{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:XS67K5ZENM3CIRB42YHJKN6PQA","short_pith_number":"pith:XS67K5ZE","schema_version":"1.0","canonical_sha256":"bcbdf577246b3624443cd60e9537cf8025eedbac465b1ccd3a48ab03b5271d9d","source":{"kind":"arxiv","id":"2606.08501","version":1},"attestation_state":"computed","paper":{"title":"Back on Track: Aligning Rewards and States for Reasoning in Diffusion Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hongchen Luo, Jie Xiao, Kai Zhu, Wei Zhai, Xueyang Fu, Yang Cao, Yawen Shao, Yu Liu, Zheng-Jun Zha","submitted_at":"2026-06-07T07:59:55Z","abstract_excerpt":"Reinforcement learning (RL) holds immense promise for enhancing the reasoning capabilities of diffusion large language models (dLLMs). However, progress is fundamentally constrained by a dual misalignment between authentic generation trajectory and the gradient update process: (i) Process-reward misalignment. Sparse, terminal rewards are indiscriminately assigned to all intermediate steps of the generation process, failing to provide discriminative credit assignment. (ii) State-trajectory misalignment. Policy updates are often diverted toward artificial, out-of-trajectory states, squandering g"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2606.08501","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-06-07T07:59:55Z","cross_cats_sorted":[],"title_canon_sha256":"233f8c8448d641f4a7727d67a09e53c3470bb42e913d27ec8ceec0c04aae1d17","abstract_canon_sha256":"bb831f2ec09f51108d373f3bdc119b6bbe14e9e1f79f74720c3981a94ed6fe1e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:05:38.452419Z","signature_b64":"AZ3i7ziX0gom8vWQy1dkYvnGFv6yuI7RSQVMzatTIPMj9o0mlkNLC4dXWAHZQ+jJzFFYVTpthnMBqHqIE/bKAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bcbdf577246b3624443cd60e9537cf8025eedbac465b1ccd3a48ab03b5271d9d","last_reissued_at":"2026-06-09T01:05:38.451957Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:05:38.451957Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Back on Track: Aligning Rewards and States for Reasoning in Diffusion Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hongchen Luo, Jie Xiao, Kai Zhu, Wei Zhai, Xueyang Fu, Yang Cao, Yawen Shao, Yu Liu, Zheng-Jun Zha","submitted_at":"2026-06-07T07:59:55Z","abstract_excerpt":"Reinforcement learning (RL) holds immense promise for enhancing the reasoning capabilities of diffusion large language models (dLLMs). However, progress is fundamentally constrained by a dual misalignment between authentic generation trajectory and the gradient update process: (i) Process-reward misalignment. Sparse, terminal rewards are indiscriminately assigned to all intermediate steps of the generation process, failing to provide discriminative credit assignment. (ii) State-trajectory misalignment. Policy updates are often diverted toward artificial, out-of-trajectory states, squandering g"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.08501","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/2606.08501/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2606.08501","created_at":"2026-06-09T01:05:38.452039+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.08501v1","created_at":"2026-06-09T01:05:38.452039+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08501","created_at":"2026-06-09T01:05:38.452039+00:00"},{"alias_kind":"pith_short_12","alias_value":"XS67K5ZENM3C","created_at":"2026-06-09T01:05:38.452039+00:00"},{"alias_kind":"pith_short_16","alias_value":"XS67K5ZENM3CIRB4","created_at":"2026-06-09T01:05:38.452039+00:00"},{"alias_kind":"pith_short_8","alias_value":"XS67K5ZE","created_at":"2026-06-09T01:05:38.452039+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XS67K5ZENM3CIRB42YHJKN6PQA","json":"https://pith.science/pith/XS67K5ZENM3CIRB42YHJKN6PQA.json","graph_json":"https://pith.science/api/pith-number/XS67K5ZENM3CIRB42YHJKN6PQA/graph.json","events_json":"https://pith.science/api/pith-number/XS67K5ZENM3CIRB42YHJKN6PQA/events.json","paper":"https://pith.science/paper/XS67K5ZE"},"agent_actions":{"view_html":"https://pith.science/pith/XS67K5ZENM3CIRB42YHJKN6PQA","download_json":"https://pith.science/pith/XS67K5ZENM3CIRB42YHJKN6PQA.json","view_paper":"https://pith.science/paper/XS67K5ZE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.08501&json=true","fetch_graph":"https://pith.science/api/pith-number/XS67K5ZENM3CIRB42YHJKN6PQA/graph.json","fetch_events":"https://pith.science/api/pith-number/XS67K5ZENM3CIRB42YHJKN6PQA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XS67K5ZENM3CIRB42YHJKN6PQA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XS67K5ZENM3CIRB42YHJKN6PQA/action/storage_attestation","attest_author":"https://pith.science/pith/XS67K5ZENM3CIRB42YHJKN6PQA/action/author_attestation","sign_citation":"https://pith.science/pith/XS67K5ZENM3CIRB42YHJKN6PQA/action/citation_signature","submit_replication":"https://pith.science/pith/XS67K5ZENM3CIRB42YHJKN6PQA/action/replication_record"}},"created_at":"2026-06-09T01:05:38.452039+00:00","updated_at":"2026-06-09T01:05:38.452039+00:00"}