{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZTBNYP2XKII7ULM5O3J6BXAASP","short_pith_number":"pith:ZTBNYP2X","schema_version":"1.0","canonical_sha256":"ccc2dc3f575211fa2d9d76d3e0dc0093f1c1862d8c5c3cdd9bdc1462c64399d1","source":{"kind":"arxiv","id":"2505.19433","version":2},"attestation_state":"computed","paper":{"title":"Can Compressed LLMs Truly Act? An Empirical Evaluation of Agentic Capabilities in LLM Compression","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bo Li, Lujun Li, Peijie Dong, Xiang Liu, Xiaowen Chu, Zhenheng Tang","submitted_at":"2025-05-26T02:49:07Z","abstract_excerpt":"Post-training compression reduces the computational and memory costs of large language models (LLMs), enabling resource-efficient deployment. However, existing compression benchmarks only focus on language modeling (e.g., perplexity) and natural language understanding tasks (e.g., GLUE accuracy), ignoring the agentic capabilities - workflow, tool use/function call, long-context understanding and real-world application. We introduce the Agent Compression Benchmark (ACBench), the first comprehensive benchmark for evaluating how compression impacts LLMs' agentic abilities. ACBench spans (1) 12 ta"},"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":"2505.19433","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-26T02:49:07Z","cross_cats_sorted":[],"title_canon_sha256":"8b742ae6ca34576ada6856938e4d80f1728f267179d2de4a89636ab2c3c61a41","abstract_canon_sha256":"8e9a6fcf41cdebf7da9a9ce7d9e011018de190f3900e2d7d65e13792e3dae097"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:13:26.917729Z","signature_b64":"ec1zEZ06U0ZqCPWEXvkr3R9YOBz0jHIoYwi8uIx43cXEyphyiCPvyYSGpp7YYa1yLwgckA166rRJIK4KTwUADw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ccc2dc3f575211fa2d9d76d3e0dc0093f1c1862d8c5c3cdd9bdc1462c64399d1","last_reissued_at":"2026-07-05T11:13:26.917209Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:13:26.917209Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Can Compressed LLMs Truly Act? An Empirical Evaluation of Agentic Capabilities in LLM Compression","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bo Li, Lujun Li, Peijie Dong, Xiang Liu, Xiaowen Chu, Zhenheng Tang","submitted_at":"2025-05-26T02:49:07Z","abstract_excerpt":"Post-training compression reduces the computational and memory costs of large language models (LLMs), enabling resource-efficient deployment. However, existing compression benchmarks only focus on language modeling (e.g., perplexity) and natural language understanding tasks (e.g., GLUE accuracy), ignoring the agentic capabilities - workflow, tool use/function call, long-context understanding and real-world application. We introduce the Agent Compression Benchmark (ACBench), the first comprehensive benchmark for evaluating how compression impacts LLMs' agentic abilities. ACBench spans (1) 12 ta"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.19433","kind":"arxiv","version":2},"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/2505.19433/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":"2505.19433","created_at":"2026-07-05T11:13:26.917272+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.19433v2","created_at":"2026-07-05T11:13:26.917272+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.19433","created_at":"2026-07-05T11:13:26.917272+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZTBNYP2XKII7","created_at":"2026-07-05T11:13:26.917272+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZTBNYP2XKII7ULM5","created_at":"2026-07-05T11:13:26.917272+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZTBNYP2X","created_at":"2026-07-05T11:13:26.917272+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.18854","citing_title":"Evaluating Memory Condensation Strategies for Coding Agents in Data-Driven Scientific Discovery","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2508.03949","citing_title":"Model Compression vs. Adversarial Robustness: An Empirical Study on Language Models for Code","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2507.13334","citing_title":"A Survey of Context Engineering for Large Language Models","ref_index":235,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11733","citing_title":"Position: LLM Inference Should Be Evaluated as Energy-to-Token Production","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22577","citing_title":"QuantClaw: Precision Where It Matters for OpenClaw","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZTBNYP2XKII7ULM5O3J6BXAASP","json":"https://pith.science/pith/ZTBNYP2XKII7ULM5O3J6BXAASP.json","graph_json":"https://pith.science/api/pith-number/ZTBNYP2XKII7ULM5O3J6BXAASP/graph.json","events_json":"https://pith.science/api/pith-number/ZTBNYP2XKII7ULM5O3J6BXAASP/events.json","paper":"https://pith.science/paper/ZTBNYP2X"},"agent_actions":{"view_html":"https://pith.science/pith/ZTBNYP2XKII7ULM5O3J6BXAASP","download_json":"https://pith.science/pith/ZTBNYP2XKII7ULM5O3J6BXAASP.json","view_paper":"https://pith.science/paper/ZTBNYP2X","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.19433&json=true","fetch_graph":"https://pith.science/api/pith-number/ZTBNYP2XKII7ULM5O3J6BXAASP/graph.json","fetch_events":"https://pith.science/api/pith-number/ZTBNYP2XKII7ULM5O3J6BXAASP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZTBNYP2XKII7ULM5O3J6BXAASP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZTBNYP2XKII7ULM5O3J6BXAASP/action/storage_attestation","attest_author":"https://pith.science/pith/ZTBNYP2XKII7ULM5O3J6BXAASP/action/author_attestation","sign_citation":"https://pith.science/pith/ZTBNYP2XKII7ULM5O3J6BXAASP/action/citation_signature","submit_replication":"https://pith.science/pith/ZTBNYP2XKII7ULM5O3J6BXAASP/action/replication_record"}},"created_at":"2026-07-05T11:13:26.917272+00:00","updated_at":"2026-07-05T11:13:26.917272+00:00"}