statement

• deep_training

    2024-06-12 init glm4v

install

• pip install -U -r requirements.txt
• 如果无法安装 , 可以切换官方源 pip install -i https://pypi.org/simple -U -r requirements.txt

weight

• glm-4v-9b

data sample

open_data https://github.com/ssbuild/open_data

{"id": 1, "conversations": [{"from": "system", "value": "You are ChatGLM4, a large language model trained by Zhipu.AI. Follow the user's instructions carefully. Respond using markdown."}, {"from": "user", "value": "图中的狗是什么品种？", "img": "../assets/demo.jpeg"}, {"from": "assistant", "value": "图中是一只拉布拉多犬。"}]}
{"id": 2, "conversations": [{"from": "user", "value": "这张图片的背景里有什么内容？", "img": "../assets/2p.png"}, {"from": "assistant", "value": "这张图片的背景是蒙蒙细雨。"}]}
{"id": 3, "conversations": [{"from": "user", "value": "这张图片的背景里有什么内容？", "img": "../assets/pig.png"}, {"from": "assistant", "value": "这张图片的背景是是虚化的。"}]}

infer

# infer.py 推理预训练模型
# infer_finetuning.py 推理微调模型
# infer_lora_finetuning.py 推理lora微调模型
 python infer.py

training

    # 制作数据
    cd scripts
    bash train_full.sh -m dataset 
    or
    bash train_lora.sh -m dataset 
    or
    bash train_ptv2.sh -m dataset 
    
    注: num_process_worker 为多进程制作数据 ， 如果数据量较大 ， 适当调大至cpu数量
    dataHelper.make_dataset_with_args(data_args.train_file,mixed_data=False, shuffle=True,mode='train',num_process_worker=0)
    
    # 全参数训练 
        bash train_full.sh -m train
        
    # lora adalora ia3 
        bash train_lora.sh -m train
        
    # ptv2
        bash train_ptv2.sh -m train

训练参数

训练参数

友情链接

• pytorch-task-example
• chatmoss_finetuning
• chatglm_finetuning
• chatglm2_finetuning
• t5_finetuning
• llm_finetuning
• llm_rlhf
• chatglm_rlhf
• t5_rlhf
• rwkv_finetuning
• baichuan_finetuning
• baichuan2_finetuning
• xverse_finetuning
• aigc_serving
• aigc_evals

纯粹而干净的代码

Reference

https://github.com/THUDM/glm4v-6B

Star History