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
55.93/100
前 6.1%
全站排名 #3,930
发表论文10 篇
平均评分
年均产出3.3 篇/年
Nikunj Saunshi
研究方向
self-supervised learning · language modeling · meta-learning · contrastive learning · representation learning
17
Efficient stagewise pretraining via progressive subnetworks
ICLR 2025Poster
二作9
StagFormer: Time Staggering Decoder only Transformers
COLM 2025Poster
10
Learning to Keep a Promise: Scaling Language Model Decoding Parallelism with Learned Asynchronous Decoding
ICML 2025Poster
16
Reasoning with Latent Thoughts: On the Power of Looped Transformers
ICLR 2025Poster
一作26
A Little Help Goes a Long Way: Efficient LLM Training by Leveraging Small LMs
ICLR 2025Rejected
15
On the Role of Depth and Looping for In-Context Learning with Task Diversity
ICLR 2025Rejected
二作13
StagFormer: A Staggered Transformer for Decoding Layers in Parallel
ICLR 2025Rejected