GitLab

* Add "ollama-telegram" to Extensions & Plugins

* Update README.md

handle ctrl+z

Fix adapter loading from SHA hash

* docker: set PATH, LD_LIBRARY_PATH, and capabilities

* example: update k8s gpu manifest

revert checksum calculation to calculate-as-you-go

* Add OllamaEmbeddings to python LangChain example

* typo

---------
Co-authored-by: Alec Hammond <alechammond@fb.com>

Co-authored-by: Cloud User <azureuser@testgpu2.hqzwom21okjenksna4y3c4ymjd.phxx.internal.cloudapp.net>

Co-authored-by: rootedbox

add new recent models as examples

progress: fix bar rate

Fix a typo in the CA update command

* add remote create to python example client

Co-authored-by: Wen Sun <iwendellsun@gmail.com>

* Support cuda build in Windows
* Enable dynamic NumGPU allocation for Windows

When CUDA peer access is enabled, multi-gpu inference will produce
garbage output. This is a known bug of llama.cpp (or nvidia). Until the
upstream bug is fixed, we can disable CUDA peer access temporarily
to ensure correct output.

See #961.

fix: gguf int type