# T2S: Tokenized Skill Scaling for Lifelong Imitation Learning

arXiv: https://arxiv.org/abs/2508.01167 · HTML: https://mlsanity.com/paper/2508.01167/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Robot manipulation imitation learning, one of 13 directions in Offline RL and imitation learning from data.
- Builds on 25 of its 41 references that are in the map.
- First posted on arXiv 2 Aug 2025.

## Placed in

- Offline RL and imitation learning from data / [Robot manipulation imitation learning](https://mlsanity.com/offlinerl/robot-manipulation-imitation-learning/)

## Builds on (most cited first, 25 of 25)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [Overcoming catastrophic forgetting in neural networks](https://mlsanity.com/paper/1612.00796/) (2016)
- [Parameter-Efficient Transfer Learning for NLP](https://mlsanity.com/paper/1902.00751/) (2019)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [The Power of Scale for Parameter-Efficient Prompt Tuning](https://mlsanity.com/paper/2104.08691/) (2021)
- [LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning](https://mlsanity.com/paper/2306.03310/) (2023)
- [PackNet: Adding Multiple Tasks to a Single Network by Iterative Pruning](https://mlsanity.com/paper/1711.05769/) (2017)
- [DeepSeek LLM: Scaling Open-Source Language Models with Longtermism](https://mlsanity.com/paper/2401.02954/) (2024)
- [Deep Recurrent Q-Learning for Partially Observable MDPs](https://mlsanity.com/paper/1507.06527/) (2015)
- [An Algorithmic Perspective on Imitation Learning](https://mlsanity.com/paper/1811.06711/) (2018)
- [Scalable agent alignment via reward modeling: a research direction](https://mlsanity.com/paper/1811.07871/) (2018)
- [Language-Conditioned Imitation Learning for Robot Manipulation Tasks](https://mlsanity.com/paper/2010.12083/) (2020)
- [Preventing Zero-Shot Transfer Degradation in Continual Learning of Vision-Language Models](https://mlsanity.com/paper/2303.06628/) (2023)
- [Bottom-Up Skill Discovery from Unsegmented Demonstrations for Long-Horizon Robot Manipulation](https://mlsanity.com/paper/2109.13841/) (2021)
- [Learning to Grow Pretrained Models for Efficient Transformer Training](https://mlsanity.com/paper/2303.00980/) (2023)
- [Decomposing the Generalization Gap in Imitation Learning for Visual Robotic Manipulation](https://mlsanity.com/paper/2307.03659/) (2023)
- [bert2BERT: Towards Reusable Pretrained Language Models](https://mlsanity.com/paper/2110.07143/) (2021)
- [LOTUS: Continual Imitation Learning for Robot Manipulation Through Unsupervised Skill Discovery](https://mlsanity.com/paper/2311.02058/) (2023)
- [Online Continual Learning For Interactive Instruction Following Agents](https://mlsanity.com/paper/2403.07548/) (2024)
- [M2Distill: Multi-Modal Distillation for Lifelong Imitation Learning](https://mlsanity.com/paper/2410.00064/) (2024)
- [Never-Ending Behavior-Cloning Agent for Robotic Manipulation](https://mlsanity.com/paper/2403.00336/) (2024)
- [TokenFormer: Rethinking Transformer Scaling with Tokenized Model Parameters](https://mlsanity.com/paper/2410.23168/) (2024)

## Built on by (most cited first, 2 of 2)

- [SPREAD: Subspace Representation Distillation for Lifelong Imitation Learning](https://mlsanity.com/paper/2603.08763/) (2026)
- [Learning New Tasks via Reusable Skills: Skill-Compositional Experts for Embodied Continual Learning](https://mlsanity.com/paper/2606.15685/) (2026)

## Abstract (arXiv)

The main challenge in lifelong imitation learning lies in the balance between mitigating catastrophic forgetting of previous skills while maintaining sufficient capacity for acquiring new ones. However, current approaches typically address these aspects in isolation, overlooking their internal correlation in lifelong skill acquisition. We address this limitation with a unified framework named Tokenized Skill Scaling (T2S). Specifically, by tokenizing the model parameters, the linear parameter mapping of the traditional transformer is transformed into cross-attention between input and learnable tokens, thereby enhancing model scalability through the easy extension of new tokens. Additionally, we introduce language-guided skill scaling to transfer knowledge across tasks efficiently and avoid linearly growing parameters. Extensive experiments across diverse tasks demonstrate that T2S: 1) effectively prevents catastrophic forgetting (achieving an average NBT of 1.0% across the three LIBERO task suites), 2) excels in new skill scaling with minimal increases in trainable parameters (needing only 8.0% trainable tokens in an average of lifelong tasks), and 3) enables efficient knowledge transfer between tasks (achieving an average FWT of 77.7% across the three LIBERO task suites), offering a promising solution for lifelong imitation learning.