{"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/arxiv-2601-20357","title":"TABED: Test-Time Adaptive Ensemble Drafting for Robust Speculative Decoding in LVLMs","arxiv_id":"2601.20357","date":"2026-01-28","proceeding":null,"authors":["Minjae Lee","Wonjun Kang","Byeongkeun Ahn","Christian Classen","Kevin Galim","Seunghyuk Oh","Minghao Yan","Hyung Il Koo","Kangwook Lee"],"abstract":"Speculative decoding (SD) has proven effective for accelerating LLM inference by quickly generating draft tokens and verifying them in parallel. However, SD remains largely unexplored for Large Vision-Language Models (LVLMs), which extend LLMs to process both image and text prompts. To address this gap, we benchmark existing inference methods with small draft models on 11 datasets across diverse input scenarios and observe scenario-specific performance fluctuations. Motivated by these findings, we propose Test-time Adaptive Batched Ensemble Drafting (TABED), which dynamically ensembles multiple drafts obtained via batch inference by leveraging deviations from past ground truths available in the SD setting. The dynamic ensemble method achieves an average robust walltime speedup of 1.74x over autoregressive decoding and a 5% improvement over single drafting methods, while remaining training-free and keeping ensembling costs negligible through parameter sharing. With its plug-and-play compatibility, we further enhance TABED by integrating advanced verification and alternative drafting methods. Code and custom-trained models are available at https://github.com/furiosa-ai/TABED.","url_abs":"https://arxiv.org/abs/2601.20357","url_pdf":"https://arxiv.org/pdf/2601.20357","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2601.20357","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2601.20357"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+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":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/furiosa-ai/TABED","reach":null}],"summary":{"unverified":6},"by_repo_kind":{"found_in_text":{"samples":6,"ran":0,"repositories":1}},"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":"b3b13886988b3982","entry":"capture_config","repo":"furiosa-ai/TABED","repo_kind":"found_in_text","path":"tabed/config.py","file_url":"https://github.com/furiosa-ai/TABED/blob/HEAD/tabed/config.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"b3b13886988b3982"}},{"code_sha256_prefix":"05d9bde1e9a1636a","entry":"clean_prompt","repo":"furiosa-ai/TABED","repo_kind":"found_in_text","path":"tabed/datamodules/dataset.py","file_url":"https://github.com/furiosa-ai/TABED/blob/HEAD/tabed/datamodules/dataset.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"05d9bde1e9a1636a"}},{"code_sha256_prefix":"4d169eb369d478cd","entry":"load_image_processor","repo":"furiosa-ai/TABED","repo_kind":"found_in_text","path":"tabed/modules/load_pretrained.py","file_url":"https://github.com/furiosa-ai/TABED/blob/HEAD/tabed/modules/load_pretrained.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"4d169eb369d478cd"}},{"code_sha256_prefix":"dd45268c213d8548","entry":"load_pretrained_model","repo":"furiosa-ai/TABED","repo_kind":"found_in_text","path":"tabed/modules/load_pretrained.py","file_url":"https://github.com/furiosa-ai/TABED/blob/HEAD/tabed/modules/load_pretrained.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"dd45268c213d8548"}},{"code_sha256_prefix":"433fce01b1f7cb60","entry":"load_tokenizer","repo":"furiosa-ai/TABED","repo_kind":"found_in_text","path":"tabed/modules/load_pretrained.py","file_url":"https://github.com/furiosa-ai/TABED/blob/HEAD/tabed/modules/load_pretrained.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"433fce01b1f7cb60"}},{"code_sha256_prefix":"b15ba266145d6889","entry":"measure_time","repo":"furiosa-ai/TABED","repo_kind":"found_in_text","path":"tabed/modules/eval_specbench.py","file_url":"https://github.com/furiosa-ai/TABED/blob/HEAD/tabed/modules/eval_specbench.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"b15ba266145d6889"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.LG","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}