Jiaming Shen
YOU?
Author Swipe
View article: IC-Cache: Efficient Large Language Model Serving via In-context Caching
IC-Cache: Efficient Large Language Model Serving via In-context Caching Open
Large language models (LLMs) have excelled in various applications, yet serving them at scale is challenging due to their substantial resource demands and high latency. Our real-world studies reveal that over 70% of user requests to LLMs h…
View article: HeMeNet: Heterogeneous Multichannel Equivariant Network for Protein Multi-task Learning
HeMeNet: Heterogeneous Multichannel Equivariant Network for Protein Multi-task Learning Open
Understanding and leveraging the 3D structures of proteins is central to a variety of biological and drug discovery tasks. While deep learning has been applied successfully for structure-based protein function prediction tasks, current met…
View article: Integrating Planning into Single-Turn Long-Form Text Generation
Integrating Planning into Single-Turn Long-Form Text Generation Open
Generating high-quality, in-depth textual documents, such as academic papers, news articles, Wikipedia entries, and books, remains a significant challenge for Large Language Models (LLMs). In this paper, we propose to use planning to gener…
View article: RRM: Robust Reward Model Training Mitigates Reward Hacking
RRM: Robust Reward Model Training Mitigates Reward Hacking Open
Reward models (RMs) play a pivotal role in aligning large language models (LLMs) with human preferences. However, traditional RM training, which relies on response pairs tied to specific prompts, struggles to disentangle prompt-driven pref…
View article: Building Math Agents with Multi-Turn Iterative Preference Learning
Building Math Agents with Multi-Turn Iterative Preference Learning Open
Recent studies have shown that large language models' (LLMs) mathematical problem-solving capabilities can be enhanced by integrating external tools, such as code interpreters, and employing multi-turn Chain-of-Thought (CoT) reasoning. Whi…
View article: Multi-omics association study identifies new potential prostate cancer- causing gene
Multi-omics association study identifies new potential prostate cancer- causing gene Open
BACKGROUND Prostate cancer is one of the most common malignant tumors among men worldwide, and surgery remains its mainstay of treatment. It is unclear how prostate cancer develops and what the most effective drug targets are for treating …
View article: Knowledge Distillation with Perturbed Loss: From a Vanilla Teacher to a Proxy Teacher
Knowledge Distillation with Perturbed Loss: From a Vanilla Teacher to a Proxy Teacher Open
Knowledge distillation is a popular technique to transfer knowledge from a large teacher model to a small student model. Typically, the student learns to imitate the teacher by minimizing the KL divergence of its output distribution with t…
View article: LAMPO: Large Language Models as Preference Machines for Few-shot Ordinal Classification
LAMPO: Large Language Models as Preference Machines for Few-shot Ordinal Classification Open
We introduce LAMPO, a novel paradigm that leverages Large Language Models (LLMs) for solving few-shot multi-class ordinal classification tasks. Unlike conventional methods, which concatenate all demonstration examples with the test instanc…
View article: Multilingual Fine-Grained News Headline Hallucination Detection
Multilingual Fine-Grained News Headline Hallucination Detection Open
The popularity of automated news headline generation has surged with advancements in pre-trained language models. However, these models often suffer from the ``hallucination'' problem, where the generated headline is not fully supported by…
View article: PLaD: Preference-based Large Language Model Distillation with Pseudo-Preference Pairs
PLaD: Preference-based Large Language Model Distillation with Pseudo-Preference Pairs Open
Large Language Models (LLMs) have exhibited impressive capabilities in various tasks, yet their vast parameter sizes restrict their applicability in resource-constrained settings. Knowledge distillation (KD) offers a viable solution by tra…
View article: HeMeNet: Heterogeneous Multichannel Equivariant Network for Protein Multitask Learning
HeMeNet: Heterogeneous Multichannel Equivariant Network for Protein Multitask Learning Open
Understanding and leveraging the 3D structures of proteins is central to a variety of biological and drug discovery tasks. While deep learning has been applied successfully for structure-based protein function prediction tasks, current met…
View article: Impact of COVID‐19 Nonpharmaceutical Interventions on Respiratory Syncytial Virus Infections in Hospitalized Children
Impact of COVID‐19 Nonpharmaceutical Interventions on Respiratory Syncytial Virus Infections in Hospitalized Children Open
Background Nonpharmaceutical interventions (NPIs) targeted at SARS‐CoV‐2 have remarkably affected the circulation of other respiratory pathogens, including respiratory syncytial virus (RSV). This study aimed to assess the changes in epidem…
View article: 17β-Estradiol, through activating the G protein-coupled estrogen receptor, exacerbates the complication of benign prostatic hyperplasia in type 2 diabetes mellitus patients by inducing prostate proliferation
17β-Estradiol, through activating the G protein-coupled estrogen receptor, exacerbates the complication of benign prostatic hyperplasia in type 2 diabetes mellitus patients by inducing prostate proliferation Open
Benign prostatic hyperplasia (BPH) is one of the major chronic complications of type 2 diabetes mellitus (T2DM), and sex steroid hormones are common risk factors for the occurrence of T2DM and BPH. The profiles of sex steroid hormones are …
View article: TELEClass: Taxonomy Enrichment and LLM-Enhanced Hierarchical Text Classification with Minimal Supervision
TELEClass: Taxonomy Enrichment and LLM-Enhanced Hierarchical Text Classification with Minimal Supervision Open
Hierarchical text classification aims to categorize each document into a set of classes in a label taxonomy, which is a fundamental web text mining task with broad applications such as web content analysis and semantic indexing. Most earli…
View article: LiPO: Listwise Preference Optimization through Learning-to-Rank
LiPO: Listwise Preference Optimization through Learning-to-Rank Open
Aligning language models (LMs) with curated human feedback is critical to control their behaviors in real-world applications. Several recent policy optimization methods, such as DPO and SLiC, serve as promising alternatives to the traditio…
View article: Liver Injury and Its Impact on Prognosis in Patients with HBV-Related Hepatocellular Carcinoma Undergoing Transarterial Chemoembolization Combined with Tyrosine Kinase Inhibitors Plus Immune Checkpoint Inhibitors
Liver Injury and Its Impact on Prognosis in Patients with HBV-Related Hepatocellular Carcinoma Undergoing Transarterial Chemoembolization Combined with Tyrosine Kinase Inhibitors Plus Immune Checkpoint Inhibitors Open
Liver injury adverse events are common in HBV-related HCC patients treated with TACE-TKIs-ICIs. Patients who developed liver injury had a poor prognosis. For HBV-related HCC patients, effective prophylactic antiviral therapy and regular li…
View article: Removal of Expression of Concern: Sodium butyrate ameliorated diabetic nephropathy-associated tubulointerstitial inflammation by modulating tight junction of renal tubular epithelial cells
Removal of Expression of Concern: Sodium butyrate ameliorated diabetic nephropathy-associated tubulointerstitial inflammation by modulating tight junction of renal tubular epithelial cells Open
Removal of Expression of Concern for ‘Sodium butyrate ameliorated diabetic nephropathy-associated tubulointerstitial inflammation by modulating tight junction of renal tubular epithelial cells’ by Tingting Yang et al. , Food Funct. , 2022,…
View article: Predicting Text Preference Via Structured Comparative Reasoning
Predicting Text Preference Via Structured Comparative Reasoning Open
Comparative reasoning plays a crucial role in text preference prediction; however, large language models (LLMs) often demonstrate inconsistencies in their reasoning. While approaches like Chain-of-Thought improve accuracy in many other set…
View article: Explanation-aware Soft Ensemble Empowers Large Language Model In-context Learning
Explanation-aware Soft Ensemble Empowers Large Language Model In-context Learning Open
Large language models (LLMs) have shown remarkable capabilities in various natural language understanding tasks. With only a few demonstration examples, these LLMs can quickly adapt to target tasks without expensive gradient updates. Commo…
View article: FCA-Net: A Fast Inference and Channel Attention Based Network for Hyperspectral Image Classification
FCA-Net: A Fast Inference and Channel Attention Based Network for Hyperspectral Image Classification Open
Hyperspectral imaging (HSI) is a competitive remote sensing technique used in various fields such as land cover mapping and environmental monitoring. Each hyperspectral imaging (HSI) scene is comprised of numerous narrow and contiguous spe…
View article: Local Boosting for Weakly-Supervised Learning
Local Boosting for Weakly-Supervised Learning Open
Boosting is a commonly used technique to enhance the performance of a set of\nbase models by combining them into a strong ensemble model. Though widely\nadopted, boosting is typically used in supervised learning where the data is\nlabeled …
View article: HiPrompt: Few-Shot Biomedical Knowledge Fusion via Hierarchy-Oriented Prompting
HiPrompt: Few-Shot Biomedical Knowledge Fusion via Hierarchy-Oriented Prompting Open
Medical decision-making processes can be enhanced by comprehensive biomedical knowledge bases, which require fusing knowledge graphs constructed from different sources via a uniform index system. The index system often organizes biomedical…
View article: Large Language Models are Effective Text Rankers with Pairwise Ranking Prompting
Large Language Models are Effective Text Rankers with Pairwise Ranking Prompting Open
Ranking documents using Large Language Models (LLMs) by directly feeding the query and candidate documents into the prompt is an interesting and practical problem. However, researchers have found it difficult to outperform fine-tuned basel…