{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:PWQKCOUNY6MMWGGOYGP7SLLWU7","short_pith_number":"pith:PWQKCOUN","schema_version":"1.0","canonical_sha256":"7da0a13a8dc798cb18cec19ff92d76a7d04d8ce30eaef8b8b1447bc7561cd47c","source":{"kind":"arxiv","id":"2305.19420","version":2},"attestation_state":"computed","paper":{"title":"What and How does In-Context Learning Learn? Bayesian Model Averaging, Parameterization, and Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Fengzhuo Zhang, Yufeng Zhang, Zhaoran Wang, Zhuoran Yang","submitted_at":"2023-05-30T21:23:47Z","abstract_excerpt":"In this paper, we conduct a comprehensive study of In-Context Learning (ICL) by addressing several open questions: (a) What type of ICL estimator is learned by large language models? (b) What is a proper performance metric for ICL and what is the error rate? (c) How does the transformer architecture enable ICL? To answer these questions, we adopt a Bayesian view and formulate ICL as a problem of predicting the response corresponding to the current covariate, given a number of examples drawn from a latent variable model. To answer (a), we show that, without updating the neural network parameter"},"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":"2305.19420","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2023-05-30T21:23:47Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"9eea85245d347f7a026846e822751260286d2603410a052b87a8a40dd5f0a8e8","abstract_canon_sha256":"487264a40b4a18e1ee55b009d589521edae85fb068117c58b726b14976fc10e0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:58:57.090888Z","signature_b64":"waZa1bDuXdHmNLWDOXna6h8e8YFzdO/xFclM/jBPRLuiddh1fRMkESmSRM0hkVyYudBJh/eMxrI9ZFg66vaTDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7da0a13a8dc798cb18cec19ff92d76a7d04d8ce30eaef8b8b1447bc7561cd47c","last_reissued_at":"2026-07-05T06:58:57.090395Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:58:57.090395Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What and How does In-Context Learning Learn? Bayesian Model Averaging, Parameterization, and Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Fengzhuo Zhang, Yufeng Zhang, Zhaoran Wang, Zhuoran Yang","submitted_at":"2023-05-30T21:23:47Z","abstract_excerpt":"In this paper, we conduct a comprehensive study of In-Context Learning (ICL) by addressing several open questions: (a) What type of ICL estimator is learned by large language models? (b) What is a proper performance metric for ICL and what is the error rate? (c) How does the transformer architecture enable ICL? To answer these questions, we adopt a Bayesian view and formulate ICL as a problem of predicting the response corresponding to the current covariate, given a number of examples drawn from a latent variable model. To answer (a), we show that, without updating the neural network parameter"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.19420","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/2305.19420/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":"2305.19420","created_at":"2026-07-05T06:58:57.090453+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.19420v2","created_at":"2026-07-05T06:58:57.090453+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.19420","created_at":"2026-07-05T06:58:57.090453+00:00"},{"alias_kind":"pith_short_12","alias_value":"PWQKCOUNY6MM","created_at":"2026-07-05T06:58:57.090453+00:00"},{"alias_kind":"pith_short_16","alias_value":"PWQKCOUNY6MMWGGO","created_at":"2026-07-05T06:58:57.090453+00:00"},{"alias_kind":"pith_short_8","alias_value":"PWQKCOUN","created_at":"2026-07-05T06:58:57.090453+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17832","citing_title":"From Drift to Coherence: Stabilizing Beliefs in LLMs","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08068","citing_title":"DICE: Entropy-Regularized Equilibrium Selection for Stable Multi-Agent LLM Coordination","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2502.17421","citing_title":"LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20730","citing_title":"Distributional Alignment as a Criterion for Designing Task Vectors in In-Context Learning","ref_index":54,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18979","citing_title":"TabQL: In-Context Q-Learning with Tabular Foundation Models","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12316","citing_title":"Autoregressive Learning in Joint KL: Sharp Oracle Bounds and Lower Bounds","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12412","citing_title":"Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05115","citing_title":"Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior","ref_index":237,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PWQKCOUNY6MMWGGOYGP7SLLWU7","json":"https://pith.science/pith/PWQKCOUNY6MMWGGOYGP7SLLWU7.json","graph_json":"https://pith.science/api/pith-number/PWQKCOUNY6MMWGGOYGP7SLLWU7/graph.json","events_json":"https://pith.science/api/pith-number/PWQKCOUNY6MMWGGOYGP7SLLWU7/events.json","paper":"https://pith.science/paper/PWQKCOUN"},"agent_actions":{"view_html":"https://pith.science/pith/PWQKCOUNY6MMWGGOYGP7SLLWU7","download_json":"https://pith.science/pith/PWQKCOUNY6MMWGGOYGP7SLLWU7.json","view_paper":"https://pith.science/paper/PWQKCOUN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.19420&json=true","fetch_graph":"https://pith.science/api/pith-number/PWQKCOUNY6MMWGGOYGP7SLLWU7/graph.json","fetch_events":"https://pith.science/api/pith-number/PWQKCOUNY6MMWGGOYGP7SLLWU7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PWQKCOUNY6MMWGGOYGP7SLLWU7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PWQKCOUNY6MMWGGOYGP7SLLWU7/action/storage_attestation","attest_author":"https://pith.science/pith/PWQKCOUNY6MMWGGOYGP7SLLWU7/action/author_attestation","sign_citation":"https://pith.science/pith/PWQKCOUNY6MMWGGOYGP7SLLWU7/action/citation_signature","submit_replication":"https://pith.science/pith/PWQKCOUNY6MMWGGOYGP7SLLWU7/action/replication_record"}},"created_at":"2026-07-05T06:58:57.090453+00:00","updated_at":"2026-07-05T06:58:57.090453+00:00"}