Llama 3.1 Swallow - Built with Llama

Llama 3.1 Swallow is a series of large language models (8B, 70B) that were built by continual pre-training on the Meta Llama 3.1 models. Llama 3.1 Swallow enhanced the Japanese language capabilities of the original Llama 3.1 while retaining the English language capabilities. We use approximately 200 billion tokens that were sampled from a large Japanese web corpus (Swallow Corpus Version 2), Japanese and English Wikipedia articles, and mathematical and coding contents, etc (see the Training Datasets section) for continual pre-training. The instruction-tuned models (Instruct) were built by supervised fine-tuning (SFT) on the synthetic data specially built for Japanese. See the Swallow Model Index section to find other model variants.

Release History

October 08, 2024: Released Llama-3.1-Swallow-8B-v0.1, Llama-3.1-Swallow-8B-Instruct-v0.1, Llama-3.1-Swallow-70B-v0.1, and Llama-3.1-Swallow-70B-Instruct-v0.1.

Swallow Model Index

Model	Llama-3.1-Swallow	Llama-3.1-Swallow-Instruct
8B	Link	Link
70B	Link	Link

The website https://swallow-llm.github.io/ provides large language models developed by the Swallow team.

Model Details

Model type: Please refer to Llama 3.1 MODEL_CARD for details on the model architecture.
Language(s): Japanese English
Library: Megatron-LM
Tokenizer: Please refer to Llama 3.1 blog for details on the tokenizer.
Contact: swallow[at]nlp.c.titech.ac.jp

Model Performance

Japanese tasks

Model	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
Qwen2-72B	0.9607	0.6399	0.5617	0.9261	0.2362	0.7560	0.2747	0.2419	0.7831	0.5567	0.5937
Qwen2.5-72B	0.9723	0.6111	0.6194	0.9301	0.2792	0.8280	0.2869	0.2521	0.8046	0.6482	0.6232
Sarashina2-70B	0.9285	0.7173	0.6681	0.9294	0.1899	0.4880	0.3129	0.2429	0.5916	0.2384	0.5307
Llama 3 70B	0.9473	0.6042	0.5965	0.9207	0.2254	0.6720	0.2855	0.2526	0.6975	0.4799	0.5682
Llama 3.1 70B	0.9482	0.6112	0.5968	0.9251	0.2284	0.6840	0.2870	0.2553	0.6690	0.4573	0.5662
Llama 3 Youko 70B	0.9455	0.6088	0.6068	0.9226	0.2428	0.6680	0.2909	0.2495	0.7038	0.4530	0.5692
Llama 3 Swallow 70B	0.9714	0.6695	0.6881	0.9218	0.2404	0.7080	0.3072	0.2548	0.7049	0.4683	0.5934
Llama 3.1 Swallow 70B	0.9553	0.6450	0.6776	0.9231	0.2722	0.6840	0.3199	0.2591	0.7088	0.4872	0.5932

English tasks

Model	OpenBookQA	TriviaQA	HellaSWAG	SQuAD2.0	XWINO	MMLU	GSM8K	BBH	HumanEval	En Avg
	4-shot	4-shot	4-shot	4-shot	4-shot	5-shot	4-shot	3-shot	0-shot
	Acc	EM acc	Acc	EM acc	Acc	Acc	EM acc	CoT EM Acc	pass@1
Qwen2-72B	0.4160	0.7890	0.6766	0.4052	0.9161	0.8428	0.8908	0.6388	0.6049	0.6867
Qwen2.5-72B	0.4160	0.7604	0.6849	0.3997	0.9015	0.8608	0.8726	0.7268	0.5543	0.6863
Sarashina2-70B	0.3920	0.5373	0.6270	0.4174	0.9178	0.6303	0.0106	0.6386	0.2799	0.4945
Llama 3 70B	0.4360	0.8263	0.6909	0.4071	0.9213	0.7870	0.8014	0.8266	0.5177	0.6905
Llama 3.1 70B	0.4420	0.8288	0.6898	0.4050	0.9196	0.7846	0.7991	0.6566	0.5476	0.6748
Llama 3 Youko 70B	0.4300	0.8291	0.6900	0.4057	0.9222	0.7862	0.7968	0.8275	0.4128	0.6778
Llama 3 Swallow 70B	0.4240	0.8231	0.6828	0.4059	0.9234	0.7745	0.8143	0.7352	0.4909	0.6749
Llama 3.1 Swallow 70B	0.4320	0.8262	0.6898	0.4018	0.9277	0.7724	0.8089	0.8063	0.5396	0.6894

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])

Training Datasets

Continual Pre-Training

The following datasets were used for continual pre-training.

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.

Acknowledgements

We thank Meta Research for releasing Llama 3.1 under a generous open license.

We received various supports including:

AIST project: "Research and Development of Foundation Models for Generative AI in the Physical Domain"
NEDO project: "Development of Artificial Intelligence Application Technology to Support Judgment in Design Risk Assessment Work Based on the Perspective of Skilled Persons" (JPNP18002) of "Development of Integration Technology as the Core of Next Generation Artificial Intelligence and Robotics"
MEXT project: "Formation of R&D center to ensure transparency and reliability of generative AI models"
AIST program: Large Generative AI Development Support Program

License

META LLAMA 3.1 COMMUNITY LICENSE

Authors

Here are the team members:

From Tokyo Institute of Technology Okazaki Laboratory, the following members:
From Tokyo Institute of Technology YOKOTA Laboratory, the following members:
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 these papers.

@inproceedings{Fujii:COLM2024,
   title={Continual Pre-Training for Cross-Lingual LLM Adaptation:
Enhancing Japanese Language Capabilities},
   author={Kazuki Fujii and Taishi Nakamura and Mengsay Loem and Hiroki
Iida and Masanari Ohi and Kakeru Hattori and Hirai Shota and Sakae
Mizuki and Rio Yokota and Naoaki Okazaki},
   booktitle="Proceedings of the First Conference on Language Modeling",
   series={COLM},
   pages="(to appear)",
   year="2024",
   month=oct,
   address={University of Pennsylvania, USA},
}

@inproceedings{Okazaki:COLM2024,
   title={Building a Large Japanese Web Corpus for Large Language Models},
   author={Naoaki Okazaki and Kakeru Hattori and Hirai Shota and Hiroki
Iida and Masanari Ohi and Kazuki Fujii and Taishi Nakamura and Mengsay
Loem and Rio Yokota and Sakae Mizuki},
   booktitle="Proceedings of the First Conference on Language Modeling",
   series={COLM},
   pages="(to appear)",
   year="2024",
   month=oct,
   address={University of Pennsylvania, USA},
}

References

@misc{dubey2024llama3herdmodels,
      title={The Llama 3 Herd of Models}, 
      author={Abhimanyu Dubey and Abhinav Jauhri and Abhinav Pandey and Abhishek Kadian and Ahmad Al-Dahle and Aiesha Letman and Akhil Mathur and Alan Schelten and Amy Yang and Angela Fan et al.},
      year={2024},
      eprint={2407.21783},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2407.21783}, 
}

tokyotech-llm
/

Llama-3.1-Swallow-70B-v0.1