mirror of
https://github.com/ollama/ollama.git
synced 2025-11-12 11:07:55 +01:00
c2f5d6662b
Worst case graph preallocation was disabled by a27462b
"ollamarunner: Temporarily disable worst case graph preallocation"
since it caused crashes with large batches when not using the GPU.
This backports upstream llama.cpp commit f057808
"ggml: Don't assert fail when tensor data changes (#13222)", which
fixes the underlying bug and allows reverting the previous workaround.
1.6 KiB
1.6 KiB