Under identical LoRA hyperparameters and training data, the 27B Gemma model outperforms the 70B LLaMA model on every computed essay-scoring metric¶

Claim · All claims
Evidence · 1 study · causal r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? With identical training data (120 essays) and identical LoRA hyperparameters, Gemma-3-27B achieves lower RMSE (0.474 vs. 0.512), lower MAE (0.335 vs. 0.379), higher QWK (0.828 vs. 0.777), and higher ±0.5 agreement (90.56% vs. 87.22%) than LLaMA-3.3-70B. → John Maurice Gayed 2026

Evidence¶

John Maurice Gayed 2026¶

John Maurice Gayed. (2026). AiAWE: An Open-Source LLM Automated Writing Evaluation System Using LoRA-Adapted Instruction-Tuned Models. A PREPRINT. https://arxiv.org/abs/2606.12801 doi:10.48550/arxiv.2606.12801

q2 · i? · causal · r2

Head-to-head comparison of the two LoRA-adapted open-weight models on the 360-essay held-out test set, both evaluated under the same Q4_K_M quantisation. The article reports that "the 27B-parameter Gemma model outperforms the 70B-parameter LLaMA model across every metric we computed."

"Under identical training data (120 essays) and identical LoRA hyperparameters, the 27B-parameter Gemma model outperforms the 70B-parameter LLaMA model across every metric we computed: lower RMSE (0.474 vs. 0.512), lower MAE (0.335 vs. 0.379), higher QWK (0.828 vs. 0.777), and higher±0.5 agreement (90.56% vs. 87.22%)."

Discussion¶

Related Claims¶

• A LoRA-adapted open-weight Gemma-3-27B model fine-tuned on 120 essays outperforms a fine-tuned GPT-3.5 baseline on TOEFL essay scoring across all reported metrics — related
• LLaMA's LoRA adaptation compresses its prediction range to roughly 2.0–4.0, while Gemma maintains a broader distribution closer to the ground truth — related
• LoRA fine-tuning hyperparameters are not model-agnostic: increasing rank above 64 destroyed LLaMA's feedback generation while Gemma remained robust — related
• Both fine-tuned models show regression toward the mean in score-wise bias, and error rises monotonically above score 3.0, with high scores hardest to predict — related
• CDPK performance scales with model size, with a sharp Pareto-frontier drop-off below around 8B parameters — related
• LoRA adaptation adds only 0.6M trainable parameters on an 86.4M frozen backbone, keeping methods lightweight relative to LLM-based scoring — related
• Model family and instruction-tuning approach appear better predictors of tutoring quality than parameter count alone — a broader claim this one bears on
• Model size is not the primary factor in outcome-prediction performance — related
• Among the three 8B LLMs, Llama 3 underperforms Apertus and Granite 3.3 on most evaluation metrics — related
• Self-hosted open-weight LLMs offer privacy, reproducibility, and environmental advantages over commercial LLM interfaces in educational ITSs — related
• Fine-tuned GPT-4o achieves state-of-the-art scoring of TOEFL argumentative essays with QWK 0.84 and RMSE 0.44 against official scores — related
• Adaptation gains grow monotonically with model capacity under the same LoRA recipe, though three data points cannot establish a scaling law — related
• Carbon per Accuracy as a sustainability metric likely rewards under-trained models because accuracy scales non-linearly with compute — related