LoganDark
LoganDark
Sorry, let me try to clarify. This library currently contains routines for converting between sRGB8 uints and linear-RGB floats. I'm asking if converting between sRGB8 and Linear RGB8 (or RGB16)...
> Hmm, linear RGB8 is very low precision, to the point where it tends to be a mistake Yeah, converting to Linear RGB8 would completely clobber the whole point of...
> IME you'll have a very hard time beating array lookup for something like this though. It doesn't SIMD well, but if the table is in the cache (and for...
I have uploaded some quantized RWKV-4-Raven models to HuggingFace at [`LoganDark/rwkv-4-raven-ggml`](https://huggingface.co/LoganDark/rwkv-4-raven-ggml/tree/master). Conversion took about 2 hours, and upload took about 24 hours and 500GB of disk space. At the time...
> I hope it won't be too breaking :) It should be possible to reimplement the current API in terms of the new one, in order to keep compatibility with...
> he abstract claims the model can be "formulated as a Transformer" "which parallelizes computations during training". I interpret this as _perhaps_ meaning that you train the model using a...
> How do you reconcile this with the perception that "it needs a serial scan to update attention scores wkv" (4.2) this is for RNN mode. RNN is run one...
Also just thought of this. Would be nice :)
> Please provide a [small code sample that reproduces the issue](https://minimum-reproduction.wtf/). will try later.