Update .gitattributes

Merge pull request ollama#4089 from ollama/mxyng/target-invalid

server: destination invalid

gpu: add 512MiB to darwin minimum, metal doesn't have partial offload…

…ing overhead (ollama#4068)

Merge pull request ollama#3968 from dhiltgen/win_generate

Fine grain control over windows generate steps

Merge pull request ollama#3925 from dhiltgen/bump

Bump llama.cpp to b2737

Merge pull request ollama#3933 from dhiltgen/ci_fixes

Move cuda/rocm dependency gathering into generate script

Merge pull request ollama#3926 from dhiltgen/ci_fixes

Fix release CI

Merge pull request ollama#3923 from ollama/mxyng/mem

only count output tensors

Merge pull request ollama#3684 from ollama/mxyng/scale-graph

scale graph based on gpu count

app: gracefully shut down `ollama serve` on windows (ollama#3641)

* app: gracefully shut down `ollama serve` on windows

* fix linter errors

* bring back `HideWindow`

* remove creation flags

* restore `windows.CREATE_NEW_PROCESS_GROUP`