{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:HSPOK5VN2LCBPFIJXBTSX45JCR","short_pith_number":"pith:HSPOK5VN","canonical_record":{"source":{"id":"2604.23862","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-26T20:09:25Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"c974ccea7fca6e2627f929eb263e0832b8eda1dd2c6feaf1cae9d80dc56ee91e","abstract_canon_sha256":"22ae36e7fb6ac610402f580714ab826f9c81da91cb3e88afd7dafb4f274cd42c"},"schema_version":"1.0"},"canonical_sha256":"3c9ee576add2c4179509b8672bf3a9145e2eddb17055a71cf5a98061cacbf45a","source":{"kind":"arxiv","id":"2604.23862","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.23862","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"arxiv_version","alias_value":"2604.23862v2","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.23862","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"pith_short_12","alias_value":"HSPOK5VN2LCB","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"pith_short_16","alias_value":"HSPOK5VN2LCBPFIJ","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"pith_short_8","alias_value":"HSPOK5VN","created_at":"2026-05-28T02:04:48Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:HSPOK5VN2LCBPFIJXBTSX45JCR","target":"record","payload":{"canonical_record":{"source":{"id":"2604.23862","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-26T20:09:25Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"c974ccea7fca6e2627f929eb263e0832b8eda1dd2c6feaf1cae9d80dc56ee91e","abstract_canon_sha256":"22ae36e7fb6ac610402f580714ab826f9c81da91cb3e88afd7dafb4f274cd42c"},"schema_version":"1.0"},"canonical_sha256":"3c9ee576add2c4179509b8672bf3a9145e2eddb17055a71cf5a98061cacbf45a","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-28T02:04:48.216901Z","signature_b64":"wlrnZGzznPYj7MHRYHQ90bYWE2aBFnpSqyiL6OfbgH+ME6U3ArKVcA9YosLwjINMDrUXv51fgbUKX0FYaroKBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3c9ee576add2c4179509b8672bf3a9145e2eddb17055a71cf5a98061cacbf45a","last_reissued_at":"2026-05-28T02:04:48.216382Z","signature_status":"signed_v1","first_computed_at":"2026-05-28T02:04:48.216382Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2604.23862","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":"2026-05-28T02:04:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"m9+hPS7u+aZ2voFyDVEetn8QRNdbaDxYR+Z7pJM3XM1mIMqP8cFVN9vcCD6Ece76iEuOlB6JxAmABjgUY4DJAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-03T23:17:19.689544Z"},"content_sha256":"d171e2a4a7f404a0f898f8299e06c4c40b2b1515ddd848b877cc60c54c41baf5","schema_version":"1.0","event_id":"sha256:d171e2a4a7f404a0f898f8299e06c4c40b2b1515ddd848b877cc60c54c41baf5"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:HSPOK5VN2LCBPFIJXBTSX45JCR","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Graph Memory Transformer (GMT)","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A learned graph of memory centroids can replace the feed-forward sublayer in a decoder-only transformer.","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Niccol\\`o Ferrari, Nicola Zanarini","submitted_at":"2026-04-26T20:09:25Z","abstract_excerpt":"We investigate whether the Feed-Forward Network (FFN) sublayer in a decoder-only transformer can be replaced by an explicit learned memory graph while preserving the surrounding autoregressive architecture. The proposed Graph Memory Transformer (GMT) keeps causal self-attention intact, but replaces the usual per-token FFN transformation with a memory cell that routes token representations over a learned bank of centroids connected by a learned directed transition matrix. In the base GMT v7 instantiation studied here, each of 16 transformer blocks contains 128 centroids, a 128 * 128 edge matrix"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"The base GMT v7 model trains stably and exposes centroid usage, transition structure, and source-to-target movement as directly inspectable quantities of the forward computation. It remains behind the larger dense baseline in validation loss and perplexity (3.5995/36.58 vs. 3.2903/26.85), while showing close zero-shot benchmark behavior under the evaluated setting.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That gravitational source routing plus token-conditioned target selection plus gated displacement readout can functionally substitute for the dense FFN transformation without requiring additional capacity or architectural changes outside the memory cell itself.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Graph Memory Transformer (GMT) swaps dense FFN sublayers for a graph of 128 centroids and a learned 128x128 transition matrix per block, yielding a 82M-parameter decoder-only LM that trains stably but trails a 103M dense baseline in perplexity.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A learned graph of memory centroids can replace the feed-forward sublayer in a decoder-only transformer.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"622e7f307e16a4a43f6e050b20f4b25425ce585f5d9c7e61d1f9238ecdf6a960"},"source":{"id":"2604.23862","kind":"arxiv","version":2},"verdict":{"id":"66cf2d24-7e13-42cf-bc27-690c491710d1","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T06:24:14.512096Z","strongest_claim":"The base GMT v7 model trains stably and exposes centroid usage, transition structure, and source-to-target movement as directly inspectable quantities of the forward computation. It remains behind the larger dense baseline in validation loss and perplexity (3.5995/36.58 vs. 3.2903/26.85), while showing close zero-shot benchmark behavior under the evaluated setting.","one_line_summary":"Graph Memory Transformer (GMT) swaps dense FFN sublayers for a graph of 128 centroids and a learned 128x128 transition matrix per block, yielding a 82M-parameter decoder-only LM that trains stably but trails a 103M dense baseline in perplexity.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That gravitational source routing plus token-conditioned target selection plus gated displacement readout can functionally substitute for the dense FFN transformation without requiring additional capacity or architectural changes outside the memory cell itself.","pith_extraction_headline":"A learned graph of memory centroids can replace the feed-forward sublayer in a decoder-only transformer."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.23862/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-21T07:43:11.410025Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T22:42:36.913523Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"f0922dd35a64782136d1f08b38c0aeb566f919a7180a5d3b6d3866065d4fb412"},"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":"66cf2d24-7e13-42cf-bc27-690c491710d1"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-28T02:04:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"sdZe68GtiFee02W7UuDATx432qMYaRPIPUVH+wfsvdzocDJzjdA0NqICaHwYF4MbnsOhXE5AiNxHI22jFO3jAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-03T23:17:19.690040Z"},"content_sha256":"f582953bd7a173ab13484d35fc0c2b11dc49a20250eccf49efb962095acc9d68","schema_version":"1.0","event_id":"sha256:f582953bd7a173ab13484d35fc0c2b11dc49a20250eccf49efb962095acc9d68"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/HSPOK5VN2LCBPFIJXBTSX45JCR/bundle.json","state_url":"https://pith.science/pith/HSPOK5VN2LCBPFIJXBTSX45JCR/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/HSPOK5VN2LCBPFIJXBTSX45JCR/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":"2026-06-03T23:17:19Z","links":{"resolver":"https://pith.science/pith/HSPOK5VN2LCBPFIJXBTSX45JCR","bundle":"https://pith.science/pith/HSPOK5VN2LCBPFIJXBTSX45JCR/bundle.json","state":"https://pith.science/pith/HSPOK5VN2LCBPFIJXBTSX45JCR/state.json","well_known_bundle":"https://pith.science/.well-known/pith/HSPOK5VN2LCBPFIJXBTSX45JCR/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:HSPOK5VN2LCBPFIJXBTSX45JCR","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":"22ae36e7fb6ac610402f580714ab826f9c81da91cb3e88afd7dafb4f274cd42c","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-26T20:09:25Z","title_canon_sha256":"c974ccea7fca6e2627f929eb263e0832b8eda1dd2c6feaf1cae9d80dc56ee91e"},"schema_version":"1.0","source":{"id":"2604.23862","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.23862","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"arxiv_version","alias_value":"2604.23862v2","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.23862","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"pith_short_12","alias_value":"HSPOK5VN2LCB","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"pith_short_16","alias_value":"HSPOK5VN2LCBPFIJ","created_at":"2026-05-28T02:04:48Z"},{"alias_kind":"pith_short_8","alias_value":"HSPOK5VN","created_at":"2026-05-28T02:04:48Z"}],"graph_snapshots":[{"event_id":"sha256:f582953bd7a173ab13484d35fc0c2b11dc49a20250eccf49efb962095acc9d68","target":"graph","created_at":"2026-05-28T02:04:48Z","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":"The base GMT v7 model trains stably and exposes centroid usage, transition structure, and source-to-target movement as directly inspectable quantities of the forward computation. It remains behind the larger dense baseline in validation loss and perplexity (3.5995/36.58 vs. 3.2903/26.85), while showing close zero-shot benchmark behavior under the evaluated setting."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That gravitational source routing plus token-conditioned target selection plus gated displacement readout can functionally substitute for the dense FFN transformation without requiring additional capacity or architectural changes outside the memory cell itself."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Graph Memory Transformer (GMT) swaps dense FFN sublayers for a graph of 128 centroids and a learned 128x128 transition matrix per block, yielding a 82M-parameter decoder-only LM that trains stably but trails a 103M dense baseline in perplexity."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"A learned graph of memory centroids can replace the feed-forward sublayer in a decoder-only transformer."}],"snapshot_sha256":"622e7f307e16a4a43f6e050b20f4b25425ce585f5d9c7e61d1f9238ecdf6a960"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-21T07:43:11.410025Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T22:42:36.913523Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2604.23862/integrity.json","findings":[],"snapshot_sha256":"f0922dd35a64782136d1f08b38c0aeb566f919a7180a5d3b6d3866065d4fb412","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We investigate whether the Feed-Forward Network (FFN) sublayer in a decoder-only transformer can be replaced by an explicit learned memory graph while preserving the surrounding autoregressive architecture. The proposed Graph Memory Transformer (GMT) keeps causal self-attention intact, but replaces the usual per-token FFN transformation with a memory cell that routes token representations over a learned bank of centroids connected by a learned directed transition matrix. In the base GMT v7 instantiation studied here, each of 16 transformer blocks contains 128 centroids, a 128 * 128 edge matrix","authors_text":"Niccol\\`o Ferrari, Nicola Zanarini","cross_cats":["cs.AI","cs.CL"],"headline":"A learned graph of memory centroids can replace the feed-forward sublayer in a decoder-only transformer.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-26T20:09:25Z","title":"Graph Memory Transformer (GMT)"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.23862","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-08T06:24:14.512096Z","id":"66cf2d24-7e13-42cf-bc27-690c491710d1","model_set":{"reader":"grok-4.3"},"one_line_summary":"Graph Memory Transformer (GMT) swaps dense FFN sublayers for a graph of 128 centroids and a learned 128x128 transition matrix per block, yielding a 82M-parameter decoder-only LM that trains stably but trails a 103M dense baseline in perplexity.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"A learned graph of memory centroids can replace the feed-forward sublayer in a decoder-only transformer.","strongest_claim":"The base GMT v7 model trains stably and exposes centroid usage, transition structure, and source-to-target movement as directly inspectable quantities of the forward computation. It remains behind the larger dense baseline in validation loss and perplexity (3.5995/36.58 vs. 3.2903/26.85), while showing close zero-shot benchmark behavior under the evaluated setting.","weakest_assumption":"That gravitational source routing plus token-conditioned target selection plus gated displacement readout can functionally substitute for the dense FFN transformation without requiring additional capacity or architectural changes outside the memory cell itself."}},"verdict_id":"66cf2d24-7e13-42cf-bc27-690c491710d1"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:d171e2a4a7f404a0f898f8299e06c4c40b2b1515ddd848b877cc60c54c41baf5","target":"record","created_at":"2026-05-28T02:04:48Z","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":"22ae36e7fb6ac610402f580714ab826f9c81da91cb3e88afd7dafb4f274cd42c","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-26T20:09:25Z","title_canon_sha256":"c974ccea7fca6e2627f929eb263e0832b8eda1dd2c6feaf1cae9d80dc56ee91e"},"schema_version":"1.0","source":{"id":"2604.23862","kind":"arxiv","version":2}},"canonical_sha256":"3c9ee576add2c4179509b8672bf3a9145e2eddb17055a71cf5a98061cacbf45a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3c9ee576add2c4179509b8672bf3a9145e2eddb17055a71cf5a98061cacbf45a","first_computed_at":"2026-05-28T02:04:48.216382Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-28T02:04:48.216382Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"wlrnZGzznPYj7MHRYHQ90bYWE2aBFnpSqyiL6OfbgH+ME6U3ArKVcA9YosLwjINMDrUXv51fgbUKX0FYaroKBw==","signature_status":"signed_v1","signed_at":"2026-05-28T02:04:48.216901Z","signed_message":"canonical_sha256_bytes"},"source_id":"2604.23862","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d171e2a4a7f404a0f898f8299e06c4c40b2b1515ddd848b877cc60c54c41baf5","sha256:f582953bd7a173ab13484d35fc0c2b11dc49a20250eccf49efb962095acc9d68"],"state_sha256":"7e77d89a9cbac7aa99861e49e13534b67b770daadf357db13caca8de33635483"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0yOjjutB8J+24bKB6GE0oPb+JeHfcLEW3qeOXzVLiL5NpksZTKjboDX96rzXh71ZC+T9lJjGQyAu1eujbx8JAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-03T23:17:19.692475Z","bundle_sha256":"17f17b4bb965be2953d86bc5b1a5d8dd49b010b19aca51ea6170445c70fe8f8b"}}