---
language:
- fr
- srr
pipeline_tag: translation
tags:
- low-resource
- nllb
- serer
- machine-translation
license: other
---

# French–Serer NLLB LoRA

Part of a benchmark of six configurations for French→Serer neural machine translation.
Serer is a critically low-resource Niger-Congo language (~1.2M speakers, Senegal/Gambia),
phylogenetically close to the well-resourced Wolof.

## Model summary

- **Experiment ID:** C
- **Kind:** final_translation_model
- **Direction:** French → Serer
- **Base model:** `facebook/nllb-200-distilled-600M` (revision: `main`)
- **Best checkpoint:** `french_serer_nllb_lora-epoch=09-val_bleu=19.1677.ckpt`
- **Random seed:** 42

## Hyperparameters

- `LR`: 0.0003
- `NUM_EPOCHS`: 10
- `LORA_R`: 16
- `LORA_ALPHA`: 32
- `LORA_DROPOUT`: 0.1
- `WARMUP_STEPS`: 500

## Critical limitation — Wolof decoding proxy

Serer (`srr_Latn`) is not a supported NLLB-200 target. This model decodes under the `wol_Latn` (Wolof) language tag as a proxy, fine-tuned on French–Serer data. Automatic metrics (especially BLEU) can be partly inflated by lexical/orthographic overlap with Wolof; see the companion proximity-probe calibration for this corpus (`Fallovski/french-serer-nllb-wolof-proximity-probe`). Human review by a qualified Serer speaker is strongly recommended before any downstream use.

## Intended use

Research on French-to-Serer machine translation on a corpus that is ~90% religious
(Bible) register, ~10% educational glossaries, primarily Siin dialect. **Not** validated
for legal, medical, emergency, or fully autonomous publication use. Private repository —
not intended for public deployment in its current state.

## Evaluation

| Metric | Value |
|---|---|
| BLEU (test, beam=5) | 17.9291 |
| chrF | 38.7191 |
| ROUGE-1 | 0.4343 |
| ROUGE-L | 0.391 |
| BERTScore-F1 | 0.8764 |
| Test loss | 1.9768 |
| BLEU (mean ± std, 3 seeds) | 19.1615 ± 1.044 |

Evaluated on the held-out test split (2890 sentence pairs, SHA-256 of the split:
`01d14d982a3c0cce172b5099e2db064d05bdef89677fe72a45f56715d6364ee2`). Metrics were computed with the project's own evaluation scripts
(not copied from the manuscript without independent reproduction); the training and
evaluation code is kept in a private repository, available on request.

## Training data and rights

Parallel corpus of 23113 train / 2889 val / 2890 test
French–Serer sentence pairs, built primarily from religious texts (Bible, ~90%) and
educational glossaries (~10%), predominantly Siin dialect. Preprocessing: Unicode
normalization, exact-duplicate removal, length-ratio filtering (1:3–3:1). Document-level
splitting was not possible (no document identifiers available); the split is at the
sentence level with a fixed seed. Full provenance, licensing, and consent documentation
are kept in a private dataset card, available on request, prior to any public release.