dsh-livebench-panel 0.2.7 → 0.2.9

package/lib/client.js

@@ -197,7 +197,14 @@ window.__ModuleLoader__.load({
 					const toggleList = (key, value) => setSel((prev) => {
 						const list = prev[key];
 						const next = list.includes(value) ? list.filter((v) => v !== value) : [...list, value];
-						return { ...prev, [key]: next };
+						const out = { ...prev, [key]: next };
+						if (key === "cats") {
+							// 取消分类时同步清除该分类下已选的任务：否则这些任务标签
+							// 会从界面消失（分类不再选中）却仍在提交列表里 —— 用户只选了
+							// coding_completion 却连带跑了 cta 就是这个原因。
+							out.tasks = prev.tasks.filter((t) => next.includes(t.split("/")[0]));
+						}
+						return out;
 					});
 
 					// 已选模型解析（provider::model）

package/lib/index.js

@@ -691,7 +691,12 @@ function apply(ctx) {
 			? null
 			: reasoningEffort;
 		const hasEffortSuffix = effort !== null && effort !== "off";
-		const displayName = displayModelName(providerId || "direct", modelId) + (hasEffortSuffix ? "@" + effort : "");
+		// 每次评测使用带时间戳的独立 display-name：答案/判分文件各自独立，
+		// 成绩表里每次评测是独立一行，可对比模型的历史变化（不再覆盖旧数据）。
+		const _now = new Date();
+		const _pad = (n) => String(n).padStart(2, "0");
+		const runStamp = `r${_now.getFullYear()}${_pad(_now.getMonth() + 1)}${_pad(_now.getDate())}-${_pad(_now.getHours())}${_pad(_now.getMinutes())}${_pad(_now.getSeconds())}`;
+		const displayName = displayModelName(providerId || "direct", modelId) + (hasEffortSuffix ? "@" + effort : "") + "__" + runStamp;
 		let cliModel = modelId;
 		let writeError = null;		// Anthropic-protocol proxies cannot go through --api-base (that path
 		// speaks OpenAI Chat Completions). Instead the generated model config

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-livebench-panel",
-  "version": "0.2.7",
+  "version": "0.2.9",
   "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",