{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:CIBIAHX5EOBV2RIILUWM6IWRCA","short_pith_number":"pith:CIBIAHX5","schema_version":"1.0","canonical_sha256":"1202801efd23835d45085d2ccf22d1100837659189ebb3ed07dd06bb027d766a","source":{"kind":"arxiv","id":"2401.00448","version":3},"attestation_state":"computed","paper":{"title":"Beyond Chinchilla-Optimal: Accounting for Inference in Language Model Scaling Laws","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Jacob Portes, Jonathan Frankle, Nikhil Sardana, Sasha Doubov","submitted_at":"2023-12-31T10:53:58Z","abstract_excerpt":"Large language model (LLM) scaling laws are empirical formulas that estimate changes in model quality as a result of increasing parameter count and training data. However, these formulas, including the popular Deepmind Chinchilla scaling laws, neglect to include the cost of inference. We modify the Chinchilla scaling laws to calculate the optimal LLM parameter count and pre-training data size to train and deploy a model of a given quality and inference demand. We conduct our analysis both in terms of a compute budget and real-world costs and find that LLM researchers expecting reasonably large"},"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":"2401.00448","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-12-31T10:53:58Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"5bf9c7213374f602d26906e4029340061839884072c3ddf692d45df36ef45f8b","abstract_canon_sha256":"0dd2dcc7cea0ff1bb04b881dadb39171f13d1dde51d3e55a84d05c391f2c39bb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:48:14.220086Z","signature_b64":"sOHRFMA0cIsjPXmLncKOI37qNLVBqmnQ9lipnVnIVX33yIPZF3NHaWbvfySkQx0Vjz43/JOT0FS9N9Q2UfDMBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1202801efd23835d45085d2ccf22d1100837659189ebb3ed07dd06bb027d766a","last_reissued_at":"2026-07-05T10:48:14.219625Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:48:14.219625Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Beyond Chinchilla-Optimal: Accounting for Inference in Language Model Scaling Laws","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Jacob Portes, Jonathan Frankle, Nikhil Sardana, Sasha Doubov","submitted_at":"2023-12-31T10:53:58Z","abstract_excerpt":"Large language model (LLM) scaling laws are empirical formulas that estimate changes in model quality as a result of increasing parameter count and training data. However, these formulas, including the popular Deepmind Chinchilla scaling laws, neglect to include the cost of inference. We modify the Chinchilla scaling laws to calculate the optimal LLM parameter count and pre-training data size to train and deploy a model of a given quality and inference demand. We conduct our analysis both in terms of a compute budget and real-world costs and find that LLM researchers expecting reasonably large"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.00448","kind":"arxiv","version":3},"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/2401.00448/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":"2401.00448","created_at":"2026-07-05T10:48:14.219680+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.00448v3","created_at":"2026-07-05T10:48:14.219680+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.00448","created_at":"2026-07-05T10:48:14.219680+00:00"},{"alias_kind":"pith_short_12","alias_value":"CIBIAHX5EOBV","created_at":"2026-07-05T10:48:14.219680+00:00"},{"alias_kind":"pith_short_16","alias_value":"CIBIAHX5EOBV2RII","created_at":"2026-07-05T10:48:14.219680+00:00"},{"alias_kind":"pith_short_8","alias_value":"CIBIAHX5","created_at":"2026-07-05T10:48:14.219680+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":16,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25986","citing_title":"The Inference-Compute Frontier and a Latency-Efficient Architecture for Limit Order Book Prediction","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2606.22189","citing_title":"L20-Edu-135M: An Auditable Single-GPU Study of Data-Efficient Small Language Modeling","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18144","citing_title":"Memory as a Wasting Asset: Pricing Flash Endurance for Embodied Agents, and the Limits of Doing So","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2606.15079","citing_title":"Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale","ref_index":68,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05610","citing_title":"Predictable Scaling Laws of Optimal Hyperparameters for LLM Continued Pre-training","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2501.05465","citing_title":"Small Language Models (SLMs) Can Still Pack a Punch: A survey (updated 2026)","ref_index":112,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19195","citing_title":"The Thermodynamic Costs of Simple Linear Regression","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2505.15134","citing_title":"The Unreasonable Effectiveness of Entropy Minimization in LLM Reasoning","ref_index":66,"is_internal_anchor":false},{"citing_arxiv_id":"2510.18245","citing_title":"Scaling Laws Meet Model Architecture: Toward Inference-Efficient LLMs","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12520","citing_title":"BoostTaxo: Zero-Shot Taxonomy Induction via Boosting-Style Agentic Reasoning and Constraint-Aware Calibration","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2404.06395","citing_title":"MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2403.04652","citing_title":"Yi: Open Foundation Models by 01.AI","ref_index":66,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06116","citing_title":"Policy-Guided Stepwise Model Routing for Cost-Effective Reasoning","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01111","citing_title":"When Less is Enough: Efficient Inference via Collaborative Reasoning","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18002","citing_title":"Neural Garbage Collection: Learning to Forget while Learning to Reason","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02853","citing_title":"Trust, but Verify: Peeling Low-Bit Transformer Networks for Training Monitoring","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CIBIAHX5EOBV2RIILUWM6IWRCA","json":"https://pith.science/pith/CIBIAHX5EOBV2RIILUWM6IWRCA.json","graph_json":"https://pith.science/api/pith-number/CIBIAHX5EOBV2RIILUWM6IWRCA/graph.json","events_json":"https://pith.science/api/pith-number/CIBIAHX5EOBV2RIILUWM6IWRCA/events.json","paper":"https://pith.science/paper/CIBIAHX5"},"agent_actions":{"view_html":"https://pith.science/pith/CIBIAHX5EOBV2RIILUWM6IWRCA","download_json":"https://pith.science/pith/CIBIAHX5EOBV2RIILUWM6IWRCA.json","view_paper":"https://pith.science/paper/CIBIAHX5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.00448&json=true","fetch_graph":"https://pith.science/api/pith-number/CIBIAHX5EOBV2RIILUWM6IWRCA/graph.json","fetch_events":"https://pith.science/api/pith-number/CIBIAHX5EOBV2RIILUWM6IWRCA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CIBIAHX5EOBV2RIILUWM6IWRCA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CIBIAHX5EOBV2RIILUWM6IWRCA/action/storage_attestation","attest_author":"https://pith.science/pith/CIBIAHX5EOBV2RIILUWM6IWRCA/action/author_attestation","sign_citation":"https://pith.science/pith/CIBIAHX5EOBV2RIILUWM6IWRCA/action/citation_signature","submit_replication":"https://pith.science/pith/CIBIAHX5EOBV2RIILUWM6IWRCA/action/replication_record"}},"created_at":"2026-07-05T10:48:14.219680+00:00","updated_at":"2026-07-05T10:48:14.219680+00:00"}