Total runs:
136.9K
Run Growth:
-124.9K
Growth Rate:
-91.24%
Sponsored by
APIMart
s-nlp / mt0-xl-detox-mpd
Model's Last Updated: March 06 2025
text-generation
Introduction of mt0-xl-detox-mpd
Model Details of mt0-xl-detox-mpd
mT0-XL (SynthDetoxM Full)
This a fine-tune of
bigscience/mt0-xl
model on multilingual text detoxification dataset
MultiParaDetox
from the NAACL 2025 Main Track paper
SynthDetoxM: Modern LLMs are Few-Shot Parallel Detoxification Data Annotators
by Daniil Moskovskiy et al.
Usage
The usage is similar to the
from transformers import pipeline
toxic_text = "Your toxic text goes here."
pipe = pipeline("text2text-generation", model="s-nlp/mt0-xl-detox-mpd")
pipe(f"Detoxify: {toxic_text}")
Training Details
The model was fine-tuned for 2 epochs on
textdetox/multilingual_paradetox
dataset with full precision (FP32) using Adafactor optimizer with
1e-4
learning rate and batch size of
4
with gradient checkpointing enabled. The full training configuration is available below:
{
"do_train": true,
"do_eval": true,
"per_device_train_batch_size": 4,
"per_device_eval_batch_size": 4,
"learning_rate": 1e-4,
"weight_decay": 0,
"num_train_epochs": 2,
"gradient_accumulation_steps": 1,
"logging_strategy": "steps",
"logging_steps": 1,
"save_strategy": "epoch",
"save_total_limit": 1,
"warmup_steps": 1,
"report_to": "wandb",
"optim": "adafactor",
"lr_scheduler_type": "linear",
"predict_with_generate": true,
"bf16": false,
"gradient_checkpointing": true,
"output_dir": "/path/",
"seed": 42,
}
Metrics
We use the multilingual detoxification evaluation setup from TextDetox 2024 Multilingual Text Detoxification Shared Task . Specifically, we use the following metrics:
-
Style Transfer Accuracy
(
STA
) is calculated with a
textdetox/xlmr-large-toxicity-classifier. -
Text Similarity
(
SIM
) is calculated as a similarity of text embeddings given by a
sentence-transformers/LaBSEencoder. - Fluency ( FL ) is calculated as a character n-gram F score - $\text{ChrF}_1$ .
These metrics are aggregated in a final Joint metric ( J ):
Evaluation Results
This model was evaluated on the test set of
textdetox/multilingual_paradetox
dataset from
TextDetox 2024 Multilingual Text Detoxification Shared Task
.
The results of the evaluation are presented below.
| German | Spanish | Russian | |
|---|---|---|---|
| Human References | 0.733 | 0.709 | 0.732 |
| Baselines | |||
| Duplicate | 0.287 | 0.090 | 0.048 |
| Delete | 0.362 | 0.319 | 0.255 |
| Backtranslation | 0.233 | 0.275 | 0.223 |
| mT0-XL supervised fine-tuning | |||
| MultiParaDetox (this model) | 0.446 | 0.344 | 0.472 |
| SynthDetoxM (Subset AVG) | 0.460 | 0.402 | 0.475 |
SynthDetoxM
s-nlp/mt0-xl-detox-sdm-full
|
0.482 | 0.470 | 0.546 |
Software
Code for replicating the results from the paper can be found on GitHub .
Citation
BibTeX:
@misc{moskovskiy2025synthdetoxmmodernllmsfewshot,
title={SynthDetoxM: Modern LLMs are Few-Shot Parallel Detoxification Data Annotators},
author={Daniil Moskovskiy and Nikita Sushko and Sergey Pletenev and Elena Tutubalina and Alexander Panchenko},
year={2025},
eprint={2502.06394},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2502.06394},
}
License
This model is licensed under the OpenRAIL++ License, which supports the development of various technologies—both industrial and academic—that serve the public good.
Model Card Authors
Model Card Contact
For any questions, please contact: Daniil Moskovskiy
Runs of s-nlp mt0-xl-detox-mpd on huggingface.co
5
Total runs
0
24-hour runs
-1
3-day runs
-1
7-day runs
-4
30-day runs
More Information About mt0-xl-detox-mpd huggingface.co Model
Url of mt0-xl-detox-mpd
mt0-xl-detox-mpd huggingface.co Url
Provider of mt0-xl-detox-mpd huggingface.co
Other API from s-nlp
Updated:November 08 2024
Total runs:
85.4K
Run Growth:
-72.7K
Growth Rate:
-89.68%
Total runs:
2.7K
Run Growth:
-1.7K
Growth Rate:
-62.85%
Total runs:
2.1K
Run Growth:
-74
Growth Rate:
-3.46%
Total runs:
578
Run Growth:
0
Growth Rate:
0.00%
Total runs:
471
Run Growth:
-58
Growth Rate:
-12.31%
Total runs:
201
Run Growth:
1
Growth Rate:
0.53%
Total runs:
198
Run Growth:
-253
Growth Rate:
-127.78%
Total runs:
179
Run Growth:
25
Growth Rate:
13.97%
Total runs:
155
Run Growth:
0
Growth Rate:
0.00%
Total runs:
154
Run Growth:
-97
Growth Rate:
-62.99%
Total runs:
107
Run Growth:
107
Growth Rate:
100.00%
Total runs:
79
Run Growth:
40
Growth Rate:
59.70%
Total runs:
55
Run Growth:
29
Growth Rate:
50.88%
Total runs:
42
Run Growth:
9
Growth Rate:
23.08%
Total runs:
35
Run Growth:
-114
Growth Rate:
-325.71%
Total runs:
32
Run Growth:
15
Growth Rate:
46.88%
Total runs:
25
Run Growth:
-7
Growth Rate:
-25.93%
Total runs:
24
Run Growth:
-31
Growth Rate:
-140.91%
Total runs:
18
Run Growth:
-6
Growth Rate:
-33.33%
Total runs:
18
Run Growth:
8
Growth Rate:
44.44%
Total runs:
16
Run Growth:
0
Growth Rate:
0.00%
Total runs:
12
Run Growth:
-2
Growth Rate:
-16.67%
Total runs:
9
Run Growth:
5
Growth Rate:
55.56%
Total runs:
8
Run Growth:
4
Growth Rate:
50.00%
Total runs:
4
Run Growth:
-4
Growth Rate:
-100.00%
Total runs:
4
Run Growth:
0
Growth Rate:
0.00%
Total runs:
3
Run Growth:
-4
Growth Rate:
-133.33%
Total runs:
3
Run Growth:
-3
Growth Rate:
-100.00%
Total runs:
3
Run Growth:
0
Growth Rate:
0.00%
Total runs:
2
Run Growth:
1
Growth Rate:
50.00%
Total runs:
1
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%