@romainlanz Je pense que รงa peut toujours รชtre intรฉressant avec l'รฉvolution de l'IA de voir les workflows de chacun. Puis justement quand l'IA fait des erreurs, d'aller plus en dรฉtails sur pourquoi c'est pas le plus "opti/structurรฉ/perf" et reguider l'IA via des ppt "https://t.co/w4xQZsfEGG"
@d4m1n It prevents context bloat from large llm.txt files and having multiple MCP servers. For me, it acts like a smart docs proxy that helps the agent fetch exactly what it needs at the right moment. I'm still experimenting with it. I may be wrong but it seems to work for now
@d4m1n I use a specific stack, I have a cursor rule that routes all documentation queries through context7. Before I had to manually paste doc links or use cursor documentation (but it's not up to date, cannot scrape properly some documentation + need to refresh to keep it up to date)
๐ Meet Your New Virtual Characters on https://t.co/r9ZcMttfzR! ๐
Create unique characters by choosing traits like gender, age, ethnicity and more. Generate images and pick your favorite, we handle the rest by creating a hight-quality dataset using Flux 1.1 [PRO]๐ค
Try it now
๐ https://t.co/bFARfDHrNX just got even better! ๐
Now you train a model of yourself or enhance your product marketing with personalized, professional-quality images!
Perfect for elevating your social media presence and brand appeal. ๐จ๐ธ
It takes only 5 minutes to get a model trained ๐
@fofrAI@ostrisai Another thought that came to my mind, is if we use auto-captioning it would be nice to have the caption generated for each image in the output folder ๐
@araminta_k@fofrAI@ostrisai Oh I see, thanks for the feedback I was trying to understand more about this parameter but haven't tested it yet on my side
@fofrAI@ostrisai Just saw @ostrisai mentioning yesterday on his discord that he will test for "content_or_style" combined with linear_timesteps, probably better to wait till his feedback
@fofrAI@ostrisai Few ones under the "train" section such as "content_or_style", "linear_timestep" and maybe "train_text_encoder" (not sure yet if this one has some impact on the caption set on the images, as experimented here lol: https://t.co/WTj4n8Ogw1)
Everyone who said captions don't do anything for Flux is wrong because I captioned a dog as a "cat" and it was ONE picture, and now the model is beautiful except whenever I prompt cat or kitten it gives me dogs.
Also shoutout to @comfydeploy for the awesome service.