Adding `safetensors` variant of this model

Update tokenizer_config.json

Update config.json

Duplicate from decapoda-research/llama-7b-hf