|
|
--- |
|
|
library_name: transformers |
|
|
license: mit |
|
|
base_model: agentlans/multilingual-e5-small-aligned-v2 |
|
|
tags: |
|
|
- generated_from_trainer |
|
|
metrics: |
|
|
- accuracy |
|
|
language: |
|
|
- ar |
|
|
- zh |
|
|
- cs |
|
|
- da |
|
|
- nl |
|
|
- fr |
|
|
- de |
|
|
- el |
|
|
- hu |
|
|
- id |
|
|
- it |
|
|
- ja |
|
|
- fa |
|
|
- pl |
|
|
- pt |
|
|
- ru |
|
|
- es |
|
|
- sv |
|
|
- tr |
|
|
- vi |
|
|
datasets: |
|
|
- agentlans/fineweb2hq-vs-c4 |
|
|
pipeline_tag: text-classification |
|
|
--- |
|
|
|
|
|
# agentlans/multilingual-e5-small-fineweb2hq-vs-c4-classifier |
|
|
|
|
|
> [!IMPORTANT] |
|
|
> **Note:** This model is provided for reference and reproducibility, not for standalone use. |
|
|
|
|
|
This model is a fine-tuned version of [agentlans/multilingual-e5-small-aligned-v2](https://huggingface.co/agentlans/multilingual-e5-small-aligned-v2) |
|
|
on the [agentlans/fineweb2hq-vs-c4](https://huggingface.co/datasets/agentlans/fineweb2hq-vs-c4) dataset. |
|
|
|
|
|
The aim is to classify text as higher quality (FineWeb 2 HQ) or lower quality (C4) for AI training. |
|
|
|
|
|
On the validation set: |
|
|
- Loss: 0.1983 |
|
|
- Accuracy: 0.9515 |
|
|
- Combined Score: 1.3494 |
|
|
- Num Input Tokens Seen: 122880000 |
|
|
|
|
|
|
|
|
## Example |
|
|
|
|
|
```python |
|
|
from transformers import pipeline |
|
|
|
|
|
classifier = pipeline("text-classification", model="agentlans/multilingual-e5-small-fineweb2hq-vs-c4-classifier") |
|
|
classifier("Your text here.") |
|
|
``` |
|
|
|
|
|
## Limitations |
|
|
- **Not trained on English data** |
|
|
- Tends to be overly permissive, labelling most texts outside training data as high quality |
|
|
- May be biased against some text types |
|
|
|
|
|
## 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: 3.0 |
|
|
|
|
|
### Training results |
|
|
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Combined Score | Input Tokens Seen | |
|
|
|:-------------:|:-----:|:------:|:---------------:|:--------:|:--------------:|:-----------------:| |
|
|
| 0.1387 | 1.0 | 40000 | 0.1983 | 0.9515 | 1.3494 | 40960000 | |
|
|
| 0.0682 | 2.0 | 80000 | 0.2264 | 0.9528 | 1.3270 | 81920000 | |
|
|
| 0.0424 | 3.0 | 120000 | 0.2598 | 0.9552 | 1.2845 | 122880000 | |
|
|
|
|
|
|
|
|
### Framework versions |
|
|
|
|
|
- Transformers 4.51.3 |
|
|
- Pytorch 2.6.0+cu124 |
|
|
- Datasets 3.2.0 |
|
|
- Tokenizers 0.21.0 |
