Some scrips 4 FineTuningLlama3 by using unsloth
算力云服务器 https://www.autodl.com/home 费用:每小时 1.66元
机器配置: CPU :14 核心 内存:45 GB GPU :NVIDIA GeForce RTX 3090, 1
预训练大模型 Llama3-Chinese is a large model trained on 500k high-quality Chinese multi-turn SFT data, 100k English multi-turn SFT data, and 2k single-turn self-cognition data, using the training methods of DORA and LORA+ based on Meta-Llama-3-8B as the base. https://www.modelscope.cn/models/seanzhang/Llama3-Chinese
模型微调框架 unsloth https://unsloth.ai/ Finetune Llama 3.1, Mistral, Phi-3 & Gemma 2-5x faster with 80% less memory! 更少的内存,更好的性能
数据集采样 客户化定制数据集采样:来自于麦库笔记 json 文件格式用例:
[ { "output": "Jason,Keep going!" }]