issues
search
davmacario
/
MDI-LLM
Implementation of Model-Distributed Inference for Large Language Models, built on top of LitGPT
MIT License
3
stars
2
forks
source link
[feat]: swap client/server nodes
#1
Closed
davmacario
closed
8 months ago
davmacario
commented
8 months ago
Now the nodes act as clients for the following ones, and as servers for the previous ones.
Improved stability and efficiency by removing the waiting delay when message queue is empty