PaperHub

Pang Wei Koh

~Pang_Wei_Koh1

29
论文总数
14.5
年均投稿
6.7
平均评分
接收情况23/29
会议分布
ICLR
10
COLM
8
NeurIPS
8
ICML
3

发表论文 (29 篇)

202520 篇

4.5
4

On Erroneous Agreements of CLIP Image Embeddings

ICLR 2025Rejected
6.3
4

JPEG-LM: LLMs as Image Generators with Canonical Codec Representations

ICLR 2025Rejected
6.6
4

S4S: Solving for a Fast Diffusion Model Solver

ICML 2025Poster
8.8
4

EvalTree: Profiling Language Model Weaknesses via Hierarchical Capability Trees

COLM 2025Poster
7.2
4

NICE Data Selection for Instruction Tuning in LLMs with Non-differentiable Evaluation Metric

ICML 2025Poster
6.0
4

Group-robust Sample Reweighting for Subpopulation Shifts via Influence Functions

ICLR 2025Poster
5.7
3

ParaPO: Aligning Language Models to Reduce Verbatim Reproduction of Pre-training Data

COLM 2025Poster
4.8
4

Conformal Reasoning: Uncertainty Estimation in Interactive Environments

ICLR 2025Rejected
4.0
4

MoSH: Modeling Multi-Objective Tradeoffs with Soft and Hard Bounds

ICLR 2025Rejected
8.0
4

The Delta Learning Hypothesis: Preference Tuning on Weak Data can Yield Strong Gains

COLM 2025Poster
7.7
3

Fluid Language Model Benchmarking

COLM 2025Poster
5.8
4

A False Sense of Privacy: Evaluating Textual Data Sanitization Beyond Surface-level Privacy Leakage

ICLR 2025Rejected
6.8
4

Precise Information Control in Long-Form Text Generation

NeurIPS 2025Poster
6.3
4

Establishing Task Scaling Laws via Compute-Efficient Model Ladders

COLM 2025Poster
7.0
4

ReasonIR: Training Retrievers for Reasoning Tasks

COLM 2025Poster
6.3
3

DataDecide: How to Predict Best Pretraining Data with Small Experiments

ICML 2025Poster
6.5
4

Language models scale reliably with over-training and on downstream tasks

ICLR 2025Poster
9.1
4

FlexOLMo: Open Language Models for Flexible Data Use

NeurIPS 2025Spotlight
8.7
3

OLMoE: Open Mixture-of-Experts Language Models

ICLR 2025Oral
7.0
4

2 OLMo 2 Furious (COLM’s Version)

COLM 2025Poster

20249 篇

6.7
6

Improving Domain Generalization with Domain Relations

ICLR 2024Spotlight
6.3
4

MediQ: Question-Asking LLMs and a Benchmark for Reliable Interactive Clinical Reasoning

NeurIPS 2024Poster
8.3
4

Information-Theoretic Distillation for Reference-less Summarization

COLM 2024Poster
6.0
4

The Unmet Promise of Synthetic Training Images: Using Retrieved Real Images Performs Better

NeurIPS 2024Poster
7.0
4

Multilingual Diversity Improves Vision-Language Representations

NeurIPS 2024Spotlight
7.2
5

Uncertainty of Thoughts: Uncertainty-Aware Planning Enhances Information Seeking in LLMs

NeurIPS 2024Poster
6.3
6

Scaling Retrieval-Based Language Models with a Trillion-Token Datastore

NeurIPS 2024Poster
7.0
4

The Generative AI Paradox: “What It Can Create, It May Not Understand”

ICLR 2024Poster
5.7
3

Language models scale reliably with over-training and on downstream tasks

NeurIPS 2024Rejected
合作者 (20)