Product
Help
Pricing
About
Jobs
Log in
Get started
Author
Add to collection
Mansheej Paul
Observed name:
Paul, Mansheej
Institution:
Databricks (United States)
Observed institutions:
Mosaic,
Databricks (United States),
Stanford University
OpenAlex ID:
A5036307641
Deep Learning on a Data Diet: Finding Important Examples Early in Training
2021
·
Paul, Mansheej
, Ganguli, Surya
, et al.
·
arXiv (Cornell University)
·
133
PDF
Deep learning versus kernel learning: an empirical study of loss landscape geometry and the time evolution of the Neural Tangent Kernel
2020
·
Fort, Stanislav
, Dziugaite, Gintare Karolina
, et al.
·
arXiv (Cornell University)
·
41
PDF
LoRA Learns Less and Forgets Less
2024
·
Biderman, Dan
, Portes, Jacob
, et al.
·
arXiv (Cornell University)
·
33
PDF
Pretraining task diversity and the emergence of non-Bayesian in-context learning for regression
2023
·
Allan Raventós
, Mansheej Paul
, et al.
·
neural information processing systems
·
13
PDF
Lottery Tickets on a Data Diet: Finding Initializations with Sparse Trainable Networks
2022
·
Mansheej Paul
, Brett Larsen
, et al.
·
neural information processing systems
·
8
PDF
Perplexed by Perplexity: Perplexity-Based Data Pruning With Small Reference Models
2024
·
Ankner, Zachary
, Blakeney, Cody
, et al.
·
arXiv (Cornell University)
·
5
PDF
Unmasking the Lottery Ticket Hypothesis: What's Encoded in a Winning Ticket's Mask?
2022
·
Paul, Mansheej
, Chen, Feng
, et al.
·
arXiv (Cornell University)
·
3
PDF
Scaling Laws for Precision
2024
·
Kumar, Tanishq
, Ankner, Zachary
, et al.
·
arXiv (Cornell University)
·
2
PDF
Critique-out-Loud Reward Models
2024
·
Ankner, Zachary
, Paul, Mansheej
, et al.
·
arXiv (Cornell University)
·
1
PDF
$μ$nit Scaling: Simple and Scalable FP8 LLM Training
2025
·
Narayan, Saaketh
, Gupta, Abhay
, et al.
·
arXiv (Cornell University)
PDF
Forget About It: Mitigating Forgetting Without Replay or Auxiliary Objectives
2025
·
Kleiman, Anat
, Dziugaite, Gintare Karolina
, et al.
·
arXiv (Cornell University)
PDF
Does your data spark joy? Performance gains from domain upsampling at the end of training
2024
·
Blakeney, Cody
, Paul, Mansheej
, et al.
·
arXiv (Cornell University)
PDF
Show more works
View as search
Works count:
12
Citation count:
239
H-index:
5
I10-index:
4
Year
Topic
Type
