dsh-livebench-panel 0.2.0 → 0.2.2

package/lib/client.js

@@ -489,7 +489,10 @@ window.__ModuleLoader__.load({
 									disabled: busy || !config?.available || sel.models.length === 0,
 								}, `开始评测（${sel.models.length} 个模型并发）`),
 								running && h("button", { className: c("btn") + " " + c("btnDanger") + " " + c("btnBar"), onClick: onStop }, "停止全部"),
-								h("button", { className: c("btnGhost") + " " + c("btn") + " " + c("btnBar"), onClick: () => { loadConfig(); loadResults(); } }, "刷新"),
+								h("button", { className: c("btnGhost") + " " + c("btn") + " " + c("btnBar"), onClick: async () => {
+									try { await api("/clear", { method: "POST" }); } catch { /* ignore */ }
+									loadConfig(); loadResults(); refreshStatus();
+								} }, "刷新"),
 							),
 							startError !== null && h("p", { className: c("error") }, startError),
 						),

package/lib/index.js

@@ -103,7 +103,7 @@ const RELEASES = [
 /** Sane cap so a runaway run cannot eat memory with its log. */
 const LOG_MAX_LINES = 600;
 /** One evaluation at a time per model, but several models may run concurrently. */
-const MAX_CONCURRENT_RUNS = 6;
+const MAX_CONCURRENT_RUNS = 9;
 
 /** Resolve the profile directory from the config-tree anchor (plugin-market pattern). */
 function resolveProfileDir(ctx) {
@@ -657,6 +657,13 @@ function apply(ctx) {
 			...process.env,
 			HF_HOME: join(layout.root, ".hf_cache"),
 			HF_HUB_DISABLE_SYMLINKS_WARNING: "1",
+			// 限制评测进程的数值库线程（numpy/pyarrow/MKL 默认吃满全核，
+			// 多模型并发时会把 node 事件循环饿死 → websocket 断连）。
+			OMP_NUM_THREADS: "1",
+			MKL_NUM_THREADS: "1",
+			OPENBLAS_NUM_THREADS: "1",
+			NUMEXPR_NUM_THREADS: "1",
+			RAYON_NUM_THREADS: "1",
 			// run_livebench.py shells out to bare `python`; without the venv on
 			// PATH that resolves to the system interpreter and dies on the first
 			// livebench import (shortuuid etc.).
@@ -694,6 +701,15 @@ function apply(ctx) {
 			env,
 			windowsHide: true,
 		});
+		// 降低评测进程优先级：评测不应与 DSH/浏览器争抢 CPU
+		try {
+			if (child.pid) {
+				spawn("powershell", [
+					"-NoProfile", "-Command",
+					`(Get-Process -Id ${child.pid}).PriorityClass = 'BelowNormal'`,
+				], { windowsHide: true, stdio: "ignore" }).on("error", () => {});
+			}
+		} catch { /* 优先级设置失败不影响评测 */ }
 
 		const record = {
 			id: `${Date.now()}-${Math.random().toString(36).slice(2, 6)}`,
@@ -875,6 +891,26 @@ function apply(ctx) {
 		},
 	}), `${name}: stop route`);
 
+	ctx.effect(() => ctx.webServer.register({
+		kind: "exact",
+		path: `${API}/clear`,
+		handler: async (req, res) => {
+			if (req.method !== "POST") {
+				sendJson(res, 405, { ok: false, error: "method not allowed" });
+				return;
+			}
+			if (!originAllowed(req)) {
+				sendJson(res, 403, { ok: false, error: "forbidden origin" });
+				return;
+			}
+			let removed = 0;
+			for (const [id, r] of [...runs]) {
+				if (r.exitCode !== null) { runs.delete(id); removed += 1; }
+			}
+			sendJson(res, 200, { ok: true, removed });
+		},
+	}), `${name}: clear route`);
+
 	ctx.effect(() => ctx.webServer.register({
 		kind: "exact",
 		path: `${API}/results`,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-livebench-panel",
-  "version": "0.2.0",
+  "version": "0.2.2",
   "description": "DSH web plugin: a LiveBench tab in the Trajectory view (right of 对话/轨迹). Run LiveBench evaluations against every model configured in the DeepSeek Harness — pick provider/model, category, task, release and question range from dropdowns, watch progress, and read scores in place.",
   "license": "MIT",
   "type": "module",