Update gemma_utils.cpp

intel · Mar 22, 2024 · bb59f1f · bb59f1f
1 parent 055f840
commit bb59f1f
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/neural_speed/models/gemma/gemma_utils.cpp b/neural_speed/models/gemma/gemma_utils.cpp
@@ -223,7 +223,7 @@ class gemma_quant_layer : public quant_layer_base {
     if ((layername.find("embedding") != std::string::npos) ||
         (layername == "token_embd.weight" || layername == "model.embed_tokens.weight")) {
       // special layer process, can be loaded by config file
-      return quant_params_internal();  // q40
+      return quant_params_internal{quant_bits::q8};  // q80
     }
     quantize &= (ne.size() == 2);
     if (quantize) {