{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:TFJ2U5NDIWP2WVU5CZ6HKJOULD","short_pith_number":"pith:TFJ2U5ND","canonical_record":{"source":{"id":"2406.20092","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-06-28T17:57:14Z","cross_cats_sorted":[],"title_canon_sha256":"6190449daeeb774374337f497b862616481c77d1de85ae18e64c76e359ca293c","abstract_canon_sha256":"bcff5bd92aa317033d4153ae26eac6d52917431f35a2416566cb2e16d5fc156a"},"schema_version":"1.0"},"canonical_sha256":"9953aa75a3459fab569d167c7525d458dd0734159282e094cf1011bb356d3418","source":{"kind":"arxiv","id":"2406.20092","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.20092","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"arxiv_version","alias_value":"2406.20092v2","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.20092","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"pith_short_12","alias_value":"TFJ2U5NDIWP2","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"pith_short_16","alias_value":"TFJ2U5NDIWP2WVU5","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"pith_short_8","alias_value":"TFJ2U5ND","created_at":"2026-07-05T09:36:12Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:TFJ2U5NDIWP2WVU5CZ6HKJOULD","target":"record","payload":{"canonical_record":{"source":{"id":"2406.20092","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-06-28T17:57:14Z","cross_cats_sorted":[],"title_canon_sha256":"6190449daeeb774374337f497b862616481c77d1de85ae18e64c76e359ca293c","abstract_canon_sha256":"bcff5bd92aa317033d4153ae26eac6d52917431f35a2416566cb2e16d5fc156a"},"schema_version":"1.0"},"canonical_sha256":"9953aa75a3459fab569d167c7525d458dd0734159282e094cf1011bb356d3418","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:36:12.863713Z","signature_b64":"wKklwBdUxXw+CqDBA2Hj9HD1EW8jLRObgIoM/fpOQwKFtT32JvicvnmBefRnRqb0Ox0eQO4bS8ICRBVHihaoBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9953aa75a3459fab569d167c7525d458dd0734159282e094cf1011bb356d3418","last_reissued_at":"2026-07-05T09:36:12.863217Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:36:12.863217Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2406.20092","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":"2026-07-05T09:36:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"t/bWb6FWRK3Z13W+CXTQ1d6kpRcYc9lLBELMXpOdqsMdPctBxhYiE25bo8cdCa7yvIYZOGU/yfe2kbSvRRweAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T19:30:49.320248Z"},"content_sha256":"545130378068328a0a4cf6232f2146203f1139160aca9fc6b100cc11723f6803","schema_version":"1.0","event_id":"sha256:545130378068328a0a4cf6232f2146203f1139160aca9fc6b100cc11723f6803"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:TFJ2U5NDIWP2WVU5CZ6HKJOULD","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Efficient Large Multi-modal Models via Visual Context Compression","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alan Yuille, Daniel Khashabi, Jieneng Chen, Ju He, Luoxin Ye, Zhao-Yang Wang","submitted_at":"2024-06-28T17:57:14Z","abstract_excerpt":"While significant advancements have been made in compressed representations for text embeddings in large language models (LLMs), the compression of visual tokens in multi-modal LLMs (MLLMs) has remained a largely overlooked area. In this work, we present the study on the analysis of redundancy concerning visual tokens and efficient training within these models. Our initial experiments show that eliminating up to 70% of visual tokens at the testing stage by simply average pooling only leads to a minimal 3% reduction in visual question answering accuracy on the GQA benchmark, indicating signific"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.20092","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/2406.20092/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:36:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vmrQkUxqcUxIxZDXK9ue6pGAZZWQF63+1M60QchJsa4xOhofLpCe1/+28L6z4sFVmFuuH7S/+Crkby1BTkb3Bg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T19:30:49.320861Z"},"content_sha256":"794d7287326c27fef969f47d2432c3629b15eb59a325bdd5a54135457f75a480","schema_version":"1.0","event_id":"sha256:794d7287326c27fef969f47d2432c3629b15eb59a325bdd5a54135457f75a480"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/TFJ2U5NDIWP2WVU5CZ6HKJOULD/bundle.json","state_url":"https://pith.science/pith/TFJ2U5NDIWP2WVU5CZ6HKJOULD/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/TFJ2U5NDIWP2WVU5CZ6HKJOULD/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":"2026-08-09T19:30:49Z","links":{"resolver":"https://pith.science/pith/TFJ2U5NDIWP2WVU5CZ6HKJOULD","bundle":"https://pith.science/pith/TFJ2U5NDIWP2WVU5CZ6HKJOULD/bundle.json","state":"https://pith.science/pith/TFJ2U5NDIWP2WVU5CZ6HKJOULD/state.json","well_known_bundle":"https://pith.science/.well-known/pith/TFJ2U5NDIWP2WVU5CZ6HKJOULD/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:TFJ2U5NDIWP2WVU5CZ6HKJOULD","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":"bcff5bd92aa317033d4153ae26eac6d52917431f35a2416566cb2e16d5fc156a","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-06-28T17:57:14Z","title_canon_sha256":"6190449daeeb774374337f497b862616481c77d1de85ae18e64c76e359ca293c"},"schema_version":"1.0","source":{"id":"2406.20092","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.20092","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"arxiv_version","alias_value":"2406.20092v2","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.20092","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"pith_short_12","alias_value":"TFJ2U5NDIWP2","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"pith_short_16","alias_value":"TFJ2U5NDIWP2WVU5","created_at":"2026-07-05T09:36:12Z"},{"alias_kind":"pith_short_8","alias_value":"TFJ2U5ND","created_at":"2026-07-05T09:36:12Z"}],"graph_snapshots":[{"event_id":"sha256:794d7287326c27fef969f47d2432c3629b15eb59a325bdd5a54135457f75a480","target":"graph","created_at":"2026-07-05T09:36:12Z","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"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2406.20092/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"While significant advancements have been made in compressed representations for text embeddings in large language models (LLMs), the compression of visual tokens in multi-modal LLMs (MLLMs) has remained a largely overlooked area. In this work, we present the study on the analysis of redundancy concerning visual tokens and efficient training within these models. Our initial experiments show that eliminating up to 70% of visual tokens at the testing stage by simply average pooling only leads to a minimal 3% reduction in visual question answering accuracy on the GQA benchmark, indicating signific","authors_text":"Alan Yuille, Daniel Khashabi, Jieneng Chen, Ju He, Luoxin Ye, Zhao-Yang Wang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-06-28T17:57:14Z","title":"Efficient Large Multi-modal Models via Visual Context Compression"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.20092","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":"sha256:545130378068328a0a4cf6232f2146203f1139160aca9fc6b100cc11723f6803","target":"record","created_at":"2026-07-05T09:36:12Z","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":"bcff5bd92aa317033d4153ae26eac6d52917431f35a2416566cb2e16d5fc156a","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-06-28T17:57:14Z","title_canon_sha256":"6190449daeeb774374337f497b862616481c77d1de85ae18e64c76e359ca293c"},"schema_version":"1.0","source":{"id":"2406.20092","kind":"arxiv","version":2}},"canonical_sha256":"9953aa75a3459fab569d167c7525d458dd0734159282e094cf1011bb356d3418","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9953aa75a3459fab569d167c7525d458dd0734159282e094cf1011bb356d3418","first_computed_at":"2026-07-05T09:36:12.863217Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:36:12.863217Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"wKklwBdUxXw+CqDBA2Hj9HD1EW8jLRObgIoM/fpOQwKFtT32JvicvnmBefRnRqb0Ox0eQO4bS8ICRBVHihaoBg==","signature_status":"signed_v1","signed_at":"2026-07-05T09:36:12.863713Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.20092","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:545130378068328a0a4cf6232f2146203f1139160aca9fc6b100cc11723f6803","sha256:794d7287326c27fef969f47d2432c3629b15eb59a325bdd5a54135457f75a480"],"state_sha256":"f9195de4b8051bfa278442daefa906c72f13f20c81e3ed49130a7e4035196cc9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ApieqYp4PcApUuXcqSMM1OfBqPedhtW9xqAuEbiiRXk6d+dVKkcirB/5TZfp6wi82e0cnKcBEF3AK9JjnHcnCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T19:30:49.326349Z","bundle_sha256":"6ead4f96ffb6aaf34c4aab5282ed1de750a597e37c531edd7cb272ffb5239f91"}}