arXiv:2601. 22146v2 Announce Type: replace-cross Abstract: Due to limited supervised training data, large language models (LLMs) are typically pre-trained via a self-supervised "predict the next word" objective on a vast amount of unstructured text data.
arXiv:2604. 13977v2 Announce Type: replace-cross Abstract: Synthetic data is a standard component in training large language models, yet systematic comparisons across design dimensions, including rephrasing strategy, generator model, and source data, remain absent.
The paper introduces DKL, a method for adding new knowledge to instruction‑tuned language models without compromising their instruction‑following abilities. DKL performs extended pre‑training on a base LLM to embed knowledge, then merges these weights into the instruction‑tuned model, avoiding costly instruction fine‑tuning. Experiments show DKL raises RAG accuracy from 54.17% to 79.26% on retrieval failure cases while using far less training data than previous approaches.
arXiv:2609.37891v1 Announce Type: cross Abstract: Current pre-training datasets are derived from web crawls, with all their issues, and were not designed to support mid- and post-training pipelines--...
The paper introduces DKL, a method that decouples knowledge learning from instruction tuning in language models. Instead of fine‑tuning the instruction‑tuned model directly, DKL first extends pre‑training on a base model to embed new knowledge, then merges these weights into the instruction‑tuned model, preserving its instruction‑following abilities. Experiments show DKL raises RAG accuracy from 54.17 % to 79.26 % on retrieval failures, outperforming prior methods while using far less training data.
arXiv:2608. 25826v1 Announce Type: cross Abstract: A recent line of synthetic-data work reconstructs the thinking behind existing text rather than rewriting the text itself, but it operates on short web passages, recovers only local thoughts, and leaves the structure of whole documents untouched.
A recent line of synthetic-data work reconstructs the thinking behind existing text rather than rewriting the text itself, but it operates on short web passages, recovers only local thoughts, and leav...
RePro is a web‑recycling technique that trains a small language model (as little as 1 B parameters) with reinforcement learning to produce high‑quality, faithful rephrasings of pretraining data. The method uses one quality reward and three faithfulness rewards to preserve core semantics and structure while converting organic data into better training examples. Experiments show that RePro boosts downstream accuracy by 3.7–14.5 % over organic‑only baselines and improves data efficiency 2–3×, outperforming prior prompting‑based recycling approaches.
arXiv:2607. 08646v1 Announce Type: cross Abstract: As available training data approaches its physical limit, gains from Scaling Laws have begun to diminish.
arXiv:2608. 20281v1 Announce Type: cross Abstract: Large language models often fail to answer questions about a bounded document collection when the source documents are not retrieved at inference time.
The paper introduces CoDIT, a contrastive decoding technique that separates instruction-following behavior from pre-trained world knowledge in large language models. By generating responses that emphasize post-training instruction capabilities while suppressing shared pre-trained knowledge, CoDIT creates instruction-tuning datasets that lead to consistently better model performance than directly generated responses or existing public datasets. The authors also provide theoretical and empirical evidence that CoDIT effectively distills instruction-tuning knowledge from model parameters into text, facilitating cross-architecture transfer.
arXiv:2608.30627v1 Announce Type: new Abstract: As language-model compute continues to scale, high-quality training data is becoming an increasingly important bottleneck. Conventional next-token pred...
arXiv:2606. 24841v1 Announce Type: new Abstract: Prompt-based learning has emerged as a dominant paradigm in natural language processing.