Views
No views yet
| Seed | AUC | MRR | NDCG@5 | NDCG@10 |
|---|---|---|---|---|
| 42 * | 0.6720 | 0.3250 | 0.3576 | 0.4206 |
1model:
2 name: lstur
3 architecture:
4 news_encoder:
5 type: cnn
6 filter_num: 400
7 kernel_size: 3
8 activation: relu
9 attention_hidden_dim: 200
10 user_encoder:
11 type: gru
12 gru_unit: 600
13 variant: ini
14 click_predictor:
15 type: dot_product
16 embedding:
17 size: 300
18 trainable: true
19 category_embedding_dim: 100
20 subcategory_embedding_dim: 100
21 dropout_rate: 0.2
22 user_embedding_dropout_rate: 0.5
23 seed: 42
24inputs:
25 title:
26 max_length: 32
27 history:
28 max_length: 50
29 impressions:
30 max_length: 5
31 process_title: true
32 process_abstract: false
33 process_category: true
34 process_subcategory: true
35 process_user_id: true
36training:
37 loss:
38 name: categorical_crossentropy
39 from_logits: true
40 reduction: sum_over_batch_size
41 label_smoothing: 0.0
42 optimizer: adam
43 learning_rate: 0.0001
44 batch_size: 256
45 num_epochs: 10
46 early_stopping:
47 patience: 3
48 min_improvement: 0.01
49 negative_sampling:
50 strategy: random
51 candidates: 4
52evaluation:
53 mode: fast
54 metrics:
55 - auc
56 - mrr
57 - ndcg@5
58 - ndcg@10
59 batch_size: 512newsrex/LSTUR-JAX-MIND-small-bert-random/
├── model.safetensors ← best seed (42)
├── test_results.json
├── training_run_summary.json
├── seed_42/model.safetensors
└── README.md1git clone https://github.com/igor17400/NewsReX.git
2cd NewsReX && uv sync
3
4# Run evaluation with best seed weights
5uv run python src/eval.py \
6 experiment=mind/glove/lstur \
7 framework=jax \
8 weights=hf://newsrex/LSTUR-JAX-MIND-small-bert-random/model.safetensors
9
10# Run evaluation with a specific seed
11uv run python src/eval.py \
12 experiment=mind/glove/lstur \
13 framework=jax \
14 weights=hf://newsrex/LSTUR-JAX-MIND-small-bert-random/seed_42/model.safetensors1@misc{newsrex2026,
2 title={NewsReX: An Open-Source Multi-Framework for Neural News Recommendation},
3 author={Igor L. R. Azevedo and Toyotaro Suzumura and Yuichiro Yasui},
4 year={2025},
5 eprint={2508.21572},
6 archivePrefix={arXiv},
7 primaryClass={cs.IR},
8 url={https://arxiv.org/abs/2508.21572},
9}