Pre-trained models have demonstrated exceptional generalization capabilities in time-series forecasting; however, adapting them to evolving data distributions remains a significant challenge. A key hurdle lies in accessing the original training data, as fine-tuning solely on new data often leads to catastrophic forgetting. To address this issue, we propose Replay Tuning (R-Tuning), a novel framework designed for the continual adaptation of pre-trained time-series models. R-Tuning constructs a unified latent space that captures both prior and current task knowledge through a frequency-aware replay strategy. Specifically, it augments model-generated samples via wavelet-based decomposition across multiple frequency bands, generating trend-preserving and fusion-enhanced variants to improve representation diversity and replay efficiency. To further reduce reliance on synthetic samples, R-Tuning introduces a latent consistency constraint that aligns new representations with the prior task space. This constraint guides joint optimization within a compact and semantically coherent latent space, ensuring robust knowledge retention and adaptation. Extensive experimental results demonstrate the superiority of R-Tuning, which reduces MAE and MSE by up to 46.9% and 46.8%, respectively, on new tasks, while preserving prior knowledge with gains of up to 5.7% and 6.0% on old tasks. Notably, under few-shot settings, R-Tuning outperforms all state-of-the-art baselines even when synthetic proxy samples account for only 5% of the new task dataset.

Can Large Language Models (LLMs) be trained to avoid hallucinating factual statements, and can Retrieval-Augmented Generation (RAG) be triggered only when necessary to reduce retrieval and computation costs? In this work, we address both challenges simultaneously. We introduce ConfQA, a fine-tuning strategy that reduces hallucination rates from 20-40% to below 5% across multiple factuality benchmarks. The approach is simple: when the model answers correctly, it is trained to output the answer; otherwise, it is trained to respond with "I am unsure". Two design choices make this training effective: (1) a dampening prompt ("answer only if you are confident") that explicitly discourages overconfident hallucinations, and (2) training data drawn from atomic factual statements (e.g., knowledge graph attribute values), which calibrates model confidence and yields robust generalization across domains and question types. Building on ConfQA, we propose ConfRAG, a triggering strategy that invokes RAG only when the model responses with unsure. This framework achieves accuracy above 95% in ideal case while reducing unnecessary external retrievals by over 30%.

Hallucinations is a major challenge for large language models (LLMs), prevents adoption in diverse fields. Uncertainty estimation could be used for alleviating the damages of hallucinations. The skeptical emotion of human could be useful for enhancing the ability of self estimation. Inspirited by this observation, we proposed a new approach called Skepticism Modeling (SM). This approach is formalized by combining the information of token and logits for self estimation. We construct the doubt emotion aware data, perform continual pre-training, and then fine-tune the LLMs, improve their ability of self estimation. Experimental results demonstrate this new approach effectively enhances a model's ability to estimate their uncertainty, and validate its generalization ability of other tasks by out-of-domain experiments.

Large language models (LLMs) have revolutionized numerous domains with their impressive performance but still face their challenges. A predominant issue is the propensity for these models to generate non-existent facts, a concern termed hallucination. Our research is motivated by the observation that previous instruction tuning methods force the model to complete a sentence no matter whether the model knows the knowledge or not. When the question is out of the parametric knowledge, it will try to make up something and fail to indicate when it lacks knowledge. In this paper, we present a new approach called Refusal-Aware Instruction Tuning (R-Tuning). This approach is formalized by first identifying the knowledge gap between parametric knowledge and the instruction tuning data. Then, we construct the refusal-aware data based on the knowledge intersection, to tune LLMs to refrain from responding to questions beyond its parametric knowledge. Experimental results demonstrate this new instruction tuning approach effectively improves a model's ability to answer known questions and refrain from answering unknown questions. Furthermore, when tested on out-of-domain datasets, the refusal ability was found to be a meta-skill that could be generalized to other tasks. Further analysis surprisingly finds that learning the uncertainty during training displays a better ability to estimate uncertainty than uncertainty-based testing. Our code will be released at https://github.com/shizhediao/R-Tuning.