Selected work

Representative Papers

Echo-LoRA: Parameter-Efficient Fine-Tuning via Cross-Layer Representation Injection

May 05, 2026

Existing LoRA-based methods are constrained to single-layer weight updates, limiting their ability to leverage intermediate representations from deep networks and thereby hindering the performance of parameter-efficient fine-tuning. This work proposes Echo-LoRA, which introduces a cross-layer representation injection mechanism: sample-specific echo representations are generated by aggregating hidden states at deep layer boundaries and then injected into shallow LoRA or DoRA modules via a lightweight projection and gating network. During training, answer masking, masked distillation, and stochastic routing are jointly employed to stabilize the auxiliary pathway and reduce the training–inference discrepancy; notably, the auxiliary structure can be entirely removed at inference time, incurring no additional overhead. Evaluated on eight commonsense reasoning benchmarks, Echo-LoRA achieves an average improvement of 5.7 percentage points over LoRA (3.0 points under unified reimplementation) and gains 2.7 points when combined with DoRA.

0 citationsRead paper

Dual-view Spatio-Temporal Feature Fusion with CNN-Transformer Hybrid Network for Chinese Isolated Sign Language Recognition

Jun 08, 2025

To address two key bottlenecks in isolated sign language recognition (ISLR)—incomplete vocabulary coverage and single-view hand occlusion—this work introduces NationalCSL-DP, the first large-scale bilingual (front + left) dataset covering the entire lexicon of China’s national sign language. We propose a hybrid CNN-Transformer architecture coupled with a lightweight, highly efficient dual-view fusion strategy, incorporating spatiotemporal feature alignment, early/late fusion, and concatenation mechanisms to overcome the limitations of sequential models in capturing cross-view complementary information. Experiments on NationalCSL-DP demonstrate that our method significantly outperforms single-view baselines and various seq2seq-based fusion approaches. The results validate both the effectiveness and generalizability of dual-view data acquisition and our minimalist fusion design, establishing a new benchmark for ISLR research.

0 citationsRead paper

Recent publications

Latest Papers

Echo-LoRA: Parameter-Efficient Fine-Tuning via Cross-Layer Representation Injection

May 05, 2026

Existing LoRA-based methods are constrained to single-layer weight updates, limiting their ability to leverage intermediate representations from deep networks and thereby hindering the performance of parameter-efficient fine-tuning. This work proposes Echo-LoRA, which introduces a cross-layer representation injection mechanism: sample-specific echo representations are generated by aggregating hidden states at deep layer boundaries and then injected into shallow LoRA or DoRA modules via a lightweight projection and gating network. During training, answer masking, masked distillation, and stochastic routing are jointly employed to stabilize the auxiliary pathway and reduce the training–inference discrepancy; notably, the auxiliary structure can be entirely removed at inference time, incurring no additional overhead. Evaluated on eight commonsense reasoning benchmarks, Echo-LoRA achieves an average improvement of 5.7 percentage points over LoRA (3.0 points under unified reimplementation) and gains 2.7 points when combined with DoRA.

0 citationsRead paper

Dual-view Spatio-Temporal Feature Fusion with CNN-Transformer Hybrid Network for Chinese Isolated Sign Language Recognition

Jun 08, 2025

To address two key bottlenecks in isolated sign language recognition (ISLR)—incomplete vocabulary coverage and single-view hand occlusion—this work introduces NationalCSL-DP, the first large-scale bilingual (front + left) dataset covering the entire lexicon of China’s national sign language. We propose a hybrid CNN-Transformer architecture coupled with a lightweight, highly efficient dual-view fusion strategy, incorporating spatiotemporal feature alignment, early/late fusion, and concatenation mechanisms to overcome the limitations of sequential models in capturing cross-view complementary information. Experiments on NationalCSL-DP demonstrate that our method significantly outperforms single-view baselines and various seq2seq-based fusion approaches. The results validate both the effectiveness and generalizability of dual-view data acquisition and our minimalist fusion design, establishing a new benchmark for ISLR research.

0 citationsRead paper