You can deploy this model and start earning money today!

YeungNLP/
bloom-2b6-zh

Name: bloom-2b6-zh
Availability: InStock

text-generation-inference

endpoints_compatible

Views

No views yet

项目地址：LLMPruner：大语言模型裁剪工具

LLMPruner是一个大语言模型裁剪工具，通过对大语言模型的冗余词表进行裁剪，减少模型参数量，降低显存占用，提升训练速度，并且能够保留预训练中学习到的知识。

本项目对Bloom进行词表裁剪，保留中文token和常用的英文token，词表由250880将至46145，缩减为原来的18.39%。裁剪得到的Bloom模型如下表：

裁剪模型	原模型	参数量比例
YeungNLP/bloom-396m-zh	bigscience/bloom-560m	70.96%
YeungNLP/bloom-820m-zh	bigscience/bloom-1b1	77.13%
YeungNLP/bloom-1b4-zh	bigscience/bloom-1b7	81.14%
YeungNLP/bloom-2b6-zh	bigscience/bloom-3b	86.48%
YeungNLP/bloom-6b4-zh	bigscience/bloom-7b1	90.81%
YeungNLP/bloomz-396m-zh	bigscience/bloomz-560m	70.96%
YeungNLP/bloomz-820m-zh	bigscience/bloomz-1b1	77.13%
YeungNLP/bloomz-1b4-zh	bigscience/bloomz-1b7	81.14%
YeungNLP/bloomz-2b6-zh	bigscience/bloomz-3b	86.48%
YeungNLP/bloomz-6b4-zh	bigscience/bloomz-7b1	90.81%
YeungNLP/bloomz-6b4-mt-zh	bigscience/bloomz-7b1-mt	90.81%

使用方法：

python

from transformers import BloomTokenizerFast, BloomForCausalLM

tokenizer = BloomTokenizerFast.from_pretrained('YeungNLP/bloom-1b4-zh')
model = BloomForCausalLM.from_pretrained('YeungNLP/bloom-1b4-zh')
print(tokenizer.batch_decode(model.generate(tokenizer.encode('长风破浪会有时', return_tensors='pt'))))

YeungNLP/bloom-2b6-zh

YeungNLP/
bloom-2b6-zh