{"as_of":"2026-08-09T18:43:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:ad67391bbb9c0f86d65cb1cf1cba5b9a287ee025fe31a187aabd49df80bdfad6","coverage":[{"denominator":39,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":39,"source":"paper_references, paper_reference_links","source_observed_at":"2026-08-04T01:44:49.125497Z","state":"measured"},{"denominator":39,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":39,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-09T06:31:02.800959+00:00","state":"measured"},{"denominator":0,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"cited_works","source_observed_at":null,"state":"measured"}],"external_citation_measurements":[],"inbound":[],"links":{"evidence":"/evidence","html":"/paper/2608.00029/citation-record","integrity":"/paper/2608.00029/integrity","json":"/paper/2608.00029/citation-record.json","paper":"/paper/2608.00029"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.077041Z","title":"By capturing both forward activations and backward gradients into a unified graph, it unlocks the global visibility required for whole-step optimizations","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.077041Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:f89bcfa8f3899cb04dd6dd919f1299167b70e88f78d691aa79e12a95a708002d","observation_id":"6b2a5a3a-3979-4bc0-b77a-c8372603e17d","resolution":{"observed_at":"2026-08-04T01:44:46.077041Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.147871Z","title":"This represents the full training step natively within 2 MLIR, serving as the definitive foundation for both cross- operator fusion and compiler-native gradient synchronization","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.147871Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:25513be865923d33730bb9b22bec32e3abda9cb93b44be34f09dae21a61c2240","observation_id":"b1c872c9-cac1-4853-a601-eadb69a4f79b","resolution":{"observed_at":"2026-08-04T01:44:46.147871Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.223514Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.223514Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:5fda7c67b06c715566ab68e3ef9c71b6c5c2c55a4a49fc3e44afeb4b486c12a9","observation_id":"d7843cfd-5aa9-45a1-8e58-da6c754cde85","resolution":{"observed_at":"2026-08-04T01:44:46.223514Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.296596Z","title":"parallel","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.296596Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:a7eb5e7e51d3e4fb2daf6177b4d698888dc93783b84d1e535eb080c3d5e61462","observation_id":"f0bf3f6f-4cef-4031-b89b-07129caa6fe8","resolution":{"observed_at":"2026-08-04T01:44:46.296596Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.371972Z","title":"Nova time-shifts nvgpu.device_async_copy operations by depth − 1 iterations, guaranteeing that itera- tion i+2 prefetches from HBM while iteration i computes on resident memory","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.371972Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:87e080fd391f2d786c2fe1fc841c051d44f6e5f00b0756f8e11bcb07693c8329","observation_id":"06d9497a-79c6-43bb-a284-a1b162ae5acc","resolution":{"observed_at":"2026-08-04T01:44:46.371972Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.444133Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.444133Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:73ae759bed3530a48dd29af7ab73ca2199d4f2d69c53b1b87bfade9fcfd31db1","observation_id":"1df89411-c041-4bcb-afde-0217c0de178e","resolution":{"observed_at":"2026-08-04T01:44:46.444133Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.516366Z","title":"The hardware ldmatrix instruction is highly rigid, and its ldmatrix.trans variant is strictly limited to 16-bit granularity","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.516366Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:5a746b8ec0e301aa9aeaec7ecf07a763a2ded0d9c467ce44d81a43d6e0ec7633","observation_id":"369a9268-8a61-445b-a7e9-19aa1300da7c","resolution":{"observed_at":"2026-08-04T01:44:46.516366Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.617036Z","title":"To optimize them, Nova pins the accumulators directly in registers and expands clustered subgroup reductions into gpu.shuffle 6 butterfly trees","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.617036Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:f66a9247f50a07d0eb14ec6c1bf41d511fc63dc49de8fd52e60d06c7ab99dab3","observation_id":"6cf57b15-4b9f-48c3-ac56-8a522ccbac7b","resolution":{"observed_at":"2026-08-04T01:44:46.617036Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.680645Z","title":"All tested implementations remained within the 10−4 relative error band","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.680645Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:6018d8259acc15ed5b65246bbddbd1b090a355ebd7d76e40197fb9386b2d0864","observation_id":"64605d85-231c-43dd-a636-47844e5f6724","resolution":{"observed_at":"2026-08-04T01:44:46.680645Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.787792Z","title":"At 144M parameters, both PyTorch and the eager baseline run out of memory (OOM) on the 12 GB card, whereas Nova and XLA successfully fit and continue training","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.787792Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:aace730ce2d873defc1f365bd694dd069a49d828f415c40ad1273d455dd1d0b6","observation_id":"43519325-4a2c-44a9-bc93-366efba83ca6","resolution":{"observed_at":"2026-08-04T01:44:46.787792Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.863106Z","title":"By operating en- tirely on a single fused graph, Nova proves that on a memory- constrained GPU, it is possible to lower the memory floor and raise throughput simultaneously","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.863106Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:27d6bb3f98001829ea8dce6252a27c093682a3597b44788a7189744233e4674b","observation_id":"4b9d8b8e-fd9b-423e-8e20-44028903e59e","resolution":{"observed_at":"2026-08-04T01:44:46.863106Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:46.960003Z","title":"At 144M parame- ters, PyTorch and XLA both encounter OOM errors, whereas Nova successfully fits the DDP step","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:46.960003Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:b2b4d1f0b32b230d1d6fa4e9e26f6e40e40ead8193e383c23c785a683d60c868","observation_id":"ad12ddab-edad-416b-b221-e01cbc77b68c","resolution":{"observed_at":"2026-08-04T01:44:46.960003Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:47.078424Z","title":"It efficiently overlaps gradient synchronization with computation","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.078424Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:542c438fe87b77977bf75b0e51b415ac53962fb72e8646d99735115491c7a072","observation_id":"908ece37-c54d-47e7-9b1b-63c6b324242e","resolution":{"observed_at":"2026-08-04T01:44:47.078424Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:47.146700Z","title":"Figure 10:Device memory consumption (MB) in Mixed preci- sion training","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.146700Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:930d6fda7b3ca72f318bb3c2b94403817bc3c61c8f7f66d02e0870d2911a7539","observation_id":"dc046a27-db3f-41ad-8649-5baed81d7b1b","resolution":{"observed_at":"2026-08-04T01:44:47.146700Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:47.207169Z","title":"7 Discussions and limitations In this section, we analyze Nova’s architectural trade-offs, evaluating the limits of its performance characteristics and open challenges for scaling","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.207169Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:e0c1b05ba0faf37d7e9e4902a28792d7d5f0b5b2bacd36489bf5dcefac471c9f","observation_id":"829c65fe-32b9-4938-8fc5-8f8289a86c6c","resolution":{"observed_at":"2026-08-04T01:44:47.207169Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:47.286458Z","title":"IREE: Intermediate Representation Execution En- vironment","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.286458Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:fe5fd7547e2d05cf94350bb760435d14960e86d7c96555c7672725aa07331997","observation_id":"a7f9ee6c-ba27-457e-beff-8eb9217e85e9","resolution":{"observed_at":"2026-08-04T01:44:47.286458Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:47.343829Z","title":"OpenXLA: A compiler for machine learning","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.343829Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:f98302e0e67affd16c44ee19894724885ac6cc693638ca551f53d2125dd61a64","observation_id":"18086e39-76ce-4999-9d8b-5358eb890999","resolution":{"observed_at":"2026-08-04T01:44:47.343829Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2301.13062","last_updated":"2023-01-30T17:03:33Z","snapshot_observed_at":"2026-08-09T11:21:19.470879Z","submitted_at":"2023-01-30T17:03:33Z","title":"Operator Fusion in XLA: Analysis and Evaluation","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2301.13062","snapshot_observed_at":"2026-08-04T01:44:47.428288Z","title":"Operator Fusion in XLA: Analysis and Eval- uation","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.428288Z"},"links":{"cited_paper":"/paper/2301.13062","citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:9b63112d66f519f86926098df989c2de2195033b294f92b4a6f851767da15cf9","observation_id":"4f824d65-e7f2-4028-9527-992194256e61","resolution":{"observed_at":"2026-08-04T01:44:47.428288Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:47.511502Z","title":"MLIR: Scaling Compiler Infrastructure for Domain- Specific Computation","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":19,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.511502Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:7c718212e3216ef5e24c05cef34c94b6a1d0db3a90127c8e61238fe03ee9f099","observation_id":"85f1d730-47f1-4e87-8d4c-8d0eecf020fc","resolution":{"observed_at":"2026-08-04T01:44:47.511502Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:47.581293Z","title":"Halide: A Language and Compiler for Opti- mizing Parallelism, Locality, and Recomputation in Image Processing Pipelines","venue":null,"work_id":null,"year":2013},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.581293Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:06c75c202da29047416768160e8a585352b5814d14e1c1302dd8ecd0d05a20ca","observation_id":"f6d47976-55fb-41d3-bf8f-cb9b0d1bc3ed","resolution":{"observed_at":"2026-08-04T01:44:47.581293Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1802.04730","last_updated":"2018-06-29T00:16:36Z","snapshot_observed_at":"2026-08-07T06:36:37.208888Z","submitted_at":"2018-02-13T16:53:01Z","title":"Tensor Comprehensions: Framework-Agnostic High-Performance Machine Learning Abstractions","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1802.04730","snapshot_observed_at":"2026-08-04T01:44:47.637724Z","title":"Tensor Comprehensions: Framework- Agnostic High-Performance Machine Learning Abstractions","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.637724Z"},"links":{"cited_paper":"/paper/1802.04730","citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:d574434d9c6ababbddbc7eb64bb3140b002afd603ec29f9344eb9e2fd6b6f877","observation_id":"23ad3ed3-a8e8-40fd-b9d4-17e31e457ec1","resolution":{"observed_at":"2026-08-04T01:44:47.637724Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1410.0759","last_updated":"2014-12-18T01:13:16Z","snapshot_observed_at":"2026-07-06T03:56:19.212671Z","submitted_at":"2014-10-03T06:16:43Z","title":"cuDNN: Efficient Primitives for Deep Learning","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1410.0759","snapshot_observed_at":"2026-08-04T01:44:47.730009Z","title":"cuDNN: Efficient Primitives for Deep Learning","venue":null,"work_id":null,"year":2014},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.730009Z"},"links":{"cited_paper":"/paper/1410.0759","citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:29bbbd3cf57d599cbc0be19f550a39cf5dfb5ff24c9cd11c8bc5b166f37aebc2","observation_id":"2be29e31-f76a-4cb9-b1f7-95e54b9dc1b8","resolution":{"observed_at":"2026-08-04T01:44:47.730009Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1802.04799","last_updated":"2018-10-05T18:47:38Z","snapshot_observed_at":"2026-08-07T20:53:31.333665Z","submitted_at":"2018-02-12T20:49:34Z","title":"TVM: An Automated End-to-End Optimizing Compiler for Deep Learning","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1802.04799","snapshot_observed_at":"2026-08-04T01:44:47.841074Z","title":"TVM: An Automated End-to-End Optimizing Compiler for Deep Learning","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.841074Z"},"links":{"cited_paper":"/paper/1802.04799","citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:4671a9ac44697b58900a9b7817b8e8bd6f5a9873fe2522345c04b249bd91e710","observation_id":"b15599c9-198f-46f8-bef1-b8d8db21aa32","resolution":{"observed_at":"2026-08-04T01:44:47.841074Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:47.962392Z","title":"PyTorch: An Imperative Style, High-Performance Deep Learning Library","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:47.962392Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:715c839fabf80fe3949ad86a731ac115748bb0482185ec2473c93ae8a49f3f54","observation_id":"6bd8cc2d-458e-4acb-9ce5-dd77b9ddddbc","resolution":{"observed_at":"2026-08-04T01:44:47.962392Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.076306Z","title":"TensorFlow: A System for Large-Scale Machine Learning","venue":null,"work_id":null,"year":2016},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":25,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.076306Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:db135b3c92a0825c8ad4ff36c97f6acf266ea8af9ea2fc3cffec9e24e7f3f08d","observation_id":"a23b4fa5-9f23-4d6d-8f9e-e7eea7f192fb","resolution":{"observed_at":"2026-08-04T01:44:48.076306Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2202.03293","last_updated":"2022-02-07T15:31:59Z","snapshot_observed_at":"2026-08-04T09:08:27.390128Z","submitted_at":"2022-02-07T15:31:59Z","title":"Composable and Modular Code Generation in MLIR: A Structured and Retargetable Approach to Tensor Compiler Construction","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2202.03293","snapshot_observed_at":"2026-08-04T01:44:48.229027Z","title":"Composable and Modular Code Generation in MLIR: A Structured and Retargetable Approach to Tensor Compiler Construction","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.229027Z"},"links":{"cited_paper":"/paper/2202.03293","citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:9f2561d6f7583eee823d97400fd9b5a2ca70ab1022339f6aa333c77a5bdbba6d","observation_id":"342254c8-20b7-42e1-9864-85bebc0e2e46","resolution":{"observed_at":"2026-08-04T01:44:48.229027Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.297577Z","title":"Triton: An Intermediate Language and Compiler for Tiled Neural Network Computations","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":27,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.297577Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:170e2ef622bee02fbd329556b46e6f5827d12a5664e6a3620e07dd9cd281e263","observation_id":"77df85a4-e998-4474-ac03-aeec1cfae9dc","resolution":{"observed_at":"2026-08-04T01:44:48.297577Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.387900Z","title":"Data Movement Is All You Need: A Case Study on Optimizing Transformers","venue":null,"work_id":null,"year":2021},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.387900Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:6d845bbc9df9be414d7fae02251cd44dae0228ef28a802612cc897caf494c111","observation_id":"81cfc8be-1b58-4ecc-8623-cd31d6781211","resolution":{"observed_at":"2026-08-04T01:44:48.387900Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.447876Z","title":"The MLIR Transform Dialect: Your Compiler Is More Powerful Than You Think","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.447876Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:e23a4797f15e2197c957f0dcd29671e645b0b721598b734c577f1dbe5ffa979c","observation_id":"da0409ad-d884-4f62-86fe-dcd626631e0a","resolution":{"observed_at":"2026-08-04T01:44:48.447876Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.511066Z","title":"Ansor: Generating High-Performance Tensor Programs for Deep Learning","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.511066Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:5d951c0606d76f6be3081fb0b856d0894b01d210fafb78e372bb65cf868b6907","observation_id":"c1ba18a3-f3fb-422b-8070-f11b82a985b9","resolution":{"observed_at":"2026-08-04T01:44:48.511066Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.599965Z","title":"cuBLAS Library User Guide","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":31,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.599965Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:3f8526821e84bed18a8193ea485b82c9f6f2ce91917c19b2020b5cf0863578ec","observation_id":"4315f462-1de6-4588-8fea-3b7d05cb5278","resolution":{"observed_at":"2026-08-04T01:44:48.599965Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.668457Z","title":"CUTLASS: CUDA Templates for Linear Algebra Subroutines","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":32,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.668457Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:55185f50782214fae32f0fdbdf9ae24e586138fd7f2653015e8ab61c54b2a12a","observation_id":"8fb5fe0b-d5fe-4d91-9cf2-f575e5be191b","resolution":{"observed_at":"2026-08-04T01:44:48.668457Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.754989Z","title":"FlashAttention: Fast and Memory-Efficient Exact Atten- tion with IO-Awareness","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":33,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.754989Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:502ff40a99bce164f99408c362cb327b53d6f14225aa6bc9f31187ee63df8e8c","observation_id":"122a1c7e-d264-42f3-b883-9a37679ff6c4","resolution":{"observed_at":"2026-08-04T01:44:48.754989Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.811955Z","title":"PyTorch Distributed: Experiences on Accelerating Data Parallel Training","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":34,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.811955Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:7d6bba77b74271fa4df0b7fa27d9cfe47d60b7ccb57c73014c71f9d3b4034967","observation_id":"1ac5a38b-9c92-439c-ba88-2e6fa9d0bfa4","resolution":{"observed_at":"2026-08-04T01:44:48.811955Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.866741Z","title":"Automatic Differentiation in Machine Learning: a Survey","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":35,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.866741Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:3b14fed52c1317c31e04b89d92c0bb8537cb642c642b531563feb1737f3ab9c1","observation_id":"b067abe4-8107-4802-9038-5fff65b18e09","resolution":{"observed_at":"2026-08-04T01:44:48.866741Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:48.965402Z","title":"JAX: Composable Transformations of Python+NumPy Programs","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":36,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:48.965402Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:ffa9a240c81a966148eaf4d043cebfabebc6d87ca2a920bee71a3424fdfcba8f","observation_id":"0b9c0e5b-2f4c-4c9d-8da3-e88fcbfe0c56","resolution":{"observed_at":"2026-08-04T01:44:48.965402Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:49.008467Z","title":"Roofline: An Insightful Visual Performance Model for Multicore Architectures","venue":null,"work_id":null,"year":2009},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":37,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:49.008467Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:16bc044fa1e9a8a5e53bfee099f5a43872734aab436f9c2725a166b53f04ebb0","observation_id":"9fbb5862-e225-4f48-9d5a-65bfe239036c","resolution":{"observed_at":"2026-08-04T01:44:49.008467Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:49.066748Z","title":"CUDA C++ Programming Guide","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":38,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:49.066748Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:bf6bafea1f15cba9b82ee7592b5604833f2f88c84a25805489c87f29a4b1850e","observation_id":"ca78294d-e0b0-4632-8395-45e14830b622","resolution":{"observed_at":"2026-08-04T01:44:49.066748Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-08-04T01:44:49.125497Z","title":"LLVM/MLIR 21.1.6; CUDA Toolkit 13.0","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning","version":1},"reference_index":39,"source":"pdf_text","source_observed_at":"2026-08-04T01:44:49.125497Z"},"links":{"citing_paper":"/paper/2608.00029"},"observation_digest":"sha256:f040c4392c64bc9df7225c7cfb84eed3fee6eff37adcc4df4f1bab9860086aba","observation_id":"3d7001f9-e9de-479c-bc51-5c1f2c5211be","resolution":{"observed_at":"2026-08-04T01:44:49.125497Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2608.00029","last_updated":"2026-07-15T05:16:28Z","latest_version":1,"primary_category":"cs.AI","snapshot_observed_at":"2026-08-07T18:32:55.347456Z","submitted_at":"2026-07-15T05:16:28Z","title":"Nova: An End-to-End MLIR Compiler for Deep Learning"},"reference_resolution":{"displayed":39,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":39,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":39},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-09T06:31:02.800959+00:00","source":"crossref"},{"observed_at":"2026-08-09T06:30:57.326959+00:00","source":"retraction_watch"}],"thesis":"As of 9 August 2026, this Paper Citation Record lists 39 of 39 outbound references and 0 inbound Pith citation observations for arXiv:2608.00029."}