README.md

metadata

library_name: transformers
license: mit
base_model: agentlans/multilingual-e5-small-aligned-v2
tags:
  - generated_from_trainer
model-index:
  - name: multilingual-e5-small-aligned-v2-text-quality-v3
    results: []
language:
  - multilingual
datasets:
  - agentlans/en-translations-quality-v3

Multilingual Text Quality Model

This model rates the quality of non-English text for AI learning. Input a text string, and it outputs a numeric quality score reflecting overall informativeness and usefulness.

Performance

On the evaluation set, it achieved:

• Loss: 0.0641
• MSE: 0.0641
• Combined Score: 0.0641
• Tokens processed during training: 1,109,813,760

Usage Example

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model_name = "agentlans/multilingual-e5-small-quality-v3"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name).to("cuda" if torch.cuda.is_available() else "cpu")

# Higher scores indicate higher text quality.
# The sign of the score has no particular meaning.
# For example, a negative score doesn't necessarily mean that the text is low quality.
def quality(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True).to(model.device)
    with torch.no_grad():
        score = model(**inputs).logits.squeeze().cpu().item()
    return score

print(quality("Your text here."))

Limitations

• Works best on non-fiction and general-purpose texts.
• Scores give an overall quality estimate but don’t explain why.
• Unlike the other quality-v3 models, this model is only trained on short non-English sentences.
• Check for biases and suitability before use.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Mse	Combined Score	Input Tokens Seen
0.0725	1.0	108381	0.0727	0.0727	0.0727	110981376
0.0603	2.0	216762	0.0675	0.0675	0.0675	221962752
0.0559	3.0	325143	0.0703	0.0703	0.0703	332944128
0.0387	4.0	433524	0.0675	0.0675	0.0675	443925504
0.0325	5.0	541905	0.0704	0.0704	0.0704	554906880
0.0276	6.0	650286	0.0672	0.0672	0.0672	665888256
0.025	7.0	758667	0.0641	0.0641	0.0641	776869632
0.0182	8.0	867048	0.0676	0.0676	0.0676	887851008
0.0154	9.0	975429	0.0647	0.0647	0.0647	998832384
0.0133	10.0	1083810	0.0643	0.0643	0.0643	1109813760

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0