i tried to make an inference script for motion streamer and i found out that there was no implementation of the two step forward infrecne in llama_model so i made one based on the training loop, and is working but the Motion Streamer is not a streamer anymore it takes 5 seconds to generate 1 second of motion with a a rtx 6000 PRO
i tried to make an inference script for motion streamer and i found out that there was no implementation of the two step forward infrecne in llama_model so i made one based on the training loop, and is working but the Motion Streamer is not a streamer anymore it takes 5 seconds to generate 1 second of motion with a a rtx 6000 PRO