Home Technology peripherals AI Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

Dec 04, 2023 pm 12:39 PM
industry codi-2

Researchers pointed out that CoDi-2 marks a major breakthrough in the field of developing comprehensive multi-modal basic models

In May of this year, North Carolina The University of Chapel Hill and Microsoft proposed a Composable Diffusion (CoDi) model, making it possible for one model to unify multiple modalities. CoDi not only supports single-modal to single-modal generation, but can also receive multiple conditional inputs and multi-modal joint generation.

Recently, many researchers from UC Berkeley, Microsoft Azure AI, Zoom, and the University of North Carolina at Chapel Hill have upgraded the CoDi system to the CoDi-2 version

Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

  • Paper address: https://arxiv.org/pdf/2311.18775.pdf

  • Project address: https://codi-2. github.io/Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

Rewrite the content without changing the original meaning. It needs to be rewritten into Chinese language and the original sentence does not need to appear.

According to Zineng Tang’s paper, CoDi-2 follows complex multi-modal interleaved contextual instructions to generate any modality (text, visual and audio) with zero- or few-shot interactions )

Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

##This link is the source of the picture: https://twitter.com/ZinengTang/status/1730658941414371820

It can be said that as a multi-functional, interactive multi-modal large language model (MLLM), CoDi-2 can perform contextual learning, reasoning, chatting, and editing with any-to-any input-output modal paradigm. Wait for the task. By aligning modalities and languages ​​during encoding and generation, CoDi-2 enables LLM not only to understand complex modal interleaving instructions and contextual examples, but also to autoregressively generate reasonable and coherent multi-modal output within a continuous feature space. .

To train CoDi-2, the researchers built a large-scale generative dataset containing contextual multimodal instructions across text, visuals, and audio. CoDi-2 demonstrates a range of zero-shot capabilities for multi-modal generation, such as contextual learning, reasoning, and any-to-any modal generation combinations through multiple rounds of interactive dialogue. Among them, it surpasses previous domain-specific models in tasks such as topic-driven image generation, visual transformation, and audio editing.

Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

Multiple rounds of dialogue between humans and CoDi-2 provide contextual multimodal instructions for image editing.

What needs to be rewritten is: Model Architecture

CoDi-2 is designed to handle text, images and audio in context and other multi-modal inputs, using specific instructions to promote contextual learning and generate corresponding text, image, and audio output. What needs to be rewritten for CoDi-2 is: The model architecture diagram is as follows.

Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

Using multi-modal large language model as the basic engine

This any-to-any basic model can digest interleaved Modal input, understanding and reasoning about complex instructions (such as multi-turn conversations, contextual examples), and interacting with multi-modal diffusers require a powerful basic engine. The researchers proposed MLLM as this engine, which was built to provide multimodal perception for text-only LLMs.

Using aligned multi-modal encoder mapping, researchers can seamlessly make LLM aware of modally interleaved input sequences. Specifically, when processing multi-modal input sequences, they first use multi-modal encoders to map multi-modal data to feature sequences, and then special tokens are added before and after the feature sequence, such as "〈audio〉 [audio feature sequence" ] 〈/audio〉”.

The basis of multi-modal generation is MLLM

Researchers proposed to integrate the diffusion model (DM) into MLLM to generate multi-modal output. During this process, detailed multimodal interleaving instructions and prompts were followed. The training objectives of the diffusion model are as follows:

Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

They then proposed training an MLLM to generate conditional features c = C_y (y), which are fed into a diffusion model to synthesize the target output x. In this way, the generative loss of the diffusion model is used to train the MLLM.

Task Type

The model demonstrates its power in the following example task types, which provide a unique way to prompt the model to generate or transform context The multi-modal content in , including text, images, audio, video and their combinations

The rewritten content is: 1. Zero-sample reasoning. Zero-shot inference tasks require the model to reason and generate new content without any previous examples

2. One-time/small number of sample prompts. One or a few sample prompts provide the model with one or several examples to learn from before performing similar tasks. This approach is evident in tasks where the model applies learned concepts from one image to another, or creates a new artwork by understanding the style described in the provided example.

(1) Example Learning Explicitly show the model examples of the desired output before asking the model to apply this learning to new instances.
(2) Concept learning involves the model learning from the shared concepts/properties of these given examples, such as art style or pattern, and then creating new content that exhibits similar concepts/properties.
(3) Topic-driven learning focuses on generating new content based on a set of provided images.

Experiments and results

Model settings

The implementation of the model in this article is based on Llama2, especially Llama-2-7b- chat-hf. The researchers used ImageBind, which has aligned image, video, audio, text, depth, thermal and IMU mode encoders. We use ImageBind to encode image and audio features and project them through a multilayer perceptron (MLP) to the input dimensions of an LLM (Llama-2-7b-chat-hf). MLP consists of linear mapping, activation, normalization and another linear mapping. When LLMs generate image or audio features, they project them back into the ImageBind feature dimension through another MLP. The image diffusion model in this article is based on StableDiffusion2.1 (stabilityai/stable-diffusion-2-1-unclip), AudioLDM2 and zeroscope v2.

In order to obtain higher fidelity original input images or audio, researchers input them into the diffusion model and generate features by concatenating diffusion noise. This method is very effective. It can retain the perceptual characteristics of the input content to the greatest extent, and can add new content or change the style and other instructions. Edit

The content that needs to be rewritten is: Image generation evaluation

The following figure shows the evaluation results of topic-driven image generation on Dreambench and the FID score on MSCOCO. The method in this paper achieves highly competitive zero-shot performance, demonstrating its generalization ability to unknown new tasks.

Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

Audio Generation Evaluation

Table 5 shows the evaluation results for audio processing tasks, i.e., adding, deleting, and replacing audio tracks Elements. From the table, it is obvious that our method shows excellent performance compared with previous methods. Notably, it achieved the lowest scores on all metrics - Log Spectral Distance (LSD), Kullback-Leibler (KL) Divergence and Fréchet Distance (FD) across all three editing tasks

Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2

Read the original article for more technical details.

The above is the detailed content of Any text, visual, audio mixed generation, multi-modal with a powerful basic engine CoDi-2. For more information, please follow other related articles on the PHP Chinese website!

Statement of this Website
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn

Hot AI Tools

Undresser.AI Undress

Undresser.AI Undress

AI-powered app for creating realistic nude photos

AI Clothes Remover

AI Clothes Remover

Online AI tool for removing clothes from photos.

Undress AI Tool

Undress AI Tool

Undress images for free

Clothoff.io

Clothoff.io

AI clothes remover

Video Face Swap

Video Face Swap

Swap faces in any video effortlessly with our completely free AI face swap tool!

Hot Tools

Notepad++7.3.1

Notepad++7.3.1

Easy-to-use and free code editor

SublimeText3 Chinese version

SublimeText3 Chinese version

Chinese version, very easy to use

Zend Studio 13.0.1

Zend Studio 13.0.1

Powerful PHP integrated development environment

Dreamweaver CS6

Dreamweaver CS6

Visual web development tools

SublimeText3 Mac version

SublimeText3 Mac version

God-level code editing software (SublimeText3)

Hot Topics

Java Tutorial
1658
14
PHP Tutorial
1257
29
C# Tutorial
1231
24
DeepMind robot plays table tennis, and its forehand and backhand slip into the air, completely defeating human beginners DeepMind robot plays table tennis, and its forehand and backhand slip into the air, completely defeating human beginners Aug 09, 2024 pm 04:01 PM

But maybe he can’t defeat the old man in the park? The Paris Olympic Games are in full swing, and table tennis has attracted much attention. At the same time, robots have also made new breakthroughs in playing table tennis. Just now, DeepMind proposed the first learning robot agent that can reach the level of human amateur players in competitive table tennis. Paper address: https://arxiv.org/pdf/2408.03906 How good is the DeepMind robot at playing table tennis? Probably on par with human amateur players: both forehand and backhand: the opponent uses a variety of playing styles, and the robot can also withstand: receiving serves with different spins: However, the intensity of the game does not seem to be as intense as the old man in the park. For robots, table tennis

The first mechanical claw! Yuanluobao appeared at the 2024 World Robot Conference and released the first chess robot that can enter the home The first mechanical claw! Yuanluobao appeared at the 2024 World Robot Conference and released the first chess robot that can enter the home Aug 21, 2024 pm 07:33 PM

On August 21, the 2024 World Robot Conference was grandly held in Beijing. SenseTime's home robot brand "Yuanluobot SenseRobot" has unveiled its entire family of products, and recently released the Yuanluobot AI chess-playing robot - Chess Professional Edition (hereinafter referred to as "Yuanluobot SenseRobot"), becoming the world's first A chess robot for the home. As the third chess-playing robot product of Yuanluobo, the new Guoxiang robot has undergone a large number of special technical upgrades and innovations in AI and engineering machinery. For the first time, it has realized the ability to pick up three-dimensional chess pieces through mechanical claws on a home robot, and perform human-machine Functions such as chess playing, everyone playing chess, notation review, etc.

Claude has become lazy too! Netizen: Learn to give yourself a holiday Claude has become lazy too! Netizen: Learn to give yourself a holiday Sep 02, 2024 pm 01:56 PM

The start of school is about to begin, and it’s not just the students who are about to start the new semester who should take care of themselves, but also the large AI models. Some time ago, Reddit was filled with netizens complaining that Claude was getting lazy. "Its level has dropped a lot, it often pauses, and even the output becomes very short. In the first week of release, it could translate a full 4-page document at once, but now it can't even output half a page!" https:// www.reddit.com/r/ClaudeAI/comments/1by8rw8/something_just_feels_wrong_with_claude_in_the/ in a post titled "Totally disappointed with Claude", full of

At the World Robot Conference, this domestic robot carrying 'the hope of future elderly care' was surrounded At the World Robot Conference, this domestic robot carrying 'the hope of future elderly care' was surrounded Aug 22, 2024 pm 10:35 PM

At the World Robot Conference being held in Beijing, the display of humanoid robots has become the absolute focus of the scene. At the Stardust Intelligent booth, the AI ​​robot assistant S1 performed three major performances of dulcimer, martial arts, and calligraphy in one exhibition area, capable of both literary and martial arts. , attracted a large number of professional audiences and media. The elegant playing on the elastic strings allows the S1 to demonstrate fine operation and absolute control with speed, strength and precision. CCTV News conducted a special report on the imitation learning and intelligent control behind "Calligraphy". Company founder Lai Jie explained that behind the silky movements, the hardware side pursues the best force control and the most human-like body indicators (speed, load) etc.), but on the AI ​​side, the real movement data of people is collected, allowing the robot to become stronger when it encounters a strong situation and learn to evolve quickly. And agile

ACL 2024 Awards Announced: One of the Best Papers on Oracle Deciphering by HuaTech, GloVe Time Test Award ACL 2024 Awards Announced: One of the Best Papers on Oracle Deciphering by HuaTech, GloVe Time Test Award Aug 15, 2024 pm 04:37 PM

At this ACL conference, contributors have gained a lot. The six-day ACL2024 is being held in Bangkok, Thailand. ACL is the top international conference in the field of computational linguistics and natural language processing. It is organized by the International Association for Computational Linguistics and is held annually. ACL has always ranked first in academic influence in the field of NLP, and it is also a CCF-A recommended conference. This year's ACL conference is the 62nd and has received more than 400 cutting-edge works in the field of NLP. Yesterday afternoon, the conference announced the best paper and other awards. This time, there are 7 Best Paper Awards (two unpublished), 1 Best Theme Paper Award, and 35 Outstanding Paper Awards. The conference also awarded 3 Resource Paper Awards (ResourceAward) and Social Impact Award (

Li Feifei's team proposed ReKep to give robots spatial intelligence and integrate GPT-4o Li Feifei's team proposed ReKep to give robots spatial intelligence and integrate GPT-4o Sep 03, 2024 pm 05:18 PM

Deep integration of vision and robot learning. When two robot hands work together smoothly to fold clothes, pour tea, and pack shoes, coupled with the 1X humanoid robot NEO that has been making headlines recently, you may have a feeling: we seem to be entering the age of robots. In fact, these silky movements are the product of advanced robotic technology + exquisite frame design + multi-modal large models. We know that useful robots often require complex and exquisite interactions with the environment, and the environment can be represented as constraints in the spatial and temporal domains. For example, if you want a robot to pour tea, the robot first needs to grasp the handle of the teapot and keep it upright without spilling the tea, then move it smoothly until the mouth of the pot is aligned with the mouth of the cup, and then tilt the teapot at a certain angle. . this

Hongmeng Smart Travel S9 and full-scenario new product launch conference, a number of blockbuster new products were released together Hongmeng Smart Travel S9 and full-scenario new product launch conference, a number of blockbuster new products were released together Aug 08, 2024 am 07:02 AM

This afternoon, Hongmeng Zhixing officially welcomed new brands and new cars. On August 6, Huawei held the Hongmeng Smart Xingxing S9 and Huawei full-scenario new product launch conference, bringing the panoramic smart flagship sedan Xiangjie S9, the new M7Pro and Huawei novaFlip, MatePad Pro 12.2 inches, the new MatePad Air, Huawei Bisheng With many new all-scenario smart products including the laser printer X1 series, FreeBuds6i, WATCHFIT3 and smart screen S5Pro, from smart travel, smart office to smart wear, Huawei continues to build a full-scenario smart ecosystem to bring consumers a smart experience of the Internet of Everything. Hongmeng Zhixing: In-depth empowerment to promote the upgrading of the smart car industry Huawei joins hands with Chinese automotive industry partners to provide

Distributed Artificial Intelligence Conference DAI 2024 Call for Papers: Agent Day, Richard Sutton, the father of reinforcement learning, will attend! Yan Shuicheng, Sergey Levine and DeepMind scientists will give keynote speeches Distributed Artificial Intelligence Conference DAI 2024 Call for Papers: Agent Day, Richard Sutton, the father of reinforcement learning, will attend! Yan Shuicheng, Sergey Levine and DeepMind scientists will give keynote speeches Aug 22, 2024 pm 08:02 PM

Conference Introduction With the rapid development of science and technology, artificial intelligence has become an important force in promoting social progress. In this era, we are fortunate to witness and participate in the innovation and application of Distributed Artificial Intelligence (DAI). Distributed artificial intelligence is an important branch of the field of artificial intelligence, which has attracted more and more attention in recent years. Agents based on large language models (LLM) have suddenly emerged. By combining the powerful language understanding and generation capabilities of large models, they have shown great potential in natural language interaction, knowledge reasoning, task planning, etc. AIAgent is taking over the big language model and has become a hot topic in the current AI circle. Au

See all articles