All public logs

Combined display of all available logs of Marovi AI. You can narrow down the view by selecting a log type, the username (case-sensitive), or the affected page (also case-sensitive).

• 03:25, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/25/es (Created page with "Registramos las trayectorias de entrenamiento del agente durante todo el proceso de formación, lo que incluye diferentes niveles de habilidad de juego. Este conjunto de trayectorias registradas constituye nuestro conjunto de datos <math>\mathcal{T}_{agent}</math>, utilizado para entrenar el modelo generativo (véase la sección [https://arxiv.org/html/2408.14837v1#S3.SS2 3.2]).")
• 03:24, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/1/es (Created page with "'''Autores:''' Dani Valevski (Google Research), Yaniv Leviathan (Google Research), Moab Arar (Universidad de Tel Aviv), Shlomi Fruchter (Google DeepMind)")
• 03:23, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/1/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 03:23, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/25/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 03:22, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/50/es (Created page with "Entrenamos todos los modelos de simulación a partir de un punto de control preentrenado de Stable Diffusion 1.4, descongelando todos los parámetros de U-Net. Utilizamos un tamaño de lote de 128 y una tasa de aprendizaje constante de 2e-5, con el optimizador Adafactor sin decaimiento de peso (Shazeer & Stern, [https://arxiv.org/html/2408.14837v1#bib.bib31 2018]) y recorte de gradiente de 1,0. Cambiamos la parametrización de la pérdida de difusión para que sea predic...")
• 03:22, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/49/es (Created page with "=== 4.2 Entrenamiento del Modelo Generativo ===")
• 03:22, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/48/es (Created page with "El modelo de agente se entrena utilizando PPO (Schulman et al., [https://arxiv.org/html/2408.14837v1#bib.bib30 2017]), con una CNN simple como red de características, siguiendo a Mnih et al. ([https://arxiv.org/html/2408.14837v1#bib.bib21 2015]). Se entrena en CPU utilizando la infraestructura de Stable Baselines 3 (Raffin et al., [https://arxiv.org/html/2408.14837v1#bib.bib24 2021]). Al agente se le proporcionan versiones reducidas de las imágenes de los fotogramas y...")
• 03:22, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/47/es (Created page with "=== 4.1 Entrenamiento del Agente ===")
• 03:21, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/46/es (Created page with "== 4 Configuración Experimental ==")
• 03:19, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/37/es (Created page with "=== 3.3 Inferencia ===")
• 03:17, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/31/es (Created page with "donde <math>T = {\{ o_{i \leq n},a_{i \leq n}\}} \sim \mathcal{T}_{agente}</math>, <math>x_{0} = {\phi{(o_{n})}}</math>, <math>t \sim {\mathcal{U}{(0,1)}}</math>, <math>\epsilon \sim {\mathcal{N}{(0,\mathbf{I})}}</math>, <math>x_{t} = {\sqrt{\overline{\alpha}_{t}}x_{0} + \sqrt{1 - \overline{\alpha}_{t}}\epsilon}</math>, <math>{v(\epsilon,x_{0},t)} = {\sqrt{\overline{\alpha}_{t}}\epsilon - \sqrt{1 - \overline{\alpha}_{t}}x_{0}}</math>, y <math>v_{\theta^{\prime}}</math> e...")
• 03:17, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/29/es (Created page with "Entrenamos el modelo para minimizar la pérdida de difusión usando la parametrización de velocidad (Salimans & Ho, [https://arxiv.org/html/2408.14837v1#bib.bib29 2022b]):")
• 03:17, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/28/es (Created page with "Reutilizamos un modelo de difusión de texto a imagen preentrenado, Stable Diffusion v1.4 (Rombach et al., [https://arxiv.org/html/2408.14837v1#bib.bib26 2022]). Condicionamos el modelo <math>f_{\theta}</math> en trayectorias <math>T \sim \mathcal{T}_{agent}</math>, es decir, en una secuencia de acciones previas <math>a_{< n}</math> y observaciones (fotogramas) <math>o_{< n}</math>, y eliminamos todo condicionamiento textual. Específicamente, para condicionar en las acc...")
• 03:16, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/26/es (Created page with "=== 3.2 Entrenamiento del Modelo Generativo de Difusión ===")
• 03:15, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/18/es (Created page with "Dado un entorno interactivo de entrada <math>\mathcal{E}</math>, y un estado inicial <math>s_{0} \in \mathcal{S}</math>, una ''simulación de mundo interactivo'' es una ''función de distribución de simulación'' <math>q \left( o_{n} \,|\, \{o_{< n}, a_{\leq n}\} \right), \; o_{i} \in \mathcal{O}, \; a_{i} \in \mathcal{A}</math>. Dada una métrica de distancia entre observaciones <math>D: \mathcal{O} \times \mathcal{O} \rightarrow \mathbb{R}</math>, una ''política'', e...")
• 03:14, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/15/es (Created page with "== 2 Simulación Interactiva del Mundo ==")
• 03:14, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/12/es (Created page with "En este trabajo, demostramos que la respuesta es afirmativa. En concreto, mostramos que un videojuego complejo, el emblemático juego DOOM, puede ejecutarse en una red neuronal (una versión aumentada de la Stable Diffusion v1.4 abierta (Rombach et al., [https://arxiv.org/html/2408.14837v1#bib.bib26 2022])), en tiempo real, logrando una calidad visual comparable a la del juego original. Aunque no es una simulación exacta, el modelo neuronal es capaz de realizar actualiz...")
• 03:13, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/8/es (Created page with "Los juegos de computadora son sistemas de software creados manualmente centrados en el siguiente ''bucle de juego'': (1) recopilar las entradas del usuario, (2) actualizar el estado del juego y (3) renderizarlo en píxeles en la pantalla. Este bucle de juego, que se ejecuta a altas velocidades de fotogramas, crea la ilusión de un mundo virtual interactivo para el jugador. Estos bucles de juego clásicamente se ejecutan en computadoras estándar, y aunque ha habido mucho...")
• 03:12, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/3/es (Created page with "'''Sitio web del proyecto:''' [https://gamengen.github.io/ https://gamengen.github.io]")
• 03:10, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/12/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 03:09, 7 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/Page display title/es (Created page with "Diffusion Models Are Real-Time Game Engines")
• 03:08, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/15/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 03:08, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/37/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 03:06, 7 September 2024 Felipefelixarias talk contribs marked Diffusion Models Are Real-Time Game Engines for translation
• 02:59, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/29/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 02:55, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/26/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 02:54, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/8/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 02:54, 7 September 2024 FuzzyBot talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/142/en (Importing a new version from external source)
• 02:54, 7 September 2024 FuzzyBot talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/141/en (Importing a new version from external source)
• 02:54, 7 September 2024 FuzzyBot talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/140/en (Importing a new version from external source)
• 02:54, 7 September 2024 FuzzyBot talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/139/en (Importing a new version from external source)
• 02:52, 7 September 2024 Felipefelixarias talk contribs removed Diffusion Models Are Real-Time Game Engines from the translation system
• 02:52, 7 September 2024 Felipefelixarias talk contribs marked Diffusion Models Are Real-Time Game Engines for translation
• 02:51, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/28/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 02:51, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/3/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 02:51, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/31/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 02:50, 7 September 2024 FuzzyBot talk contribs deleted page Translations:Diffusion Models Are Real-Time Game Engines/18/es (Part of translation page "Diffusion Models Are Real-Time Game Engines/es": Reuploading the translation from source file)
• 07:36, 6 September 2024 Felipefelixarias talk contribs created page Glossary (Created page with "AI-relevant Terminology") Tag: Visual edit
• 21:57, 2 September 2024 Felipefelixarias talk contribs created page File:2408.14837.figure.8.png (Auto-regressive evaluation of the simulation model: Sample #1. Top row: Context frames. Middle row: Ground truth frames. Bottom row: Model predictions.)
• 21:57, 2 September 2024 Felipefelixarias talk contribs uploaded File:2408.14837.figure.8.png (Auto-regressive evaluation of the simulation model: Sample #1. Top row: Context frames. Middle row: Ground truth frames. Bottom row: Model predictions.)
• 22:03, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/45/es (Created page with "Observamos que es trivial aumentar aún más la tasa de generación de imágenes de forma sustancial paralelizando la generación de varios fotogramas en hardware adicional, de manera similar a la técnica clásica SLI Alternate Frame Rendering (AFR) de NVidia. De manera similar al AFR, la tasa de simulación real no aumentaría y el input lag no se reduciría.")
• 22:02, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/44/es (Created page with "Dado que observamos degradación cuando utilizamos un único paso de muestreo, también experimentamos con la destilación de modelos de forma similar a (Yin et al., [https://arxiv.org/html/2408.14837v1#bib.bib39 2024]; Wang et al., [https://arxiv.org/html/2408.14837v1#bib.bib36 2023]) en el escenario de un solo paso. La destilación ayuda sustancialmente en este caso (permitiéndonos alcanzar los 50 FPS como en el caso anterior), pero aún conlleva cierto coste para la...")
• 22:01, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/43/es (Created page with "Utilizar sólo 4 pasos de eliminación de ruido supone un coste total de U-Net de 40 ms (y un coste total de inferencia de 50 ms, incluido el codificador automático) o 20 fotogramas por segundo. Nuestra hipótesis es que el impacto insignificante en la calidad con pocos pasos en nuestro caso se debe a una combinación de: (1) un espacio de imágenes restringido, y (2) un fuerte condicionamiento por los fotogramas anteriores.")
• 22:00, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/42/es (Created page with "Durante la inferencia, tenemos que ejecutar tanto el denoiser U-Net (para un número de pasos) como el auto-codificador. En nuestra configuración de hardware (una TPU-v5), un solo paso del denoiser y una evaluación del auto-codificador tardan 10 ms. Si ejecutáramos nuestro modelo con un único paso de eliminación de ruido, la latencia total mínima posible en nuestra configuración sería de 20 ms por fotograma, o 50 fotogramas por segundo. Por lo general, los modelo...")
• 21:58, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/41/es (Created page with "==== 3.3.2 Pasos de Muestreo del Denoizador ====")
• 21:56, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/40/es (Created page with "También experimentamos generando 4 muestras en paralelo y combinando los resultados, con la esperanza de evitar que se acepten predicciones extremas poco frecuentes y reducir la acumulación de errores. Experimentamos tanto con el promedio de las muestras como con la elección de la muestra más próxima a la mediana. El promedio fue ligeramente peor que el de un solo fotograma, y la elección de la más cercana a la mediana fue solo levemente mejor. Dado que ambos aume...")
• 21:53, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/39/es (Created page with "Utilizamos el muestreo DDIM (Song et al., [https://arxiv.org/html/2408.14837v1#bib.bib34 2022]). Empleamos Classifier-Free Guidance (Ho & Salimans, [https://arxiv.org/html/2408.14837v1#bib.bib12 2022]) solo para la condición de observaciones pasadas <math>o_{< n}</math>. No encontramos que la guía para la condición de acciones pasadas <math>a_{< n}</math> mejore la calidad. La ponderación que utilizamos es relativamente pequeña (1,5), ya que ponderaciones mayores cr...")
• 21:52, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/38/es (Created page with "==== 3.3.1 Configuración ====")
• 21:51, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/37/es (Created page with " === 3.3 Inferencia ===")
• 21:51, 1 September 2024 Felipefelixarias talk contribs created page Translations:Diffusion Models Are Real-Time Game Engines/36/es (Created page with "El auto codificador preentrenado de Stable Diffusion v1.4, que comprime parches de 8x8 píxeles en 4 canales latentes, produce artefactos significativos al predecir los fotogramas del juego, lo que afecta a los pequeños detalles y, en particular, a la barra inferior HUD ("heads up display"). Para aprovechar el conocimiento preentrenado y mejorar al mismo tiempo la calidad de la imagen, entrenamos solo el decodificador del auto codificador latente utilizando una pérdida...")