diff --git a/official/recommend/naml/MINDlarge_config_cpu.yaml b/official/recommend/naml/MINDlarge_config_cpu.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..377e55e6921974200667204cf152a9d2bb81941c
--- /dev/null
+++ b/official/recommend/naml/MINDlarge_config_cpu.yaml
@@ -0,0 +1,107 @@
+# Builtin Configurations(DO NOT CHANGE THESE CONFIGURATIONS unless you know exactly what you are doing)
+enable_modelarts: False
+# Url for modelarts
+data_url: ""
+train_url: ""
+checkpoint_url: ""
+# Path for local
+data_path: "/cache/data"
+output_path: "/cache/train"
+load_path: "/cache/checkpoint_path"
+device_target: "CPU"
+need_modelarts_dataset_unzip: False
+modelarts_dataset_unzip_name: ""
+
+# ==============================================================================
+# options
+n_categories: 19
+n_sub_categories: 286
+n_words: 74308
+default_epochs: 1
+epochs: -1
+lr: 0.001
+print_times: 1000
+embedding_file: "MINDlarge_utils/embedding_all.npy"
+word_dict_path: "MINDlarge_utils/word_dict_all.pkl"
+category_dict_path: "MINDlarge_utils/vert_dict.pkl"
+subcategory_dict_path: "MINDlarge_utils/subvert_dict.pkl"
+uid2index_path: "MINDlarge_utils/uid2index.pkl"
+train_dataset_path: "MINDlarge_train"
+eval_dataset_path: "MINDlarge_dev"
+
+# default option
+seed: 1
+platform: "CPU"
+save_graphs: False
+dataset: "large"
+dataset_path: ""
+n_browsed_news: 50
+n_words_title: 16
+n_words_abstract: 48
+word_embedding_dim: 304
+category_embedding_dim: 112
+query_vector_dim: 208
+n_filters: 400
+window_size: 3
+checkpoint_path: "" # change to naml_large_new.ckpt path when export
+batch_size: 64 # change to 16 when export or infer
+
+# train option
+beta1: 0.9
+beta2: 0.999
+epsilon: 0.00000001 # 1e-8
+neg_sample: 4 #when training, neg_sample=4, when test, neg_sample=-1
+mixed: True
+sink_mode: True
+weight_decay: True
+save_checkpoint: True
+save_checkpoint_path: "./checkpoint"
+dropout_ratio: 0.2
+
+# eval option
+eval_neg_sample: -1
+
+# export option
+export_file_dir: "./"
+file_format: "MINDIR"
+export_neg_sample: -1
+
+# infer option
+preprocess_path: "./"
+result_path: "./"
+label_path: "./"
+
+---
+
+# Help description for each configuration
+# default option
+seed: "random seed"
+platform: "run platform, only support Ascend"
+save_graphs: "whether save graphs, default is False."
+dataset: "MIND dataset, support large, small and demo."
+dataset_path: "MIND dataset path."
+n_browsed_news: "number of browsed news per user"
+n_words_title: "number of words per title"
+n_words_abstract: "number of words per abstract"
+word_embedding_dim: "dimension of word embedding vector"
+category_embedding_dim: "dimension of category embedding vector"
+query_vector_dim: "dimension of the query vector in attention"
+n_filters: "number of filters in CNN"
+window_size: "size of filter in CNN"
+checkpoint_path: "Pre trained checkpoint path, default is None."
+batch_size: "size of each batch"
+
+# train option
+beta1: "ADAM beta1"
+beta2: "ADAM beta2"
+epsilon: "ADAM epsilon for numerical stability"
+neg_sample: "number of negative samples in negative sampling"
+mixed: "whether use mixed precision, default is True."
+sink_mode: "whether use dataset sink, default is True."
+weight_decay: "whether use weight decay, default is True."
+save_checkpoint: "whether save checkpoint, default is True."
+save_checkpoint_path: "Save checkpoint path, default is checkpoint."
+dropout_ratio: "ratio of dropout"
+
+# export option
+file_format: "choices in ['AIR', 'ONNX', 'MINDIR']"
diff --git a/official/recommend/naml/README.md b/official/recommend/naml/README.md
index 604c249e7c168248ddf160d90e62f12678a03535..cf71da9d83e645b511b2f9c66cbbbb26fc8736c5 100644
--- a/official/recommend/naml/README.md
+++ b/official/recommend/naml/README.md
@@ -74,6 +74,7 @@ You can download the dataset and put the directory in structure as follows:
鈹溾攢鈹€ MINDdemo_config.yaml # Configurations for demo
鈹溾攢鈹€ MINDlarge_config.yaml # Configurations for large
鈹溾攢鈹€ MINDsmall_config.yaml # Configurations for small
+ 鈹溾攢鈹€ MINDslarge_config_cpu.yaml # Configurations for large on cpu
鈹溾攢鈹€ ascend310_infer # application for 310 inference
鈹溾攢鈹€ train.py # training script
鈹溾攢鈹€ eval.py # evaluation script
@@ -112,6 +113,36 @@ You can start training using python or shell scripts. The usage of shell scripts
<https://gitee.com/mindspore/models/tree/master/utils/hccl_tools>.
+- running on CPU
+
+ ```shell
+ # train using python
+ python train.py --config_path=[CONFIG_PATH]
+ --platform=[PLATFORM]
+ --dataset=[DATASET]
+ --dataset_path=[DATASET_PATH]
+ --save_checkpoint_path=[SAVE_CHECKPOINT_PATH]
+ --weight_decay=False
+ --sink_mode=False
+ # example
+ python train.py --config_path=MINDlarge_config_cpu.yaml --platform=CPU --dataset=large --dataset_path=MINDlarge --save_checkpoint_path=./script/checkpoint --weight_decay=False --sink_mode=False
+ # evaluation using python
+ python eval.py --config_path=[CONFIG_PATH]
+ --platform=[PLATFORM]
+ --dataset=[DATASET]
+ --dataset_path=[DATASET_PATH]
+ --checkpoint_path=[CHECKPOINT_PATH]
+ # example
+ python eval.py --config_path=MINDlarge_config_cpu.yaml --platform=CPU --dataset=large --dataset_path=MINDlarge --checkpoint_path=./script/checkpoint/naml_last.ckpt
+ ```
+
+ - `PLATFORM` should be CPU.
+ - `DEVICE_ID` is the device id you want to run the network.
+ - `DATASET` MIND dataset, support large.
+ - `DATASET_PATH` is the dataset path, the structure as [Dataset](#dataset).
+ - `SAVE_CHECKPOINT_PATH` is a pre-trained checkpoint path to save.
+ - `CHECKPOINT_PATH` is a pre-trained checkpoint path.
+
- ModelArts (If you want to run in modelarts, please check the official documentation of [modelarts](https://support.huaweicloud.com/modelarts/), and you can start training as follows)
- Train large dataset 1p/8p on ModelArts
@@ -260,6 +291,19 @@ bash run_infer_310.sh [NEWS_MODEL] [USER_MODEL] [DEVICE_ID]
| outputs | probability |
| Accuracy | AUC: 0.6669 |
+### Inference on CPU Performance
+
+| Parameters | CPU |
+| ----------------- | ------------------------- |
+| Model Version | NAML |
+| Resource | CPU |
+| Uploaded Date | 8/9/2022 (month/day/year) |
+| MindSpore Version | 1.8 |
+| Dataset | MINDlarge |
+| batch_size | 64 |
+| outputs | probability |
+| Accuracy | AUC: 0.6727 |
+
# [Description of Random Situation](#contents)
<!-- In dataset.py, we set the seed inside 鈥渃reate_dataset" function. We also use random seed in train.py. -->
diff --git a/official/recommend/naml/train.py b/official/recommend/naml/train.py
index cdb715033eb5e7a271079a7e0365506c7421720b..dfbac890f7c2caebaa28865fc791c93b774644b1 100644
--- a/official/recommend/naml/train.py
+++ b/official/recommend/naml/train.py
@@ -1,4 +1,4 @@
-# Copyright 2021 Huawei Technologies Co., Ltd
+# Copyright 2021-2022 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
@@ -141,7 +141,10 @@ def run_train():
for _, cell in net_with_loss.cells_and_names():
if isinstance(cell, (nn.Embedding, nn.Softmax, nn.SoftmaxCrossEntropyWithLogits)):
cell.to_float(mstype.float32)
- model = Model(net_with_loss, optimizer=opt, loss_scale_manager=loss_scale_manager)
+ if config.platform == 'CPU':
+ model = Model(net_with_loss, optimizer=opt, loss_scale_manager=None)
+ else:
+ model = Model(net_with_loss, optimizer=opt, loss_scale_manager=loss_scale_manager)
else:
model = Model(net_with_loss, optimizer=opt)