An implementation of model parallel autoregressive transformers on GPUs, based on the DeepSpeed library.
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).