4d ago · llama.cpp releases →
ggml-zdnn: accept non-zdnn tensors, invalidate compute tensors after…
Includes: llama.cpp b12219, llama.cpp b12220, llama.cpp b12221, b12223: models : Attach Granite NVFP4 weight scales (#31449), b12224: ggml-backend: don't upload inputs that are not read (#31419), llama.cpp b12226, b12227: tests : keep dummy ffn.act scales away from zero (#31940), b12228: ggml : implement the backward pass for TRI (#30712), llama.cpp b12229, llama.cpp b12230