File size: 1,655 Bytes
5326d7f
 
 
c8158be
5326d7f
 
c8158be
 
 
5326d7f
 
c8158be
 
 
 
5326d7f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
language:
- en
license: apache-2.0
size_categories:
- 1M<n<10M
task_categories:
- text-retrieval
- feature-extraction
---

# F2LLM Dataset

[Paper](https://huggingface.co/papers/2510.02294) | [Code](https://github.com/codefuse-ai/CodeFuse-Embeddings/tree/main/F2LLM)

The F2LLM dataset includes 6 million query-document-negative tuples curated solely from open-source, non-synthetic data, serving as a strong, budget-friendly baseline for training embedding models.

## Data Format

Data are compiled into three categories: retrieval, classification, and clustering. Each retrieval and clustering data sample is accompanied by 24 hard negatives. Each classification data sample is accompanied by 1 hard negative.

The data fields are:
```json
{
  "query": ...
  "passage": ...
  "negative_1": ...
  ...
  "negative_n": ...
}
```

For more details, please refer to our [technical report](https://arxiv.org/abs/2510.02294).

## Usage

Code for training embedding models on the F2LLM data is available in our [Github repo](https://github.com/codefuse-ai/CodeFuse-Embeddings/tree/main/F2LLM).

## Citation

If you use the F2LLM models, data, or code, please cite the following technical report.

```
@article{2025F2LLM,
  title={F2LLM Technical Report: Matching SOTA Embedding Performance with 6 Million Open-Source Data}, 
  author={Ziyin Zhang and Zihan Liao and Hang Yu and Peng Di and Rui Wang},
  journal      = {CoRR},
  volume       = {abs/2510.02294},
  year         = {2025},
  url          = {https://doi.org/10.48550/arXiv.2510.02294},
  doi          = {10.48550/ARXIV.2510.02294},
  eprinttype    = {arXiv},
  eprint       = {2510.02294}
}
```