FlashMap acelera la decodificación de LLM hasta 6,14 veces mediante el uso de mapas de flujo continuos, ofreciendo una alternativa de código abierto más rápida para una inferencia de IA eficiente.
Resumen generado con IA. Consulta el original para comprobar el contexto y la precisión.
Este es el extracto disponible del feed. Puede estar abreviado o en el idioma original.
Researchers from Polymathic AI, Princeton University, USC ISI, NYU and the Flatiron Institute have released FlashMap. FlashMap speculative decoding replaces the discrete diffusion drafter with a continuous flow map. It drafts a full 16-token block in 1 forward pass. Tokens in that block stay consistent with each other. On Qwen3-8B, the research team reports up to 6.14x lossless speedup on GSM8K. Drafter weights for Qwen3-8B and Qwen3-4B are on Hugging Face, with vLLM and SGLang plugins. TL;DR Size: 0.5B-parameter drafter (500M for Qwen3-8B, 475M for Qwen3-4B); drafts blocks of K = 16 tokens. Runs on: the same GPU as the target model. Paper speedups use 1 NVIDIA H200 NVL at concurrency 1, in bf16. Performance: fastest drafter tested on Qwen3-8B, about 7% above a retrained DFlash on average. Best: 6.14x speedup on GSM8K at temperature 0 (acceptance length 7.69 tokens). Worst: 2.28x speedup on Alpaca at temperature 1. On Qwen3-4B (temperature 0), it trails DFlash on 4 of 9 tasks. What is FlashMap? FlashMap is a speculative decoding drafter that maps Gaussian noise to a token block in 1 step. A small drafter proposes K tokens. The target model then checks all K in 1 parallel pass and keeps the longest matching prefix. Output stays exactly what the target would produce. Speed depends on 2 things. How many drafted tokens get accepted, and how cheaply they were drafted. EAGLE-3 drafts autoregressively, so cost grows with block length. DFlash drafts the full block in 1 pass with block diffusion. Why do diffusion drafters lose acceptance? DFlash predicts each slot in the block independently. The research team calls this the factorized approximation. Picture a continuation that is “new york” or “san diego” with equal odds. An independent drafter proposes “new diego” or “san york” half the time. The target rejects those at slot 2. FlashMap builds on Flow Map Language Models. Each noise sample maps to 1 full block. The noise picks the mode for every slot jointly. H
Una ayuda automática para ordenar, basada en fuentes, palabras clave y entidades, en una escala de 1–12. Alta: 9–12; media: 6–8; baja: 1–5. No mide urgencia, verifica noticias ni representa respaldo editorial.
La integración de Genie 4 con un planificador Gemini en su generación de mundo en tiempo real con audio ofrece un plano arquitectónico crítico para la construcción de entornos de entrenamiento de IA avanzados.
El desarrollo de máquinas con una audición superior permite aplicaciones críticas, como la localización rápida de accidentes, mejorando significativamente la percepción artificial más allá de las capacidades humanas.
LEANN reduce drásticamente el almacenamiento RAG al recomputar los embeddings al instante, lo que permite búsquedas semánticas a gran escala en laptops estándar.