---
title: 'TencentGR-10M: Industrial-Scale GR Dataset'
url: https://www.emergentmind.com/topics/tencentgr-10m-dataset
type: topic
---

# TencentGR-10M: Industrial-Scale GR Dataset

TencentGR-10M is a large-scale, industrial-grade benchmark dataset for all-modality generative recommendation (GR), released as part of the final track of the Tencent Advertising Algorithm Challenge 2025. Built from real, de-identified Tencent Ads logs, TencentGR-10M distinguishes itself through its scale, the integration of rich collaborative categorical identifiers and multi-modal content (such as text and vision embeddings), and a task and evaluation protocol custom-designed for generative modeling of sequential user–item interactions in online advertising [2604.04976].

## 1. Dataset Construction and Scale

TencentGR-10M comprises 10,139,575 users, each with up to 100 logged sequential ad interactions, yielding approximately 986 million exposure, click, and conversion events. The ad corpus spans 17,487,676 distinct items, with a global candidate pool of 3,637,720 ads relevant for retrieval ranking. User sequences $S_u$ begin with a user-profile token, followed by item-interaction tokens $x_{u,1}, ..., x_{u,T_u}$ (average $T_u = 97.29$). Each $x_{u,t}$ is annotated with an action-type label $r_{u,t} \in \{0 = \mathrm{exposure}, 1 = \mathrm{click}, 2 = \mathrm{conversion}\}$.

The prediction target construction is aligned with industry-standard conversion attribution: if a conversion post-$t_\mathrm{begin}$ can be attributed within the attribution window, this conversion is mapped back to the causal click and impression; otherwise, the earliest post-$t_\mathrm{begin}$ click (and its impression) is taken as the target. Conversions thus appear both within the history and as possible targets. The overall action-type distribution is 94.63% exposures, 2.85% clicks, and 2.52% conversions.

All user and item identifiers, along with sensitive content (names, text, videos, raw images), are de-identified or removed; only hashed categorical IDs and pre-computed embeddings derived from content remain. No user sequence exceeds 100 item tokens.

## 2. Feature Schema and Modalities

TencentGR-10M merges sparse collaborative (ID-based) features with dense, pre-extracted multi-modal embeddings for each ad, yielding a fully all-modality dataset.

### Sparse Features

Ad-side categorical fields (feature IDs 100–122) include single-valued features such as format (6 values, 99.96% coverage), advertiser type (53 values), campaign ID (173,463 values), business line (30 values), device type (33 values), and category (e.g., second-level category, 988 values, 29% coverage). User-side features (IDs 103–110) include age-bucket (87 values), gender (2), region (7), long-term and short-term interest tags (multi-valued, 14 and 19 tags, respectively), device OS (3), and other demographic categorizations.

### Multi-Modal Embeddings

Each ad interaction can include up to six embeddings from high-capacity pretrained models:

| Embedding ID | Model                          | Modality | Size      | Output Dim |
|--------------|-------------------------------|----------|-----------|------------|
| 81           | Bert-finetune                 | Text     | 0.3B      | 32         |
| 82           | Conan-embedding-v1            | Text     | 0.3B      | 1024       |
| 83           | gte-Qwen2-7B-instruct         | Text     | 7B        | 3584       |
| 84           | hunyuan_mm_7B_finetune        | Image    | 7B        | 32         |
| 85           | QQMM-embed-v1                 | Image    | 8B        | 3584       |
| 86           | UniME-LLaVA-OneVision-7B      | Image    | 8B        | 3584       |

Text models 81 and 83 are further finetuned with contrastive loss using Tencent’s click/conversion logs. Most embeddings have $>90\%$ coverage in the dataset.

### Record Schema

Each interaction record consists of hashed 32-bit categorical IDs, an 8-bit action code, and float32 multi-modal vectors. For each user entry in Parquet format, fields include user_id, user_feats, a sequence of item-interaction dictionaries (with ad_id, campaign_id, etc.), candidate_pool of ads, target_ad, and target_action.

## 3. Data Format, Access, and Partitioning

TencentGR-10M is distributed via HuggingFace at https://huggingface.co/datasets/TAAC2025/TencentGR-10M as compressed Parquet files. Directory layout follows: `train/part-xxxx.parquet`, `validation/*.parquet`, `test/*.parquet`. Users are shuffled and split chronologically by timestamps $t_\mathrm{begin}$, $t_\mathrm{end}$: 80% for training, 10% for validation, 10% for test. Each Parquet partition consists of approximately 100,000 full user sequences.

The candidate pool and prediction targets are provided explicitly, consistent with the all-modality recommendation protocol. A private test set, withheld by organizers, is used for final evaluation; only training and validation labels are public.

## 4. Evaluation Protocol

The evaluation framework weights high-value conversions more heavily, employing “weighted” variants of standard metrics. For user $u$, ground-truth $G_u$, and predicted $\hat{y}_{u,1…K}$:

- Action weights: $w(\text{exposure})=0$, $w(\text{click})=1$, $w(\text{conversion})=2.5$.
- Weighted HitRate@K:
$$
\text{w-HitRate@K}(u) = \frac{ \sum_{k=1}^K w(\hat{y}_{u,k}) \cdot \mathbb{1}\{\hat{y}_{u,k} \in G_u\} }{ \sum_{i \in G_u} w(i) }
$$
- Weighted DCG@K, IDCG@K, and NDCG@K defined by substituting the above weighting into standard definitions, with final leaderboard score:
$$
0.31 \cdot \text{w-HitRate@10} + 0.69 \cdot \text{w-NDCG@10}
$$
Weights are calibrated such that both terms have comparable magnitude.

The private test set labels are not disclosed; only train and validation sets feature public labels for model development and hyperparameter tuning.

## 5. Baseline Model and Training Objective

The baseline for TencentGR-10M is a single-block ($L=1$) causal Transformer with hidden size $d=32$, a single attention head, and dropout of 0.2. Input sequences can be up to 1 user token plus 100 item tokens.

#### Input Representation

Sparse fields are embedded via learnable lookups; continuous multi-modal embeddings are concatenated. User token $x_u^0$ is obtained through a multilayer perceptron (MLP) over all user-side embeddings. Item token $x_{u,t}^0$ uses an MLP over both ad-side categorical fields and all available multimodal embeddings. Positional encodings $p_t$ are added, and the resulting sequence $H^0 = [x_u^0 + p_0, x_{u,1}^0 + p_1, ..., x_{u,T_u}^0 + p_{T_u}]$ is passed through the Transformer layer to yield $H^1$ with user state $h_{u,t} = H^1[t]$.

#### Contrastive Loss

Training employs the InfoNCE loss over one positive target and sampled negatives:
$$
L = -\sum_{(u,t,a)}w_a \cdot \log \frac{\exp(s_{u,t,i^+})}{\exp(s_{u,t,i^+}) + \sum_{i^-}\exp(s_{u,t,i^-})}
$$
where $s_{u,t,i} = h_{u,t}^\top \cdot \mathrm{item\_embedding}(i)$, and $w_\text{exposure}=0$, $w_\text{click}=1$, $w_\text{conversion}=2.5$ matching the evaluation weighting.

#### Inference

All item embeddings are precomputed and indexed via ANN (e.g., Faiss). For each user, the most recent state $h_{u,T_u}$ is used to fetch the top-K nearest items. Baseline code and instructions are available at https://github.com/TencentAdvertisingAlgorithmCompetition/baseline_2025. No absolute performance numbers for the 10M track baseline are published.

## 6. Key Findings from the Challenge

Analysis of top-ranked solutions in the final round identified several reproducible trends:

- **Action-conditioning**: Modeling exposure, click, and conversion as explicit conditioning signals (via gated FiLM layers or special tokens) enables superior intent disentanglement.
- **Semantic IDs via quantization**: Multi-modal quantization (e.g., RQ-KMeans, SVD-based quantizers) transforming continuous embeddings to discrete codes enhances long-tail item coverage and reduces memory footprint.
- **Hierarchical time/session features**: Employing representations such as absolute timestamps, inter-event gaps, session segmentation, cross-day indicators, and Fourier decompositions yields measurable improvements.
- **Contrastive learning scale**: Utilizing large-scale negative banks (order $10^5 – 10^6$ negatives per batch) provides greater performance gains than increasing model depth/width under constant compute.
- **Two-stage training**: Pretraining on exposures for self-supervised next-item prediction, followed by finetuning on clicks/conversions, better aligns models with high-value action distributions.
- **Efficiency techniques**: Adoption of mixed-precision (AMP), FlashAttention, grouped GEMMs, KV cache optimizations, separate sparse/dense optimizers, and single-model (non-ensembled) submissions endows implementations with latency and memory properties suitable for industrial environments.

TencentGR-10M and its companion codebase aim to provide a reproducible and challenging foundation for research on all-modality generative recommendation in advertising at industrial scale [2604.04976].

Source: https://www.emergentmind.com/topics/tencentgr-10m-dataset