Views
No views yet
| Seed | AUC | MRR | NDCG@5 | NDCG@10 |
|---|---|---|---|---|
| 123 | 0.6655 | 0.3105 | 0.3460 | 0.4091 |
| 42 | 0.6588 | 0.3031 | 0.3378 | 0.4007 |
| 456 * | 0.6667 | 0.3128 | 0.3484 | 0.4106 |
| mean ± std | 0.6636±0.0035 | 0.3088±0.0041 | 0.3440±0.0045 | 0.4068±0.0044 |
1model:
2 name: tccm
3 architecture:
4 news_encoder:
5 type: multi_head_self_attention
6 news_dim: 400
7 num_heads: 20
8 head_dim: 20
9 co_num_heads: 5
10 co_head_dim: 40
11 attention_hidden_dim: 200
12 entity_embedding_dim: 100
13 category_embedding_dim: 200
14 popularity_embedding_bins: 200
15 popularity_embedding_dim: 400
16 popularity_encoder:
17 token_embedding_bins: 200
18 token_embedding_dim: 200
19 num_heads: 1
20 head_dim: 400
21 content_dims:
22 - 256
23 - 256
24 - 128
25 time_module:
26 embedding_bins: 505
27 embedding_dim: 100
28 dense_dims:
29 - 64
30 - 64
31 lambda_exp: 2.0
32 activity_gate_dims:
33 - 128
34 - 64
35 use_entity: true
36 use_activity_gate: true
37 use_causal_intervention: false
38 intervention_value: 0.5
39 embedding:
40 size: 300
41 trainable: true
42 dropout_rate: 0.2
43 seed: 42
44inputs:
45 title:
46 max_length: 30
47 history:
48 max_length: 50
49 impressions:
50 max_length: 5
51 max_entities: 5
52 process_title: true
53 process_abstract: false
54 process_category: false
55 process_subcategory: false
56 process_user_id: false
57training:
58 loss:
59 name: categorical_crossentropy
60 from_logits: true
61 reduction: sum_over_batch_size
62 label_smoothing: 0.0
63 optimizer: adam
64 learning_rate: 0.0001
65 batch_size: 256
66 num_epochs: 10
67 early_stopping:
68 patience: 3
69 min_improvement: 0.01
70 negative_sampling:
71 strategy: random
72 candidates: 1
73evaluation:
74 mode: fast
75 evaluator: tccm
76 metrics:
77 - auc
78 - mrr
79 - ndcg@5
80 - ndcg@10
81 batch_size: 256newsrex/TCCM-PYTORCH-MIND-small-glove-random/
├── model.safetensors ← best seed (456)
├── test_results.json
├── training_run_summary.json
├── seed_123/model.safetensors
├── seed_42/model.safetensors
├── seed_456/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/tccm \
7 framework=pytorch \
8 weights=hf://newsrex/TCCM-PYTORCH-MIND-small-glove-random/model.safetensors
9
10# Run evaluation with a specific seed
11uv run python src/eval.py \
12 experiment=mind/glove/tccm \
13 framework=pytorch \
14 weights=hf://newsrex/TCCM-PYTORCH-MIND-small-glove-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}