HF RL Explorer

I'm trying to train the conf branch method with a resnet18 backbone on CIFAR and it blows up with a shape…

I'm trying to train the conf branch method with a resnet18 backbone on CIFAR and it blows up with a shape…: a task in MiMo-V2.6-RL-oss: Agentic RL Environments (MiMo RL release). The configs/train conf branch.yml as-is doesn't seem to work for anything other than whatever tiny backbone it was…

The task

The configs/train_conf_branch.yml as-is doesn't seem to work for anything other than whatever tiny backbone it was originally written against. Also while poking around I noticed postprocessor.name: opengan just throws a KeyError saying it's not registered, and trainer.name…

Part of XiaomiMiMo/MiMo-V2.6-RL-oss.