<\/p>\n\n\n\n
While standard Next-Token Prediction (NTP) lays the foundation of language model pre- training, its teacher-forced training paradigm may not be optimal for long-horizon reasoning and planning. Recent works such as Multi-Token Prediction (MTP) and Next-Latent prediction (NextLat) try to mitigate the problem through predicting multiple future tokens and self-supervised prediction in the latent space. However, those auxiliary objectives either have a limited horizon or suffer from compounding error from multi-step rollout. We introduce Hierarchical Latent Prediction (HiLP), which introduces an auxiliary higher-level abstract latent to help reduce the error accumulation effect in latent-space rollouts. Experiments show that HiLP can lead to longer-horizon coherent belief state representation and demonstrate the effectiveness of our method across coding and multi-step reasoning benchmarks, and offers more speculative decoding efficiency.<\/p>\n","protected":false},"excerpt":{"rendered":"
While standard Next-Token Prediction (NTP) lays the foundation of language model pre- training, its teacher-forced training paradigm may not be optimal for long-horizon reasoning and planning. Recent works such as Multi-Token Prediction (MTP) and Next-Latent prediction (NextLat) try to mitigate the problem through predicting multiple future tokens and self-supervised prediction in the latent space. However, […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":[{"type":"text","value":"Changyan Shi","user_id":0},{"type":"user_nicename","value":"Tim Pearce","user_id":"44054"},{"type":"user_nicename","value":"Manan Tomar","user_id":"43986"},{"type":"user_nicename","value":"Siddhartha Sen","user_id":"33656"},{"type":"user_nicename","value":"John Langford","user_id":"32204"}],"msr_publishername":"","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"","msr_editors":"","msr_how_published":"arXiv","msr_isbn":"","msr_issue":"","msr_journal":"","msr_number":"","msr_organization":"","msr_pages_string":"","msr_page_range_start":"","msr_page_range_end":"","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"","msr_doi":"","msr_arxiv_id":"2608.05806","msr_mag_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_release_tracker_id":"","msr_highlight_type":"","msr_date_display_format":"","msr_main_download_label":"","msr_external_link_label":"","msr_doi_label":"","msr_published_date":"2026-08-06","msr_startdate":"","msr_presentation_date":"","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_year":2026,"msr_month":8,"msr_day":6,"msr_microsoftintellectualproperty":false,"msr_pub_id":"fcdb6618a32e1be6e51f70ca02a8934d637f1f52","msr_publication_uploader":[{"type":"url","title":"https:\/\/arxiv.org\/abs\/2608.05806","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_original_fields_of_study":[],"msr_s2_paper_id":"fcdb6618a32e1be6e51f70ca02a8934d637f1f52","msr_s2_pdf_url":"","msr_citation_count_updated":"","msr_citation_count":0,"msr_influential_citations":0,"msr_reference_count":19,"msr_s2_open_access":false,"msr_s2_author_ids":[],"msr_pub_ids":[{"provider":"s2","id":"fcdb6618a32e1be6e51f70ca02a8934d637f1f52"},{"provider":"arxiv","id":"2608.05806"},{"provider":"corpusid","id":"290852271"}],"msr_hide_image_in_river":0,"footnotes":""},"msr-research-highlight":[],"research-area":[13556,13545],"msr-publication-type":[270373],"msr-publisher":[],"msr-publication-cta":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[246691,265497,267249],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-1181843","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-artificial-intelligence","msr-research-area-human-language-technologies","msr-locale-en_us","msr-field-of-study-computer-science","msr-field-of-study-machine-learning-296","msr-field-of-study-natural-language-processing-184"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2026-08-06","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"arXiv","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":0,"msr_main_download":"","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"url","title":"https:\/\/arxiv.org\/abs\/2608.05806","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_citation_count":0,"msr_citation_count_updated":"","msr_s2_paper_id":"fcdb6618a32e1be6e51f70ca02a8934d637f1f52","msr_influential_citations":0,"msr_reference_count":19,"msr_arxiv_id":"2608.05806","msr_s2_author_ids":[],"msr_s2_open_access":false,"msr_s2_pdf_url":null,"msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"Changyan Shi","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Tim Pearce","user_id":44054,"rest_url":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Tim Pearce"},{"type":"user_nicename","value":"Manan Tomar","user_id":43986,"rest_url":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Manan Tomar"},{"type":"user_nicename","value":"Siddhartha Sen","user_id":33656,"rest_url":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Siddhartha Sen"},{"type":"user_nicename","value":"John Langford","user_id":32204,"rest_url":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=John Langford"}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[],"msr_project":[],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"misc","related_content":[],"_links":{"self":[{"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1181843","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":2,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1181843\/revisions"}],"predecessor-version":[{"id":1184384,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1181843\/revisions\/1184384"}],"wp:attachment":[{"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/media?parent=1181843"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=1181843"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=1181843"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=1181843"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=1181843"},{"taxonomy":"msr-publication-cta","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-cta?post=1181843"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=1181843"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=1181843"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=1181843"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=1181843"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=1181843"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=1181843"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=1181843"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=1181843"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}