AI Paper Atlas

Research map / Reinforcement learning / Policy gradient and actor-critic

Chunking the Critic: A Transformer-based Soft Actor-Critic with N-Step Returns

Dong Tian, Onur Celik, Gerhard Neumann

ICLR 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Actor-critic theory · 13 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Transformer critic with N-step supervision to SACSAC · Box-Pushing densesuccess rate (%): <=85 -> 96.8Improved
sparse-reward Box-Pushing, T-SAC vs TOP-ERLTOP-ERL · Box-Pushing sparsesuccess rate (%): 70 -> 60Hurt
replace Transformer critic with multi-step MLP critic (ablation)T-SAC · Box-Pushing densesuccess rateHurt
remove causal mask from Transformer critic (ablation)T-SAC · Box-Pushing densesuccess rateImproved
replace Transformer with GRU or LSTM critic (ablation)T-SAC · Box-Pushing densesuccess rateHurt
use fixed horizon instead of sampled N-step horizons (ablation)T-SAC · Box-Pushing densesuccess rateHurt
add multiple supervision windows per step (up to 4) (ablation)T-SAC · Box-Pushing densesuccess rateImproved
add policy-side chunking to MLP critic (ablation)MLP critic · Box-Pushing densesuccess rateNo clear effect
add policy chunking length 4 to Transformer critic (ablation)T-SAC · Box-Pushing densefinal success rate (%): 99.5 final successImproved
increase critic freezing interval K to very large values (ablation)T-SAC · Walker2dreturnHurt
replace Polyak target with critic-freezing hard copy (ablation)T-SAC · MuJoCo locomotionreturnNo clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy