mirror of
https://github.com/hanzoai/ml.git
synced 2026-11-20 22:54:34 +00:00
Three-way per file through the rename: base 39355c6c91, theirs the upstream
merge of main with 35d7ae7c (via 7ef6a694d), ours origin/main.
Lands the barrier thread pool, runtime CPU feature dispatch, aarch64 and x86
repacked quantized kernels (indexed_gemv, gemv_fused_shared_lhs), quantized
embedding on CPU/Metal/CUDA, paged flash-attn kernels, cutile fused MoE, the
in-place op aliasing API, parking_lot storage locks, Metal buffer builders and
debug labels, safetensors 0.8, hf-hub 1.0, tokenizers 0.23.
Ours keeps ROCm, Vulkan, wgpu, the IQ/MXFP4/NVFP4/TQ dtypes, and the i-quant
weight/activation block split in matmul. Our backends ride InplaceOpN and the
quantized embedding; Metal get_rows gains the IQ kernels. ug leaves, as upstream.
Co-authored-by: Hanzo Dev <[email protected]>
hanzo-ml-dinov2
[Depth Anything V2] is a model for Monocular Depth Estimation (MDE, i.e. just using a single image) which builds on the DINOv2 vision transformer.
This example first instantiates the DINOv2 model and then proceeds to create DepthAnythingV2 and run it.
Running an example with color map and CUDA
cargo run --features cuda,depth_anything_v2 --package hanzo-ml-examples --example depth_anything_v2 -- --color-map --image hanzo-ml-examples/examples/yolo-v8/assets/bike.jpg