Popular repositories Loading

1. mlxfast-qwen38-tf-work mlxfast-qwen38-tf-work Public

   Optimisation work for the Yukon mlx.fast Qwen3.8-27B TensorFold challenge: in-place ring-cache attention for the DFlash drafter, patches, results and experiment journal.

   Python

2. TensorFold TensorFold Public

   Forked from ashhart/TensorFold

   LLM Inference Engine for Metal, CUDA and Vulkan.

   Zig

3. mlxfast-qwen38-27b-tf-engine mlxfast-qwen38-27b-tf-engine Public

   Forked from Layr-Labs/mlxfast-qwen38-27b-tf-engine

   Qwen 3.8 27B TensorFold decode-speed challenge on Apple silicon (MLX): engine, scoring contract and pinned benchd harness for Yukon

   Shell