7 @@ You can run our packages with vanilla JS, without any bundler, by using a CDN or
 
 ```html
 <script type="module">
-    import { InferenceClient } from 'https://cdn.jsdelivr.net/npm/@huggingface/inference@3.9.2/+esm';
+    import { InferenceClient } from 'https://cdn.jsdelivr.net/npm/@huggingface/inference@3.10.0/+esm';
     import { createRepo, commit, deleteRepo, listFiles } from "https://cdn.jsdelivr.net/npm/@huggingface/hub@1.1.2/+esm";
 </script>
 ```
diff --git a/packages/inference/README.md b/packages/inference/README.md
@@ -54,6 +54,7 @@ Currently, we support the following providers:
 - [Nebius](https://studio.nebius.ai)
 - [Novita](https://novita.ai/?utm_source=github_huggingface&utm_medium=github_readme&utm_campaign=link)
 - [Nscale](https://nscale.com)
+- [OVHcloud](https://endpoints.ai.cloud.ovh.net/)
 - [Replicate](https://replicate.com)
 - [Sambanova](https://sambanova.ai)
 - [Together](https://together.xyz)
@@ -85,6 +86,7 @@ Only a subset of models are supported when requesting third-party providers. You
 - [Hyperbolic supported models](https://huggingface.co/api/partners/hyperbolic/models)
 - [Nebius supported models](https://huggingface.co/api/partners/nebius/models)
 - [Nscale supported models](https://huggingface.co/api/partners/nscale/models)
+- [OVHcloud supported models](https://huggingface.co/api/partners/ovhcloud/models)
 - [Replicate supported models](https://huggingface.co/api/partners/replicate/models)
 - [Sambanova supported models](https://huggingface.co/api/partners/sambanova/models)
 - [Together supported models](https://huggingface.co/api/partners/together/models)
diff --git a/packages/inference/package.json b/packages/inference/package.json
@@ -1,6 +1,6 @@
 {
 	"name": "@huggingface/inference",
-	"version": "3.9.2",
+	"version": "3.10.0",
 	"packageManager": "pnpm@8.10.5",
 	"license": "MIT",
 	"author": "Hugging Face and Tim Mikeladze <tim.mikeladze@gmail.com>",
diff --git a/packages/inference/src/lib/getProviderHelper.ts b/packages/inference/src/lib/getProviderHelper.ts
@@ -12,6 +12,7 @@ import * as Nebius from "../providers/nebius";
 import * as Novita from "../providers/novita";
 import * as Nscale from "../providers/nscale";
 import * as OpenAI from "../providers/openai";
+import * as OvhCloud from "../providers/ovhcloud";
 import type {
 	AudioClassificationTaskHelper,
 	AudioToAudioTaskHelper,
@@ -130,6 +131,10 @@ export const PROVIDERS: Record<InferenceProvider, Partial<Record<InferenceTask,
 	openai: {
 		conversational: new OpenAI.OpenAIConversationalTask(),
 	},
+	ovhcloud: {
+		conversational: new OvhCloud.OvhCloudConversationalTask(),
+		"text-generation": new OvhCloud.OvhCloudTextGenerationTask(),
+	},
 	replicate: {
 		"text-to-image": new Replicate.ReplicateTextToImageTask(),
 		"text-to-speech": new Replicate.ReplicateTextToSpeechTask(),
diff --git a/packages/inference/src/providers/consts.ts b/packages/inference/src/providers/consts.ts
@@ -33,6 +33,7 @@ export const HARDCODED_MODEL_INFERENCE_MAPPING: Record<
 	novita: {},
 	nscale: {},
 	openai: {},
+	ovhcloud: {},
 	replicate: {},
 	sambanova: {},
 	together: {},
diff --git a/packages/inference/src/providers/ovhcloud.ts b/packages/inference/src/providers/ovhcloud.ts
@@ -0,0 +1,75 @@
+/**
+ * See the registered mapping of HF model ID => OVHcloud model ID here:
+ *
+ * https://huggingface.co/api/partners/ovhcloud/models
+ *
+ * This is a publicly available mapping.
+ *
+ * If you want to try to run inference for a new model locally before it's registered on huggingface.co,
+ * you can add it to the dictionary "HARDCODED_MODEL_ID_MAPPING" in consts.ts, for dev purposes.
+ *
+ * - If you work at OVHcloud and want to update this mapping, please use the model mapping API we provide on huggingface.co
+ * - If you're a community member and want to add a new supported HF model to OVHcloud, please open an issue on the present repo
+ * and we will tag OVHcloud team members.
+ *
+ * Thanks!
+ */
+
+import { BaseConversationalTask, BaseTextGenerationTask } from "./providerHelper";
+import type { ChatCompletionOutput, TextGenerationOutput, TextGenerationOutputFinishReason } from "@huggingface/tasks";
+import { InferenceOutputError } from "../lib/InferenceOutputError";
+import type { BodyParams } from "../types";
+import { omit } from "../utils/omit";
+import type { TextGenerationInput } from "@huggingface/tasks";
+
+const OVHCLOUD_API_BASE_URL = "https://oai.endpoints.kepler.ai.cloud.ovh.net";
+
+interface OvhCloudTextCompletionOutput extends Omit<ChatCompletionOutput, "choices"> {
+	choices: Array<{
+		text: string;
+		finish_reason: TextGenerationOutputFinishReason;
+		logprobs: unknown;
+		index: number;
+	}>;
+}
+
+export class OvhCloudConversationalTask extends BaseConversationalTask {
+	constructor() {
+		super("ovhcloud", OVHCLOUD_API_BASE_URL);
+	}
+}
+
+export class OvhCloudTextGenerationTask extends BaseTextGenerationTask {
+	constructor() {
+		super("ovhcloud", OVHCLOUD_API_BASE_URL);
+	}
+
+	override preparePayload(params: BodyParams<TextGenerationInput>): Record<string, unknown> {
+		return {
+			model: params.model,
+			...omit(params.args, ["inputs", "parameters"]),
+			...(params.args.parameters
+				? {
+						max_tokens: (params.args.parameters as Record<string, unknown>).max_new_tokens,
+						...omit(params.args.parameters as Record<string, unknown>, "max_new_tokens"),
+				  }
+				: undefined),
+			prompt: params.args.inputs,
+		};
+	}
+
+	override async getResponse(response: OvhCloudTextCompletionOutput): Promise<TextGenerationOutput> {
+		if (
+			typeof response === "object" &&
+			"choices" in response &&
+			Array.isArray(response?.choices) &&
+			typeof response?.model === "string"
+		) {
+			const completion = response.choices[0];
+			return {
+				generated_text: completion.text,
+			};
+		}
+		throw new InferenceOutputError("Expected OVHcloud text generation response format");
+	}
+}
diff --git a/packages/inference/src/types.ts b/packages/inference/src/types.ts
@@ -52,6 +52,7 @@ export const INFERENCE_PROVIDERS = [
 	"novita",
 	"nscale",
 	"openai",
+	"ovhcloud",
 	"replicate",
 	"sambanova",
 	"together",
diff --git a/packages/inference/test/InferenceClient.spec.ts b/packages/inference/test/InferenceClient.spec.ts
@@ -1875,7 +1875,109 @@ describe.skip("InferenceClient", () => {
 		},
 		TIMEOUT
 	);
+	describe.concurrent(
+		"OVHcloud",
+		() => {
+			const client = new HfInference(env.HF_OVHCLOUD_KEY ?? "dummy");
+
+			HARDCODED_MODEL_INFERENCE_MAPPING["ovhcloud"] = {
+				"meta-llama/llama-3.1-8b-instruct": {
+					hfModelId: "meta-llama/llama-3.1-8b-instruct",
+					providerId: "Llama-3.1-8B-Instruct",
+					status: "live",
+					task: "conversational",
+				},
+			};
+
+			it("chatCompletion", async () => {
+				const res = await client.chatCompletion({
+					model: "meta-llama/llama-3.1-8b-instruct",
+					provider: "ovhcloud",
+					messages: [{ role: "user", content: "A, B, C, " }],
+					seed: 42,
+					temperature: 0,
+					top_p: 0.01,
+					max_tokens: 1,
+				});
+				expect(res.choices && res.choices.length > 0);
+				const completion = res.choices[0].message?.content;
+				expect(completion).toContain("D");
+			});
+
+			it("chatCompletion stream", async () => {
+				const stream = client.chatCompletionStream({
+					model: "meta-llama/llama-3.1-8b-instruct",
+					provider: "ovhcloud",
+					messages: [{ role: "user", content: "A, B, C, " }],
+					stream: true,
+					seed: 42,
+					temperature: 0,
+					top_p: 0.01,
+					max_tokens: 1,
+				}) as AsyncGenerator<ChatCompletionStreamOutput>;
+
+				let fullResponse = "";
+				for await (const chunk of stream) {
+					if (chunk.choices && chunk.choices.length > 0) {
+						const content = chunk.choices[0].delta?.content;
+						if (content) {
+							fullResponse += content;
+						}
+					}
+				}
 
+				// Verify we got a meaningful response
+				expect(fullResponse).toBeTruthy();
+				expect(fullResponse).toContain("D");
+			});
+
+			it("textGeneration", async () => {
+				const res = await client.textGeneration({
+					model: "meta-llama/llama-3.1-8b-instruct",
+					provider: "ovhcloud",
+					inputs: "A B C ",
+					parameters: {
+						seed: 42,
+						temperature: 0,
+						top_p: 0.01,
+						max_new_tokens: 1,
+					},
+				});
+				expect(res.generated_text.length > 0);
+				expect(res.generated_text).toContain("D");
+			});
+
+			it("textGeneration stream", async () => {
+				const stream = client.textGenerationStream({
+					model: "meta-llama/llama-3.1-8b-instruct",
+					provider: "ovhcloud",
+					inputs: "A B C ",
+					stream: true,
+					parameters: {
+						seed: 42,
+						temperature: 0,
+						top_p: 0.01,
+						max_new_tokens: 1,
+					},
+				}) as AsyncGenerator<ChatCompletionStreamOutput>;
+
+				let fullResponse = "";
+				for await (const chunk of stream) {
+					if (chunk.choices && chunk.choices.length > 0) {
+						const content = chunk.choices[0].text;
+						if (content) {
+							fullResponse += content;
+						}
+					}
+				}
+
+				// Verify we got a meaningful response
+				expect(fullResponse).toBeTruthy();
+				expect(fullResponse).toContain("D");
+			});
+		},
+		TIMEOUT
+	);
 	describe.concurrent(
 		"CentML",
 		() => {
@@ -1958,8 +2060,7 @@ describe.skip("InferenceClient", () => {
 					expect(fullResponse.length).toBeGreaterThan(0);
 				});
 			});
-
 		},
 		TIMEOUT
 	);
-});
+});
diff --git a/packages/tasks/package.json b/packages/tasks/package.json
@@ -1,7 +1,7 @@
 {
 	"name": "@huggingface/tasks",
 	"packageManager": "pnpm@8.10.5",
-	"version": "0.18.11",
+	"version": "0.19.0",
 	"description": "List of ML tasks for huggingface.co/tasks",
 	"repository": "https://github.com/huggingface/huggingface.js.git",
 	"publishConfig": {
diff --git a/packages/tasks/src/inference-providers.ts b/packages/tasks/src/inference-providers.ts
@@ -7,6 +7,7 @@ const INFERENCE_PROVIDERS = [
 	"fireworks-ai",
 	"hf-inference",
 	"hyperbolic",
+	"ovhcloud",
 	"replicate",
 	"sambanova",
 	"together",
diff --git a/packages/tasks/src/model-libraries-snippets.ts b/packages/tasks/src/model-libraries-snippets.ts
diff --git a/packages/tasks/src/tasks/visual-document-retrieval/data.ts b/packages/tasks/src/tasks/visual-document-retrieval/data.ts
