---
license: apache-2.0
language:
- fr
library_name: transformers
tags:
- t5
- orfeo
- pytorch
- pictograms
- translation
metrics:
- bleu
widget:
- text: "je mange une pomme"
  example_title: "A simple sentence"
- text: "je ne pense pas à toi"
  example_title: "Sentence with a negation"
- text: "il y a 2 jours, les gendarmes ont vérifié ma licence"
  example_title: "Sentence with a polylexical term"
---

# t2p-t5-large-orféo

*t2p-t5-large-orféo* is a text-to-pictograms translation model built by fine-tuning the [t5-large](https://huggingface.co/google-t5/t5-large) model on a dataset of pairs of transcriptions / pictogram token sequence (each token is linked to a pictogram image from [ARASAAC](https://arasaac.org/)).

## Training details

### Datasets

We used the [Propicto-orféo dataset](https://www.ortolang.fr/market/corpora/propicto), which was created from the CEFC-Orféo corpus. 
This dataset was presented in the research paper titled ["A Multimodal French Corpus of Aligned Speech, Text, and Pictogram Sequences for Speech-to-Pictogram Machine Translation](https://aclanthology.org/2024.lrec-main.76/)" at LREC-Coling 2024. The dataset was split into training, validation, and test sets.

| **Split** | **Number of utterances** |
|-----------|:-----------------------:|
| train | 231,374 |
| valid | 28,796 |
| test | 29,009 |

### Parameters

A full list of the parameters is available in the config.json file. We specified the following arguments in the training pipeline :

```python
training_args = Seq2SeqTrainingArguments(
    output_dir="checkpoints_orfeo/",
    evaluation_strategy="epoch",
    save_strategy="epoch",
    learning_rate=2e-5,
    per_device_train_batch_size=32,
    per_device_eval_batch_size=32,
    weight_decay=0.01,
    save_total_limit=3,
    num_train_epochs=40,
    predict_with_generate=True,
    fp16=True,
    load_best_model_at_end=True
)
```

### Evaluation

The model was evaluated with [sacreBLEU](https://huggingface.co/spaces/evaluate-metric/sacrebleu/blob/d94719691d29f7adf7151c8b1471de579a78a280/sacrebleu.py), where we compared the reference pictogram translation with the model hypothesis.

### Results

Comparison to other translation models :
| **Model** | **validation** | **test** |
|-----------|:-----------------------:|:-----------------------:|
| **t2p-t5-large-orféo** | 85.2 | 85.8 |
| t2p-nmt-orféo | **87.2** | **87.4** | 
| t2p-mbart-large-cc25-orfeo | 75.2 | 75.6 |
| t2p-nllb-200-distilled-600M-orfeo | 86.3 | 86.9 |

### Environmental Impact



## Using t2p-t5-large-orféo model with HuggingFace transformers

```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

source_lang = "fr"
target_lang = "frp"
max_input_length = 128
max_target_length = 128

tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint)

inputs = tokenizer("Je mange une pomme", return_tensors="pt").input_ids
outputs = model.generate(inputs.to("cuda:0"), max_new_tokens=40, do_sample=True, top_k=30, top_p=0.95)
pred = tokenizer.decode(outputs[0], skip_special_tokens=True)
```

## Information

- **Language(s):** French
- **License:** Apache-2.0
- **Developed by:** Cécile Macaire
- **Funded by**
  - GENCI-IDRIS (Grant 2023-AD011013625R1)
  - PROPICTO ANR-20-CE93-0005
- **Authors**
  - Cécile Macaire
  - Chloé Dion
  - Emmanuelle Esperança-Rodier
  - Benjamin Lecouteux
  - Didier Schwab


## Citation

If you use this model for your own research work, please cite as follows:

```bibtex

```