---
license: mit
tags:
  - relbench
  - relational-deep-learning
  - self-supervised
  - event-sequences
---

# LEDGER — RelBench checkpoints

**L**earned **E**vent **D**istributions for **G**eneric **E**ntity **R**eadout

Checkpoints behind the LEDGER entry on the
[RelBench leaderboard](https://star-project.stanford.edu/relbench/leaderboard/).
A relational database is read as a ledger of timestamped events, and one
self-supervised objective — the sufficient statistics of the window
`(t_q, t_q + D]` — covers every entity task. The task enters only at readout.
**No task labels are used in training.**

Code: https://github.com/ShantanuAnant/ledger

| Board | Metric | LEDGER | Position |
|---|---|---|---|
| Recommendation | MAP | **8.75** | **3 of 7** |
| Classification | AUROC | **72.89** | 7 of 13 |
| Regression | NMAE | **0.3391** | 8 of 14 |

All 31 tasks scored by `python -m relbench.submit` (relbench 3.0.1).

## Layout

```
recommendation/   WHO head      → link-prediction tasks
window/           WINDOW head   → classification + regression tasks
MANIFEST.md       task → checkpoint, all 31 tasks
```

Files are named `{dataset}__{entity}.pt`. One checkpoint often serves several
tasks; `MANIFEST.md` is the authoritative mapping.

## Usage

```python
import torch
from ledger.model.load import load_checkpoint     # from the GitHub repo

ckpt = torch.load("recommendation/rel-amazon__customer.pt",
                  map_location="cpu", weights_only=False)
print(ckpt["args"])      # the exact training configuration
```

Evaluate with `scripts/eval_rec.py` (recommendation) or `scripts/eval_entity.py`
/ `scripts/icl_readout.py` (classification, regression). Both need a corpus
built at the matching cutoff — see the repository README.

## Note

These are **slimmed for inference**: `model` and `args` only, with optimizer
state removed. They score identically to the originals and can be fine-tuned
from, but training cannot be *resumed* from them.

Trained on a single NVIDIA B200 (183 GB), CUDA 12.8, PyTorch 2.11.