xj-zhu98 / krdn

KRDN

Source code for Knowledge-refined Denoising Network for Robust Recommendation

Environment Requirements

• Ubuntu OS
• Python >= 3.8 (Anaconda3 is recommended)
• PyTorch 1.7+
• A Nvidia GPU with cuda 11.1+

Datasets

We use three processed datasets: Alibaba-iFashion, Yelp2018 and Last-FM.

• We follow the paper "Learning Intents behind Interactions with Knowledge Graph for Recommendation" to process data.
• You can find the full version of recommendation datasets via Alibaba-iFashion, Yelp2018 and Last-FM.

Training

• Alibaba-iFashion dataset

python main.py --dataset alibaba-ifashion --lr 0.0001 --context_hops 3 --num_neg_sample 200 --margin 0.6 --max_iter 2

• Yelp2018 dataset

python main.py --dataset yelp2018 --lr 0.0001 --context_hops 2 --num_neg_sample 400 --margin 0.8 --max_iter 1

• Last-FM dataset

python main.py --dataset last-fm --lr 0.0001 --context_hops 2 --num_neg_sample 400 --margin 0.7 --max_iter 2

krdn's Issues

请问MCCLK这个基线是如何实现的

关于KRDN的公式4和公式2的代码实现

作者您好，感谢您的团队提供的KRDN代码！现在有一个问题想请教：在下载的代码中，我认为class Aggregator的KG_forward方法实现的是论文中的公式4，不知是否正确？
但是在KG_forward我未能找到公式4中的m(h,t)，也就是在公式2中定义的符合伯努利分布的m(i)。

希望得到您的解答与纠正，谢谢！

Something went wrong...

@xj-zhu98
`The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "main.py", line 106, in
all_feed_data = get_feed_data(train_cf_pairs, user_dict['train_user_set']) # {'user': [n,], 'pos_item': [n,], 'neg_item': [n, n_sample]}
File "main.py", line 54, in get_feed_data
feed_dict['neg_items'] = torch.LongTensor(negative_sampling(entity_pairs,train_user_set))
File "main.py", line 46, in negative_sampling
neg_items = pool.map(get_neg_one, user_item.cpu().numpy()[:, 0])
File "Anaconda3\pytorch\lib\multiprocessing\pool.py", line 364, in map
return self._map_async(func, iterable, mapstar, chunksize).get()
File "Anaconda3\pytorch\lib\multiprocessing\pool.py", line 771, in get
raise self._value
ValueError: high <= 0

进程已结束，退出代码为 1
`

datasets preprocess

Can you provide data preprocessing code?Thank you very much

代码爆显存

RuntimeError: CUDA out of memory. Tried to allocate 170.00 MiB (GPU 0; 23.70 GiB total capacity; 21.01 GiB already allocated; 136.56 MiB free; 21.41 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF感觉是够的运行就会报这个错误