Welcome to the [AI Daily] column! This is your daily guide to exploring the world of artificial intelligence, where we present the hottest content in the AI field every day, focusing on developers to help you understand technical trends and innovative AI product applications.
Fresh AI Products Click to Learn:https://top.aibase.com/
Kunlun's "Tiangong 3.0" and "Tiangong SkyMusic" Music Large Models Open for Public Testing
Kunlun's "Tiangong 3.0" and "Tiangong SkyMusic" music large models are open for public testing, showcasing their technical strength and innovative capabilities in the AI field. These large models exhibit outstanding performance in semantic understanding and music generation, providing users with a more powerful AI experience.
Highlights:
⚙️ "Tiangong 3.0" has over 4000 billion parameters, making it the world's largest open-source MoE large model, with breakthrough performance improvements.
🎵 "Tiangong SkyMusic" is China's first music SOTA model, supporting multiple dialects and end-to-end music generation capabilities.
🔍 "Tiangong 3.0" features multi-round search, chart drawing, research mode, and other AI capabilities, while "Tiangong SkyMusic" supports reference music generation and dialect song generation.
Details:🔗 https://top.aibase.com/tool/tiangongskymusic
Tongyi Qianwen Releases Code Model CodeQwen1.5 Based on Qwen1.5
Based on the Qwen language model, the code expert model CodeQwen1.5 has 7B parameters, a GQA architecture, and supports 92 programming languages. CodeQwen1.5 excels in code generation, long sequence modeling, code modification, and SQL capabilities, greatly enhancing developers' work efficiency.
Highlights:
⭐ CodeQwen1.5 has 7B parameters and supports 92 programming languages.
⭐ It performs exceptionally in code generation, long sequence modeling, code modification, and SQL capabilities.
⭐ It showcases outstanding capabilities in multilingual programming, long sequence support, and code modification.
Details:🔗 https://qwenlm.github.io/zh/blog/codeqwen1.5/
Spline Releases 3D Generation Tool AI3D Generation
Spline's latest 3D generation tool marks a significant advancement in the 3D design field, offering an intuitive interface, a rich resource library and templates, advanced AI-assisted design, and team collaboration features.
Highlights:
🌟 Intuitive interface suitable for 3D design beginners to quickly get started
🎨 Rich resource library and templates accelerate the design process
🤖 Advanced AI-assisted design enhances the professional quality of the work
Details:🔗 https://top.aibase.com/tool/spline-ai
CTRL-F-VIDEO: Allows Users to Search for Specific Words or Phrases in Videos
In the digital age, video has become an important channel for information acquisition. The CTRL-F-VIDEO project provides in-video search functionality through a Chrome extension, saving users time and using OpenAI technology to convert audio to text for precise searching. The project demonstrates excellent processing capabilities.
Highlights:
⚙️ Improved video search efficiency: users can search for specific content within YouTube videos, saving time on retrieval.
🔍 Core technology: text search and timeline marking, audio-to-text conversion, matching vocabulary marking.
💡 Innovative possibilities: the project opens up new ideas for video content search and management, although in its early stages, it performs excellently.
Details:🔗 https://top.aibase.com/tool/ctrl-f-video
Bilibili UP Master Uses AI Technology to "Revive" the Giant Race, Sparking Wide Attention
This article introduces a Bilibili UP Master who uses AI technology to create seemingly authentic historical myth videos, sparking 1.3 million views and discussions. Viewers discuss the authenticity of AI-generated content, with netizens launching a "Identify AI Contest" to explore video details.
Highlights:
🎬 Bilibili UP Master creates seemingly authentic historical myth videos through AI technology
🤔 Viewers discuss the authenticity of AI-generated content, with the boundary between reality and fiction becoming a focus
💬 Netizens launch an "Identify AI Contest" to explore video details
Details: https://www.chinaz.com/2024/0417/1610688.shtml
Free AI Music Generation Tool Sonauto
Sonauto is a powerful online music creation tool that enables everyone to become a music creator through artificial intelligence technology. Its core advantages lie in editing functions and user-friendly operation interfaces, simplifying the music creation process.
Highlights:
⚙️ Users do not need professional music production skills, and can generate complete music works through text prompts or lyrics
🎵 Provides various music styles to choose from, allowing users to further edit and adjust the generated music works
🔓 Suitable for various groups, including music lovers, independent musicians, video content creators, etc.
Details:🔗 https://top.aibase.com/tool/sonauto-
MiniMax XiYu Technology Releases Trillion MoE Model abab6.5
MiniMax XiYu Technology has launched the abab6.5 series models, a trillion MoE model, showcasing leading large language model capabilities. The abab6.5 and abab6.5s will be updated to products, and users can experience advanced language models through the Hailuo AI and MiniMax Open Platform.
Highlights:
⚙️ The abab6.5 series includes the abab6.5 and abab6.5s models, supporting a context length of 200k tokens.
💡 The abab6.5s is more efficient and can process nearly 30,000 words of text in 1 second, approaching the level of the world's leading large language models.
🚀 The abab6.5 and abab6.5s will be updated to MiniMax products, including the Hailuo AI and MiniMax Open Platform.
Multimodal Language Model Reka Core: Analyzes Images, Videos, Audio
Reka Core is a cutting-edge multimodal language model with powerful performance and flexible deployment, comparable to industry-leading models. It has key capabilities such as multimodal understanding, ultra-long context windows, and strong reasoning, and is collaborating with leading technology platforms like Snowflake and Oracle to advance multimodal technology development.
Highlights:
🌟 Reka Core has excellent performance and flexible deployment, capable of analyzing images, videos, and audio.
🔍 Core has multimodal understanding, ultra-long context windows, and strong reasoning capabilities.
🤝 Reka collaborates with Snowflake, Oracle, etc., to advance multimodal technology development, helping customers maximize data value.
Details:🔗 https://top.aibase.com/tool/reka-core
Performance Exceeds LLaMA2-7B! AI Model JetMoE-8B Training Cost Less Than $100,000
JetMoE-8B is an AI model with a sparse activation architecture that outperforms LLaMA2-7B, LLaMA-13B, and DeepseekMoE-16B, with a training cost of less than $100,000. Its unique architecture design has low computational costs and high parameter activation efficiency, with training fully reliant on public data and completely open source. It performs excellently in evaluations, demonstrating efficient performance and cost advantages.
Highlights:
⚙️ JetMoE-8B uses a sparse activation architecture, outperforming multiple AI models
💡 Unique design makes computational costs low, with high efficiency in activating parameters per input token
🔗 JetMoE-8B training fully relies on public data, with the entire process completely open source
Details:🔗 https://huggingface.co/jetmoe/jetmoe-8b
Ollama Software Update: Performance Improvement and New Model Support
The latest Ollama software update brings significant performance improvements and support for multiple popular models. It optimizes VRAM and GPU utilization, adds model support for WizardLM2, Snowflake Arctic Embed, etc., enhancing software stability and flexibility. Users will be able to complete tasks more efficiently and explore new possibilities in artificial intelligence.
Highlights:
🚀 Optimized VRAM and GPU utilization, reducing memory errors and improving software performance.
💡 Added model support for WizardLM2, Snowflake Arctic Embed, etc., improving text generation quality.
🔗 Read detailed introductions and model download addresses to learn more about the updates.
Detailed introduction: https://github.com/ollama/ollama/releases
Model download: https://top.aibase.com/tool/ollama
37% of Professionals Save 5-10 Hours per Week Using Generative AI Tools
Generative artificial intelligence tools bring efficiency improvements, helping professionals work more efficiently. Through Generative AI tools, professionals achieve time savings in their work.
Highlights:
⭐ 37% of professionals save 5-10 hours per week
⭐ Generative AI tools improve work efficiency
⭐ Professionals save work time through Generative AI tools
——————
Daily midjourney prompt: ID Photo
Image source: Image generated by AI, image licensed by Midjourney
Image Prompt + A handsome 20-year-old Chinese man wearing a white shirt, exquisite Asian face, a delicate appearance, Korean style short hair, eyebrows and ears exposed, no makeup, front, one-inch photo, stock photo, blue background, studio lighting --ar5:7--no necklace or earrings --s250--iw2
A handsome 20-year-old Chinese man in a white shirt, with an exquisite Asian face, a delicate appearance, Korean style short hair, exposed eyebrows and ears, no makeup, front, one-inch photo, stock photo, blue background, studio lighting --ar5:7--no necklace or earrings
Image source: Image generated by AI, image licensed by Midjourney