wangguanzhong
wangguanzhong
Add train_imagenet and reader_imagenet
Does the single scale version of CenterNet contain origin image and flipped image? I found that image is flipped as default in test/coco.py and it may cause double cost in...
I have trained CornerNet-Squeeze on P40 4gpu with default config and met some problems: 1. The training process costed around 10 days and I got mAP=33.6, but the released mAP...
Could you please provide the test speed info ?
## PaddleMIX 2.0版本正式发布 https://github.com/PaddlePaddle/PaddleMIX/tree/v2.0.0 * 多模态理解:新增LLaVA系列,Qwen-VL等;新增Auto模块统一SFT训练流程;新增mixtoken训练策略,SFT吞吐量提升5.6倍。 * 多模态生成:发布[PPDiffusers 0.24.1](./ppdiffusers/README.md)版本,支持视频生成能力,文生图模型新增LCM。新增飞桨版peft,accelerate后端。提供基于飞桨开发的ComfyUI插件。 * 多模态数据处理工具箱[DataCopilot](./paddlemix/datacopilot/):支持自定义数据结构,数据转换,离线格式检查;支持基本的统计信息,数据可视化功能。 欢迎开发者们使用并反馈!