English(EN) Measured independently, not self-reported: Qwen3.5 2B (Non-reasoning) hits 43.8% GPQA, but only 5% HLE, 15% Long Context Reasoning, and 7.2% SciCode. The collap

Qwen3.5 2B在独立基准测试中表现参差不齐

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-02 19:00

独立基准测试显示，Qwen3.5 2B（一个非推理模型）在GPQA基准测试中取得了43.8%的成绩。然而，在更复杂的任务上，其表现显著下降，在HLE上仅得5%，在长上下文推理上仅得15%，在SciCode上仅得7.2%。这表明在具有挑战性的评估中存在显著的性能差距。 AI

影响突出了小型开源模型在复杂推理任务上的性能局限性。

排序理由开源模型的独立基准测试结果。[lever_c_demoted from research: ic=1 ai=1.0]

在 Mastodon — mastodon.social 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 3 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 开源模型的独立基准测试结果。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: model release, paper; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Same-day; Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

Mastodon — mastodon.social TIER_1 English(EN) · opensourceaitech · 2026-09-02 19:00

独立测量，非自报：Qwen3.5 2B（非推理）GPQA 达 43.8%，但 HLE 仅 5%，长上下文推理仅 15%，SciCode 仅 7.2%。崩

Measured independently, not self-reported: Qwen3.5 2B (Non-reasoning) hits 43.8% GPQA, but only 5% HLE, 15% Long Context Reasoning, and 7.2% SciCode. The collapse on harder tests is the real story. See the full data: https:// olud.ai/leaderboard.html # LLM # Benchmarks # OpenSour…

链接 olud.ai/leaderboard.html

本文如何被排名

报道来源 [1]

独立测量，非自报：Qwen3.5 2B（非推理）GPQA 达 43.8%，但 HLE 仅 5%，长上下文推理仅 15%，SciCode 仅 7.2%。崩

相关实体

相关话题