Qualcomm AI Engine Direct - Replace private variable with function (#8724)

Summary:
- To improve readability, replace is_bert_ with is_bert()

Author: shewu-quic

Diff for: examples/qualcomm/oss_scripts/llama/runner/io_manager.cpp

@@ -81,8 +81,7 @@ ShiftPointerIoMgr::ShiftPointerIoMgr(
       eval_mode_(eval_mode),
       prefill_forward_name_(prefill_forward_name),
       kv_forward_name_(kv_forward_name),
-      use_int64_token_(use_int64_token),
-      is_bert_(prefill_cache_len_ == 0) {
+      use_int64_token_(use_int64_token) {
   if (!prefill_forward_name_.empty()) {
     input_tensors_[prefill_forward_name_] =
         std::vector<std::vector<executorch::aten::TensorImpl*>>(modules.size());
@@ -391,7 +390,7 @@ void ShiftPointerIoMgr::prepare_prefill_io(
   input_tensors_[prefill_forward_name_][0].push_back(
       prefill_attention_mask_.get());
 
-  if (!is_bert_) {
+  if (!is_bert()) {
     // [I]: prefill_input_pos
     Result<TensorInfo> prefill_input_pos =
         methods_meta[0]->input_tensor_meta(2);
@@ -544,7 +543,7 @@ void ShiftPointerIoMgr::update_prefill_to_kv_io(
   size_t copied_size = pos * sizeof(uint8_t);
   for (int i = 0; i < k_cache_in.size(); ++i) {
     uint8_t* ptr_in = k_cache_in[i]->mutable_data<uint8_t>();
-    if (is_bert_) {
+    if (is_bert()) {
       const uint8_t* ptr_out = k_cache_out[i]->data<uint8_t>();
       for (size_t j = 0, offset = kv_cache_len_; j < head_dim_;
            ++j, offset += kv_cache_len_) {
@@ -620,7 +619,7 @@ void ShiftPointerIoMgr::update_prefill_io(
   (void)cur_token;
   (void)output_tensors;
 
-  if (!is_bert_) {
+  if (!is_bert()) {
     // update v_cache
     auto& v_cache_in = v_cache_in_[prefill_forward_name_];
     auto& v_cache_out = v_cache_out_[prefill_forward_name_];
@@ -668,7 +667,7 @@ void ShiftPointerIoMgr::fill_prefill_toks(
     std::vector<uint64_t>& prompt_tokens) {
   IO* ptr = static_cast<IO*>(get_mutable_ptr());
   for (int i = 0; i < prefill_ar_len_; i++) {
-    if (!is_bert_) {
+    if (!is_bert()) {
       ptr->prefill_input_pos[i] = start_pos + i;
     }
 
@@ -735,8 +734,7 @@ SmartMaskIoMgr::SmartMaskIoMgr(
       eval_mode_(eval_mode),
       prefill_forward_name_(prefill_forward_name),
       kv_forward_name_(kv_forward_name),
-      use_int64_token_(use_int64_token),
-      is_bert_(prefill_cache_len == 0) {
+      use_int64_token_(use_int64_token) {
   if (!prefill_forward_name_.empty()) {
     input_tensors_[prefill_forward_name_] =
         std::vector<std::vector<executorch::aten::TensorImpl*>>(modules.size());
@@ -1196,7 +1194,7 @@ void SmartMaskIoMgr::prepare_prefill_io(
       executorch::aten::ScalarType::Bits16,
       prefill_attention_mask.get());
 
-  if (!is_bert_) {
+  if (!is_bert()) {
     // [I]: prefill_input_pos
     Result<TensorInfo> prefill_input_pos =
         methods_meta[0]->input_tensor_meta(2);
@@ -1319,7 +1317,7 @@ void SmartMaskIoMgr::update_prefill_to_kv_io(
     ptr->kv_attention_mask[i] = 65535;
   }
 
-  if (is_bert_) {
+  if (is_bert()) {
     // update v_cache
     auto& v_cache_in = v_cache_in_[kv_forward_name_];
     auto& v_cache_out = v_cache_out_[prefill_forward_name_];
@@ -1366,7 +1364,7 @@ void SmartMaskIoMgr::update_prefill_io(
     std::vector<std::vector<Tensor>>& output_tensors) {
   (void)output_tensors;
 
-  if (!is_bert_) {
+  if (!is_bert()) {
     // update v_cache
     auto& v_cache_in = v_cache_in_[prefill_forward_name_];
     auto& v_cache_out = v_cache_out_[prefill_forward_name_];
@@ -1400,7 +1398,7 @@ void SmartMaskIoMgr::fill_prefill_toks(
     std::vector<uint64_t>& prompt_tokens) {
   IO* ptr = static_cast<IO*>(get_mutable_ptr());
   for (int i = 0; i < prefill_ar_len_; i++) {
-    if (!is_bert_) {
+    if (!is_bert()) {
       ptr->prefill_input_pos[i] = start_pos + i;
     }
 

Diff for: examples/qualcomm/oss_scripts/llama/runner/io_manager.h

@@ -151,6 +151,11 @@ class ShiftPointerIoMgr : public IoMgrBase {
   };
 
  private:
+  // If the cache length is zero, it indicates a BERT model, which does not use
+  // position ids or KV cache inputs.
+  bool is_bert() const {
+    return prefill_cache_len_ == 0;
+  }
   std::unique_ptr<executorch::aten::TensorImpl> kv_input_toks_;
   std::unique_ptr<executorch::aten::TensorImpl> kv_input_pos_;
   std::unique_ptr<executorch::aten::TensorImpl> kv_attention_mask_;
@@ -189,7 +194,6 @@ class ShiftPointerIoMgr : public IoMgrBase {
   std::string prefill_forward_name_;
   std::string kv_forward_name_;
   const bool use_int64_token_{false};
-  const bool is_bert_{false};
 };
 
 class SmartMaskIoMgr : public IoMgrBase {
@@ -288,6 +292,11 @@ class SmartMaskIoMgr : public IoMgrBase {
   };
 
  private:
+  // If the cache length is zero, it indicates a BERT model, which does not use
+  // position ids or KV cache inputs.
+  bool is_bert() const {
+    return prefill_cache_len_ == 0;
+  }
   std::unique_ptr<executorch::aten::TensorImpl> kv_input_toks_;
   std::unique_ptr<executorch::aten::TensorImpl> kv_input_pos_;
   std::unique_ptr<executorch::aten::TensorImpl> kv_attention_mask_;
@@ -326,9 +335,6 @@ class SmartMaskIoMgr : public IoMgrBase {
   std::string prefill_forward_name_;
   std::string kv_forward_name_;
   const bool use_int64_token_{false};
-  // If the cache length is zero, it indicates a BERT model, which does not use
-  // position ids or KV cache inputs.
-  const bool is_bert_{false};
 };
 
 } // namespace example