LoRA adaptation adds only 0.6M trainable parameters on an 86.4M frozen backbone, keeping methods lightweight relative to LLM-based scoring¶

Claim · All claims
Evidence · 1 study · design 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? LoRA updates only a small subset of parameters: 0.6M trainable parameters on the fixed 86.4M ViT backbone, while confidence-aware variants add no further trainable parameters. → Fang 2026

Evidence¶

Fang 2026¶

Fang, L., Zhang, Y., Park, J., Wang, Z., Ma, P., Zhai, X. (2026). Confidence-Aware Automated Assessment of Student-Drawn Scientific Models. https://arxiv.org/abs/2606.20264 doi:10.48550/arxiv.2606.20264

q2 · i? · design · r2

Model complexity analysis (Table 3): all approaches share the same 86.4M ViT backbone, so "performance differences are not attributable to model capacity"; confidence-aware scoring raises inference latency to about 20.5 ms versus about 1.03 ms.

"LoRA introduces only 0.6M additional trainable parameters while keeping the backbone fixed, and the confidence-aware variants add no further trainable parameters."

Discussion¶

Related Claims¶

• Under identical LoRA hyperparameters and training data, the 27B Gemma model outperforms the 70B LLaMA model on every computed essay-scoring metric — related
• 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
• LoRA fine-tuning hyperparameters are not model-agnostic: increasing rank above 64 destroyed LLaMA's feedback generation while Gemma remained robust — 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
• Confidence-aware selective test-time scoring achieves the best average agreement with expert rubric scoring across six NGSS drawing items — related