lllyasviel/ControlNet

Low VRAM tests (now 8GB ok, begin to solve 6gb)

Open

#3 创建于 2023年2月11日

在 GitHub 查看
 (21 评论) (3 反应) (0 负责人)Python (27,526 star) (2,500 fork)batch import
help wanted

描述

RuntimeError: CUDA out of memory. Tried to allocate 58.00 MiB (GPU 0; 8.00 GiB total capacity; 7.14 GiB already allocated; 0 bytes free; 7.26 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

贡献者指南

技术栈
pythonpytorch
领域
machine learningperformance
议题类型
performance
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
4
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
3-5 days
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
active
清晰度议题是否清楚说明期望改动、验收标准和下一步。
mostly clear
前置要求
PythonPyTorchCUDA basics
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
40
研究方向
Investigate memory allocation in PyTorch and identify areas where memory can be saved, such as using gradient checkpointing, reducing batch size, or optimizing model layers. Look at current memory usage with nvidia smi and PyTorch memory profiler. Explore settings like max split size mb as mentioned in the error. The goal is to make the code work on 6GB GPUs.