Unverified50% confidenceFactExact time
Autoregressive models suffer from the exposure bias problem, whereby the model sees real prefixes during training but faces self-generated erroneous prefixes during inference, causing errors to accumulate progressively.
1
Sources
50%
Confidence
Long-term
Relevance
9/9/2026
First Seen
Sources
Related Entities
Related Claims
Unverified提示注入利用的是模型无法可靠区分「数据」与「指令」的天然缺陷，这一缺陷根植于LLM的训练机制本身79% similarUnverified蒸馏训练范式导致开源模型学习到经过高度过滤、几乎不包含失败和纠错过程的数据分布，使模型形成不愿自我纠正的性格77% similarVerifiedIf the synthetic data generation process omits correlations between key variables, models trained on such data often perform poorly in real-world deployment environments.76% similarUnverifiedBehavioral Cloning, as the most fundamental imitation learning method, suffers from the compounding error problem, where models tend to produce cascading failures on states outside the training distribution.76% similarUnverifiedModel poisoning involves implanting backdoors or biases during the training phase of an AI model, causing the model to produce erroneous judgments under specific conditions.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/885773API
curl https://kongchang.com/api/v1/knowledge/claims/885773MCP
get_claim(id=885773)