{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:4GEIPNFJHGX6GROLZI5OCZPFDU","short_pith_number":"pith:4GEIPNFJ","schema_version":"1.0","canonical_sha256":"e18887b4a939afe345cbca3ae165e51d01d748e93dae4b18886e417628b44fe4","source":{"kind":"arxiv","id":"2304.15004","version":2},"attestation_state":"computed","paper":{"title":"Are Emergent Abilities of Large Language Models a Mirage?","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Brando Miranda, Rylan Schaeffer, Sanmi Koyejo","submitted_at":"2023-04-28T17:52:11Z","abstract_excerpt":"Recent work claims that large language models display emergent abilities, abilities not present in smaller-scale models that are present in larger-scale models. What makes emergent abilities intriguing is two-fold: their sharpness, transitioning seemingly instantaneously from not present to present, and their unpredictability, appearing at seemingly unforeseeable model scales. Here, we present an alternative explanation for emergent abilities: that for a particular task and model family, when analyzing fixed model outputs, emergent abilities appear due to the researcher's choice of metric rath"},"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":"2304.15004","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.AI","submitted_at":"2023-04-28T17:52:11Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"f3b6a1d30fe0f1119ef3bcecd0dd8f85c7d01a8c0c1961a8c1eaa64e7fa54c1f","abstract_canon_sha256":"2988666406454e5d448ea6a888d691d09ecda833d0b37e8043f4d59092b7be69"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:12:24.415195Z","signature_b64":"ZPIYYhr5/liZaVMEulA/3OkwSXDhEmxFMfIYriHCv5cZfmVVKRMCdOj03moEQ9M091xtNwtQdKUeI9zhufJ0Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e18887b4a939afe345cbca3ae165e51d01d748e93dae4b18886e417628b44fe4","last_reissued_at":"2026-07-05T06:12:24.414726Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:12:24.414726Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Are Emergent Abilities of Large Language Models a Mirage?","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Brando Miranda, Rylan Schaeffer, Sanmi Koyejo","submitted_at":"2023-04-28T17:52:11Z","abstract_excerpt":"Recent work claims that large language models display emergent abilities, abilities not present in smaller-scale models that are present in larger-scale models. What makes emergent abilities intriguing is two-fold: their sharpness, transitioning seemingly instantaneously from not present to present, and their unpredictability, appearing at seemingly unforeseeable model scales. Here, we present an alternative explanation for emergent abilities: that for a particular task and model family, when analyzing fixed model outputs, emergent abilities appear due to the researcher's choice of metric rath"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.15004","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/2304.15004/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":"2304.15004","created_at":"2026-07-05T06:12:24.414786+00:00"},{"alias_kind":"arxiv_version","alias_value":"2304.15004v2","created_at":"2026-07-05T06:12:24.414786+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.15004","created_at":"2026-07-05T06:12:24.414786+00:00"},{"alias_kind":"pith_short_12","alias_value":"4GEIPNFJHGX6","created_at":"2026-07-05T06:12:24.414786+00:00"},{"alias_kind":"pith_short_16","alias_value":"4GEIPNFJHGX6GROL","created_at":"2026-07-05T06:12:24.414786+00:00"},{"alias_kind":"pith_short_8","alias_value":"4GEIPNFJ","created_at":"2026-07-05T06:12:24.414786+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":27,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26050","citing_title":"Natural Ungrokking: Asymmetric Control of Which Rules Survive Pretraining","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2606.25010","citing_title":"Emergent Capabilities Arise Randomly from Learning Sparse Attention Patterns","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.24119","citing_title":"When Top-1 Fails: Calibrating LoRA Monitors for Masked Diffusion LMs","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2607.02464","citing_title":"Will Scaling Improve Social Simulation with LLMs?","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2606.10877","citing_title":"XtrAIn: Training-Guided Occlusion for Feature Attribution","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00913","citing_title":"Two AI Metrics Diverged: Will it Make All the Difference?","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05106","citing_title":"Arithmetic Pedagogy for Language Models","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02953","citing_title":"Linguistic Productivity in Large Language Models: Models Coerce, but do not Preempt","ref_index":179,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28235","citing_title":"Govern the Repository, Not the Agent: Measuring Ecosystem-Level Risk in AI-Native Software","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2606.30815","citing_title":"When transformers learn \"impossible\" languages, what do they learn?","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07568","citing_title":"A Systematic Study of Behavioral Cloning for Scientific Data Annotation","ref_index":215,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27914","citing_title":"Does Capability Transfer to Subjective Behavior -- and Would Our Instruments Tell Us? A Self-Evolving, Trust-by-Construction Evaluation Paradigm","ref_index":100,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07623","citing_title":"Finite Certificates for In-Context Determinacy and a Threshold Theory of Emergence in Language Models","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23312","citing_title":"Towards Generalizable and Efficient Large-Scale Generative Recommenders","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20441","citing_title":"Weight Decay Regimes in Grokking Transformers: Cheap Online Diagnostics","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2601.14053","citing_title":"LLMOrbit: A Circular Taxonomy of Large Language Models -From Scaling Walls to Agentic AI Systems","ref_index":132,"is_internal_anchor":false},{"citing_arxiv_id":"2309.16797","citing_title":"Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution","ref_index":286,"is_internal_anchor":false},{"citing_arxiv_id":"2603.00883","citing_title":"Knowledge without Wisdom: Measuring Misalignment between LLMs and Intended Impact","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12809","citing_title":"Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spaces","ref_index":258,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04539","citing_title":"RLearner-LLM: Balancing Logical Grounding and Fluency in Large Language Models via Hybrid Direct Preference Optimization","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04539","citing_title":"RLearner-LLM: Balancing Logical Grounding and Fluency in Large Language Models via Hybrid Direct Preference Optimization","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08904","citing_title":"OPT-BENCH: Evaluating the Iterative Self-Optimization of LLM Agents in Large-Scale Search Spaces","ref_index":91,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04539","citing_title":"RLearner-LLM: Balancing Logical Grounding and Fluency in Large Language Models via Hybrid Direct Preference Optimization","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01420","citing_title":"Artificial Jagged Intelligence as Uneven Optimization Energy Allocation Capability Concentration, Redistribution, and Optimization Governance","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01134","citing_title":"To Use AI as Dice of Possibilities with Timing Computation","ref_index":64,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4GEIPNFJHGX6GROLZI5OCZPFDU","json":"https://pith.science/pith/4GEIPNFJHGX6GROLZI5OCZPFDU.json","graph_json":"https://pith.science/api/pith-number/4GEIPNFJHGX6GROLZI5OCZPFDU/graph.json","events_json":"https://pith.science/api/pith-number/4GEIPNFJHGX6GROLZI5OCZPFDU/events.json","paper":"https://pith.science/paper/4GEIPNFJ"},"agent_actions":{"view_html":"https://pith.science/pith/4GEIPNFJHGX6GROLZI5OCZPFDU","download_json":"https://pith.science/pith/4GEIPNFJHGX6GROLZI5OCZPFDU.json","view_paper":"https://pith.science/paper/4GEIPNFJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2304.15004&json=true","fetch_graph":"https://pith.science/api/pith-number/4GEIPNFJHGX6GROLZI5OCZPFDU/graph.json","fetch_events":"https://pith.science/api/pith-number/4GEIPNFJHGX6GROLZI5OCZPFDU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4GEIPNFJHGX6GROLZI5OCZPFDU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4GEIPNFJHGX6GROLZI5OCZPFDU/action/storage_attestation","attest_author":"https://pith.science/pith/4GEIPNFJHGX6GROLZI5OCZPFDU/action/author_attestation","sign_citation":"https://pith.science/pith/4GEIPNFJHGX6GROLZI5OCZPFDU/action/citation_signature","submit_replication":"https://pith.science/pith/4GEIPNFJHGX6GROLZI5OCZPFDU/action/replication_record"}},"created_at":"2026-07-05T06:12:24.414786+00:00","updated_at":"2026-07-05T06:12:24.414786+00:00"}