Unverified50% confidenceFactExact time
Warmup策略在训练初期使用极小学习率，待参数进入合理范围后再提升，被大型语言模型训练广泛采用
1
Sources
50%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
梯度下降与反向传播详解：神经网络学习机制从零拆解
Related Claims
UnverifiedIn transfer learning, there are typically two strategies: freezing most layers of the pretrained model and only training the classification head (Feature Extraction), or initializing with pretrained weights and fine-tuning all layers with a smaller learning rate (Fine-Tuning).78% similarUnverified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76% similarUnverifiedSimply stacking parameters without sufficient training data results in a model that is large but not capable; a leap in parameters must be accompanied by a proportionally expanded training corpus and computing power to realize capability gains.76% similarUnverifiedTraditional large model training relies heavily on manually annotated data and external corpus collection, which are typically the bottlenecks constraining iteration speed.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/558140API
curl https://kongchang.com/api/v1/knowledge/claims/558140MCP
get_claim(id=558140)