Giter Site home page Giter Site logo

Comments (21)

ScottishFold007 avatar ScottishFold007 commented on July 30, 2024 12

目前感觉中文开源大模型效果最好的就是chatglm6b,大佬有内部测试过吗?

你是来砸场子的么?

from moss.

MrToy avatar MrToy commented on July 30, 2024 2

亲测,ChatGLM-6B强一些。例如你让它写一段react代码,感觉这个模型在胡言乱语

from moss.

Aida-yy avatar Aida-yy commented on July 30, 2024 1

实际上的效果6b大于16B

from moss.

CodePothunter avatar CodePothunter commented on July 30, 2024 1

亲测,一个略显尴尬的事实,ChatGLM-6B强不少

from moss.

rayvzn119 avatar rayvzn119 commented on July 30, 2024 1

实测了一下,和chatglm6b效果有不少差距,例如在做prompt tuning时,moss经常会无法准确理解用户的意图,导致回答很奇怪。不过考虑到moss基座只有16b,glm的基座是130b,语义理解能力有差距很正常。感谢moss团队的开源,期待后续尝试更大的基座,看能否有进一步提高

from moss.

rayvzn119 avatar rayvzn119 commented on July 30, 2024 1

理论上来说16B的模型效果肯定比6B的好

不是参数量越大越好的,moss16b实测效果不如glm6b,感觉跟基座模型能力有关

from moss.

Siegfried-qgf avatar Siegfried-qgf commented on July 30, 2024

我怀疑你在搞事情

from moss.

upupc avatar upupc commented on July 30, 2024

(⊙_⊙)

from moss.

iron-js avatar iron-js commented on July 30, 2024

+1

from moss.

zzcgithub avatar zzcgithub commented on July 30, 2024

哪位大佬用gpt4给来个测评

from moss.

terrifyzhao avatar terrifyzhao commented on July 30, 2024

理论上来说16B的模型效果肯定比6B的好

from moss.

YYGe01 avatar YYGe01 commented on July 30, 2024

能细细说下吗

from moss.

Ricardokevins avatar Ricardokevins commented on July 30, 2024

确实对效果挺好奇的。。。评测应该很快就有人做了

from moss.

CodePothunter avatar CodePothunter commented on July 30, 2024

实测了一下,和chatglm6b效果有不少差距,例如在做prompt tuning时,moss经常会无法准确理解用户的意图,导致回答很奇怪。不过考虑到moss基座只有16b,glm的基座是130b,语义理解能力有差距很正常。感谢moss团队的开源,期待后续尝试更大的基座,看能否有进一步提高

chatglm 6B的基座就是6B

from moss.

ykk648 avatar ykk648 commented on July 30, 2024

我看来看去,哪里写了16B?

from moss.

Ricardokevins avatar Ricardokevins commented on July 30, 2024

我看来看去,哪里写了16B?

看不懂readme的中文啊

from moss.

rayvzn119 avatar rayvzn119 commented on July 30, 2024

实测了一下,和chatglm6b效果有不少差距,例如在做prompt tuning时,moss经常会无法准确理解用户的意图,导致回答很奇怪。不过考虑到moss基座只有16b,glm的基座是130b,语义理解能力有差距很正常。感谢moss团队的开源,期待后续尝试更大的基座,看能否有进一步提高

chatglm 6B的基座就是6B

glm有个130b的基座,6b是基于这个优化训练而来的

from moss.

CodePothunter avatar CodePothunter commented on July 30, 2024

实测了一下,和chatglm6b效果有不少差距,例如在做prompt tuning时,moss经常会无法准确理解用户的意图,导致回答很奇怪。不过考虑到moss基座只有16b,glm的基座是130b,语义理解能力有差距很正常。感谢moss团队的开源,期待后续尝试更大的基座,看能否有进一步提高

chatglm 6B的基座就是6B

glm有个130b的基座,6b是基于这个优化训练而来的

“基于”其实基于数据和经验,并非基于130B的模型(6B不是130B蒸馏得到的),所以从模型规模来说,这就是6B。

from moss.

ykk648 avatar ykk648 commented on July 30, 2024

我看来看去,哪里写了16B?

看不懂readme的中文啊

这个README我是看不懂,一堆没用的中文,ctrl+F 16b 在哪呢?

from moss.

rayvzn119 avatar rayvzn119 commented on July 30, 2024

我看来看去,哪里写了16B?

看不懂readme的中文啊

这个README我是看不懂,一堆没用的中文,ctrl+F 16b 在哪呢?

16b就是160亿的意思,moss介绍里说了,它的参数量是160亿

from moss.

txsun1997 avatar txsun1997 commented on July 30, 2024

@MrToy @rayvzn119 @ykk648 @CodePothunter @YYGe01 @Aida-yy @ScottishFold007 @Siegfried-qgf @upupc @iron-js
之前README中给出的repetition penalty会导致代码生成质量低下,参考#42

from moss.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.