rinna/nekomata-14b

Overview

We conduct continual pre-training of qwen-14b on 66B tokens from a mixture of Japanese and English datasets. The continual pre-training significantly improves the model's performance on Japanese tasks. It also enjoys the following great features provided by the original Qwen model.

• The inclusive Qwen vocabulary (vocab size > 150k) enables the model to processs Japanese texts much more efficiently than the previously released youri series.
• The model supports a maximum sequence length of 8192.

The name nekomata comes from the Japanese word 猫又/ねこまた/Nekomata, which is a kind of Japanese mythical creature (妖怪/ようかい/Youkai).

• Library

  The model was trained using code based on aws-neuron/neuronx-nemo-megatron.

• Model architecture

  A 40-layer, 5120-hidden-size transformer-based language model. Please refer to the Qwen paper for architecture details.

• Continual pre-training

  The model was initialized with the qwen-14b model and continually trained on around 66B tokens from a mixture of the following corpora

  • Japanese CC-100
  • Japanese C4
  • Japanese OSCAR
  • The Pile
  • Wikipedia
  • rinna curated Japanese dataset

• Training Infrastructure

  nekomata-14B was trained on 16 nodes of Amazon EC2 trn1.32xlarge instance powered by AWS Trainium purpose-built ML accelerator chip. The pre-training job was completed within a timeframe of approximately 7 days.

• Contributors

  • Tianyu Zhao
  • Akio Kaga
  • Kei Sawada

---

Benchmarking

Please refer to rinna's LM benchmark page.

---

How to use the model

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("rinna/nekomata-14b", trust_remote_code=True)

# Use GPU with bf16
# model = AutoModelForCausalLM.from_pretrained("rinna/nekomata-14b", device_map="auto", trust_remote_code=True, bf16=True)

# Use GPU with fp16
# model = AutoModelForCausalLM.from_pretrained("rinna/nekomata-14b", device_map="auto", trust_remote_code=True, fp16=True)

# Use CPU
# model = AutoModelForCausalLM.from_pretrained("rinna/nekomata-14b", device_map="cpu", trust_remote_code=True)

# Automatically select device and precision
model = AutoModelForCausalLM.from_pretrained("rinna/nekomata-14b", device_map="auto", trust_remote_code=True)

text = "西田幾多郎は、"
token_ids = tokenizer.encode(text, add_special_tokens=False, return_tensors="pt")

with torch.no_grad():
    output_ids = model.generate(
        token_ids.to(model.device),
        max_new_tokens=200,
        min_new_tokens=200,
        do_sample=True,
        temperature=1.0,
        top_p=0.95,
        pad_token_id=tokenizer.pad_token_id,
        bos_token_id=tokenizer.bos_token_id,
        eos_token_id=tokenizer.eos_token_id
    )

output = tokenizer.decode(output_ids.tolist()[0])
print(output)

---

Tokenization

The model uses the original Qwen tokenizer. It augments the cl100k tiktoken tokenizer and has a vocabulary size of 151,936. The inclusive vocabulary helps the model to reach a better tokenization efficiency, especially for Japanese texts.

We compared the Qwen tokenizer (as used in nekomata) and the llama-2 tokenizer (as used in youri) on different text collections and found that the Qwen tokenizer achieves a much better byte2token rate (i.e. the average number of tokens produced from 1 byte of text) as following. A lower byte2token rate indicates a better tokenization efficiency.

Tokenizer	Japanese	English	Multilingual
Qwen	0.24	0.27	0.27
llama-2	0.40	0.29	0.36

---

How to cite

@misc{rinna-nekomata-14b,
    title = {rinna/nekomata-14b},
    author = {Zhao, Tianyu and Kaga, Akio and Sawada, Kei},
    url = {https://huggingface.co/rinna/nekomata-14b}
}

@inproceedings{sawada2024release,
    title = {Release of Pre-Trained Models for the {J}apanese Language},
    author = {Sawada, Kei and Zhao, Tianyu and Shing, Makoto and Mitsui, Kentaro and Kaga, Akio and Hono, Yukiya and Wakatsuki, Toshiaki and Mitsuda, Koh},
    booktitle = {Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)},
    month = {5},
    year = {2024},
    pages = {13898--13905},
    url = {https://aclanthology.org/2024.lrec-main.1213},
    note = {\url{https://arxiv.org/abs/2404.01657}}
}

---

License

Tongyi Qianwen LICENSE AGREEMENT