{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:TPTOYNHDRI5WNUW4EQWJ7YGGN7","short_pith_number":"pith:TPTOYNHD","schema_version":"1.0","canonical_sha256":"9be6ec34e38a3b66d2dc242c9fe0c66ff530ab5ce5926aa69e2889bb370a6ca8","source":{"kind":"arxiv","id":"2410.14157","version":3},"attestation_state":"computed","paper":{"title":"Beyond Autoregression: Discrete Diffusion for Complex Reasoning and Planning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Jiacheng Ye, Jiahui Gao, Lingpeng Kong, Lin Zheng, Shansan Gong, Xin Jiang, Zhenguo Li","submitted_at":"2024-10-18T03:48:53Z","abstract_excerpt":"Autoregressive language models, despite their impressive capabilities, struggle with complex reasoning and long-term planning tasks. We introduce discrete diffusion models as a novel solution to these challenges. Through the lens of subgoal imbalance, we demonstrate how diffusion models effectively learn difficult subgoals that elude autoregressive approaches. We propose Multi-Granularity Diffusion Modeling (MGDM), which prioritizes subgoals based on difficulty during learning. On complex tasks like Countdown, Sudoku, and Boolean Satisfiability Problems, MGDM significantly outperforms autoregr"},"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":"2410.14157","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-18T03:48:53Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"0582ae514146e3710df3c493e93776fe95243acbbbb64040c6407759e9168591","abstract_canon_sha256":"d88c3a4c87cde8668ee4f35fffc52bd5c6bc4fbbb2bb09bedea4a9042568eeee"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:15:46.554913Z","signature_b64":"Y6BaciQe2P/7SVTuzCUCRcx6FGCPBhIDcshioiGNfbAW3ToxMUdaXRZxXpUaQonIHOZJvKU727VDuYyTlAR0Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9be6ec34e38a3b66d2dc242c9fe0c66ff530ab5ce5926aa69e2889bb370a6ca8","last_reissued_at":"2026-07-05T10:15:46.554189Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:15:46.554189Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Beyond Autoregression: Discrete Diffusion for Complex Reasoning and Planning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Jiacheng Ye, Jiahui Gao, Lingpeng Kong, Lin Zheng, Shansan Gong, Xin Jiang, Zhenguo Li","submitted_at":"2024-10-18T03:48:53Z","abstract_excerpt":"Autoregressive language models, despite their impressive capabilities, struggle with complex reasoning and long-term planning tasks. We introduce discrete diffusion models as a novel solution to these challenges. Through the lens of subgoal imbalance, we demonstrate how diffusion models effectively learn difficult subgoals that elude autoregressive approaches. We propose Multi-Granularity Diffusion Modeling (MGDM), which prioritizes subgoals based on difficulty during learning. On complex tasks like Countdown, Sudoku, and Boolean Satisfiability Problems, MGDM significantly outperforms autoregr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.14157","kind":"arxiv","version":3},"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/2410.14157/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":"2410.14157","created_at":"2026-07-05T10:15:46.554397+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.14157v3","created_at":"2026-07-05T10:15:46.554397+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.14157","created_at":"2026-07-05T10:15:46.554397+00:00"},{"alias_kind":"pith_short_12","alias_value":"TPTOYNHDRI5W","created_at":"2026-07-05T10:15:46.554397+00:00"},{"alias_kind":"pith_short_16","alias_value":"TPTOYNHDRI5WNUW4","created_at":"2026-07-05T10:15:46.554397+00:00"},{"alias_kind":"pith_short_8","alias_value":"TPTOYNHD","created_at":"2026-07-05T10:15:46.554397+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":18,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24773","citing_title":"Posterior Refinement: Fast Language Generation via Any-Order Flow Maps","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17924","citing_title":"PearlVLA: Progressive Embodied Action-Plan Refinement in Latent Space","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08501","citing_title":"Back on Track: Aligning Rewards and States for Reasoning in Diffusion Large Language Models","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19349","citing_title":"Where to Place the Query? Unveiling and Mitigating Positional Bias in In-Context Learning for Diffusion LLMs via Decoding Dynamics","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26106","citing_title":"Looped Diffusion Language Models","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25638","citing_title":"Reinforcement Learning from Denoising Feedback","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00295","citing_title":"Adaptive Order Policies for Masked Diffusion","ref_index":124,"is_internal_anchor":false},{"citing_arxiv_id":"2602.18176","citing_title":"Improving Sampling for Masked Diffusion Models via Information Gain","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22765","citing_title":"Uniform Diffusion Models Revisited: Leave-One-Out Denoiser and Absorbing State Reformulation","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17497","citing_title":"Self-Supervised On-Policy Distillation for Reasoning Language Models","ref_index":88,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17174","citing_title":"Beyond Execution: Static-Analysis Rewards and Hint-Conditioned Diffusion RL for Code Generation","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2508.14685","citing_title":"SSA: Improving Performance With a Better Scoring Function","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2601.12538","citing_title":"Agentic Reasoning for Large Language Models","ref_index":149,"is_internal_anchor":false},{"citing_arxiv_id":"2603.12554","citing_title":"Reinforcement Learning for Diffusion LLMs with Entropy-Guided Step Selection and Stepwise Advantages","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11726","citing_title":"Block-R1: Rethinking the Role of Block Size in Multi-domain Reinforcement Learning for Diffusion Large Language Models","ref_index":69,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11726","citing_title":"Block-R1: Rethinking the Role of Block Size in Multi-domain Reinforcement Learning for Diffusion Large Language Models","ref_index":69,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06548","citing_title":"Continuous Latent Diffusion Language Model","ref_index":103,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07933","citing_title":"How to Train Your Latent Diffusion Language Model Jointly With the Latent Space","ref_index":54,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TPTOYNHDRI5WNUW4EQWJ7YGGN7","json":"https://pith.science/pith/TPTOYNHDRI5WNUW4EQWJ7YGGN7.json","graph_json":"https://pith.science/api/pith-number/TPTOYNHDRI5WNUW4EQWJ7YGGN7/graph.json","events_json":"https://pith.science/api/pith-number/TPTOYNHDRI5WNUW4EQWJ7YGGN7/events.json","paper":"https://pith.science/paper/TPTOYNHD"},"agent_actions":{"view_html":"https://pith.science/pith/TPTOYNHDRI5WNUW4EQWJ7YGGN7","download_json":"https://pith.science/pith/TPTOYNHDRI5WNUW4EQWJ7YGGN7.json","view_paper":"https://pith.science/paper/TPTOYNHD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.14157&json=true","fetch_graph":"https://pith.science/api/pith-number/TPTOYNHDRI5WNUW4EQWJ7YGGN7/graph.json","fetch_events":"https://pith.science/api/pith-number/TPTOYNHDRI5WNUW4EQWJ7YGGN7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TPTOYNHDRI5WNUW4EQWJ7YGGN7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TPTOYNHDRI5WNUW4EQWJ7YGGN7/action/storage_attestation","attest_author":"https://pith.science/pith/TPTOYNHDRI5WNUW4EQWJ7YGGN7/action/author_attestation","sign_citation":"https://pith.science/pith/TPTOYNHDRI5WNUW4EQWJ7YGGN7/action/citation_signature","submit_replication":"https://pith.science/pith/TPTOYNHDRI5WNUW4EQWJ7YGGN7/action/replication_record"}},"created_at":"2026-07-05T10:15:46.554397+00:00","updated_at":"2026-07-05T10:15:46.554397+00:00"}