diff --git a/vowpalwabbit/core/src/reductions/cbzo.cc b/vowpalwabbit/core/src/reductions/cbzo.cc
index ace84e507dd..ea0fff5fe33 100644
--- a/vowpalwabbit/core/src/reductions/cbzo.cc
+++ b/vowpalwabbit/core/src/reductions/cbzo.cc
@@ -46,22 +46,22 @@ class linear_update_data
 };
 
 // uint64_t index variant of VW::get_weight
-inline float get_weight(VW::workspace& all, uint64_t index, uint32_t offset)
+inline float get_weight(VW::workspace& all, uint64_t index)
 {
-  return (&all.weights[(index) << all.weights.stride_shift()])[offset];
+  return all.weights[(index) << all.weights.stride_shift()];
 }
 
 // uint64_t index variant of VW::set_weight
-inline void set_weight(VW::workspace& all, uint64_t index, uint32_t offset, float value)
+inline void set_weight(VW::workspace& all, uint64_t index, float value)
 {
-  (&all.weights[(index) << all.weights.stride_shift()])[offset] = value;
+  all.weights[(index) << all.weights.stride_shift()] = value;
 }
 
 float l1_grad(VW::workspace& all, uint64_t fi)
 {
   if (all.no_bias && fi == VW::details::CONSTANT) { return 0.0f; }
 
-  float fw = get_weight(all, fi, 0);
+  float fw = get_weight(all, fi);
   return fw >= 0.0f ? all.l1_lambda : -all.l1_lambda;
 }
 
@@ -69,7 +69,7 @@ float l2_grad(VW::workspace& all, uint64_t fi)
 {
   if (all.no_bias && fi == VW::details::CONSTANT) { return 0.0f; }
 
-  float fw = get_weight(all, fi, 0);
+  float fw = get_weight(all, fi);
   return all.l2_lambda * fw;
 }
 
@@ -77,7 +77,7 @@ inline void accumulate_dotprod(float& dotprod, float x, float& fw) { dotprod +=
 
 inline float constant_inference(VW::workspace& all)
 {
-  float wt = get_weight(all, VW::details::CONSTANT, 0);
+  float wt = get_weight(all, VW::details::CONSTANT);
   return wt;
 }
 
@@ -102,7 +102,7 @@ float inference(VW::workspace& all, VW::example& ec)
 template <bool feature_mask_off>
 void constant_update(cbzo& data, VW::example& ec)
 {
-  float fw = get_weight(*data.all, VW::details::CONSTANT, 0);
+  float fw = get_weight(*data.all, VW::details::CONSTANT);
   if (feature_mask_off || fw != 0.0f)
   {
     float action_centroid = inference<CONSTANT_POLICY>(*data.all, ec);
@@ -110,19 +110,19 @@ void constant_update(cbzo& data, VW::example& ec)
     float update =
         -data.all->eta * (grad + l1_grad(*data.all, VW::details::CONSTANT) + l2_grad(*data.all, VW::details::CONSTANT));
 
-    set_weight(*data.all, VW::details::CONSTANT, 0, fw + update);
+    set_weight(*data.all, VW::details::CONSTANT, fw + update);
   }
 }
 
 template <bool feature_mask_off>
 void linear_per_feature_update(linear_update_data& upd_data, float x, uint64_t fi)
 {
-  float fw = get_weight(*upd_data.all, fi, 0);
+  float fw = get_weight(*upd_data.all, fi);
 
   if (feature_mask_off || fw != 0.0f)
   {
     float update = upd_data.mult * (upd_data.part_grad * x + (l1_grad(*upd_data.all, fi) + l2_grad(*upd_data.all, fi)));
-    set_weight(*upd_data.all, fi, 0, fw + update);
+    set_weight(*upd_data.all, fi, fw + update);
   }
 }
 
@@ -236,7 +236,7 @@ void save_load(cbzo& data, VW::io_buf& model_file, bool read, bool text)
   if (read)
   {
     VW::details::initialize_regressor(all);
-    if (data.all->initial_constant != 0.0f) { set_weight(all, VW::details::CONSTANT, 0, data.all->initial_constant); }
+    if (data.all->initial_constant != 0.0f) { set_weight(all, VW::details::CONSTANT, data.all->initial_constant); }
   }
   if (model_file.num_files() > 0) { save_load_regressor(all, model_file, read, text); }
 }
@@ -368,7 +368,6 @@ std::shared_ptr<VW::LEARNER::learner> VW::reductions::cbzo_setup(VW::setup_base_
 
   auto l = make_bottom_learner(std::move(data), get_learn(all, policy, feature_mask_off), get_predict(all, policy),
       stack_builder.get_setupfn_name(cbzo_setup), prediction_type_t::PDF, label_type_t::CONTINUOUS)
-               .set_params_per_weight(0)
                .set_save_load(save_load)
                .set_output_example_prediction(output_example_prediction_cbzo)
                .set_print_update(print_update_cbzo)