中级5 Bungee jumping

本文探讨了蹦极作为一种极限运动的魅力所在。作者描述了自己对于蹦极既期待又害怕的复杂心情,并表达了宁愿选择短暂而充满刺激的生活也不愿过漫长而乏味日子的人生态度。

摘要生成于 C知道 ,由 DeepSeek-R1 满血版支持, 前往体验 >

wild boar
it originated in new Zealand
look,taste,sound ,smell
taste delicious
see,watch,look at,hear,listen to feel
I watch him doing it
I watch him do it
i'd like to have a appointment with you doctor.
will you be available then?
be tied to= be bound to
That poor dog is tied to the fence.
I hate his guts.
he scared me to death.
be scared of = be afraid of.
He looked into the mirror and found he looked much older.
be frightened of..fear.
get somewhere =get nowhere
if you work hard ,you'll get somewhere one day.
agree with sb on sth
i agree with you on this point,but disagree with you on that point.
agreeable
I like Mary because of her agreeable personality.
he is a very agreeable person.he has an agreeable personality.

蹦极看起来蛮好玩的。光是看别人做这件事就会让我紧张。
双腿只用一条绳索绑着,从水面上一千尺的高度跳下的确需要很大的胆量。光是想到这点就够让我害怕的了。
可是,这是我总有一天真的想做的事。
有些人认为我疯了。他们说去跳就已经够蠢了,而还要付钱则不啻为疯狂的行为。

我并不同意。对我来说,过个短暂而剌激的生活比过漫长却无聊的日子好得多。你认为呢?


Bungee jumping looks like fun.It makes me nervous just to watch someone do it.

It certainly takes a lot of guts to jump one thousand feet above the water with only a rope tied to your legs.

It scares me just to think about it.But it is something I really want to do one day.

Some people think I'm crazy.They say to jump is too foolish,but to have to pay for it is madness.

I don't agree.For me,to live a short and exciting life is far better than to live a long and boring one.

What do you think?

资源下载链接为: https://pan.quark.cn/s/140386800631 通用大模型文本分类实践的基本原理是,借助大模型自身较强的理解和推理能力,在使用时需在prompt中明确分类任务目标,并详细解释每个类目概念,尤其要突出类目间的差别。 结合in-context learning思想,有效的prompt应包含分类任务介绍及细节、类目概念解释、每个类目对应的例子和待分类文本。但实际应用中,类目和样本较多易导致prompt过长,影响大模型推理效果,因此可先通过向量检索缩小范围,再由大模型做最终决策。 具体方案为:离线时提前配置好每个类目的概念及对应样本;在线时先对给定query进行向量召回,再将召回结果交给大模型决策。 该方法不更新任何模型参数,直接使用开源模型参数。其架构参考GPT-RE并结合相关实践改写,加入上下文学习以提高准确度,还使用BGE作为向量模型,K-BERT提取文本关键词,拼接召回的相似例子作为上下文输入大模型。 代码实现上,大模型用Qwen2-7B-Instruct,Embedding采用bge-base-zh-v1.5,向量库选择milvus。分类主函数的作用是在向量库中召回相似案例,拼接prompt后输入大模型。 结果方面,使用ICL时accuracy达0.94,比bert文本分类的0.98低0.04,错误类别6个,处理时添加“家居”类别,影响不大;不使用ICL时accuracy为0.88,错误58项,可能与未修改prompt有关。 优点是无需训练即可有较好结果,例子优质、类目界限清晰时效果更佳,适合围绕通用大模型api打造工具;缺点是上限不高,仅针对一个分类任务部署大模型不划算,推理速度慢,icl的token使用多,用收费api会有额外开销。 后续可优化的点是利用key-bert提取的关键词,因为核心词语有时比语意更重要。 参考资料包括
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值