config_fl.yaml

# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

runner:
  sync_mode: "geo" # 可选, string: sync/async/geo
  #with_coodinator: 1
  geo_step: 100 # 可选, int, 在geo模式下控制本地的迭代次数
  split_file_list: True # 可选, bool, 若每个节点上都拥有全量数据，则需设置为True
  thread_num: 1 # 多线程配置

  # reader类型，分布式下推荐QueueDataset
  reader_type: "QueueDataset" # DataLoader / QueueDataset / RecDataset
  pipe_command: "python queuedataset_reader.py" # QueueDataset 模式下的数据pipe命令
  dataset_debug: False # QueueDataset 模式下 Profiler 开关

  train_data_dir: "../../../datasets/movielens_pinterest_NCF/fl_data/fl_train_data"
  train_reader_path: "movielens_reader"  # importlib format
  train_batch_size: 512
  model_save_path: "output_model_ncf"

  use_gpu: False
  epochs: 2
  print_interval: 50
  
  test_data_dir: "../../../datasets/movielens_pinterest_NCF/fl_data/fl_test_data"
  infer_reader_path: "movielens_reader"  # importlib format
  infer_batch_size: 1
  infer_load_path: "output_model_ncf"
  infer_start_epoch: 2
  infer_end_epoch: 3
  
  need_dump: True
  dump_fields_path: "/home/wangbin/the_one_ps/ziyoujiyi_PaddleRec/PaddleRec/models/recall/ncf"
  dump_fields: ['item_input', 'user_input']
  dump_param: []
  local_sparse: ['embedding_0.w_0']
  remote_sparse: ['embedding_1.w_0']

hyper_parameters:
  optimizer: 
    class: adam
    learning_rate: 0.001
  num_users: 6040
  num_items: 3706
  mf_dim: 8
  mode: "NCF_MLP"  # optional: NCF_NeuMF, NCF_GMF, NCF_MLP
  fc_layers: [64, 32, 16, 8]