Add support for HF summarization endpoint in the websearch (#319)

src/lib/server/websearch/summarizeWeb.ts

@@ -1,7 +1,28 @@
+import { HF_ACCESS_TOKEN } from "$env/static/private";
+import { HfInference } from "@huggingface/inference";
 import { generateFromDefaultEndpoint } from "../generateFromDefaultEndpoint";
 import type { BackendModel } from "../models";
 
 export async function summarizeWeb(content: string, query: string, model: BackendModel) {
+	// if HF_ACCESS_TOKEN is set, we use a HF dedicated endpoint for summarization
+	try {
+		if (HF_ACCESS_TOKEN) {
+			const summary = (
+				await new HfInference(HF_ACCESS_TOKEN).summarization({
+					model: "facebook/bart-large-cnn",
+					inputs: content,
+					parameters: {
+						max_length: 512,
+					},
+				})
+			).summary_text;
+			return summary;
+		}
+	} catch (e) {
+		console.log(e);
+	}
+
+	// else we use the LLM to generate a summary
 	const summaryPrompt =
 		model.userMessageToken +
 		content