You can deploy this model and start earning money today!

Ksenia-Dmitrieva/
Corpus-E-opus-mt-en-fr-finetuned-en-to-fr

Name: Corpus-E-opus-mt-en-fr-finetuned-en-to-fr
Availability: InStock

Helsinki-NLP/opus-mt-en-fr

finetune

apache-2.0

endpoints_compatible

Views

No views yet

This model was fine-tuned on a modified corpus of English-French chat translation texts from WMT24 Chat Shared Task.

Original model: Helsinki-NLP/opus-mt-en-fr

Training corpus: parallel training corpus of WMT texts with artificially created "noise". F. e. incorrectly processed html tags ("'" instead of '), indexes ("problem11" instead of "problem") and list markers ("-reason" instead of "•reason")

Training parameters:

training_args = Seq2SeqTrainingArguments(

f"TEST-{model_name}-finetuned-{source_lang}-to-{target_lang}",

report_to="none",

eval_strategy = "epoch",

save_strategy = "epoch"

learning_rate=1.278740919668083e-06,

per_device_train_batch_size=2,

per_device_eval_batch_size=2,

weight_decay=0.07280588826629236,

num_train_epochs=3,

warmup_steps=499

)

Model quality

Estimated average quality from model's translations of test set

BLEU: 50.5

ChrF++: 75.6

COMET: 91.5

COMET Kiwi: 84.9

Ksenia-Dmitrieva/Corpus-E-opus-mt-en-fr-finetuned-en-to-fr

Training parameters:

Model quality

Ksenia-Dmitrieva/
Corpus-E-opus-mt-en-fr-finetuned-en-to-fr