{"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/12-in-1-multi-task-vision-and-language","title":"12-in-1: Multi-Task Vision and Language Representation Learning","arxiv_id":"1912.02315","date":"2019-12-05","proceeding":"CVPR 2020 6","authors":["Jiasen Lu","Vedanuj Goswami","Marcus Rohrbach","Devi Parikh","Stefan Lee"],"abstract":"Much of vision-and-language research focuses on a small but diverse set of independent tasks and supporting datasets often studied in isolation; however, the visually-grounded language understanding skills required for success at these tasks overlap significantly. In this work, we investigate these relationships between vision-and-language tasks by developing a large-scale, multi-task training regime. Our approach culminates in a single model on 12 datasets from four broad categories of task including visual question answering, caption-based image retrieval, grounding referring expressions, and multi-modal verification. Compared to independently trained single-task models, this represents a reduction from approximately 3 billion parameters to 270 million while simultaneously improving performance by 2.05 points on average across tasks. We use our multi-task framework to perform in-depth analysis of the effect of joint training diverse tasks. Further, we show that finetuning task-specific models from our single multi-task model can lead to further improvements, achieving performance at or above the state-of-the-art.","url_abs":"https://arxiv.org/abs/1912.02315v2","url_pdf":"https://arxiv.org/pdf/1912.02315v2.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":"12-in-1-multi-task-vision-and-language","repo_url":"https://github.com/facebookresearch/vilbert-multi-task","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"12-in-1-multi-task-vision-and-language","repo_url":"https://github.com/Cloud-CV/vilbert-multi-task","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"12-in-1-multi-task-vision-and-language","repo_url":"https://github.com/jialinwu17/tmpimgs","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"12-in-1-multi-task-vision-and-language","repo_url":"https://github.com/jiasenlu/vilbert_beta","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"12-in-1-multi-task-vision-and-language","repo_url":"https://github.com/johntiger1/multitask_multimodal","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"10-shot-image-generation","task_name":"10-shot image generation"},{"task_slug":"image-retrieval","task_name":"Image Retrieval"},{"task_slug":"question-answering","task_name":"Question Answering"},{"task_slug":"representation-learning","task_name":"Representation Learning"},{"task_slug":"retrieval","task_name":"Retrieval"},{"task_slug":"visual-question-answering-1","task_name":"Visual Question Answering"},{"task_slug":"visual-question-answering","task_name":"Visual Question Answering (VQA)"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/10-shot-image-generation-on-mead","task":"10-shot image generation","dataset":"MEAD","model":"12","rank_in_archive_order":1,"of":1,"metrics":{"12k":"12"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1912.02315","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1912.02315"}},"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":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/jialinwu17/tmpimgs","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/jiasenlu/vilbert_beta","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/facebookresearch/vilbert-multi-task","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/johntiger1/multitask_multimodal","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/Cloud-CV/vilbert-multi-task","reach":{"status":"ok","spdx":"NOASSERTION"}}],"summary":{"ran_draft_wrong":2,"ran":1,"ran_honours":1,"unverified":16},"by_repo_kind":{"official":{"samples":14,"ran":3,"repositories":1},"listed":{"samples":6,"ran":1,"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":20,"samples":[{"code_sha256_prefix":"0f786c407fb1ee4c","entry":"swish","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/basebert.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/basebert.py","link_basis":"harvester_set","language":"python","status":"ran_draft_wrong","verification_level":2,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"0f786c407fb1ee4c"}},{"code_sha256_prefix":"89bdd6decba2b415","entry":"compute_score_with_logits","repo":"johntiger1/multitask_multimodal","repo_kind":"listed","path":"evaluation/eval_vqa.py","file_url":"https://github.com/johntiger1/multitask_multimodal/blob/HEAD/evaluation/eval_vqa.py","link_basis":"harvester_set","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"89bdd6decba2b415"}},{"code_sha256_prefix":"fdc64f4c72036ae4","entry":"gelu","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/basebert.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/basebert.py","link_basis":"harvester_set","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"fdc64f4c72036ae4"}},{"code_sha256_prefix":"5eff22fa0a651276","entry":"url_to_filename","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/utils.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/utils.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":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"5eff22fa0a651276"}},{"code_sha256_prefix":"4c86b7a9d48a484c","entry":"ForwardModelsTrain","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/task_utils.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/task_utils.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":false,"mcp_get_code":{"code_sha256":"4c86b7a9d48a484c"}},{"code_sha256_prefix":"ce5a514305bf7e1b","entry":"ForwardModelsVal","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/task_utils.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/task_utils.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":false,"mcp_get_code":{"code_sha256":"ce5a514305bf7e1b"}},{"code_sha256_prefix":"8a9074f29e11ae33","entry":"LoadLosses","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/task_utils.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/task_utils.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":false,"mcp_get_code":{"code_sha256":"8a9074f29e11ae33"}},{"code_sha256_prefix":"5731b925b4c10219","entry":"cached_path","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/utils.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/utils.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":false,"mcp_get_code":{"code_sha256":"5731b925b4c10219"}},{"code_sha256_prefix":"7a3860e69193cd1e","entry":"deserialize_lmdb","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/datasets/concept_cap_dataset.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/datasets/concept_cap_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":false,"mcp_get_code":{"code_sha256":"7a3860e69193cd1e"}},{"code_sha256_prefix":"2454efa3ee5d7949","entry":"evaluate","repo":"johntiger1/multitask_multimodal","repo_kind":"listed","path":"evaluation/eval_coco_retrieval.py","file_url":"https://github.com/johntiger1/multitask_multimodal/blob/HEAD/evaluation/eval_coco_retrieval.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"2454efa3ee5d7949"}},{"code_sha256_prefix":"cee4c7bd2fc134c9","entry":"evaluate","repo":"johntiger1/multitask_multimodal","repo_kind":"listed","path":"evaluation/eval_concap_retrieval.py","file_url":"https://github.com/johntiger1/multitask_multimodal/blob/HEAD/evaluation/eval_concap_retrieval.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"cee4c7bd2fc134c9"}},{"code_sha256_prefix":"701dca7f22198e4f","entry":"evaluate","repo":"johntiger1/multitask_multimodal","repo_kind":"listed","path":"evaluation/eval_refer_expression.py","file_url":"https://github.com/johntiger1/multitask_multimodal/blob/HEAD/evaluation/eval_refer_expression.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"701dca7f22198e4f"}},{"code_sha256_prefix":"a60fbe82ff2a8bc4","entry":"filename_to_url","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/utils.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/utils.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":false,"mcp_get_code":{"code_sha256":"a60fbe82ff2a8bc4"}},{"code_sha256_prefix":"32d8e4939405d6e2","entry":"get_annotations","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/datasets/flickr_grounding_dataset.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/datasets/flickr_grounding_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":false,"mcp_get_code":{"code_sha256":"32d8e4939405d6e2"}},{"code_sha256_prefix":"7310149bac03cbe2","entry":"get_sentence_data","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/datasets/flickr_grounding_dataset.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/datasets/flickr_grounding_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":false,"mcp_get_code":{"code_sha256":"7310149bac03cbe2"}},{"code_sha256_prefix":"2d56ebf472887b22","entry":"instance_bce_with_logits","repo":"johntiger1/multitask_multimodal","repo_kind":"listed","path":"evaluation/eval_refer_expression.py","file_url":"https://github.com/johntiger1/multitask_multimodal/blob/HEAD/evaluation/eval_refer_expression.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"2d56ebf472887b22"}},{"code_sha256_prefix":"a1bd52099cee7c87","entry":"iou","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/datasets/flickr_grounding_dataset.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/datasets/flickr_grounding_dataset.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"a1bd52099cee7c87"}},{"code_sha256_prefix":"9499ed1255d97132","entry":"iou","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/datasets/concept_cap_dataset.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/datasets/concept_cap_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":false,"mcp_get_code":{"code_sha256":"9499ed1255d97132"}},{"code_sha256_prefix":"baa5766f4566aafc","entry":"load_tf_weights_in_bert","repo":"facebookresearch/vilbert-multi-task","repo_kind":"official","path":"vilbert/vilbert.py","file_url":"https://github.com/facebookresearch/vilbert-multi-task/blob/HEAD/vilbert/vilbert.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"baa5766f4566aafc"}},{"code_sha256_prefix":"50cecb34b46ee0b0","entry":"lr_lambda_update","repo":"johntiger1/multitask_multimodal","repo_kind":"listed","path":"evaluation/eval_vqa.py","file_url":"https://github.com/johntiger1/multitask_multimodal/blob/HEAD/evaluation/eval_vqa.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":false,"mcp_get_code":{"code_sha256":"50cecb34b46ee0b0"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}