Use quantization-specific activation init in Phi3 models

Author: orionpapadakis

src/main/java/org/beehive/gpullama3/inference/state/Phi3State.java

@@ -80,7 +80,11 @@ protected StateFields createStateFields(Configuration config) {
         fields.valueCache = Stream.generate(() -> ArrayFloatTensor.allocate(contextLength, kvDim)).limit(nLayers).toArray(FloatTensor[]::new);
 
         // TornadoVM wrapper arrays for GPU acceleration
-        fields.embeddingX = new HalfFloatArray(config.dim());
+        switch (config.modelType()) {
+            case "FP16" -> fields.createActivationFP16(config.dim());
+            case "Q8_0" -> fields.createActivationQ8_0(config.dim());
+            default -> throw new UnsupportedOperationException("Quantization format " + config.modelType());
+        }
         fields.wrapX = new FloatArray(dim);
         fields.wrapXb = new FloatArray(dim);
         fields.wrapXb2 = new FloatArray(dim);