Add debug script to check inference snippets + update Python text-to-image snippets

.github/workflows/inference-check-snippets.yml

@@ -0,0 +1,42 @@
+name: Inference check snippets
+on:
+  pull_request:
+    paths:
+      - "packages/tasks/src/snippets/**"
+      - ".github/workflows/inference-check-snippets.yml"
+
+jobs:
+  check-snippets:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v3
+
+      - run: corepack enable
+
+      - uses: actions/setup-node@v3
+        with:
+          node-version: "20"
+          cache: "pnpm"
+          cache-dependency-path: "**/pnpm-lock.yaml"
+      - run: |
+          cd packages/tasks
+          pnpm install
+
+      # TODO: Find a way to run on all pipeline tags
+      # TODO: print snippet only if it has changed since the last commit on main (?)
+      # TODO: (even better: automated message on the PR with diff)
+      - name: Print text-to-image snippets
+        run: |
+          cd packages/tasks
+          pnpm run check-snippets --pipeline-tag="text-to-image"
+
+      - name: Print simple text-generation snippets
+        run: |
+          cd packages/tasks
+          pnpm run check-snippets --pipeline-tag="text-generation"
+
+      - name: Print conversational text-generation snippets
+        run: |
+          cd packages/tasks
+          pnpm run check-snippets --pipeline-tag="text-generation" --tags="conversational"

packages/tasks/package.json

@@ -32,7 +32,8 @@
 		"check": "tsc",
 		"inference-codegen": "tsx scripts/inference-codegen.ts && prettier --write src/tasks/*/inference.ts",
 		"inference-tgi-import": "tsx scripts/inference-tgi-import.ts && prettier --write src/tasks/text-generation/spec/*.json && prettier --write src/tasks/chat-completion/spec/*.json",
-		"inference-tei-import": "tsx scripts/inference-tei-import.ts && prettier --write src/tasks/feature-extraction/spec/*.json"
+		"inference-tei-import": "tsx scripts/inference-tei-import.ts && prettier --write src/tasks/feature-extraction/spec/*.json",
+		"check-snippets": "tsx scripts/check-snippets.ts"
 	},
 	"type": "module",
 	"files": [

packages/tasks/scripts/check-snippets.ts

@@ -0,0 +1,55 @@
+/*
+ * Generates inference snippets as they would be shown on the Hub for Curl, JS and Python.
+ * Snippets will only be printed to the terminal to make it easier to debug when making changes to the snippets.
+ *
+ * Usage:
+ *   pnpm run check-snippets --pipeline-tag="text-generation" --tags="conversational"
+ *   pnpm run check-snippets --pipeline-tag="image-text-to-text" --tags="conversational"
+ *   pnpm run check-snippets --pipeline-tag="text-to-image"
+ *
+ * This script is meant only for debug purposes.
+ */
+import { python, curl, js } from "../src/snippets/index";
+import type { InferenceSnippet, ModelDataMinimal } from "../src/snippets/types";
+import type { PipelineType } from "../src/pipelines";
+
+// Parse command-line arguments
+const args = process.argv.slice(2).reduce(
+	(acc, arg) => {
+		const [key, value] = arg.split("=");
+		acc[key.replace("--", "")] = value;
+		return acc;
+	},
+	{} as { [key: string]: string }
+);
+
+const accessToken = "hf_**********";
+const pipelineTag = (args["pipeline-tag"] || "text-generation") as PipelineType;
+const tags = (args["tags"] || "").split(",");
+
+const modelMinimal: ModelDataMinimal = {
+	id: "llama-6-1720B-Instruct",
+	pipeline_tag: pipelineTag,
+	tags: tags,
+	inference: "****",
+};
+
+const printSnippets = (snippets: InferenceSnippet | InferenceSnippet[], language: string) => {
+	const snippetArray = Array.isArray(snippets) ? snippets : [snippets];
+	snippetArray.forEach((snippet) => {
+		console.log(`\n\x1b[33m${language} ${snippet.client}\x1b[0m`);
+		console.log(`\n\`\`\`${language === "JS" ? "js" : language.toLowerCase()}\n${snippet.content}\n\`\`\`\n`);
+	});
+};
+
+const generateAndPrintSnippets = (
+	generator: (model: ModelDataMinimal, token: string) => InferenceSnippet | InferenceSnippet[],
+	language: string
+) => {
+	const snippets = generator(modelMinimal, accessToken);
+	printSnippets(snippets, language);
+};
+
+generateAndPrintSnippets(curl.getCurlInferenceSnippet, "Curl");
+generateAndPrintSnippets(python.getPythonInferenceSnippet, "Python");
+generateAndPrintSnippets(js.getJsInferenceSnippet, "JS");

packages/tasks/src/snippets/python.ts

@@ -4,6 +4,11 @@ import { stringifyGenerationConfig, stringifyMessages } from "./common.js";
 import { getModelInputSnippet } from "./inputs.js";
 import type { InferenceSnippet, ModelDataMinimal } from "./types.js";
 
+const snippetImportInferenceClient = (model: ModelDataMinimal, accessToken: string): string =>
+	`from huggingface_hub import InferenceClient
+
+client = InferenceClient(${model.id}, token="${accessToken || "{API_TOKEN}"}")`;
+
 export const snippetConversational = (
 	model: ModelDataMinimal,
 	accessToken: string,
@@ -168,18 +173,31 @@ export const snippetFile = (model: ModelDataMinimal): InferenceSnippet => ({
 output = query(${getModelInputSnippet(model)})`,
 });
 
-export const snippetTextToImage = (model: ModelDataMinimal): InferenceSnippet => ({
-	content: `def query(payload):
+export const snippetTextToImage = (model: ModelDataMinimal, accessToken: string): InferenceSnippet[] => {
+	return [
+		{
+			client: "requests",
+			content: `def query(payload):
 	response = requests.post(API_URL, headers=headers, json=payload)
 	return response.content
+
 image_bytes = query({
 	"inputs": ${getModelInputSnippet(model)},
 })
 # You can access the image with PIL.Image for example
 import io
 from PIL import Image
 image = Image.open(io.BytesIO(image_bytes))`,
-});
+		},
+		{
+			client: "huggingface_hub",
+			content: `${snippetImportInferenceClient(model, accessToken)}
+
+# output is a PIL.Image object
+image = client.text_to_image(${getModelInputSnippet(model)})`,
+		},
+	];
+};
 
 export const snippetTabular = (model: ModelDataMinimal): InferenceSnippet => ({
 	content: `def query(payload):
@@ -284,6 +302,9 @@ export function getPythonInferenceSnippet(
 	if (model.tags.includes("conversational")) {
 		// Conversational model detected, so we display a code snippet that features the Messages API
 		return snippetConversational(model, accessToken, opts);
+	} else if (model.pipeline_tag == "text-to-image") {
+		// TODO: factorize this logic
+		return snippetTextToImage(model, accessToken);
 	} else {
 		let snippets =
 			model.pipeline_tag && model.pipeline_tag in pythonSnippets