Sequential Fine-Tuning Approach Improves Automated Essay Scoring by Modeling Discourse Structure
Researchers found that fine-tuning LLaMA-3.1-8B sequentially—training it on essay discourse elements in a structured order—outperforms both independent task-specific models and a general-purpose LLaMA-70B baseline on key scoring tasks. The study used the PERSUADE 2.0 corpus and parameter-efficient LoRA with 4-bit quantization to compare three training curricula: sequential, independent, and randomized. The findings suggest that curriculum design aligned with discourse structure can enable smaller, cost-effective models to compete with much larger language models in educational assessment.
A study submitted to arXiv investigates how training order affects the performance of LLaMA-3.1-8B on Automated Essay Scoring (AES), a task requiring evaluation of interdependent discourse elements such as leads, claims, evidence, and conclusions. The researchers compared three fine-tuning approaches—sequential (training on discourse elements in a logical order), independent (separate task-specific models), and randomized (shuffled multi-task training)—using parameter-efficient LoRA with 4-bit quantization on the PERSUADE 2.0 corpus. Sequential fine-tuning achieved the strongest overall results, including F1 scores of 65% for evidence and 87% for conclusion, and outperformed a general-purpose LLaMA-70B model on conclusion scoring despite having nearly nine times fewer parameters. Randomized training showed some benefit for position scoring (57% F1) but was inconsistent across other tasks. The authors argue that aligning curriculum design with the inherent structure of discourse can materially improve AES performance and offer a practical, scalable path for educational NLP applications. Code templates and implementation details are being released to support reproducibility and further research.
What's missing
It is unclear whether the sequential ordering tested reflects the only effective ordering or merely one of several possible beneficial orderings, as ablations over alternative sequences are not reported. Human rater agreement baselines are not provided, making it difficult to contextualize the absolute F1 and accuracy figures.
What different sources said
- arXiv cs.LGCenter
The Order Matters: Sequential Fine-Tuning of LLaMA for Coherent Automated Essay Scoring
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.