Code

April 2024 Binary Update#662

Code

Merged

martindevans merged 7 commits intoSciSharp:masterSciSharp/LLamaSharp:masterfrom martindevans:april_2024_binary_updatemartindevans/LLamaSharp:april_2024_binary_updateCopy head branch name to clipboardApr 16, 2024

Commits

Commits on Apr 12, 2024

• Updated binaries, using [this build](https://github.com/SciSharp/LLamaSharp/actions/runs/8654672719/job/23733195669) for llama.cpp commit f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7.

  Show description for f5524e8
  martindevans

  committed
  f5524e8

  Copy full SHA for f5524e8

• Moved native methods to more specific locations.

  Show description for 56e7e41
  martindevans

  committed
  56e7e41

  Copy full SHA for 56e7e41

• Removed exception if GpuLayerCount > 0 when GPU is not supported.

  martindevans

  committed
  ae5ad71

  Copy full SHA for ae5ad71

• - Added low level wrapper methods for new per-sequence state load/save in SafeLLamaContextHandle

  Show description for c933206
  martindevans

  committed
  c933206

  Copy full SHA for c933206

Commits on Apr 13, 2024

• Added update and defrag methods for KV cache in SafeLLamaContextHandle

  martindevans

  committed
  869f389

  Copy full SHA for 869f389

• Updated submodule to f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7

  martindevans

  committed
  0139b43

  Copy full SHA for 0139b43

Commits on Apr 15, 2024

• Passing the sequence ID when saving a single sequence state

  martindevans

  committed
  fd98941

  Copy full SHA for fd98941