Skip to content

Himank-J/GPT2-CustomTrained

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Training GPT-2 from Scratch using Tiny Shakespeare dataset

In this repo you can find the Tiny Shakespeare dataset (input.txt) and 17 files each representing a step in training GPT-2 model from scratch. In files train_get2-1 through train_get2-8 we setup our code for training. From train_get2-9-speedup-1 thorugh speedup-9 we implement different techniques to speed up our training process.

Objective

Target - loss less than 0.099

Training Params -

  • Epochs - 6000
  • Batch size = 8
  • Number of tokens = 1024

Results

As seen in below image, after runnning for 6000 epohcs with batch size 8 and tokens 1024, we are able to get our loss to 0.06

Screenshot 2024-06-27 at 5 28 10 PM

App

Link to app - https://huggingface.co/spaces/HimankJ/GPT2_CustomTrained

Screenshot 2024-06-28 at 8 54 01 PM

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages