Picture for Michael Y. Hu

Michael Y. Hu

Between Circuits and Chomsky: Pre-pretraining on Formal Languages Imparts Linguistic Biases

Add code
Feb 26, 2025
Viaarxiv icon

Findings of the Second BabyLM Challenge: Sample-Efficient Pretraining on Developmentally Plausible Corpora

Add code
Dec 06, 2024
Viaarxiv icon

Aioli: A Unified Optimization Framework for Language Model Data Mixing

Add code
Nov 08, 2024
Viaarxiv icon

[Call for Papers] The 2nd BabyLM Challenge: Sample-efficient pretraining on a developmentally plausible corpus

Add code
Apr 09, 2024
Figure 1 for [Call for Papers] The 2nd BabyLM Challenge: Sample-efficient pretraining on a developmentally plausible corpus
Viaarxiv icon

Comparing Abstraction in Humans and Large Language Models Using Multimodal Serial Reproduction

Add code
Feb 06, 2024
Figure 1 for Comparing Abstraction in Humans and Large Language Models Using Multimodal Serial Reproduction
Figure 2 for Comparing Abstraction in Humans and Large Language Models Using Multimodal Serial Reproduction
Figure 3 for Comparing Abstraction in Humans and Large Language Models Using Multimodal Serial Reproduction
Figure 4 for Comparing Abstraction in Humans and Large Language Models Using Multimodal Serial Reproduction
Viaarxiv icon

Latent State Models of Training Dynamics

Add code
Aug 18, 2023
Figure 1 for Latent State Models of Training Dynamics
Figure 2 for Latent State Models of Training Dynamics
Figure 3 for Latent State Models of Training Dynamics
Figure 4 for Latent State Models of Training Dynamics
Viaarxiv icon

Using Natural Language and Program Abstractions to Instill Human Inductive Biases in Machines

Add code
May 23, 2022
Figure 1 for Using Natural Language and Program Abstractions to Instill Human Inductive Biases in Machines
Figure 2 for Using Natural Language and Program Abstractions to Instill Human Inductive Biases in Machines
Figure 3 for Using Natural Language and Program Abstractions to Instill Human Inductive Biases in Machines
Figure 4 for Using Natural Language and Program Abstractions to Instill Human Inductive Biases in Machines
Viaarxiv icon