Hugging Face's logo

aisingapore
/
llama3-8b-cpt-sea-lionv2.1-instruct

Text Generation
Transformers
Safetensors
llama
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
llama3-8b-cpt-sea-lionv2.1-instruct / README.md
tainc's picture
tainc
Upload LlamaForCausalLM
6f60058 verified
preview code
|
raw
history blame
12.7 kB
metadata
language:
  - en
  - id
  - ta
  - th
  - vi
license: llama3

LLaMA3 8B CPT SEA-LIONv2 Instruct

SEA-LION is a collection of Large Language Models (LLMs) which has been pretrained and instruct-tuned for the Southeast Asia (SEA) region.

LLaMA3 8B CPT SEA-LIONv2 Instruct is a multilingual model which has been fine-tuned with thousands of English and Indonesian instruction-completion pairs alongside a smaller pool of instruction-completion pairs from other ASEAN languages. These instructions have been carefully curated and rewritten to ensure the model was trained on truly open, commercially permissive and high quality datasets.

SEA-LION stands for Southeast Asian Languages In One Network.

  • Developed by: Products Pillar, AI Singapore
  • Funded by: Singapore NRF
  • Model type: Decoder
  • Languages: English, Indonesian, Thai, Vietnamese, Tamil
  • License: LLaMA3 Community License

Model Details

Base model

We performed instruction tuning in English and Indonesian on our continued pre-trained LLaMA3 CPT 8B SEA-LIONv2, a decoder model using the LLaMA3 architecture, to create LLaMA3 8B SEA-LIONv2 Instruct.

Benchmark Performance

We evaluated LLaMA3 8B CPT SEA-LIONv2 Instruct on the BHASA benchmark (arXiv and GitHub) across a variety of tasks.

BHASA stands out amongst other evaluations for SEA languages for its holistic approach to evaluation, including not just traditional Natural Language Processing (NLP) benchmarking tasks (such as sentiment analysis and question answering), but also linguistic and cultural diagnostic tests which are meticulously handcrafted.

The evaluation was done zero-shot with Indonesian prompts and only a sample of 100-1000 instances for each dataset was used as per the setting described in the BHASA paper. The scores shown in the table below have been adjusted to only consider answers provided in the appropriate language.

General Language Capabilities (BHASA)

QA Sentiment Toxicity Eng>Lang Lang>Eng Summary NLI Causal LINDSEA
Language Model Win-rate F1 F1 Macro-F1 ChrF++ ChrF++ F1 Accuracy Accuracy Accuracy
ID llama3-8b-cpt-sealionv2-instruct 76.39% 72.23 84.72 54.64 66.71 65.29 18.70 68.90 87.40 39.91
ID gemma-2-9b-it 76.39% 54.77 78.83 53.37 66.56 65.15 18.20 72.00 94.20 72.14
ID aya-23-8B 61.11% 64.51 82.61 45.40 64.60 63.91 22.15 44.40 89.00 50.45
ID SeaLLM3-7B-Chat 51.39% 45.42 74.58 50.42 64.03 63.44 17.44 58.20 92.00 65.22
ID Qwen2-7B-Instruct 45.83% 45.77 81.97 42.92 58.83 62.79 13.66 63.70 90.80 65.32
ID Meta-Llama-3.1-8B-Instruct 41.67% 63.98 61.34 37.10 63.90 65.35 19.44 29.40 83.20 57.12
ID Sailor-7B-Chat 41.67% 36.93 85.17 42.67 66.61 63.34 14.16 59.50 85.20 54.10
ID Meta-Llama-3-8B-Instruct 36.11% 55.49 72.27 44.68 56.54 55.63 15.35 71.80 82.40 59.25
ID Mistral-7B-Instruct-v0.3 19.44% 40.69 78.84 40.33 49.88 57.89 15.74 59.60 71.80 34.48
VI gemma-2-9b-it 78.91% 48.11 64.23 50.08 57.21 59.20 17.18 52.40 92.60 -
VI llama3-8b-cpt-sealionv2-instruct 64.84% 57.05 54.09 21.99 58.60 58.97 18.28 52.40 87.80 -
VI SeaLLM3-7B-Chat 57.81% 48.71 51.36 27.60 55.05 57.64 16.40 54.50 89.40 -
VI Qwen2-7B-Instruct 54.69% 43.21 61.94 38.44 52.02 56.99 13.10 60.00 88.60 -
VI aya-23-8B 54.69% 73.69 42.14 21.17 56.70 57.02 22.40 50.80 86.80 -
VI Meta-Llama-3.1-8B-Instruct 50.00% 63.49 61.43 7.02 55.91 60.07 18.78 33.20 78.40 -
VI Sailor-7B-Chat 40.62% 31.00 13.13 30.66 58.85 59.02 11.85 49.20 85.80 -
VI Meta-Llama-3-8B-Instruct 25.00% 35.42 70.44 20.91 48.42 52.90 9.65 41.10 83.00 -
VI Mistral-7B-Instruct-v0.3 23.44% 36.13 51.01 41.30 36.89 49.06 13.22 34.70 69.60 -
TH gemma-2-9b-it 82.81% 76.33 49.01 65.49 43.49 56.48 25.79 38.90 90.40 -
TH llama3-8b-cpt-sealionv2-instruct 73.44% 72.41 52.51 38.25 44.84 56.05 18.73 48.80 85.80 -
TH Qwen2-7B-Instruct 62.50% 39.47 50.85 65.89 36.99 52.58 21.32 47.40 88.00 -
TH SeaLLM3-7B-Chat 56.25% 45.01 40.24 55.48 41.80 54.58 23.33 36.40 90.20 -
TH Sailor-7B-Chat 48.44% 31.44 48.11 33.10 44.26 56.03 15.24 45.30 85.60 -
TH Meta-Llama-3.1-8B-Instruct 42.19% 82.16 32.46 25.48 39.65 55.47 24.92 6.20 73.40 -
TH Meta-Llama-3-8B-Instruct 40.62% 68.57 38.80 48.63 35.03 47.74 14.21 54.30 78.20 -
TH Mistral-7B-Instruct-v0.3 29.69% 29.78 45.91 55.58 22.90 41.85 18.65 41.70 59.20 -
TH aya-23-8B 14.06% 43.29 28.84 27.64 19.10 40.29 19.53 33.60 50.60 -
TA gemma-2-9b-it 81.84% 39.04 97.70 0.85 0.86 11.98 89.20 - 38.30 -
TA llama3-8b-cpt-sealionv2-instruct 70.51% 29.35 97.19 0.87 0.86 6.80 76.80 - 34.50 -
TA SeaLLM3-7B-Chat 56.25% 31.79 91.69 0.69 0.78 11.88 51.80 - 34.60 -
TA Qwen2-7B-Instruct 53.12% 25.13 86.39 0.47 0.71 7.49 57.60 - 37.20 -
TA Meta-Llama-3.1-8B-Instruct 48.83% 51.86 88.51 0.81 0.85 9.34 56.60 - 30.80 -
TA aya-23-8B 43.75% 41.89 41.71 0.47 0.74 6.47 43.40 - 40.60 -
TA Sailor-7B-Chat 37.50% 17.46 32.65 0.46 0.70 5.60 11.00 - 0.00 -
TA Meta-Llama-3-8B-Instruct 37.50% 20.88 67.40 0.71 0.70 0.74 58.60 - 41.30 -
TA Mistral-7B-Instruct-v0.3 20.70% 13.85 0.00 0.37 0.52 5.31 14.20 - 0.80 -

Instruction-following Capabilities (IFEval)

| | Indonesian | Vietnamese | English | |--- |:---: |:---: |:---: | | Model | Lang normalised score | Lang normalised score | Lang normalised score | | gemma-2-9b-it | 0.88 | 0.77 | 0.85 | | Meta-Llama-3.1-8B-Instruct | 0.68 | 0.68 | 0.85 | | Qwen2-7B-Instruct | 0.63 | 0.65 | 0.70 | | llama3-8b-cpt-sealionv2-instruct | 0.61 | 0.66 | 0.70 | | aya-23-8B | 0.58 | 0.56 | 0.67 | | SeaLLMs-v3-7B-Chat | 0.55 | 0.52 | 0.67 | | Mistral-7B-Instruct-v0.3 | 0.43 | 0.39 | 0.70 | | Meta-Llama-3-8B-Instruct | 0.27 | 0.21 | 0.80 | | Sailor-7B-Chat | 0.26 | 0.25 | 0.42 |

Multi-turn Capatbilities (MT-Bench)

Indonesian Vietnamese English
Model Weighted Win Rate Weighted Win Rate Weighted Win Rate
gemma-2-9b-it 0.684 0.674 0.638
SeaLLMs-v3-7B-Chat 0.583 0.656 0.429
Qwen2-7B-Instruct 0.498 0.556 0.597
llama3-8b-cpt-sealionv2-instruct 0.531 0.517 0.510
Meta-Llama-3.1-8B-Instruct 0.411 0.477 0.618
aya-23-8B 0.499 0.546 0.416
Meta-Llama-3-8B-Instruct 0.403 0.437 0.564
Mistral-7B-Instruct-v0.3 0.347 0.202 0.524
Sailor-7B-Chat 0.290 0.314 0.190

Usage

SEA-LION can be run using the 🤗 Transformers library 

# Please use transformers==4.43.2

from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("aisingapore/llama3-8b-cpt-sealionv2-instruct")
model = AutoModelForCausalLM.from_pretrained("aisingapore/llama3-8b-cpt-sealionv2-instruct")

prompt_template = "### USER:\n{human_prompt}\n\n### RESPONSE:\n"
prompt = """Apa sentimen dari kalimat berikut ini?
Kalimat: Buku ini sangat membosankan.
Jawaban: """
full_prompt = prompt_template.format(human_prompt=prompt)

tokens = tokenizer(full_prompt, return_tensors="pt")
output = model.generate(tokens["input_ids"], max_new_tokens=20, eos_token_id=tokenizer.eos_token_id)
print(tokenizer.decode(output[0], skip_special_tokens=True))

Prompting Guide

Coming soon

Caveats

It is important for users to be aware that our model exhibits certain limitations that warrant consideration. Firstly, like many LLMs, the model can hallucinate and occasionally generates irrelevant content, introducing fictional elements that are not grounded in the provided context. Users should also exercise caution in interpreting and validating the model's responses due to the potential inconsistencies in its reasoning. Finally, it should be noted that the model has not been optimized for multi-turn dialogue interactions, which may result in reduced effectiveness in extended conversations.

Limitations

Safety

Current SEA-LION models, including this commercially permissive release, have not been aligned for safety. Developers and users should perform their own safety fine-tuning and related security measures. In no event shall the authors be held liable for any claim, damages, or other liability arising from the use of the released weights and codes.

Technical Specifications

Fine-Tuning Details

The LLaMA3 8B CPT SEA-LIONv2 Instruct was fine-tuned using 8x A100-40GB using parameter efficient fine tuning in the form of LoRA.

Data

LLaMA3 8B CPT SEA-LIONv2 Instruct was trained on a wide range of instructions that were manually and stringently verified by our team. A large portion of the effort was dedicated to ensuring that each instruction-completion pair that the model sees is of a high quality and any errors were corrected and rewritten by native speakers or else dropped from our mix.

In addition, special care was taken to ensure that the datasets used had commercially permissive licenses through verification with the original data source.

Link to dataset: coming soon

Call for Contributions

We encourage researchers, developers, and language enthusiasts to actively contribute to the enhancement and expansion of SEA-LION. Contributions can involve identifying and reporting bugs, sharing pre-training, instruction, and preference data, improving documentation usability, proposing and implementing new model evaluation tasks and metrics, or training versions of the model in additional Southeast Asian languages. Join us in shaping the future of SEA-LION by sharing your expertise and insights to make these models more accessible, accurate, and versatile. Please check out our GitHub for further information on the call for contributions.

The Team

Brandon Ong
Bryan Siow
Esther Choa
Huang Yuli
Lee Chwan Ren
Leong Wai Yi
Leong Wei Qi
Li Yier
Liu Bing Jie Darius
Lovenia Holy
Montalan Jann Railey
Ng Boon Cheong Raymond
Ngui Jian Gang
Nguyen Thanh Ngan
Nicholas Cheng
Ong Tat-Wee David
Ong Zhi Hao
Rengarajan Hamsawardhini
Susanto Yosephine
Tai Ngee Chia
Tan Choon Meng
Teo Jin Howe
Teo Eng Sipp Leslie
Teo Wei Yi
Tjhi William
Walter Teng
Wayne Lau
Yeo Yeow Tong
Yong Xianbin

Acknowledgements

AI Singapore is a national programme supported by the National Research Foundation, Singapore and hosted by the National University of Singapore. Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not reflect the views of the National Research Foundation or the National University of Singapore.

Contact

For more info, please contact us using this SEA-LION Inquiry Form

Link to SEA-LION's GitHub repository

Disclaimer

This is the repository for the commercial instruction-tuned model. The model has not been aligned for safety. Developers and users should perform their own safety fine-tuning and related security measures. In no event shall the authors be held liable for any claims, damages, or other liabilities arising from the use of the released weights and codes.