Mirrors/photoprism
1
0
Fork 0
mirror of https://github.com/photoprism/photoprism.git synced 2026-07-17 16:49:04 +00:00
Code Issues Projects Releases Wiki Activity

Docker: Tune Ollama defaults: flash-attention on, telemetry opt-outs

Browse source
This commit is contained in:
Cathie Integra 2026-05-23 09:41:12 +02:00
parent 0544f71c12
commit 771bc3688c
5 changed files with 30 additions and 15 deletions
Download patch file Download diff file Expand all files Collapse all files

9
setup/docker/arm64/compose.yaml
View file

@ -183,12 +183,15 @@ services:
OLLAMA_MULTIUSER_CACHE: "false" # optimize prompt caching for multi-user scenarios
OLLAMA_NOPRUNE: "false" # disables pruning of model blobs at startup
OLLAMA_NOHISTORY: "true" # disables readline history
OLLAMA_FLASH_ATTENTION: "false" # enables the experimental flash attention feature
OLLAMA_KV_CACHE_TYPE: "f16" # cache quantization (f16, q8_0, or q4_0)
OLLAMA_FLASH_ATTENTION: "true" # required for OLLAMA_KV_CACHE_TYPE quantization
OLLAMA_KV_CACHE_TYPE: "f16" # cache precision: f16 (default), q8_0, q4_0
OLLAMA_SCHED_SPREAD: "false" # allows scheduling models across all GPUs.
OLLAMA_NEW_ENGINE: "true" # enables the new Ollama engine
# OLLAMA_DEBUG: "true" # shows additional debug information
# OLLAMA_INTEL_GPU: "true" # enables experimental Intel GPU detection
## Telemetry / privacy opt-outs (containers do not inherit /etc/environment):
DO_NOT_TRACK: "true"
HF_HUB_DISABLE_TELEMETRY: "1"
# OLLAMA_NO_CLOUD: "1" # uncomment to disable Ollama Cloud models/features
volumes:
- "./ollama:/root/.ollama"

9
setup/docker/compose.yaml
View file

@ -188,12 +188,15 @@ services:
OLLAMA_MULTIUSER_CACHE: "false" # optimize prompt caching for multi-user scenarios
OLLAMA_NOPRUNE: "false" # disables pruning of model blobs at startup
OLLAMA_NOHISTORY: "true" # disables readline history
OLLAMA_FLASH_ATTENTION: "false" # enables the experimental flash attention feature
OLLAMA_KV_CACHE_TYPE: "f16" # cache quantization (f16, q8_0, or q4_0)
OLLAMA_FLASH_ATTENTION: "true" # required for OLLAMA_KV_CACHE_TYPE quantization
OLLAMA_KV_CACHE_TYPE: "f16" # cache precision: f16 (default), q8_0, q4_0
OLLAMA_SCHED_SPREAD: "false" # allows scheduling models across all GPUs.
OLLAMA_NEW_ENGINE: "true" # enables the new Ollama engine
# OLLAMA_DEBUG: "true" # shows additional debug information
# OLLAMA_INTEL_GPU: "true" # enables experimental Intel GPU detection
## Telemetry / privacy opt-outs (containers do not inherit /etc/environment):
DO_NOT_TRACK: "true"
HF_HUB_DISABLE_TELEMETRY: "1"
# OLLAMA_NO_CLOUD: "1" # uncomment to disable Ollama Cloud models/features
## NVIDIA GPU Hardware Acceleration (see https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html):
# NVIDIA_VISIBLE_DEVICES: "all"
# NVIDIA_DRIVER_CAPABILITIES: "compute,utility"

9
setup/docker/nvidia/compose.yaml
View file

@ -188,12 +188,15 @@ services:
OLLAMA_MULTIUSER_CACHE: "false" # optimize prompt caching for multi-user scenarios
OLLAMA_NOPRUNE: "false" # disables pruning of model blobs at startup
OLLAMA_NOHISTORY: "true" # disables readline history
OLLAMA_FLASH_ATTENTION: "false" # enables the experimental flash attention feature
OLLAMA_KV_CACHE_TYPE: "f16" # cache quantization (f16, q8_0, or q4_0)
OLLAMA_FLASH_ATTENTION: "true" # required for OLLAMA_KV_CACHE_TYPE quantization
OLLAMA_KV_CACHE_TYPE: "f16" # cache precision: f16 (default), q8_0, q4_0
OLLAMA_SCHED_SPREAD: "false" # allows scheduling models across all GPUs.
OLLAMA_NEW_ENGINE: "true" # enables the new Ollama engine
# OLLAMA_DEBUG: "true" # shows additional debug information
# OLLAMA_INTEL_GPU: "true" # enables experimental Intel GPU detection
## Telemetry / privacy opt-outs (containers do not inherit /etc/environment):
DO_NOT_TRACK: "true"
HF_HUB_DISABLE_TELEMETRY: "1"
# OLLAMA_NO_CLOUD: "1" # uncomment to disable Ollama Cloud models/features
## NVIDIA GPU Hardware Acceleration (see https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html):
NVIDIA_VISIBLE_DEVICES: "all"
NVIDIA_DRIVER_CAPABILITIES: "compute,utility"