面壁智能与清华发布 PostTrainBench：智能体执行强但策略锁定
An agent that can run a full post-training pipeline almost never reconsiders its initial strategy.
面壁智能与清华团队基于 PostTrainBench 分析 1338 条完整训练轨迹，覆盖 7 个基准、4 个基座模型和 Claude Code、Codex CLI、OpenCode，在单张 H100 上运行 10 小时。结果显示智能体平均分随经验从 33.34 提升到 48.55，但 3557 组相邻实验中仅 2.1% 真正切换策略，执行层建议全部被采纳而策略层建议无一被采纳。同一检查点下，一句人类引导切换策略即可超过自主继续，GSM8K 提升 17.44pp、HumanEval 提升 10.80pp、AIME 2025 提升 6.67pp。
正文
An agent that can run a full post-training pipeline almost never reconsiders its initial strategy. We analyze 1,338 complete training trajectories from PostTrainBench — 7 benchmarks, 4 base models, Claude Code / Codex CLI / OpenCode, 10h on one H100 — and split capability into EXECUTION vs STRATEGY.
1️⃣ Qualified executor, locked strategy. Agents improve over base models on every benchmark; with experience, avg score goes 33.34 → 48.55. ALL execution-level suggestions were adopted, but NONE of the strategy-level suggestions (e.g., switching to RL, adding SFT warmup) were adopted. Only 2.1% of 3,557 adjacent experiment pairs actually switch strategy. Defaults follow the agent, not the task: Claude Code starts full-parameter SFT 71.9% of the time; Codex CLI starts PEFT/LoRA 89.5%.
2️⃣ Experience and compute don't unlock it. Extra experience and memory raise execution but never leave the locked strategy; more reasoning tokens buy finer execution rather than better direction — on AIME 2025, ~14M extra tokens solve just 1 more problem.
3️⃣ A mid-run human guidance wins everywhere. At the same checkpoint, same remaining budget, one sentence switching to the alternative strategy beats autonomous continuation: GSM8K +17.44pp, HumanEval +10.80pp, AIME 2025 +6.67pp.
The missing ability isn't executing a new strategy — it's the trigger to spontaneously restart strategy decisions mid-trajectory.
📄 Paper: https://t.co/yuI2gosUXC #AIAgent #PostTraining #LLM #RLHF #GRPO #PPO #RSI #AgentBenchmark #PostTrainBench #THUNLP