{"ok":true,"nota":{"id":"a5398b76b1582d9b","slug":"una-forma-de-comprimir-modelos-de-lenguaje-con-menos-memoria-a5398b","titulo":"Una forma de comprimir modelos de lenguaje con menos memoria","resumen":"Trabajo sobre representaciones de baja precisión para embeddings de n-gramas en modelos de lenguaje. Sin cifras de compresión ni de desempeño en el material recibido.","fuente":"arxiv-ia","fuenteNombre":"arXiv · Inteligencia artificial","color":"#b31b1b","autor":"Stephen Chung","url":"https://arxiv.org/abs/2610.31646v1","imagen":"/hoy/portada/a5398b76b1582d9b","tema":"tecnologia","subtemas":["compresión de modelos","embeddings","modelos de lenguaje","eficiencia","cuantización"],"entidades":["arXiv"],"formato":"dato","entretenimiento":false,"pubTs":"2026-10-22T17:56:13.000Z","indexTs":"2026-10-23T02:32:17.920Z","hash":"f763f6ddf1c150ff3d126b8d5767a2a445a2a2ef3214ee4924e83b5ddad8fe0d","tsa":{"time":"Oct 23 02:32:40 2026 GMT","provider":"freetsa.org","sobre":"sello"},"stats":{"views":0,"likes":0,"clicks":0,"dwell":0},"tipo":"normal","registro":"hallazgo","porQueImporta":"Comprimir sus componentes sin perder mucha calidad permitiría usarlos en equipos más modestos. Para quien tiene una notebook común, eso puede significar acceso más barato. Sin los números del trabajo, el alcance real queda abierto.","quienFirma":"arXiv (preprint, institución no informada)","confianza":"preprint","consulta":null},"sello":{"hash":"f763f6ddf1c150ff3d126b8d5767a2a445a2a2ef3214ee4924e83b5ddad8fe0d","sello":"c8c167284fca0b3c5874d8520ffcbaeb9b789114ca7fe900c3f82e0f98d730ff","beacon":{"drand":{"cadena":"8990e7a9aaed2ffed73dbd7092123d6f289930540d7651336225dc172e51b2ce","ronda":6576390,"aleatorio":"1d3e1d214bdf3ef8eddcba7f18593c0574622aab3780a3fcf3cb6a38dbf4c6f8","tiempo":"2026-10-23T02:32:00.000Z"},"nist":{"cadena":2,"pulso":1995394,"salida":"3F6D985D1327226AE4457CDD8F996DA2081E15665A7E38EFF78EDFFDD0693E0FFF244359B0BF6D6DE415CC30037EF9A3867957EA0F7FC4D1204E1012C22BD83D","tiempo":"2026-10-23T02:31:00.000Z"},"noAntes":"2026-10-23T02:32:00.000Z"},"verificacion":{"drand":true,"nist":true},"tsa":{"time":"Oct 23 02:32:40 2026 GMT","provider":"freetsa.org","sobre":"sello"},"original":{"titulo":"Low-Bit N-Gram Embeddings for Language Models","descripcion":"","pubDate":"2026-10-22T17:56:13Z","url":"https://arxiv.org/abs/2610.31646v1"},"indexado":"2026-10-23T02:32:17.920Z"}}