Amirkeivan Mohtashami
7 papers · 2021–2025 · 4 conferences · across top CS/AI conferences
Achievements
Jump to papers ↓+2 more ↓ Show less ↑
πΊοΈ Taxonomy Completionist (12) π£ Hot Topic Early Bird π Conference Polyglot (4) π§ Keyword Pioneer π Cross-Pollinator (10)
π
Interdisciplinary Bridge
π₯
Unstoppable
(5)
Conferences
NIPS (3)
AISTATS (2)
ICLR (1)
ICML (1)
Top co-authors
Keywords
stochastic gradient descent
(3)
memory efficiency
(2)
model compression
(2)
non-convex optimization
(1)
attention mechanism
(1)
distributed learning
(1)
neural network optimization
(1)
asynchronous optimization
(1)
learning rate
(1)
context length
(1)
global minimum
(1)
information flow
(1)
batch size
(1)
gradient staleness
(1)
large language model
(1)
neural network
(1)
asynchronous update
(1)
partial gradient
(1)
communication efficient training
(1)
landmark token
(1)
Papers
CoTFormer: A Chain of Thought Driven Architecture with Budget-Adaptive Computation Cost at Inference
ICLR 2025
QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs
NIPS 2024
DenseFormer: Enhancing Information Flow in Transformers via Depth Weighted Averaging
NIPS 2024
Special Properties of Gradient Descent with Large Learning Rates
ICML 2023
Random-Access Infinite Context Length for Transformers
NIPS 2023
Masked Training of Neural Networks with Partial Gradients
AISTATS 2022
Critical Parameters for Scalable Distributed Learning with Large Batches and Asynchronous Updates
AISTATS 2021