---
language:
  - de
  - en
  - es
  - fr
---

# Model Card for `passage-ranker-v1-L-multilingual`

This model is a passage ranker developed by Sinequa. It produces a relevance score given a query-passage pair and is used to order search results.

Model name: `passage-ranker-v1-L-multilingual`

## Supported Languages

The model was trained and tested in the following languages:

- English
- French
- German
- Spanish

## Scores

| Metric              | Value |
|:--------------------|------:|
| Relevance (NDCG@10) | 0.471 |

Note that the relevance score is computed as an average over 14 retrieval datasets (see
[details below](#evaluation-metrics)).

## Inference Times

| GPU                                       | Quantization type |  Batch size 1  |  Batch size 32 |
|:------------------------------------------|:------------------|---------------:|---------------:|
| NVIDIA A10                                | FP16              |           2 ms |          31 ms |
| NVIDIA A10                                | FP32              |           4 ms |          82 ms |
| NVIDIA T4                                 | FP16              |           3 ms |          65 ms |
| NVIDIA T4                                 | FP32              |          14 ms |         364 ms |
| NVIDIA L4                                 | FP16              |           2 ms |          38 ms |
| NVIDIA L4                                 | FP32              |           5 ms |         124 ms |

## Gpu Memory usage

| Quantization type                                |   Memory   |
|:-------------------------------------------------|-----------:|
| FP16                                             |    550 MiB |
| FP32                                             |   1050 MiB |

Note that GPU memory usage only includes how much GPU memory the actual model consumes on an NVIDIA T4 GPU with a batch
size of 32. It does not include the fix amount of memory that is consumed by the ONNX Runtime upon initialization which
can be around 0.5 to 1 GiB depending on the used GPU.

## Requirements

- Minimal Sinequa version: 11.10.0
- Minimal Sinequa version for using FP16 models and GPUs with CUDA compute capability of 8.9+ (like NVIDIA L4): 11.11.0
- [Cuda compute capability](https://developer.nvidia.com/cuda-gpus): above 5.0 (above 6.0 for FP16 use)

## Model Details

### Overview

- Number of parameters: 124 million
- Base language model: Homegrown Sinequa BERT-Base ([Paper](https://arxiv.org/abs/1810.04805)) pretrained in the four
  supported languages
- Insensitive to casing and accents
- Training procedure: [MonoBERT](https://arxiv.org/abs/1901.04085)

### Training Data

- Probably-Asked Questions
  ([Paper](https://arxiv.org/abs/2102.07033),
  [Official Page](https://github.com/facebookresearch/PAQ))
    - Original English dataset
    - Translated datasets for the other three supported languages

### Evaluation Metrics

To determine the relevance score, we averaged the results that we obtained when evaluating on the datasets of the
[BEIR benchmark](https://github.com/beir-cellar/beir). Note that all these datasets are in English.

| Dataset           | NDCG@10 |
|:------------------|--------:|
| Average           |   0.471 |
|                   |         |
| Arguana           |   0.583 |
| CLIMATE-FEVER     |   0.150 |
| DBPedia Entity    |   0.366 |
| FEVER             |   0.734 |
| FiQA-2018         |   0.288 |
| HotpotQA          |   0.698 |
| MS MARCO          |   0.341 |
| NFCorpus          |   0.345 |
| NQ                |   0.483 |
| Quora             |   0.766 |
| SCIDOCS           |   0.142 |
| SciFact           |   0.654 |
| TREC-COVID        |   0.711 |
| Webis-Touche-2020 |   0.334 |

We evaluated the model on the datasets of the [MIRACL benchmark](https://github.com/project-miracl/miracl) to test its multilingual capacities. Note that not all training languages are part of the benchmark, so we only report the metrics for the existing languages.

| Language | NDCG@10 |
|:---------|--------:|
| French   |   0.401 |
| German   |   0.396 |
| Spanish  |   0.453 |