Categories
News

El modelo Samsung Gauss2 Multimodal AI con soporte para hasta 14 idiomas se presentó en SDC24

[ad_1]

Samsung El jueves presentó su modelo de inteligencia artificial (IA) gaussiana de segunda generación. Se dice que el nuevo modelo de IA multimodal, denominado Gauss2, presenta un rendimiento y una eficiencia mejorados y casos de uso de integración de aplicaciones mejorados. El nuevo modelo de lenguaje grande (LLM) se presentó durante el discurso de apertura de la Samsung Developer Conference Korea 2024 (SDC24 Korea) celebrada en línea. El gigante tecnológico destacó que está utilizando las capacidades del modelo de IA con fines de investigación y desarrollo de software.

Detección del modelo Samsung Gauss2 AI

en un Trabajo de redacciónel gigante tecnológico surcoreano ha detallado la segunda generación de su modelo Gauss AI que se lanzó el año pasado. La nueva versión del modelo base ahora viene con varias actualizaciones. La compañía afirmó que se ha mejorado la capacidad multimodal, que implica el manejo de conjuntos de datos en diferentes modalidades.

Además, el modelo de IA ahora también soporta entre 9 y 14 lenguajes, además de varios lenguajes de programación. También se dice que mejora el rendimiento en lenguaje, código e imágenes.

Samsung Gauss2 está disponible en tres modelos distintos, según el tamaño de sus parámetros: Compact, Balanced y Supreme. Compact es un modelo compacto diseñado para brindar eficiencia y operación en un entorno computacional limitado. El modelo equilibrado está optimizado tanto para el rendimiento como para la eficiencia, y el modelo superior puede manejar tareas informáticas de alto nivel aprovechando la tecnología Mixture of Experts (MoE).

Samsung afirma que los modelos Balanced y Super pueden superar a los “modelos líderes de IA de código abierto” en tareas basadas en idiomas en inglés y coreano, así como en tareas relacionadas con la programación. La compañía también afirma que los formularios brindan tiempos de espera más bajos, velocidades de procesamiento más rápidas y un mejor manejo de tareas en comparación con los formularios de código abierto.

Samsung Gauss2 se utiliza actualmente en la división Device eXperience (DX) de la empresa y en institutos de investigación externos. El caso de uso más común de un modelo de IA es un asistente de codificación interno llamado code.i que ayuda en el desarrollo de software. El gigante tecnológico afirmó que el 60 por ciento de su división DX utiliza esta herramienta. Los empleados del centro de llamadas de la empresa también utilizan esta tecnología para categorizar y resumir las llamadas de los clientes.

Si bien Gauss2 se utiliza actualmente internamente, la compañía también planea enviarlo con sus productos. Samsung también cree que el modelo de IA puede mejorar la capacidad de personalización de las funciones de IA existentes.

Para lo último Noticias de tecnología y ReseñasSiga Gadgets 360 en incógnita, Facebook, WhatsApp, Temas y noticias de google. Para ver los últimos vídeos sobre gadgets y tecnología, suscríbete a nuestro canal. canal de youtube. Si quieres saber todo sobre los top influencers, sigue nuestra web ¿Quién es ese 360? en Instagram y YouTube.


Cinco acusados ​​en EE.UU. por piratear criptomonedas utilizando el método 'Scattered Spider': detalles



[ad_2]

Source Article Link

Categories
News

Se informa que Amazon está trabajando en un Chatbot de IA multimodal para contrarrestar el ChatGPT de OpenAI

[ad_1]

Amazonas Según se informa, la compañía está desarrollando un chatbot impulsado por inteligencia artificial (IA) que puede competir con ChatGPT de OpenAI. Se dice que el nombre en código del proyecto interno de la empresa es Metis. Según el informe, el chatbot en desarrollo podrá realizar todas las tareas comunes de la IA generativa, como generar contenido de texto, responder consultas y más. Según se informa, también admitirá la creación de imágenes y tendrá acceso a Internet. El informe también afirma que la nueva plataforma de inteligencia artificial podría lanzarse en septiembre de 2024, aproximadamente cuando Amazon organiza su evento anual de hardware y servicios.

Amazon podría lanzar pronto su propio chatbot impulsado por inteligencia artificial

Según Business Insider un informeel gigante del comercio electrónico pretende competir directamente con él ChatGPT Con su propio modelo interno de IA. Citando a personas familiarizadas con el proyecto, la publicación afirma que el proyecto lleva el nombre en código Metis, en honor a la diosa griega de la sabiduría, la prudencia y el pensamiento profundo. El chatbot está diseñado para ser accesible a través de un navegador web, similar a los chatbots más comunes con tecnología de inteligencia artificial.

Según un documento interno obtenido por el periódico, afirmó que el chatbot Metis funcionaría con el modelo de inteligencia artificial interno de la compañía llamado Olympus. Se dice que es más avanzado que el actual Titan Large Language Model (LLM) que impulsa algunos productos de Amazon.

En términos de su funcionalidad, se dice que el chatbot habilitado para IA es capaz de realizar tareas basadas en texto, como mantener conversaciones, responder consultas y crear contenido. Además, también puede crear imágenes, según el informe. Esto indica que Metis utilizará un modelo de IA multimodal. En particular, ChatGPT no puede crear imágenes, pero los usuarios pueden obtener una suscripción a Dall-E y utilizarla a través de ChatGPT.

Metis utiliza un marco de IA generativa de recuperación aumentada

Según se informa, el chatbot Metis se ejecutará en un marco basado en inteligencia artificial (RAG). El mecanismo mejora la calidad de las respuestas mediante una combinación de generación de texto y recuperación de información a partir de grandes conjuntos de datos. Según se informa, el chatbot podrá acceder y recuperar información de Internet.

Por ejemplo, podrá mostrar actualizaciones del precio de las acciones casi en tiempo real, algo con lo que muchos chatbots de IA tienen problemas. Sin embargo, tanto ChatGPT como mellizo Puedes hacerlo.

El informe afirma que el proyecto está siendo desarrollado por la división de Inteligencia General Artificial (AGI) de la compañía, dirigida por el vicepresidente senior y científico jefe Rohit Prasad. Mientras tanto, también se dice que el director ejecutivo de Amazon, Andy Jassy, ​​está muy involucrado en el proyecto.

El informe destaca que a algunos empleados les preocupa que Amazon pueda llegar demasiado tarde en la carrera de los chatbots impulsados ​​por IA, que ya se está volviendo un poco concurrida.

[ad_2]

Source Article Link

Categories
News

How to Use Apple’s Ferret 7B Multi-modal Large Language Model

Apple's Ferret 7B

Apple’s recent unveiling of the Ferret 7B model has caught the attention of tech enthusiasts and professionals alike. Developed by Jarvis Labs, this multi-modal Large Language Model (LLM) is breaking new ground by combining image processing with text-based instructions to produce comprehensive responses. If you’re curious about how this model works and how you can leverage it for your projects, you’re in the right place. Let’s dive into the details of Ferret 7B and explore its capabilities, setup process, and practical applications.

Understanding Ferret 7B’s Capabilities

At its core, Ferret 7B is designed to understand and interact with both visual and textual information. This dual capability allows it to process images through points, bounding boxes, or sketches, and respond to text instructions with an understanding of the content and context of the images. Imagine asking detailed questions about an image, and receiving precise answers as if you were discussing it with a human expert. This level of interaction is now possible with Ferret 7B, thanks to its innovative integration of technologies.

The model is built on a foundation that includes components from renowned models like Vicuna and OpenCLIP, enriched by a novel instruction-following mechanism. This architecture allows Ferret to excel in tasks requiring a deep understanding of both visual elements and textual descriptions. The research paper accompanying Ferret’s release introduces key concepts such as “referring” and “grounding,” pivotal for the model’s understanding of multi-modal inputs.

Getting Started with Ferret 7B

If you’re eager to experiment with Ferret 7B, Vishnu Subramaniam from Jarvis Labs offers a comprehensive guide to get you started. The setup involves a few essential steps:

  1. Environment Setup: Begin by creating a Python environment tailored for Ferret. This ensures that all dependencies and libraries are correctly aligned with the model’s requirements.
  2. Cloning Repositories: Next, clone the necessary repositories. This step is crucial for accessing the model’s architecture and scripts essential for its operation.
  3. Downloading Model Weights: Model weights, released shortly after Ferret’s announcement, are vital for harnessing the full potential of the model. Download and integrate these weights as per the instructions.
  4. Configuration Adjustments: Before diving into Ferret’s capabilities, make sure to adjust configurations according to your project’s needs. This fine-tuning is key to optimizing performance.

Vishnu’s walkthrough doesn’t stop at setup; it also includes troubleshooting tips for common issues you might encounter. This ensures a smooth experience as you explore Ferret’s capabilities.

Practical Applications of Ferret 7B

The potential applications for Ferret 7B are vast, spanning various fields from academic research to creative industries. Whether you’re analyzing images for detailed insights, generating content based on visual prompts, or developing interactive educational tools, Ferret can enhance your projects with its nuanced understanding of combined visual and textual data.

Exploring Further

As you embark on your journey with Ferret 7B, remember that the learning curve is part of the adventure. Experiment with different types of visual inputs and textual instructions to fully grasp the model’s versatility. The integration of grounding and referring mechanisms offers a unique opportunity to explore multi-modal AI in ways that were previously unimaginable.

Ferret 7B represents a significant step forward in the field of multi-modal AI. Its ability to process and respond to a blend of visual and textual information opens up new avenues for innovation and creativity. By following the guidance provided by experts like Vishnu Subramaniam, you can unlock the full potential of this model and explore a wide range of applications. With Ferret 7B, the future of multi-modal interaction is in your hands.

Source JarvisLabs AI

Filed Under: Apple, Guides





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.

Categories
News

Apple releases Ferret 7B multimodal large language model (MLLM)

Apple releases Ferret 7B multimodal large language model

Apple has recently introduced the Ferret 7B, a sophisticated large language model (LLM) that represents a significant step forward in the realm of artificial intelligence. This new technology is a testament to Apple’s commitment to advancing AI and positions the company as a formidable player in the tech industry. The Ferret 7B is engineered to integrate smoothly with both iOS and macOS, taking full advantage of Apple’s powerful silicon to ensure users enjoy a fluid experience.

The standout feature of the Ferret 7B is its multimodal capabilities, which allow it to interpret and create content that combines images and text. This breakthrough goes beyond what traditional text-based AI models can do. The Ferret 7B’s capabilities are showcased in systems like the Google 5.2 coding model and MixL 8X 7B, which are built on Apple’s MLX platform and utilize its unique tools.

  • Ferret Model – Hybrid Region Representation + Spatial-aware Visual Sampler enable fine-grained and open-vocabulary referring and grounding in MLLM.
  • GRIT Dataset (~1.1M) – A Large-scale, Hierarchical, Robust ground-and-refer instruction tuning dataset.
  • Ferret-Bench – A multimodal evaluation benchmark that jointly requires Referring/Grounding, Semantics, Knowledge, and Reasoning.

There’s buzz around the upcoming iOS 18, which is expected to incorporate AI more comprehensively, potentially transforming how users interact with Apple devices. The collaboration between AI advancements and Apple’s silicon architecture is likely to result in a more cohesive and powerful ecosystem for both iOS and macOS users.

Apple Ferret 7B MLLM

Here are some other articles you may find of interest on the subject of  multimodal large language models :

For those interested in the technical performance of the Ferret 7B, Apple has developed the Ferret Bench, a benchmarking tool specifically for this model. This tool will help developers and researchers evaluate the model’s efficiency and flexibility in various situations.

Apple’s approach to AI is centered on creating practical applications that provide tangible benefits to users of its devices. The company’s dedication to this strategy is clear from its decision to make the Ferret 7B open-source, offering the code and checkpoints for research purposes. This move encourages further innovation and collaboration within the AI community.

Training complex models like the Ferret 7B requires considerable resources, and Apple has invested in this by using NVIDIA A100 GPUs. This reflects the company’s deep investment in AI research and development.

Apple multimodal large language model (MLLM)

It’s important to note the differences between the 7B and the larger 13B versions of the model. The 7B is likely tailored for iOS devices, carefully balancing performance with the constraints of mobile hardware. This strategic decision is in line with Apple’s focus on the user experience, ensuring that AI improvements directly benefit the user.

# 7B
python3 -m ferret.model.apply_delta \
    --base ./model/vicuna-7b-v1-3 \
    --target ./model/ferret-7b-v1-3 \
    --delta path/to/ferret-7b-delta
# 13B
python3 -m ferret.model.apply_delta \
    --base ./model/vicuna-13b-v1-3 \
    --target ./model/ferret-13b-v1-3 \
    --delta path/to/ferret-13b-delta

Usage and License Notices: The data, and code is intended and licensed for research use only. They are also restricted to uses that follow the license agreement of LLaMA, Vicuna and GPT-4. The dataset is CC BY NC 4.0 (allowing only non-commercial use) and models trained using the dataset should not be used outside of research purposes.

With the release of the Ferret 7B LLM, Apple has made a bold move in the AI space. The launch showcases the company’s technical prowess and its commitment to creating powerful, user-friendly AI. This development is set to enhance device functionality and enrich user interactions. As Apple continues to invest in AI, we can expect to see more innovations that will significantly impact how we interact with technology.

Filed Under: Apple, Technology News, Top News





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.

Categories
News

What is Multimodal Artificial Intelligence (AI)?

What is Multimodal Artificial Intelligence

If you have engaged with the latest ChatGPT-4 AI model or perhaps the latest Google search engine, you will of already used multimodal artificial intelligence.  However just a few years ago such easy access to multimodal AI was only a dream. In this guide will explain more about what this new technology is and how it is truly revolutionizing our world on a daily basis.

AI technologies that specialized in one form of data analysis, perhaps text-based chatbots or image recognition software is Single-Modality Learning . But now AI can combine different forms of data such as images, text, photographs, graphs, reports and more for a richer, more insightful analysis. These AI applications are multimodal AI in the already making their mark across many different areas of our lives.

For example in autonomous vehicles, multimodal AI helps in collecting data from cameras, LiDAR, and radar, combined it all for better situational awareness. In healthcare, AI can combine textual medical records with imaging data for more accurate diagnoses. In conversational agents such as ChatGPT-4, multimodal AI can interpret both the text and the tone of voice to provide more nuanced responses.

Multimodal Artificial Intelligence

  • Single-Modality Learning: Handles only one type of input.
  • Multimodal Learning: Can process multiple types of inputs like text, audio, and images.

Older machine learning models were unimodal, meaning they capable of only handling one type of input. For instance, text-based models like the Transformer architecture focus exclusively on textual data. Similarly, Convolutional Neural Networks (CNNs) are geared for visual data like images.

One area of multimodal AI technology you can try is  within OpenAI’s ChatGPT. Now capable of interpreting inputs from text, files and imagery. Another is  Google’s multimodal search engine. In essence, multimodal artificial intelligence (AI) systems are engineered to comprehend, interpret, and integrate multiple forms of data, be it text, images, audio, or even video. This versatile approach enhances the AI’s contextual understanding, thus making its outputs much more accurate.

What is Multimodal Artificial Intelligence?

The limitation here is evident—these models cannot naturally handle a mix of inputs, such as both audio and text. For example, you might have a conversational model that understands the text but fails to account for the tone or intonation captured in the audio, leading to misinterpretation.

In contrast, multimodal learning aims to build models that can process various types of inputs and possibly create a unified representation. This unification is beneficial because learning from one modality can enhance the model’s performance on another. Imagine a language model trained on both books and accompanying audiobooks; it might better understand the sentiment or context by aligning the text with the spoken words’ tone.

Another remarkable feature is the ability to generate common responses irrespective of the input type. In practical terms, this means the AI system could understand a query whether it’s typed in as text, spoken aloud, or even conveyed through a sequence of images. This has profound implications for accessibility, user experience, and the development of more robust systems. Let’s delve deeper into the facets of multimodal learning in machine learning models, a subfield that is garnering significant attention for its versatile applications and improved performance metrics. Key facets of multimodal AI include :

  • Data Types: Includes text, images, audio, video, and more.
  • Specialized Networks: Utilizes specialized neural networks like Convolutional Neural Networks (CNNs) for images and Recurrent Neural Networks (RNNs) or Transformers for text.
  • Data Fusion: The integration of different data types through fusion techniques like concatenation, attention mechanisms, etc.

Simply put, integrating multiple data types allows for a more nuanced interpretation of complex situations. Imagine a healthcare scenario where a textual medical report might be ambiguous. Add to this X-ray images, and the AI system can arrive at a more definitive diagnosis. So, to enhance your experience with AI applications, multimodal systems offer a holistic picture by amalgamating disparate chunks of data.

In a multimodal architecture, different modules or neural networks are generally specialized for processing specific kinds of data. For example, a Convolutional Neural Network (CNN) might be used for image processing, while a Recurrent Neural Network (RNN) or Transformer might be employed for text. These specialized networks can then be combined through various fusion techniques, like concatenation, attention mechanisms, or more complex operations, to generate a unified representation.

In case you’re curious how these systems function, they often employ a blend of specialized networks designed for each data type. For instance, a CNN processes image data to extract relevant features, while a Transformer may process text data to comprehend its semantic meaning. These isolated features are then fused to create a holistic representation that captures the essence of the multifaceted input.

Fusion Techniques:

  • Concatenation: Simply stringing together features from different modalities.
  • Attention Mechanisms: Weighing the importance of features across modalities.
  • Hybrid Architectures: More complex operations that dynamically integrate features during processing.

Simplified Analogies

he Orchestra Analogy: Think of multimodal AI as an orchestra. In a traditional, single-modal AI model, it’s as if you’re listening to just one instrument—say, a violin. That’s beautiful, but limited. With a multimodal approach, it’s like having an entire orchestra—violins, flutes, drums, and so on—playing in harmony. Each instrument (or data type) brings its unique sound (or insight), and when combined, they create a richer, fuller musical experience (or analysis).

The Swiss Army Knife Analogy: A traditional, single-modal AI model is like a knife with just one tool—a blade for cutting. Multimodal AI is like a Swiss Army knife, equipped with various tools for different tasks—scissors, screwdrivers, tweezers, etc. Just as you can tackle a wider range of problems with a Swiss Army knife, multimodal AI can handle more complex queries by utilizing multiple types of data.

Real-World Applications

To give you an idea of its vast potential, let’s delve into a few applications:

  • Autonomous Vehicles: Sensor fusion leverages data from cameras, LiDAR, and radar to provide an exhaustive situational awareness.
  • Healthcare: Textual medical records can be complemented by imaging data for a more thorough diagnosis.
  • E-commerce: Recommender systems can incorporate user text reviews and product images for enhanced recommendations.

Google, with its multimodal capabilities in search algorithms, leverages both text and images to give you a more complete set of search results. Similarly, Tesla excels in implementing multimodal sensor fusion in its self-driving cars, capturing a 360-degree view of the car’s surroundings.

The importance of multimodal learning primarily lies in its ability to generate common representations across diverse inputs. For instance, in a healthcare application, a multimodal model might align a patient’s verbal description of symptoms with medical imaging data to provide a more accurate diagnosis. These aligned representations enable the model to understand the subject matter more holistically, leveraging complementary information from different modalities for a more rounded view.

Multimodal AI has immense promise but is also subject to ongoing research to solve challenges like data alignment and modality imbalance. However, with advancements in deep learning and data science, this field is poised for significant growth.
So there you have it, a sweeping yet accessible view of what multimodal AI entails. With the ability to integrate a medley of data types, this technology promises a future where AI is not just smart but also insightful and contextually aware.

Multimodal Artificial Intelligence (AI) summary:

  • Single-Modality Learning: Handles only one type of input.
  • Multimodal Learning: Can process multiple types of inputs like text, audio, and images.
  • Cross-Modality Benefits: Learning from one modality can enhance performance in another.
  • Common Responses: Capable of generating unified outputs irrespective of input type.
  • Common Representations: Central to the multimodal approach, allowing for a holistic understanding of diverse data types.

Multimodal learning offers an evolved, nuanced approach to machine learning. By fostering common representations across a spectrum of inputs, these models are pushing the boundaries of what AI can perceive, interpret, and act upon.

Filed Under: Guides, Top News





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.