Llama3 Swallow

Our Swallow model has undergone continual pre-training from the Llama 3 family, primarily with the addition of Japanese language data. The Instruct versions use supervised fine-tuning (SFT) and Chat Vector.

Model Release Updates

We are excited to share the release schedule for our latest models:

• July 1, 2024: Released the Llama-3-Swallow-8B-v0.1, Llama-3-Swallow-8B-Instruct-v0.1, Llama-3-Swallow-70B-v0.1, and Llama-3-Swallow-70B-Instruct-v0.1.

Swallow Model Index

Model	Llama-3-Swallow	Llama3 Swallow Instruct
8B	Link	Link
70B	Link	Link

This repository provides large language models developed by Swallow-LLM. Read our blog post.

Third-Party Community Consideration

This model is not owned or developed by NVIDIA. This model has been developed and built to a third-party’s requirements for this application and use case; see link to Non-NVIDIA Model Card.

License

META LLAMA 3 COMMUNITY LICENSE

Model Details

• Model type: Please refer to Llama 3 MODEL_CARD for details on the model architecture.
• Languages: Japanese English
• Library: Megatron-LM
• Tokenizer: Please refer to Llama 3 blog for details on the tokenizer.
• Contact: swallow[at]nlp.c.titech.ac.jp

Model Architecture:

Architecture Type: Transformer

Risks and Limitations

The models released here are still in the early stages of our research and development and have not been tuned to ensure outputs align with human intent and safety considerations.

Input:

Input Type(s): Text
Input Format(s): String
Input Parameters: One Dimensional (1D)

Output:

Output Type(s): Text
Output Format: String
Output Parameters: 1D

Supported Hardware Microarchitecture Compatibility:

• NVIDIA Ampere
  
• NVIDIA Hopper
  
• NVIDIA Lovelace
  

Preferred Operating System(s):

• Linux
  

Training Dataset:

Instruction Tuning

The following datasets were used for the instruction tuning.

• OpenAssistant Conversations Dataset EN top-1 thread
• OpenAssistant Conversations Dataset was used, where human utterances are included but the responses are not used. Instead, the responses were generated using the Mixtral-8x7B-Instruct-v0.1 model.

Model Performance

Japanese tasks

Model	Size	JCom.	JEMHopQA	NIILC	JSQuAD	XL-Sum	MGSM	WMT20-en-ja	WMT20-ja-en	JMMLU	JHumanEval	Ja Avg
		4-shot	4-shot	4-shot	4-shot	1-shot	4-shot	4-shot	4-shot	5-shot	0-shot
		EM acc	Char-F1	Char-F1	Char-F1	ROUGE-2	EM acc	BLEU	BLEU	EM acc	pass@1
karakuri-lm-70b-chat-v0.1	70B	0.8847	0.5139	0.5668	0.9096	0.1369	0.2800	0.2526	0.2095	0.4648	0.2354	0.4454
Meta-Llama-3-70B-Instruct	70B	0.9419	0.6114	0.5506	0.9164	0.1912	0.7200	0.2708	0.2350	0.6789	0.6610	0.5777
Llama-3-Swallow-70B-Instruct-v0.1	70B	0.9607	0.6188	0.6026	0.9236	0.1389	0.6560	0.2724	0.2532	0.6572	0.6000	0.5683
Qwen2-72B-Instruct	72B	0.9634	0.6268	0.5418	0.9210	0.1644	0.7840	0.2592	0.2327	0.7713	0.6909	0.5955

English tasks

Model	Size	OpenBookQA	TriviaQA	HellaSWAG	SQuAD2.0	XWINO	MMLU	GSM8K	BBH	HumanEval	EnAvg
		4-shot	4-shot	4-shot	4-shot	4-shot	5-shot	4-shot	3-shot	0-shot
		Acc	EMacc	Acc	EMacc	Acc	Acc	EMacc	CoTEMAcc	pass@1
karakuri-lm-70b-chat-v0.1	70B	0.4100	0.6873	0.6315	0.3677	0.9049	0.5941	0.3882	0.5724	0.2305	0.5319
Meta-Llama-3-70B-Instruct	70B	00.4400	0.7999	0.6552	0.4024	0.9127	0.7992	0.9052	0.8326	0.7555	0.7225
Llama-3-Swallow-70B-Instruct-v0.1	70B	0.4520	0.8174	0.6758	0.4050	0.9230	0.7883	0.8688	0.8152	0.6890	0.7150
Qwen2-72B-Instruct	72B	0.4360	0.7588	0.6857	0.3913	0.9110	0.8391	0.8499	0.2436	0.6939	0.6455

MT-Bench JA

Model	Size	coding	extraction	humanities	math	reasoning	roleplay	stem	writing	JMTAvg
karakuri-lm-70b-chat-v0.1	70B	0.2804	0.5862	0.6240	0.2934	0.4183	0.5530	0.4859	0.5964	0.4797
Meta-Llama-3-70B-Instruct	70B	0.5969	0.8410	0.7120	0.4481	0.4884	0.7117	0.6510	0.6900	0.6424
Llama-3-Swallow-70B-Instruct-v0.1	70B	0.5269	0.7250	0.5690	0.4669	0.6121	0.6238	0.5533	0.5698	0.5809
Qwen2-72B-Instruct	72B	0.5699	0.7858	0.8222	0.5096	0.7032	0.7963	0.7728	0.8223	0.7228
GPT-3.5(gpt-3.5-turbo-0125)		0.6851	0.7641	0.7414	0.5522	0.5128	0.7104	0.6266	0.7361	0.6661
GPT-4o(gpt-4o-2024-05-13)		0.7296	0.8540	0.8646	0.6641	0.6661	0.8274	0.8184	0.8085	0.7791

Evaluation Benchmarks

Japanese evaluation benchmarks

We used llm-jp-eval(v1.3.0), JP Language Model Evaluation Harness(commit #9b42d41) and Code Generation LM Evaluation Harness(commit #0261c52). The details are as follows:

• Multiple-choice question answering (JCommonsenseQA [Kurihara et al., 2022])
• Open-ended question answering (JEMHopQA [Ishii et al., 2024])
• Open-ended question answering (NIILC [関根, 2003])
• Machine reading comprehension (JSQuAD [Kurihara et al., 2022])
• Automatic summarization (XL-Sum [Hasan et al., 2021])
• Machine translation (WMT2020 ja-en [Barrault et al., 2020])
• Machine translation (WMT2020 en-ja [Barrault et al., 2020])
• Mathematical reasoning (MGSM [Shi et al., 2023])
• Academic exams (JMMLU [尹ら, 2024])
• Code generation (JHumanEval [佐藤ら, 2024])

English evaluation benchmarks

We used the Language Model Evaluation Harness(v.0.4.2) and Code Generation LM Evaluation Harness(commit #0261c52). The details are as follows:

• Multiple-choice question answering (OpenBookQA [Mihaylov et al., 2018])
• Open-ended question answering (TriviaQA [Joshi et al., 2017])
• Machine reading comprehension (SQuAD2 [Rajpurkar et al., 2018])
• Commonsense reasoning (XWINO [Tikhonov and Ryabinin, 2021])
• Natural language inference (HellaSwag [Zellers et al., 2019])
• Mathematical reasoning (GSM8K [Cobbe et al., 2021])
• Reasoning (BBH (BIG-Bench-Hard) [Suzgun et al., 2023])
• Academic exams (MMLU [Hendrycks et al., 2021])
• Code generation (HumanEval [Chen et al., 2021])

MT-Bench JA

We used Japanese MT-Bench to assess the instruction-following capabilities of models. We utilized the following settings:

• Implemantation: FastChat [Zheng+, 2023] (commit #e86e70d0)
• Question: Nejumi LLM-Leaderboard NEO, mtbench_ja_question_v3
• Reference Answer: Nejumi LLM-Leaderboard NEO, mtbench_ja_referenceanswer_v1
• Prompt for Judge: Nejumi LLM-Lederboard NEO, mtbench_ja_prompt_v1
• Judge: gpt-4-1106-preview
• Scoring: Absolute scale normalized to a 0-1 range, averaged over five runs.

Inference:

Engine: TensorRT-LLM
Test Hardware:

• NVIDIA H100x4
  

Ethical Considerations:

NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.

Please report security vulnerabilities or NVIDIA AI Concerns here.

Authors

Here are the team members:

• From Tokyo Institute of Technology Okazaki Laboratory, the following members:
  • Naoaki Okazaki
  • Sakae Mizuki
  • Youmi Ma
  • Koki Maeda
  • Kakeru Hattori
  • Masanari Ohi
  • Taihei Shiotani
  • Koshiro Saito
• From Tokyo Institute of Technology YOKOTA Laboratory, the following members:
  • Rio Yokota
  • Kazuki Fujii
  • Taishi Nakamura
  • Takumi Okamoto
  • Ishida Shigeki
• From Artificial Intelligence Research Center, AIST, Japan, the following members:
  • Hiroya Takamura

How to Cite

If you find our work helpful, please feel free to cite us.

Copy Code
@misc{llama3swallow,
      title={Llama 3 Swallow},
      url={https://swallow-llm.github.io/llama3-swallow.en.html},
      author={Swallow LLM},
      year={2024},
}

Citations

Copy Code
@article{llama3modelcard,
    title={Llama 3 Model Card},
    author={AI@Meta},
    year={2024},
    url = {https://github.com/meta-llama/llama3/blob/main/MODEL_CARD.md}
}

Acknowledgements

We thank Meta Research for releasing Llama 3 under an open license for others to build on.

Our project is supported by the Large Generative AI Development Support Program of the National Institute of Advanced Industrial Science and Technology.