File size: 3,532 Bytes
e03ae76
a38661d
e03ae76
 
 
 
 
 
 
2720d3e
e03ae76
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f6f8cac
 
 
 
 
 
 
 
 
 
 
 
 
 
e03ae76
 
 
90dfb5b
 
e03ae76
90dfb5b
 
 
 
 
 
 
 
 
 
 
 
 
e03ae76
90dfb5b
 
e03ae76
90dfb5b
 
 
 
e03ae76
 
a38661d
 
e03ae76
 
 
f6f8cac
 
ee26814
f6f8cac
 
 
 
 
 
 
 
ee26814
f6f8cac
 
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
94
95
96
97
---
license: odc-by
language:
- en
tags:
- pretrained
---
# Amber-Data

<img src="amber_logo.png" alt="amber logo" width="300"/>

This dataset contains the fully prepared data sequence used to train Amber, an
LLM360 model. 

## About LLM360
LLM360 is an initiative for comprehensive and fully open-sourced LLMs, 
where all training details, model checkpoints, intermediate results, and 
additional analyses are made available to the community. Our goal is to advance 
the field by inviting the community to deepen the understanding of LLMs 
together. As the first step of the project LLM360, we release all intermediate 
model checkpoints, our fully-prepared pre-training dataset, all source code and
configurations, and training details. We are
committed to continually pushing the boundaries of LLMs through this open-source 
effort.

Get access now at [LLM360 site](https://www.llm360.ai/)

## Data Description

- **Data Format:** 360 tokenized data chunks, each instance has 2049 token indexes.
- **License:** Apache 2.0
- **Resources for more information:**
  - [Code to produce data](https://github.com/LLM360/amber-data-prep)
  - [Amber Model](https://huggingface.co/LLM360/Amber)

## DataMix
The amber dataset uses the following data mix.

| Subset      | Tokens (Billion) |
| ----------- | ----------- |
| Arxiv      | 30.00       |
| Book   | 28.86        |
| C4   | 197.67        |
| Refined-Web   | 665.01        |
| StarCoder   | 291.92        |
| StackExchange   | 21.75        |
| Wikipedia   | 23.90        |
| Total | 1259.13 |


# Loading Amber's Pretraining Data

Below is an example of how to download, sample, and detokenize any subset of AmberDatasets corresponding to an Amber checkpoint. Just set the `CHECKPOINT_NUM` to the subset you are interested in (0-359) and point `CHECKPOINT_PATH` to the local checkpoint folder. 

```python
import random
from transformers import AutoTokenizer
from datasets import load_dataset

CHECKPOINT_NUM = 0  # Pretraining dataset for checkpoint
NUM_SAMPLES = 10  # Number of random samples to decode
CHECKPOINT_PATH = "/path/to/ckpt_000/"  # Local path to a Amber checkpoint

dataset = load_dataset(
    "LLM360/AmberDatasets",
    data_files=f"train/train_{CHECKPOINT_NUM:03}.jsonl",
    split=None,
)

tokenizer = AutoTokenizer.from_pretrained(CHECKPOINT_PATH)
samples = set(random.choices(range(len(dataset["train"])), k=NUM_SAMPLES))

for i, line in enumerate(dataset["train"]):
    if i in samples:
        tokens = line["token_ids"]
        print(f"{i}:{tokenizer.decode(tokens)}")
```

# License
We release our work under [ODC-BY](https://opendatacommons.org/licenses/by/1-0/), hence granting the rights over the dataset, but not the contents of the dataset individually.


# Citation
To cite LLM360, you can cite the following:

```
@misc{liu2023llm360,
      title={LLM360: Towards Fully Transparent Open-Source LLMs}, 
      author={Zhengzhong Liu and Aurick Qiao and Willie Neiswanger and Hongyi Wang and Bowen Tan and Tianhua Tao and Junbo Li and Yuqi Wang and Suqi Sun and Omkar Pangarkar and Richard Fan and Yi Gu and Victor Miller and Yonghao Zhuang and Guowei He and Haonan Li and Fajri Koto and Liping Tang and Nikhil Ranjan and Zhiqiang Shen and Xuguang Ren and Roberto Iriondo and Cun Mu and Zhiting Hu and Mark Schulze and Preslav Nakov and Tim Baldwin and Eric P. Xing},
      year={2023},
      eprint={2312.06550},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}
```

If you only uses the original dataset, please cite the original datasets.