Picture for Cyril Zhang

Cyril Zhang

Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone

Apr 23, 2024
Viaarxiv icon

Can large language models explore in-context?

Mar 22, 2024
Figure 1 for Can large language models explore in-context?
Figure 2 for Can large language models explore in-context?
Figure 3 for Can large language models explore in-context?
Figure 4 for Can large language models explore in-context?
Viaarxiv icon

Butterfly Effects of SGD Noise: Error Amplification in Behavior Cloning and Autoregression

Add code
Oct 17, 2023
Viaarxiv icon

Pareto Frontiers in Neural Feature Learning: Data, Compute, Width, and Luck

Sep 07, 2023
Figure 1 for Pareto Frontiers in Neural Feature Learning: Data, Compute, Width, and Luck
Figure 2 for Pareto Frontiers in Neural Feature Learning: Data, Compute, Width, and Luck
Figure 3 for Pareto Frontiers in Neural Feature Learning: Data, Compute, Width, and Luck
Figure 4 for Pareto Frontiers in Neural Feature Learning: Data, Compute, Width, and Luck
Viaarxiv icon

Exposing Attention Glitches with Flip-Flop Language Modeling

Add code
Jun 01, 2023
Figure 1 for Exposing Attention Glitches with Flip-Flop Language Modeling
Figure 2 for Exposing Attention Glitches with Flip-Flop Language Modeling
Figure 3 for Exposing Attention Glitches with Flip-Flop Language Modeling
Figure 4 for Exposing Attention Glitches with Flip-Flop Language Modeling
Viaarxiv icon

Learning Hidden Markov Models Using Conditional Samples

Feb 28, 2023
Figure 1 for Learning Hidden Markov Models Using Conditional Samples
Figure 2 for Learning Hidden Markov Models Using Conditional Samples
Viaarxiv icon

Neural Active Learning on Heteroskedastic Distributions

Add code
Nov 02, 2022
Figure 1 for Neural Active Learning on Heteroskedastic Distributions
Figure 2 for Neural Active Learning on Heteroskedastic Distributions
Figure 3 for Neural Active Learning on Heteroskedastic Distributions
Figure 4 for Neural Active Learning on Heteroskedastic Distributions
Viaarxiv icon

Transformers Learn Shortcuts to Automata

Add code
Oct 19, 2022
Viaarxiv icon

Recurrent Convolutional Neural Networks Learn Succinct Learning Algorithms

Sep 01, 2022
Figure 1 for Recurrent Convolutional Neural Networks Learn Succinct Learning Algorithms
Figure 2 for Recurrent Convolutional Neural Networks Learn Succinct Learning Algorithms
Figure 3 for Recurrent Convolutional Neural Networks Learn Succinct Learning Algorithms
Viaarxiv icon

Hidden Progress in Deep Learning: SGD Learns Parities Near the Computational Limit

Jul 18, 2022
Figure 1 for Hidden Progress in Deep Learning: SGD Learns Parities Near the Computational Limit
Figure 2 for Hidden Progress in Deep Learning: SGD Learns Parities Near the Computational Limit
Figure 3 for Hidden Progress in Deep Learning: SGD Learns Parities Near the Computational Limit
Figure 4 for Hidden Progress in Deep Learning: SGD Learns Parities Near the Computational Limit
Viaarxiv icon