Content pfp
Content
@
0 reply
0 recast
2 reactions

eggman 🔵 pfp
eggman 🔵
@eggman.eth
gm all, def been less trading and more dev time since the market rinse for eggy recently, flux dropped their gen ai image model - and I’m having a blast building out a dev pipeline for it. Hoping it’ll culminate in something v cool under the /imgnai banner to show off on /warpcast and /base in the near future!
9 replies
1 recast
17 reactions

Niko pfp
Niko
@nikomfer.eth
sounds fun! what is the pipeline being used for? 100 $DEGEN
2 replies
0 recast
2 reactions

eggman 🔵 pfp
eggman 🔵
@eggman.eth
So far I’ve set up systems for mass image outputs with varying prefixes, guidance, and steps across a set list of prompts. Coupled with that then, I’ve built some new tooling for image gathering/transform & auto-tagging, which I’m using to build datasets for LoRA’s and fine-tunes. Lastly, a telegram bot frontend for interfacing with the live server and testing out the model+lora while it’s running (or inviting others to!) Should hopefully see us hammer out some serious quality LoRA’s and fine-tunes!
2 replies
0 recast
2 reactions

Niko pfp
Niko
@nikomfer.eth
the part of LoRA's went above my head, but i love how you explained everything technically! I'm reading that LoRA's is a method for fine tuning LLMs. Does this also work for images?
0 reply
0 recast
1 reaction