From ea9b20ae42ce8bc22e7bfbbbf90aef5fcfe83917 Mon Sep 17 00:00:00 2001 From: Onur Gündüz Date: Sat, 27 Jun 2026 07:08:37 +0200 Subject: chore: update README.md --- README.md | 27 ++++++++++++++++----------- 1 file changed, 16 insertions(+), 11 deletions(-) (limited to 'README.md') diff --git a/README.md b/README.md index dcb0c5e..958eb1d 100644 --- a/README.md +++ b/README.md @@ -3,24 +3,24 @@ ffwd is a high-performance text embedding inference library and toolkit for transformer-based models. -It runs natively on Apple Silicon and NVIDIA GPUs, with CPU support for both x86 -and ARM architectures. +It runs natively on Apple silicon (M1 or later) and NVIDIA GPUs, with CPU +support for both x86 and ARM architectures. ## Supported models Tested architectures include: -- Qwen3: Qwen3-Embedding and pplx-embed families -- Qwen2: GTE-Qwen2 -- BERT/RoBERTa: BERT, BGE, MiniLM, and XLM-R multilingual encoders such as - multilingual-E5 and Snowflake Arctic +- **Qwen3:** `Qwen3-Embedding` and `pplx-embed` families +- **Qwen2:** `GTE-Qwen2` +- **BERT/RoBERTa:** `BERT`, `BGE`, `MiniLM`, and `XLM-R` multilingual encoders + (such as `multilingual-E5` and `Snowflake Arctic`) ## Components ### libffwd -[libffwd](./libffwd) is the core embedding inference library, providing -optimized kernels, tokenization, and model loading. +[libffwd](./libffwd) is the core embedding inference library. It provides +optimized kernels, tokenizers, and model loading. ### ffwd-server @@ -37,10 +37,15 @@ computing cosine similarity. Before building, install the dependencies for your platform: -- On Linux, install `libopenblas-dev` and `libyyjson-dev`. -- On macOS, install `mlx`, `mlx-c` and `yyjson` through Homebrew. +- **Linux:** Install `libopenblas-dev` (required for both CPU and GPU builds). +- **macOS:** Install `mlx` and `mlx-c` via Homebrew if you're targeting + Metal/GPU. Otherwise, no extra dependencies needed. -Once the dependencies are in place, run `make` to build the libraries and tools. +Once the dependencies are in place, build the project by running: + +```sh +make # or make gpu for GPU build +``` ## Acknowledgements -- cgit v1.3.1