terra-quantum-public / tqcompressedgpt2 Goto Github PK
View Code? Open in Web Editor NEWLicense: Other
License: Other
Hey,
I am struggling a bit to understand how the embedding matrix is constructed back from the following factors:
If I understood right from the proposed decomposition in the paper, B should have a shape [1,4]? Simply put, how do I construct back the "approximate" embedding matrix, from these factors?
Also, it appears that the other paper (https://arxiv.org/abs/2110.08152) do not count the output matrix (the softmax) in the parameter count, (which would technically make the model of size 120M?!). And I noticed in your checkpoint a parameter named lm_head.weight
, is it the output layer? and if yes, why don't you count it?
Looking forward to you clarifications, and thanks for providing the code!
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.