Papers by Shell Xu Hu
Distilling Knowledge from Text-to-Image Generative Models Improves Visio-Linguistic Reasoning in CLIP (2024.emnlp-main)
Copied to clipboard
| Challenge: | Image-text contrastive models like CLIP struggle on compositional visio-linguistic tasks where their performance is no better than random chance. |
| Approach: | They propose a distillation method to enhance CLIP's compositional visio-linguistic reasoning by using a model-derived distillation objective borrowed from large text-to-image generative models like Stable-Diffusion. |
| Outcome: | The proposed method improves CLIP models' visio-linguistic performance on the Winoground benchmark by 7% while on the ARO dataset, it boosts performance by 3%. |
MobileQuant: Mobile-friendly Quantization for On-device Language Models (2024.findings-emnlp)
Copied to clipboard
Fuwen Tan, Royson Lee, Łukasz Dudziak, Shell Xu Hu, Sourav Bhattacharya, Timothy Hospedales, Georgios Tzimiropoulos, Brais Martinez
| Challenge: | Large language models (LLMs) have revolutionized language processing, but deployment on edge devices is costly in terms of memory, computation and energy. |
| Approach: | They propose to reduce the number of bits used to represent weights and activations . they propose to use 8-bit activations to enable LLMs to fully exploit mobile-friendly hardware . |
| Outcome: | The proposed method reduces the number of bits used to represent weights and activations . 8-bit activations are attractive for on-device deployment as they would exploit mobile-friendly hardware . |