Unverified50% confidenceFactExact time
针对工具调用的强化学习训练通常采用执行反馈机制，将实际执行的成功/失败信号作为奖励，可大规模自动化
1
Sources
50%
Confidence
Long-term
Relevance
7/7/2026
First Seen
Sources
Related Claims
Unverified行为克隆预训练解决了强化学习的冷启动问题：纯强化学习在稀疏奖励环境中往往需要数亿次交互才能学到基本可用策略74% similarVerifiedIf reinforcement learning rewards are based solely on whether a call executes successfully rather than on whether it conforms to general schema specifications, the model will learn shortcut strategies targeting specific tools, which is known as reward hacking.73% similarUnverifiedSelf-supervised learning is the core paradigm of the pre-training phase, where the model constructs training signals from the data itself through a next-token prediction task.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/179001API
curl https://kongchang.com/api/v1/knowledge/claims/179001MCP
get_claim(id=179001)