7 3 1

Yongchang Hao

yongchanghao

https://yongchanghao.github.io

AI & ML interests

None yet

Recent Activity

upvoted a paper 3 days ago

Qwen2.5 Technical Report

reacted to their post with 🔥 about 2 months ago

We just released a paper (NeuZip) that compresses VRAM in a lossless manner to run larger models. This should be particularly useful when VRAM is insufficient during training/inference. Specifically, we look inside each floating number and find that the exponents are highly compressible (as shown in the figure below). Read more about the work at https://huggingface.co/papers/2410.20650

authored a paper about 2 months ago

Teacher Forcing Recovers Reward Functions for Text Generation

View all activity

Organizations

yongchanghao's activity

upvoted a paper 3 days ago

Qwen2.5 Technical Report

Paper • 2412.15115 • Published 5 days ago • 322

reacted to their post with 🔥 about 2 months ago

Post

3746

We just released a paper (NeuZip) that compresses VRAM in a lossless manner to run larger models. This should be particularly useful when VRAM is insufficient during training/inference. Specifically, we look inside each floating number and find that the exponents are highly compressible (as shown in the figure below).

Read more about the work at NeuZip: Memory-Efficient Training and Inference with Dynamic Compression of Neural Networks (2410.20650)

authored a paper about 2 months ago

Teacher Forcing Recovers Reward Functions for Text Generation

Paper • 2210.08708 • Published Oct 17, 2022

posted an update about 2 months ago

Post

3746

We just released a paper (NeuZip) that compresses VRAM in a lossless manner to run larger models. This should be particularly useful when VRAM is insufficient during training/inference. Specifically, we look inside each floating number and find that the exponents are highly compressible (as shown in the figure below).

Read more about the work at NeuZip: Memory-Efficient Training and Inference with Dynamic Compression of Neural Networks (2410.20650)