WaltonFuture/vscode-bench-results
VSCode Bench Results 图片生成与编辑 / Image generation and editing 14 组本地原生图像模型 + 1 组 coding agent(gpt-6.1-sol / high),10 套 benchmark,每组固定 539 道图片题。 共 8,074 条有效评分、9 条内容过滤跳过、2 条源答案缺失;跳过和缺失不补零。 共同题目对比:纯文本分数表 · 宽表 CSV 本地模型原始汇总 · Coding agent 原始汇总 全部逐题结果 Parquet · 主要指标 Parquet 范围、口径、字段与审计说明 共同题目对比统一使用全部 15 组都有有效评分的 534 道题。各指标按 0–100 展示,保留原评分规则,不计算跨 benchmark 总分。上方数据查看器的 image_metrics 配置提供指标表,image_scores 提供逐题数据。
| Author / organization | WaltonFuture |
|---|---|
| License (as declared) | mit |
| Downloads | 1.2K (as reported by Hugging Face) |
| Likes | 0 |
| Created | 14 October 2026 |
| Last updated at source | 15 October 2026 |
Popularity, last 2 days
1,189 now (+7 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
More from WaltonFuture
- WaltonFuture/DrivingVQADataset
- WaltonFuture/Multimodal-Cold-StartDataset
- WaltonFuture/SpatialVQADataset
- WaltonFuture/VQA-MultiTaskDataset
- WaltonFuture/clevr-mathDataset
- WaltonFuture/REC-VQADataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.