klozehsu/gdcn-criteo-ctr

GDCN-P: Gated Deep Cross Network for CTR Prediction

Results on Avazu (200,000 samples)

Metric	This Run	Paper (full 40M)
AUC	0.749974	0.7905
LogLoss	0.396100	0.3739
Epoch	2/8	-
Params	1,812,017	-

Paper

GDCN (CIKM 2024)

Architecture

22 fields → Embed(16d) → [352d]
├── Gated Cross Net (3 layers) → [352d]
└── DNN (400→400→400) → [400d]
Concat → Linear(1) → σ → P(click)

Gate: σ(W_g·x) filters each cross layer's interactions.

Dataset

reczoo/Avazu_x4

Generated by ML Intern

This model repository was generated by ML Intern, an agent for machine learning research and development on the Hugging Face Hub.

• Try ML Intern: https://smolagents-ml-intern.hf.space
• Source code: https://github.com/huggingface/ml-intern

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = 'klozehsu/gdcn-criteo-ctr'
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

For non-causal architectures, replace AutoModelForCausalLM with the appropriate AutoModel class.