Paper Detail

Looped Language Models Improve Compositional Tool Calling

Andrei Cristian Popescu, Haitz Sáez de Ocáriz Borde, Pietro Liò

huggingface Score 22.8

Published 2026-08-17 · First seen 2026-08-20

General AI

Abstract

Looped language models have shown promising results on reasoning benchmarks, yet their potential for agentic tool use remains largely unexplored. We study this question in compositional tool-calling settings, where models must coordinate multiple API calls, maintain intermediate state, and preserve dependencies across tool interactions. We evaluate native and retrofitted looped language models on API-Bank, BFCL, and NESTful, comparing looped and non-looped models trained under matched supervised fine-tuning recipes and varying recurrent depth at inference time. In controlled experiments, recurrent computation generally benefits compositional and dependency-aware tool use, while providing smaller and more model-dependent gains on isolated API invocation. Accuracy on multi-step tool use generally increases with recurrent depth; adaptive inference, however, achieves a more favorable compute-performance trade-off by allocating additional computation only when needed. Our results suggest that looped language models are a promising architecture for agentic systems that require reliable planning, coordination, and execution of compositional tool use workflows.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
now
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@misc{popescu2026looped,
  title = {Looped Language Models Improve Compositional Tool Calling},
  author = {Andrei Cristian Popescu and Haitz Sáez de Ocáriz Borde and Pietro Liò},
  year = {2026},
  abstract = {Looped language models have shown promising results on reasoning benchmarks, yet their potential for agentic tool use remains largely unexplored. We study this question in compositional tool-calling settings, where models must coordinate multiple API calls, maintain intermediate state, and preserve dependencies across tool interactions. We evaluate native and retrofitted looped language models on API-Bank, BFCL, and NESTful, comparing looped and non-looped models trained under matched supervised},
  url = {https://huggingface.co/papers/2608.18171},
  keywords = {looped language models, compositional tool-calling, recurrent depth, adaptive inference, API-Bank, BFCL, NESTful, multi-step tool use, agentic systems, huggingface daily},
  eprint = {2608.18171},
  archiveprefix = {arXiv},
}

Metadata

{}