mirror of
https://github.com/wassname/ray.git
synced 2026-09-28 14:30:40 +08:00
Cleanup: TFPolicyGraph now automatically adds loss input entries for state_in_*, so that graph sub-classes don't need to worry about it. Multi-GPU support: Allow setting up model tower replicas with existing state input tensors Truncate the per-device minibatch slices so that they are always a multiple of max_seq_len.