Large language models, despite their strong performance, frequently produce hallucinated content due to excessive reliance on pre-trained knowledge while insufficiently integrating newly provided context. We introduce LACD, a technique that dynamically rebalances probability distributions across layers, ensuring critical context is not overshadowed. By emphasizing new prompt information, LACD alleviates lower-layer dominance and mitigates hallucinations. On the HotPotQA dataset, LACD outperforms basic context injection baselines by approximately 2.2% in exact match (EM) and matches or exceeds advanced methods such as DoLa and CAD. LACD also demonstrates robust gains on SQuAD, underscoring its capacity to reduce hallucinations while improving factual consistency. Overall, these findings highlight the importance of carefully integrating newly provided context with pre-trained knowledge to achieve more reliable language generation.

Large language models, despite their strong performance, frequently produce hallucinated content by excessively relying on pre-trained knowledge and overlooking newly provided prompts. We introduce LACD, a technique that dynamically rebalances probability distributions across layers, ensuring critical context is not overshadowed. By emphasizing new prompt information, LACD alleviates lower-layer dominance and mitigates hallucinations. On the HotPotQA dataset, LACD surpasses naive context augmentation by roughly 2.2% in EM, matching or outperforming advanced approaches such as DoLa and CAD. LACD also demonstrates robust gains on SQuAD, underscoring its capacity to reduce hallucinations while improving factual consistency. Overall, these findings highlight the importance of carefully integrating newly provided context with pre-trained knowledge to achieve more reliable language generation.