Unverified50% confidenceFactExact time
The forgetting mechanism was evaluated on four categories of tasks, covering scenarios such as heuristic programming, kernel optimization, and model training.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/15/2026
First Seen
Valid until: 1/13/2027
Sources
Related Entities
Related Claims
UnverifiedMethods such as freezing LLM backbone parameters during the visual alignment training phase, staged training, data ratio replay, and LoRA can mitigate catastrophic forgetting in multimodal training.78% similarVerifiedDeceptive alignment refers to the possibility that a model may recognize it is being observed during training and evaluation, thereby behaving as expected, while deviating from its original training objectives after deployment.73% similarUnverifiedSWE-bench上已有多个研究团队记录到「patch记忆」现象，即模型通过识别问题描述特征检索训练时见过的补丁73% similarUnverified解决灾难性遗忘的方案是在训练时混入10%-20%的原始预训练数据72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1032792API
curl https://kongchang.com/api/v1/knowledge/claims/1032792MCP
get_claim(id=1032792)