Replace compression sidecar with Hermes micro-compaction

This commit is contained in:
Mikei386
2026-08-26 06:18:51 +02:00
parent 438bd34b72
commit 896c12193e
8 changed files with 23 additions and 140 deletions
+1 -9
View File
@@ -43,8 +43,7 @@ required=(AI_HOSTNAME ADMIN_USER MODEL_DIR FAST_MODEL_FILE
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
COMPRESSION_MODEL_FILE COMPRESSION_MODEL_URL COMPRESSION_MODEL_SHA256)
VISION_PROJECTOR_FILE VISION_PROJECTOR_URL VISION_PROJECTOR_SHA256)
for name in "${required[@]}"; do
[[ -n "${!name:-}" ]] || die "Pflichtwert $name fehlt."
done
@@ -340,9 +339,6 @@ UNCENSORED_MODEL_FILE=$UNCENSORED_MODEL_FILE
UNCENSORED_PROJECTOR_FILE=$UNCENSORED_PROJECTOR_FILE
EXPERIMENTAL_MODEL_FILE=$EXPERIMENTAL_MODEL_FILE
VISION_PROJECTOR_FILE=$VISION_PROJECTOR_FILE
COMPRESSION_MODEL_FILE=$COMPRESSION_MODEL_FILE
COMPRESSION_CONTEXT=${COMPRESSION_CONTEXT:-65536}
COMPRESSION_GPU_DEVICE=${COMPRESSION_GPU_DEVICE:-GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b}
FAST_CONTEXT=${FAST_CONTEXT:-76800}
MEDIUM_CONTEXT=${MEDIUM_CONTEXT:-160000}
LARGE_CONTEXT=${LARGE_CONTEXT:-192000}
@@ -404,7 +400,6 @@ $UNCENSORED_MODEL_FILE|$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
$COMPRESSION_MODEL_FILE|$COMPRESSION_MODEL_URL|$COMPRESSION_MODEL_SHA256
EOF
}
@@ -500,9 +495,6 @@ build_and_start() {
docker compose --env-file "$SECRETS_DIR/stack.env" --profile image create flux-worker
docker compose --env-file "$SECRETS_DIR/stack.env" up -d --build \
xtts piper tts-gateway profile-controller router open-webui hermes
docker compose --env-file "$SECRETS_DIR/stack.env" --profile compression up -d \
llama-compression
if [[ ${WIREGUARD_MODE:-container} == container ]]; then
systemctl restart mike-ai-container-vpn-guard.service
fi