{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:2ZTXNN37CZTH3F3SJPXCVLZYVX","short_pith_number":"pith:2ZTXNN37","canonical_record":{"source":{"id":"2606.23443","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-22T14:59:57Z","cross_cats_sorted":["cs.AI","physics.chem-ph"],"title_canon_sha256":"0d621d68b1f8e5f7e796828fc59e728a9f463f83952dd1adeb535291d3f3060d","abstract_canon_sha256":"2c90c7a6a1ddf27384f9db2466b9fc587dd0b8c7f41ca7180e30dab053a3059f"},"schema_version":"1.0"},"canonical_sha256":"d66776b77f16667d97724bee2aaf38adf49c7891030c7e816588ec5b41311245","source":{"kind":"arxiv","id":"2606.23443","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.23443","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"arxiv_version","alias_value":"2606.23443v1","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.23443","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"pith_short_12","alias_value":"2ZTXNN37CZTH","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"pith_short_16","alias_value":"2ZTXNN37CZTH3F3S","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"pith_short_8","alias_value":"2ZTXNN37","created_at":"2026-06-23T03:14:20Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:2ZTXNN37CZTH3F3SJPXCVLZYVX","target":"record","payload":{"canonical_record":{"source":{"id":"2606.23443","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-22T14:59:57Z","cross_cats_sorted":["cs.AI","physics.chem-ph"],"title_canon_sha256":"0d621d68b1f8e5f7e796828fc59e728a9f463f83952dd1adeb535291d3f3060d","abstract_canon_sha256":"2c90c7a6a1ddf27384f9db2466b9fc587dd0b8c7f41ca7180e30dab053a3059f"},"schema_version":"1.0"},"canonical_sha256":"d66776b77f16667d97724bee2aaf38adf49c7891030c7e816588ec5b41311245","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T03:14:20.104153Z","signature_b64":"RJrT9MeMXyLxsJkBqSycD92hBuZHOWwCkfD/advBfJX7pyYe38DXt1xMN4CJDVJrfvd4lt4GU6/yimRI+nqmDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d66776b77f16667d97724bee2aaf38adf49c7891030c7e816588ec5b41311245","last_reissued_at":"2026-06-23T03:14:20.103748Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T03:14:20.103748Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.23443","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-23T03:14:20Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"yAIEx9yIhlJBJqVvAYChmTJPkgi/EG9AxBIe7FKntMzPmpqIkew0N50KrekoRl88chxW7H3lSMFkrZ9JcDk3Ag==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-29T22:16:47.262769Z"},"content_sha256":"01fbd12ee958ed43e074873c8aae7073b1b9337d5c3436c7533e75179dc7c828","schema_version":"1.0","event_id":"sha256:01fbd12ee958ed43e074873c8aae7073b1b9337d5c3436c7533e75179dc7c828"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:2ZTXNN37CZTH3F3SJPXCVLZYVX","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"What Does a Chemical Language Model Know About Molecules?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","physics.chem-ph"],"primary_cat":"cs.LG","authors_text":"Christian Kenneth, Etowah Adams, Gerard JP van Westen, Liam Bai","submitted_at":"2026-06-22T14:59:57Z","abstract_excerpt":"Chemical language models (cLMs) are widely assumed to learn surface-level syntactic patterns rather than learning meaningful molecular semantics. Here, we apply sparse autoencoders (SAEs) to MolFormer, an encoder-only cLM, to mechanistically examine how molecular representations are built across layers. We discover that early layers rely on position-tracking latents to parse molecular grammar, while later layers encode atom-in-substructure and pharmacologically relevant features. Additionally, we show that non-canonical SMILES produce more disruptive representation shifts than invalid SMILES, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.23443","kind":"arxiv","version":1},"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/2606.23443/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-23T03:14:20Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"byRvN2pTnKYEnD2FqFd0isAOY4uktAQuTYYNiTiIoPm/d8Pb1ShCP173FGcm0Ht0lsXeaT57HQcEcINRtBh7Bg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-29T22:16:47.263137Z"},"content_sha256":"065c521fb13094efdb9c89fa5a0f8cc36dfeb8d66cc19f9c23415be78ecb0b25","schema_version":"1.0","event_id":"sha256:065c521fb13094efdb9c89fa5a0f8cc36dfeb8d66cc19f9c23415be78ecb0b25"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/2ZTXNN37CZTH3F3SJPXCVLZYVX/bundle.json","state_url":"https://pith.science/pith/2ZTXNN37CZTH3F3SJPXCVLZYVX/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/2ZTXNN37CZTH3F3SJPXCVLZYVX/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":"2026-06-29T22:16:47Z","links":{"resolver":"https://pith.science/pith/2ZTXNN37CZTH3F3SJPXCVLZYVX","bundle":"https://pith.science/pith/2ZTXNN37CZTH3F3SJPXCVLZYVX/bundle.json","state":"https://pith.science/pith/2ZTXNN37CZTH3F3SJPXCVLZYVX/state.json","well_known_bundle":"https://pith.science/.well-known/pith/2ZTXNN37CZTH3F3SJPXCVLZYVX/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:2ZTXNN37CZTH3F3SJPXCVLZYVX","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":"2c90c7a6a1ddf27384f9db2466b9fc587dd0b8c7f41ca7180e30dab053a3059f","cross_cats_sorted":["cs.AI","physics.chem-ph"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-22T14:59:57Z","title_canon_sha256":"0d621d68b1f8e5f7e796828fc59e728a9f463f83952dd1adeb535291d3f3060d"},"schema_version":"1.0","source":{"id":"2606.23443","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.23443","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"arxiv_version","alias_value":"2606.23443v1","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.23443","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"pith_short_12","alias_value":"2ZTXNN37CZTH","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"pith_short_16","alias_value":"2ZTXNN37CZTH3F3S","created_at":"2026-06-23T03:14:20Z"},{"alias_kind":"pith_short_8","alias_value":"2ZTXNN37","created_at":"2026-06-23T03:14:20Z"}],"graph_snapshots":[{"event_id":"sha256:065c521fb13094efdb9c89fa5a0f8cc36dfeb8d66cc19f9c23415be78ecb0b25","target":"graph","created_at":"2026-06-23T03:14:20Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2606.23443/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Chemical language models (cLMs) are widely assumed to learn surface-level syntactic patterns rather than learning meaningful molecular semantics. Here, we apply sparse autoencoders (SAEs) to MolFormer, an encoder-only cLM, to mechanistically examine how molecular representations are built across layers. We discover that early layers rely on position-tracking latents to parse molecular grammar, while later layers encode atom-in-substructure and pharmacologically relevant features. Additionally, we show that non-canonical SMILES produce more disruptive representation shifts than invalid SMILES, ","authors_text":"Christian Kenneth, Etowah Adams, Gerard JP van Westen, Liam Bai","cross_cats":["cs.AI","physics.chem-ph"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-22T14:59:57Z","title":"What Does a Chemical Language Model Know About Molecules?"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.23443","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:01fbd12ee958ed43e074873c8aae7073b1b9337d5c3436c7533e75179dc7c828","target":"record","created_at":"2026-06-23T03:14:20Z","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":"2c90c7a6a1ddf27384f9db2466b9fc587dd0b8c7f41ca7180e30dab053a3059f","cross_cats_sorted":["cs.AI","physics.chem-ph"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-22T14:59:57Z","title_canon_sha256":"0d621d68b1f8e5f7e796828fc59e728a9f463f83952dd1adeb535291d3f3060d"},"schema_version":"1.0","source":{"id":"2606.23443","kind":"arxiv","version":1}},"canonical_sha256":"d66776b77f16667d97724bee2aaf38adf49c7891030c7e816588ec5b41311245","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d66776b77f16667d97724bee2aaf38adf49c7891030c7e816588ec5b41311245","first_computed_at":"2026-06-23T03:14:20.103748Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-23T03:14:20.103748Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"RJrT9MeMXyLxsJkBqSycD92hBuZHOWwCkfD/advBfJX7pyYe38DXt1xMN4CJDVJrfvd4lt4GU6/yimRI+nqmDA==","signature_status":"signed_v1","signed_at":"2026-06-23T03:14:20.104153Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.23443","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:01fbd12ee958ed43e074873c8aae7073b1b9337d5c3436c7533e75179dc7c828","sha256:065c521fb13094efdb9c89fa5a0f8cc36dfeb8d66cc19f9c23415be78ecb0b25"],"state_sha256":"f5bfa61b1d8bb06dc93870c8e2994db446c6c3327705c2eb5b436ca4490fd1c6"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6rbQs8z1ToJS4M5OT/x5cAnaE1Uak0TE6nNXTGgjVlAXlXCDomtJMJ9gq6Xa0OIJw5DGFflP6jY1SC6pNzGaAg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-29T22:16:47.265054Z","bundle_sha256":"cde985a8b0829ea03334fbb645890a6f744767dd0c68d26c1115e294b69220a7"}}