Jingyi Chen

i solve my problem by changing the picture storage format from JPEG to PNG. As the jpg format does not guarantee the binarization of images.

i load the pretrained model of inpainting_places and use 7w+ data, only after serveral epochs it perform great

有没有朋友知道怎么解决 用脚本跑也行呀 I0000 00:00:1721828901.767197 89162 subchannel.cc:806] subchannel 0x5d0f7f0 {address=ipv6:%5B::1%5D:42359, args={grpc.client_channel_factory=0x41690c0, grpc.default_authority=localhost:42359, grpc.internal.channel_credentials=0x41690a0, grpc.internal.client_channel_call_destination=0x7f0e6adce3d0, grpc.internal.event_engine=0x5cefd30, grpc.internal.security_connector=0x5cee5d0, grpc.internal.subchannel_pool=0x5cedbd0, grpc.max_receive_message_length=268435456, grpc.primary_user_agent=grpc-python/1.65.1, grpc.resource_quota=0x5ceeab0, grpc.server_uri=dns:///localhost:42359}}: connect failed (UNKNOWN:Failed to connect to remote host: connect: Connection...

自己搞定了 config设置一个toml 然后用train_by_toml.sh即可