Viewing profile — irt24

"GGUF quantization" is the most popular tech stack for quantizing Llama-like models for CPU. But the documentation is very sparse, and the maintainers made it clear that writing a …

We're working on a similar feature (that we call Textify) that is more production-ready: https://storia.ai/textify

Unfortunately the default prompt start on HF ("My name is X...") is completely out of domain from the training set (i.e., nobody starts an image prompt like that) and the model goe…

To build this model, we scraped 4 weeks' worth of user-issued text prompts from Midjourney's public Discord server. We released the dataset on Kaggle: https://www.kaggle.com/datase…

I think this view is heavily influenced by the fact that the writer is a project manager -- I can see how having full visibility into all communications could make their particular…

Resistance is futile. Humans will adapt, as we've been doing for millenia. Apparently people also panicked when books were invented, assuming that future generations would be compl…