{"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/long-text-generation-via-adversarial-training","title":"Long Text Generation via Adversarial Training with Leaked Information","arxiv_id":"1709.08624","date":"2017-09-24","proceeding":null,"authors":["Jiaxian Guo","Sidi Lu","Han Cai","Wei-Nan Zhang","Yong Yu","Jun Wang"],"abstract":"Automatically generating coherent and semantically meaningful text has many\napplications in machine translation, dialogue systems, image captioning, etc.\nRecently, by combining with policy gradient, Generative Adversarial Nets (GAN)\nthat use a discriminative model to guide the training of the generative model\nas a reinforcement learning policy has shown promising results in text\ngeneration. However, the scalar guiding signal is only available after the\nentire text has been generated and lacks intermediate information about text\nstructure during the generative process. As such, it limits its success when\nthe length of the generated text samples is long (more than 20 words). In this\npaper, we propose a new framework, called LeakGAN, to address the problem for\nlong text generation. We allow the discriminative net to leak its own\nhigh-level extracted features to the generative net to further help the\nguidance. The generator incorporates such informative signals into all\ngeneration steps through an additional Manager module, which takes the\nextracted features of current generated words and outputs a latent vector to\nguide the Worker module for next-word generation. Our extensive experiments on\nsynthetic data and various real-world tasks with Turing test demonstrate that\nLeakGAN is highly effective in long text generation and also improves the\nperformance in short text generation scenarios. More importantly, without any\nsupervision, LeakGAN would be able to implicitly learn sentence structures only\nthrough the interaction between Manager and Worker.","url_abs":"http://arxiv.org/abs/1709.08624v2","url_pdf":"http://arxiv.org/pdf/1709.08624v2.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":"long-text-generation-via-adversarial-training","repo_url":"https://github.com/CR-Gjx/LeakGAN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"long-text-generation-via-adversarial-training","repo_url":"https://github.com/liyzcj/leakgan-py3","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"long-text-generation-via-adversarial-training","repo_url":"https://github.com/nurpeiis/LeakGAN-PyTorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"long-text-generation-via-adversarial-training","repo_url":"https://github.com/rupes438/CodeGen","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"long-text-generation-via-adversarial-training","repo_url":"https://github.com/universebh/text_generation_fsa_gan","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"long-text-generation-via-adversarial-training","repo_url":"https://github.com/valko073/LyricsGANs","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"sentence","task_name":"Sentence"},{"task_slug":"text-generation","task_name":"Text Generation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/text-generation-on-coco-captions","task":"Text Generation","dataset":"COCO Captions","model":"LeakGAN","rank_in_archive_order":1,"of":5,"metrics":{"BLEU-2":"0.950","BLEU-3":"0.880","BLEU-4":"0.778","BLEU-5":"0.686"},"uses_additional_data":false},{"leaderboard":"/sota/text-generation-on-coco-captions","task":"Text Generation","dataset":"COCO Captions","model":"partGAN","rank_in_archive_order":2,"of":5,"metrics":{"BLEU-2":"0.910","BLEU-3":"0.713","BLEU-4":"O.753","BLEU-5":"0.590"},"uses_additional_data":false},{"leaderboard":"/sota/text-generation-on-chinese-poems","task":"Text Generation","dataset":"Chinese Poems","model":"LeakGAN","rank_in_archive_order":3,"of":3,"metrics":{"BLEU-2":"0.456"},"uses_additional_data":false},{"leaderboard":"/sota/text-generation-on-emnlp2017-wmt","task":"Text Generation","dataset":"EMNLP2017 WMT","model":"LeakGAN","rank_in_archive_order":1,"of":5,"metrics":{"BLEU-2":"0.956","BLEU-3":"0.819","BLEU-4":"0.627","BLEU-5":"0.498"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1709.08624","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1709.08624"}},"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/rupes438/CodeGen","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/nurpeiis/LeakGAN-PyTorch","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/valko073/LyricsGANs","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/liyzcj/leakgan-py3","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/CR-Gjx/LeakGAN","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/universebh/text_generation_fsa_gan","reach":{"status":"unanswered"}}],"summary":{"ran_honours":2},"by_repo_kind":{},"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":2,"samples":[{"code_sha256_prefix":"d742e892dc2e5813","entry":"pre_train_epoch","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"d742e892dc2e5813"}},{"code_sha256_prefix":"336fb3340370addd","entry":"target_loss","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"336fb3340370addd"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}