{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:ZFSINJKNCOLEDE6JVIYNBEYIFL","short_pith_number":"pith:ZFSINJKN","schema_version":"1.0","canonical_sha256":"c96486a54d13964193c9aa30d093082adb73a458a58968868c1ab5579a2639d6","source":{"kind":"arxiv","id":"2205.12255","version":1},"attestation_state":"computed","paper":{"title":"TALM: Tool Augmented Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Aaron Parisi, Noah Fiedel, Yao Zhao","submitted_at":"2022-05-24T17:58:13Z","abstract_excerpt":"Transformer based language models (LMs) demonstrate increasing performance with scale across a wide variety of tasks. Scale alone however cannot enable models to solve tasks that require access to ephemeral, changing, or private data that was unavailable at training time. Many useful tasks may also benefit from LMs being able to access APIs that read or modify state. In this work, we present Tool Augmented Language Models (TALM), combining a text-only approach to augment language models with non-differentiable tools, and an iterative \"self-play\" technique to bootstrap performance starting from"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2205.12255","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-24T17:58:13Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ebab42715ff715594fe6a765654fddf9bf4cdcd00337598f2113ac1cd0a3a5b5","abstract_canon_sha256":"a1936c18d28b213986a56a95eabe8ce3c571ebea1831d5fe924ea8ed10a9bfcf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:26:12.624403Z","signature_b64":"qwk2jnrwlU/c5wgJNb1trORFUiR6VMVFp5EqtL70Uk+89zjW+HiP+IhsCYMxtM70hNvb3E3vpUtjs9xIc91uDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c96486a54d13964193c9aa30d093082adb73a458a58968868c1ab5579a2639d6","last_reissued_at":"2026-07-05T04:26:12.624018Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:26:12.624018Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TALM: Tool Augmented Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Aaron Parisi, Noah Fiedel, Yao Zhao","submitted_at":"2022-05-24T17:58:13Z","abstract_excerpt":"Transformer based language models (LMs) demonstrate increasing performance with scale across a wide variety of tasks. Scale alone however cannot enable models to solve tasks that require access to ephemeral, changing, or private data that was unavailable at training time. Many useful tasks may also benefit from LMs being able to access APIs that read or modify state. In this work, we present Tool Augmented Language Models (TALM), combining a text-only approach to augment language models with non-differentiable tools, and an iterative \"self-play\" technique to bootstrap performance starting from"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2205.12255","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2205.12255/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2205.12255","created_at":"2026-07-05T04:26:12.624073+00:00"},{"alias_kind":"arxiv_version","alias_value":"2205.12255v1","created_at":"2026-07-05T04:26:12.624073+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.12255","created_at":"2026-07-05T04:26:12.624073+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZFSINJKNCOLE","created_at":"2026-07-05T04:26:12.624073+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZFSINJKNCOLEDE6J","created_at":"2026-07-05T04:26:12.624073+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZFSINJKN","created_at":"2026-07-05T04:26:12.624073+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":26,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.08768","citing_title":"UniClawBench: A Universal Benchmark for Proactive Agents on Real-World Tasks","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2607.06157","citing_title":"LLM Agents for Deliberative Collaboration: A Study on Joint Decision Making Under Partial Observability","ref_index":121,"is_internal_anchor":true},{"citing_arxiv_id":"2606.22385","citing_title":"MetaPS: Adaptive Programmatic Strategy Selection for Market Agents","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06566","citing_title":"NTILC: Neural Tool Invocation via Learned Compression","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05263","citing_title":"Policy-Conditioned Counterfactual Credit for Verifiable Reinforcement Learning of Long-Horizon Language Agents","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03858","citing_title":"PyraMathBench: Evaluating and Improving Mathematical Capability in Large Language Models","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2503.22693","citing_title":"Bridging Language Models and Financial Analysis","ref_index":74,"is_internal_anchor":false},{"citing_arxiv_id":"2510.06824","citing_title":"Efficient numeracy in language models through single-token number embeddings","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21299","citing_title":"Tracing the ongoing emergence of human-like reasoning in Large Language Models","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14038","citing_title":"Model-Adaptive Tool Necessity Reveals the Knowing-Doing Gap in LLM Tool Use","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17318","citing_title":"RooAgent: An LLM Agent for Root-Based High Energy Physics Analysis","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2307.06435","citing_title":"A Comprehensive Overview of Large Language Models","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2309.17452","citing_title":"ToRA: A Tool-Integrated Reasoning Agent for Mathematical Problem Solving","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2506.19500","citing_title":"NaviAgent: Bilevel Planning on Tool Navigation Graph for Large-Scale Orchestration","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2510.23853","citing_title":"Your LLM Agents are Temporally Blind: The Misalignment Between Tool Use Decisions and Human Time Perception","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2303.08128","citing_title":"ViperGPT: Visual Inference via Python Execution for Reasoning","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2309.02427","citing_title":"Cognitive Architectures for Language Agents","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2303.09014","citing_title":"ART: Automatic multi-step reasoning and tool-use for large language models","ref_index":166,"is_internal_anchor":false},{"citing_arxiv_id":"2601.13262","citing_title":"CURE-Med: Curriculum-Informed Reinforcement Learning for Multilingual Medical Reasoning","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2306.13549","citing_title":"A Survey on Multimodal Large Language Models","ref_index":193,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14038","citing_title":"Model-Adaptive Tool Necessity Reveals the Knowing-Doing Gap in LLM Tool Use","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2404.13208","citing_title":"The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27586","citing_title":"Trace-Level Analysis of Information Contamination in Multi-Agent Systems","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19299","citing_title":"Rethinking Scale: Deployment Trade-offs of Small Language Models under Agent Paradigms","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08519","citing_title":"Cram Less to Fit More: Training Data Pruning Improves Memorization of Facts","ref_index":67,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZFSINJKNCOLEDE6JVIYNBEYIFL","json":"https://pith.science/pith/ZFSINJKNCOLEDE6JVIYNBEYIFL.json","graph_json":"https://pith.science/api/pith-number/ZFSINJKNCOLEDE6JVIYNBEYIFL/graph.json","events_json":"https://pith.science/api/pith-number/ZFSINJKNCOLEDE6JVIYNBEYIFL/events.json","paper":"https://pith.science/paper/ZFSINJKN"},"agent_actions":{"view_html":"https://pith.science/pith/ZFSINJKNCOLEDE6JVIYNBEYIFL","download_json":"https://pith.science/pith/ZFSINJKNCOLEDE6JVIYNBEYIFL.json","view_paper":"https://pith.science/paper/ZFSINJKN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2205.12255&json=true","fetch_graph":"https://pith.science/api/pith-number/ZFSINJKNCOLEDE6JVIYNBEYIFL/graph.json","fetch_events":"https://pith.science/api/pith-number/ZFSINJKNCOLEDE6JVIYNBEYIFL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZFSINJKNCOLEDE6JVIYNBEYIFL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZFSINJKNCOLEDE6JVIYNBEYIFL/action/storage_attestation","attest_author":"https://pith.science/pith/ZFSINJKNCOLEDE6JVIYNBEYIFL/action/author_attestation","sign_citation":"https://pith.science/pith/ZFSINJKNCOLEDE6JVIYNBEYIFL/action/citation_signature","submit_replication":"https://pith.science/pith/ZFSINJKNCOLEDE6JVIYNBEYIFL/action/replication_record"}},"created_at":"2026-07-05T04:26:12.624073+00:00","updated_at":"2026-07-05T04:26:12.624073+00:00"}