{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:5L7B7RSIOTNF5LRHXOF7LCM6BN","short_pith_number":"pith:5L7B7RSI","schema_version":"1.0","canonical_sha256":"eafe1fc64874da5eae27bb8bf5899e0b6f31a59bd2c372db35fe23fa62487273","source":{"kind":"arxiv","id":"2405.00208","version":3},"attestation_state":"computed","paper":{"title":"A Primer on the Inner Workings of Transformer-based Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Arianna Bisazza, Gabriele Sarti, Javier Ferrando, Marta R. Costa-juss\\`a","submitted_at":"2024-04-30T21:20:17Z","abstract_excerpt":"The rapid progress of research aimed at interpreting the inner workings of advanced language models has highlighted a need for contextualizing the insights gained from years of work in this area. This primer provides a concise technical introduction to the current techniques used to interpret the inner workings of Transformer-based language models, focusing on the generative decoder-only architecture. We conclude by presenting a comprehensive overview of the known internal mechanisms implemented by these models, uncovering connections across popular approaches and active research directions in"},"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":"2405.00208","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-30T21:20:17Z","cross_cats_sorted":[],"title_canon_sha256":"0954876a7d0723626b727167c8dd742f2c7e7eb9989492bf1e2c6292fd8446e8","abstract_canon_sha256":"3c192f6aa431361f4019b6048890fbaa3260eb2df5892248579a4bd34e8433e9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:19:56.928125Z","signature_b64":"4QjoqCS57ooSOXx+225eSCccnG5dtUybSmP5EII5umEoTFegCZOmYHXPNgKyzoWmdVI+szTTevbWVE1xKk2JAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eafe1fc64874da5eae27bb8bf5899e0b6f31a59bd2c372db35fe23fa62487273","last_reissued_at":"2026-07-05T09:19:56.927668Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:19:56.927668Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Primer on the Inner Workings of Transformer-based Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Arianna Bisazza, Gabriele Sarti, Javier Ferrando, Marta R. Costa-juss\\`a","submitted_at":"2024-04-30T21:20:17Z","abstract_excerpt":"The rapid progress of research aimed at interpreting the inner workings of advanced language models has highlighted a need for contextualizing the insights gained from years of work in this area. This primer provides a concise technical introduction to the current techniques used to interpret the inner workings of Transformer-based language models, focusing on the generative decoder-only architecture. We conclude by presenting a comprehensive overview of the known internal mechanisms implemented by these models, uncovering connections across popular approaches and active research directions in"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.00208","kind":"arxiv","version":3},"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/2405.00208/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":"2405.00208","created_at":"2026-07-05T09:19:56.927725+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.00208v3","created_at":"2026-07-05T09:19:56.927725+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.00208","created_at":"2026-07-05T09:19:56.927725+00:00"},{"alias_kind":"pith_short_12","alias_value":"5L7B7RSIOTNF","created_at":"2026-07-05T09:19:56.927725+00:00"},{"alias_kind":"pith_short_16","alias_value":"5L7B7RSIOTNF5LRH","created_at":"2026-07-05T09:19:56.927725+00:00"},{"alias_kind":"pith_short_8","alias_value":"5L7B7RSI","created_at":"2026-07-05T09:19:56.927725+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":21,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24026","citing_title":"Can Language Model Agents be Helpful Circuit Explainers in Mechanistic Interpretability?","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09466","citing_title":"DECSELFMASK: Leveraging Unlabeled Text via Self-Relevance-Guided Masking for Decoder-Only Classification","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00012","citing_title":"PRA-RAG: Provably Robust Aggregation in Retrieval-Augmented Generation against Retrieval Corruption","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03085","citing_title":"Multi-component Causal Tracing in Large Language Models","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22532","citing_title":"Relational Linear Properties in Language Models: An Empirical Investigation","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07604","citing_title":"Contribution Weights: A Geometrical Analysis of Self-Attention Transformers","ref_index":158,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07604","citing_title":"Contribution Weights: A Geometrical Analysis of Self-Attention Transformers","ref_index":72,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22532","citing_title":"Relational Linear Properties in Language Models: An Empirical Investigation","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2602.16608","citing_title":"Explainable AI: Context-Aware Layer-Wise Integrated Gradients for Explaining Transformer Models","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2509.18127","citing_title":"Safe-SAIL: Towards a Fine-grained Safety Landscape of Large Language Models via Sparse Autoencoder Interpretation Framework","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2601.14004","citing_title":"Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models","ref_index":72,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14075","citing_title":"Rethinking Layer Relevance in Large Language Models Beyond Cosine Similarity","ref_index":54,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11206","citing_title":"Instructions Shape Production of Language, not Processing","ref_index":138,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11093","citing_title":"Enabling Performant and Flexible Model-Internal Observability for LLM Inference","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12412","citing_title":"Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space","ref_index":74,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11206","citing_title":"Instructions Shape Production of Language, not Processing","ref_index":138,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09129","citing_title":"Data-driven Circuit Discovery for Interpretability of Language Models","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04177","citing_title":"Are LLMs Ready for Conflict Monitoring? Empirical Evidence from West Africa","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07019","citing_title":"ConceptTracer: Interactive Analysis of Concept Saliency and Selectivity in Neural Representations","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06356","citing_title":"In-Context Learning in Speech Language Models: Analyzing the Role of Acoustic Features, Linguistic Structure, and Induction Heads","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17761","citing_title":"Contrastive Attribution in the Wild: An Interpretability Analysis of LLM Failures on Realistic Benchmarks","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5L7B7RSIOTNF5LRHXOF7LCM6BN","json":"https://pith.science/pith/5L7B7RSIOTNF5LRHXOF7LCM6BN.json","graph_json":"https://pith.science/api/pith-number/5L7B7RSIOTNF5LRHXOF7LCM6BN/graph.json","events_json":"https://pith.science/api/pith-number/5L7B7RSIOTNF5LRHXOF7LCM6BN/events.json","paper":"https://pith.science/paper/5L7B7RSI"},"agent_actions":{"view_html":"https://pith.science/pith/5L7B7RSIOTNF5LRHXOF7LCM6BN","download_json":"https://pith.science/pith/5L7B7RSIOTNF5LRHXOF7LCM6BN.json","view_paper":"https://pith.science/paper/5L7B7RSI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.00208&json=true","fetch_graph":"https://pith.science/api/pith-number/5L7B7RSIOTNF5LRHXOF7LCM6BN/graph.json","fetch_events":"https://pith.science/api/pith-number/5L7B7RSIOTNF5LRHXOF7LCM6BN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5L7B7RSIOTNF5LRHXOF7LCM6BN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5L7B7RSIOTNF5LRHXOF7LCM6BN/action/storage_attestation","attest_author":"https://pith.science/pith/5L7B7RSIOTNF5LRHXOF7LCM6BN/action/author_attestation","sign_citation":"https://pith.science/pith/5L7B7RSIOTNF5LRHXOF7LCM6BN/action/citation_signature","submit_replication":"https://pith.science/pith/5L7B7RSIOTNF5LRHXOF7LCM6BN/action/replication_record"}},"created_at":"2026-07-05T09:19:56.927725+00:00","updated_at":"2026-07-05T09:19:56.927725+00:00"}