File size: 2,371 Bytes
b676a2a
 
 
 
 
 
 
 
fed64ef
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b676a2a
 
fed64ef
b676a2a
fed64ef
 
b676a2a
fed64ef
 
 
 
 
 
b676a2a
 
 
 
 
 
fed64ef
b676a2a
fed64ef
 
b676a2a
fed64ef
 
 
b676a2a
fed64ef
 
 
 
b676a2a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fed64ef
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
---
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