{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:ZJPWWNRLHXMHCR5S7BC33BQE3W","short_pith_number":"pith:ZJPWWNRL","schema_version":"1.0","canonical_sha256":"ca5f6b362b3dd87147b2f845bd8604ddb6061931d8116244d2a969c39ee1f264","source":{"kind":"arxiv","id":"2107.05431","version":2},"attestation_state":"computed","paper":{"title":"CoBERL: Contrastive BERT for Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Adri\\`a Puidomenech Badia, Andrea Banino, Charles Blundell, Jacob Walker, Jovana Mitrovic, Tim Scholtes","submitted_at":"2021-07-12T13:54:18Z","abstract_excerpt":"Many reinforcement learning (RL) agents require a large amount of experience to solve tasks. We propose Contrastive BERT for RL (CoBERL), an agent that combines a new contrastive loss and a hybrid LSTM-transformer architecture to tackle the challenge of improving data efficiency. CoBERL enables efficient, robust learning from pixels across a wide range of domains. We use bidirectional masked prediction in combination with a generalization of recent contrastive methods to learn better representations for transformers in RL, without the need of hand engineered data augmentations. We find that Co"},"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":"2107.05431","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-07-12T13:54:18Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"73970c4d01137d1e010d69350f1b59fa5118b741bc9b67916f25c2ded215c572","abstract_canon_sha256":"f38d0a9bc7d52dc629a83b13c41db4dd2d6d3ee80c26f8f18d20dcde54000a34"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:58:50.611148Z","signature_b64":"VsXrDHoqtf1OOmJGnCvFo77ceTVhmT4EzLPZRPDmavvY4/ffYThW08eA9vE0BiXjYaNxKqHfx/QgUo/9c0RDBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ca5f6b362b3dd87147b2f845bd8604ddb6061931d8116244d2a969c39ee1f264","last_reissued_at":"2026-07-05T03:58:50.610611Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:58:50.610611Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CoBERL: Contrastive BERT for Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Adri\\`a Puidomenech Badia, Andrea Banino, Charles Blundell, Jacob Walker, Jovana Mitrovic, Tim Scholtes","submitted_at":"2021-07-12T13:54:18Z","abstract_excerpt":"Many reinforcement learning (RL) agents require a large amount of experience to solve tasks. We propose Contrastive BERT for RL (CoBERL), an agent that combines a new contrastive loss and a hybrid LSTM-transformer architecture to tackle the challenge of improving data efficiency. CoBERL enables efficient, robust learning from pixels across a wide range of domains. We use bidirectional masked prediction in combination with a generalization of recent contrastive methods to learn better representations for transformers in RL, without the need of hand engineered data augmentations. We find that Co"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2107.05431","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2107.05431/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":"2107.05431","created_at":"2026-07-05T03:58:50.610672+00:00"},{"alias_kind":"arxiv_version","alias_value":"2107.05431v2","created_at":"2026-07-05T03:58:50.610672+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2107.05431","created_at":"2026-07-05T03:58:50.610672+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZJPWWNRLHXMH","created_at":"2026-07-05T03:58:50.610672+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZJPWWNRLHXMHCR5S","created_at":"2026-07-05T03:58:50.610672+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZJPWWNRL","created_at":"2026-07-05T03:58:50.610672+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.05208","citing_title":"Transformer-Enhanced Reinforcement Learning: Fundamentals and Applications in Communication Networks","ref_index":83,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZJPWWNRLHXMHCR5S7BC33BQE3W","json":"https://pith.science/pith/ZJPWWNRLHXMHCR5S7BC33BQE3W.json","graph_json":"https://pith.science/api/pith-number/ZJPWWNRLHXMHCR5S7BC33BQE3W/graph.json","events_json":"https://pith.science/api/pith-number/ZJPWWNRLHXMHCR5S7BC33BQE3W/events.json","paper":"https://pith.science/paper/ZJPWWNRL"},"agent_actions":{"view_html":"https://pith.science/pith/ZJPWWNRLHXMHCR5S7BC33BQE3W","download_json":"https://pith.science/pith/ZJPWWNRLHXMHCR5S7BC33BQE3W.json","view_paper":"https://pith.science/paper/ZJPWWNRL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2107.05431&json=true","fetch_graph":"https://pith.science/api/pith-number/ZJPWWNRLHXMHCR5S7BC33BQE3W/graph.json","fetch_events":"https://pith.science/api/pith-number/ZJPWWNRLHXMHCR5S7BC33BQE3W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZJPWWNRLHXMHCR5S7BC33BQE3W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZJPWWNRLHXMHCR5S7BC33BQE3W/action/storage_attestation","attest_author":"https://pith.science/pith/ZJPWWNRLHXMHCR5S7BC33BQE3W/action/author_attestation","sign_citation":"https://pith.science/pith/ZJPWWNRLHXMHCR5S7BC33BQE3W/action/citation_signature","submit_replication":"https://pith.science/pith/ZJPWWNRLHXMHCR5S7BC33BQE3W/action/replication_record"}},"created_at":"2026-07-05T03:58:50.610672+00:00","updated_at":"2026-07-05T03:58:50.610672+00:00"}