mirror of
https://github.com/hanzoai/ml.git
synced 2026-11-20 23:08:29 +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]>
PaliGemma
HuggingFace Model Card - Model Page
cargo run --features cuda --release --example paligemma -- \
--prompt "caption fr" --image hanzo-ml-examples/examples/yolo-v8/assets/bike.jpg
loaded image with shape Tensor[dims 1, 3, 224, 224; bf16, cuda:0]
loaded the model in 1.267744448s
caption fr. Un groupe de cyclistes qui sont dans la rue.
13 tokens generated (56.52 token/s)
cargo run --features cuda --release --example paligemma -- \
--prompt "caption fr" --image hanzo-ml-examples/examples/flux/assets/flux-robot.jpg
loaded image with shape Tensor[dims 1, 3, 224, 224; bf16, cuda:0]
loaded the model in 1.271492621s
caption fr une image d' un robot sur la plage avec le mot rouillé
15 tokens generated (62.78 token/s)