I want the streaming evaluate command to support speech-to-speech streaming evaluation through --task…
I want the streaming evaluate command to support speech-to-speech streaming evaluation through --task…: a task in MiMo-V2.6-RL-harbor-code: MiMo-V2.6-RL Code (Harbor) (Harbor dataset). I also need an expressive speech-to-speech mode on the same command. When I add --expressive to the --task s2st…
The task
I also need an expressive speech-to-speech mode on the same command. When I add `--expressive` to the `--task s2st` invocation, the command should select the expressive speech-to-speech agent instead of the standard streaming speech-to-speech agent while keeping the S2ST evaluation defaults above. A valid `--task…
Part of FineEnvs/MiMo-V2.6-RL-harbor-code.