add fine-tuned model trained on multilingual conversations based on decapoda-research/llama-13b-hf model

committing original decapoda-research/llama-13b-hf model