{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:52YMVX2E3PTADRRW2TGBBHRIV2","short_pith_number":"pith:52YMVX2E","schema_version":"1.0","canonical_sha256":"eeb0cadf44dbe601c636d4cc109e28aeb9327361760a60383e57f54e2254e38b","source":{"kind":"arxiv","id":"2412.13549","version":2},"attestation_state":"computed","paper":{"title":"EscapeBench: Towards Advancing Creative Intelligence of Language Model Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bingxiang He, Cheng Qian, Denghui Zhang, Heng Ji, Hongyi Du, Jiarui Yao, Peixuan Han, Qinyu Luo, Xiaocheng Yang, Xiusi Chen, Yuji Zhang, Yunzhu Li","submitted_at":"2024-12-18T06:50:39Z","abstract_excerpt":"Language model agents excel in long-session planning and reasoning, but existing benchmarks primarily focus on goal-oriented tasks with explicit objectives, neglecting creative adaptation in unfamiliar environments. To address this, we introduce EscapeBench, a benchmark suite of room escape game environments designed to challenge agents with creative reasoning, unconventional tool use, and iterative problem-solving to uncover implicit goals. Our results show that current LM models, despite employing working memory and Chain-of-Thought reasoning, achieve only 15% average progress without hints,"},"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":"2412.13549","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-12-18T06:50:39Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"975351d4a72f5edd602fd0e9234c39931bc46d67fa9bd38be36e6316aa3d8c74","abstract_canon_sha256":"5ba0faef346c4e747a642d3d361844ba8cab33c653abb4796f627b7444a2d7a0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:08:39.444854Z","signature_b64":"CgD731gd5co4Yyw2NnfXN/7wqz3t7khNIsHyqNaMSEfzqCVDJwyCBg0akMK7Lyske/9Iz+0MXo0l4GC0iXpKCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eeb0cadf44dbe601c636d4cc109e28aeb9327361760a60383e57f54e2254e38b","last_reissued_at":"2026-07-05T11:08:39.444406Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:08:39.444406Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"EscapeBench: Towards Advancing Creative Intelligence of Language Model Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bingxiang He, Cheng Qian, Denghui Zhang, Heng Ji, Hongyi Du, Jiarui Yao, Peixuan Han, Qinyu Luo, Xiaocheng Yang, Xiusi Chen, Yuji Zhang, Yunzhu Li","submitted_at":"2024-12-18T06:50:39Z","abstract_excerpt":"Language model agents excel in long-session planning and reasoning, but existing benchmarks primarily focus on goal-oriented tasks with explicit objectives, neglecting creative adaptation in unfamiliar environments. To address this, we introduce EscapeBench, a benchmark suite of room escape game environments designed to challenge agents with creative reasoning, unconventional tool use, and iterative problem-solving to uncover implicit goals. Our results show that current LM models, despite employing working memory and Chain-of-Thought reasoning, achieve only 15% average progress without hints,"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.13549","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/2412.13549/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":"2412.13549","created_at":"2026-07-05T11:08:39.444463+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.13549v2","created_at":"2026-07-05T11:08:39.444463+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.13549","created_at":"2026-07-05T11:08:39.444463+00:00"},{"alias_kind":"pith_short_12","alias_value":"52YMVX2E3PTA","created_at":"2026-07-05T11:08:39.444463+00:00"},{"alias_kind":"pith_short_16","alias_value":"52YMVX2E3PTADRRW","created_at":"2026-07-05T11:08:39.444463+00:00"},{"alias_kind":"pith_short_8","alias_value":"52YMVX2E","created_at":"2026-07-05T11:08:39.444463+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2504.13958","citing_title":"ToolRL: Reward is All Tool Learning Needs","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/52YMVX2E3PTADRRW2TGBBHRIV2","json":"https://pith.science/pith/52YMVX2E3PTADRRW2TGBBHRIV2.json","graph_json":"https://pith.science/api/pith-number/52YMVX2E3PTADRRW2TGBBHRIV2/graph.json","events_json":"https://pith.science/api/pith-number/52YMVX2E3PTADRRW2TGBBHRIV2/events.json","paper":"https://pith.science/paper/52YMVX2E"},"agent_actions":{"view_html":"https://pith.science/pith/52YMVX2E3PTADRRW2TGBBHRIV2","download_json":"https://pith.science/pith/52YMVX2E3PTADRRW2TGBBHRIV2.json","view_paper":"https://pith.science/paper/52YMVX2E","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.13549&json=true","fetch_graph":"https://pith.science/api/pith-number/52YMVX2E3PTADRRW2TGBBHRIV2/graph.json","fetch_events":"https://pith.science/api/pith-number/52YMVX2E3PTADRRW2TGBBHRIV2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/52YMVX2E3PTADRRW2TGBBHRIV2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/52YMVX2E3PTADRRW2TGBBHRIV2/action/storage_attestation","attest_author":"https://pith.science/pith/52YMVX2E3PTADRRW2TGBBHRIV2/action/author_attestation","sign_citation":"https://pith.science/pith/52YMVX2E3PTADRRW2TGBBHRIV2/action/citation_signature","submit_replication":"https://pith.science/pith/52YMVX2E3PTADRRW2TGBBHRIV2/action/replication_record"}},"created_at":"2026-07-05T11:08:39.444463+00:00","updated_at":"2026-07-05T11:08:39.444463+00:00"}