WizardLM is participating 2021 OSC China Open Source Project Selection , please vote for it!
WizardLM in 2021 OSC China Open Source Project Selection {{projectVoteCount} has been obtained in, please vote for it!
2021 OSC China Open Source Project Selection It is in hot progress. Come and vote for your favorite open source project!
2021 OSC China Open Source Project Selection>>> Midfield Review
WizardLM won the 2021 OSC China Open Source Project Selection "The Best Popularity Project" !
Authorization Agreement unknown
development language Python View source code »
operating system Cross platform
Software type Open source software
Open source organizations nothing
region Unknown
deliverer game
intended for unknown
Recording time 2023-05-09

Software Introduction

WizardLM is a fine tuned 7B LLaMA model. It uses a large number of instructions with different difficulties to follow the dialogue to fine tune. The novelty of this model is that it uses LLM to automatically generate training data.

The WizardLM model uses a new method called Evol Instruct (a new method that uses LLM generation of human beings to independently batch generate open instructions of various difficulty levels and technical ranges to improve the LLM ability) to train through 70k computer generated instructions. This method generates instructions with different difficulty levels.

Evol Inspect uses the following five operations to expand prompts:

  • Add Constraint

  • deepen

  • Concretization

  • Add reasoning steps

  • Complex input

These operations are sequentially applied to the initial instruction to make it more complex, and the reply is generated by LLM.

 

Expand to read the full text

code

Gitee index of is
exceed Items for

comment

Click to lead the topic 📣 Post and join the discussion 🔥
No content temporarily
{{o.pubDate | formatDate}}

{{formatAllHtml(o.title)}}

{{parseInt(o.replyCount) | bigNumberTransform}}
{{parseInt(o.viewCount) | bigNumberTransform}}
No more
No content temporarily
Issued a question and answer
{{o.pubDate | formatDate}}

{{formatAllHtml(o.title)}}

{{parseInt(o.replyCount) | bigNumberTransform}}
{{parseInt(o.viewCount) | bigNumberTransform}}
No more
No content temporarily
No content temporarily
zero comment
three Collection
 OSCHINA
Log in to view more high-quality content
 Back to top
Top