影响力指数
论文质量、代表作、近期表现、广度与样本量置信度综合计算
62.86/100
前 4%
全站排名 #2,606
发表论文9 篇
平均评分
年均产出4.5 篇/年
Jun Suzuki
研究方向
Data Selection · Eficient Transformers · Interpretability and Explainability in NLP · Language Models · Grammatical Error Correction · Neural Encoder-Decoder Models · Neural Word Embeddings · Model Compression and Efficient Models for NLP · Semi-supervised Learning · Machine Translation · Question Answering · Natural Language Processing / Understanding · Kernel Methods
15
Optimal Sparsity of Mixture-of-Experts Language Models for Reasoning Tasks
ICLR 2026Oral
19
Pre-training LLM without Learning Rate Decay Enhances Supervised Fine-Tuning
ICLR 2026Poster
通讯10
Vertical Attention: Automatic Exploration of Inter-Layer Connections in Transformer-based Language Models
ICLR 2026Rejected
16
Transformer Key-Value Memories Are Nearly as Interpretable as Sparse Autoencoders
NeurIPS 2025Poster
二作21
Drop-Upcycling: Training Sparse Mixture of Experts with Partial Re-initialization
ICLR 2025Poster
通讯20
Layerwise Importance Analysis of Feed-Forward Networks in Transformer-based Language Models
COLM 2025Poster
通讯19
Efficient Construction of Model Family through Progressive Training Using Model Expansion
COLM 2025Poster
通讯17
Spike No More: Stabilizing the Pre-training of Large Language Models
COLM 2025Poster
通讯22
Spike No More: Stabilizing the Pre-training of Large Language Models
ICLR 2025Rejected
通讯