{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/scaling-vision-pre-training-to-4k-resolution","title":"Scaling Vision Pre-Training to 4K Resolution","arxiv_id":"2503.19903","date":"2025-03-25","proceeding":"CVPR 2025 1","authors":["Baifeng Shi","Boyi Li","Han Cai","Yao Lu","Sifei Liu","Marco Pavone","Jan Kautz","Song Han","Trevor Darrell","Pavlo Molchanov","Hongxu Yin"],"abstract":"High-resolution perception of visual details is crucial for daily tasks. Current vision pre-training, however, is still limited to low resolutions (e.g., 378 x 378 pixels) due to the quadratic cost of processing larger images. We introduce PS3 that scales CLIP-style vision pre-training to 4K resolution with a near-constant cost. Instead of contrastive learning on global image representation, PS3 is pre-trained by selectively processing local regions and contrasting them with local detailed captions, enabling high-resolution representation learning with greatly reduced computational overhead. The pre-trained PS3 is able to both encode the global image at low resolution and selectively process local high-resolution regions based on their saliency or relevance to a text prompt. When applying PS3 to multi-modal LLM (MLLM), the resulting model, named VILA-HD, significantly improves high-resolution visual perception compared to baselines without high-resolution vision pre-training such as AnyRes and S^2 while using up to 4.3x fewer tokens. PS3 also unlocks appealing scaling properties of VILA-HD, including scaling up resolution for free and scaling up test-time compute for better performance. Compared to state of the arts, VILA-HD outperforms previous MLLMs such as NVILA and Qwen2-VL across multiple benchmarks and achieves better efficiency than latest token pruning approaches. Finally, we find current benchmarks do not require 4K-resolution perception, which motivates us to propose 4KPro, a new benchmark of image QA at 4K resolution, on which VILA-HD outperforms all previous MLLMs, including a 14.5% improvement over GPT-4o, and a 3.2% improvement and 2.96x speedup over Qwen2-VL.","url_abs":"https://arxiv.org/abs/2503.19903v1","url_pdf":"https://arxiv.org/pdf/2503.19903v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"scaling-vision-pre-training-to-4k-resolution","repo_url":"https://github.com/efficient-large-model/vila","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"scaling-vision-pre-training-to-4k-resolution","repo_url":"https://github.com/nvlabs/vila","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"4k","task_name":"4k"},{"task_slug":"contrastive-learning","task_name":"Contrastive Learning"}],"methods":[{"method_slug":"contrastive-learning","method_name":"Contrastive Learning"},{"method_slug":"pruning","method_name":"Pruning"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2503.19903","atlas_url":"https://app.syntology.ai/?focus=2503.19903","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2503.19903"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/efficient-large-model/vila","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/nvlabs/vila","reach":null}],"summary":{"ran_violates":1,"ran_fixture":1,"ran":5,"ran_draft_wrong":3,"ran_honours":1,"unverified":3},"by_repo_kind":{"listed":{"samples":14,"ran":11,"repositories":2}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"8dd6b65f72c07727","entry":"check_dot_in_model_path","repo":"efficient-large-model/vila","repo_kind":"listed","path":"llava/remote_code/modeling_vila.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/llava/remote_code/modeling_vila.py","link_basis":"harvester_set","language":"python","status":"ran_violates","verification_level":2,"contract_check":"VIOLATES","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"8dd6b65f72c07727"}},{"code_sha256_prefix":"30d7eec482ebf6b1","entry":"repeat_kv","repo":"efficient-large-model/vila","repo_kind":"listed","path":"llava/eval/vision_niah_vila/zigzag_ring_attn/modeling_qwen2.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/llava/eval/vision_niah_vila/zigzag_ring_attn/modeling_qwen2.py","link_basis":"harvester_set","language":"python","status":"ran_fixture","verification_level":2,"contract_check":"RAISES","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"30d7eec482ebf6b1"}},{"code_sha256_prefix":"232455d9c7b1d476","entry":"VisionTower","repo":"nvlabs/vila","repo_kind":"listed","path":"llava/model/multimodal_encoder/ps3_encoder.py","file_url":"https://github.com/nvlabs/vila/blob/HEAD/llava/model/multimodal_encoder/ps3_encoder.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"232455d9c7b1d476"}},{"code_sha256_prefix":"5d81bffdb6022427","entry":"apply_rotary_pos_emb","repo":"efficient-large-model/vila","repo_kind":"listed","path":"llava/eval/vision_niah_vila/zigzag_ring_attn/modeling_qwen2.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/llava/eval/vision_niah_vila/zigzag_ring_attn/modeling_qwen2.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"5d81bffdb6022427"}},{"code_sha256_prefix":"554c742ab091b991","entry":"get_vila_version","repo":"efficient-large-model/vila","repo_kind":"listed","path":"llava/remote_code/modeling_vila.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/llava/remote_code/modeling_vila.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"554c742ab091b991"}},{"code_sha256_prefix":"70759df34aae94c5","entry":"load_haystack","repo":"efficient-large-model/vila","repo_kind":"listed","path":"llava/eval/vision_niah_vila/eval_vision_niah.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/llava/eval/vision_niah_vila/eval_vision_niah.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"70759df34aae94c5"}},{"code_sha256_prefix":"1c69e761fdf05a26","entry":"load_image","repo":"efficient-large-model/vila","repo_kind":"listed","path":"server.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/server.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"1c69e761fdf05a26"}},{"code_sha256_prefix":"fca9a6d2afa23288","entry":"load_video","repo":"efficient-large-model/vila","repo_kind":"listed","path":"server.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/server.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"fca9a6d2afa23288"}},{"code_sha256_prefix":"b99eea6376d1e212","entry":"rotate_half","repo":"efficient-large-model/vila","repo_kind":"listed","path":"llava/eval/vision_niah_vila/zigzag_ring_attn/modeling_qwen2.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/llava/eval/vision_niah_vila/zigzag_ring_attn/modeling_qwen2.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"b99eea6376d1e212"}},{"code_sha256_prefix":"300bb78b66d35115","entry":"safe_tokenize","repo":"efficient-large-model/vila","repo_kind":"listed","path":"llava/eval/vision_niah_vila/eval_vision_niah.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/llava/eval/vision_niah_vila/eval_vision_niah.py","link_basis":"harvester_set","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"300bb78b66d35115"}},{"code_sha256_prefix":"7af38f0a2846676a","entry":"with_opencv","repo":"efficient-large-model/vila","repo_kind":"listed","path":"data_prepare/panda_split.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/data_prepare/panda_split.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"7af38f0a2846676a"}},{"code_sha256_prefix":"ef6066f5b1710c5d","entry":"PS3VisionTower","repo":"nvlabs/vila","repo_kind":"listed","path":"llava/model/multimodal_encoder/ps3_encoder.py","file_url":"https://github.com/nvlabs/vila/blob/HEAD/llava/model/multimodal_encoder/ps3_encoder.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ef6066f5b1710c5d"}},{"code_sha256_prefix":"aa7f4da9de0dc7cc","entry":"build_mm_projector","repo":"efficient-large-model/vila","repo_kind":"listed","path":"llava/remote_code/modeling_vila.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/llava/remote_code/modeling_vila.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"aa7f4da9de0dc7cc"}},{"code_sha256_prefix":"f8e6d235cb12b138","entry":"get_literal_values","repo":"efficient-large-model/vila","repo_kind":"listed","path":"server.py","file_url":"https://github.com/efficient-large-model/vila/blob/HEAD/server.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"f8e6d235cb12b138"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}