{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2016:S6TTYZZVXZG5AY6SDVSH4VDEPO","short_pith_number":"pith:S6TTYZZV","canonical_record":{"source":{"id":"1612.00563","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2016-12-02T04:37:22Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"55b67be8cae328edce2fb0c89400ce41408e965bb6b765c8a4e3feca9d87095f","abstract_canon_sha256":"17083da582779221ff85508061dcf62fbb35c31797e4127169972d84d8ecf3ec"},"schema_version":"1.0"},"canonical_sha256":"97a73c6735be4dd063d21d647e54647bbd66b9624bf5fd45e863e3700ca959ac","source":{"kind":"arxiv","id":"1612.00563","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1612.00563","created_at":"2026-05-18T00:30:28Z"},{"alias_kind":"arxiv_version","alias_value":"1612.00563v2","created_at":"2026-05-18T00:30:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1612.00563","created_at":"2026-05-18T00:30:28Z"},{"alias_kind":"pith_short_12","alias_value":"S6TTYZZVXZG5","created_at":"2026-05-18T12:30:44Z"},{"alias_kind":"pith_short_16","alias_value":"S6TTYZZVXZG5AY6S","created_at":"2026-05-18T12:30:44Z"},{"alias_kind":"pith_short_8","alias_value":"S6TTYZZV","created_at":"2026-05-18T12:30:44Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2016:S6TTYZZVXZG5AY6SDVSH4VDEPO","target":"record","payload":{"canonical_record":{"source":{"id":"1612.00563","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2016-12-02T04:37:22Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"55b67be8cae328edce2fb0c89400ce41408e965bb6b765c8a4e3feca9d87095f","abstract_canon_sha256":"17083da582779221ff85508061dcf62fbb35c31797e4127169972d84d8ecf3ec"},"schema_version":"1.0"},"canonical_sha256":"97a73c6735be4dd063d21d647e54647bbd66b9624bf5fd45e863e3700ca959ac","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:30:28.603120Z","signature_b64":"A41ICapw80X9Ze+JF2ubdWu6JR0fs4ApCKpnm3xdfCeZ/nuCk7pFXk2MpQyB+TY/748r9KafSlBKHjloiDdJAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"97a73c6735be4dd063d21d647e54647bbd66b9624bf5fd45e863e3700ca959ac","last_reissued_at":"2026-05-18T00:30:28.602647Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:30:28.602647Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1612.00563","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T00:30:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"LBkAqxnT4gkRJVafEMBqnB6ViAd34QBHywPhvUJhlbIOnXb92Wczu9MbEOr3mn0nnkL/Z0hX/Z5gOHj2dHSyBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-04T17:08:23.194522Z"},"content_sha256":"745397dda2c6beffc8b9c1335c4f716f9b94462328ef75717445fd0b431c8cd5","schema_version":"1.0","event_id":"sha256:745397dda2c6beffc8b9c1335c4f716f9b94462328ef75717445fd0b431c8cd5"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2016:S6TTYZZVXZG5AY6SDVSH4VDEPO","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Self-critical Sequence Training for Image Captioning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Etienne Marcheret, Jarret Ross, Steven J. Rennie, Vaibhava Goel, Youssef Mroueh","submitted_at":"2016-12-02T04:37:22Z","abstract_excerpt":"Recently it has been shown that policy-gradient methods for reinforcement learning can be utilized to train deep end-to-end systems directly on non-differentiable metrics for the task at hand. In this paper we consider the problem of optimizing image captioning systems using reinforcement learning, and show that by carefully optimizing our systems using the test metrics of the MSCOCO task, significant gains in performance can be realized. Our systems are built using a new optimization approach that we call self-critical sequence training (SCST). SCST is a form of the popular REINFORCE algorith"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1612.00563","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T00:30:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"QD6gJpk17uBa+Q5SsqCcdoDQ4VF8cccZ+75z8XRbPyAOoXAzfyWpxcM2gdQKrbPCsg4/oWpgV3wuIEWRasZQBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-04T17:08:23.194868Z"},"content_sha256":"ef20411d25ce77a9a2bd778fe17348e3344d41383efce924bdc6a0d2dba1cb3a","schema_version":"1.0","event_id":"sha256:ef20411d25ce77a9a2bd778fe17348e3344d41383efce924bdc6a0d2dba1cb3a"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/S6TTYZZVXZG5AY6SDVSH4VDEPO/bundle.json","state_url":"https://pith.science/pith/S6TTYZZVXZG5AY6SDVSH4VDEPO/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/S6TTYZZVXZG5AY6SDVSH4VDEPO/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":"2026-06-04T17:08:23Z","links":{"resolver":"https://pith.science/pith/S6TTYZZVXZG5AY6SDVSH4VDEPO","bundle":"https://pith.science/pith/S6TTYZZVXZG5AY6SDVSH4VDEPO/bundle.json","state":"https://pith.science/pith/S6TTYZZVXZG5AY6SDVSH4VDEPO/state.json","well_known_bundle":"https://pith.science/.well-known/pith/S6TTYZZVXZG5AY6SDVSH4VDEPO/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2016:S6TTYZZVXZG5AY6SDVSH4VDEPO","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":"17083da582779221ff85508061dcf62fbb35c31797e4127169972d84d8ecf3ec","cross_cats_sorted":["cs.AI","cs.CV"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2016-12-02T04:37:22Z","title_canon_sha256":"55b67be8cae328edce2fb0c89400ce41408e965bb6b765c8a4e3feca9d87095f"},"schema_version":"1.0","source":{"id":"1612.00563","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1612.00563","created_at":"2026-05-18T00:30:28Z"},{"alias_kind":"arxiv_version","alias_value":"1612.00563v2","created_at":"2026-05-18T00:30:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1612.00563","created_at":"2026-05-18T00:30:28Z"},{"alias_kind":"pith_short_12","alias_value":"S6TTYZZVXZG5","created_at":"2026-05-18T12:30:44Z"},{"alias_kind":"pith_short_16","alias_value":"S6TTYZZVXZG5AY6S","created_at":"2026-05-18T12:30:44Z"},{"alias_kind":"pith_short_8","alias_value":"S6TTYZZV","created_at":"2026-05-18T12:30:44Z"}],"graph_snapshots":[{"event_id":"sha256:ef20411d25ce77a9a2bd778fe17348e3344d41383efce924bdc6a0d2dba1cb3a","target":"graph","created_at":"2026-05-18T00:30:28Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Recently it has been shown that policy-gradient methods for reinforcement learning can be utilized to train deep end-to-end systems directly on non-differentiable metrics for the task at hand. In this paper we consider the problem of optimizing image captioning systems using reinforcement learning, and show that by carefully optimizing our systems using the test metrics of the MSCOCO task, significant gains in performance can be realized. Our systems are built using a new optimization approach that we call self-critical sequence training (SCST). SCST is a form of the popular REINFORCE algorith","authors_text":"Etienne Marcheret, Jarret Ross, Steven J. Rennie, Vaibhava Goel, Youssef Mroueh","cross_cats":["cs.AI","cs.CV"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2016-12-02T04:37:22Z","title":"Self-critical Sequence Training for Image Captioning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1612.00563","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:745397dda2c6beffc8b9c1335c4f716f9b94462328ef75717445fd0b431c8cd5","target":"record","created_at":"2026-05-18T00:30:28Z","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":"17083da582779221ff85508061dcf62fbb35c31797e4127169972d84d8ecf3ec","cross_cats_sorted":["cs.AI","cs.CV"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2016-12-02T04:37:22Z","title_canon_sha256":"55b67be8cae328edce2fb0c89400ce41408e965bb6b765c8a4e3feca9d87095f"},"schema_version":"1.0","source":{"id":"1612.00563","kind":"arxiv","version":2}},"canonical_sha256":"97a73c6735be4dd063d21d647e54647bbd66b9624bf5fd45e863e3700ca959ac","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"97a73c6735be4dd063d21d647e54647bbd66b9624bf5fd45e863e3700ca959ac","first_computed_at":"2026-05-18T00:30:28.602647Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T00:30:28.602647Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"A41ICapw80X9Ze+JF2ubdWu6JR0fs4ApCKpnm3xdfCeZ/nuCk7pFXk2MpQyB+TY/748r9KafSlBKHjloiDdJAg==","signature_status":"signed_v1","signed_at":"2026-05-18T00:30:28.603120Z","signed_message":"canonical_sha256_bytes"},"source_id":"1612.00563","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:745397dda2c6beffc8b9c1335c4f716f9b94462328ef75717445fd0b431c8cd5","sha256:ef20411d25ce77a9a2bd778fe17348e3344d41383efce924bdc6a0d2dba1cb3a"],"state_sha256":"d588fd6ccc91c28b6da7fa2da7bf7d6b66cfa8b6efd543f5704b7d92433ee6fd"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7Ckdrn8kPEmHBmBrWQJUopnnQy8cLpjPdCAepRK8hLpayhm/F1JYfAguQGlfwDzJvMXnaTlyNLefEFCtWtxfDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-04T17:08:23.196981Z","bundle_sha256":"4589fb0334ab34425d55b81a199a2643eb282f7fa78d5c16d5c97fc63ca48ce2"}}