Generation result

[mosvlad]

Generation result looks not so good... For example:

prompt = 'monkey eat banana'

Maybe someone get better generation result?

[JavSensei]

That model is a quick training in colab, don't expect good results

[bitcoin5000]

where get good model?

[nikuson]

Everything will be later, be patient

[bitcoin5000]

and how long wait? im sick of dalle2 waiting already

[Desm0nt]

and how long wait? im sick of dalle2 waiting already

Training models of this scale is not an easy task, not fast at all, and most importantly, not cheap. Unless you are Google or Amazon with a bunch of data centers with flagship GPUs and TPUs - training such models can cost hundreds of thousands of USD. And the more you try to reduce costs, the slower will be the training process (on weaker GPUs in smaller quantities with less optimal parameters).

And we do not even mention the preparation of dataset for all this goodness (LAION probably already have good enough).

And in this case, the model so far has only been recreated by Google's Paper. It will still be changed by trial and error (with repeated attempts at training and playing with parameters)

So, if you are so impatient - go and buy 1 month in midjourney - their network is good enough to get very interesting and beautiful results, and the basic package will cost you no more than 1 month of Colab Pro.