Ollama is an application built on llama.cpp, enabling direct interaction with LLMs through your computer. It supports any GGUF quant models created by the community, such as those from bartowski, MaziyarPanahi, and many others, available on Hugging Face. With Ollama, there’s no need to create a new Modelfile—you can use any of the 45,000+ public … Read more
A cutting-edge 12B model featuring a 128k context length, developed by Mistral AI in partnership with NVIDIA. Mistral NeMo is a 12B model built in collaboration with NVIDIA. Mistral NeMo offers a large context window of up to 128k tokens. Its reasoning, world knowledge, and coding accuracy are state-of-the-art in its size category. As it … Read more
Llama 3.2 is available on Ollama! It’s lightweight and multimodal! It’s so fast and good! Meta’s Llama 3.2 goes small with 1B and 3B models. The Meta Llama 3.2 series of multilingual large language models (LLMs) consists of pretrained and instruction-tuned generative models available in 1B and 3B parameter configurations (text in/text out). Optimized for … Read more
Codestral is Mistral AI‘s inaugural code model, built specifically for code generation tasks, featuring a 22-billion parameter architecture. Proficient in over 80 programming languages. Codestral has been trained on a diverse dataset encompassing more than 80 languages, such as Python, Java, C, C++, JavaScript, Swift, Fortran, and Bash. It can complete functions, generate tests, and … Read more
LLaVA is an innovative end-to-end large multimodal model, integrating a vision encoder with Vicuna to enable comprehensive visual and language understanding. It has been updated to version 1.6. Run LLava LLaVA, which stands for Large Language and Vision Assistant, is a multimodal model that combines both visual and language processing abilities. It integrates a vision … Read more
A set of Mixture of Experts (MoE) model with open weights by Mistral AI in 8x7b and 8x22b parameter sizes. The Mixtral large language models (LLMs) are a set of pretrained generative Sparse Mixture of Experts (SMoE). Sizes: Mixtral 8x22B: To run: ollama run mixtral:8x22b Mixtral 8x22B sets a new benchmark for performance and efficiency … Read more
Qwen2.5 is the latest generation of the Qwen language models, offering a variety of base and instruction-tuned models with parameter sizes ranging from 0.5 to 72 billion. Key improvements in Qwen2.5 compared to Qwen2 include: Note: All models except the 3B and 72B are released under the Apache 2.0 license, while the 3B and 72B … Read more
Mistral is a 7B parameter model, distributed with the Apache license. It is available in both instruct (instruction following) and text completion. The Mistral AI team has noted that Mistral 7B: Performance in details We compared Mistral 7B to the Llama 2 family, and re-run all model evaluations ourselves for fair comparison. The performance comparison … Read more
Phi-3.5-mini is a lightweight, cutting-edge open model based on the same datasets as Phi-3, utilizing synthetic data and carefully curated publicly available websites with an emphasis on high-quality, reasoning-rich information. Part of the Phi-3 model family, it supports a 128K token context length. The model has undergone significant improvements through supervised fine-tuning, proximal policy optimization, … Read more
Qwen2 is trained on data across 29 languages, including English and Chinese. It comes in four different parameter sizes: 0.5B, 1.5B, 7B, and 72B. The 7B and 72B models feature an extended context length of up to 128k tokens. Model Qwen2-0.5B Qwen2-1.5B Qwen2-7B Qwen2-72B Params 0.49B 1.54B 7.07B 72.71B Non-Emb Params 0.35B 1.31B 5.98B 70.21B … Read more
