HuFY-dev/SAELens
Training Sparse Autoencoders on Language Models
Training Sparse Autoencoders on Language Models
Evaluating and reproducing real-world robot manipulation policies (e.g., RT-1, RT-1-X, Octo) in simulation under common setups (e.g., Google Robot, WidowX+Bridge)
Sparse Autoencoder for Mechanistic Interpretability
本地部署chatglm生成并以语音回复你的bilibili直播弹幕 Use chatglm to generate and reply your bilibili live danmu with voice
The nnsight package enables interpreting and manipulating the internals of deep learned models.
Mastering Diverse Domains through World Models
Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities
Use ChatGPT to summarize the arXiv papers. 全流程加速科研,利用chatgpt进行论文总结+润色+审稿+审稿回复
This is a toy example of GPT
DGL implementation of GCN for link prediction, trained on ogbl-colab dataset.
SoftVC VITS Singing Voice Conversion
so-vits-svc fork with REALTIME support (voice changer) and greatly improved interface.
An easy to understand TTS / SVS / SVC framework