Search results

Untuned compile/torch throughput regressed 30-60% vs prior README on humanoid/halfcheetah

Status: Open.

#70 In vmoens/mujoco-torch;

· vmoens opened on Apr 21, 2026

TorchRL collector + LazyTensorStorage(ndim=2) + extend_buffer=True hangs on multi-env sweep

Status: Open.

#69 In vmoens/mujoco-torch;

· vmoens opened on Apr 21, 2026

Ahead-of-time compilation: ship pre-compiled models to eliminate warmup

Status: Open.

#27 In vmoens/mujoco-torch;

· vmoens opened on Mar 2, 2026

Sensor group tensors are invisible to Model.to(device)

Status: Open.

#24 In vmoens/mujoco-torch;

· vmoens opened on Mar 1, 2026

Leverage TensorDict batch_size semantics with torch.vmap

Status: Open.

#23 In vmoens/mujoco-torch;

· vmoens opened on Feb 28, 2026

torch.compile compatibility: progress and remaining work

Status: Open.

#1 In vmoens/mujoco-torch;

· vmoens opened on Feb 12, 2026