3 1

wujing

wjmcat

http://wj-Mcat.github.io

AI & ML interests

student who wants to search for nlp models

Recent Activity

published a dataset 2 minutes ago

ernie-research/MEnvData-SWE-Trajectory

updated a dataset 17 minutes ago

ernie-research/MEnvData-SWE

updated a dataset 24 minutes ago

ernie-research/MEnvBench

View all activity

Organizations

published a dataset 2 minutes ago

ernie-research/MEnvData-SWE-Trajectory

Updated 2 minutes ago

updated a dataset 17 minutes ago

ernie-research/MEnvData-SWE

Preview • Updated 9 minutes ago • 5

updated a dataset 24 minutes ago

ernie-research/MEnvBench

Viewer • Updated 24 minutes ago • 1k • 13

published a dataset 4 days ago

ernie-research/MEnvData-SWE

Preview • Updated 9 minutes ago • 5

published a dataset 14 days ago

ernie-research/MEnvBench

Viewer • Updated 24 minutes ago • 1k • 13

New activity in zai-org/GLM-4.6 4 months ago

What is the Time Range Selected in the LCB V6 Benchmark?

#15 opened 4 months ago by

wjmcat

liked a model 5 months ago

baidu/ERNIE-4.5-21B-A3B-Thinking

Text Generation • 22B • Updated Nov 26, 2025 • 331 • • 772

New activity in deepseek-ai/DeepSeek-V3.1-Base 5 months ago

the pass@1 of deepseek-v3.1-base in lcb benchmark

#57 opened 5 months ago by

wjmcat

reacted to yushun0410's post with 🚀🔥 over 1 year ago

Post

4656

Hi Huggingfacers!

Thrilled to introduce Adam-mini, an optimizer that achieves on-par or better performance than AdamW with 45% to 50% less memory footprint. Adam-mini can also achieve 49.5% higher throughput than AdamW on Llama2-7B pre-training.

The design of Adam-mini is inspired by certain Hessian structures we observed on Transformers.

Feel free to try it out! Try switching to Adam-mini with the same hyperparams of AdamW, it would work with only half memory. Hope Adam-mini can help save time, cost, and energy in your tasks!

Paper: "Adam-mini: Use Fewer Learning Rates To Gain More" https://arxiv.org/abs/2406.16793

Code: https://github.com/zyushun/Adam-mini