note

llama.cpp Setup Guide for DiffusionGemma (Windows Environment)

Last time, I tried running DiffusionGemma on Windows 11 with an RTX 4070. In the end, due to insufficient VRAM, it only ran at 1/4 the speed of standard Gemma, but I would like to leave a record of ...
搜狐

刚开源！Meta Llama 4 来了，10M 超长记忆 + 免费可商用，开发者已经炸了

Llama 4 是 Meta 于 2025 年 4 月发布的多模态大语言模型系列，采用混合专家（MoE）架构，旗下包含 Scout（109B 总参数）、Maverick（400B 总参数）两个已开放权重的模型，以及仍在训练中的超旗舰 Behemoth（约 2T 总参数）。这一代模型原生支持图文多模态输入，最长 ...
theregister

Tinker with LLMs in the privacy of your own home using Llama.cpp

HANDS ON Training large language models (LLMs) may require millions or even billion of dollars of infrastructure, but the fruits of that labor are often more accessible than you might think. Many ...
GitHub

llama-cpp-python cuBLAS wheels

MacOS 11 and Windows ROCm wheels are unavailable for 0.2.21+. This is due to build issues with llama.cpp that are not yet resolved. ROCm builds for AMD GPUs: https ...
note

Switching from OpenAI to Local LLMs with Minimal Effort Using Llama.cpp's OpenAI-Compatible ...

There is something called a local LLM that allows you to run something like a simplified ChatGPT on your home computer. Companies that handle high-level confidential information, such as in finance, ...
GitHub

Releases: jllllll/llama-cpp-python-cuBLAS-wheels

python -m pip install llama-cpp-python --prefer-binary --extra-index-url=https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels/basic/cpu 0.1.85 builds likely won't ...