提交 b5e37224 编写于 作者: W wangguibao

Elastic CTR Serving code

上级 fe4115a2
...@@ -37,7 +37,7 @@ using baidu::paddle_serving::predictor::elastic_ctr::Request; ...@@ -37,7 +37,7 @@ using baidu::paddle_serving::predictor::elastic_ctr::Request;
const int VARIABLE_NAME_LEN = 256; const int VARIABLE_NAME_LEN = 256;
const int CTR_PREDICTION_DENSE_DIM = 13; const int CTR_PREDICTION_DENSE_DIM = 13;
const int CTR_PREDICTION_EMBEDDING_SIZE = 10; const int CTR_PREDICTION_EMBEDDING_SIZE = 9;
bthread::Mutex ElasticCTRPredictionOp::mutex_; bthread::Mutex ElasticCTRPredictionOp::mutex_;
int64_t ElasticCTRPredictionOp::cube_time_us_ = 0; int64_t ElasticCTRPredictionOp::cube_time_us_ = 0;
...@@ -143,7 +143,7 @@ int ElasticCTRPredictionOp::inference() { ...@@ -143,7 +143,7 @@ int ElasticCTRPredictionOp::inference() {
} }
} }
#if 0 #if 1
rec::mcube::CubeAPI *cube = rec::mcube::CubeAPI::instance(); rec::mcube::CubeAPI *cube = rec::mcube::CubeAPI::instance();
predictor::KVManager &kv_manager = predictor::KVManager::instance(); predictor::KVManager &kv_manager = predictor::KVManager::instance();
const predictor::KVInfo *kvinfo = const predictor::KVInfo *kvinfo =
...@@ -269,7 +269,7 @@ int ElasticCTRPredictionOp::inference() { ...@@ -269,7 +269,7 @@ int ElasticCTRPredictionOp::inference() {
instance.slots(i).slot_name().c_str()); instance.slots(i).slot_name().c_str());
lod_tensor.name = std::string(name); lod_tensor.name = std::string(name);
lod_tensors[i].dtype = paddle::PaddleDType::INT64; lod_tensors[i].dtype = paddle::PaddleDType::FLOAT32;
std::vector<std::vector<size_t>> &lod = lod_tensors[i].lod; std::vector<std::vector<size_t>> &lod = lod_tensors[i].lod;
lod.resize(1); lod.resize(1);
lod[0].push_back(0); lod[0].push_back(0);
...@@ -296,10 +296,15 @@ int ElasticCTRPredictionOp::inference() { ...@@ -296,10 +296,15 @@ int ElasticCTRPredictionOp::inference() {
int idx = base + j; int idx = base + j;
if (values[idx].buff.size() != if (values[idx].buff.size() !=
sizeof(float) * CTR_PREDICTION_EMBEDDING_SIZE) { sizeof(float) * CTR_PREDICTION_EMBEDDING_SIZE) {
#if 0
LOG(ERROR) << "Embedding vector size not expected"; LOG(ERROR) << "Embedding vector size not expected";
fill_response_with_message( fill_response_with_message(
res, -1, "Embedding vector size not expected"); res, -1, "Embedding vector size not expected");
return 0; return 0;
#else
// sizeof(float) * CTR_PREDICTION_EMBEDDING_SIZE = 36
values[idx].buff.append(36, '0');
#endif
} }
memcpy(data_ptr, values[idx].buff.data(), values[idx].buff.size()); memcpy(data_ptr, values[idx].buff.data(), values[idx].buff.size());
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册