Llama 2 Everywhere (L2E)

Step by step explanation/tutorial of llama2.c

minimal C implementation of speculative decoding based on llama2.c

Llama 2 Everywhere (L2E)

llama2.c running on the PSP

TinyLlama on Cortex-M55 using CMSIS-DSP and Helium vector instructions

Automatic Neural Network Assistant (ANNA) is a very powerful cross-platform AI toolkit. It contains versatile CLI/GUI LLM tools optimized for local inference, and supports optional remote offloading.

Integrating LLMs into reMarkable could unlock transformative features like auto-completion, grammar/style corrections, contextual suggestions, and note summarization or paraphrasing. These tools would enhance productivity and elevate the user experience, making the device a powerful ally for creative and professional tasks.

使用opengl计算着色器实现的transformer (llama2)模型推理。implementation of the transformer (llama2) model,using OpenGL Compute Shader

C++ Implementation of Meta's LLaMA v2 Engine. Credited to ggerganov/llama.cpp

Inference Llama 2 in one file of pure C

A minimal hybrid architecture combining llama2.c with a symbolic concept-slot memory for persistent, editable knowledge and neural-symbolic integration.

基于OpenGL计算着色器的常见大模型推理（目前已支持llama2和rwkv4）

Adaptation of llama2.c (https://github.com/karpathy/llama2.c) to run on a PULP microcontroller (https://github.com/pulp-platform/pulp)