Ollama tweeks
This commit is contained in:
@@ -68,11 +68,13 @@ services:
|
|||||||
- HCC_AMDGPU_TARGET=gfx906
|
- HCC_AMDGPU_TARGET=gfx906
|
||||||
- HIP_VISIBLE_DEVICES=0,1
|
- HIP_VISIBLE_DEVICES=0,1
|
||||||
- ROCR_VISIBLE_DEVICES=0,1
|
- ROCR_VISIBLE_DEVICES=0,1
|
||||||
- HSA_ENABLE_SDMA=0
|
- HSA_ENABLE_SDMA=1
|
||||||
|
- OLLAMA_MAX_LOADED_MODELS=1
|
||||||
|
- OLLAMA_MAX_QUEUE=512
|
||||||
- OLLAMA_HOST=0.0.0.0
|
- OLLAMA_HOST=0.0.0.0
|
||||||
- OLLAMA_DEBUG=1
|
- OLLAMA_DEBUG=1
|
||||||
- OLLAMA_FLASH_ATTENTION=0
|
- OLLAMA_FLASH_ATTENTION=1
|
||||||
- OLLAMA_NUM_PARALLEL=2
|
- OLLAMA_NUM_PARALLEL=1
|
||||||
devices:
|
devices:
|
||||||
# Map the render nodes and KFD for ROCm to work inside the container
|
# Map the render nodes and KFD for ROCm to work inside the container
|
||||||
- /dev/kfd:/dev/kfd
|
- /dev/kfd:/dev/kfd
|
||||||
|
|||||||
Reference in New Issue
Block a user