We provide a comprehensive review of the FRUIT (Faithfully Reflecting Updated Information in Text) task, which formalizes the challenge of accurately updating textual information with large language models (LLMs). Our work begins with an in-depth analysis of the FRUIT dataset, revealing key structural insights. We also investigate the unsupervised capabilities of LLMs—such as zero-shot learning, chain-of-thought reasoning, self-reflection, and evidence ordering. Experimental results demonstrate that unsupervised approaches perform competitively with supervised methods in faithful text updating. Qualitative analysis shows that updates utilizing table-structured evidence outperform those based on unstructured text. We also discuss important limitations, including the need for new datasets and the risks of information leakage in this domain. These findings have significant implications for applications requiring precise document updates, such as software engineering, technical documentation, and legal document maintenance.

Large Language Models often achieve strong performance by generating long intermediate chain-of-thought (CoT) reasoning. However, it remains unclear when a model’s final answer is actually determined during generation. If the answer is already fixed at an intermediate stage, subsequent reasoning tokens may constitute post-decision explanation, increasing inference cost and latency without improving correctness. We study the evolution of predicted answers over reasoning steps using *forced answer completion*, which elicits a model’s intermediate prediction at partial reasoning prefixes. Across a range of models and datasets, we find that the final answer is fixed from the earliest stages of generation in most cases, with **only 32% of answers changing after CoT**. Moreover, once the final answer switch does occur, models continue to generate substantial reasoning, producing **760 additional tokens per query** on average and accounting for around **40% of the total reasoning**. Motivated by these findings, we explore early-stopping strategies that halt generation once the answer has stabilized. We show that simple probe-based stopping can reduce reasoning token usage by **500 tokens per query** while incurring only a **2% drop in accuracy**. Together, our results indicate that a large portion of chain-of-thought generation functions as **post hoc explanation** and can be reduced with minimal impact on performance.

We propose a novel approach for machine-generated text detection using a RoBERTa model with weighted layer averaging and AdaLoRA for parameter-efficient fine-tuning. Our method incorporates information from all model layers, capturing diverse linguistic cues beyond those accessible from the final layer alone. To mitigate potential overfitting and improve generalizability, we leverage AdaLoRA, which injects trainable low-rank matrices into each Transformer layer, significantly reducing the number of trainable parameters. Furthermore, we employ data mixing to ensure our model encounters text from various domains and generators during training, enhancing its ability to generalize to unseen data. This work highlights the potential of combining layer-wise information with parameter-efficient fine-tuning and data mixing for effective machine-generated text detection.