Update entrypoint.sh

entrypoint.sh

@@ -1,7 +1,7 @@
 #!/bin/sh
 set -eu
 
-MODEL_SPEC="${MODEL_SPEC:-google/gemma-4-E2B-it-qat-q4_0-gguf:Q4_0}"
+MODEL_SPEC="${MODEL_SPEC:-unsloth/gemma-4-E2B-it-qat-GGUF:UD-Q4_K_XL}"
 HOST="${HOST:-0.0.0.0}"
 PORT="${PORT:-7860}"
 CTX_SIZE="${CTX_SIZE:-131072}"
@@ -21,3 +21,5 @@ exec /app/llama-server \
   --cache-type-k "$CACHE_TYPE_K" \
   --cache-type-v "$CACHE_TYPE_V" \
   --reasoning "$REASONING_MODE"
+  --spec-type draft-mtp --spec-draft-n-max 4 \
+  -fa on