Pith Number

pith:DCDPKVRX

pith:2023:DCDPKVRXTDIYEJSDLF7FJSWYFP

not attested not anchored not stored refs resolved

Tool Learning with Foundation Models

Bokai Xu, Bowen Li, Chaojun Xiao, Cheng Qian, Cheng Yang, Chi Han, Dahai Li, Ganqu Cui, Heng Ji, Huadong Wang, Jason Phang, Jing Yi, Junxi Yan, Kunlun Zhu, Lan Yan, Maosong Sun, Ning Ding, Runchu Tian, Shengding Hu, Shihao Liang, Tongshuang Wu, Weilin Zhao, Weize Chen, Xian Sun, Xin Cong, Xingyu Shen, Xu Han, Yankai Lin, Yaxi Lu, Yining Ye, Yi Ren Fung, Yufei Huang, Yujia Qin, Yusheng Su, Yuxiang Huang, Yuzhang Zhu, Zheni Zeng, Zhenning Dai, Zhen Zhang, Zhiyuan Liu, Ziwei Tang

Foundation models learn tool use by decomposing tasks into subtasks, reasoning to adjust plans, and selecting the right tools for each step.

arxiv:2304.08354 v3 · 2023-04-17 · cs.CL · cs.AI · cs.LG

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{DCDPKVRXTDIYEJSDLF7FJSWYFP}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

We formulate a general tool learning framework: starting from understanding the user instruction, models should learn to decompose a complex task into several subtasks, dynamically adjust their plan through reasoning, and effectively conquer each sub-task by selecting appropriate tools.

C2weakest assumption

That foundation models can be effectively trained and prompted to follow the proposed decomposition-reasoning-tool-selection process at scale, and that experiments with 18 tools sufficiently demonstrate this potential without detailed methodology or baselines in the provided abstract.

C3one line summary

The paper reviews tool learning research, formulates a framework for foundation models to decompose tasks and use tools via reasoning, and demonstrates current models' capabilities through experiments with 18 tools.

References

15 extracted · 15 resolved · 1 Pith anchors

[1] Fine-Tuning Language Models from Human Preferences 1909 · arXiv:1909.08593

[2] tag: Banana Pie Recipes, type: recipe

[3] tag: Custard and Cream Pies, type: recipes

[4] tag: Mexican, type: recipe

[5] tag: No-Bake Pie Recipes, type: recipe

Formal links

2 machine-checked theorem links

Cited by

22 papers in Pith

Learning to Ask: When LLM Agents Meet Unclear Instruction

AgentClinic: a multimodal agent benchmark to evaluate AI in simulated clinical environments

An Empirical Study of Privacy Leakage Chains via Prompt Injection in Black-Box Chatbot Environments

Personal LLM Agents: Insights and Survey about the Capability, Efficiency and Security

Mind2Web: Towards a Generalist Agent for the Web

Receipt and verification

First computed	2026-05-17T23:38:47.770663Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

1886f5563798d1822643597e54cad82bc3209fdbb2bc6bc31a5549acfed333f8

Aliases

arxiv: 2304.08354 · arxiv_version: 2304.08354v3 · doi: 10.48550/arxiv.2304.08354 · pith_short_12: DCDPKVRXTDIY · pith_short_16: DCDPKVRXTDIYEJSD · pith_short_8: DCDPKVRX

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/DCDPKVRXTDIYEJSDLF7FJSWYFP \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 1886f5563798d1822643597e54cad82bc3209fdbb2bc6bc31a5549acfed333f8

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "b52d902d9c90a4c8075368c709429af6c316cb03eae703881a1705641afc0d71",
    "cross_cats_sorted": [
      "cs.AI",
      "cs.LG"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2023-04-17T15:16:10Z",
    "title_canon_sha256": "4cc0b63cd7974af7923f4c96d284caa7265f6d56a40398ea3cf20a4d8d82977c"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2304.08354",
    "kind": "arxiv",
    "version": 3
  }
}