BIG-Bench Hard

BIG-Bench Hard (BBH): 23 challenging BIG-Bench tasks

由 23 个 BIG-Bench 任务(6,511 个样例)组成的套件,2022 年的语言模型在这些任务上得分低于人类评分者平均水平:多步算术、逻辑推演、追踪被打乱的物体、因果判断、日期理解等。得分为各任务精确匹配准确率的平均值,通常采用 3-shot 并附带思维链示例。它证明了 CoT 提示能带来涌现式提升,并成为标准推理检验;前沿模型如今在多数任务上已超过最佳人类评分者。

87.5% · 人类 67.7%
DeepSeek-V4-Pro-Base
厂商自报
2022-10-17 → 2026-04-22: 73.9% → 87.5%
发布
2022-10
维护者
Google Research / Stanford (Suzgun et al.)
状态
saturated
污染风险
high
指标
accuracy (percent, ↑)
题量
6,511
领域
reasoning
人类
67.7% average BIG-Bench human rater (max rater: 94.4%)

备注. Answer-only vs CoT prompting changes scores by 10-20 points; the Open LLM Leaderboard v2 reported a 3-shot answer-only variant, so numbers across harnesses are not comparable.

完整账本

系统开发者分数日期来源条件
DeepSeek-V4-Pro-BaseDeepSeek87.5%厂商自报shots: 3
DeepSeek-V4 model card, base-model table (HF repo created 2026-04-22). Base model, exact match, 3-shot; V4-Flash-Base 86.9, V3.2-Base 87.6.
Codex (code-davinci-002), CoTOpenAI (evaluated by Suzgun et al.)73.9%论文shots: 3
Table 2 of the BBH paper, all 23 tasks with chain-of-thought prompting (answer-only 56.6; PaLM 540B CoT 65.2; average human rater 67.7, max 94.4).