Pith. sign in

REVIEW 2 cited by

Inverse Dynamics Pretraining Learns Good Representations for Multitask Imitation

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2305.16985 v2 pith:VXMWSED4 submitted 2023-05-26 cs.LG

Inverse Dynamics Pretraining Learns Good Representations for Multitask Imitation

classification cs.LG
keywords pretrainingdynamicsinversecontextcorpusdemonstrationdemonstrationsdimensional
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

In recent years, domains such as natural language processing and image recognition have popularized the paradigm of using large datasets to pretrain representations that can be effectively transferred to downstream tasks. In this work we evaluate how such a paradigm should be done in imitation learning, where both pretraining and finetuning data are trajectories collected by experts interacting with an unknown environment. Namely, we consider a setting where the pretraining corpus consists of multitask demonstrations and the task for each demonstration is set by an unobserved latent context variable. The goal is to use the pretraining corpus to learn a low dimensional representation of the high dimensional (e.g., visual) observation space which can be transferred to a novel context for finetuning on a limited dataset of demonstrations. Among a variety of possible pretraining objectives, we argue that inverse dynamics modeling -- i.e., predicting an action given the observations appearing before and after it in the demonstration -- is well-suited to this setting. We provide empirical evidence of this claim through evaluations on a variety of simulated visuomotor manipulation problems. While previous work has attempted various theoretical explanations regarding the benefit of inverse dynamics modeling, we find that these arguments are insufficient to explain the empirical advantages often observed in our settings, and so we derive a novel analysis using a simple but general environment model.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Learning to Move Before Learning to Do: Task-Agnostic pretraining for VLAs

    cs.RO 2026-07 unverdicted novelty 6.0

    TAP uses two-stage pretraining on unlabeled data to learn physical competence before language grounding, matching 1M-expert models with far less labeled data and showing robustness on real robots.

  2. ATM: Action-Consistency Transfer Matrix for Diagnosing and Improving Latent World Models

    cs.CV 2026-06 unverdicted novelty 6.0

    ATM is a post-hoc probe-based transfer matrix that diagnoses action consistency in latent world models and serves as a training signal via AITS, enabling fast reliable ranking with claimed 100x speedup over CEM planne...