影响力指数

69.88/100

前 2.7%

全站排名 #1,718

发表论文15 篇

平均评分5.3

年均产出5.0 篇/年

Leshem Choshen

Postdoc@Massachusetts Institute of Technology·美国·OpenReview

研究方向

NLP · language · open · Games · Scaling laws · loss spaces · model merging · recycling · Generalization · OOD · Evaluation · sentence semantics · translation · grammatical error correction · learner language · UCCA · argumentation · reinforcement learning

5.0

Beyond Binary Rewards: Training LMs to Reason About Their Uncertainty

ICLR 2026Poster

4.0

ErrorMap and ErrorAtlas: Charting the Failure Landscape of Large Language Models

ICLR 2026Rejected

通讯

3.5

Resolving Interference (RI): Disentangling Models for Improved Model Merging

ICLR 2026Rejected

2.5

Stop Guessing When to Stop Testing: Efficient Model Evaluation with Just Enough Data

ICLR 2026Withdrawn

通讯

6.8

Sloth: scaling laws for LLM skills to predict multi-benchmark performance across families

NeurIPS 2025Poster

三作

6.6

Sloth: scaling laws for LLM skills to predict multi-benchmark performance across families

合作者 (20)

Leshem Choshen

Beyond Binary Rewards: Training LMs to Reason About Their Uncertainty

ErrorMap and ErrorAtlas: Charting the Failure Landscape of Large Language Models

Resolving Interference (RI): Disentangling Models for Improved Model Merging

Stop Guessing When to Stop Testing: Efficient Model Evaluation with Just Enough Data

Sloth: scaling laws for LLM skills to predict multi-benchmark performance across families

Compress then Serve: Serving Thousands of LoRA Adapters with Little Overhead

Model merging with SVD to tie the Knots

A Hitchhiker's Guide to Scaling Law Estimation

LiveXiv - A Multi-Modal live benchmark based on Arxiv papers content

Compress then Serve: Serving Thousands of LoRA Adapters with Little Overhead

A Hitchhiker's Guide to Scaling Law Estimation

Sloth: scaling laws for LLM skills to predict multi-benchmark performance across families