[Feature] Add ACTModel and ACTLoss for robot learning

[theap06]

Implements Action Chunking with Transformers (Zhao et al., 2023, arxiv.org/abs/2304.13705) as a composable model + loss pair.

Implements - #3636
ACTModel (torchrl/modules/models/act.py):

• CVAE encoder: [CLS | action_tokens | obs_token] → z_mu, z_log_var
• DETR-style Transformer decoder: learned action queries attend to [obs_token, z_token] → action chunk of shape (chunk_size, action_dim)
• Training mode: pass action_chunk to activate encoder and sample z
• Inference mode: omit action_chunk; defaults to z=0 (prior mean)
• Sinusoidal positional encoding, configurable hidden_dim / nheads / num_encoder_layers / num_decoder_layers / latent_dim / dropout

ACTLoss (torchrl/objectives/act.py):

• loss = L1(action_pred, action_chunk) + kl_weight * KL(q(z|o,a)||N(0,I))
• Returns loss_act (differentiable), loss_reconstruction and loss_kl (detached, for logging)
• kl_weight defaults to 10.0 per the original paper
• Follows the LossModule / convert_to_functional / to_module pattern established by DiffusionBCLoss

Also:

• Registers ACTModel in torchrl.modules.models and ACTLoss in torchrl.objectives
• Adds 31 tests covering shapes, training/inference modes, KL decomposition, reductions, and gradient flow
• Adds docs entries in objectives_other.rst and modules_models.rst

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/3639

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• Rolling out OSDC (ARC) runners on pull & trunk workflows in PyTorch main

⚠️ 17 Awaiting Approval

As of commit 92ee098 with merge base eb90c5d ():

AWAITING APPROVAL - The following workflows need approval before CI can run:

• Build Aarch64 Linux Wheels (gh)
• Build Linux Wheels (gh)
• Build M1 Wheels (gh)
• Build Windows Wheels (gh)
• Continuous Benchmark (gh)
• Continuous Benchmark (PR) (gh)
• Generate documentation (gh)
• Habitat Tests on Linux (gh)
• Libs Tests on Linux (gh)
• Lint (gh)
• LLM Tests on Linux (gh)
• Push Binary Nightly (gh)
• SOTA Tests on Linux (gh)
• Tutorials Tests on Linux (gh)
• Unit-tests on Linux (gh)
• Unit-tests on Windows (gh)
• Validate Test Partitioning (gh)

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[theap06]

@vmoens lmk if you need any outputs before mergings.

[theap06]

@vmoens lgtm