Welcome to the AI Daily section! Here, you'll find your daily guide to exploring the world of artificial intelligence. Every day, we bring you the hottest topics in the AI field, focusing on developers, helping you understand technological trends and innovative AI product applications.

Fresh AI Products Click to Learn More: https://top.aibase.com/

1. Keling AI Unleashes New Features: Web-based Interface, Enhanced Frame and Camera Controls

Keling AI has recently released significant new features, including the launch of a web-based interface, improved image quality, and the addition of start/end frame and camera control functions. The duration for text-to-video generation has been extended to 10 seconds. Updates cover upgrades to the foundational model, start/end frame control, camera control, increased text-to-video duration, enhanced image-to-video capabilities, and the web-based interface. Keling AI can generate high-definition videos, supporting 1080p resolution, and offers personalized video control options, allowing users to create more diverse content. The web service is available for free, with future support for voice-lip synchronization and other features.

1.jpg

AiBase Summary:

🚀 Upgraded foundational model, supporting 1080p resolution, with film-quality image quality.

💡 Start/end frame control allows users to customize the beginning and ending of videos, enhancing personalized creative experiences.

🎥 Increased camera control functions, including pan and tilt, making videos more vivid and interesting.

Details link: https://top.aibase.com/tool/keling-ai

2. Kuaishou Open-Sources Image Generation Model Kolors

Kuaishou has released a significant announcement today, open-sourcing its image generation model "Kolors," which has been trained on billions of text-image pairs. It is equipped with a General Language Model (GLM) as a text encoder, supports both Chinese and English prompts, and has the ability to handle long texts and extensive data training. Kolors has been specially optimized for Chinese cultural elements, supports Chinese text generation, and demonstrates strong technical support and cultural heritage.

QQ截图20240708111705.jpg

AiBase Summary:

🌟 Supports both Chinese and English: Uses General Language Model (GLM) as a text encoder, supports both Chinese and English prompts, and can handle up to 256 tokens of context.

🚀 Long text processing capability: Supports up to 256 tokens of context length, allowing creators to meticulously describe their thoughts, whether complex scenes or rich stories.

💡 Extensive data training: Trained on billions of text-image pairs, the model has a vast knowledge base, capable of generating diverse and accurate images.

Kolors entry: https://top.aibase.com/tool/kuaishouketudamoxingkolors

Detailed content introduction: https://www.aibase.com/news/10085

3. Kuaishou Launches AIGC Micro-Short Film "The Mirror of the Mountains and Seas: Shattering Waves"

Kuaishou has released the first AIGC original fantasy micro-short film in China, "The Mirror of the Mountains and Seas: Shattering Waves," combining traditional charm with modern technology to deliver a stunning viewing experience. With the support of large model technology, it presents upgraded visual effects, promoting the development of the micro-short film industry and leading the new trend of "AIGC + Micro-Short Films."

AiBase Summary:

🎬 Kuaishou launches the first AIGC original fantasy micro-short film in China, "The Mirror of the Mountains and Seas: Shattering Waves"

💡 Inspired by "The Classic of Mountains and Seas," the micro-short film recreates the ancient mythological world with cyber style, featuring divine creatures and exotic plants.

🌟 Kuaishou launches the "Starlight Short Film × Keling Large Model" creator incubation plan, supporting the creation of AIGC micro-short films.

Detailed content: https://www.aibase.com/news/10075

4. Moonshot AI Launches Kimi Browser Extension with Point-and-Ask Pen and Summarizer Features

Moonshot AI's Kimi browser extension offers users two major features: the Point-and-Ask Pen and the Summarizer, optimizing the user experience on web pages and applications. The extension supports global floating window and sidebar modes, making it convenient for users to engage in continuous dialogue and search while writing. Additionally, Kimi has been optimized for experience, including support for opening PDF files, search citation tracing, content copying, and more. Dual-end synchronized updates have also added a calculator and question recommendation features.

image.png

AiBase Summary:

🖊️ The Point-and-Ask Pen feature allows users to get instant explanations and answers by selecting text.

📝 The Summarizer, located in the lower right corner of the web page, helps users quickly summarize the full content.

🔗 Supports shortcut keys to summon Kimi, providing convenient operations and feature recommendations.

Details link: https://kimi.moonshot.cn/extension/download

5. Alibaba DAMO Academy's "Shimmer" Revolutionizes AI Video Creation Workflow

The field of AI video creation has undergone a revolutionary change. Alibaba DAMO Academy's "Shimmer" platform made a stunning debut at WAIC, providing creators with a one-stop AI video creation solution that greatly improves creation efficiency and reshapes the video creation workflow.

AiBase Summary:

✨ Launched a one-stop AI video creation platform "Shimmer," integrating script creation, storyboard design, and video material editing, simple and efficient.

🔥 AI technology applications enable one-click completion of lens angle adjustments, target elimination and modification, improving creation efficiency.

💡 The Shimmer platform supports script creation assistance, AI editing functions, camera control, target addition/elimination/modification, and other powerful functions.

Details link: https://top.aibase.com/tool/xunguangshipinchuangzuopingtai

6. Shusheng Puyu 2.5 - InternLM2.5-7B Model Officially Open-Sourced

On July 3, 2024, the Shanghai Artificial Intelligence Laboratory, together with SenseTime, The Chinese University of Hong Kong, and Fudan University, officially released the new generation of large language model InternLM2.5-7B. The model has significantly improved reasoning capabilities, long-text support, and autonomous planning and tool calling.

image.png

AiBase Summary:

🚀 The InternLM2.5-7B model performs exceptionally in reasoning capabilities, especially in the math evaluation set MATH, achieving a 100% performance improvement with an accuracy rate of 60%.

💬 The model supports processing up to 1M tokens of context, optimizing long document understanding and agent interaction.

🔍 Possesses the ability to search and integrate information from hundreds of web pages, effectively integrating network information through the MindSearch multi-agent framework.

Details link: https://github.com/InternLM/InternLM

7. Alibaba Tongyi Lab Open-Sources FunAudioLLM Audio Generation Large Model, Supporting Emotional Voice Dialogue, Audiobooks, and Other Scenarios

Alibaba Tongyi Lab has recently open-sourced the FunAudioLLM audio generation large model project, aiming to enhance the natural voice interaction experience between humans and large language models (LLMs). The project includes two core models: SenseVoice and CosyVoice, dedicated to voice generation and voice recognition, respectively. FunAudioLLM supports various human-machine interaction application scenarios, such as multilingual translation, emotional voice dialogue, interactive podcasts, and audiobooks.

image.png

AiBase Summary:

🔊 CosyVoice focuses on natural voice generation, supports multiple languages, voice tones, and emotional control, and performs exceptionally well.

🔍 SenseVoice is dedicated to high-precision multilingual voice recognition and emotional identification, supporting over 50 languages.

🔗 The FunAudioLLM project combines SenseVoice, LLMs, and CosyVoice, supporting seamless voice-to-voice translation and emotional voice chat applications.

Details link: https://github.com/FunAudioLLM

8. Tsinghua University Open-Sources CodeGeeX4-ALL-9B: Multilingual Code Generation Model Surpassing Major Competitors

The Knowledge Engineering Group and Data Mining Team at Tsinghua University have introduced CodeGeeX4-ALL-9B, marking a milestone in the development of code generation models. With unparalleled performance, comprehensive features, and user-friendly integration, it will drive the efficiency and innovation of software development.

image.png

AiBase Summary:

🚀 CodeGeeX4-ALL-9B is the latest innovation in the CodeGeeX series, representing the pinnacle of multilingual code generation, setting new standards for performance and efficiency.

💡 The model has 940 million parameters, making it one of the most powerful in its category, with outstanding performance and repository-level code Q&A capabilities, enhancing developers' interaction with code repositories.

🔗 CodeGeeX4-ALL-9B performs exceptionally well in performance benchmarks, surpassing larger models, establishing itself as a leading model.

Details link: https://huggingface.co/THUDM/codegeex4-all-9b

9. Anti-AI Image Theft Tool Glaze Sees Dramatic Increase in Demand, Attracting Many Artists

The Glaze tool has emerged to protect artists' styles from being copied by AI image generators. With Meta planning to use user data for AI training, the demand for Glaze has surged. However, security researchers have discovered methods to bypass Glaze's protection, raising questions about its effectiveness.

image.png

AiBase Summary:

🖼️ Artists are flocking to the Glaze tool to prevent AI image theft.

🔒 Glaze demand has surged due to Meta's plan to use user data for AI training.

⚙️ Security researchers have found ways to bypass Glaze protection, casting doubt on its effectiveness.

Details link: https://top.aibase.com/tool/glaze

10. Science Fiction Becomes Reality? Open-TeleVision Supports Remote Control of Robots

This article introduces the Open-TeleVision project developed by researchers from the University of California, San Diego, and the Massachusetts Institute of Technology, which realizes the high-tech scenario of remotely controlling robots, reminiscent of the movie "Avatar." The system supports multiple devices, provides an immersive experience, and enhances the convenience and realism of operations through VR headsets.

AiBase Summary: