{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2019:ZKUCKW7RUB6BDQIVXZWSPXKUSJ","short_pith_number":"pith:ZKUCKW7R","canonical_record":{"source":{"id":"1905.11946","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-05-28T17:05:32Z","cross_cats_sorted":["cs.CV","stat.ML"],"title_canon_sha256":"5d57ed40bf99234b9f21dec3ed51799244e63fef2bb7a81fafadb22a131f417e","abstract_canon_sha256":"ed3605569195a209763a2b9cfe22d6d49f16a66c354514def888be6f55d12b8d"},"schema_version":"1.0"},"canonical_sha256":"caa8255bf1a07c11c115be6d27dd54926a42ce8404039ca9728347695b74beb1","source":{"kind":"arxiv","id":"1905.11946","version":5},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1905.11946","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"arxiv_version","alias_value":"1905.11946v5","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1905.11946","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"pith_short_12","alias_value":"ZKUCKW7RUB6B","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"ZKUCKW7RUB6BDQIV","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"ZKUCKW7R","created_at":"2026-05-18T12:33:33Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2019:ZKUCKW7RUB6BDQIVXZWSPXKUSJ","target":"record","payload":{"canonical_record":{"source":{"id":"1905.11946","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-05-28T17:05:32Z","cross_cats_sorted":["cs.CV","stat.ML"],"title_canon_sha256":"5d57ed40bf99234b9f21dec3ed51799244e63fef2bb7a81fafadb22a131f417e","abstract_canon_sha256":"ed3605569195a209763a2b9cfe22d6d49f16a66c354514def888be6f55d12b8d"},"schema_version":"1.0"},"canonical_sha256":"caa8255bf1a07c11c115be6d27dd54926a42ce8404039ca9728347695b74beb1","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:47.894555Z","signature_b64":"tkIx2K+tOb2zTRH+t7yanVktIosCwbYOr9X8ZpzVhu8+5CoZef7PEYBD6QutmNsQovpW9junqDpDoem7MGcjBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"caa8255bf1a07c11c115be6d27dd54926a42ce8404039ca9728347695b74beb1","last_reissued_at":"2026-05-17T23:38:47.894023Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:47.894023Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1905.11946","source_version":5,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"MV0iZdh1hfuHKTqnXI9OljTlE0YqzDK8fHg+UjfxQ0kq9Dcg0l3/Bla1x8eDaRBaq3eBrKVo/o6cHGQYhrK0Cg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-04T19:20:41.709927Z"},"content_sha256":"f68128bfb39ec2f304039502dce4f054d4b02b65628de2b3a34ebbf94a7417a9","schema_version":"1.0","event_id":"sha256:f68128bfb39ec2f304039502dce4f054d4b02b65628de2b3a34ebbf94a7417a9"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2019:ZKUCKW7RUB6BDQIVXZWSPXKUSJ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Scaling depth, width, and resolution together with one compound coefficient produces more accurate and efficient convolutional networks than scaling any single dimension.","cross_cats":["cs.CV","stat.ML"],"primary_cat":"cs.LG","authors_text":"Mingxing Tan, Quoc V. Le","submitted_at":"2019-05-28T17:05:32Z","abstract_excerpt":"Convolutional Neural Networks (ConvNets) are commonly developed at a fixed resource budget, and then scaled up for better accuracy if more resources are available. In this paper, we systematically study model scaling and identify that carefully balancing network depth, width, and resolution can lead to better performance. Based on this observation, we propose a new scaling method that uniformly scales all dimensions of depth/width/resolution using a simple yet highly effective compound coefficient. We demonstrate the effectiveness of this method on scaling up MobileNets and ResNet.\n  To go eve"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"our EfficientNet-B7 achieves state-of-the-art 84.3% top-1 accuracy on ImageNet, while being 8.4x smaller and 6.1x faster on inference than the best existing ConvNet.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The scaling ratios found by grid search on the small baseline network remain near-optimal when applied to much larger models and across different datasets and tasks.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"EfficientNet scales network depth, width, and resolution uniformly via a compound coefficient to deliver state-of-the-art accuracy and efficiency on image classification.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Scaling depth, width, and resolution together with one compound coefficient produces more accurate and efficient convolutional networks than scaling any single dimension.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c600afcf4b0f88b48bdcaaf76357800f3408f65170f7dd1bdc29dc279b4192e9"},"source":{"id":"1905.11946","kind":"arxiv","version":5},"verdict":{"id":"7580ba5b-e30e-4b61-9d1a-f4a6fee8f472","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T12:15:52.068716Z","strongest_claim":"our EfficientNet-B7 achieves state-of-the-art 84.3% top-1 accuracy on ImageNet, while being 8.4x smaller and 6.1x faster on inference than the best existing ConvNet.","one_line_summary":"EfficientNet scales network depth, width, and resolution uniformly via a compound coefficient to deliver state-of-the-art accuracy and efficiency on image classification.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The scaling ratios found by grid search on the small baseline network remain near-optimal when applied to much larger models and across different datasets and tasks.","pith_extraction_headline":"Scaling depth, width, and resolution together with one compound coefficient produces more accurate and efficient convolutional networks than scaling any single dimension."},"references":{"count":52,"sample":[{"doi":"","year":2011,"title":"Berg, T., Liu, J., Woo Lee, S., Alexander, M. L., Jacobs, D. W., and Belhumeur, P. N. Birdsnap: Large-scale fine-grained visual categorization of birds. CVPR, pp.\\ 2011--2018, 2014","work_id":"7e0ed155-6002-40fe-845a-3c5d2e0cb48b","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2014,"title":"Food-101--mining discriminative components with random forests","work_id":"0c135045-d9c9-42d0-a750-3b3b57618704","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"Proxylessnas: Direct neural architecture search on target task and hardware","work_id":"08465e82-fa22-4794-9994-bef8a30900a0","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Xception: Deep learning with depthwise separable convolutions","work_id":"4b286089-2fcf-408c-a148-9b7c981be9d0","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"D., Zoph, B., Mane, D., Vasudevan, V., and Le, Q","work_id":"09a7ad71-285d-47e0-a077-67b96e9c1bb7","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":52,"snapshot_sha256":"acecf4bebb971b1c5af10f33105892479730e53011cad04443c4e95d20692619","internal_anchors":7},"formal_canon":{"evidence_count":3,"snapshot_sha256":"d3b5a4cb994d6c2dbfa4802ba6c71f853ceaa677ebb4320e7c8b4b3836fb1a71"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"7580ba5b-e30e-4b61-9d1a-f4a6fee8f472"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"sUqf7s5ge32MBni4TvNOsdFB5zExaZBLv1pTOAkoPtPSTZjvFdaMQZx7e3zoTtO4z5XP60bHN+IFTpyLkKgLDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-04T19:20:41.710468Z"},"content_sha256":"9eafb8340f7cf5d69cda18397b7830dff968147b5326f74478769572df432f91","schema_version":"1.0","event_id":"sha256:9eafb8340f7cf5d69cda18397b7830dff968147b5326f74478769572df432f91"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ZKUCKW7RUB6BDQIVXZWSPXKUSJ/bundle.json","state_url":"https://pith.science/pith/ZKUCKW7RUB6BDQIVXZWSPXKUSJ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ZKUCKW7RUB6BDQIVXZWSPXKUSJ/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-04T19:20:41Z","links":{"resolver":"https://pith.science/pith/ZKUCKW7RUB6BDQIVXZWSPXKUSJ","bundle":"https://pith.science/pith/ZKUCKW7RUB6BDQIVXZWSPXKUSJ/bundle.json","state":"https://pith.science/pith/ZKUCKW7RUB6BDQIVXZWSPXKUSJ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ZKUCKW7RUB6BDQIVXZWSPXKUSJ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:ZKUCKW7RUB6BDQIVXZWSPXKUSJ","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"ed3605569195a209763a2b9cfe22d6d49f16a66c354514def888be6f55d12b8d","cross_cats_sorted":["cs.CV","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-05-28T17:05:32Z","title_canon_sha256":"5d57ed40bf99234b9f21dec3ed51799244e63fef2bb7a81fafadb22a131f417e"},"schema_version":"1.0","source":{"id":"1905.11946","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1905.11946","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"arxiv_version","alias_value":"1905.11946v5","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1905.11946","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"pith_short_12","alias_value":"ZKUCKW7RUB6B","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"ZKUCKW7RUB6BDQIV","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"ZKUCKW7R","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:9eafb8340f7cf5d69cda18397b7830dff968147b5326f74478769572df432f91","target":"graph","created_at":"2026-05-17T23:38:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"our EfficientNet-B7 achieves state-of-the-art 84.3% top-1 accuracy on ImageNet, while being 8.4x smaller and 6.1x faster on inference than the best existing ConvNet."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The scaling ratios found by grid search on the small baseline network remain near-optimal when applied to much larger models and across different datasets and tasks."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"EfficientNet scales network depth, width, and resolution uniformly via a compound coefficient to deliver state-of-the-art accuracy and efficiency on image classification."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Scaling depth, width, and resolution together with one compound coefficient produces more accurate and efficient convolutional networks than scaling any single dimension."}],"snapshot_sha256":"c600afcf4b0f88b48bdcaaf76357800f3408f65170f7dd1bdc29dc279b4192e9"},"formal_canon":{"evidence_count":3,"snapshot_sha256":"d3b5a4cb994d6c2dbfa4802ba6c71f853ceaa677ebb4320e7c8b4b3836fb1a71"},"paper":{"abstract_excerpt":"Convolutional Neural Networks (ConvNets) are commonly developed at a fixed resource budget, and then scaled up for better accuracy if more resources are available. In this paper, we systematically study model scaling and identify that carefully balancing network depth, width, and resolution can lead to better performance. Based on this observation, we propose a new scaling method that uniformly scales all dimensions of depth/width/resolution using a simple yet highly effective compound coefficient. We demonstrate the effectiveness of this method on scaling up MobileNets and ResNet.\n  To go eve","authors_text":"Mingxing Tan, Quoc V. Le","cross_cats":["cs.CV","stat.ML"],"headline":"Scaling depth, width, and resolution together with one compound coefficient produces more accurate and efficient convolutional networks than scaling any single dimension.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-05-28T17:05:32Z","title":"EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks"},"references":{"count":52,"internal_anchors":7,"resolved_work":52,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Berg, T., Liu, J., Woo Lee, S., Alexander, M. L., Jacobs, D. W., and Belhumeur, P. N. Birdsnap: Large-scale fine-grained visual categorization of birds. CVPR, pp.\\ 2011--2018, 2014","work_id":"7e0ed155-6002-40fe-845a-3c5d2e0cb48b","year":2011},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Food-101--mining discriminative components with random forests","work_id":"0c135045-d9c9-42d0-a750-3b3b57618704","year":2014},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Proxylessnas: Direct neural architecture search on target task and hardware","work_id":"08465e82-fa22-4794-9994-bef8a30900a0","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Xception: Deep learning with depthwise separable convolutions","work_id":"4b286089-2fcf-408c-a148-9b7c981be9d0","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"D., Zoph, B., Mane, D., Vasudevan, V., and Le, Q","work_id":"09a7ad71-285d-47e0-a077-67b96e9c1bb7","year":2019}],"snapshot_sha256":"acecf4bebb971b1c5af10f33105892479730e53011cad04443c4e95d20692619"},"source":{"id":"1905.11946","kind":"arxiv","version":5},"verdict":{"created_at":"2026-05-16T12:15:52.068716Z","id":"7580ba5b-e30e-4b61-9d1a-f4a6fee8f472","model_set":{"reader":"grok-4.3"},"one_line_summary":"EfficientNet scales network depth, width, and resolution uniformly via a compound coefficient to deliver state-of-the-art accuracy and efficiency on image classification.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Scaling depth, width, and resolution together with one compound coefficient produces more accurate and efficient convolutional networks than scaling any single dimension.","strongest_claim":"our EfficientNet-B7 achieves state-of-the-art 84.3% top-1 accuracy on ImageNet, while being 8.4x smaller and 6.1x faster on inference than the best existing ConvNet.","weakest_assumption":"The scaling ratios found by grid search on the small baseline network remain near-optimal when applied to much larger models and across different datasets and tasks."}},"verdict_id":"7580ba5b-e30e-4b61-9d1a-f4a6fee8f472"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:f68128bfb39ec2f304039502dce4f054d4b02b65628de2b3a34ebbf94a7417a9","target":"record","created_at":"2026-05-17T23:38:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"ed3605569195a209763a2b9cfe22d6d49f16a66c354514def888be6f55d12b8d","cross_cats_sorted":["cs.CV","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-05-28T17:05:32Z","title_canon_sha256":"5d57ed40bf99234b9f21dec3ed51799244e63fef2bb7a81fafadb22a131f417e"},"schema_version":"1.0","source":{"id":"1905.11946","kind":"arxiv","version":5}},"canonical_sha256":"caa8255bf1a07c11c115be6d27dd54926a42ce8404039ca9728347695b74beb1","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"caa8255bf1a07c11c115be6d27dd54926a42ce8404039ca9728347695b74beb1","first_computed_at":"2026-05-17T23:38:47.894023Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:47.894023Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"tkIx2K+tOb2zTRH+t7yanVktIosCwbYOr9X8ZpzVhu8+5CoZef7PEYBD6QutmNsQovpW9junqDpDoem7MGcjBA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:47.894555Z","signed_message":"canonical_sha256_bytes"},"source_id":"1905.11946","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f68128bfb39ec2f304039502dce4f054d4b02b65628de2b3a34ebbf94a7417a9","sha256:9eafb8340f7cf5d69cda18397b7830dff968147b5326f74478769572df432f91"],"state_sha256":"e0758b47b618524a6736a667aacca85f2fae45e150d07884d9261b3512e8a71a"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PS/18Qs8PtDwk6lP3B9kTVkOPCxFyFQDSU6ElSTqfHkTWvf2kZygYTcdueIysiFOzWfeh1EanrCGyUXw0zCzCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-04T19:20:41.712900Z","bundle_sha256":"47cddb1c103aad248d0794f13c6f8dd3f872cc5f4820b0dbef9b8c0319a1ab3a"}}