PolyLM多语言-智能服务-文本生成模型

模型简介

PolyLM是一个通晓多种语言的大规模语言模型，涵盖中文、英文、西班牙语、法语、德语、俄语、葡萄牙语、意大利语、阿拉伯语、日语、韩语、泰语、越南语和印尼语等18个语言。该模型可以应用于对话问答、文本生成、机器翻译和情感分析等领域，能够自动生成高质量的多语言文本，从而为跨语言、文化的交流提供便利。

Abstract in English

Large language models (LLMs) demonstrate remarkable ability to comprehend, reason, and generate following nature language instructions. However, the development of LLMs has been primarily focused on high-resource languages, such as English, thereby limiting their applicability and research in other languages. Consequently, we present PolyLM, a multilingual LLM trained on 640 billion (B) tokens, avaliable in two model sizes: 1.7B and 13B. To enhance its multilingual capabilities, we 1) integrate bilingual data into training data; and 2) adopt a curriculum learning strategy that increases the proportion of non-English data from 30% in the first stage to 60% in the final stage during pre-training. Further, we propose a multilingual self-instruct method which automatically generates 132.7K diverse multilingual instructions for model fine-tuning. To assess the model’s performance, we collect several existing multilingual tasks, including multilingual understanding, question answering, generation, and translation. Extensive experiments show that PolyLM surpasses other open-source models such as LLaMA and BLOOM on multilingual tasks while maintaining comparable performance in English.

Our models, alone with the multilingual instruction data, are available at Github and Huggingface.

模型版本

本项目提供了一系列不同规模和用途的模型，参数规模包括1.7B/13B版本（当前模型为13B版本），同时涵盖了预训练底座模型以及指令精调后的Chat版本（即MultiAlpaca系列）。全部版本如下表所示：

Model	Precision	Layers	Heads	Hidden	Max_length	LR	Batch	Type
PolyLM-1.7B	bfloat16	24	16	2048	2048	1.0e-4	4M	Pretrain Model
PolyLM-13B	bfloat16	40	40	5120	2048	6.0e-5	4M	Pretrain Model
PolyLM-MultiAlpaca-13B	bfloat16	40	40	5120	2048	6.0e-5	4M	Chat Model
PolyLM-Assistant-13B	bfloat16	40	40	5120	2048	6.0e-5	4M	Chat Model

训练数据

该模型以PolyLM-13B预训练模型为底座，在以下数据上指令微调得到：

名称	数量	构建方式	备注
code_alpaca	28	GPT 3.5 self-instruct	为了正确展示，对代码做了格式过滤，要求输入、输出中至少有一端可以找出一对```
dolly	15,011	人工编写
flan_v2	100,000	各类NLP任务、CoT任务	从 flan_v2中采样，全量数据非常大
gpt4_alpaca (英文)	52,002	GPT-4 self-instruct
gpt4_alpaca (中文)	48,818	GPT-4 self-instruct
multilingual_alpaca	132,701	GPT-3.5 self-instruct
open_assistant	55,668	人工编写
share_gpt	140,591	ChatGPT聊天记录
gpteacher_codegen	4,535

模型下载

 git lfs install
 git clone https://www.modelscope.cn/damo/nlp_polylm_assistant_13b_text_generation.git

模型使用

# git clone https://github.com/modelscope/modelscope
# cd modelscope
# pip install .

from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
from modelscope import snapshot_download

polylm_13b_model_id = 'damo/nlp_polylm_assistant_13b_text_generation'
revision = 'v1.0.0'

model_dir = snapshot_download(polylm_13b_model_id, revision)

input_text = f"Beijing is the capital of China.\nTranslate this sentence from English to Chinese."
input_text = "<|user|>\n" + f"{input_text}\n" + "<|assistant|>\n"

kwargs = {"do_sample": False, "num_beams": 4, "max_new_tokens": 128, "early_stopping": True, "eos_token_id": 2}
pipeline_ins = pipeline(Tasks.text_generation, model=model_dir)

result = pipeline_ins(input_text, **kwargs)
print(result['text'])

论文引用

如果你觉得这个该模型对有所帮助，请考虑引用下面的相关的论文：

@misc{wei2023polylm,
      title={PolyLM: An Open Source Polyglot Large Language Model}, 
      author={Xiangpeng Wei and Haoran Wei and Huan Lin and Tianhao Li and Pei Zhang and Xingzhang Ren and Mei Li and Yu Wan and Zhiwei Cao and Binbin Xie and Tianxiang Hu and Shangjie Li and Binyuan Hui and Bowen Yu and Dayiheng Liu and Baosong Yang and Fei Huang and Jun Xie},
      year={2023},
      eprint={2307.06018},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}