NLP, Machine Learning

insight - NLP, Machine Learning

QASE Enhanced PLMs: Improved Control in Text Generation for MRC

QASE module enhances generative PLMs for better MRC performance.

KeNet: Knowledge-Enhanced Doc-Label Attention Network for Multi-Label Text Classification

외부 지식을 활용한 KeNet 모델은 다중 레이블 텍스트 분류에서 우수한 성능을 보입니다.

IMPOSSIBLE DISTILLATION: Paraphrasing and Summarization Framework

IMPOSSIBLE DISTILLATION distills high-quality paraphrase datasets and models from low-quality LMs using paraphrastic proximity and critic-guided filtering.

QASE Enhanced PLMs: Improved Control in Text Generation for MRC

Generative PLMs can be enhanced with the lightweight QASE module to improve text generation quality and factual consistency in MRC tasks.

MiniLLM: Knowledge Distillation of Large Language Models

Reverse KLD is proposed for distilling LLM knowledge into smaller models, improving performance and reducing exposure bias.

SOCIALITE-LLAMA: An Instruction-Tuned Model for Social Scientific Tasks

Instruction tuning improves the social understanding of large language models for social scientific tasks.

BeLLM: Backward Dependency Enhanced Large Language Model for Sentence Embeddings

Incorporating backward dependencies in large language models enhances sentence embeddings.

Hyper-CL: Conditioning Sentence Representations with Hypernetworks

Efficiently conditioning sentence representations using hypernetworks and contrastive learning.

LoRA-SP: Streamlined Partial Parameter Adaptation for Efficient Fine-Tuning of Large Language Models

Selective parameter freezing in LoRA-SP optimizes fine-tuning efficiency without compromising model performance.

Knowledge-Enhanced Doc-Label Attention Network for Multi-Label Text Classification

The author introduces KeNet, a Knowledge-enhanced Doc-Label Attention Network, to address challenges in Multi-Label Text Classification by incorporating external knowledge and attention mechanisms.

About

Products | Resources

Insights