{"_canonicalization":{"envelope_id":"axm_ + sha256(envelope minus {signature, axiom_id, anchors})","envelope_signature":"ed25519(envelope minus {signature, axiom_id})","json":"sort_keys=True, separators=(',',':'), ensure_ascii=False, allow_nan=False, utf-8","leaf_hash":"sha256(0x00 || canonical_json(envelope_full))","seal_signature":"ed25519(seal minus {signature, sig_algorithm})"},"axiom_id":"axm_3fce9d96f66888c8cf2af4e8c2767a3247d6c24a7db8a86be2dc6d539ec7dcb7","bitcoin_anchor":{"bitcoin_attestations":[],"calendar_attestations":[],"ots_url":"","stamped_at":"","status":"pending_next_stamp"},"envelope":{"anchors":[{"chain":"crovia.axiom_graph","height":0,"merkle_proof":"spider_vendor_press_v1","root_at_anchor":"spider_vendor_press_v1"}],"axiom_id":"axm_3fce9d96f66888c8cf2af4e8c2767a3247d6c24a7db8a86be2dc6d539ec7dcb7","axiom_type":"AX.OBS","body":{"axiom_subtype":"news.vendor_press.v1","category":"news","fingerprint":"d0eef1f7e6c862c9bbc4d5443e753fcca9bf4457ed16010411e358be521855ef","published":"Tue, 30 Jun 2026 00:00:00 -0400","receipt_hash":"d0eef1f7e6c862c9bbc4d5443e753fcca9bf4457ed16010411e358be521855ef","schema":"spider.news.vendor_press.v1","spider":"vendor_press","spider_record":{"axiom_subtype":"news.vendor_press.v1","category":"news","decision_hint":"POSITIVE","envelope_target":"AX.OBS","fingerprint":"d0eef1f7e6c862c9bbc4d5443e753fcca9bf4457ed16010411e358be521855ef","observed_at":"2026-06-30T04:43:04.087680Z","parent_run_hash":"74f7ab392cc702044101fe24a76a2fdad11164cd79ce725aad6c446a477e89c5","published":"Tue, 30 Jun 2026 00:00:00 -0400","runtime_version":"0.1.0","schema":"spider.news.vendor_press.v1","source_status":200,"source_url":"https://export.arxiv.org/rss/cs.AI","spider":"vendor_press","summary_excerpt":"arXiv:2606.19317v2 Announce Type: replace-cross \nAbstract: A longstanding goal of research on interpretable deep learning is to replace opaque neural computations with human-meaningful symbolic descriptions. In this paper, we propose an approach for approximating the behavior of components of deep networks with executable programs. We focus on attention heads in transformer language models. For a given head, we first compute its associated attention matrices on a collection of randomly selected training examples. Next, we prompt a pre-trained language model with a summary of these matrices, and instruct it to generate a set of Python programs that can reproduce the associated attention patterns given only text from the input sentence. Finally, we re-rank programs according to how well our final set of programs predict behavior on held-out inputs. We demonstrate that a set of fewer than 1,000 such generated programs can reproduce the attention patterns of heads in GPT-2, TinyLlama-1.1B,","title":"Explaining Attention with Program Synthesis","url":"https://arxiv.org/abs/2606.19317","vendor":"arxiv_cs_ai"},"summary":"arXiv:2606.19317v2 Announce Type: replace-cross \nAbstract: A longstanding goal of research on interpretable deep learning is to replace opaque neural computations with human-meaningful symbolic descriptions. In this paper, we propose an approach for approximating the behavior of components of deep networks with executable programs. We focus on attention heads in transformer language models. For a given head, we first compute its associated attention matrices on a collection of randomly selected training examples. Next, we prompt a pre-trained language model with a summary of these matrices, and instruct it to generate a set of Python programs that can reproduce the associated attention patterns given only text from the input sentence. Finally, we re-rank programs according to how well our final set of programs predict behavior on held-out inputs. We demonstrate that a set of fewer than 1,000 such generated programs can reproduce the attention patterns of heads in GPT-2, TinyLlama-1.1B,","title":"Explaining Attention with Program Synthesis","vendor":"arxiv_cs_ai"},"confidence":{"method":"deterministic"},"decision":"POSITIVE","issued_at":"2026-06-30T04:43:04Z","notes":"Spider vendor_press (news) news.vendor_press.v1","object":{"captured_by":"crovia.spider.vendor_press","primary_source_url":"https://arxiv.org/abs/2606.19317"},"predecessors":[],"schema":"crovia.axiom.v1","signature":"ed25519:8f8625ecdf035b8c05c661f3c6b5010b8759fe1d5eb40c45eae1f1ec6c4487d4c8bf4dc54535e284973e12f4e325e9ec5c4d8ac13458c057f2ec1f3b5018f105","signer":"crovia.substrate","subject":{"observed_at":"2026-06-30T04:43:04Z","source_collector":"spider:vendor_press","target_id":"https://arxiv.org/abs/2606.19317"},"tsa":{"authority":"crovia.substrate.bootstrap","rfc3161_token":"{\"kind\":\"crovia.bootstrap.tsa\",\"source_jsonl\":\"/opt/crovia/spider/data/news/vendor_press_v1.jsonl\",\"source_seal_merkle_root\":\"spider_vendor_press_v1\",\"upgrade_path\":\"Sessione H \\u2014 OpenTimestamps weekly anchor\"}"},"zk_mode":"clear","zk_proof":null},"ledger":{"leaf_hash":"2d55ab9121b5e1d254e0fcf5c27fa56c2e3e248a87e4d4023ac4ee26d217ab0b","leaf_index":265241,"ledger_path":"/opt/crovia/substrate/axiom_ledger.jsonl"},"merkle_proof":{"hash_alg":"sha256","leaf_prefix":"0x00","node_prefix":"0x01","odd_leaf_rule":"duplicate_last","path":[{"sibling":"16921d2dc86fe577e0534b78d32cbe0596e299bbf0bb320a7296d9aae628f14a","side":"left"},{"sibling":"074d02ed0619d9cf8cca05c6d95b4b043e13d6c6cb712f401b70678c33183f14","side":"right"},{"sibling":"1425370febdb4e4854a8770e3600626b2dc9597ba086f077046150175390d46a","side":"right"},{"sibling":"e28f49dbf6d6489b1d057f7dc03774cfa9dbd9ec7aea66b2fb112e449c854a55","side":"left"},{"sibling":"6ef353e6b7056d8c2c01c777cf533698651b30b5f0e2028a344a98ce34d3ca49","side":"left"},{"sibling":"a495e190d223e69b24e0d5ad23a15e735fff86f2c73b282efad1181b007e3623","side":"right"},{"sibling":"ff273c8201c286328de95b4c3f7ad04c9fb005c67fda3de45c54b72f9f5db5b3","side":"right"},{"sibling":"3cde187de7516743e6cd60012ce336384e260530220514cdd49599ef710e14e5","side":"right"},{"sibling":"2da116eff026ee371a88b8eb1a0aeb87679b487f1fbf1623083121809f754234","side":"right"},{"sibling":"eef13ac8a36eb836ad77846f6ef08d9afa4dcc380a64b88c542b57231522db0a","side":"right"},{"sibling":"a2cf20d97c095bc4fa7c4f5bb4f2266dd953e3d733d7e185aa164763491b312c","side":"left"},{"sibling":"f9b4bed84fa6990c71ad2887c91bda183001f05f6b648f21d1273045a26b11fd","side":"left"},{"sibling":"173d2dc4b29ee04ea41d6d0ebc334c4bc2d46e7ee4230c94765413f24fb4bc42","side":"right"},{"sibling":"112461f7c0ec411116fb5c6c90fe95cea9d8f188b9fe08afe25a837ac02d0071","side":"right"},{"sibling":"ea9488204352c49db8f7daf05eefcd7628ecf9413830346674801a99d0654a94","side":"right"},{"sibling":"6261c13b9922cb657f10d1e5d36ec15d8771cf8766e36c61dcbffb7bed57e396","side":"right"},{"sibling":"fa19aa3faf287618b820bcfceebb366152ad521dd20ef9f51e977816663e448b","side":"right"},{"sibling":"c32f943406b62d1fc59b7f7e243492174c8e1caba8c8a2705f86c773315736e0","side":"right"},{"sibling":"1cecb7f447febd025aac272837c80de218aecc6485d2395a509b2a1f1b9c746e","side":"left"}]},"schema":"crovia.axiom_proof.v1","seal":{"first_collector_run_id":"","first_receipt_hash":"","jsonl_path":"/opt/crovia/substrate/axiom_ledger.jsonl","key_id":"430895f101d38164","last_collector_run_id":"","last_receipt_hash":"","leaf_count":265374,"merkle_root":"9636001ecab173cb6af10dc7c71eb14585daa62f9c0a6f027046f05633156891","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","run_id":"hourly_json_retrofit_20260630T053701Z","schema":"crovia.seal.v1","seal_family_version":"crovia-seal-family/1","seal_kind":"substrate_batch","sealed_at":"2026-06-30T05:38:03Z","sig_algorithm":"ed25519","signature":"6d4b8fd9b9da856cbb5fba7540877c6a63fa18a5ec3eaf28dc4d6d1c64921c9c0565f8c95f4b7aec0e7744fd7754844f051baf863db708cad768765b416a7b0c","signer_version":"1.1.0"},"trust_root":{"key_id":"430895f101d38164","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","signature_algorithm":"ed25519","url":"/registry/canon/TRUST_ROOT.md"},"verifier":{"spec":"/registry/canon/AXIOM_RECEIPT_v1.md","url":"/v/axm_3fce9d96f66888c8cf2af4e8c2767a3247d6c24a7db8a86be2dc6d539ec7dcb7"}}