Merge pull request #2 from RUCAIBox/0.2.x

merge from 0.2.x

.github/workflows/python-package.yml

@@ -1,4 +1,4 @@
-name: Python package
+name: RecBole tests
 
 on:
 - push
@@ -22,12 +22,25 @@ jobs:
       run: |
         python -m pip install --upgrade pip
         pip install pytest
+        pip install dgl
         if [ -f requirements.txt ]; then pip install -r requirements.txt; fi
 
     # Use "python -m pytest" instead of "pytest" to fix imports
     - name: Test metrics
       run: |
-        python -m pytest tests/metrics
+        python -m pytest -v tests/metrics
     - name: Test evaluation_setting
       run: |
-        python -m pytest tests/evaluation_setting
+        python -m pytest -v tests/evaluation_setting
+    - name: Test model
+      run: |
+        python -m pytest -v tests/model/test_model_auto.py
+    - name: Test config
+      run: |
+        python -m pytest -v tests/config/test_config.py
+        export PYTHONPATH=.
+        python tests/config/test_command_line.py --use_gpu=False --valid_metric=Recall@10 --split_ratio=[0.7,0.2,0.1] --metrics=['Recall@10'] --epochs=200 --eval_setting='LO_RS' --learning_rate=0.3
+    - name: Test evaluation_setting
+      run: |
+        python -m pytest -v tests/evaluation_setting
+

recbole/data/dataloader/general_dataloader.py

@@ -156,6 +156,7 @@ def _neg_sampling(self, inter_feat):
         return self.sampling_func(uid_field, iid_field, neg_iids, inter_feat)
 
     def _neg_sample_by_pair_wise_sampling(self, uid_field, iid_field, neg_iids, inter_feat):
+        inter_feat = pd.concat([inter_feat] * self.times, ignore_index=True)
         inter_feat.insert(len(inter_feat.columns), self.neg_item_id, neg_iids)
 
         if self.dataset.item_feat is not None:

recbole/data/dataloader/neg_sample_mixin.py

@@ -96,8 +96,6 @@ def __init__(self, config, dataset, sampler, neg_sample_args,
                  batch_size=1, dl_format=InputType.POINTWISE, shuffle=False):
         if neg_sample_args['strategy'] != 'by':
             raise ValueError('neg_sample strategy in GeneralInteractionBasedDataLoader() should be `by`')
-        if dl_format == InputType.PAIRWISE and neg_sample_args['by'] != 1:
-            raise ValueError('Pairwise dataloader can only neg sample by 1')
 
         self.user_inter_in_one_batch = (sampler.phase != 'train') and (config['eval_type'] != EvaluatorType.INDIVIDUAL)
         self.neg_sample_by = neg_sample_args['by']
@@ -109,7 +107,7 @@ def __init__(self, config, dataset, sampler, neg_sample_args,
             self.label_field = config['LABEL_FIELD']
             dataset.set_field_property(self.label_field, FeatureType.FLOAT, FeatureSource.INTERACTION, 1)
         elif dl_format == InputType.PAIRWISE:
-            self.times = 1
+            self.times = self.neg_sample_by
             self.sampling_func = self._neg_sample_by_pair_wise_sampling
 
             neg_prefix = config['NEG_PREFIX']

requirements.txt

@@ -6,4 +6,4 @@ hyperopt>=0.2.4
 pandas>=1.0.5
 tqdm>=4.48.2
 scikit_learn>=0.23.2
-pyyaml>=5.1.0
+pyyaml>=5.1.0

tests/config/test_overall.py

@@ -3,7 +3,10 @@
 # @Author  :   Kaiyuan Li
 # @email   :   tsotfsk@outlook.com
 
-
+# UPDATE:
+# @Time   : 2020/11/17
+# @Author : Xingyu Pan
+# @Email  : panxy@ruc.edu.cn
 import os
 import sys
 import unittest
@@ -119,7 +122,7 @@ def test_split_ratio(self):
         }
 
         self.assertTrue(run_parms({'split_ratio':[  # [0.8, 0.2], 
-        [0.8, 0.1, 0.1], [7, 1, 1, 1], [16, 2, 2]]}))
+        [0.8, 0.1, 0.1],  [16, 2, 2]]}))
 
     def test_leave_one_num(self):
         settings = {