✦ Claude-mini incoming. 》》》》 Also model doesn't work properly [!] RNN are commonly used upon sequential data, this model is a revised version of GeminiLM, and GeminiAI Gen 1.9m, where is rather uses probability than modifing into expected values, this result in better contextual understanding, but less performance, and more intensive. This model is untrained, will take a long time. Learn more about RNNs- https://en.wikipedia.org/wiki/Recurrent_neural_network --- I will be fixing the model for the size limit by cutting vocab size by half, and rounding up floats i am planning to make a VAE for text to image w/ this use turbowarp want to try generating text on one of my working models? - https://scratch.mit.edu/projects/1002139855/ - ---ver logs--- v.1 - release, untrained b - fixed some training issue, idk if it will work fully v.2 - fixed some derivatives calculations, diff intit v.3 - fixed a bunch of things in training, output compute, the model actually slightly learns now! Text Gen - Gemini Development Framework Series (GDFSfTG) (this might replace GeminiLM, if it outpeforms, and that will replace the official release of GeminiAI [gen1.9max])
