{"_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_9542b0a86de8d03fb3ea4baf97f3d965d74d8d3eeb386e6a5623d6dbc0fae32c","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_9542b0a86de8d03fb3ea4baf97f3d965d74d8d3eeb386e6a5623d6dbc0fae32c","axiom_type":"AX.OBS","body":{"axiom_subtype":"news.vendor_press.v1","category":"news","fingerprint":"c48bdafa5401e5297d2693fee5084c9d1e5c4782bd2236d988f3ff7885fb11aa","published":"Mon, 01 Jun 2026 00:00:00 -0400","receipt_hash":"c48bdafa5401e5297d2693fee5084c9d1e5c4782bd2236d988f3ff7885fb11aa","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":"c48bdafa5401e5297d2693fee5084c9d1e5c4782bd2236d988f3ff7885fb11aa","observed_at":"2026-06-01T04:43:13.859018Z","parent_run_hash":"8993bbc535dae8c9669e099af3624cb39166b8d9bbfd66f26ae5c338cbb21be2","published":"Mon, 01 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:2605.30861v1 Announce Type: new \nAbstract: Post-training for reasoning models typically combines supervised fine-tuning with reinforcement learning from verifiable rewards, most commonly with GRPO. However, this algorithm suffers from sparse rewards, limited exploration, and mode collapse. Building upon recent works on self-distillation, we propose Feedback Distillation, a training method where the model is trained to match, at the token level, its own distribution conditioned on privileged feedback produced by a language model. Feedback Distillation offers token-level supervision and can inject external knowledge. Evaluating our method for Lean4 theorem-proving, we find that Feedback Distillation maintains greater diversity in generated trajectories than GRPO, yielding higher policy entropy and better pass@k scaling. The two methods are complementary: initializing GRPO from a Feedback Distillation checkpoint outperforms either method alone. All in all, our results suggest a prom","title":"Distilling LLM Feedback for Lean Theorem Proving","url":"https://arxiv.org/abs/2605.30861","vendor":"arxiv_cs_ai"},"summary":"arXiv:2605.30861v1 Announce Type: new \nAbstract: Post-training for reasoning models typically combines supervised fine-tuning with reinforcement learning from verifiable rewards, most commonly with GRPO. However, this algorithm suffers from sparse rewards, limited exploration, and mode collapse. Building upon recent works on self-distillation, we propose Feedback Distillation, a training method where the model is trained to match, at the token level, its own distribution conditioned on privileged feedback produced by a language model. Feedback Distillation offers token-level supervision and can inject external knowledge. Evaluating our method for Lean4 theorem-proving, we find that Feedback Distillation maintains greater diversity in generated trajectories than GRPO, yielding higher policy entropy and better pass@k scaling. The two methods are complementary: initializing GRPO from a Feedback Distillation checkpoint outperforms either method alone. All in all, our results suggest a prom","title":"Distilling LLM Feedback for Lean Theorem Proving","vendor":"arxiv_cs_ai"},"confidence":{"method":"deterministic"},"decision":"POSITIVE","issued_at":"2026-06-01T04:43:13Z","notes":"Spider vendor_press (news) news.vendor_press.v1","object":{"captured_by":"crovia.spider.vendor_press","primary_source_url":"https://arxiv.org/abs/2605.30861"},"predecessors":[],"schema":"crovia.axiom.v1","signature":"ed25519:d04a806aaa8fad885c2f7279a1f4fe6949b6ad92463823c36d60f678b46977bc43bab1abf3c52e4c0f2185776f716799d0fa40d4ad7b3f61226651e7b18faa07","signer":"crovia.substrate","subject":{"observed_at":"2026-06-01T04:43:13Z","source_collector":"spider:vendor_press","target_id":"https://arxiv.org/abs/2605.30861"},"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":"09a83942cef99f5b7f3001a59c161fdc202b278f82b622c385140906f6d4cf27","leaf_index":163781,"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":"414721c7a4bc8fdf56f9ee594ea6d5f372deb1f18015869676084aef908a2eb3","side":"left"},{"sibling":"8011b072a3e2f46ac6c40d287bc4465b893873eefedfb922c3805abf6e969351","side":"right"},{"sibling":"6c246a48a59556923511405f4dc265c0ae1f2ddac07440afd7334c71d1c2fd4b","side":"left"},{"sibling":"469540337da04f0a33e8bd3d05dcef93e1553224b2245e240921ba56ca9d7deb","side":"right"},{"sibling":"12f06b9003cd0759b5a6b217c74a5d3407ce952c9f3d1f5ffb2f2b610ea9cf2b","side":"right"},{"sibling":"14a28c2982687dd26393e7e927addebf2d5a5101d99d05872284131df7a2f44a","side":"right"},{"sibling":"6f506d4e4c7fb170ea5ab20efffbcbb69464bf6db4cde5e90925fede53e31584","side":"left"},{"sibling":"764e4312cdb701ef8613acdc2311e1724d6a379f21c70c30ed832e7ede3d2d33","side":"left"},{"sibling":"5a0517e6c348bc5a70f1aef04c1415d23980c712e375838fff54dfa281f98760","side":"left"},{"sibling":"d6986f4b6a5bd07cba1de43f660d527af780ad1465d0f4af5f137bd514e95f7e","side":"left"},{"sibling":"aff54f89d4445cf32bb05b2ece532d190200cb92524489ddfa648b5cc36e21a4","side":"left"},{"sibling":"f028fad1771bff7dceff3a83baf90249f4eb410ebedfe92dda1bb2b89d7093c8","side":"left"},{"sibling":"59c6490072e8a1d357ece10bb08d7f449a3acbae58e130e3e7469cbea0314c65","side":"left"},{"sibling":"66331bac84ca0f8983eb09fac7eaf95af234f1b82680b793eabff4ee25caac40","side":"left"},{"sibling":"35ca36cee447f0ef7064a25d55f59357c66901e31427729c4c1d8b14aa8adb6c","side":"left"},{"sibling":"c9ac00eed8c475f1e525869bb329af03d052181b19ff8234edefe12f6ecec154","side":"right"},{"sibling":"ce41d9b82f34b16efd653dfb3552acc4e2512939e47903e5fc979fbed00c5764","side":"right"},{"sibling":"d841ad93efda0869e5eb97678f348f03f5caab4353e05ff4bf18f47fb945b822","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":164217,"merkle_root":"a1098816aea1b60b8fe37b62410469bc5024a2c335bbec4f6ef2add7875dbdf2","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","run_id":"hourly_json_retrofit_20260601T053701Z","schema":"crovia.seal.v1","seal_family_version":"crovia-seal-family/1","seal_kind":"substrate_batch","sealed_at":"2026-06-01T05:37:41Z","sig_algorithm":"ed25519","signature":"d7f91db1d54b9495c499440c2828f4bd53360555391ce6e25adea5183bc1fa0f697d80708a099d0b0429e6f8cb6c71e7fccf82acb3c84481149974fb26074708","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_9542b0a86de8d03fb3ea4baf97f3d965d74d8d3eeb386e6a5623d6dbc0fae32c"}}