Remove obsolete experimental profile

.env.example

@@ -19,7 +19,6 @@ 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
-EXPERIMENTAL_MODEL_FILE=qwen/Qwen3.8-27B-UD-IQ4_XS.gguf
 VISION_PROJECTOR_FILE=qwen/mmproj-BF16.gguf
 
 FAST_CONTEXT=76800
@@ -37,7 +36,6 @@ ULTRA_UBATCH_SIZE=128
 UNCENSORED_CONTEXT=80000
 UNCENSORED_BATCH_SIZE=2048
 UNCENSORED_UBATCH_SIZE=128
-EXPERIMENTAL_CONTEXT=76800
 FAST_GPU_DEVICES=0,1
 MEDIUM_GPU_DEVICES=0,1
 MEDIUM_TENSOR_SPLIT=85,15
@@ -48,7 +46,6 @@ ULTRA_TENSOR_SPLIT=80,20
 UNCENSORED_GPU_DEVICES=0,1
 UNCENSORED_TENSOR_SPLIT=90,10
 UNCENSORED_MTP_MAX=2
-EXPERIMENTAL_GPU_DEVICES=0
 LLAMA_THREADS=6
 LLAMA_THREADS_BATCH=6
 FAST_PARALLEL_SLOTS=1
@@ -57,4 +54,3 @@ MEDIUM_PARALLEL_SLOTS=1
 LARGE_PARALLEL_SLOTS=1
 ULTRA_PARALLEL_SLOTS=1
 UNCENSORED_PARALLEL_SLOTS=1
-EXPERIMENTAL_PARALLEL_SLOTS=1

compose.yaml

@@ -471,55 +471,6 @@ services:
       - --spec-draft-type-v
       - f16
 
-  llama-experimental:
-    <<: *llama-common
-    container_name: mike-ai-llama-experimental
-    labels:
-      com.mike-ai.llama-profile: experimental
-    environment:
-      NVIDIA_VISIBLE_DEVICES: ${EXPERIMENTAL_GPU_DEVICES:-0}
-      NVIDIA_DRIVER_CAPABILITIES: compute,utility
-    command:
-      - --model
-      - "/models/${EXPERIMENTAL_MODEL_FILE:?EXPERIMENTAL_MODEL_FILE is required}"
-      - --alias
-      - qwen-experimental
-      - --ctx-size
-      - "${EXPERIMENTAL_CONTEXT:-76800}"
-      - --flash-attn
-      - "on"
-      - --cache-type-k
-      - q4_0
-      - --cache-type-v
-      - q4_0
-      - --cache-prompt
-      - --cache-reuse
-      - "${LLAMA_CACHE_REUSE:-256}"
-      - --cache-ram
-      - "${LLAMA_CACHE_RAM_MIB:-24576}"
-      - --parallel
-      - "${EXPERIMENTAL_PARALLEL_SLOTS:-1}"
-      - --kv-unified
-      - --jinja
-      - --reasoning
-      - auto
-      - --reasoning-preserve
-      - --host
-      - 0.0.0.0
-      - --port
-      - "8080"
-      - --metrics
-      - --fit
-      - "off"
-      - --n-gpu-layers
-      - all
-      - --no-mmap
-      - --no-ui
-      - --device
-      - CUDA0
-      - --split-mode
-      - none
-
   profile-controller:
     build: ./platform/docker/profile-controller
     image: mike-ai/profile-controller:local
@@ -531,7 +482,7 @@ services:
       - /var/run/docker.sock:/var/run/docker.sock
     environment:
       CONTROLLER_TOKEN: "${CONTROLLER_TOKEN:?CONTROLLER_TOKEN is required}"
-      ALLOWED_PROFILES: fast,medium,large,ultra,uncensored,experimental
+      ALLOWED_PROFILES: fast,medium,large,ultra,uncensored
       IMAGE_WORKER: image
     networks: [control]
     security_opt: ["no-new-privileges:true"]

config/install.env.example

@@ -65,9 +65,6 @@ UNCENSORED_MODEL_SHA256=5d53637a59cfcd3a4d8354e254ffd44943e5a693da2405a3e228c629
 UNCENSORED_PROJECTOR_FILE=qwen3.8-27b-abliterated/mmproj-Qwen3.8-27B-ABLITERATED-F16.gguf
 UNCENSORED_PROJECTOR_URL=https://huggingface.co/Blackfrost-AI/Qwen3.8-27B-ABLITERATED-GGUF/resolve/main/mmproj-Qwen3.8-27B-ABLITERATED-F16.gguf
 UNCENSORED_PROJECTOR_SHA256=2284099ce864f1023d721e6ef5eaef32bb56abdbc1dc561c6d91300f12ef2e4b
-EXPERIMENTAL_MODEL_FILE=qwen/Qwen3.8-27B-UD-IQ4_XS.gguf
-EXPERIMENTAL_MODEL_URL=https://huggingface.co/unsloth/Qwen3.8-27B-GGUF/resolve/main/Qwen3.8-27B-UD-IQ4_XS.gguf
-EXPERIMENTAL_MODEL_SHA256=40fac4050e940397dbf13087afd50f4734a11805bf9d65ef8ddd7483470e6199
 VISION_PROJECTOR_FILE=qwen/mmproj-BF16.gguf
 VISION_PROJECTOR_URL=https://huggingface.co/unsloth/Qwen3.8-27B-GGUF/resolve/main/mmproj-BF16.gguf
 VISION_PROJECTOR_SHA256=83ee4f4f205fa514161778c41df1ea14144faa0f713510893b63c2395f5c2d53
@@ -95,7 +92,6 @@ UNCENSORED_BATCH_SIZE=2048
 UNCENSORED_UBATCH_SIZE=128
 UNCENSORED_TENSOR_SPLIT=90,10
 UNCENSORED_MTP_MAX=2
-EXPERIMENTAL_CONTEXT=76800
 LLAMA_THREADS=6
 LLAMA_THREADS_BATCH=6
 FAST_PARALLEL_SLOTS=1
@@ -104,7 +100,6 @@ MEDIUM_PARALLEL_SLOTS=1
 LARGE_PARALLEL_SLOTS=1
 ULTRA_PARALLEL_SLOTS=1
 UNCENSORED_PARALLEL_SLOTS=1
-EXPERIMENTAL_PARALLEL_SLOTS=1
 PIPER_TTS_VERSION=1.6.0
 PIPER_VOICE=de_DE-thorsten-high
 XTTS_IMAGE=ghcr.io/coqui-ai/xtts-streaming-server:latest-cuda121@sha256:f7fb3b1f9d4bc88af94da1b5959d8002f1e0b003c97557164034eb8a29f01b90

config/profile-matrix.json

@@ -19,7 +19,7 @@
       "id": "medium",
       "alias": "qwen-medium",
       "context": 160000,
-      "parallel_slots": 2,
+      "parallel_slots": 1,
       "model_env": "MEDIUM_MODEL_FILE",
       "model_family": "Qwen3.8-27B IQ4 XS Pure",
       "gpu_split": "85:15",

docs/STANDARD_PROFILE_MATRIX.md

@@ -7,7 +7,7 @@ Standardprofil: **medium** · globales Ausgabelimit: **8192 Token**
 | Profil | API-Alias | Gesamtkontext | Slots | Modell | GPU-Verteilung | Vision | MTP |
 |---|---|---:|---:|---|---|---|---:|
 | fast | `qwen-fast` | 76,800 | 1 | Qwen3.8-27B IQ4 Mix | 5080 only | ja | 2 |
-| medium | `qwen-medium` | 160,000 | 2 | Qwen3.8-27B IQ4 XS Pure | 85:15 | ja | 3 |
+| medium | `qwen-medium` | 160,000 | 1 | Qwen3.8-27B IQ4 XS Pure | 85:15 | ja | 3 |
 | large | `qwen-large` | 192,000 | 1 | Qwen3.8-27B IQ4 XS Pure | 86:14 | ja | 3 |
 | ultra | `qwen-ultra` | 262,144 | 1 | Qwen3.8-27B IQ4 XS Pure | 80:20 | nein | 2 |
 | uncensored | `qwen-uncensored` | 80,000 | 1 | Qwen3.8-27B Abliterated Q4_K_M | 90:10 | ja | 2 |

install.sh

@@ -42,7 +42,6 @@ required=(AI_HOSTNAME ADMIN_USER MODEL_DIR FAST_MODEL_FILE
   FAST_MODEL_URL FAST_MODEL_SHA256 MEDIUM_MODEL_FILE MEDIUM_MODEL_URL
   MEDIUM_MODEL_SHA256 LARGE_MODEL_FILE LARGE_MODEL_URL LARGE_MODEL_SHA256
   ULTRA_MODEL_FILE ULTRA_MODEL_URL ULTRA_MODEL_SHA256
-  EXPERIMENTAL_MODEL_FILE EXPERIMENTAL_MODEL_URL EXPERIMENTAL_MODEL_SHA256
   VISION_PROJECTOR_FILE VISION_PROJECTOR_URL VISION_PROJECTOR_SHA256)
 for name in "${required[@]}"; do
   [[ -n "${!name:-}" ]] || die "Pflichtwert $name fehlt."
@@ -321,14 +320,12 @@ LARGE_MODEL_FILE=$LARGE_MODEL_FILE
 ULTRA_MODEL_FILE=$ULTRA_MODEL_FILE
 UNCENSORED_MODEL_FILE=$UNCENSORED_MODEL_FILE
 UNCENSORED_PROJECTOR_FILE=$UNCENSORED_PROJECTOR_FILE
-EXPERIMENTAL_MODEL_FILE=$EXPERIMENTAL_MODEL_FILE
 VISION_PROJECTOR_FILE=$VISION_PROJECTOR_FILE
 FAST_CONTEXT=${FAST_CONTEXT:-76800}
 MEDIUM_CONTEXT=${MEDIUM_CONTEXT:-160000}
 LARGE_CONTEXT=${LARGE_CONTEXT:-192000}
 ULTRA_CONTEXT=${ULTRA_CONTEXT:-262144}
 UNCENSORED_CONTEXT=${UNCENSORED_CONTEXT:-80000}
-EXPERIMENTAL_CONTEXT=${EXPERIMENTAL_CONTEXT:-76800}
 FAST_GPU_DEVICES=${TEXT_GPU_DEVICES:-0}${SECONDARY_GPU_DEVICES:+,$SECONDARY_GPU_DEVICES}
 MEDIUM_GPU_DEVICES=${TEXT_GPU_DEVICES:-0}${SECONDARY_GPU_DEVICES:+,$SECONDARY_GPU_DEVICES}
 MEDIUM_TENSOR_SPLIT=${MEDIUM_TENSOR_SPLIT:-90,10}
@@ -339,7 +336,6 @@ ULTRA_TENSOR_SPLIT=${ULTRA_TENSOR_SPLIT:-80,20}
 UNCENSORED_GPU_DEVICES=${TEXT_GPU_DEVICES:-0}${SECONDARY_GPU_DEVICES:+,$SECONDARY_GPU_DEVICES}
 UNCENSORED_TENSOR_SPLIT=${UNCENSORED_TENSOR_SPLIT:-90,10}
 UNCENSORED_MTP_MAX=${UNCENSORED_MTP_MAX:-2}
-EXPERIMENTAL_GPU_DEVICES=${TEXT_GPU_DEVICES:-0}
 IMAGE_GPU_DEVICES=${IMAGE_GPU_DEVICES:-${TEXT_GPU_DEVICES:-0}}
 FLUX_MODEL_DIR=${FLUX_MODEL_DIR:-/data/models/FLUX.2-klein-4B}
 LLAMA_THREADS=${LLAMA_THREADS:-6}
@@ -382,7 +378,6 @@ $LARGE_MODEL_FILE|$LARGE_MODEL_URL|$LARGE_MODEL_SHA256
 $ULTRA_MODEL_FILE|$ULTRA_MODEL_URL|$ULTRA_MODEL_SHA256
 $UNCENSORED_MODEL_FILE|$UNCENSORED_MODEL_URL|$UNCENSORED_MODEL_SHA256
 $UNCENSORED_PROJECTOR_FILE|$UNCENSORED_PROJECTOR_URL|$UNCENSORED_PROJECTOR_SHA256
-$EXPERIMENTAL_MODEL_FILE|$EXPERIMENTAL_MODEL_URL|$EXPERIMENTAL_MODEL_SHA256
 $VISION_PROJECTOR_FILE|$VISION_PROJECTOR_URL|$VISION_PROJECTOR_SHA256
 EOF
 }