Recommended-System-PyTorch

Recommended system(2018-2019)

参考腾讯开源工具PyTorch On Angel, arming PyTorch with a powerful Parameter Server, which enable PyTorch to train very big models.

Data

(Fin)

1. movielen data
   • ml-latest
   • ml-100k
   • ml-1m
   • ml-10m
   • ml-20m
2. Criteo data
   • dac

Embedding

(Fin)

1. sparse features
2. sequence features
3. dense features

CTR 模型

(WIP)

model	structure
FM
FFM
DeepFM-201703
xDeepFM-2018
AFM-201708
NFM-201708
FGCNN-201904
MLR
NFFM
WDL
PNN-201611
CIN
CCPM-201510
AutoInt-201810
DCN-201708
DSIN
FNN-201601
DIEN
DIN-201706

Refrences

1. 《推荐系统实践》
2. git@github.com:dawenl/vae_cf.git
3. git@github.com:eelxpeng/CollaborativeVAE.git
4. git@github.com:hidasib/GRU4Rec.git
5. git@github.com:hexiangnan/neural_collaborative_filtering.git
6. git@github.com:NVIDIA/DeepRecommender.git
7. shenweichen/DeepCTR