Continued pre-training of Llama 3.1 70B with expanded tokenizer on Irish human text collection (output_model/irish_llama31_lora) or both human text and machine-translated text (output_model/irish_llama31_lora_data_v3) using LoRA