Knowledge Tree
Relations (1)
related 3.70 — strongly supporting 12 facts
Exposure bias is a fundamental structural issue in Large Language Models that causes hallucinations by creating a mismatch between training and inference distributions, as described in [1] and [2]. This phenomenon leads to compounding errors during generation [3] and remains a persistent challenge that is not fully resolved by techniques like retrieval augmentation or RLHF [4], [5].
Facts (12)
Sources
Hallucination Causes: Why Language Models Fabricate Facts mbrenndoerfer.com 12 facts
claimExposure bias in large language models creates compounding errors where a small factual inaccuracy or semantic drift at a specific position changes the conditioning context for subsequent positions, leading the model to generate statistically likely continuations based on erroneous premises.
claimExposure bias is not unique to large language models; it arises in any sequence-to-sequence system trained with teacher forcing, including neural machine translation systems from the pre-transformer era.
claimExposure bias in large language models is the discrepancy between the distribution of conditioning contexts seen during training, which uses ground-truth tokens via teacher forcing, and the distribution seen during inference, which uses model-generated tokens.
claimBenchmarks for large language models that test only high-frequency factual questions fail to reveal tail entity hallucination, and benchmarks that test only short responses fail to reveal exposure bias accumulation.
claimExposure bias is a cause of hallucination in large language models that arises from a mismatch between training efficiency and inference realism.
claimExposure bias in large language models does not require the model to lack the correct answer; rather, hallucinations arise because an error changes the input distribution, activating incorrect associations despite the model potentially possessing reliable knowledge.
claimUncertainty calibration through Reinforcement Learning from Human Feedback (RLHF) addresses the surface expression of completion pressure in large language models but does not change the underlying lack of a world model or the exposure bias structure.
claimThe max_new_tokens parameter controls sequence length in large language models, and longer generations face higher cumulative exposure bias divergence, which increases hallucination risk as the sequence grows.
claimRetrieval augmentation, which adds external knowledge at inference time, addresses knowledge gaps in large language models but leaves exposure bias and completion pressure untouched.
claimExposure bias in large language models causes errors to be self-reinforcing, where each subsequent token conditions on an initial incorrect context rather than the ground truth.
claimLarge language models exhibit a 3% floor of irreducible hallucination even at high training frequencies, which is caused by exposure bias, completion pressure, and conflicting signals in training data.
claimFinetuning large language models modifies the model's response style regarding expressed confidence, but the underlying knowledge gaps and exposure bias patterns remain encoded in the base model from pretraining.