{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:R56UD5BHBFZAID2RLEPHSQRWHL","short_pith_number":"pith:R56UD5BH","schema_version":"1.0","canonical_sha256":"8f7d41f4270972040f51591e7942363ac5fd6fd697e564f54590189628531c28","source":{"kind":"arxiv","id":"2402.14830","version":1},"attestation_state":"computed","paper":{"title":"Orca-Math: Unlocking the potential of SLMs in Grade School Math","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ahmed Awadallah, Arindam Mitra, Corby Rosset, Hamed Khanpour","submitted_at":"2024-02-16T23:44:38Z","abstract_excerpt":"Mathematical word problem-solving has long been recognized as a complex task for small language models (SLMs). A recent study hypothesized that the smallest model size, needed to achieve over 80% accuracy on the GSM8K benchmark, is 34 billion parameters. To reach this level of performance with smaller models, researcher often train SLMs to generate Python code or use tools to help avoid calculation errors. Additionally, they employ ensembling, where outputs of up to 100 model runs are combined to arrive at a more accurate result. Result selection is done using consensus, majority vote or a sep"},"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":"2402.14830","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-16T23:44:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c49887e8ccc0ab4d0c82771e9eb5d13401c129f9e3592ed86665b992954d4e09","abstract_canon_sha256":"3cd47cc5557565c74b29702590beebb93cdf657e1863d2e6b7f84a56760ec786"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:48:18.370285Z","signature_b64":"t28+Hm7JNpPCM/ncKp7e+sL68X2dqq7f9CdBxLBC6pFNbN3/uFfzm5zzCqx87lzOE+9kFlYB+3Zmeuj35W8cBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8f7d41f4270972040f51591e7942363ac5fd6fd697e564f54590189628531c28","last_reissued_at":"2026-07-05T07:48:18.369857Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:48:18.369857Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Orca-Math: Unlocking the potential of SLMs in Grade School Math","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ahmed Awadallah, Arindam Mitra, Corby Rosset, Hamed Khanpour","submitted_at":"2024-02-16T23:44:38Z","abstract_excerpt":"Mathematical word problem-solving has long been recognized as a complex task for small language models (SLMs). A recent study hypothesized that the smallest model size, needed to achieve over 80% accuracy on the GSM8K benchmark, is 34 billion parameters. To reach this level of performance with smaller models, researcher often train SLMs to generate Python code or use tools to help avoid calculation errors. Additionally, they employ ensembling, where outputs of up to 100 model runs are combined to arrive at a more accurate result. Result selection is done using consensus, majority vote or a sep"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.14830","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/2402.14830/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":"2402.14830","created_at":"2026-07-05T07:48:18.369918+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.14830v1","created_at":"2026-07-05T07:48:18.369918+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.14830","created_at":"2026-07-05T07:48:18.369918+00:00"},{"alias_kind":"pith_short_12","alias_value":"R56UD5BHBFZA","created_at":"2026-07-05T07:48:18.369918+00:00"},{"alias_kind":"pith_short_16","alias_value":"R56UD5BHBFZAID2R","created_at":"2026-07-05T07:48:18.369918+00:00"},{"alias_kind":"pith_short_8","alias_value":"R56UD5BH","created_at":"2026-07-05T07:48:18.369918+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":18,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.28551","citing_title":"DataComp-VLM: Improved Open Datasets for Vision-Language Models","ref_index":216,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28551","citing_title":"DataComp-VLM: Improved Open Datasets for Vision-Language Models","ref_index":216,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27365","citing_title":"LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2406.11354","citing_title":"Preserving Knowledge in Large Language Model with Model-Agnostic Self-Decompression","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07711","citing_title":"SimCT: Recovering Lost Supervision for Cross-Tokenizer On-Policy Distillation","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2406.18629","citing_title":"Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2509.07379","citing_title":"DuoServe-MoE: Dual-Phase Expert Prefetch and Caching for LLM Inference QoS Assurance","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2509.17677","citing_title":"EngiBench: A Benchmark for Evaluating Large Language Models on Engineering Problem Solving","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2512.20856","citing_title":"NVIDIA Nemotron 3: Efficient and Open Intelligence","ref_index":172,"is_internal_anchor":false},{"citing_arxiv_id":"2511.01831","citing_title":"Routing-Based Continual Learning for Multimodal Large Language Models","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2406.16860","citing_title":"Cambrian-1: A Fully Open, Vision-Centric Exploration of Multimodal LLMs","ref_index":101,"is_internal_anchor":false},{"citing_arxiv_id":"2602.20816","citing_title":"Don't Ignore the Tail: Decoupling top-K Probabilities for Efficient Language Model Distillation","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10171","citing_title":"When Reviews Disagree: Fine-Grained Contradiction Analysis in Scientific Peer Reviews","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04078","citing_title":"Validity-Calibrated Reasoning Distillation","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04992","citing_title":"You Snooze, You Lose: Automatic Safety Alignment Restoration through Neural Weight Translation","ref_index":97,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04078","citing_title":"Validity-Calibrated Reasoning Distillation","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07711","citing_title":"SimCT: Recovering Lost Supervision for Cross-Tokenizer On-Policy Distillation","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2412.05271","citing_title":"Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling","ref_index":189,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/R56UD5BHBFZAID2RLEPHSQRWHL","json":"https://pith.science/pith/R56UD5BHBFZAID2RLEPHSQRWHL.json","graph_json":"https://pith.science/api/pith-number/R56UD5BHBFZAID2RLEPHSQRWHL/graph.json","events_json":"https://pith.science/api/pith-number/R56UD5BHBFZAID2RLEPHSQRWHL/events.json","paper":"https://pith.science/paper/R56UD5BH"},"agent_actions":{"view_html":"https://pith.science/pith/R56UD5BHBFZAID2RLEPHSQRWHL","download_json":"https://pith.science/pith/R56UD5BHBFZAID2RLEPHSQRWHL.json","view_paper":"https://pith.science/paper/R56UD5BH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.14830&json=true","fetch_graph":"https://pith.science/api/pith-number/R56UD5BHBFZAID2RLEPHSQRWHL/graph.json","fetch_events":"https://pith.science/api/pith-number/R56UD5BHBFZAID2RLEPHSQRWHL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R56UD5BHBFZAID2RLEPHSQRWHL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R56UD5BHBFZAID2RLEPHSQRWHL/action/storage_attestation","attest_author":"https://pith.science/pith/R56UD5BHBFZAID2RLEPHSQRWHL/action/author_attestation","sign_citation":"https://pith.science/pith/R56UD5BHBFZAID2RLEPHSQRWHL/action/citation_signature","submit_replication":"https://pith.science/pith/R56UD5BHBFZAID2RLEPHSQRWHL/action/replication_record"}},"created_at":"2026-07-05T07:48:18.369918+00:00","updated_at":"2026-07-05T07:48:18.369918+00:00"}