HF RL Explorer

Complete the MIL implementation of reverse sequence so sequence models can be evaluated and validated…

Complete the MIL implementation of reverse sequence so sequence models can be evaluated and validated…: a task in MiMo-V2.6-RL-oss: Agentic RL Environments (MiMo RL release). The current public builder can lower the operation, but materialized inputs do not produce a compile-time value and…

The task

The current public builder can lower the operation, but materialized inputs do not produce a compile-time value and malformed lengths or axes are allowed to reach later stages. For mb.reverse_sequence, lengths is a rank-1 tensor with one entry for every position of the batch…

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