Ollama 0.2 is here! Concurrency is now enabled by default.
https://t.co/EaOnzdFBJV
This unlocks 2 major features:
Parallel requests
Ollama can now serve multiple requests at the same time, using only a little bit of additional memory for each request. This enables use cases such as:
- Handling multiple chat sessions at the same time
- Hosting code completion LLMs for your team
- Processing different parts of a document simultaneously
- Running multiple agents at the same time
Run multiple models
Ollama now supports loading different models at the same time. This improves several use cases:
- Retrieval Augmented Generation (RAG): both the embedding and text completion models can be loaded into memory simultaneously.
- Agents: multiple versions of an agent can now run simultaneously
- Running large and small models side-by-side
Models are automatically loaded and unloaded based on requests and how much GPU memory is available.
The evolution of information retrieval and semantic technology,From initial keyword matching to today's intelligent agent models, each step of progress represents our deepening cognition of the nature of information and the ongoing exploration of artificial intelligence potential
New Relic and NVIDIA have launched the first observability integration to monitor AI applications built with NVIDIA NIM, offering features such as full AI stack integration, in-depth response tracking insights, model inventory, detailed GPU insights, and enhanced data security.
KolorsPrompts, a comprehensive evaluation dataset, was used to benchmark against other state-of-the-art models, showing Kolors' competitive performance. In human assessments by 50 imagery experts, Kolors achieved the highest scores in visual appeal and overall satisfaction.
he Kolors team released the Kolors model, a large-scale text-to-image generation model based on latent diffusion. Trained on billions of text-image pairs, Kolors excels in visual quality, semantic accuracy, and text rendering for both Chinese and English.
We’re at the beginning of a new Industrial Revolution. But instead of generating electricity, we’re generating intelligence… [Open source] activated every single company. Made it possible for every company to be an AI company.--Jensen Huang, CEO of NVIDIA
Rask AI is a comprehensive localization tool designed for content creators and companies. Its key features include "text-to-speech" and "voice cloning," allowing users to efficiently translate their videos into over 130 languages.
George Zhao, CEO of Honor, Emphasizes the Importance of Data Privacy in Artificial Intelligence
In an exclusive interview with CNBC, Zhao highlighted Honor's commitment to keeping all AI operations confined within devices to safeguard user data.
Critical thinking consists of two aspects. One is the critical and solemn examination, and the other is to deeply contemplate the issue. Both aspects are all that you require.
significantly streamlining developer workflows. MarsCode also includes AI-native features like intelligent code completion, bug fixing, and unit test generation, boosting programming efficiency and quality.
ByteDance has recently launched an innovative AI development tool, BeanBag MarsCode, designed to revolutionize the coding process with unparalleled efficiency and convenience. This tool supports multiple programming languages and mainstream IDEs,
offering outstanding performance in code writing, optimization, and testing. Its standout feature is enabling cloud-based coding without complex environment setup; users can perform programming, debugging, and other tasks simply through their web browsers,
DingTalk's AI Search focuses on solving information dispersion issues with features like personalized search, natural language input, and content traceability. It also supports multi-Agent workflows with AI assistants, enhancing enterprise information mgr and work efficiency.
Following OpenAI's announcement to end API services in China, DingTalk released version 7.6, integrating seven domestic models: Tongyi, MiniMax, MoonDark, Zhipu AI, ZeroOne, BaiChuan, and Orion Star. Users can switch models based on needs.