{"_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_a1b415faa75450f738bb78d278593a733a5d458c491902827451aa87f6061a5e","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_a1b415faa75450f738bb78d278593a733a5d458c491902827451aa87f6061a5e","axiom_type":"AX.OBS","body":{"axiom_subtype":"news.vendor_press.v1","category":"news","fingerprint":"46ce1b9fa70b3de8edb5bff540058e3af5953a9f20c81f8133c765cfc849e8ea","published":"Tue, 21 Jul 2026 00:00:00 -0400","receipt_hash":"46ce1b9fa70b3de8edb5bff540058e3af5953a9f20c81f8133c765cfc849e8ea","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":"46ce1b9fa70b3de8edb5bff540058e3af5953a9f20c81f8133c765cfc849e8ea","observed_at":"2026-07-21T04:43:35.036805Z","parent_run_hash":"03e944014de2697434479833d15ea9303e014945afc230ecc7f207824493b589","published":"Tue, 21 Jul 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:2602.21492v2 Announce Type: replace-cross \nAbstract: Reinforcement learning (RL) has become a central post-training paradigm for large language models (LLMs), but its performance is highly sensitive to the quality of training problems. This sensitivity stems from the non-stationarity of RL: rollouts are generated by an evolving policy, and learning is shaped by exploration and reward feedback, unlike supervised fine-tuning (SFT) with fixed trajectories. As a result, prior work often relies on manual curation or simple heuristic filters (e.g., accuracy), which can admit incorrect or low-utility problems. We propose GradAlign, a gradient-aligned data selection method for LLM reinforcement learning that uses a small, trusted validation set to prioritize training problems whose policy gradients align with validation gradients, yielding an adaptive curriculum. We evaluate GradAlign across three challenging data regimes: unreliable reward signals, distribution imbalance, and low-utilit","title":"GradAlign: Gradient-Aligned Data Selection for LLM Reinforcement Learning","url":"https://arxiv.org/abs/2602.21492","vendor":"arxiv_cs_ai"},"summary":"arXiv:2602.21492v2 Announce Type: replace-cross \nAbstract: Reinforcement learning (RL) has become a central post-training paradigm for large language models (LLMs), but its performance is highly sensitive to the quality of training problems. This sensitivity stems from the non-stationarity of RL: rollouts are generated by an evolving policy, and learning is shaped by exploration and reward feedback, unlike supervised fine-tuning (SFT) with fixed trajectories. As a result, prior work often relies on manual curation or simple heuristic filters (e.g., accuracy), which can admit incorrect or low-utility problems. We propose GradAlign, a gradient-aligned data selection method for LLM reinforcement learning that uses a small, trusted validation set to prioritize training problems whose policy gradients align with validation gradients, yielding an adaptive curriculum. We evaluate GradAlign across three challenging data regimes: unreliable reward signals, distribution imbalance, and low-utilit","title":"GradAlign: Gradient-Aligned Data Selection for LLM Reinforcement Learning","vendor":"arxiv_cs_ai"},"confidence":{"method":"deterministic"},"decision":"POSITIVE","issued_at":"2026-07-21T04:43:35Z","notes":"Spider vendor_press (news) news.vendor_press.v1","object":{"captured_by":"crovia.spider.vendor_press","primary_source_url":"https://arxiv.org/abs/2602.21492"},"predecessors":[],"schema":"crovia.axiom.v1","signature":"ed25519:cae5e5d8c1ed352dbc2d266a73047900be805af3413702e9a40444d2e26eda2da9c58fc5091f8044120d8b58ade044a22b0914a6439b950a5f6939b93fe5d20d","signer":"crovia.substrate","subject":{"observed_at":"2026-07-21T04:43:35Z","source_collector":"spider:vendor_press","target_id":"https://arxiv.org/abs/2602.21492"},"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":"b404dfc31649047e0618d76356cd12195831a6a9abcd16f407f8bb2edcb1b89c","leaf_index":336951,"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":"2fbc62887b7160ebb123a2ab3a42e584f8c74a5d1aa43e88cad49e3f382ecb92","side":"left"},{"sibling":"d405f805f312d2bc6bc54f6f7539ecace078cbca5e0d8170efb739cebfa2e908","side":"left"},{"sibling":"4a73c7e2a1cbbf479dcc9ca2be4089ffe91de8914f17b0a6f4a6443123e7f906","side":"left"},{"sibling":"0327fb284254e71440c0cb0792a27e2f6469accbf98fcf8fa3da049eb74a2c6c","side":"right"},{"sibling":"24d5ee4978268aeb8f558ef20aad366b0953b3bf3f0cf558fab499d1678ecc80","side":"left"},{"sibling":"bfabfb8c7ddd2bdc7b69e0e90130c2fa87e5e86c2ab6eb510187bf548726b65c","side":"left"},{"sibling":"950013dbd7f9f5861e4c5ff582a4f28f41529fee5ece541adb6d424c9cbd1f75","side":"right"},{"sibling":"45255096c5a08e740a615b6e3c262589a170dbfc373bdeaceae175916dd2ab97","side":"right"},{"sibling":"1298242aa509bc13b92160250ead2d95a721941ca6310e33b9eb0bf0821557d0","side":"right"},{"sibling":"61d44c6d16b87de577db53cff0dba2f003c50c036d92dc335120aa70124c8788","side":"right"},{"sibling":"5d0b792dbe69fd0a7cd8d79b467aa10204d1f25be602c06190f3423bc22727b8","side":"left"},{"sibling":"9eb5077edfb3dc553857d4794b925bfce117e0f8a1d049af5d0dd9026b470eef","side":"right"},{"sibling":"414b1a70fd1dcb25489a194714b97492b066684b15d0b7a48a176c4b9b5bc713","side":"right"},{"sibling":"21d66dd41003813f710b7617944f1bfba3258658a5d3370c21cad8f9e945bc99","side":"left"},{"sibling":"613f015699131eb89bd755dee67133be95af25cf5f16c1c8ce4b99d963b8dd86","side":"right"},{"sibling":"a729b574b1135956436ded5eef1fe8f08014ff6a0729749d307ab1bca93fcdc9","side":"right"},{"sibling":"77025bcb374a7ad74f520643e20a8ae1205a7ee78507b0beb93117760f1c29d3","side":"left"},{"sibling":"4bf21052e085e8ac81f1dec1d2b310bd12bf948992de6177d12e9d2fda8d39f0","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":337144,"merkle_root":"5e969cc01afa67e4dbe5d37b712cdb10f4aa1fd74404e02eab724cf487c8d6d9","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","run_id":"hourly_json_retrofit_20260721T053701Z","schema":"crovia.seal.v1","seal_family_version":"crovia-seal-family/1","seal_kind":"substrate_batch","sealed_at":"2026-07-21T05:38:38Z","sig_algorithm":"ed25519","signature":"5c0c1a8dd2793d787ccd5e49e8b4d70eed136352555518589f05c74be357fc171702e42a76c3a556d90e3d51ff36cb3d292aaac83c66566de7b943f318bda50c","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_a1b415faa75450f738bb78d278593a733a5d458c491902827451aa87f6061a5e"}}