{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2022:PFZHD3TFIWOEKN4ZU3CDZP3APP","short_pith_number":"pith:PFZHD3TF","canonical_record":{"source":{"id":"2211.15661","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-11-28T18:59:51Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"6caaadf80372d564f29b370e6c37cced5e9e052c272d7b54f59c819c6937e6d2","abstract_canon_sha256":"e8bab37bc0e56fc58a49bf89d5cf2bbb25839cce6a80f49b692916efa402136b"},"schema_version":"1.0"},"canonical_sha256":"797271ee65459c453799a6c43cbf607be37dc8edbd11f2091c68b86538df2cc0","source":{"kind":"arxiv","id":"2211.15661","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2211.15661","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"arxiv_version","alias_value":"2211.15661v3","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2211.15661","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"pith_short_12","alias_value":"PFZHD3TFIWOE","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"PFZHD3TFIWOEKN4Z","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"PFZHD3TF","created_at":"2026-05-18T12:33:33Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2022:PFZHD3TFIWOEKN4ZU3CDZP3APP","target":"record","payload":{"canonical_record":{"source":{"id":"2211.15661","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-11-28T18:59:51Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"6caaadf80372d564f29b370e6c37cced5e9e052c272d7b54f59c819c6937e6d2","abstract_canon_sha256":"e8bab37bc0e56fc58a49bf89d5cf2bbb25839cce6a80f49b692916efa402136b"},"schema_version":"1.0"},"canonical_sha256":"797271ee65459c453799a6c43cbf607be37dc8edbd11f2091c68b86538df2cc0","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:13.931043Z","signature_b64":"iXTn5qfVogpRzkUSBB5mxvpBlDxXc45GBFFGypaEHl5d/EiyMBEYqhr8pE81SAlMayUg5JI7ocKWQ45AgeQEDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"797271ee65459c453799a6c43cbf607be37dc8edbd11f2091c68b86538df2cc0","last_reissued_at":"2026-05-17T23:38:13.930507Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:13.930507Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2211.15661","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:13Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"a6APmRjkU3Pcd+CbbBGUvFBdA6luY7R2K6B1hDTzqILbNvyYZULNAymzZdGyMnXypokH9mN6QKBt/8WCAyPECw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-20T05:31:14.735929Z"},"content_sha256":"b1b8576dcb92128468b00555563abe30ea7c3bce161589a3155bef7a6dfae522","schema_version":"1.0","event_id":"sha256:b1b8576dcb92128468b00555563abe30ea7c3bce161589a3155bef7a6dfae522"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2022:PFZHD3TFIWOEKN4ZU3CDZP3APP","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"What learning algorithm is in-context learning? Investigations with linear models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Transformers implement gradient descent and ridge regression implicitly when doing in-context learning on linear tasks.","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Dale Schuurmans, Denny Zhou, Ekin Aky\\\"urek, Jacob Andreas, Tengyu Ma","submitted_at":"2022-11-28T18:59:51Z","abstract_excerpt":"Neural sequence models, especially transformers, exhibit a remarkable capacity for in-context learning. They can construct new predictors from sequences of labeled examples $(x, f(x))$ presented in the input without further parameter updates. We investigate the hypothesis that transformer-based in-context learners implement standard learning algorithms implicitly, by encoding smaller models in their activations, and updating these implicit models as new examples appear in the context. Using linear regression as a prototypical problem, we offer three sources of evidence for this hypothesis. Fir"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Trained in-context learners closely match the predictors computed by gradient descent, ridge regression, and exact least-squares regression, transitioning between different predictors as transformer depth and dataset noise vary, and converging to Bayesian estimators for large widths and depths.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That results on linear regression as a prototypical problem will extend to the more complex, non-linear tasks typical of real in-context learning in language models.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Transformers performing in-context learning implicitly implement gradient descent, ridge regression, and least-squares predictors for linear models, with behavior shifting based on model depth, width, and data noise.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Transformers implement gradient descent and ridge regression implicitly when doing in-context learning on linear tasks.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ec988c53f7351edc99d0a73d498306e283a9c543a189fdf77b8434b3bdc8d8e9"},"source":{"id":"2211.15661","kind":"arxiv","version":3},"verdict":{"id":"49e18b99-281f-43c3-890a-82ea371269f5","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T13:34:58.805779Z","strongest_claim":"Trained in-context learners closely match the predictors computed by gradient descent, ridge regression, and exact least-squares regression, transitioning between different predictors as transformer depth and dataset noise vary, and converging to Bayesian estimators for large widths and depths.","one_line_summary":"Transformers performing in-context learning implicitly implement gradient descent, ridge regression, and least-squares predictors for linear models, with behavior shifting based on model depth, width, and data noise.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That results on linear regression as a prototypical problem will extend to the more complex, non-linear tasks typical of real in-context learning in language models.","pith_extraction_headline":"Transformers implement gradient descent and ridge regression implicitly when doing in-context learning on linear tasks."},"references":{"count":31,"sample":[{"doi":"","year":2016,"title":"Understanding intermediate layers using linear classifier probes","work_id":"bdc944db-4be2-44f7-950b-eaef12fab00e","ref_index":1,"cited_arxiv_id":"1610.01644","is_internal_anchor":true},{"doi":"","year":2016,"title":"Hoffman, David Pfau, Tom Schaul, and Nando de Freitas","work_id":"82defb4b-6d1f-4494-9004-339e87d67d22","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"Layer Normalization","work_id":"20a2d720-0046-4c7c-bcd6-327ec8143f69","ref_index":3,"cited_arxiv_id":"1607.06450","is_internal_anchor":true},{"doi":"","year":2020,"title":"Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert - Voss, Gretc","work_id":"9f711468-19ce-4b64-a999-bb8e3e934e66","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Thread: circuits","work_id":"d9cad1f6-2e0a-4543-b594-9bf6a98592a4","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":31,"snapshot_sha256":"95c00cd09b87214b570278c5b5195ba3abf0a61b2cc2cc5439201f91320208a0","internal_anchors":6},"formal_canon":{"evidence_count":1,"snapshot_sha256":"6f2385ee3ca73ec023a6e04cdedc30a4c5b02f3ae2b55e4abffc5fe10fa38c47"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"49e18b99-281f-43c3-890a-82ea371269f5"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:13Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"jlUDg1OPn7HiIlSRmDMzH/tS7kxJSSn3+QM/bo9g0TLoxCWk+qUKCx/DnrQuxQvE6zumLyDkdVmjm8PXE0XmDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-20T05:31:14.737020Z"},"content_sha256":"21f3c6b5f40e97f96ff4aa67da9138cf88e793a45a8b785fa11efcb685b082da","schema_version":"1.0","event_id":"sha256:21f3c6b5f40e97f96ff4aa67da9138cf88e793a45a8b785fa11efcb685b082da"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/PFZHD3TFIWOEKN4ZU3CDZP3APP/bundle.json","state_url":"https://pith.science/pith/PFZHD3TFIWOEKN4ZU3CDZP3APP/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/PFZHD3TFIWOEKN4ZU3CDZP3APP/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-20T05:31:14Z","links":{"resolver":"https://pith.science/pith/PFZHD3TFIWOEKN4ZU3CDZP3APP","bundle":"https://pith.science/pith/PFZHD3TFIWOEKN4ZU3CDZP3APP/bundle.json","state":"https://pith.science/pith/PFZHD3TFIWOEKN4ZU3CDZP3APP/state.json","well_known_bundle":"https://pith.science/.well-known/pith/PFZHD3TFIWOEKN4ZU3CDZP3APP/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:PFZHD3TFIWOEKN4ZU3CDZP3APP","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"e8bab37bc0e56fc58a49bf89d5cf2bbb25839cce6a80f49b692916efa402136b","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-11-28T18:59:51Z","title_canon_sha256":"6caaadf80372d564f29b370e6c37cced5e9e052c272d7b54f59c819c6937e6d2"},"schema_version":"1.0","source":{"id":"2211.15661","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2211.15661","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"arxiv_version","alias_value":"2211.15661v3","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2211.15661","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"pith_short_12","alias_value":"PFZHD3TFIWOE","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"PFZHD3TFIWOEKN4Z","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"PFZHD3TF","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:21f3c6b5f40e97f96ff4aa67da9138cf88e793a45a8b785fa11efcb685b082da","target":"graph","created_at":"2026-05-17T23:38:13Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Trained in-context learners closely match the predictors computed by gradient descent, ridge regression, and exact least-squares regression, transitioning between different predictors as transformer depth and dataset noise vary, and converging to Bayesian estimators for large widths and depths."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That results on linear regression as a prototypical problem will extend to the more complex, non-linear tasks typical of real in-context learning in language models."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Transformers performing in-context learning implicitly implement gradient descent, ridge regression, and least-squares predictors for linear models, with behavior shifting based on model depth, width, and data noise."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Transformers implement gradient descent and ridge regression implicitly when doing in-context learning on linear tasks."}],"snapshot_sha256":"ec988c53f7351edc99d0a73d498306e283a9c543a189fdf77b8434b3bdc8d8e9"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"6f2385ee3ca73ec023a6e04cdedc30a4c5b02f3ae2b55e4abffc5fe10fa38c47"},"paper":{"abstract_excerpt":"Neural sequence models, especially transformers, exhibit a remarkable capacity for in-context learning. They can construct new predictors from sequences of labeled examples $(x, f(x))$ presented in the input without further parameter updates. We investigate the hypothesis that transformer-based in-context learners implement standard learning algorithms implicitly, by encoding smaller models in their activations, and updating these implicit models as new examples appear in the context. Using linear regression as a prototypical problem, we offer three sources of evidence for this hypothesis. Fir","authors_text":"Dale Schuurmans, Denny Zhou, Ekin Aky\\\"urek, Jacob Andreas, Tengyu Ma","cross_cats":["cs.CL"],"headline":"Transformers implement gradient descent and ridge regression implicitly when doing in-context learning on linear tasks.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-11-28T18:59:51Z","title":"What learning algorithm is in-context learning? Investigations with linear models"},"references":{"count":31,"internal_anchors":6,"resolved_work":31,"sample":[{"cited_arxiv_id":"1610.01644","doi":"","is_internal_anchor":true,"ref_index":1,"title":"Understanding intermediate layers using linear classifier probes","work_id":"bdc944db-4be2-44f7-950b-eaef12fab00e","year":2016},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Hoffman, David Pfau, Tom Schaul, and Nando de Freitas","work_id":"82defb4b-6d1f-4494-9004-339e87d67d22","year":2016},{"cited_arxiv_id":"1607.06450","doi":"","is_internal_anchor":true,"ref_index":3,"title":"Layer Normalization","work_id":"20a2d720-0046-4c7c-bcd6-327ec8143f69","year":2016},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert - Voss, Gretc","work_id":"9f711468-19ce-4b64-a999-bb8e3e934e66","year":2020},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Thread: circuits","work_id":"d9cad1f6-2e0a-4543-b594-9bf6a98592a4","year":2020}],"snapshot_sha256":"95c00cd09b87214b570278c5b5195ba3abf0a61b2cc2cc5439201f91320208a0"},"source":{"id":"2211.15661","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-17T13:34:58.805779Z","id":"49e18b99-281f-43c3-890a-82ea371269f5","model_set":{"reader":"grok-4.3"},"one_line_summary":"Transformers performing in-context learning implicitly implement gradient descent, ridge regression, and least-squares predictors for linear models, with behavior shifting based on model depth, width, and data noise.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Transformers implement gradient descent and ridge regression implicitly when doing in-context learning on linear tasks.","strongest_claim":"Trained in-context learners closely match the predictors computed by gradient descent, ridge regression, and exact least-squares regression, transitioning between different predictors as transformer depth and dataset noise vary, and converging to Bayesian estimators for large widths and depths.","weakest_assumption":"That results on linear regression as a prototypical problem will extend to the more complex, non-linear tasks typical of real in-context learning in language models."}},"verdict_id":"49e18b99-281f-43c3-890a-82ea371269f5"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:b1b8576dcb92128468b00555563abe30ea7c3bce161589a3155bef7a6dfae522","target":"record","created_at":"2026-05-17T23:38:13Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"e8bab37bc0e56fc58a49bf89d5cf2bbb25839cce6a80f49b692916efa402136b","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-11-28T18:59:51Z","title_canon_sha256":"6caaadf80372d564f29b370e6c37cced5e9e052c272d7b54f59c819c6937e6d2"},"schema_version":"1.0","source":{"id":"2211.15661","kind":"arxiv","version":3}},"canonical_sha256":"797271ee65459c453799a6c43cbf607be37dc8edbd11f2091c68b86538df2cc0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"797271ee65459c453799a6c43cbf607be37dc8edbd11f2091c68b86538df2cc0","first_computed_at":"2026-05-17T23:38:13.930507Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:13.930507Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"iXTn5qfVogpRzkUSBB5mxvpBlDxXc45GBFFGypaEHl5d/EiyMBEYqhr8pE81SAlMayUg5JI7ocKWQ45AgeQEDA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:13.931043Z","signed_message":"canonical_sha256_bytes"},"source_id":"2211.15661","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b1b8576dcb92128468b00555563abe30ea7c3bce161589a3155bef7a6dfae522","sha256:21f3c6b5f40e97f96ff4aa67da9138cf88e793a45a8b785fa11efcb685b082da"],"state_sha256":"004435dd79f8d47debd668ffc526e8e3556f94f20a3c7832a74fff80d5a45e69"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"nr60rdkoGQAa0/Tw7vwmxfEf2J/7N7bdsAU6JhU7FOsXsk0ddEqC5GhR9AE6bAHZ46wM7ge8JLE+wrckL3urBA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-20T05:31:14.742488Z","bundle_sha256":"9313c0abc1240ac20a92cb7382cbbc5b0ff3973c350cc21dd4f69868404f4f28"}}