add train multiple negative and infer (#1422)

* fix readme2.0

* add tagspace infer

fluid/PaddleRec/gru4rec/README.md

@@ -21,7 +21,7 @@ GRU4REC模型的介绍可以参阅论文[Session-based Recommendations with Recu
 
 论文的贡献在于首次将RNN（GRU）运用于session-based推荐，相比传统的KNN和矩阵分解，效果有明显的提升。
 
-论文的核心思想史在一个session中，用户点击一系列item的行为看做一个序列，用来训练RNN模型。预测阶段，给定已知的点击序列作为输入，预测下一个可能点击的item。
+论文的核心思想是在一个session中，用户点击一系列item的行为看做一个序列，用来训练RNN模型。预测阶段，给定已知的点击序列作为输入，预测下一个可能点击的item。
 
 session-based推荐应用场景非常广泛，比如用户的商品浏览、新闻点击、地点签到等序列数据。
 

fluid/PaddleRec/TagSpace/README.md → fluid/PaddleRec/tagspace/README.md

@@ -6,6 +6,7 @@
 .
 ├── README.md            # 文档
 ├── train.py             # 训练脚本
+├── infer.py             # 预测脚本
 ├── utils                # 通用函数
 ├── small_train.txt      # 小样本训练集
 └── small_test.txt       # 小样本测试集
@@ -26,7 +27,6 @@ TagSpace模型的介绍可以参阅论文[#TagSpace: Semantic Embeddings from Ha
 "3","Wall St. Bears Claw Back Into the Black (Reuters)","Reuters - Short-sellers, Wall Street's dwindling\band of ultra-cynics, are seeing green again."
 ```
 
-
 ## 训练
 '--use_cuda 1' 表示使用gpu, 缺省表示使用cpu 
 
@@ -41,10 +41,9 @@ CPU 环境
 python train.py small_train.txt small_test.txt
 ```
 
-## 未来工作
-
-添加预测部分
-
-添加多种负例采样方式
+## 预测
 
+```
+CUDA_VISIBLE_DEVICES=0 python infer.py model/ 1 10 small_train.txt small_test.txt --use_cuda 1
+```
 

fluid/PaddleRec/tagspace/infer.py

+import sys
+import time
+import math
+import unittest
+import contextlib
+import numpy as np
+import six
+import paddle.fluid as fluid
+import paddle
+import utils
+
+def infer(test_reader, vocab_tag, use_cuda, model_path):
+    """ inference function """
+    place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
+    exe = fluid.Executor(place)
+
+    with fluid.scope_guard(fluid.core.Scope()):
+        infer_program, feed_target_names, fetch_vars = fluid.io.load_inference_model(
+            model_path, exe)
+        t0 = time.time()
+        step_id = 0
+        true_num = 0 
+        all_num = 0
+        size = len(vocab_tag)
+        value = []
+        for data in test_reader():
+            step_id += 1
+            lod_text_seq = utils.to_lodtensor([dat[0] for dat in data], place)
+            lod_tag = utils.to_lodtensor([dat[1] for dat in data], place)
+            lod_pos_tag = utils.to_lodtensor([dat[2] for dat in data], place)
+            para = exe.run(
+                infer_program,
+                feed={
+                    "text": lod_text_seq,
+                    "pos_tag": lod_tag},
+                fetch_list=fetch_vars,
+                return_numpy=False)
+            value.append(para[0]._get_float_element(0))
+            if step_id % size == 0 and step_id > 1:
+                all_num += 1
+                true_pos = [dat[2] for dat in data][0][0]
+                if value.index(max(value)) == int(true_pos):
+                    true_num += 1
+                value = []
+            if step_id % 1000 == 0:
+                print(step_id, 1.0 * true_num / all_num)
+        t1 = time.time()
+
+if __name__ == "__main__":
+    if len(sys.argv) != 6:
+        print(
+            "Usage: %s model_dir start_epoch last_epoch(inclusive) train_file test_file"
+        )
+        exit(0)
+    train_file = ""
+    test_file = ""
+    model_dir = sys.argv[1]
+    try:
+        start_index = int(sys.argv[2])
+        last_index = int(sys.argv[3])
+        train_file = sys.argv[4]
+        test_file = sys.argv[5]
+    except:
+        print(
+            "Usage: %s model_dir start_ipoch last_epoch(inclusive) train_file test_file"
+        )
+        exit(-1)
+    vocab_text, vocab_tag, train_reader, test_reader = utils.prepare_data(
+        train_file,
+        test_file,
+        batch_size=1,
+        buffer_size=1000,
+        word_freq_threshold=0)
+
+    for epoch in xrange(start_index, last_index + 1):
+        epoch_path = model_dir + "/epoch_" + str(epoch)
+        infer(test_reader=test_reader, vocab_tag=vocab_tag, use_cuda=False, model_path=epoch_path)

fluid/PaddleRec/TagSpace/train.py → fluid/PaddleRec/tagspace/train.py

@@ -24,7 +24,7 @@ def parse_args():
     args = parser.parse_args()
     return args
 
-def network(vocab_text_size, vocab_tag_size, emb_dim=10, hid_dim=1000, win_size=5, margin=0.1):
+def network(vocab_text_size, vocab_tag_size, emb_dim=10, hid_dim=1000, win_size=5, margin=0.1, neg_size=5):
     """ network definition """
     text = io.data(name="text", shape=[1], lod_level=1, dtype='int64')
     pos_tag = io.data(name="pos_tag", shape=[1], lod_level=1, dtype='int64')
@@ -44,12 +44,14 @@ def network(vocab_text_size, vocab_tag_size, emb_dim=10, hid_dim=1000, win_size=
             act="tanh",
             pool_type="max",
             param_attr="cnn")
-    
     text_hid = fluid.layers.fc(input=conv_1d, size=emb_dim, param_attr="text_hid")
-
     cos_pos = nn.cos_sim(pos_tag_emb, text_hid)
-    cos_neg = nn.cos_sim(neg_tag_emb, text_hid)
-    
+    mul_text_hid = fluid.layers.sequence_expand_as(x=text_hid, y=neg_tag_emb)
+    mul_cos_neg = nn.cos_sim(neg_tag_emb, mul_text_hid)
+    cos_neg_all = fluid.layers.sequence_reshape(input=mul_cos_neg, new_dim=neg_size)
+    #choose max negtive cosine
+    cos_neg = nn.reduce_max(cos_neg_all, dim=1, keep_dim=True)
+    #calculate hinge loss 
     loss_part1 = nn.elementwise_sub(
             tensor.fill_constant_batch_size_like(
                 input=cos_pos,
@@ -63,22 +65,20 @@ def network(vocab_text_size, vocab_tag_size, emb_dim=10, hid_dim=1000, win_size=
                 input=loss_part2, shape=[-1, 1], value=0.0, dtype='float32'),
             loss_part2)
     avg_cost = nn.mean(loss_part3)
-
     less = tensor.cast(cf.less_than(cos_neg, cos_pos), dtype='float32')
     correct = nn.reduce_sum(less)
     return text, pos_tag, neg_tag, avg_cost, correct, cos_pos
 
-def train(train_reader, vocab_text, vocab_tag, base_lr, batch_size,
+def train(train_reader, vocab_text, vocab_tag, base_lr, batch_size, neg_size,
           pass_num, use_cuda, model_dir):
     """ train network """
-
     args = parse_args()
     vocab_text_size = len(vocab_text)
     vocab_tag_size = len(vocab_tag)
 
     place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
     # Train program
-    text, pos_tag, neg_tag, avg_cost, correct, pos_cos = network(vocab_text_size, vocab_tag_size)
+    text, pos_tag, neg_tag, avg_cost, correct, cos_pos = network(vocab_text_size, vocab_tag_size, neg_size=neg_size)
 
     # Optimization to minimize lost
     sgd_optimizer = fluid.optimizer.Adagrad(learning_rate=base_lr)
@@ -117,8 +117,8 @@ def train(train_reader, vocab_text, vocab_tag, base_lr, batch_size,
               (epoch_idx, batch_id, total_time / epoch_idx))
         save_dir = "%s/epoch_%d" % (model_dir, epoch_idx)
         feed_var_names = ["text", "pos_tag"]
-        fetch_vars = [pos_cos]
-        fluid.io.save_inference_model(save_dir ,feed_var_names, fetch_vars, exe)
+        fetch_vars = [cos_pos]
+        fluid.io.save_inference_model(save_dir, feed_var_names, fetch_vars, exe)
     print("finish training")
 
 def train_net():
@@ -128,17 +128,19 @@ def train_net():
     test_file = args.test_file
     use_cuda = True if args.use_cuda else False
     batch_size = 100
+    neg_size = 3
     vocab_text, vocab_tag, train_reader, test_reader = utils.prepare_data(
-        train_file, test_file, batch_size=batch_size, buffer_size=batch_size*100, word_freq_threshold=0)
+        train_file, test_file, neg_size=neg_size, batch_size=batch_size, buffer_size=batch_size*100, word_freq_threshold=0)
     train(
         train_reader=train_reader,
         vocab_text=vocab_text,
         vocab_tag=vocab_tag,
         base_lr=0.01,
         batch_size=batch_size,
+        neg_size=neg_size,
         pass_num=10,
         use_cuda=use_cuda,
-        model_dir="model_dim10_2")
+        model_dir="model")
 
 
 if __name__ == "__main__":

fluid/PaddleRec/TagSpace/utils.py → fluid/PaddleRec/tagspace/utils.py

@@ -38,12 +38,13 @@ def prepare_data(train_filename,
     train_reader = sort_batch(
         paddle.reader.shuffle(
             train(
-                train_filename, vocab_text, vocab_tag, buffer_size, data_type=DataType.SEQ),
+                train_filename, vocab_text, vocab_tag, neg_size,
+                buffer_size, data_type=DataType.SEQ),
             buf_size=buffer_size),
         batch_size, batch_size * 20)
     test_reader = sort_batch(
         test(
-            test_filename, vocab_text, vocab_tag, buffer_size, data_type=DataType.SEQ),
+            test_filename, vocab_text, vocab_tag, neg_size, buffer_size, data_type=DataType.SEQ),
         batch_size, batch_size * 20)
     return vocab_text, vocab_tag, train_reader, test_reader
 
@@ -123,7 +124,7 @@ def build_dict(column_num=2, min_word_freq=50, train_filename="", test_filename=
     word_idx = dict(list(zip(words, six.moves.range(len(words)))))
     return word_idx
 
-def reader_creator(filename, text_idx, tag_idx, n, data_type):
+def train_reader_creator(filename, text_idx, tag_idx, neg_size, n, data_type):
     def reader():
         with open(filename) as input_file:
             data_file = csv.reader(input_file)
@@ -138,7 +139,7 @@ def reader_creator(filename, text_idx, tag_idx, n, data_type):
                 max_iter = 100
                 now_iter = 0
                 sum_n = 0
-                while(sum_n < 1) :
+                while(sum_n < neg_size) :
                     now_iter += 1
                     if now_iter > max_iter:
                         print("error : only one class")
@@ -152,8 +153,26 @@ def reader_creator(filename, text_idx, tag_idx, n, data_type):
                 yield text, pos_tag, neg_tag
     return reader
 
-def train(filename, text_idx, tag_idx, n, data_type=DataType.SEQ):
-    return reader_creator(filename, text_idx, tag_idx, n, data_type)
+def test_reader_creator(filename, text_idx, tag_idx, n, data_type):
+    def reader():
+        with open(filename) as input_file:
+            data_file = csv.reader(input_file)
+            for row in data_file:
+                text_raw = re.split(r'\W+', row[2].strip())
+                text = [text_idx.get(w) for w in text_raw]
+                tag_raw = re.split(r'\W+', row[0].strip())
+                pos_index = tag_idx.get(tag_raw[0])
+                pos_tag = []
+                pos_tag.append(pos_index)
+                for ii in range(len(tag_idx)):
+                    tag = []
+                    tag.append(ii)
+                    yield text, tag, pos_tag
+    return reader
+
+
+def train(filename, text_idx, tag_idx, neg_size, n, data_type=DataType.SEQ):
+    return train_reader_creator(filename, text_idx, tag_idx, neg_size, n, data_type)
 
-def test(filename, text_idx, tag_idx, n, data_type=DataType.SEQ):
-    return reader_creator(filename, text_idx, tag_idx, n, data_type)
+def test(filename, text_idx, tag_idx, neg_size, n, data_type=DataType.SEQ):
+    return test_reader_creator(filename, text_idx, tag_idx, n, data_type)