LLM

Getting Started with OpenChat 3.5

To quick start, you can run Orca or a list of other models with just one single command on your own device. The command tool automatically downloads and installs the WasmEdge runtime, the model files, and the portable Wasm apps for inference. The OpenChat 13B model is fine-tuned on llama2 13B base model for conversation / chat applications. It has a novel fine-tuning method that is more effective than SFT but less expensive than RLFT.…

Getting Started with Mistral-7b-Instruct-v0.1

The mistral-7b-instruct-v0.1 model is a 7B instruction-tuned LLM released by Mistral AI. It is a true open source model licensed under Apache 2.0. It has a context length of 8,000 tokens and performs on par with 13B llama2 models. It is great for generating prose, summarizing documents, and writing code. In this article, we will cover How to run mistral-7b-instruct-v0.1 on your own device How to create an OpenAI-compatible API service for mistral-7b-instruct-v0.…

Getting Started with Dolphin-2.2-yi-34b

The dolphin-2.2-yi-34b model is based on the 34B LLM, Yi, released by the 01.AI team. Yi is converted to the llama2 format by Charles Goddard and then further fine-tuned by Eric Hartford. In this article, we will cover How to run dolphin-2.2-yi-34b on your own device How to create an OpenAI-compatible API service for dolphin-2.2-yi-34b We will use the Rust + Wasm stack to develop and deploy applications for this model.…

Fast and Portable Llama2 Inference on the Heterogeneous Edge

The Rust+Wasm stack provides a strong alternative to Python in AI inference. Compared with Python, Rust+Wasm apps could be 1/100 of the size, 100x the speed, and most importantly securely run everywhere at full hardware acceleration without any change to the binary code. Rust is the language of AGI. We created a very simple Rust program to run inference on llama2 models at native speed. When compiled to Wasm, the binary application (only 2MB) is completely portable across devices with heterogeneous hardware accelerators.…

Wasm as the runtime for LLMs and AGI

Large Language Model (LLM) AI is the hottest thing in tech today. With the advancement of open source LLMs, new fine-tuned and domain-specific LLMs are emerging everyday in areas as diverse as coding, education, medical QA, content summarization, writing assistance, and role playing. Don't you want to try and chat with those LLMs on your computers and even IoT devices? But, Python / PyTorch, which are traditionally required to run those models, consists of 3+GB of fragile inter-dependent packages.…

Rust + WebAssembly: Building Infrastructure for Large Language Model Ecosystems

This is a talk at the track “The Programming Languages Shaping the Future of Software Development” at QCon 2023 Beijing on Sept 6th, 2023. The session aims to address the challenges faced by the current mainstream Python and Docker approach in building infrastructure for large language model(LLM) applications. It introduced the audience to the advantages of the Rust + WebAssembly approach, emphasizing its potential in addressing the performance, security, and efficiency concerns associated with the traditional approach.…

How do I create a GGUF model file?

The llama2 family of LLMs are typically trained and fine-tuned in PyTorch. Hence, they are typically distributed as PyTorch projects on Huggingface. However, when it comes to inference, we are much more interested in the GGUF model format for three reasons. Python is not a great stack for AI inference. We would like to get rid of PyTorch and Python dependency in production systems. GGUF can support very efficient zero-Python inference using tools like llama.…