clean up pretty_print

Author: guocuimi

src/layers/embedding.h

@@ -33,10 +33,6 @@ class EmbeddingImpl : public Module {
     return F::embedding(input, weight_);
   }
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " " << weight_.sizes() << " " << weight_.device();
-  }
-
   // return the weight (for testing)
   torch::Tensor weight() const { return weight_; }
 
@@ -84,10 +80,6 @@ class ParallelEmbeddingImpl : public Module {
     return output;
   }
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " " << weight_.sizes() << " " << weight_.device();
-  }
-
   // return the weight (for testing)
   torch::Tensor weight() const { return weight_; }
 
@@ -147,10 +139,6 @@ class VocabParallelEmbeddingImpl : public Module {
     return reduce_from_model_parallel_region(output, parallel_args_);
   }
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " " << weight_.sizes() << " " << weight_.device();
-  }
-
   // return the weight (for testing)
   torch::Tensor weight() const { return weight_; }
 

src/layers/linear_impl.h

@@ -42,10 +42,6 @@ class ColumnParallelLinearImpl : public ParallelLinearImpl {
         << "bias is not loaded for " << prefix + "bias";
   }
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " " << weight_.sizes() << " " << weight_.device();
-  }
-
   // return the weight (for testing)
   torch::Tensor weight() const { return weight_; }
 
@@ -95,10 +91,6 @@ class RowParallelLinearImpl : public ParallelLinearImpl {
         << "bias is not loaded for " << prefix + "bias";
   }
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " " << weight_.sizes() << " " << weight_.device();
-  }
-
   // return the weight (for testing)
   torch::Tensor weight() const { return weight_; }
 

src/layers/normalization.h

@@ -94,10 +94,6 @@ class LayerNormImpl : public Module {
         input, normalized_shape_, weight_, bias_, eps_);
   }
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " " << weight_.sizes() << " " << weight_.device();
-  }
-
  private:
   // parameter members, must be registered
   torch::Tensor weight_{nullptr};
@@ -131,10 +127,6 @@ class RMSNormImpl : public Module {
     return detail::rms_norm(input, weight_, eps_);
   }
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " " << weight_.sizes() << " " << weight_.device();
-  }
-
  private:
   // parameter members, must be registered
   torch::Tensor weight_{nullptr};
@@ -163,10 +155,6 @@ class GemmaRMSNormImpl : public Module {
     return detail::gemma_rms_norm(input, weight_, eps_);
   }
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " " << weight_.sizes() << " " << weight_.device();
-  }
-
  private:
   // parameter members, must be registered
   torch::Tensor weight_{nullptr};
@@ -208,10 +196,6 @@ class RMSNormResidualImpl : public Module {
     return detail::rms_norm(input, weight_, eps_);
   }
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " " << weight_.sizes() << " " << weight_.device();
-  }
-
  private:
   // parameter members, must be registered
   torch::Tensor weight_{nullptr};

src/quantization/qlinear_gptq_marlin_impl.h

@@ -35,11 +35,6 @@ class ColumnParallelQLinearGPTQMarlinImpl : public ParallelLinearImpl {
   void load_state_dict(const StateDict& state_dict,
                        const std::vector<std::string>& prefixes) override;
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " qweight=" << qweight_.sizes()
-           << " scales=" << scales_.sizes() << " device=" << qweight_.device();
-  }
-
  private:
   // parameter members, must be registered
   DEFINE_FUSED_WEIGHT(qweight);
@@ -91,11 +86,6 @@ class RowParallelQLinearGPTQMarlinImpl : public ParallelLinearImpl {
   // whether the weight is loaded
   void verify_loaded_weights(const std::string& prefix = "") const override;
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " qweight=" << qweight_.sizes()
-           << " scales=" << scales_.sizes() << " device=" << qweight_.device();
-  }
-
  private:
   // parameter members, must be registered
   DEFINE_WEIGHT(qweight);

src/quantization/qlinear_impl.h

@@ -76,12 +76,6 @@ class ColumnParallelQLinearImpl : public ParallelLinearImpl {
   void load_state_dict(const StateDict& state_dict,
                        const std::vector<std::string>& prefixes) override;
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " qweight=" << qweight_.sizes()
-           << " qzeros=" << qzeros_.sizes() << " scales=" << scales_.sizes()
-           << " device=" << qweight_.device();
-  }
-
  private:
   // parameter members, must be registered
   DEFINE_FUSED_WEIGHT(qweight);
@@ -148,12 +142,6 @@ class RowParallelQLinearImpl : public ParallelLinearImpl {
   // whether the weight is loaded
   void verify_loaded_weights(const std::string& prefix = "") const override;
 
-  void pretty_print(std::ostream& stream) const override {
-    stream << name() << " qweight=" << qweight_.sizes()
-           << " qzeros=" << qzeros_.sizes() << " scales=" << scales_.sizes()
-           << " device=" << qweight_.device();
-  }
-
  private:
   // parameter members, must be registered
   DEFINE_WEIGHT(qweight);