llama-2

Training procedure

The following bitsandbytes quantization config was used during training:

The following bitsandbytes quantization config was used during training:

Framework versions