57 lines
2.3 KiB
Bash
57 lines
2.3 KiB
Bash
# Generated as /etc/mike-ai/stack.env by install.sh. Never commit real values.
|
|
AI_BIND_ADDRESS=10.77.0.2
|
|
MODEL_DIR=/data/models
|
|
ROUTER_API_KEY=GENERATED_BY_INSTALLER
|
|
CONTROLLER_TOKEN=GENERATED_BY_INSTALLER
|
|
FLUX_COMPONENT_DIR=/data/models/FLUX.2-klein-9B-components
|
|
FLUX_TRANSFORMER_DIR=/data/models/FLUX.2-klein-9B-fp8
|
|
IMAGE_GPU_DEVICES=GPU-8ad38c6c-5a01-9d8e-1dfa-ed662ad78fbe
|
|
QWEN3_TTS_IMAGE=ghcr.io/malaiwah/qwen3-tts-server:latest@sha256:b363a01d08b1bbecbfc3ca6f585368fae2cfdc591f9ecca6643738369f9a9d98
|
|
QWEN3_TTS_CACHE_DIR=/data/models/qwen3-tts-cache
|
|
QWEN3_TTS_VOICES_DIR=/data/models/qwen3-tts-voices
|
|
QWEN3_TTS_GPU_DEVICE=GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b
|
|
AI_DNS=192.168.1.1
|
|
DEFAULT_REASONING_EFFORT=off
|
|
|
|
FAST_MODEL_FILE=qwen-mix/Qwen3.8-27B-IQ4-MIX.gguf
|
|
MEDIUM_MODEL_FILE=qwen-pure/qwen3.8-27b-IQ4_XS-pure.gguf
|
|
LARGE_MODEL_FILE=qwen-pure/qwen3.8-27b-IQ4_XS-pure.gguf
|
|
ULTRA_MODEL_FILE=qwen-pure/qwen3.8-27b-IQ4_XS-pure.gguf
|
|
UNCENSORED_MODEL_FILE=qwen3.8-27b-abliterated/Qwen3.8-27B-ABLITERATED-Q4_K_M.gguf
|
|
UNCENSORED_PROJECTOR_FILE=qwen3.8-27b-abliterated/mmproj-Qwen3.8-27B-ABLITERATED-F16.gguf
|
|
VISION_PROJECTOR_FILE=qwen/mmproj-BF16.gguf
|
|
|
|
FAST_CONTEXT=76800
|
|
FAST_BATCH_SIZE=64
|
|
FAST_UBATCH_SIZE=32
|
|
MEDIUM_CONTEXT=160000
|
|
MEDIUM_BATCH_SIZE=2048
|
|
MEDIUM_UBATCH_SIZE=128
|
|
LARGE_CONTEXT=192000
|
|
LARGE_BATCH_SIZE=2048
|
|
LARGE_UBATCH_SIZE=128
|
|
ULTRA_CONTEXT=262144
|
|
ULTRA_BATCH_SIZE=2048
|
|
ULTRA_UBATCH_SIZE=128
|
|
UNCENSORED_CONTEXT=80000
|
|
UNCENSORED_BATCH_SIZE=2048
|
|
UNCENSORED_UBATCH_SIZE=128
|
|
FAST_GPU_DEVICES=GPU-8ad38c6c-5a01-9d8e-1dfa-ed662ad78fbe,GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b
|
|
MEDIUM_GPU_DEVICES=GPU-8ad38c6c-5a01-9d8e-1dfa-ed662ad78fbe,GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b
|
|
MEDIUM_TENSOR_SPLIT=85,15
|
|
LARGE_GPU_DEVICES=GPU-8ad38c6c-5a01-9d8e-1dfa-ed662ad78fbe,GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b
|
|
LARGE_TENSOR_SPLIT=86,14
|
|
ULTRA_GPU_DEVICES=GPU-8ad38c6c-5a01-9d8e-1dfa-ed662ad78fbe,GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b
|
|
ULTRA_TENSOR_SPLIT=80,20
|
|
UNCENSORED_GPU_DEVICES=GPU-8ad38c6c-5a01-9d8e-1dfa-ed662ad78fbe,GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b
|
|
UNCENSORED_TENSOR_SPLIT=90,10
|
|
UNCENSORED_MTP_MAX=2
|
|
LLAMA_THREADS=6
|
|
LLAMA_THREADS_BATCH=6
|
|
FAST_PARALLEL_SLOTS=1
|
|
LLAMA_CACHE_RAM_MIB=32768
|
|
MEDIUM_PARALLEL_SLOTS=1
|
|
LARGE_PARALLEL_SLOTS=1
|
|
ULTRA_PARALLEL_SLOTS=1
|
|
UNCENSORED_PARALLEL_SLOTS=1
|