Language Diffusion Model Experiment w/ Sudoku
- 7 Devlogs
- 35 Total hours
I wanted to research whether Language Diffusion Models are more effective at solving puzzles which are not solved in a left to right fashion.
Added a demo on HF Spaces
.png)
This needs changes before we can consider it shipped. Review the Shipwright feedback, make updates, then ship again.
Feedback
Hello, your project has the following issues, please fix them for approval:
- Your demo is either hosted on an On-Activity Wake-Up hosting service, these take a very long time to load or it's hosted on a local tunneled server which can go down anytime or it's hosted on huggingface, which we do not allow or it's hosted on youtube, which we do not allow video demos., please switch to an Always-On hosting service such as Hack Club Nest, Railway and Vercel.
If you have any questions, DM @kaboom or create a ticket in #ask-the-shipwrights.
Shipwright walkthrough
This needs changes before we can consider it shipped. Review the Shipwright feedback, make updates, then ship again.
Feedback
Hello, your project has the following issues, please fix them for approval:
- Your demo is either hosted on an On-Activity Wake-Up hosting service, these take a very long time to load or it's hosted on a local tunneled server which can go down anytime or it's hosted on huggingface, which we do not allow or it's hosted on youtube, which we do not allow video demos., please switch to an Always-On hosting service such as Hack Club Nest, Railway and Vercel.
- Your Stardance project banner is either a logo or not relevant to your project, the banner needs to be of your projects running and working, please update it.
If you have any questions, DM @kaboom or create a ticket in #ask-the-shipwrights.
Shipwright walkthrough
Again I got feedback that the code and README looked too AI generated so this is Attempt 3 at restructuring and rewriting the README.
Idk what to put for the image so I put the README, which is now 100% written by me (I changed it to sort of match the Stardance format)
This needs changes before we can consider it shipped. Review the Shipwright feedback, make updates, then ship again.
Feedback
Hello, again.
Resetting your git commit history to hide AI usage, is not acceptable.
Comparing your current code to our previous review, we find that you have not changed much or anything for that matter, and most of the code remains unchanged.
Your README is still AI-generated.
If you have any questions, DM @kaboom.
Shipwright walkthrough
Writing code myself for the first time to refactor the codebase (jk I did write most of it myself the first time… but we obey the orders of the Shipwrights)
This needs changes before we can consider it shipped. Review the Shipwright feedback, make updates, then ship again.
Feedback
Hello,
Simply removing the comments and rewriting a few words in your README won't be accepted.
You need to re-write the code and README yourself.
If you have any questions, DM @kaboom or create a ticket in #ask-the-shipwrights
Shipwright walkthrough
Cleaning up code
This needs changes before we can consider it shipped. Review the Shipwright feedback, make updates, then ship again.
Feedback
Hello,
Your project codebase is almost entirely AI-generated with no human effort whatsoever, please re-write majority of your codebase including your README by hand for approval.
If you have any questions, DM @kaboom or create a ticket in #ask-the-shipwrights.
Shipwright walkthrough
This needs changes before we can consider it shipped. Review the Shipwright feedback, make updates, then ship again.
Feedback
Hi arushg! Your project looks great; however, we only allow AI up to 30%, which seems to be exceeded in this project
If you have any questions, ask them in #ask-the-shipwrights or in DM to @mathias
Shipwright walkthrough
Ship #1
Pending review
Generating output example
Cleaned up training and eval code
Results from my experiment – with fine-tuning, diffusion models can beat pre-trained models almost 9 times larger, and autoregressive fine-tuned models 1.5 times larger
