🚀 beomi/Yi-Ko-6B
The Yi-Ko series models are advanced iterations of 01-ai/Yi models, featuring an expanded vocabulary and Korean/English corpus in further pretraining. They operate within the range of 6 - 34 billion parameters.
🚀 Quick Start
Yi-Ko series models serve as advanced iterations of 01-ai/Yi models, benefiting from an expanded vocabulary and the inclusion of Korean/English corpus in its further pretraining. Just like its predecessor, Yi-Ko series models operate within the broad range of generative text models that stretch from 6 billion to 34 billion parameters. This repository focuses on the 6B pretrained version, which is tailored to fit the Hugging Face Transformers format. For access to the other models, feel free to consult the index provided below.
✨ Features
- Advanced Iteration: Yi-Ko series models are advanced versions of 01-ai/Yi models.
- Expanded Vocabulary: Benefit from an expanded vocabulary and the inclusion of Korean/English corpus in further pretraining.
- Multiple Parameter Sizes: Available in 6B and 34B parameter sizes.
📚 Documentation
Model Details
- Model Developers: Junbum Lee (Beomi)
- Variations: Yi-Ko series will come in a range of parameter sizes — 6B and 34B variations.
- Input: Models input text only.
- Output: Models generate text only.
- Model Architecture: Yi-Ko series models are an auto-regressive language model that uses an optimized transformer architecture based on Llama-2*.
*Yi model architecture is based on Llama2, so it can be loaded via
LlamaForCausalLM class on HF.
| Model Name |
Training Data |
Params |
Context Length |
GQA |
Trained Tokens |
LR |
Batch Size(per step) |
| Yi-Ko-6B |
A mix of Korean + English online data |
6B |
4k |
O |
>60B |
5e-5 |
2048 |
Vocab Expansion
| Model Name |
Vocabulary Size |
Description |
| Original Yi-Series |
64000 |
Sentencepiece BPE |
| Expanded Yi-Ko Series |
78464 |
Sentencepiece BPE. Added Korean vocab and merges |
Tokenizing "안녕하세요, 오늘은 날씨가 좋네요.ㅎㅎ"
| Model |
# of tokens |
Tokens |
| Original Yi-Series |
47 |
['<0xEC>', '<0x95>', '<0x88>', '<0xEB>', '<0x85>', '<0x95>', '하', '<0xEC>', '<0x84>', '<0xB8>', '<0xEC>', '<0x9A>', '<0x94>', ',', '▁', '<0xEC>', '<0x98>', '<0xA4>', '<0xEB>', '<0x8A>', '<0x98>', '은', '▁', '<0xEB>', '<0x82>', '<0xA0>', '<0xEC>', '<0x94>', '<0xA8>', '가', '▁', '<0xEC>', '<0xA2>', '<0x8B>', '<0xEB>', '<0x84>', '<0xA4>', '<0xEC>', '<0x9A>', '<0x94>', '.', '<0xE3>', '<0x85>', '<0x8E>', '<0xE3>', '<0x85>', '<0x8E>'] |
| Expanded Yi-Ko Series |
10 |
['▁안녕', '하세요', ',', '▁오늘은', '▁날', '씨가', '▁좋네요', '.', 'ㅎ', 'ㅎ'] |
| *Equal Korean vocab with Llama-2-Ko Series |
|
|
Tokenizing "Llama 2: Open Foundation and Fine-Tuned Chat Models"
| Model |
# of tokens |
Tokens |
| Original Yi-Series |
21 |
['The', '▁Y', 'i', '▁series', '▁models', '▁are', '▁large', '▁language', '▁models', '▁trained', '▁from', '▁scratch', '▁by', '▁developers', '▁at', '▁', '0', '1', '.', 'AI', '.'] |
| Expanded Yi-Ko Series |
21 |
['▁The', '▁Y', 'i', '▁series', '▁models', '▁are', '▁large', '▁language', '▁models', '▁trained', '▁from', '▁scratch', '▁by', '▁developers', '▁at', '▁', '0', '1', '.', 'AI', '.'] |
*Since Expanded Yi-Ko Series prepends _ at the beginning of the text(to ensure same tokenization for Korean sentences), it shows negilible difference for the first token on English tokenization. |
|
|
Model Benchmark
LM Eval Harness - Korean (polyglot branch)
| beomi/Yi-Ko-6B |
0 |
5 |
10 |
50 |
| kobest_boolq (macro_f1) |
0.705806 |
0.79905 |
0.814299 |
0.81704 |
| kobest_copa (macro_f1) |
0.775604 |
0.808899 |
0.816866 |
0.842943 |
| kobest_hellaswag (macro_f1) |
0.500876 |
0.498673 |
0.493507 |
0.492183 |
| kobest_sentineg (macro_f1) |
0.404371 |
0.967254 |
0.982368 |
0.974811 |
| kohatespeech (macro_f1) |
0.353428 |
0.351804 |
0.402423 |
0.503764 |
| kohatespeech_apeach (macro_f1) |
0.337667 |
0.498679 |
0.471962 |
0.608401 |
| kohatespeech_gen_bias (macro_f1) |
0.124535 |
0.484745 |
0.474475 |
0.461714 |
| korunsmile (f1) |
0.382804 |
0.349344 |
0.391383 |
0.432875 |
| nsmc (acc) |
0.55064 |
0.8801 |
0.89866 |
0.9071 |
| pawsx_ko (acc) |
0.5145 |
0.54 |
0.538 |
0.5165 |
Detailed results can be found here
| Metric |
Value |
| Avg. |
50.27 |
| AI2 Reasoning Challenge (25-Shot) |
48.89 |
| HellaSwag (10-Shot) |
74.48 |
| MMLU (5-Shot) |
55.72 |
| TruthfulQA (0-shot) |
37.09 |
| Winogrande (5-shot) |
72.93 |
| GSM8k (5-shot) |
12.51 |
📄 License
Apache 2.0 (for research)
For commercial purpose,
mailto: jun@beomi.net to acquire Yi-Ko sereis commercial license.
🔗 Citation
Please use this bibtex below:
@misc {lee_junbum_2024,
author = { {Lee Junbum} },
title = { Yi-Ko-6B (Revision 205083a) },
year = 2024,
url = { https://huggingface.co/beomi/Yi-Ko-6B },
doi = { 10.57967/hf/1708 },
publisher = { Hugging Face }
}
🙏 Acknowledgement
The training is supported by TPU Research Cloud program.
🆕 Updates
%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
