We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries
