[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-a-new-training-trick-cuts-ai-memory-use-by-up-to-23x":10,"sections":34},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},10946,"a-new-training-trick-cuts-ai-memory-use-by-up-to-23x","A New Training Trick Cuts AI Memory Use by Up to 23x","Phase-HDC trains compact AI classifiers via gradient thresholds instead of optimizer memory, cutting storage up to 23x for about five points of accuracy.","A new training method strips out the memory-hungry optimizer that most AI models carry during training.\n\nResearchers built Phase-HDC for a type of compact classifier called a hyperdimensional model, where each learned parameter is just a low-bit angle - think of it as a dial with a handful of settings instead of a precise decimal number. Normally, training such a model with the Adam optimizer requires storing a running history of past gradients for every parameter, and that history can take up several times more memory than the model itself. Phase-HDC skips the history entirely: it nudges each angle one step opposite its current gradient, and only does so when that gradient is big enough to clear a threshold. The authors show this threshold rule is the exact answer to a simplified version of the training problem where every parameter change carries a fixed cost.\n\nAcross eleven datasets spanning images, tabular data, and text, Phase-HDC used 16 to 23 times less memory than standard 32-bit Adam and 4 to 6 times less than a lower-precision 8-bit version of Adam. That comes at a cost of roughly five accuracy points on average versus full-precision Adam - but Phase-HDC actually beat 8-bit Adam on six of the eleven datasets, including byte-level text prediction, where 8-bit Adam broke down completely.\n\nThat last result is the real finding here: for these discrete-grid models, the optimizer's memory was mostly deciding whether a parameter should move at all - a decision a simple gradient threshold can make for free. Compressing that memory into fewer bits, as 8-bit Adam does, apparently destroys the fine distinctions needed for that decision on sparser inputs like text.\n\nWorth noting: the memory savings reported are logical state, meaning parameter counts, not measured reductions on actual hardware. And this only applies to hyperdimensional models with angle-only parameters, not the dense neural networks most AI teams actually train.","[\"ai\",\"model-training\",\"optimizers\",\"hyperdimensional-computing\"]","2026-10-09T04:00:00.000Z","2026-10-09T22:54:05.164Z","2026-10-09T22:54:10.718Z","published",null,[],"ai",[24,26,27,28],"model-training","optimizers","hyperdimensional-computing",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.10630",0,{"sections":35},[36,39,43,48,53,57,61,66,71,76,81,86,91,96],{"name":37,"slug":24,"count":38,"latest_published_at":18},"AI",6708,{"name":40,"slug":41,"count":42,"latest_published_at":18},"Security","security",931,{"name":44,"slug":45,"count":46,"latest_published_at":47},"Policy","policy",486,"2026-10-08T22:40:11.000Z",{"name":49,"slug":50,"count":51,"latest_published_at":52},"Deals","deals",474,"2026-10-08T22:00:00.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":18},"Hardware","hardware",231,{"name":58,"slug":59,"count":60,"latest_published_at":18},"Science","science",192,{"name":62,"slug":63,"count":64,"latest_published_at":65},"Consumer Tech","consumer-tech",181,"2026-10-08T23:26:35.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Startups","startups",117,"2026-10-08T16:45:00.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Software","software",114,"2026-10-08T17:57:01.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Dev Tools","dev-tools",105,"2026-10-07T16:59:11.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"General","general",66,"2026-10-09T04:46:11.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"Gaming","gaming",58,"2026-10-08T20:08:45.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Reviews","reviews",34,"2026-10-08T14:00:22.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]