Growing Through Experience: Scaling Episodic Grounding in Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Chunhui, Sirui, Wang, Ouyang, Zhongyu, Yuan, Xiangchi, Vosoughi, Soroush
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918042849509376
author Zhang, Chunhui
Sirui
Wang
Ouyang, Zhongyu
Yuan, Xiangchi
Vosoughi, Soroush
author_facet Zhang, Chunhui
Sirui
Wang
Ouyang, Zhongyu
Yuan, Xiangchi
Vosoughi, Soroush
contents Language models (LMs) require robust episodic grounding-the capacity to learn from and apply past experiences-to excel at physical planning tasks. Current episodic grounding approaches struggle with scalability and integration, limiting their effectiveness, especially for medium-sized LMs (7B parameters). While larger LMs (70-405B parameters) possess superior hierarchical representations and extensive pre-trained knowledge, they encounter a fundamental scale paradox: despite their advanced abstraction capabilities, they lack efficient mechanisms to leverage experience streams. We propose a scalable weak-to-strong episodic learning framework that effectively transfers episodic behaviors from smaller to larger LMs. This framework integrates Monte Carlo tree search for structured experience collection with a novel distillation method, preserving the inherent LM capabilities while embedding episodic memory. Experiments demonstrate our method surpasses state-of-the-art proprietary LMs by 3.45% across diverse planning and question-answering tasks. Layer-wise probing further indicates significant improvements in task alignment, especially within deeper LM layers, highlighting stable generalization even for previously unseen scenarios with increased planning complexity-conditions where baseline methods degrade markedly.
format Preprint
id arxiv_https___arxiv_org_abs_2506_01312
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Growing Through Experience: Scaling Episodic Grounding in Language Models
Zhang, Chunhui
Sirui
Wang
Ouyang, Zhongyu
Yuan, Xiangchi
Vosoughi, Soroush
Computation and Language
Language models (LMs) require robust episodic grounding-the capacity to learn from and apply past experiences-to excel at physical planning tasks. Current episodic grounding approaches struggle with scalability and integration, limiting their effectiveness, especially for medium-sized LMs (7B parameters). While larger LMs (70-405B parameters) possess superior hierarchical representations and extensive pre-trained knowledge, they encounter a fundamental scale paradox: despite their advanced abstraction capabilities, they lack efficient mechanisms to leverage experience streams. We propose a scalable weak-to-strong episodic learning framework that effectively transfers episodic behaviors from smaller to larger LMs. This framework integrates Monte Carlo tree search for structured experience collection with a novel distillation method, preserving the inherent LM capabilities while embedding episodic memory. Experiments demonstrate our method surpasses state-of-the-art proprietary LMs by 3.45% across diverse planning and question-answering tasks. Layer-wise probing further indicates significant improvements in task alignment, especially within deeper LM layers, highlighting stable generalization even for previously unseen scenarios with increased planning complexity-conditions where baseline methods degrade markedly.
title Growing Through Experience: Scaling Episodic Grounding in Language Models
topic Computation and Language
url https://arxiv.org/abs/2506.01312