Skip to content

b12572

Pre-release
Pre-release

Choose a tag to compare

@github-actions github-actions released this 20 Nov 14:07
652d559

llama : move the graph node budget into the model classes (#32648)

graph_max_nodes() picked the nodes-per-tensor factor from lists of
architectures. Each model class now gives it through a new virtual,
llama_model::graph_nodes_per_tensor(), which defaults to 8. The
architectures that need 32 or 64 override it in models.h, and DFlash
keeps its hparams-based rule in dflash.cpp.

LoRA adapter nodes are now added for every architecture, not only for
the ones on the default factor.

Website:

Attestations:

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

  • DISABLED
  • openEuler x86 (310p)
  • openEuler x86 (910b, ACL Graph)
  • openEuler aarch64 (310p)
  • openEuler aarch64 (910b, ACL Graph)

UI: