{"paper":{"title":"From Tokens to Concepts: Leveraging SAE for SPLADE","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Replacing the token vocabulary in SPLADE with semantic concepts from Sparse Auto-Encoders yields comparable retrieval performance with improved efficiency.","cross_cats":["cs.CL"],"primary_cat":"cs.IR","authors_text":"Basile Van Cooten, Benjamin Piwowarski, Laure Soulier, Mathias Vast, Yuxuan Zong","submitted_at":"2026-04-23T10:13:21Z","abstract_excerpt":"Learned Sparse IR models, such as SPLADE, offer an excellent efficiency-effectiveness tradeoff. However, they rely on the underlying backbone vocabulary, which might hinder performance (polysemicity and synonymy) and pose a challenge for multi-lingual and multi-modal usages. To solve this limitation, we propose to replace the backbone vocabulary with a latent space of semantic concepts learned using Sparse Auto-Encoders (SAE). Throughout this paper, we study the compatibility of these 2 concepts, explore training approaches, and analyze the differences between our SAE-SPLADE model and traditio"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our experiments demonstrate that SAE-SPLADE achieves retrieval performance comparable to SPLADE on both in-domain and out-of-domain tasks while offering improved efficiency.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the SAE-learned latent concepts form a drop-in replacement for the original token vocabulary without losing retrieval-critical information and that the two training regimes remain compatible.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"SAE-SPLADE substitutes SPLADE's backbone vocabulary with SAE-derived semantic concepts and matches standard SPLADE performance with better efficiency on in- and out-of-domain tasks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Replacing the token vocabulary in SPLADE with semantic concepts from Sparse Auto-Encoders yields comparable retrieval performance with improved efficiency.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e4bf3c4da65a7997001e3d16f869f0941c1e86bb6c883182365d10d344cc2bd4"},"source":{"id":"2604.21511","kind":"arxiv","version":2},"verdict":{"id":"88771eaa-bb96-4963-9333-6be14991a1b1","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-09T20:26:30.758694Z","strongest_claim":"Our experiments demonstrate that SAE-SPLADE achieves retrieval performance comparable to SPLADE on both in-domain and out-of-domain tasks while offering improved efficiency.","one_line_summary":"SAE-SPLADE substitutes SPLADE's backbone vocabulary with SAE-derived semantic concepts and matches standard SPLADE performance with better efficiency on in- and out-of-domain tasks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the SAE-learned latent concepts form a drop-in replacement for the original token vocabulary without losing retrieval-critical information and that the two training regimes remain compatible.","pith_extraction_headline":"Replacing the token vocabulary in SPLADE with semantic concepts from Sparse Auto-Encoders yields comparable retrieval performance with improved efficiency."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.21511/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-21T12:39:54.431790Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-20T00:56:40.743827Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"91559a2c424ad4746806fa56964f845e6a8d759b1e0ac150f6a3077573b0fbbb"},"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"}