Researchers Develop Method to Mine General Data for Improving Language Models in Specialized Domains
Researchers have proposed Hierarchical Active Region Pruning (HARP), a data selection method for fine-tuning large language models that reduces the number of required training examples by roughly seven times. HARP bridges the gap between computationally cheap but imprecise train-free selectors and accurate but costly train-based selectors by organizing training data into a hierarchy and using empirical Bayes inference to estimate unmeasured data utility. The method could significantly lower the cost and resource requirements of adapting large language models to specific tasks.
Fine-tuning large language models (LLMs) for specific downstream tasks requires carefully selecting which training examples to use, a process that traditionally forces a tradeoff between scalability and accuracy. Existing train-free approaches rely on proxies like embedding similarity or clustering that may not reflect actual task performance, while train-based methods are more accurate but demand many expensive training and evaluation cycles. HARP addresses this by organizing the training pool into a node-leaf hierarchy, evaluating only representative data subsets, and inferring the utility of unevaluated data using empirical Bayes posteriors. The framework offers two complementary selection strategies: HARP-C, which conservatively reduces redundancy, and HARP-E, which rewards data from complementary regions. Theoretical guarantees show that HARP controls selection error under local smoothness and bounded estimation error assumptions. Empirically, HARP variants outperform the strongest baseline by up to 8.9 points while using approximately seven times fewer training examples, suggesting meaningful practical gains in fine-tuning efficiency.
What different sources said
- arXiv cs.LGCenter
Rethinking Local Learning: A Cheaper and Faster Recipe for LLM Post-Training
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.