michael/AI-Profile-Router
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Synchronize repository with Athena deployment

Browse Source
This commit is contained in:
Mikei386
2026-09-13 20:01:36 +02:00
parent 040a2df48b
commit fce9900389
60 changed files with 2570 additions and 607 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.env.example
+2 -9
View File
@@ -3,10 +3,9 @@ AI_BIND_ADDRESS=10.77.0.2
MODEL_DIR=/data/models
ROUTER_API_KEY=GENERATED_BY_INSTALLER
CONTROLLER_TOKEN=GENERATED_BY_INSTALLER
FLUX_MODEL_DIR=/data/models/FLUX.2-klein-4B
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
PIPER_TTS_VERSION=1.6.0
PIPER_VOICE=de_DE-thorsten-high
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
@@ -16,7 +15,6 @@ 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
BETA1_MODEL_FILE=qwen3.8-27b-gsq-rco-test/Qwen3.8-27B-GSQ-RCO-IQ3_XXS-mtp.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
@@ -29,9 +27,6 @@ FAST_UBATCH_SIZE=32
MEDIUM_CONTEXT=160000
MEDIUM_BATCH_SIZE=2048
MEDIUM_UBATCH_SIZE=128
BETA1_CONTEXT=192000
BETA1_BATCH_SIZE=2048
BETA1_UBATCH_SIZE=128
LARGE_CONTEXT=192000
LARGE_BATCH_SIZE=2048
LARGE_UBATCH_SIZE=128
@@ -44,7 +39,6 @@ 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
BETA1_GPU_DEVICES=GPU-8ad38c6c-5a01-9d8e-1dfa-ed662ad78fbe,GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b
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
@@ -57,7 +51,6 @@ LLAMA_THREADS_BATCH=6
FAST_PARALLEL_SLOTS=1
LLAMA_CACHE_RAM_MIB=32768
MEDIUM_PARALLEL_SLOTS=1
BETA1_PARALLEL_SLOTS=1
LARGE_PARALLEL_SLOTS=1
ULTRA_PARALLEL_SLOTS=1
UNCENSORED_PARALLEL_SLOTS=1